Search is not available for this dataset
article stringlengths 4.36k 149k | summary stringlengths 32 3.35k | section_headings listlengths 1 91 | keywords listlengths 0 141 | year stringclasses 13
values | title stringlengths 20 281 |
|---|---|---|---|---|---|
The mainstay of current schistosomiasis control programs is mass preventive chemotherapy of school-aged children with praziquantel . This treatment is delivered through school-based , community-based , or combined school- and community-based systems . Attaining very high coverage rates for children is essential in mass schistosomiasis treatment programs , as is ensuring that there are no persistently untreated subpopulations , a potential challenge for school-based programs in areas with low school enrollment . This review sought to compare the different treatment delivery methods based both on their coverage of school-aged children overall and on their coverage specifically of non-enrolled children . In addition , qualitative community or programmatic factors associated with high or low coverage rates were identified , with suggestions for overall coverage improvement . This review was registered prospectively with PROSPERO ( CRD 42015017656 ) . Five hundred forty-nine publication of potential relevance were identified through database searches , reference lists , and personal communications . Eligible studies included those published before October 2015 , written in English or French , containing quantitative or qualitative data about coverage rates for MDA of school-aged children with praziquantel . Among the 22 selected studies , combined community- and school-based programs achieved the highest median coverage rates ( 89% ) , followed by community-based programs ( 72% ) . School-based programs had both the lowest median coverage of children overall ( 49% ) and the lowest coverage of the non-enrolled subpopulation of children . Qualitatively , major factors affecting program success included fear of side effects , inadequate education about schistosomiasis , lack of incentives for drug distributors , and inequitable distribution to minority groups . This review provides an evidence-based framework for the development of future schistosomiasis control programs . Based on our results , a combined community and school-based delivery system should maximize coverage for both in- and out-of-school children , especially when combined with interventions such as snacks for treated children , educational campaigns , incentives for drug distributors , and active inclusion of marginalized groups . ClinicalTrials . gov CRD42015017656
The Schistosomiasis Consortium for Operational Research and Evaluation ( SCORE ) was established in December 2008 to answer strategic questions about schistosomiasis control and elimination [1] . SCORE’s goal is to find answers that will help current and future schistosomiasis control program managers to do their job better , with its main focus being control of Schistosoma haematobium and S . mansoni infections in sub-Saharan Africa . This includes learning what approaches to controlling and eliminating schistosomiasis work best , and developing and evaluating new tools for program managers to use . SCORE research is intended inform efforts to gain control of schistosomiasis in high-prevalence areas , to sustain control and move towards elimination in areas of moderate prevalence , and ultimately to eliminate schistosomiasis . As part of the SCORE program , the present systematic review was undertaken to improve understanding of the factors that affect an individual’s adherence to treatment for schistosomiasis where large-scale mass drug administration ( MDA ) programs are implemented . Schistosomiasis is a disease caused by water-borne , parasitic trematodes of Schistosoma species that infect either the genitourinary tract or the intestines and liver . Sequelae of chronic Schistosoma infection include anemia , malnutrition , growth retardation , poor school performance , infertility , and potentially fatal complications such as portal hypertension , renal failure , and bladder cancer [2] . Globally , more than 290 million individuals are estimated to be infected , with another 600–780 million at risk [3 , 4] . The most prevalent species , Schistosoma haematobium , is found predominantly in sub-Saharan Africa and the Middle East , where it is transmitted by Bulinus species snails . S . mansoni is transmitted in Africa and South America via Biomphalaria species , and the third most prevalent parasite species , S . japonicum , is transmitted in East Asia and in the Philippines by Oncomelania species snails [1] . Because of SCORE’s focus on schistosome species found in sub-Saharan Africa , results for S . japonicum were not included in this review . Historically , schistosomiasis control has employed a number of methods that limit the impact of disease . Approaches have included molluscicides that target the parasite’s intermediate snail host [5]; interventions to provide clean water , sanitation , and hygiene [6]; and selective or mass treatment with anti-schistosomal agents [7 , 8] . In the past few decades , control efforts have primarily focused on the use of praziquantel for mass preventive chemotherapy of school-aged children ( SAC ) and other high-risk adult groups ( e . g . , fishermen , car washers , laundry workers ) in endemic areas , through the use of MDA [7 , 8] . A variety of MDA methods are currently used for schistosomiasis control . Delivery modes can broadly be divided into community-based and school-based programs , or a combination of the two . Community-based distribution strategies include house-to-house distribution , distribution from a health facility or other central location , or a combination of the two . Other program characteristics that may vary from study to study include: i ) which personnel distribute the drugs , ii ) how communities are educated and mobilized , and iii ) whether the anti-schistosomal programs are “vertical” or combined with community-based campaigns against lymphatic filariasis and onchocerciasis . With respect to school-based programs , most countries affected by schistosomiasis have significant percentages of SAC who are not enrolled in school , and not all children who are enrolled regularly attend school . From data collected between 2000 and 2006 , UNICEF reported that in Eastern/Southern Africa , only 70% of children attended primary school , and that in West/Central Africa , only 62% of children attended primary school [9] . If school-based programs leave these sizeable populations of un-enrolled children persistently untreated , this will significantly impede the success of schistosomiasis control and elimination programs . World Health Organization ( WHO ) guidelines recommend that in all endemic areas , at least 75% of children are treated either annually ( in high endemicity zones with ≥ 50% SAC prevalence ) , every two years ( in moderate endemicity zones having 10–49% SAC prevalence ) , or upon entering and leaving primary school ( in low endemicity zones having < 10% SAC prevalence ) [8] . Modeling has suggested that , due to the high risk of infection resurgence upon cessation of control programs [10–12] , an even higher level of population coverage may be required for the effects of preventive chemotherapy to be sustained . For example , it is estimated that in “high-risk” villages ( >50% SAC prevalence at baseline ) , greater than 90% coverage of SAC will be required in order to achieve 3–4 subsequent years of sustained low prevalence ( i . e . , at < 10% ) [13] . With these coverage target numbers in mind , an understanding of what features contribute to the most successful MDA programs is essential . This review sought to assess the overall SAC coverage rates as well as coverage rates of non-enrolled children obtained by MDA through respective school , community , or a combination of school and community delivery systems . We discuss the factors believed to be associated with high or low coverage rates , and summarize different national and regional programs’ suggestions for improving overall SAC coverage .
The data used in this project were aggregated , anonymized data from previously published studies; as such , this study does not constitute human subjects research according to U . S . Department of Health and Human Services guidelines ( https://www . hhs . gov/ohrp/regulations-and-policy/guidance ) . The protocol for this study was developed prospectively by the authors . It was registered and published in the International Prospective Register of Systematic Reviews ( PROSPERO ) online database , https://www . crd . york . ac . uk/PROSPERO/#index . php , number CRD42015017656 , on 06 April 2015 . A copy of the registered protocol is found in S2 File . In order to evaluate the effectiveness of MDA programs for schistosomiasis , we aimed to include any studies published before October 2015 that reported quantitative or qualitative data about coverage rates of SAC for MDA with praziquantel in sub-Saharan Africa , or in other low- or middle-income countries in the middle-east or South America . Articles in English or French were included . Relevant ‘gray’ literature ( project reports , white papers ) was also obtained and reviewed when possible . The National Library of Medicine’s MEDLINE , along with Elsevier’s EMBASE , Google Scholar , African Journals Online , and Web of Science were used to identify published studies . The reference lists of these identified papers were then used to search for additional unindexed reports . When relevant , attempts were made to contact authors to clarify information and to obtain unpublished literature . Literature searches were performed using combinations of the following key words: ‘schistosomiasis’ , ‘mansoni’ , ‘haematobium/haematobia’ , ‘prevention’ , ‘control’ , ‘compliance’ , ‘adherence’ , ‘uptake’ , ‘coverage’ , ‘non-enrolled’ , ‘non-attendance’ , ‘community-directed treatment’ , and ‘MDA’ . A search log containing citation information for each study was kept using Microsoft Excel . The titles and abstracts of all studies identified in the literature search were screened for relevance to the present systematic review . The full texts of the studies not excluded after the screening phase of the study were obtained from online or library sources when possible . These reports were then reviewed in full to determine eligibility for the present systematic review . Data from the included papers was abstracted and indexed in a Microsoft Access database . Besides citation information and year of publication , information was collected on study location , study design , MDA delivery method ( community-based , school-based , or combined ) , and population characteristics . The fraction of all eligible SAC treated , the fraction of non-enrolled versus enrolled SAC treated , and the methods used to measure the number of eligible and treated individuals ( e . g . censuses , household surveys , drug distributor registers ) were also recorded . When a study reported coverage rates using both distributor registers and household surveys , the household survey coverage rate was used . Qualitative data extracted included factors positively or negatively associated with success of MDA , as well as reported side effects of the medication . Percentage of SAC coverage ( including enrolled and non-enrolled children ) was the primary measure used to compare methods of delivery of MDA; when available , coverage specifically of non-enrolled children was also compared between studies . Due to the heterogeneity in which studies reported MDA coverage , it was necessary to use multiple strategies to generate data that would be comparable for analysis: i ) when an article reported an aggregate coverage rate for SAC and adults , the coverage rate of SAC was assumed to be at least that of the reported aggregate coverage rate; ii ) when an article reported coverage rates by individual district , the coverage rate extracted was a weighted average of the individual district averages based on targeted population size; iii ) when coverage rates were reported for multiple years , these were averaged and population-based weighting was again used , where appropriate , in calculating these blended averages . Due to the heterogeneity of study sizes as well as methods of measuring coverage , it was not possible to perform formal meta-analyses , i . e . , no summary point estimates or confidence intervals have been calculated , nor have statistical comparisons been made among or between distribution methods . Qualitative analysis was performed to determine other factors , apart from method of delivery , that were commonly cited as affecting coverage rates in the included studies . The studies reviewed in this paper were mostly not protocol-driven randomized trials . They were primarily one-group observational , or pre- vs . post-modification evaluation trials , and they generally did not report on aspects of the study that might help to determine risk of bias . Publication bias could potentially have influenced data availability if reports of innovative MDA strategies were published only if successful , or if standard implementation programs were reported only if unsuccessful in reaching coverage targets .
Fig 1 shows a flow diagram of the study selection process for this systematic review . 623 references were identified through systematic database searches and 39 more through reference lists and personal communications . After screening of titles and abstracts , 114 papers were selected for full-text review; however , the full text could only be accessed for 111 . S1 Text contains a list of the studies considered for full review . Reasons for exclusion ( n = 89 ) included: did not measure or report coverage ( 31 ) , MDA not employed ( 29 ) , modeling study/no intervention ( 15 ) , not schistosomiasis-related ( 6 ) , duplicate report ( 3 ) , only adults included ( 2 ) , conflicting data ( 1 ) , inadequate age category breakdown ( 1 ) , or other ( 1 ) . Twenty-two studies were included in the systematic review . Fourteen of those studies ( Table 1 ) reported coverage rates for populations of , or including , SAC . Of those fourteen , five reported on combined community- and school-based delivery , four reported on community-based delivery , two reported on school-based delivery , and three compared school-based and community-based delivery . Three of the quantitative studies also reported separate coverage rates for enrolled and non-enrolled children . Several of the studies reporting overall coverage rates , plus eight others ( Table 2 ) , provided qualitative analysis of individual , community , or programmatic factors influencing coverage . By delivery strategy , the highest SAC coverage rates were achieved with combined community- and school-based delivery ( 78% to 95% , median 89% ) [14–18] . The second highest rates were achieved with community-only delivery , including the community arms of the comparative trials . ( 53% to 85% , median 72% ) [19–22 , 25–28] . The lowest SAC coverage rates were achieved with school-only delivery , including the school arms of the comparative trials ( 28% to 81% , median 49% ) [23–28] . These results are shown in Fig 2 . Three studies reported separate data breakdowns on coverage rates for enrolled and non-enrolled children . In combined community/school-delivery programs , coverage results for non-enrolled children followed the trends for enrolled SAC , yielding at 90% coverage for non-enrolled SAC vs . 95% for enrolled SAC in Yemen [14] and 88% ( non-enrolled ) vs . 96% ( in school ) coverage in Burkina Faso [15] . In the comparative trial of community-based vs . school-based implementation in Tanzania , the community-based arm reached 80% non-enrolled SAC coverage [27 , 28] , whereas the school-based arm of the same trial reached only 59% coverage of non-enrolled-SAC [27 , 28] .
The first aim of this systematic review was to examine the coverage rates achieved with different delivery approaches during mass drug administration ( MDA ) against schistosomiasis among school-age children ( SAC ) . Among the different observational and pre-treatment/post-treatment studies included in this review , combined community- and school-based delivery achieved the highest median SAC coverage , followed by community-only delivery , then school-only delivery [14–24] . The three comparative studies found similar results , with community-based delivery either outperforming or roughly equaling school-based delivery in every case [2 , 25–28] . Few studies reported specifically on coverage of non-enrolled SAC , but those that did found that offering community-based treatment ( either alone or as part of a combined program ) resulted in substantially better coverage for non-enrolled SAC than school-only treatment did [14 , 27 , 28] . However , even with community treatment , coverage was consistently lower among non-enrolled children compared to enrolled children [14 , 15 , 27 , 28] . WHO guidelines recommend an SAC coverage rate of at least 75% in schistosomiasis-endemic areas [8] . Each of the five studies that used combined community and school distribution were able to achieve that coverage level [14–18] . On the other hand , 75% coverage was not consistently attained across the studies that used community-only or school-only delivery systems [19–28] . This suggests that for future MDA programs , combined treatment may be the most reliable choice for achieving an acceptable coverage rate . In future studies , it will be especially important to look closely at MDA coverage for non-enrolled versus enrolled SAC , in order to ensure that all children are equitably targeted . The trend of higher coverage rates in community-based programs versus school-based programs might be explained in part by large numbers of non-enrolled children missing out on school-based distribution . If elimination of transmission is to be the eventual target for control , it is essential that there remains no persistently untreated sub-population within the community . The second aim of this review was to identify qualitative factors affecting MDA coverage rates . On the recipient level , successful distribution was hindered by fear of medication side effects [14 , 20 , 23 , 29 , 30 , 32 , 36] and inadequate education about the need for MDA , despite the absence of overt symptoms [14 , 23 , 24 , 29 , 30 , 32 , 35] . Some of praziquantel’s gastrointestinal side effects may be minimized by giving the drug after food , and provision of porridge or other snacks may yield improved MDA uptake [14 , 34 , 36] , and this aspect deserves greater clinical study . With regard to the efficacy of drug distributors , absence of financial or material incentives [17 , 24 , 26 , 29–32 , 36] and unrealistically high ratios of recipients to drug distributors [14 , 17 , 20 , 26 , 31 , 32 , 35] led to decreased motivation and staff attrition . Within communities , political tensions and differences among minority groups resulted , at times , in unequal distribution rates [21 , 22 , 26 , 31] . On the other hand , several studies showed that snacks for recipients [34] , t-shirts and small monetary rewards for teachers [24] , and educational media campaigns about schistosomiasis [33] have led to more successful MDA programs . A number of the issues affecting delivery of schistosomiasis control , namely: i ) limited disease knowledge , ii ) fear of side effects , iii ) unequal age and gender uptake , iv ) lack of drug-distributor motivation and v ) local political effects are also common to other MDA programs for control of NTDs . Each of these problems was noted by Krentel and colleagues [37] in their recent systematic review of factors affecting MDA participation in lymphatic filariasis elimination programs . With this evidence in mind , it appears that small , targeted program modifications can have a meaningful positive impact on program success . Table 3 includes a listing of suggested strategies for increasing coverage rates . Implementation of these suggestions in future studies would help determine which of these adjunctive strategies , singly or in combination , lead to the most significant improvements in coverage . In terms of limitations , our systematic review was limited by the small number of studies containing detailed information about schistosomiasis MDA coverage in SAC , especially with regard to non-enrolled children . Even within the eligible studies , heterogeneity in design , size , and methods used to measure and report coverage made it difficult to compare results or develop over-all estimates of the impact of individual program factors . As we have noted , there could potentially be bias in study reporting , which might then influence our current assessment . In order to compare the efficacy of different MDA control strategies more definitively , additional randomized control trials are needed , along with more consistent reporting of target population sizes and coverage rates . Overall , this study provides a systematic first look at how to design the most effective schistosomiasis MDA programs . From a quantitative perspective , the limited data suggest that the best means to maximize SAC coverage for both enrolled and non-enrolled children is to use a combined community-based and school-based approach . Qualitative program features that are expected to maximize coverage include provision of food for treated children , educational campaigns about schistosomiasis and its treatment , increased CDD training and incentivization , and active inclusion of marginalized populations . In the push towards elimination of schistosomiasis , consideration of these factors will be essential in the development of future MDA programs . However , beyond attaining high coverage , it is also crucial to continually assure treatment efficacy through follow-up monitoring and evaluation of infection prevalence and intensity [38] , and the corresponding Schistosoma infection-associated morbidities [39] . Optimal design of schistosomiasis surveillance strategies remains a very active area of study in operational research [40] | Schistosomiasis is a chronic inflammatory condition , caused by parasitic flukes , that affects over 290 million people worldwide . Consequences of infection include anemia , stunted growth , liver abnormalities , and subfertility . Currently , the main approach to schistosomiasis control involves mass preventive treatment of school-aged children in endemic areas . The treatment , praziquantel , can be distributed through school-based , community-based , or combined school- and community-based systems . The first part of this systematic review compared the three delivery modes and found that combined delivery resulted in the best overall coverage of school-aged children , with community-only delivery the next-best approach . School-only delivery not only had the lowest overall coverage , but especially fell behind in targeting children not enrolled in school . As a whole , these results support the more frequent use of a combined approach to delivery in order to achieve the highest coverage rates and ensure that out-of-school children are not left persistently untreated . In the second part of this review the qualitative factors affecting program success were examined . The results indicate that overall treatment coverage can be improved via small interventions , such as snacks for participating children to reduce drug side effects , educational campaigns about schistosomiasis , incentives for drug distributors , and active inclusion of marginalized groups . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"children",
"medicine",
"and",
"health",
"sciences",
"education",
"sociology",
"tropical",
"diseases",
"social",
"sciences",
"teachers",
"parasitic",
"diseases",
"health",
"care",
"age",
"groups",
"pharmaceutics",
"neglected",
"tropical",
"diseases",
"pharmacology",
"fa... | 2017 | Systematic review of community-based, school-based, and combined delivery modes for reaching school-aged children in mass drug administration programs for schistosomiasis |
Recent work has shown that much of the missing heritability of complex traits can be resolved by estimates of heritability explained by all genotyped SNPs . However , it is currently unknown how much heritability is missing due to poor tagging or additional causal variants at known GWAS loci . Here , we use variance components to quantify the heritability explained by all SNPs at known GWAS loci in nine diseases from WTCCC1 and WTCCC2 . After accounting for expectation , we observed all SNPs at known GWAS loci to explain more heritability than GWAS-associated SNPs on average ( ) . For some diseases , this increase was individually significant: for Multiple Sclerosis ( MS ) ( ) and for Crohn's Disease ( CD ) ( ) ; all analyses of autoimmune diseases excluded the well-studied MHC region . Additionally , we found that GWAS loci from other related traits also explained significant heritability . The union of all autoimmune disease loci explained more MS heritability than known MS SNPs ( ) and more CD heritability than known CD SNPs ( ) , with an analogous increase for all autoimmune diseases analyzed . We also observed significant increases in an analysis of Rheumatoid Arthritis ( RA ) samples typed on ImmunoChip , with more heritability from all SNPs at GWAS loci ( ) and more heritability from all autoimmune disease loci ( ) compared to known RA SNPs ( including those identified in this cohort ) . Our methods adjust for LD between SNPs , which can bias standard estimates of heritability from SNPs even if all causal variants are typed . By comparing adjusted estimates , we hypothesize that the genome-wide distribution of causal variants is enriched for low-frequency alleles , but that causal variants at known GWAS loci are skewed towards common alleles . These findings have important ramifications for fine-mapping study design and our understanding of complex disease architecture .
While association studies have been successful in finding a large number of significant variants for many complex traits , they have individually explained relatively little of the total heritability , motivating analyses that seek to identify this so-called “missing” heritability [1]–[3] . One hypothesis is that additional causal variation is present at the known GWAS loci but not fully quantified by individual GWAS markers [1] , [2] , [4]–[7] . This scenario may arise if the true causal variant is poorly tagged by any single GWAS marker [8] or if multiple independent causal variants exist at the locus [9] . In this case , the variance explained by the most-significant marker would only provide a lower bound on the local contribution , and some of the “missing” heritability would in fact be hidden at the previously discovered loci . If we consider “local” heritability to be the measure of aggregate variance from all causal variants at a locus , its quantification is an important step towards fully understanding the contributions made by association studies . Moreover , estimating components of local heritability indirectly from the vast amount of GWAS-level data already available would enrich our current understanding of complex disease architecture and provide insights into further study-design for post-GWAS fine-mapping studies . Here , we investigate methods for inferring components of local heritability at previously identified GWAS loci . As study sample sizes continue to grow , researchers have focused on quantifying the amount of heritability explained by individually significant single-marker associations [4] , [10]–[14] . In well-powered GWAS , one can also look for secondary variants that are conditionally independent of the leading SNP and estimate the joint contribution to phenotype . This conditional analysis has recently proven effective in GWAS for height [4] , [15] , [16] and multiple case-control traits [17] , where a handful of loci were found to contain independent secondary associations . This strategy inherently focuses on a small number of independent markers and the outcome strongly depends on power to detect the primary association as well as any secondary variants . Such complexities make it difficult to compare this estimate across different studies and disease architectures . With additional resources , one can fine-map implicated loci using denser genotyping or sequencing platforms and look for more strongly significant markers . Recent studies involving re-sequencing around known GWAS-associated regions have identified additional variants explaining significant heritability in several complex traits [5] , [18]–[20] . Looking beyond individual traits , a fine-mapping study of Celiac disease examined loci associated with other autoimmune diseases and nearly doubled the number of significant associations [21] . This approach can leverage the shared genetic architecture observed in some groups of related traits [22]–[24] . Still , such studies have not always yielded significant associations; a targeted re-sequencing analysis of Type 2 Diabetes did not yield any additional variants beyond what was known from GWAS [25] and recent work with dense genotyping did not uncover significant additional heritability at known loci for Type 2 Diabetes and Coronary Artery Disease [20] . Overall , these findings motivate methods that can infer components of additional local heritability using available GWAS data to guide fine-mapping analysis for identifying additional risk variants . We propose to address this challenge by making use of all observed markers in a variance-component analysis , which optimizes a single measure of effect-size over a sample relatedness matrix . When sample relatedness is computed directly from the observed markers - referred to as the genetic relatedness matrix ( GRM ) - this variance-component can be used to infer the narrow-sense heritability explained by these markers . This measurement of narrow-sense heritability represents the aggregate effect of all causal variants observed or tagged in the data assuming additive , normally-distributed effect sizes . Recent work in variance-components analysis has shown that the contribution of all genotyped SNPs and any markers in LD with them , denoted , can be estimated directly from large-sample GWAS data in this way [26]–[29] . Similarly , our aim is to apply the variance-component model locally , by constructing the GRM from all typed SNPs at known GWAS regions and estimating the corresponding local . The excess of this quantity over the variance explained by known associations provides a lower bound on additional heritability at the locus . Uniquely , this method allows the analysis of loci that have no known association in the focal trait but have been associated with other related traits , quantifying sources of missing heritability implicated by shared disease architecture . In this study , we apply these methods to both simulated and real phenotypes . Using simulations involving real genotypes , we find that LD between typed markers can significantly bias the estimate and propose a correction to the GRM calculation , which we compare to a recently proposed approach [30] . In local analysis , we observe higher estimates of heritability with the adjusted variance-component strategy compared to traditional association and conditional analysis , particularly when the locus harbors multiple causal variants . Importantly , our LD residual correction ensures these statistics are not inflated under the range of disease architectures considered ( unlike the correction of [30] ) . We estimate local at known loci for nine common diseases finding a significant average increase vs . the variance explained by known associations , with individually significant increases for three of the traits . We also estimate local heritability at loci identified only in other related traits , showing significant enrichment in autoimmune disease for within-trait heritability at cross-trait loci . For RA , we analyze dense genotypes from samples typed on the ImmunoChip data as part of the Rheumatoid Arthritis Consortium International ( RACI ) . This significantly larger sample-size and deep genotyping empowers us to provide precise estimates on the significant increases in local heritability within RA and across non-RA autoimmune traits . Our results have important implications for fine-mapping study-design as well as the broader understanding of disease architecture and allelic heterogeneity .
Our fundamental goal is to explain as much of the local heritability as possible without upward bias . We consider four different estimators with unique individual properties: , the variance explained by the single most associated SNP at a locus , computed directly from the effect-size of a univariate regression; , the variance explained by a conditional linear model of significant SNPs constructed by step-wise regression over all SNPs in the locus as described by [15]–[17]; , the heritability inferred with a standard variance-component constructed from all SNPs in the locus; and , the heritability inferred with an LD-residual adjusted variance-component constructed from all SNPs in the locus . The LD adjustment is crucial in scenarios where LD patterns that are systematically different at causal variants can distort the observed sample relatedness and bias traditional estimates of , as previously demonstrated by [30] . Our proposed correction uses linear regression to transform each SNP into an “LD residual” of any correlated preceding markers and construct the GRM from these residuals . We compare this correction to LDAK , the re-weighing solution of [30] , as well as other strategies ( see Methods ) . We first analyze the genome-wide heritability explained by genotyped SNPs in the nine WTCCC1 and WTCCC2 traits ( Table S1 ) . Figure 1 shows the results of this analysis for unadjusted and LD-adjusted estimates performed over genotyped and genotyped+imputed SNPs ( 2 . 1 million 1000 Genomes [31] SNPs on average; see Table S1 ) separately . Results are shown on the observed scale . ( Results on the liability scale are provided in Figure S4; all numerical values are provided in Table S14 , S15 . ) We note that stringent quality-control is imperative for heritability analysis , where many small artifacts can compound into significant inflation of the genome-wide estimate [23]; this effect can be exacerbated by LD-adjustment methods , which will tend to promote low frequency variants that may be especially prone to QC issues . As in other studies [23] , [30] , we use a series of highly conservative QC filters to stem this problem , at the cost of filtering out many potentially informative markers ( see Methods ) . The absence of any significant false heritability between the two control cohorts , particularly after LD-adjustment , indicates that genotyping artifacts are unlikely to be substantial ( Figure 1 ) . We note that in the presence of strong artifacts [30] , propose an elegant solution of estimating SNP weighing scores from an independent population , and a similar strategy can be applied to the LD-residual adjustment . For all traits we see that the LD-adjusted estimate from typed SNPs is higher than the corresponding unadjusted estimate , with an average of for genotyped SNPs . Previous work has shown the standard estimate to be robust when the trait is infinitesimal , i . e . where all SNPs are causal with normally distributed effect-sizes [32] , [33] . However , as demonstrated in our simulations and in [30] , non-infinitesimal traits with systematically less LD between rare and low-frequency variants will underrepresent those variants in the un-adjusted kinship , resulting in deflated estimates when a majority of the causal variants are low-frequency ( Figure S1 ) . The increase in adjusted estimates on real data therefore implies a genome-wide genetic architecture for these traits that is generally shifted towards low-frequency variants . As in our simulations , the effect of LD-adjustment is even stronger when imputed SNPs are included ( more on average , comparing dark-green to light-green bars ) , demonstrating the downwards bias introduced by an abundance of imputed markers without LD adjustment . Indeed , without adjustment , all of the traits exhibit lower after imputation . Interestingly , even though imputation increases the total number of markers by , the adjusted estimate from imputed SNPs is , on average , only higher than the corresponding estimate from genotyped SNPs . Because the LD adjustment effectively removes any new SNP that is a linear combination of nearby SNPs , this would be consistent with imputation providing information similar to such linear combinations [34] . This is further supported by the fact that the sum of LD-adjusted SNP variances ( roughly corresponding to the independent number of SNPs ) for imputed SNPs was only higher than that of typed SNPs . These findings do not minimize the utility of imputation for mapping , where individual effect sizes are important , but does imply that imputed variants are not explaining dramatically more missing heritability . Based on these findings and our previous simulations with imputed variants , we restrict our subsequent variance-components analysis to the genotyped data only . Next , we infer the amount of local around the GWAS loci for the nine traits and compare to the corresponding and values ( Figure 2 , Table S16 ) . When computing the increase in ( and its statistical significance ) , we always account for and the local expectation , i . e . the increase that would be expected by chance based on the total genome-wide and the fraction of genome covered by the variance-component ( see Methods ) . Across all the nine traits we find a consistent excess of local heritability , with an average increase of over the local expectation ( combined ) . These results were consistent with the LDAK-based adjustment , which had a mean increase of ( Table S18 ) . P-values were computed using a z-test and consistent to different definitions of ( see Methods ) , but an analysis involving a comparison to random regions of the genome also produced similar results ( see Methods , Table S17 , Figure S5 ) . Three of these traits ( CD , UC , and MS ) show individually significant increases ( ; ; and respectively ) . The regression-based analysis of jointly significant markers ( ) yields an average of more heritability than . In instances where there are multiple known associations at a locus , only the leading SNP is included in but all of the known associated SNPs are automatically included in , demonstrating that previously known locus heterogeneity still does not explain as much heritability as the estimate . On average , these loci are explaining 11% of the genome-wide with 1 . 1% of the genome . Interestingly , the estimate with no LD-adjustment also yields increased local heritability for all phenotypes with an even higher average increase ( Table S18 ) . Given that our simulations show an increase in unadjusted estimates only when the underlying causal variant is common ( Table 1 ) , this increase in real data suggests that most causal variation in these GWAS loci originates from common causal variants ( in contrast to the rest of the genome; see above ) . The presence of significant additional heritability in individual traits raises the question of whether it is coming from a single poorly-tagged causal variant or multiple independent causal variants . In our previous simulations , an increase in local heritability is not expected under the single causal-variant model and the ratio of to has a direct relationship to the number of causal variants . For the WTCCC2 data , a single rare or common untyped causal variant is expected to yield an of and , respectively ( Table 1 C , D ) . Both are lower than our observed average of in real data , and much lower than significant increases of and in UC and MS ( Table S16 ) . These results are therefore unlikely to arise simply due to all loci harboring a single poorly-tagged causal variant , with the point estimate of 1 . 29 indicating a likely architecture of 2–3 causal variants at the average locus . However , we caution that the variance of this ratio observed in simulations is very high ( for example , 18% of the single common causal simulations have a local increase greater than 1 . 29 ) , making it difficult to reject the single-causal variant hypothesis at this sample-size . From our previous power estimates ( Table S13 ) , we observe that at a sample-size of 15 , 000 power to detect multiple causal variants approaches 100% , allowing us to distinguish between these two scenarios . We note that some of the GWAS loci we analyzed were genome-wide significant in the WTCCC data and could potentially exhibit inflated effect-sizes due to winner's curse if discovered in this cohort . However , because the heritability from variance-components and GWAS SNPs are inferred in the same data , we expect any effect-size inflation to impact both estimates equally , making our relative comparisons robust even in the presence of biases . In light of this and the small fraction of such loci actually present ( 8% averaged over the 7 WTCCC1 traits ) we do not believe winner's curse to have had an impact on these results . Recent analyses of multiple phenotypes have demonstrated significant correlations in genetic architecture for certain groups of related traits [23] , [24] , [35] , [36] . Unique to the local variance-components approach , we can also compute components of heritability at known GWAS loci from multiple related traits without having genotypes for those traits . This measure provides an estimate of the additional variation that would be explained by fine-mapping loci associated with one trait within the affected samples of another; for example , analyzing known Ulcerative Colitis loci in a study of Crohn's Disease . We expect this to be informative when the traits have correlated genetic architectures , with causal variants that only reached statistical significance in one trait potentially explaining heritability in the other . One example of such related traits is the class of autoimmune disorders , which are known to have a shared disease architecture as well as many instances of overlapping GWAS loci [22] , [37]–[40] . For each of the nine traits , we consider the amount of heritability explained by loci that were previously associated to one or more other autoimmune diseases but not to the focal trait . By definition , the for these loci is zero , and so we compare to the local expectation , i . e . what would be expected by chance from the genome-wide and locus size ( see Methods ) . As with all other analyses , we specifically exclude the MHC for all autoimmune diseases so as to investigate the patterns of shared heritability outside of this well-studied region . Figure 3 ( numerical results in Table S19 , S20 ) shows the results of this analysis , as well as the increase in heritability explained compared to the local expectation . The five autoimmune traits have the highest relative increases and are unique in being statistically significant . On average , the loci in the autoimmune traits explain more heritability than the local expectation ( combined ) , compared to more for the non-autoimmune traits ( combined ) . Both results were consistent with the LDAK-adjusted estimate of and respectively ( Table S20 ) . We again confirmed all significant z-test results using an empirical expectation by sampling random regions of the genome ( see Methods , Table S17 , Figure S5 ) . Importantly , these results were not substantially different after accounting for increased heritability in coding regions , with the average increase after correction still significant at ( see Methods , Table S21 ) . We stress that these estimates specifically exclude any known loci for the respective disease; for example , the results from RA represent analysis of known autoimmune disease loci not identified in RA , and likewise for all of the other traits . As such , the additional heritability we identify would not have been found in a traditional targeted fine-mapping study that focuses only on trait-specific loci . Combining these results with the trait-specific analysis , we observe an average of more than at the union of autoimmune and disease-specific loci , individually significant across all the autoimmune traits ( Table S22 ) . On average , these loci are explaining 27% of the genome-wide . Most significant are the increases for MS and CD , with ( ) and ( ) more local , respectively . Overall , we find that the class of autoimmune traits has a shared genetic architecture at known GWAS loci that can be leveraged to explain significant additional heritability . Loci found in one autoimmune trait are expected to harbor significantly more for other traits ( beyond what is expected from lying near coding regions ) and can therefore be important targets for fine-mapping analysis . We estimate components of local heritability for Rheumatoid Arthritis in 23 , 092 samples of European origin typed on the ImmunoChip platform , recently analyzed for association by Eyre et al . [41] . The increased SNP density of this data is expected to provide higher power for local heritability analyses , and we again compare , , , and using simulated phenotypes from ImmunoChip genotypes ( see Methods ) . We again observe an inflated and un-inflated , though the latter is more conservative than in previous simulations ( Table S23 ) . Overall , the higher density ImmunoChip results in a greater expected increase when considering all SNPs , particularly when variants are low-frequency . We now consider real RA phenotypes . Of the 13 RA GWAS loci analyzed in the WTCCC1 data , 10 are also present on the ImmunoChip and we re-estimate local at this subset of 10 loci in both studies for comparison ( Table 2A ) . The ImmunoChip data exhibits an increase in additional heritability explained over local expectation of ( ) , compared to ( non-significant at ) in the corresponding WTCCC1 loci . The ImmunoChip also exhibits a significant increase in heritability explained compared to and local expectation , with an increase of ( ) . The ImmunoChip also contains 17 of the 24 non-RA autoimmune disease loci , also allowing us to perform the analysis of non-RA autoimmune loci . Again , we observe the local heritability to increase between the WTCCC1 and ImmunoChip data from 0 . 012 to 0 . 018 , with the latter resulting in an increase of compared to local expectation ( , Table 2B ) . Examining all relevant loci on the ImmunoChip , which are more likely to come from studies performed after the WTCCC , both local increases were lower but more significant due to the additional data analyzed . For consistency , we have assumed the same total of 0 . 14 in both of the data-sets when computing the local heritability expected by chance , though this is likely an underestimate for the dense typing on the ImmunoChip . Likewise , the densely typed ImmunoChip sites also tag some markers outside of the variance-component region , effectively increasing the local expectation . Using 1 , 000 Genomes data , we find that a sequenced variant within 500 kbp of the studied regions is tagged with an average of 0 . 33 by the ImmunoChip sites in these loci , so we also consider a local expectation where each region is increased by of “flanking” length . However , irrespective of whether we use a total of 0 . 40 ( the total estimated in previous studies excluding MHC [42] ) and/or include the flanking regions , the local heritability identified at these loci remains strongly significant ( Table S24 ) . Overall , the ImmunoChip data shows local for RA at 27 known ( RA+other ) autoimmune loci to be 0 . 032 , higher than that explained by the individual RA GWAS SNPs ( 0 . 006 ) and higher than the joint GWAS model ( 0 . 009 ) . The variance-component method allows us to estimate local at regions that are suggestive of harboring a secondary signal in this data . Specifically , Eyre et al . [41] analyzed these samples for conditional association and identified six loci that had a significant secondary signal . Predictably , when we restrict our analysis to these loci we confirm that the joint model increases heritability by over the associated SNP , but we also find the local to be even higher with a increase over the associated SNP and highly significant compared to local expectation ( Table S25 ) . Though the joint analysis has high power in this large cohort , the variance-components model still reveals additional hidden heritability . Similarly , Diogo et al . [43] fine-mapped 25 known RA loci and searched for the presence of secondary associations driven by variants in the protein-coding sequence of biological candidate genes , identifying strong enrichment of association at 10 coding variants ( 9 loci ) but no individually significant variant . We examine these 9 loci in the ImmunoChip data and again observe an increase in heritability from the joint analysis of compared to the leading SNPs , but an even higher increase in local of which is more significant at than the permutation-based reported by Diogo et al . ( Table S25 ) . Overall , the higher density and sample-size of the ImmunoChip data empowers us to identify the presence of significant additional at known RA loci as well as known non-RA autoimmune loci , beyond the heritability explained by standard mapping approaches analyzing the same data .
In this work we have sought to explain additional heritability at known GWAS loci by using large-sample SNP data . Specifically , we have utilized variance-components models that estimate the total contribution of all typed markers in the sample and do not require individual markers to be genome-wide significant . In applying these methods we have quantified biases in the standard estimate when the underlying disease architecture is non-infinitesimal and LD is systematically different at causal variants ( as recently identified by [30] ) . To address this , we have proposed and compared several methods that seek to adjust the covariance matrix such that this correlation between markers is accounted for . In particular , we find the method of using LD residuals in computing the kinship to provide accurate estimates with no observed upward bias , in contrast to the proposed LDAK strategy [30] which yielded upward bias in our genome-wide simulations ( though it exhibited lower mean error in imputed data ) . We thus recommend that the LD-residual approach be used in preference to LDAK when one is seeking lower bounds on the estimate of , as we are here . Applying the LD-residual to known GWAS loci for nine WTCCC1 and WTCCC2 traits , we see that LD-adjusted estimates are nearly always higher than the unadjusted estimates , suggesting that the disease architecture is indeed shifted towards low-frequency variants for most traits . Understanding this phenomenon and applying and LD-adjustment method is therefore important for accurate estimation of in future studies . An alternative framework is the Bayesian sparse linear mixed model , which attempts to infer the underlying genetic architecture jointly with the and can provide more accurate estimates under certain disease architectures but requires significant computational resources ( e . g . running time of 77 hours for a data set with 3 , 925 samples ) [44] . Looking at previously known GWAS loci , we showed by simulation that the LD-residual adjusted variance-components approach is not inflated and can uncover additional heritability beyond that observed by the leading tag SNP , particularly when there are multiple underlying causal variants or tags . In analysis of nine dichotomous traits , we find a significant average increase in heritability explained of ( combined ) , with three traits exhibiting individually significant increases consistent with the presence of multiple causal variants on average . The latter finding is supported by previous work showing that loci with a single causal variant are unlikely to explain substantially more heritability then the GWAS SNP and hypothesizing multiple underlying causal variants [8] . However , though our simulations show that increased heritability is an indicator of multiple causal variants on average , the current sample size is not sufficient to reject the possibility that this local increase is caused by a single causal variant being poorly tagged by the leading GWAS SNP . We extrapolate that as sample sizes reach the tens of thousands our method can conclusively draw distinctions between these two scenarios . Because the LD-unadjusted method tends to be deflated when the underlying causal variant is low-frequency ( Table 1 ) , we can use the unadjusted estimate as an indicator of the causal allele frequency . The fact that all but one of these traits exhibit an unadjusted local that is higher than the strongly suggests that the bulk of causal variation at these known loci does not lie in low-frequency variants . This is consistent with the recent findings of Hunt et al . [45] in a large-scale sequencing study that demonstrated minimal rare-variant heritability for 25 known auto-immune disease risk genes . This is in contrast to our genome-wide analysis that yielded additional heritability after LD-adjustment , indicative of a shift toward low-frequency markers . Taken together , we hypothesize that the causal frequency spectrum at these known loci is substantially different from that of the rest of the genome . In light of this finding , we caution against extrapolating the genome-wide disease architecture from known GWAS loci , as done in Hunt et al . and other studies [45]–[48] . We also applied this technique to loci that have been discovered in related traits but not in the focal trait . Additional variation would be found in instances where causal loci are shared across multiple traits but have only been mapped in one trait , allowing us to estimate the efficacy of a fine-mapping study design incorporating these loci . For autoimmune diseases we see a significant amount of excess heritability at such related-trait loci with an average of more than expected by chance . Relative to the known , the greatest increase from the union of trait-specific and related-trait loci is observed in MS ( ) and CD ( ) . This finding is substantiated by the fact that non-autoimmune traits exhibit no such significant increase and serve as negative controls . Where previous studies have documented overlap between causal variants from autoimmune disease [22] , [40] , we show that this is a wide-spread phenomenon expected to account for an average of 27% of total over five auto-immune traits . Our analysis is complementary to recent methods that construct multivariate variance-components models which directly estimate the genetic correlation between multiple traits [23] , [24] . In contrast to those studies , our approach requires only the genetic information from a single trait of interest , allowing us to analyze components of heritability between many autoimmune traits without having their genetic data . Looking forward , this strategy can be used to analyze other classes of related phenotypes such as metabolic traits [24] and psychiatric disorders [36] . Given that we observe GWAS loci to have fundamentally different disease architectures from the rest of the genome , our method will still not capture the genome-wide correlation between the two traits . A potential future application is local heritability analysis with the multivariate variance-components model , merging these two strategies . For RA , we repeated our analysis in a much larger cohort typed on the ImmunoChip and found significant additional heritability . Where the GWAS analysis of this data by Eyre et al . [41] found 6/45 loci containing a secondary marker , we quantify the overall amount of additional heritability to be than . While Eyre et al . identified a significant correlation between their associated loci and genes with auto-immune function , we additionally observe more heritability than expected by chance in non-RA auto-immune loci ( Table 2 ) , a highly significant increase . These findings demonstrate the effectiveness of our method in quantifying components of heritability from high-density data . Loci from the other traits we examined have also recently been analyzed large fine-mapping studies . Jostins et al . [40] found that 30/163 loci associated with Crohn's Disease or Ulcerative Colitis exhibit significant secondary effects , and all loci have an higher chance of being associated with immune-function genes . Likewise , we observe significant local and related-trait heritability for Crohn's Disease . On the other hand , Shea et al . [25] re-sequenced one locus for T2D and Maller et al . [20] densely genotyped 11 loci for CAD and T2D , with neither study identifying significantly more heritability . This too is consistent with our failure to observe significant increases in heritability for these traits , though both sets of negative results may be due to the small number of loci and samples examined . Two recent publications by Ehret et al . and Ke [16] , [17] propose methods to quantify the amount of recoverable heritability at known loci by selecting a conditional linear model . The conceptual distinction between these methods and our approach is that they explicitly focus on a pruned and p-value restricted set of markers and are therefore limited by power to detect association within the analyzed sample . The Ke strategy differs from that of Ehret et al . in the specific threshold values and that it does not depend on an external set of samples for estimating unbiased effects; as such , it is likely to be the less conservative estimate of local heritability and the one we selected for comparison . Because these strategies only focus on loci where conditionally nominal SNPs are present , they do not provide a complete analysis of all known loci together . While it is possible to incorporate many more SNPs into a complex multiple regression and estimate the total fraction of phenotypic variance explained , this estimate will be highly biased proportional to the effective number of SNPs divided by the effective number of samples , a difficult ratio to quantify in the presence of LD between SNPs and sample structure . On the other hand , the local variance-components model provides an approximately unbiased estimate of the total heritability explained by all SNPs , allowing us estimate components from putative loci without significant associations , as we do here with related traits . Both in simulations and in real data , we find that our strategy identifies more additional variation than the standard linear model . One limitation of the current variance-components strategy is that analysis of ascertained case-control traits can lead to underestimates of when the ratio of SNPs to samples is low ( A . L . P . , unpublished data ) , as can be the case when analyzing a small number of loci . This would lower the power to detect significant additional heritability and yield local estimates that are a conservative lower bound . Quantifying and correcting for this phenomena in case-control traits is an important area of future study . Other future directions for this work include the estimation of local heritability over more complex annotations of putative regions [49] as well as the use of local heritability for mapping previously unknown loci akin to group-wise tests [50] , [51] . The torrent of large-scale sequencing studies will do much to inform our understanding of the genetic architecture of common diseases , but the design of such studies also motivates the inference of disease architecture from currently available data . The strategies outlined here demonstrate a great diversity of allelic heterogeneity within and between traits , informing our assumptions for future GWAS and fine-mapping analysis .
We examined data from the Wellcome Trust Case Control Consortium ( WTCCC ) versions 1 and 2 . These datasets have been outlined in [13] and [12] , [40] , and we provide summary details in Table S1 . Unlike GWAS studies , heritability estimates can be particularly sensitive to individually small artifacts/batch-effects [52] , which can add up over many SNPs to exhibit false heritability [29] . To account for this , we apply several additional layers of quality control . We also examined 23 , 092 samples of European origin typed on the ImmunoChip platform ( 32% cases for Rhematoid Arthritis ) , recently analyzed for association by Eyre et al . [41] . For this data , we followed the QC protocol of Eyre et al . [41] and also excluded any SNPs below 1% allele frequency . The variance-components model assumes an idealized infinitesimal genetic architecture where every marker is causal and effect-sizes are normally distributed over the normalized variants . [33] showed that the model remains unbiased when causal variants are randomly sampled from the typed SNPs ( though the analytical standard error on the estimate does exhibit bias as the number of causal variants becomes very low [30] ) . However , as demonstrated in [33] , when causal variants are not randomly drawn from the typed SNPs , LD between markers can lead to over-representation of certain SNPs in the sample GRM and distort the estimated relationships between individuals , thereby distorting the final estimate of SNP-heritability . We describe and evaluate several methods that account for correlations between markers when constructing a GRM . In all cases , the goal is to reweigh or transform each SNP so that it is equally represented in a new adjusted genotype matrix . We caution that our simulations do not explore the robustness of this model in the presence of very rare variants ( e . g . whole-genome sequence ) where assumptions of normality may be strongly violated . Open-source software implementing the LD residual adjustment we have described is implemented in EIGENSOFT 5 . 0 at http://www . hsph . harvard . edu/alkes-price/software . HAPI-UR software is available at https://code . google . com/p/hapi-ur/ GCTA software is available at http://www . complextraitgenomics . com/software/gcta/ | Heritable diseases have an unknown underlying “genetic architecture” that defines the distribution of effect-sizes for disease-causing mutations . Understanding this genetic architecture is an important first step in designing disease-mapping studies , and many theories have been developed on the nature of this distribution . Here , we evaluate the hypothesis that additional heritable variation lies at previously known associated loci but is not fully explained by the single most associated marker . We develop methods based on variance-components analysis to quantify this type of “local” heritability , demonstrating that standard strategies can be falsely inflated or deflated due to correlation between neighboring markers and propose a robust adjustment . In analysis of nine common diseases we find a significant average increase of local heritability , consistent with multiple common causal variants at an average locus . Intriguingly , for autoimmune diseases we also observe significant local heritability in loci not associated with the specific disease but with other autoimmune diseases , implying a highly correlated underlying disease architecture . These findings have important implications to the design of future studies and our general understanding of common disease . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2013 | Quantifying Missing Heritability at Known GWAS Loci |
Voltage-gated cation channels regulate neuronal excitability through selective ion flux . NALCN , a member of a protein family that is structurally related to the α1 subunits of voltage-gated sodium/calcium channels , was recently shown to regulate the resting membrane potentials by mediating sodium leak and the firing of mouse neurons . We identified a role for the Caenorhabditis elegans NALCN homologues NCA-1 and NCA-2 in the propagation of neuronal activity from cell bodies to synapses . Loss of NCA activities leads to reduced synaptic transmission at neuromuscular junctions and frequent halting in locomotion . In vivo calcium imaging experiments further indicate that while calcium influx in the cell bodies of egg-laying motorneurons is unaffected by altered NCA activity , synaptic calcium transients are significantly reduced in nca loss-of-function mutants and increased in nca gain-of-function mutants . NCA-1 localizes along axons and is enriched at nonsynaptic regions . Its localization and function depend on UNC-79 , and UNC-80 , a novel conserved protein that is also enriched at nonsynaptic regions . We propose that NCA-1 and UNC-80 regulate neuronal activity at least in part by transmitting depolarization signals to synapses in C . elegans neurons .
Neurons generate and propagate electrical signals along nerve processes , which are converted into chemical communication through neurotransmitter release at synapses . By allowing selective ion flux across the plasma membrane , cation channels regulate the excitation and function of neurons . In most nervous systems , action potentials , the traveling and rapidly reversing membrane potentials , are induced by the opening of voltage-gated sodium channels and are modulated by voltage-gated sodium , potassium , and occasionally calcium ( Ca2+ ) channels [1 , 2] . Action potential–induced depolarization at presynaptic termini triggers the opening of voltage-gated calcium channels ( VGCCs ) , leading to an influx of Ca2+ that allows for Ca2+-dependent synaptic vesicle exocytosis and the release of neurotransmitters [3] . Voltage-gated sodium channels consist of a pore-forming α1 subunit and variable numbers of auxiliary β subunits [4] . They display similar properties and have similar functions in establishing membrane thresholds , and generating and propagating action potentials . In contrast , multiple neuronal VGCCs differ in composition , property , localization , and function . All known VGCCs are composed of a pore-forming α1 subunit , which associates with various accessory α2δ , β , and γ subunits that modulate the property of the channel [4–6] . Vertebrates have at least six subfamilies of VGCCs with different opening probabilities and kinetics [4–6] . Among them , P/Q- and N-type VGCCs are components of the active zone , the presynaptic subcellular structure where synaptic vesicles are released [7 , 8] . They mediate the Ca2+ influx that triggers the membrane fusion between synaptic vesicles and presynaptic termini [9] . Other VGCCs can also participate in the modulation of neuronal excitation , affecting the duration of action potentials of specific neurons [2 , 10] . C . elegans does not encode voltage-gated sodium channel orthologues or display typical voltage-gated sodium currents [11–16] . Therefore , C . elegans cells either do not have action potentials , or generate and propagate atypical action potentials through alternative mechanisms such as VGCCs in muscles [13 , 15 , 17] . In C . elegans neurons , the nature of the excitation signals that lead to the depolarization at synapses , and how they are transmitted , are unknown . It was proposed that their membrane properties allow the passive spreading of electrical signals along axons in the sensory neurons [12] . Alternatively , they may also generate atypical action potentials . C . elegans encodes a single P/Q- , N- , and R-family VGCC α1 subunit ( UNC-2 ) , one L-type α1 subunit ( EGL-19 ) , and one T-type α1 subunit ( CCA-1 ) [17–19] . UNC-2 is proposed to localize at presynaptic active zones and affects neurotransmitter release [20] . The loss of UNC-2 function leads to slow and abnormal locomotion , failure in neuronal migration , and abnormal sensitivity to dopamine and serotonin [19–21] . In pharyngeal muscles , the excitability threshold is set by CCA-1 , which initiates an atypical action potential in response to depolarization [15 , 17] . EGL-19 generates Ca2+ transients that define sarcomere excitability [13 , 15 , 18] . It also contributes to the Ca2+ transients in cultured mechanosensory neuron cell bodies [22] . A rat cDNA clone encoding a protein with homology to the α1 subunit of voltage-gated calcium and sodium channels was first isolated by a degenerative oligo-based PCR screening [23] . Homologues of this protein are present in various animals , namely NCA-1 and NCA-2 in C . elegans [24] , Dmα1U/CG1517 in Drosophila [25] , and Vgcnl1/NALCN [23 , 26] in mouse , rat , and human . Unlike all known sodium and calcium channel α1 subunits , whose ion selectivity filter motifs are DEKA [27] and EEEE [28 , 29] , respectively , these proteins contain EEKE at corresponding positions . They also display divergence from the known voltage-gated sodium/calcium channels by a reduction of charged amino acids in the voltage-sensing fourth transmembrane domains , suggesting that they may form channels with unique properties . Indeed , a recent paper showed that the rat NALCN forms a voltage-insensitive and poorly selective cation leak channel in HEK293T cells [26] . Drosophila Dmα1U mutants are viable but display altered sensitivity to anesthetics and abnormal circadian rhythm [25] . C . elegans nca-1;nca-2 double knockout mutants also display abnormal halothane sensitivity and more frequent pauses during locomotion , a phenotype termed “fainter” [24] . The physiological basis for these defects , however , is unknown . NALCN knockout mice are neonatal lethal due to a disrupted respiratory rhythm [26] . Mutant hippocampal neurons display reduced background Na+ leak currents and decreased firing , suggesting that NALCN functions as a Na+ leak channel and regulates neuronal excitability by affecting membrane potentials [26] . In this study , we describe the physiological and cell biological characterization of the NCA proteins in C . elegans . Our genetic and phenotypic analyses of nca loss- and gain-of-function mutants show that NCA proteins affect synaptic function by modulating the transmission of depolarization signals . This function depends on two novel auxiliary proteins: UNC-79 and UNC-80 . Thus , a putative NCA channel regulates neuronal activity in C . elegans neurons , at least in part by facilitating axonal conductance of depolarizing signals from the cell body to the synapse .
To investigate the function of NCA channels in C . elegans , we identified and analyzed the phenotypes of animals carrying dominant and recessive mutations in nca-1 and its homologue , nca-2 . Both dominant and recessive mutations in NCAs have clear effects on C . elegans behavior . gk9 and gk5 , the two deletion alleles for nca-1 and nca-2 , respectively , were generated by the C . elegans gene knockout consortium . Removing part of the essential pore-forming domain of NCA-1 and NCA-2 , both mutations are predicted to cause severe losses of protein functions ( Figure 1A ) . While either single deletion mutants display normal locomotion , nca-1 ( gk9 ) ;nca-2 ( gk5 ) double mutants are fainters that fail to sustain sinusoidal locomotion and succumb to long periods of halting ( [24 , 30] , Videos S1 and S2 ) . The fainter phenotype of gk9;gk5 mutants is recessive and fully penetrant . This synergism , together with our results presented in later sections , suggest that the phenotypes of gk9;gk5 mutants represent the physiological outcome of the complete loss of NCA activity , which we will henceforth refer to as nca ( lf ) . We identified two gain-of-function alleles of nca-1 ( see Materials and Methods , Figure 1A ) . One of these mutants , hp102 , was isolated in a screen for developmental defects in active zone markers in GABAergic neurons [31] ( Figure 1A ) ; whereas the other allele , e625 , was isolated as a locomotion-abnormal mutant originally named unc-77 [32] . We identified a single missense mutation that alters residues at positions flanking IS6 , the sixth transmembrane domain in the first repeat ( R403Q ) , or within IIS6 , the sixth transmembrane domain of the second repeat ( A717V ) of NCA-1 in hp102 and e625 mutants , respectively . Both affected amino acids are conserved in the protein family ( Figure 1B ) . Unlike the recessive fainter phenotype of nca ( lf ) ( Video S2 ) , both hp102 and e625 showed semi-dominant , uncoordinated , and exaggerated body bends during either spontaneous or stimulated locomotion ( referred to as “coiler” phenotype henceforth ) ( Video S3 ) . Moreover , the expression of a nca-1 genomic fragment that harbors the hp102 mutation in wild-type animals induced locomotion defects similar to that in hp102 mutants ( Video S4 , see Materials and Methods ) . In summary , both hp102 and e625 represent nca-1 gain-of-function alleles , which may induce elevated , misregulated , or altered NCA activities . They will henceforth be referred to as nca ( gf ) . The locomotion defects of both nca ( lf ) and nca ( gf ) mutants suggest that NCA activity regulates synapse function . To address this possibility , we recorded spontaneous and evoked postsynaptic currents in body wall muscles as an indirect measure for presynaptic activities of GABAergic and cholinergic neurons NMJs [33 , 34] . In the presence of both high and low concentrations of extracellular Ca2+ , nca ( lf ) mutants displayed a significant decrease in the frequency of spontaneous release ( miniature postsynaptic current , mPSC ) ( 29 . 4 ± 5 . 3 Hz , p < 0 . 01 at 5mM Ca2+ and 11 . 8 ± 2 . 5 Hz , p < 0 . 001 at 1mM Ca2+ ) as compared to wild-type animals ( 55 . 6 ± 5 . 3 Hz at 5mM Ca2+ and 39 . 7 ± 6 . 5 Hz at 1mM Ca2+ ) ( Figure 2A and 2C ) . They also displayed significantly reduced evoked responses . Electric stimulation of the ventral nerve cord in wild-type animals elicited currents ( evoked postsynaptic current , ePSC ) of 1234 . 1 ± 57 . 7 pA in amplitude at 5mM Ca2+ , and 1080 ± 161 . 3 pA at 1mM Ca2+ ( Figure 2B and 2D ) . In nca ( lf ) mutants , the amplitude of ePSC was reduced by 60% at 5mM Ca2+ ( 523 . 9 ± 57 . 7 pA , p < 0 . 001 ) , and by 75% at 1 mM Ca2+ ( 278 . 6 ± 109 . 2 pA , p = 0 . 01 ) ( Figure 2B and 2D ) . The decreased mPSC frequency and ePSC amplitude suggest a reduction of synaptic transmission at NMJs in nca ( lf ) mutants . We also examined how nca ( gf ) mutations affect synaptic transmission . At 5 mM extracellular Ca2+ , some nca ( gf ) animals ( Figure 2E and 2G , population 1 ) displayed normal frequency of mPSC ( 59 . 1 ± 6 . 0 pA , n = 7 versus 65 . 4 ± 5 . 4 pA , n = 10 for wild-type ) , others ( Figure 2E and 2G , population 2 ) had no mPSC at all ( 7 . 2 ± 1 . 4 pA , n = 6 ) . No ePSC could be evoked in any of the two groups ( Figure 2F and 2H ) . Although the cause of these abnormalities was not clear , these results indicate that nca ( gf ) animals also show aberrant synaptic activity and further establish the link between NCA channels and synaptic function . To investigate how altered NCA activity regulates presynaptic function , we examined neuronal excitation directly with cameleon , a genetically encoded Ca2+ sensor , in live C . elegans [35] . We focused on the serotonergic HSN motoneurons , where we also observed both morphological ( abnormal active zone marker distribution ) and behavioral ( constitutive egg-laying ) defects associated with their synapses in nca ( gf ) mutants ( Figure S1D–S1F ) . Most importantly , the unusually large size of HSN synapses provided us the unique opportunity to perform in vivo simultaneous Ca2+ imaging at both soma and the presynaptic regions ( Figure 3A ) . When C . elegans is immersed in solutions that constitutively activate egg-laying ( see Materials and Methods ) , HSNs—the motoneurons that innervate the egg-laying vulval muscles—autonomously initiate periodic trains of Ca2+ transients in cell bodies that are independent of presynaptic inputs ( M . Zhang et al . , unpublished data , Figures S2 and S3 , Video S5 ) . These transients temporally correlated with the Ca2+ spikes in the presynaptic region ( Figure 3A , blue and red traces; Figures S2 and S3 ) . The Ca2+ transients at presynaptic regions and cell bodies displayed similar spike frequency ( 2 . 6 ± 0 . 8 spikes/min at synapses versus 3 . 2 ± 0 . 6 spikes/min at cell bodies , p > 0 . 05 , Figure 3C and 3F ) and similar time intervals between spikes in the trains ( 7 . 5 ± 0 . 5 s at synapses versus 8 . 6 ± 0 . 2 s at cell bodies , p > 0 . 05 , Figure 3D and 3G ) , suggesting that the depolarization signals were generated at the cell bodies and quickly spread to the presynaptic regions . Under the same conditions , in nca ( lf ) mutants , HSN cell bodies generated trains of Ca2+ spikes undistinguishable from those in wild-type soma for spike frequency ( 3 . 5 ± 0 . 6 spikes/min versus 3 . 2 ± 0 . 6 spikes/min for wild-type , p > 0 . 05 ) , interval ( 5 . 8 ± 0 . 2 s versus 8 . 6 ± 0 . 2 s for wild-type , p > 0 . 05 ) and amplitude ( 5 . 3 ± 0 . 6% versus 6 . 9 ± 0 . 4% for wild-type , p > 0 . 05 ) ( Figure 3C–3H ) . At synapses , whereas Ca2+ transients were present in all wild-type animals , half of the nca ( lf ) mutants showed no Ca2+ transients at all ( Figure 3B , nca ( lf ) , synapse , top trace ) . The rest of the nca ( lf ) mutants retained Ca2+ transient trains ( Figure 3B , nca ( lf ) , synapse , bottom trace ) . This resulted in an overall significant decrease of synaptic spike frequency ( 1 . 2 ± 0 . 7 spikes/min in nca ( lf ) ) compared to wild-type synapses ( 2 . 6 ± 0 . 8 spikes/min , p = 0 . 029 ) , and to the spike frequency of nca ( lf ) cell bodies ( 3 . 5 ± 0 . 6 spikes/min , p = 0 . 005 ) . Remarkably , the remaining trains of Ca2+ transients in the nca ( lf ) mutants maintained temporally correlated in spike interval ( 5 . 2 ± 0 . 4 s ) with the cell body of nca ( lf ) mutants ( 5 . 8 ± 0 . 2 s , p > 0 . 05 ) . They also display comparable amplitude ( 6 . 8 ± 0 . 7% ) to those in wild-type synapses ( 5 . 2 ± 0 . 7% , p > 0 . 05 ) ( Figure 3C–3H ) . Thus the loss of NCA function disrupts the initiation of Ca2+ transients at synapses . In nca ( gf ) mutants , HSN cell bodies also displayed trains of calcium spikes similar to those in wild-type animals in their frequency ( 3 . 3 ± 1 . 0 spikes/min , versus 3 . 2 ± 0 . 6 spikes/min for wild-type , p > 0 . 05 ) , interval ( 5 . 3 ± 0 . 3 s , versus 8 . 6 ± 0 . 2 s for wild-type , p > 0 . 05 ) and amplitude ( 9 . 2 ± 1 . 4% versus 6 . 9 ± 0 . 4% for wild-type , p > 0 . 05 ) ( Figure 3B–3E ) . At synapses , they all displayed trains of Ca2+ spikes that temporally correlated in frequency ( 2 . 7± 0 . 8 spikes/min versus 3 . 3 ± 1 . 0 spikes/min , p > 0 . 05 ) , and interval ( 6 . 4 ±0 . 3 s for synapses versus 5 . 3 ± 0 . 3 s for cell bodies , p > 0 . 05 ) with those in nca ( gf ) cell bodies . However , the amplitude of Ca2+ transients was significantly increased at synapses ( 9 . 5 ± 1 . 3% for nca ( gf ) versus 5 . 2 ± 0 . 7% for wild-type , p = 0 . 029 ) . Although the mean amplitude appears only moderately bigger than in wild-type animals , nca ( gf ) mutants exhibited a fraction of unusually large Ca2+ transients at synapses that were well above the range seen in wild-type animals ( Figure 3H , red box ) . In summary , both nca ( lf ) and ( gf ) mutants specifically altered Ca2+ transients at the presynaptic regions , indicating that under our assay conditions , NCA activity does not alter the excitation at HSN soma but affects presynaptic activity . The decrease of Ca2+ transients in nca ( lf ) suggests that NCA is required to initiate presynaptic activation in response to depolarization signals . The elevated Ca2+ transients in nca ( gf ) mutants further suggests that the gain-of-function mutations enhance NCA's activity in presynaptic activation . To identify proteins that modulate NCA activities , we performed a genetic suppressor screen for mutations that reverted locomotion defects of nca ( gf ) mutants ( see Materials and Methods ) . We identified two extragenic suppressors that reverted nca ( gf ) coilers to fainters and fully suppressed their synaptic morphology defects ( Figure S1 ) . One suppressor , hp424 , corresponds to unc-79 , a gene encoding a large , novel protein [24] . Another suppressor , hp369 , failed to complement unc-80 , an uncloned mutant previously isolated by its locomotion phenotype [32] and later shown to confer hypersensitivity to halothane [30] . unc-80 ( hp369 ) , as well as two previously identified unc-80 alleles , e1272 and e1069 , exhibit recessive and fully penetrant fainter phenotypes identical to that of the nca ( lf ) double mutant ( Video S6 ) . We found that nca ( lf ) ;unc-80 triple mutants are indistinguishable from either nca ( lf ) double mutants or unc-80 single mutants in behavior ( Video S7 ) . Furthermore , all nca ( gf ) ;unc-80 double mutants display the same fainter phenotype as unc-80 single mutants ( Video S8 ) . Therefore NCA and UNC-80 function in the same genetic pathway , with unc-80 mutations epistatic to nca ( lf ) alleles , suggesting that NCA activity depends on UNC-80 . unc-80 was recently cloned based on the observation that RNAi knockdown of an open reading frame F25C8 . 3 in wild-type animals resulted in a fainter phenotype and the identification of missense mutations in F25C8 . 3 from unc-80 alleles [36] . We confirmed that genomic fragments containing only F25C8 . 3 rescued the fainter phenotype of unc-80 mutants ( Video S9 ) and reverted the unc-80;nca ( gf ) mutants from fainters to nca ( gf ) locomotion patterns ( Video S10 ) . Nonsense or splice junction mutations , which are all predicted to result in the loss of the protein function , were identified in three unc-80 alleles ( Figure S4A ) , confirming that unc-80 corresponds to F25C8 . 3 . The unc-80 gene is predicted to encode multiple isoforms of a large protein that contain no known protein motifs . Uncharacterized UNC-80 homologues are present in Drosophila , mouse , rat , and human ( Figure S4B ) , suggesting that UNC-80 is a member of a novel but conserved protein family . We confirmed that unc-80 also regulates calcium transients at synapses . The Ca2+ dynamics of unc-80 mutants were essentially identical to those observed in nca ( lf ) . The HSN cell bodies displayed trains of Ca2+ transients with normal frequency ( 5 . 1 ± 1 . 3 spikes/min for unc-80 versus 3 . 2 ± 0 . 6 spikes/min for wild-type , p > 0 . 05 ) , interval ( 5 . 8 ± 0 . 2 s for unc-80 versus 8 . 6 ± 0 . 2 s for wild-type , p > 0 . 05 ) and amplitude ( 7 . 7 ± 0 . 7% for unc-80 versus 6 . 9 ± 0 . 4% for wild-type , p > 0 . 05 ) ( Figure 3B–3E ) . Likewise , half of these animals showed silencing of Ca2+ transients at synapse regions ( Figure 3B , unc-80 , top trace ) , with an overall reduction in frequency ( 0 . 9 ± 0 . 7 spikes/min ) when compared to unc-80 cell bodies ( 5 . 1 ± 1 . 3 spike/min , p = 0 . 037 ) , and to wild-type synapses ( 2 . 6 ± 0 . 8 spike/min , p = 0 . 032 ) . The remaining trains of Ca2+ transients at synapses maintained temporally correlated with cell body transients in spike interval ( 5 . 6 ± 0 . 5 s for synapses versus 5 . 8 ± 0 . 2 s for cell bodies , p > 0 . 05 ) . They were also comparable in amplitude with wild-type synaptic transients ( 4 . 9 ± 0 . 5% versus 5 . 2 ± 0 . 7% for wild-type , p > 0 . 05 ) ( Figure 3B , unc-80 , bottom trace , Figure 3C–3E ) . Therefore in addition to sharing behavioral phenotypes with nca ( lf ) mutants , unc-80 mutants also displayed identical changes in presynaptic activation . This indicates that UNC-80 either mediates or functions together with the putative NCA channel to regulate presynaptic activation . To determine how UNC-80 regulates the NCA activity , we first examined if they are both expressed or function in the same tissue . Green fluorescent protein ( GFP ) promoter reporter constructs , which contain their predicted upstream genomic sequences , revealed similar expression patterns in the nervous system , including many sensory neurons and all motoneurons , for both the unc-80 and nca-1 genes ( Figure 4A ) . Expression of nca-1 or unc-80 by a pan-neural promoter ( Text S1 ) was able to rescue the fainter phenotype of nca ( lf ) and unc-80 mutants , respectively ( Videos S11 and S12 ) . Therefore , consistent with their expression patterns , both NCA-1 and UNC-80 are required in neurons . Furthermore , specific expression of NCA-1 by a GABAergic promoter Punc-25 [37] rescued the active zone marker defects in GABAergic neurons of nca ( gf ) mutants ( Figure S5 ) , suggesting that NCA-1 functions cell-autonomously . Hence both NCA-1 and UNC-80 function in neurons . NCA and UNC-80 may regulate presynaptic activation through either conducting Ca2+ transients at synapses , or transmitting depolarization signals along axons . To investigate these possibilities , we further examined their subcellular localization . With an NCA-1–specific antibody , we observed dense and punctate staining in the nerve ring , a synapse-rich region at the central nervous system of C . elegans , termed nerve ring , and along the ventral and dorsal nerve cords that are comprised of inter- and motoneuron processes in wild-type animals ( Figure 4B , upper left panel ) . These staining signals disappeared completely in nca-1 ( gk9 ) deletion mutants ( Figure 4B , lower left panel ) . The punctate staining pattern suggests a subcellular enrichment of NCA-1 protein along axons . We therefore examined the localization of NCA-1 relative to the presynaptic termini using antibodies against a vesicle protein , SNB-1; an active zone protein , UNC-10; and a presynaptic kinase , SAD-1 . Along both the dorsal and ventral nerve cords , we observed mostly non-colocalizing staining patterns between NCA-1 and all presynaptic proteins ( Figure 4C and Figure S6A ) , suggesting that NCA-1 is enriched at specific regions along motoneuron axons but not at synapses . The subcellular localization of UNC-80 was examined using a functional Punc-80-UNC-80::RFP construct that rescued the fainter phenotype to the same degree as untagged genomic unc-80 ( Text S1 and unpublished data ) . unc-80 mutants carrying hpIs98 , an integrated transgenic array of Punc-80-UNC-80::mRFP , were stained with antibodies against RFP . We observed specific and punctate staining signals at the nerve ring and along the nerve processes ( Figure 4B , wild-type as negative controls , Figure 4D ) that do not colocalize with presynaptic proteins ( Figure 4D and Figure S6B ) . This UNC-80::RFP staining pattern is highly reminiscent to that of NCA-1 , suggesting that both NCA-1 and UNC-80 proteins are enriched at non-synaptic regions along nerve processes ( Figure 4C and 4D ) . This expression pattern is most consistent with NCA-1 and UNC-80 functioning together to transduce depolarization signals from neuronal cell bodies . To further determine how UNC-80 regulates NCA-1 activity , we examined the distribution of NCA-1 in unc-80 mutants , and vice versa . NCA-1 staining was eliminated or greatly reduced in multiple unc-80 alleles ( Figure 5A , unc-80 panel ) . hpIs98 ( Punc-80-UNC-80::RFP ) restored NCA-1 expression at the nerve ring and along the nerve cords in unc-80 mutants ( Figure S7A ) , indicating that UNC-80 is both necessary and sufficient to localize NCA-1 along axons . While ample NCA-1 staining signals were present in nca ( gf ) mutants , the staining was also eliminated or greatly reduced in nca ( gf ) ;unc-80 mutants ( Figure 5A ) , suggesting that both wild-type and gain-of-function NCA-1 proteins depend on UNC-80 to localize along the nerve processes . nca-1 transcripts were present at wild-type level in unc-80 mutants ( Figure S7B ) . Together with the fact that no obvious UNC-80::RFP signal was detected in neuronal cell bodies ( unpublished data ) , these data indicate that UNC-80 regulates NCA-1 post-transcriptionally , perhaps through reduced translation of NCA-1 proteins or defective trafficking , clustering , or stabilization of NCA along axons . In nca ( lf ) mutants , UNC-80::RFP staining was also significantly reduced ( Figure 5B , nca ( lf ) ;hpIs98 panel ) , suggesting that UNC-80 localization along the axon is also dependent on the presence of NCA protein . This NCA-1–dependent localization of UNC-80::RFP , together with the fact that no transmembrane motifs are present in UNC-80 , is consistent with the possibility that UNC-80 functions as an auxiliary subunit that regulates the transport , stability , or clustering of NCA at the membrane . UNC-79 is another large protein with no known motif that has been implicated in the processes controlled by NCA-1 , NCA-2 , and UNC-80 . unc-79 loss-of-function mutants also have a fainter phenotype , and have been reported to contain lower than normal levels of NCA-1 protein by Western blot analyses [24] . As for unc-80 mutants , we observed reduced or completely diminished NCA-1 staining in the unc-79 mutants ( Figure 5A , lower panels ) . Interestingly , UNC-80::RFP axonal staining was also absent in the unc-79 mutants ( Figure 5B , lower panels ) , suggesting that UNC-79 is another auxiliary protein that facilitates NCA-1 localization along the axon . We generated an antibody against the UNC-79 protein , and observed punctate staining in ventral cord and nerve ring processes , consistent with the possibility of coexpression with UNC-79 and NCA-1 ( Figure 4B and Figure S8 ) . When the same antibody was used to stain nca ( lf ) , unc-80 , and unc-80;nca ( lf ) mutants , no UNC-79 staining was detectable in neuronal processes ( Figure S9 ) . Thus , NCA-1/2 and UNC-80 appear to also facilitate the localization of UNC-79 protein . These results are consistent with the possibility that UNC-79 , like UNC-80 , also functions as an accessory subunit or another regulatory interactor with the NCA channel . To further investigate whether NCA-1 and UNC-80 proteins might function together to promote NCA channel activity , we analyzed NCA function in a heterologous cell system . It was shown previously that the expression of mammalian NALCN induced constitutive cation leak currents when transfected in HEK293T cells [26] . These currents were attributed to the NALCN channel activity , because they were inhibited by verapamil or gadolinium , two blockers for the endogenous , NALCN-mediated Na+ leak currents in hippocampal neurons [26] . In our experiments , these currents appeared to induce cell death in the transfected cells , because significantly increased cell death was observed 48 h after HEK293T cells were transfected with cDNAs expressing NALCN ( 144 . 3 ± 1 . 8% , normalized against untransfected cells , p < 0 . 01 ) . This effect was not induced by the expression of other channels ( e . g . , for Kv4 . 2 , 113 . 7 ± 13 . 9% , p > 0 . 05 ) , and was abolished when transfected cells were incubated with 100 μM verapamil or 10 μM gadolinium ( Figure 6 ) ( see Materials and Methods ) , suggesting that the cell death was indeed associated with the NALCN channel activity . Using this same assay , we examined whether C . elegans NCA-1 , alone or together with UNC-80 , exhibited similar activities in HEK293T cells ( Figure 6 ) . cDNAs encoding the longest isoform for NCA-1 and UNC-80 were maintained in a low–copy number expression vector ( Text S1 ) . Transfecting with either the NCA-1 or UNC-80 expression construct alone did not cause an increase in the lethality of the host cells ( NCA-1: 119 . 3 ± 4 . 1% , and UNC-80: 114 . 3 ± 6 . 7% , p > 0 . 05 ) . In contrast , co-transfection of NCA-1 and UNC-80 constructs induced significant cell death ( 159 . 3 ± 2 . 9% , p < 0 . 01 ) . This effect was abolished when the UNC-80 expression vector was co-transfected with a NCA-1 clone carrying a deletion in the coding region ( 122 . 7 ± 11 . 6% , p > 0 . 05 ) . Moreover , the increased cell death in NCA-1 and UNC-80 co-transfected cells was also blocked in the presence of 100 mM verapamil ( 120 . 3 ± 3 . 3% , p > 0 . 05 ) or 10 mM gadolinium ( 112 . 7 ± 3 . 7% , p > 0 . 05 ) . Therefore the co-expression of UNC-80 and NCA-1 induced the same effect , with similar blocker responses as NALCN in HEK293T cells , suggesting that the putative NCA/UNC-80 channel complex likely shares similar ion leak properties .
We have shown here that the NCA-1 and NCA-2 proteins are required redundantly for synaptic activity . Both the reduction of postsynaptic currents at GABAergic and cholinergic NMJs , and the decrease of Ca2+ transients at serotonergic NMJs in nca ( lf ) mutants suggest a decreased presynaptic activity in the absence of the putative NCA channels . The calcium imaging analyses further suggest that this synaptic defect is related to a failure to initiate presynaptic activity . In wild-type animals , the calcium spikes at HSN cell bodies and synapses are temporally correlated . In both nca ( lf ) and unc-80 mutants , at least under our assay conditions , despite the normal calcium dynamics in cell bodies , the number of Ca2+ transients was reduced at synapses . The NCA channel is unlikely to conduct Ca2+ transients at synapses , since the remaining transients in nca ( lf ) mutants were normal in amplitude and maintained temporal correlation with the depolarization signals in cell bodies . Together with their nonsynaptic localization along nerve processes , these results strongly indicate that NCA channel activity is required to transmit depolarization signals to synapses ( Figure 7 ) . Depolarization signals may propagate actively or spread passively along axons . Lacking typical voltage-gated sodium currents , the passive model , conceivable for neurons with short axons or axons with a high input resistance membrane property [38] , was proposed for C . elegans sensory neurons [12] . Mouse NALCN mediates Na+ leak in hippocampal neurons [26] . A similar property for the NCA channel would allow it to drive the membrane potential close to its excitation threshold at specific regions along C . elegans neurites , facilitating the activation of other channels along axons or around synapses . This model is consistent with the presence of the active propagation of depolarization signals in C . elegans motoneurons . Interestingly , the silencing of Ca2+ transients in nca ( lf ) and unc-80 mutants is incomplete; however , the molecular lesions in these mutants predict severe loss of protein functions . All nca ( lf ) and unc-80 alleles are behaviorally indistinguishable from each other and fully penetrant for the fainter phenotype , which strongly argues against an allelic effect on phenotype penetrance . The partial loss of Ca2+ transients and the variable degree of the decrease of mPSC frequency in nca ( lf ) mutants therefore more likely suggest that while some depolarization signals depend on NCA channels to induce presynaptic activation , other signals reach synapses independently of NCA activity ( Figure 7 ) . It is worth noting that although we detected two distinct , active versus quiescent populations in nca ( lf ) and unc-80 mutants in our physiological analyses , there is little behavioral variability among individual animals . Because every animal alternates between a state of normal sinusoidal movement and quiescence , we speculate that C . elegans neurons alternately fire NCA/UNC-80–dependent and -independent depolarization signals . Perhaps due to the necessary experimental manipulation ( such as immobilization of the animal ) and the short assay time , we measured neuronal activity fixed in one “mode , ” resulting in the appearance of two distinct populations . In addition to the synaptic phenotypes observed in the loss-of-function mutants , we also observed behavioral and synaptic phenotypes in the nca gain-of-function mutants . Specifically , we found that these mutant animals showed a coiler uncoordinated phenotype , and exhibited larger calcium transients at synaptic sites . Gain-of-function mutations in NCA-1 do not affect the temporal correlation of calcium transients between HSN cell bodies and synapses . Whole-mount staining with antibodies against NCA-1 showed no obvious changes in the subcellular distribution or intensity of the staining signals in nca ( gf ) mutants , indicating that these mutations likely alter the activity rather than the abundance of the NCA-1 protein . The calcium imaging phenotype is consistent with the NCA ( gf ) channel further increasing the membrane excitability , which leads to enhanced activation of calcium channels at HSN synapses ( Figure 7 ) . The hp102 mutation alters a conserved amino acid flanking the IS6 transmembrane domain . This coincided with a hot spot region for identified gain-of-function alleles for several VGCCs . In several cases , these gain-of-function mutations lead to slowed inactivation , subsequently prolonging the duration of the corresponding currents [18 , 39 , 40] . If the hp102 mutation leads to a further increase of the leak through the NCA channel , it could indeed bring the neuronal membrane to a hyper-excitable state . Expressing the mouse NALCN carrying the hp102 equivalent mutation ( NALCN ( R329Q ) ) was able to induce similar locomotion defects as NCA-1 ( gf ) proteins in C . elegans ( Text S1 and Video S13 ) , suggesting that hp102 mutation may induce similar property changes in all NCA family channels . Another gene with a loss-of-function fainter phenotype , unc-80 , encodes a novel protein with a critical role in NCA channel function . Based on behavioral and physiological characterization of mutants , UNC-80 appears exclusively to be exclusively involved in NCA-mediated functions . Not only do unc-80 mutants show identical phenotypes as nca ( lf ) mutants , they do not enhance nca ( lf ) mutants , and they suppress defects exhibited by nca ( gf ) mutant . By contrast , unc-80 mutants do not phenocopy VGCC loss-of-function mutants or display obvious genetic epistasis with VGCC gain-of-function mutants ( Table S1 and Video S14 ) . These genetic interactions indicate that UNC-80 function is specifically required for NCA channels . UNC-80 regulates NCA channel function at least in part by localizing the putative pore-forming NCA-1 subunit to the membrane . nca-1 transcripts are present in normal levels in unc-80 mutants , suggesting that UNC-80 regulates NCA-1 post-transcriptionally . The similar and interdependent subcellular localization pattern of NCA-1 and UNC-80 implies that UNC-80 is a likely subunit of the NCA channel to transport , anchor , or stabilize the pore-forming subunit NCA-1 along axons . With close homologues present in all animals , proteins in the UNC-80 family likely play a conserved role in regulating the localization of the NCA family channels . We observed identical genetic interaction between nca and unc-79 mutants , and identical interdependent localization of NCA-1 and UNC-79 proteins . unc-79 encodes another large but evolutionarily conserved protein with no known protein motifs [24] . Similar to unc-80 , loss-of-function mutations in the unc-79 gene lead to not only the same fainter phenotype as nca ( lf ) mutants , but also a complete suppression of nca ( gf ) locomotion defects and the disappearance of NCA-1 and UNC-80 along nerve processes . Furthermore , UNC-79 is dependent on the presence of both NCA and UNC-80 for its localization along neurites . Therefore , both UNC-80 and UNC-79 are likely conserved auxiliary components of the NCA family channels . NCA-1 and NCA-2 have close sequence homologues in other invertebrate and vertebrate , including human . The mammalian member of this family , NALCN , has recently been characterized physiologically in HEK293 cells [26] . In spite of its sequence homology and similar topology to the pore-forming α1 subunits of VGCCs , NALCN forms a voltage-insensitive and nonselective cation channel . Two lines of indirect evidence support the hypothesis that C . elegans NCA and its mammalian homologues share common functional properties . First , C . elegans NCA proteins show at least similar properties to NALCN proteins when heterologously expressed in mammalian cell culture . In HEK293T cells , transfecting NALCN , or co-transfecting NCA-1 and UNC-80 induced cell death that was blocked by the NALCN blockers verapamil and gadolinium . Conversely , expressing mammalian NALCN proteins in C . elegans could substitute functionally for the NCA proteins . Specifically , wild-type C . elegans expressing a mouse cDNA that carries the hp102 equivalent mutation ( NALCN ( R329Q ) ) in neurons exhibited a locomotion pattern with exaggerated body bends , reminiscent of the nca ( gf ) mutants ( Text S1 and Video S13 ) . Thus C . elegans NCA and its mammalian homologues can mediate similar physiological functions , consistent with the possibility that NCA family proteins share similar channel properties . Given the conservation in the functional properties of NCA family members , it is reasonable to speculate that these channels may also carry out similar functions in neurons . Our current studies suggest a specific function for the NCA channel in transmitting and regulating excitability along C . elegans neuronal processes , but do not rule out the possibility that NCA also controls neuronal firing . Since our calcium imaging analysis was performed under conditions that stimulated the constitutive firing of HSNs , an altered firing ability could be masked by the hyperactivation of neurons . While the enriched localization of NCA-1 and UNC-80 at nonsynaptic regions along axons is consistent with the propagation role of the NCA channel , we do not exclude the possibility that the reduced synaptic transmission at GABAergic and cholinergic NMJs in nca ( lf ) mutants may result from a combination of deficits in the propagation of depolarization signals , neuronal firing , and even vesicle release . The mouse NALCN affects the resting potential and controls the excitability/firing rate of hippocampal neurons [26]; whether it is also involved in excitation propagation , however , is not examined . Therefore , it will be interesting to determine the subcellular localization of mouse NALCN channels and to examine whether they are also involved in such processes in mammalian neurons .
All strains were cultured at 22 °C unless specified otherwise . hp102 was originally identified in a genetic screen for hpIs3 defective mutants [31] and was outcrossed eight times against wild-type N2 . unc-80 ( hp369 ) and unc-79 ( hp424 ) were identified in a hp102 suppressor screen and outcrossed three times against N2 . e625 , e1069 , and e1272 were identified through abnormal locomotion in previous C . elegans screens [32] . gk9 and gk5 were generated by the Gene Knockout Consortium and were outcrossed three times against N2 . tm1591 was generated by National Bioresource Project for the Nematode and was outcrossed once against N2 . hp102 mutants were identified from an active zone marker hpIs3 screen [31] . Based on both of its abnormal active zone marker distribution and locomotion defects , hp102 was rough mapped to Chromosome IV between E03H12 ( 1 . 40 cM ) and D2096 ( 3 . 74 cM ) by SNP mapping against CB4856 . During the mapping , we noticed that unc ( uncoordinated ) -77 ( e625 ) , an uncloned , previously identified locomotion defective mutant [32] that was linked to a similar region on Chromosome IV , showed similar locomotion and active zone marker defects as hp102 mutants ( unpublished data ) . We determined that e625 and hp102 were alleles of the same gene due to the genetic interactions displayed by these two mutants: while hp102/e625 heterozygous animals showed fully penetrant coiling locomotion as either homozygous mutants , hp102/+ or e625/+ heterozygous mutants showed only slightly more exaggerated body bends compared with wild-type animals . This conclusion was confirmed when we mapped both mutants to the same genetic locus , rescued both mutants with the same genetic fragments , and identified mutations in the same open reading frame ( see below ) . hp102 and e625 mutations were then further fine-mapped between B0273 ( 1 . 74 cM ) and C49A9 ( 3 . 08 cM ) based on the following data . From dpy-13unc-77/CB4865 , three out of 20 Unc non Dpy animals had their recombination breakpoints between B0273 ( 1 . 74 cM ) and F38A5 ( 3 . 21 cM ) , placing unc-77 to the right of B0273 ( 1 . 74 cM ) . From unc-77bli-6/CB4856 animals , three out of three Unc non Bli recombinants and two out of two Bli non Unc recombinants had their recombination breakpoints between C49A9 ( 3 . 08 cM ) and F38A5 ( 3 . 21 cM ) , placing unc-77 to the left of C49A9 ( 3 . 08 cM ) . From unc-5unc-77/CB4856 heterozygous animals , two out of two unc-77 non unc-5 recombinants had their recombination breakpoints between C31H1 ( 2 . 56 cM ) and C49A9 ( 3 . 08 cM ) , placing unc-77 between C31H1 ( 2 . 56 cM ) and C49A9 ( 3 . 08 cM ) . Cosmids and PCR fragments amplified from the genomic sequence covering this region ( for details , see Molecular biology section below ) were injected into hp102;hpIs3 and e625;hpIs3 animals . Only DNA fragments covering the C11D2 . 6 ( nca-1 ) genomic region rescued the locomotion and hpIs3 marker defects . We further confirmed that unc-77 corresponds to nca-1 by sequencing the entire predicted genomic regions ( all exons and introns ) of hp102 and e625 mutants and identifying a single missense mutation in the coding region of each mutant ( Figure 1 ) . Both hp102 and e625 mutants harbor gain-of-function mutations for nca-1 because they both behaved as semi-dominant mutations; hp102/+ or e625/+ heterozygous mutants showed more exaggerated body bends compared to wild-type animals , but much less severe than homozygous or hp102/e625 heterozygous animals . They also behaved dominantly over nca-2 loss-of-function mutations , as hp102;nca-2 ( gk5 ) and e625;nca-2 ( gk5 ) mutants displayed the same behavior as hp102 and e625 homozygous mutants . Furthermore , PCR fragments amplified from the nca-1 genomic region from hp102 when expressed in wild-type animals induced the same coiling locomotory defects as hp102 mutants ( Video S4 ) . Lastly , overexpression of the wild-type copy of nca-1 rescued phenotypes induced by a gain-of-function mutation likely by replacing the mutated NCA-1 protein from its putative channel complex . hp102;hpIs3 mutants were mutagenized by EMS , and F2 progenies displaying noncoiler locomotion patterns were recovered as candidate suppressors . Each candidate suppressor line was rescreened and confirmed by their rescuing of hpIs3 marker defects . We backcrossed each suppressor line to wild-type animals: if coilers could be recovered from the F2 generation , the suppressor was considered as extragenic . If the coilers could not be recovered , the suppressor line was then crossed into the nca-2 ( gk9 ) background . If they showed fainter phenotype , the suppressor was confirmed as intragenic revertants . From the progenies of 13 , 000 mutagenized F1 hermaphrodites ( equivalent to 26 , 000 mutagenized haploid genomes ) , we identified four intragenic suppressors reverting hp102 mutants to wild-type locomotion . We also identified multiple alleles of two different extragenic suppressors that reverted hp102 animals to fainters , and failed to complement unc-80 and unc-79 mutants , respectively . unc-80 was first rough mapped between F21D9 ( 21 . 82 cM ) and F38A6 ( 27 . 08 cM ) on Chromosome V through two-factor SNP mapping against CB4856 . Two out of four unc-80 recombinants from unc-51unc-80/CB4856 broke between Y113G7A ( 24 . 71 cM ) and F38A6 ( 27 . 08 cM ) , placing unc-80 to the right of Y113G7A ( 24 . 71 cM ) . 0/9 Rol non Unc recombinants from unc-51rol-9/unc-80 animals segregated unc-80 , placing it to the right of rol-9 ( 25 . 124 cM , pKP5057 ) . In two-factor mapping against pha-4 and CB4856 , we could not find breakage between unc-80 and pha-4 ( 25 . 60 cM ) , placing unc-80 between 25 . 12 cM and 27 . 08 cM , tentatively near 25 . 60 cM . Clones that cover this region were generated by PCR to be tested for rescuing of unc-80 mutants . Kim Schuske ( University of Utah ) determined that RNAi knockdown of F25C8 . 3 ( which lies within this region ) in wild-type animals was able to induce a fainter phenotype ( K . Schuske , personal communication ) . We generated and shared with the Schuske group DNA fragments spanning F25C8 . 3 ( see Text S1 , Molecular biology; Trasngenic strains section ) that rescued the fainter phenotypes in unc-80 and unc-80;hp102 mutants . See Text S1 . Antibodies against aa1731–1914 of the predicted NCA-1d isoform , and a combination of aa506–608 and aa1205–1851 of UNC-79 were generated in rat ( Covance ) . Whole-mount immunofluorescent staining was carried out as previously described [41] . Antibodies against NCA-1 , UNC-79 , mRFP ( Clontech ) , SAD-1 , SNB-1 , and UNC-10 ( M . Nonet , Washington University , St . Louis ) were used in 1:10 , 1:10 , 1:200 , 1:200 , 1:100 and 1:2000 dilutions , respectively . Dissections on young adult C . elegans were performed as described [14 , 42] . The integrity of the anterior ventral medial body muscle and the ventral nerve cord was visually examined , and muscle cells were then patched using fire-polished 4-MΩ resistant borosilicate pipettes ( World Precision Instruments ) . They were clamped at −60 mV using an Axopatch 1D amplifier throughout experiments ( Molecular Devices ) , and recorded using the whole-cell patch-clamp technique in previously described recording solutions [43] within 5 min following the dissection . Signals were filtered at 5 kHz , and digitized via a Digidata 1322A acquisition card ( Molecular Devices ) . Data were acquired and analyzed using the pClamp software ( Molecular Devices ) . After 10–60 s of recoding of spontaneous events , a highly resistant fire-polished electrode filled with 3 M KCl was brought close to the ventral nerve cord region anterior to the recorded muscle cell , and a 1-ms depolarizing current , generated by a S11B GRASS stimulator ( Astromed ) was applied to induce an evoked response . All recordings were performed between 5 and 10 min after the beginning of the dissection process . Pcat-1-cameleon was used to reveal relative Ca2+ concentrations in HSN cell bodies and synapses corresponding to those on vm2 muscles . Adults 24 h post L4 stage were immobilized by surgical glue on 2% agarose pads on microscope slides and covered with 1 ml of 10 mM HEPES ( pH 7 . 1 ) , a condition that stimulates constitutive egg-laying thus spontaneous activation of HSN neurons . Recording was carried out as previously described [35] . All recordings started within 2 min after animals were glued and lasted for 10 min . Data from HSN cell bodies and synapses were obtained simultaneously . Due to slight body movements during the recordings , some synapse datasets were incomplete and were not included in analysis . Spike detection , data analysis , and statistic analysis by Kolmogorov-Smirnov rank test ( due to abnormal data distribution ) were carried out as described previously ( [44] and Text S1 ) . HEK293T cells were grown in α-MEM ( GIBCO ) medium supplemented with 10% FBS ( GIBCO ) at 37 °C in a humidified atmosphere of 5% CO2 , 95% air . Lipofectamine 2000 was used to transfect the HEK293T cells following the standard procedure ( Invitrogen ) . 0 . 4 μg total DNA was used for each transfection experiment . Medium was replaced 4 h after transfection , during which the culture was split into three sets with equal density , two sets were exposed to 100 μM verapamil or 10 μM Gd3+ , respectively . Cell death assays were performed 48 h after transfection . Culture medium was replaced by extracellular solution containing 50 μg/ml of propidium iodide ( PI ) ( Invitrogen ) . After 30 min incubation at 37 °C , fluorescence intensity in each well was measured with a plate reader ( Victor3; PerkinElmer ) as described previously [45 , 46] . The fraction of dead cells was normalized against the mock-transfected cultures . | Neurons communicate to their targets through synapses that are activated by the electrical signals conveyed along neuronal processes . The tightly regulated ion flux across the cell membrane drives the generation of these electrical signals; it is therefore important to identify ion channels that regulate the excitability of neurons . In the C . elegans nervous system , we reveal that a putative channel complex , consisting of ion-conducting , pore-forming proteins called NCAs and two auxiliary components called UNC-79 and UNC-80 , regulates neuronal function . We first show that an increase or decrease of the activity of this channel causes physiological changes that indicate corresponding alterations in neuronal activity . We then demonstrate by in vivo calcium imaging that the NCA channel , localizing along axons , specifically regulates excitation of synapses . We speculate that this channel participates in the propagation of electric signals that activate synapses . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neuroscience",
"developmental",
"biology"
] | 2008 | A Putative Cation Channel, NCA-1, and a Novel Protein, UNC-80, Transmit Neuronal Activity in C. elegans |
The V3 loop of the HIV-1 Env protein is the primary determinant of viral coreceptor usage , whereas the V1V2 loop region is thought to influence coreceptor binding and participate in shielding of neutralization-sensitive regions of the Env glycoprotein gp120 from antibody responses . The functional properties and antigenicity of V1V2 are influenced by changes in amino acid sequence , sequence length and patterns of N-linked glycosylation . However , how these polymorphisms relate to HIV pathogenesis is not fully understood . We examined 5185 HIV-1 gp120 nucleotide sequence fragments and clinical data from 154 individuals ( 152 were infected with HIV-1 Subtype B ) . Sequences were aligned , translated , manually edited and separated into V1V2 , C2 , V3 , C3 , V4 , C4 and V5 subregions . V1-V5 and subregion lengths were calculated , and potential N-linked glycosylation sites ( PNLGS ) counted . Loop lengths and PNLGS were examined as a function of time since infection , CD4 count , viral load , and calendar year in cross-sectional and longitudinal analyses . V1V2 length and PNLGS increased significantly through chronic infection before declining in late-stage infection . In cross-sectional analyses , V1V2 length also increased by calendar year between 1984 and 2004 in subjects with early and mid-stage illness . Our observations suggest that there is little selection for loop length at the time of transmission; following infection , HIV-1 adapts to host immune responses through increased V1V2 length and/or addition of carbohydrate moieties at N-linked glycosylation sites . V1V2 shortening during early and late-stage infection may reflect ineffective host immunity . Transmission from donors with chronic illness may have caused the modest increase in V1V2 length observed during the course of the pandemic .
The gp120 portion of the HIV-1 envelope protein ( Env ) mediates attachment prior to fusion with the host cell membrane during target cell infection . gp120 has five hypervariable regions ( V1–V5 ) bounded by cysteine residues and separated by four relatively “constant” regions ( C1–C4 ) [1]–[3] . Gp120 is notable for its sequence variation , which may arise through recombination and point mutation , as well as by insertion and deletion of one or more nucleotides . Insertion and deletion events ( indels ) occur throughout env but are maintained through positive selection particularly within the hypervariable loops , which thereby may acquire significant length variation [4] , The third hypervariable region is known to encode the primary determinants of coreceptor usage specificity [5]–[7] , as well as epitopes recognized by humoral [8] , [9] and cellular [10] , [11] immune responses . V3 loop sequence variation has been extensively studied , and correlated with changes in host cell range , cytopathogenicity , and disease progression [12]–[14] . The V1V2 region in particular is characterized by a high degree of length polymorphism , sequence variation , and predicted N-linked glycosylation sites ( PNLGS ) [15]–[20] , each of which may affect viral attachment , coreceptor usage and recognition by neutralizing antibodies [20] , [21] . Comparison of structural models of gp120 and gp120 bound to CD4 and a chemokine coreceptor have yielded considerable insight into the functional roles played by V1V2 and V3 during viral attachment [22] , [23] . In the unbound gp120 conformation , the V2 loop partially obscures V3 and other gp120 residues involved in coreceptor binding . Binding to CD4 induces conformational changes that expose the coreceptor binding site on gp120 , including residues from V1V2 , V3 and other regions [22] , [24] . Numerous studies have suggested that sequence variation in V1V2 influences host cell range and/or syncytium-inducing ( SI ) phenotype [25]–[31] . For example , Toohey demonstrated that recombinant chimeric clones with a V1V2 region from macrophage-tropic HIV-1 strains replicated efficiently in macrophages , whereas clones with the V1V2 region from lymphotropic strains did not [31] . However , not all studies have been concordant on the role of V1V2 in viral replication kinetics , cell range and transmission [15]–[19] , [32] . For example , Pastore showed that sequence changes in V1V2 could rescue otherwise lethal mutations in V3 associated with a change in coreceptor usage [33] , and V2 polymorphisms have also been linked with restriction to CCR5 coreceptor usage [16] . In contrast , Wang et al found no relationship between SI phenotype and V1V2 sequence , length , distribution of PNLGS or charge [32] . The V1V2 region also appears to be an important determinant of sensitivity to neutralizing antibodies [34]–[38] . The V1V2 region evolves under positive natural selection in vivo [4] , [39]–[41] , and an inverse relationship between V1–V4 length and neutralization susceptibility has been demonstrated in subtypes A [20] , B [34]–[38] and C [42] . Tellingly , laboratory strains lacking V1V2 may still replicate efficiently in vitro , but appear to be especially sensitive to antibody neutralization [43] , [44] . Consistent with this observation , viral strains with shorter and less glycosylated V1V4 regions have been reported to preferentially replicate in subjects newly infected with HIV-1 subtype C [45] ( where presumably an effective neutralizing antibody response has not had time to emerge ) , and similar observations have been made concerning the V1V2 loop in individuals recently infected by HIV-1 subtype A [19] . However , we and others have not observed this effect in HIV-1 subtype B [19] , [46] , [47] . Despite these reports , the relationship between V1V2 region length polymorphism and disease progression remains unclear . In two small longitudinal studies , elongation of V1 and V2 was noted in long-term nonprogressors ( LTNP ) , but not within individuals progressing rapidly to AIDS [15]–[19] . In a third study , no clear relationship between V1V2 length variation and disease progression was observed [48] . Lastly , some investigators postulate that V1V2 length changes positively correlate with the pace of disease progression [16] , [19] , while others have suggested that V1V2 length increase may be a correlate of delayed progression to AIDS [18] . Thus , our understanding of the role of the V1V2 loop in influencing HIV pathogenesis remains incomplete and is challenged by several contradictory observations . To more fully characterize HIV envelope subregion variability and to clarify the associations between subregion length variation , glycosylation , and disease progression , we have comprehensively examined length and glycosylation of each gp120 subregion as a function of clinical parameters in a large collection HIV-1 subtype B infected individuals .
This study was performed using publicly available data from the Los Alamos database , and previously unpublished experimental data obtained at the University of Washington . Unpublished data were obtained and analyzed with written informed consent of study participants , and approval by the University of Washington Institutional Review Board . We analyzed new and published HIV-1 envelope gene sequences and associated clinical data from all available subjects in the Seattle Primary Infection Cohort ( PIC ) [49] , the Multicenter AIDS Cohort Study ( MACS ) [50] , and from the Los Alamos National Laboratories HIV database ( HIVDB ) ( http://www . hiv . lanl . gov/content/hiv-db/mainpage . html ) not meeting pre-specified exclusion criteria . Subjects were excluded from this study if younger than 18 years of age or if there was any history of antiretroviral therapy prior to sampling as determined by patient report and clinical records ( MACS , PIC ) or as indicated in the methods section of published reports ( HIVDB ) , unless otherwise noted . All subjects considered in the cross-sectional and longitudinal analyses were infected with HIV-1 subtype B , except for two subjects infected with HIV-1 subtype A who were included in longitudinal analyses , but were excluded from cross-sectional analyses . ( Additional subtypes were considered in analyses of env subregion length change during transmission , presented in Text S1 , Section 8 ) . Clinical data retrieved included CD4 count , viral load , time since infection , and treatment history . Sequence data were only accepted if directly derived from plasma or PBMC without an intervening step involving viral propagation in vitro . In some cases , individual authors were consulted to resolve clinical or methodological ambiguities . Accession numbers for published sequences are provided in Table S1 . Gene sequence data used in this study are available at http://mullinslab . microbiol . washington . edu/publications/curlin_2010/ . Viral gene sequence data were considered in both cross-sectional ( Table 1 ) and longitudinal analyses ( Table 2 ) . The cross-sectional dataset included only plasma and PBMC sequences derived from individuals infected with subtype B ( see results , and Table 1 ) . Sequences were triaged by author , database identifier and associated clinical data to exclude duplicate entries . To assess the role of stage of illness on loop length variation , subjects were divided into four non-overlapping groups; group Cx1 subjects were sampled within two months of the estimated time of infection . Group Cx2 subjects were sampled between two months and three years following infection . Group Cx3 subjects were sampled at times >3 years post infection . Group Cx4 was comprised of all individuals meeting 1993 CDC criteria for AIDS when sampling occurred ( generally CD4 count <200/mm3 ) , regardless of time since infection . The longitudinal dataset was derived from 20 subjects infected with subtype B and 2 individuals infected with subtype A , from the PIC cohort and from previous reports [18] , [51]–[55] , in whom data were available from two or more timepoints ( see results , and Table 2 ) . All intra-individual longitudinal comparisons were made between sequences obtained from the same compartment ( e . g . , plasma vs . plasma ) . Individuals partitioned into group L1 ( N = 15 ) did not meet criteria for AIDS at any time prior to the final sample ( median follow-up 3 . 25 years , range 1 to 20 . 8 years ) , whereas subjects in group L2 ( N = 7 ) were reported to have an AIDS-defining illness or peripheral CD4 count <200/mm3 between the first and second samples ( median follow-up 2 . 75 years , range 2 to 4 years ) . Sequences from the PIC and MACS cohorts ( Tables 1 & 2 ) were obtained from plasma or PBMC by standard methods [56] , [57] , using safeguards to prevent contamination and template resampling [58] . Briefly , PCR amplification was performed using Taq polymerase ( Bioline ) with primers ED3 and BH2 [59] ( first round ) followed by ED5 and DR7 ( second round ) [60] . PCR products were cloned into a TA TOPO vector ( Invitrogen ) and selected colonies sequenced under contract using Big Dye dye-terminator protocols . Genbank accession numbers pending submission . Deduced amino acid sequences were aligned using ClustalW [61] and divided into seven subregions; V1V2 ( HXB2 nucleotide positions 6615–6812 ) , C2 ( HXB2 6813-7109 ) , V3 ( HXB2 7110–7217 ) , C3 ( HXB2 7218–7376 ) , V4 ( HXB2 7377–7478 ) , C4 ( HXB2 7479–7556 ) , and V5 ( HXB2 7557–7637 ) . Alignments were manually edited and subregion lengths were counted using MacClade . PNLGS were counted using NetNGlyc . 1 ( http://www . cbs . dtu . dk/services/NetNGlyc/ ) . Coreceptor usage ( CCR5 vs . CXCR4 tropism ) was predicted for all available subtype B V3 loop sequences , using the Position-Specific Substitution Method ( PSSM ) [62] , Geno2pheno [63] and two other machine learning algorithms [64] , [65] ( hereafter denoted PSSM , G2P , PGRC and BMLC , respectively ) . For G2P coreceptor usage predictions , we selected the standard 10% false positivity threshold , and PGRC predictions were based on the support vector machine ( SVR ) user option . Estimated time since infection was calculated for all data entries . When time was reported as time since onset of symptoms or time post seroconversion ( SC ) , symptoms and seroconversion were assumed to occur at 14 days and 42 days after infection , respectively [66] , [67] . Date of seroconversion was assumed to occur at the midpoint between most recent negative serological test and first reported positive test , unless additional information was available . For cross-sectional analyses , univariate and multivariate regressions were conducted assessing subregion lengths and number of glycosylation sites as a function of time since infection , stage of disease , CD4 count , HIV viral load , adjusting for sample source ( plasma vs . PBMC ) , and date of sampling ( calendar year ) . In regression analyses , to allow direct comparisons of the effect of each variable on V1V2 length and/or glycosylation , we compared β values ( i . e . , regression coefficients scaled such that each variable is equivalent to having a mean value of 0 and a standard deviation of 1 ) . Generalized estimating equations ( GEE ) were utilized to account for non-independence of data points [68]–[70] , and an exchangeable correlation structure was assumed . This method adjusts for the correlation of multiple sequences nested within a sample as well as multiple samples per patient . As an additional means of verifying that analysis outcomes were not influenced by data linkage , regression analyses were performed on replicate data subsets reconstituted from the original data by random resampling , including analyses on 100 data subsets each obtained by using one randomly selected sequence from each individual ( See Text S1 section S2 ) . To ensure that results were not unduly influenced by outlying sequences with extremely short or long loop lengths , analyses were repeated after excluding sequences representing the shortest 5% and longest 5% of the V1V2 loops in the dataset . For the longitudinal dataset , multivariate linear regressions were conducted assessing V1V2 length and number of glycosylation sites as a function of time since infection within a person , and the mean rate of change per year was estimated . Statistical analyses were performed using SAS version 9 . 1 ( SAS Institute , Cary , NC ) .
We obtained 5185 partial length HIV-1 env gene sequences for cross-sectional and longitudinal analysis by the methods described above ( Tables 1 & 2 ) . Sequences were isolated from 475 samples obtained from 154 individuals , including 27 from the MACS , 43 from the Seattle PIC and 84 from the HIVDB . Study subjects resided in North America ( N = 116 ) , Western Europe ( N = 25 ) , East Africa ( N = 2 ) , and Asia ( N = 11 ) , contributed a median of 14 sequences ( range 1–287 ) and included persons in stages 1 ( N = 41 ) , 2 ( N = 62 ) , 3 ( N = 40 ) , and 4 ( N = 27 ) of infection ( note that some subjects contributing to the longitudinal analysis were included at more than one stage of infection ) . Sequences were derived from plasma ( N = 2495 ) , PBMC ( N = 2620 ) and other sites ( N = 70 ) . Sequences were of subtype B ( N = 5013 ) and subtype A ( N = 172 ) . All subtype A sequences and sequences derived from sites other than blood were excluded from cross-sectional analyses , but were considered as special cases under longitudinal analyses ( sequence data available at: *webaddress pending acceptance* ) . In the longitudinal dataset , significant V1V2 length increases between first and second timepoints were noted in 10 of 22 subjects , a significant V1V2 length decrease over time occurred in one subject , and no significant V1V2 length changes over time were seen in the remaining 11 subjects . These findings appeared to vary by stage of infection ( t-test p = 0 . 03 ) . In the 15 patients from the L1 group ( individuals not meeting AIDS criteria at any time prior to final sampling ) , the mean increase of V1V2 length per subjects was 1 . 69 amino acids per year , and 9 subjects experienced significant V1V2 length increases over time ( Figures 4 and 5 ) . In contrast , of the seven subjects in the L2 group ( individuals progressing to AIDS between first and final sample ) , the mean V1V2 length decreased by an average of 0 . 10 amino acids per year , with only one having a significant trend of increasing length , while one individual showed a significant decrease in length ( Figure 6 ) . The distribution of V1V2 length change ( increase or decrease ) by group was therefore asymmetric ( Fisher's exact test , p = 0 . 02 ) , reflecting a trend of increasing length in asymptomatic individuals ( group L1 ) and stable or decreasing length in individuals with AIDS ( group L2 ) ( Table 4 ) . Three subjects in group L1 had extensive longitudinal sampling ( Figure 5 ) ; in 1362 and Q23 [51] , there was a period of V1V2 length stability of approximately 2 years , followed by increase through 4 . 5 years . V1V2 length increase over time was also seen in CC1 . In the case of CC1 , a pseudotyped virus was created using the gp120 coding region from the initial timepoint from this individual in a HIV-1 NL4-3 background , and cultured in vitro [54] . In contrast to the patterns observed in vivo , V1V2 length and number of glycosylation sites both declined rapidly over 20 generations in vitro ( p<0 . 001 ) .
We have systematically examined gp120 subregion length variation , and the relationship between length polymorphism , N-linked glycosylation sites , and clinical markers of disease progression . Although V1V2 , V4 and V5 all displayed remarkable length heterogeneity , and V1V2 , C3 and V4 were also quite variable with respect to glycosylation , the most significant associations between virological and clinical variables localized to the V1V2 region . We found that V1V2 length and glycosylation increased significantly over time during chronic infection , and then declined in late-stage illness . In regression analyses , time since infection was the most influential factor in determining V1V2 length . In addition , there was a modest but significant increase in V1V2 length over the period from 1984–2004 . V5 loop length was highly variable , but tended to decrease slightly in length over the course of infection . In SIV infection , the number of PNLGS in gp120 increases over time in vivo following inoculation of a cell-passaged strain [71] . In one earlier study in humans , Bunnik et al noted expansion in gp120 length followed by contraction over time in 4 of 5 individuals receiving antiretroviral therapy , and similar changes in glycosylation in 3 subjects [72] . Others have noted a relationship between early infection and reduced V1V2 length and glycosylation in subtypes C and A [19] , [45] . In contrast , a comparison of early and chronic HIV-1 subtype B sequences from the HIV sequence database failed to reveal any significant difference in V1V2 length [19] , suggesting that these effects may be subtype-specific . Data on length/glycosylation changes during transmission have been conflicting . Derdeyn et al [45] demonstrated reduced length and glycosylation in V1–V4 following heterosexual transmission in HIV-1 subtype C . However , Frost et al failed to note similar findings in a study of eight subtype B homosexual transmission pairs [47] , and in our examination of these and 10 additional subtype B infected homosexual transmission pairs , we found no consistent pattern of change in V1–V2 or V1–V4 length or glycosylation upon transmission [46] . Interpretation of the data presented here may be affected by several methodological factors . There is probably some variation in the accuracy of the reported time of infection for sequences obtained from previous reports . In some cases , sequences obtained from prior publications may have been obtained under conditions permitting template resampling [73] , and a systematic error due to evolving laboratory methods could result in bias . Also , in our analyses , we have not formally corrected for multiple comparisons . Physiological factors are also likely to introduce some noise , particularly in cross-sectional analyses of parameters with respect to time since infection . The individuals included here represent a broad spectrum of clinical scenarios , diverse host immune response profiles and varying disease progression rates . Plasma sequences may receive contributions from both recently infected target cells and older reservoirs , and therefore imperfectly reflect selective pressures prevailing at the time of infection . Finally , length and glycosylation phenotypes are likely to be affected by chance events and unknown factors not considered in our analyses . Therefore , the effects we describe are influential rather than deterministic , and reflect important selective forces that can be discerned against a background of high inter-individual variation . Despite these limitations , the analyses presented here and the work of others [40] , [45]–[47] , [72] provide the outlines of an overall pattern characterized by transmission of randomly selected V1V2 loop lengths from viruses present in the donor pool , a brief decline in loop size during the initial months immediately following infection , gradual selection for bulkier V1V2 loops during chronic infection , and finally , reversion to more compact loops during late stage illness . Structural studies [22] , [23] , neutralization studies [20] , [34]–[38] , [42] , and in vitro data on viruses lacking V1 and V2 [43] , [44] suggest that one major function of the V1V2 region may be to permit evasion from humoral immune responses in the host . Thus , the trends outlined above support the hypothesis that HIV populations may evolve to escape humoral selective pressure by increasing V1V2 loop size . According to this view , the newly infected , immunologically naïve host might be expected to harbor relatively short V1V2 loops that eventually lengthen in response to an effective humoral response at some fitness cost ( Figure S9 ) . Experimental evidence indicating that relaxation of antibody-mediated selective pressure during early infection is associated with shorter loops is provided by Derdeyn , who demonstrated significantly greater neutralization sensitivity among five recipients during early infection , than in the corresponding donors [45] . The decline in V1V2 size observed in advanced disease probably reflects waning effectiveness of humoral immunity in hosts with late-stage illness and profound immune dysregulation ( Figure 7 ) . This decline is also congruent with previous findings of an inverse relationship between the rate of HIV genetic evolution and the rate of CD4 T cell decline in some individuals [74] . The dramatic reduction in V1V2 length associated with transfer to the in vitro environment [54] represents the extreme case of absent host immunity , where viruses without an unnecessarily bulky V1V2 loop achieve maximum replicative fitness . As would be expected , the patterns we observe are most pronounced in plasma sequences , which most directly reflect the selective forces present at the time of sampling . In contrast , a significant increase in V1V2 length over time was not seen in the PBMC compartment . These observations are consistent with the presence of archived genotypes from earlier times during the course of infection within the PBMC compartment . We also note that genotypes present in plasma may emanate from other cellular compartments in addition to PBMC , and may therefore reflect somewhat different evolutionary pressures . However , a considerably greater number of V1V2 sequences were derived from plasma , and sample size may also account for some of the differences observed between these compartments . Our model may help to explain a failure to find any significant difference in V1V2 length in a comparison of early and chronic HIV-1 subtype B sequences ( including sequences from late-stage individuals ) [19] . When we reanalyzed the data presented by Chohan [19] after separating subjects with stable chronic illness from subjects with AIDS ( Figure S13 ) , we observed a pattern of lengthening over time , followed by decline in late-stage illness , as reported here ( See Text S1 , section S7 ) . Similarly , we may explain discordant results obtained on V1V2 length variation during transmission of HIV-1 subtypes C and B . While a trend towards shorter loops in recipients was seen in subtype C [45] but not B [46] , [47] , it is likely for methodological reasons that the subjects studied by Derdeyn were sampled at somewhat later times than those of Frost and Liu . Thus the sequences in the latter two studies would be expected to be a random sampling from the donor pool , while those of Derdeyn might reflect the expected shortening prior to the onset of an effective antibody response . Indeed , when we examine a much larger set of subtype A and C transmission pairs from East Africa with more precisely known sampling times obtained soon after transmission , it is difficult to appreciate any consistent pattern of V1V2 length change ( See Text S1 , section S8 and Figure S14 ) . Thus there may be no need to infer separate mechanisms for different HIV-1 subtypes and modes of transmission . In addition , we may also explain a trend of increasing V1V2 length by calendar year . If shorter and less glycosylated V1V2 were always selected during transmission , transmission from donors in early infection would maintain a constant V1V2 length within the epidemic , whereas if all new cases were acquired from chronically infected hosts , this increase of V1V2 length by calendar year could be dramatic . However , most studies suggest that about half of transmission events involve subjects in early infection [46] , [75] , [76] , consistent with the moderate trend we observed . Alternatively , the temporal trends we have observed could represent a gradual adaptation by HIV-1 to host the host environment at the population level , a hypothesis that has been proposed by several investigators with respect to mutational escape from HLA-restricted CTL epitopes [77]–[79] . Finally , our results imply that the polymorphisms seen in V1V2 reflect the ability of the host to mount a meaningful immunological response , rather than virologic features that dictate the course of illness . That is , we argue that V1V2 length change is a consequence of environmental selective pressure rather than a causative factor in disease progression . | The HIV envelope gene ( env ) encodes viral surface proteins ( Env ) that are vital to the basic processes used by the virus to infect and cause disease in humans . Adaptations in env determine which cells the virus can infect , and permit the virus to avoid elimination by the immune system . Env is one of the most variable genes known , and it can change dramatically over time in a single individual . However , Env-host cell interactions are complex and incompletely understood , and changes in this viral protein during infection have not yet been systematically described . We examined a large number of env sequences from 154 individuals at various stages of HIV infection but who had never received antiretroviral treatment . We found that the env V1V2 region lengthens during chronic infection and becomes more heavily glycosylated . However , these changes partially reverse during late-stage illness , possibly in response to a weakening host immune system . V1V2 lengths are also increasing over time in the epidemic at large , possibly related to the epidemiology of HIV transmission within the subtype B epidemic . These results provide fundamental insights into the biology of HIV . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"evolutionary",
"biology/microbial",
"evolution",
"and",
"genomics",
"infectious",
"diseases",
"infectious",
"diseases/sexually",
"transmitted",
"diseases",
"virology/virus",
"evolution",
"and",
"symbiosis",
"virology/virion",
"structure,",
"assembly,",
"and",
"egress",
"viro... | 2010 | HIV-1 Envelope Subregion Length Variation during Disease Progression |
A potent therapeutic T-cell vaccine may be an alternative treatment of chronic hepatitis B virus ( HBV ) infection . Previously , we developed a DNA prime-adenovirus ( AdV ) boost vaccination protocol that could elicit strong and specific CD8+ T-cell responses to woodchuck hepatitis virus ( WHV ) core antigen ( WHcAg ) in mice . In the present study , we first examined whether this new prime-boost immunization could induce WHcAg-specific T-cell responses and effectively control WHV replication in the WHV-transgenic mouse model . Secondly , we evaluated the therapeutic effect of this new vaccination strategy in chronically WHV-infected woodchucks in combination with a potent antiviral treatment . Immunization of WHV-transgenic mice by DNA prime-AdV boost regimen elicited potent and functional WHcAg-specific CD8+ T-cell response that consequently resulted in the reduction of the WHV load below the detection limit in more than 70% of animals . The combination therapy of entecavir ( ETV ) treatment and DNA prime-AdV boost immunization in chronic WHV carriers resulted in WHsAg- and WHcAg-specific CD4+ and CD8+ T-cell responses , which were not detectable in ETV-only treated controls . Woodchucks receiving the combination therapy showed a prolonged suppression of WHV replication and lower WHsAg levels compared to controls . Moreover , two of four immunized carriers remained WHV negative after the end of ETV treatment and developed anti-WHs antibodies . These results demonstrate that the combined antiviral and vaccination approach efficiently elicited sustained immunological control of chronic hepadnaviral infection in woodchucks and may be a new promising therapeutic strategy in patients .
Chronic hepatitis B virus ( HBV ) infection is still one of the major public health problems . Two billion people worldwide have been infected with HBV , of whom more than 360 million have developed chronic infection . Approximately one million patients die from HBV-associated liver diseases such as cirrhosis and hepatocellular carcinoma ( HCC ) every year . Over the past 10 years , the treatment options of chronic HBV infection have improved greatly . Currently , the two types of antiviral therapies are approved: treatment with pegylated interferon alpha 2a ( PEG-IFNα ) or nucleot ( s ) ide analogues , such as entecavir ( ETV ) and tenofovir . However , these therapies have still several limitations . The treatment with PEG-IFNα leads to a sustained antiviral response in only one third of patients [1] , regardless of combining the therapy with nucleot ( s ) ide analogues [2] , and it is frequently associated with serious side effects . The treatment with nucleot ( s ) ide analogues significantly suppresses HBV replication but cannot completely eradicate the virus . After withdrawal of the treatment , a rebound of viremia is observed in the majority of patients . Therefore , the alternative strategies to treat chronic HBV infection are still urgently needed . The host immune response determines whether acute HBV infection will progress to resolution or chronicity . An early and multi-specific immune response to HBV antigens is associated with the clearance of HBV [3] , [4] . In contrast , a weak or often undetectable HBV-specific immune response correlates with HBV persistence [5]–[8] . Thus , it is assumed that therapeutic vaccination could enhance the virus-specific immune responses contributing to control or even clearance of chronic HBV infection . Early therapeutic vaccines were based on the recombinant HBV surface antigen ( HBsAg ) protein vaccines [9]–[19] . These vaccines proved to be excellent in their prophylactic potential , but were unfortunately not effective in chronically infected patients . A DNA vaccine expressing small and middle HBV envelope proteins was also tested in chronic HBV carriers but failed to elicit sustained HBV-specific cellular immune response [20] . Moreover , administration of this vaccine to HBV carriers pre-treated with nucleoside analogues did not induce any therapeutic effect in a recent clinical trial ( S . Pol , personal communication ) . A vigorous T-cell response against HBV core antigen is crucial for the resolution of the infection but is predominantly absent in chronic hepadnaviral infections [21]–[26] . Thus , using T-cell vaccines targeting the core protein may be a potent therapeutic strategy . We hypothesized that improved WHcAg-based T-cell vaccines might be crucial to achieve sustained antiviral immunological responses . Therefore , we developed new vaccines in the woodchuck model , a proven preclinical model to study innovative prophylactic and therapeutic strategies against HBV infection . We constructed a new DNA plasmid ( pCGWHc ) and adenoviral vectors serotype 5 ( Ad5WHc ) and chimeric Ad5 displaying Ad35 fiber ( Ad35WHc ) showing high expression levels of WHcAg . These new vaccines were tested recently in mice and naïve woodchucks [27] . We showed that the DNA prime – AdV boost immunization significantly improved the magnitude of WHcAg-specific T-cell responses in mice , far beyond the previously tested by us strategies . Moreover , for the first time , we were able to induce detectable WHcAg-specific proliferative and cytotoxic T-cell responses in naïve woodchucks using these optimized vaccines . We demonstrated that heterologous Ad5WHc-Ad35WHc regimen was superior in priming of WHcAg-specific T-cell responses , compared to only DNA immunization in the woodchuck model [27] . Nevertheless , using of only recombinant adenoviruses in vaccinations regime limits the number of immunizations , due to induction of neutralizing antibodies against the structural components of the vector . Therefore , DNA prime-AdV boost immunization seems to be a rational approach which combines increased efficacy of the vaccination regime with the possibility of using multiple immunizations necessary to break immune tolerance to the targeted antigens in chronically-infected individuals . In the present study we evaluated the optimized DNA prime – AdV boost immunization first in WHV transgenic mice and then in chronically WHV-infected woodchucks . The WHV transgenic mouse , carrying a 1 . 3 fold overlength WHV transgenome , is a new animal model with well established immunological tools to determine virus-specific T-cell response . WHV replication occurs specifically in the liver and WHV particles could be produced and released into the bloodstream . We found that WHV transgenic mice are not completely tolerant to WHV proteins and WHV-specific T-cell responses could be primed by DNA vaccines , though at a low level ( our unpublished results ) . Thus , this model was ideal to evaluate our new prime-boost immunization regimen prior to the therapeutic vaccination experiment performed in chronically WHV-infected woodchucks . We could show that our new immunization strategy was able to induce effective immune responses to WHV antigens and reduce WHV replication WHV transgenic mice . In chronically WHV-infected woodchucks we combined WHcAg-based DNA prime-AdV boost vaccinations with WHsAg-expressing plasmid , in order to achieve the most favorable therapeutic effect . Following the idea that the reduction of viral loads by the nucleoside analogues pre-treatment could enhance the effect of therapeutic immunization [28]–[30] , we used a potent antiviral drug entecavir ( ETV ) . In contrast to lamivudine therapy [31] , [32] , treatment with ETV proved to efficiently suppress WHV replication in chronically infected woodchucks [33] , however , does not lead to the resolution of the infection . The results showed that our new combination therapy improved WHV-specific immune responses and led to long term viral control , induction of neutralizing anti-WHs antibodies , and viral clearance in some animals .
We have shown recently that immunization of naïve C57BL/6 mice in heterologous DNA prime – AdV boost manner using vaccines expressing WHcAg induces remarkably vigorous and potent WHcAg-specific response . [27] . We investigated whether this new vaccination protocol is able to break the WHV-specific immune tolerance and reduce the WHV replication in WHV transgenic mouse model . Thereby , mice were primed twice with the pCGWHc plasmid in a two-week interval and afterwards were boosted once with Ad5WHc or pCGWHc , or boosted twice with Ad5WHc followed by Ad35WHc . As shown in Fig . 1A , the levels of WHcAg-specific antibodies ( anti-WHc ) were significantly higher in WHV Tg mice that received boosting immunization with Ad5WHc than in group of mice immunized three times with DNA vaccine ( P<0 . 05 ) . The level of anti-WHc increased additionally in the group of mice after the fourth ( second boost ) immunization with Ad35WHc ( P<0 . 005 ) . As expected control mice mice did not induce any anti-WHc antibodies ( P<0 . 0005 , compared to all groups of immunized mice ) . The levels of anti-WHc antibodies were comparable in all mice immunized with pCGWHc plasmid either once or twice ( data not shown ) . Detection of IgG isotypes demonstrated that all tested immunization protocols induced predominantly IgG2a antibodies ( Fig . 1B ) . The anti-WHc antibodies of IgG1 subclass were only observed in group of mice immunized in DNA-Ad5WHc-Ad35WHc manner ( Fig . 1C ) ( P<0 . 005 , compared to other vaccination groups ) . Interestingly , we could detect WHsAg-specific antibodies ( anti-WHs ) in groups of mice that were immunized with DNA prime – AdV boost regimens , but not in mice immunized only with DNA ( Fig . 1D ) . The anti-WHs antibodies were detected in the sera of 14 out of 17 mice after the boosting immunization with Ad5WHc ( P<0 . 005 ) . The levels of anti-WHs increased additionally after the second boosting immunization with Ad35WHc ( P<0 . 005 ) . The control experiment in WHV transgenic mouse strain 1218 ( harbouring a mutated WHV transgenome lacking WHsAg ) showed that DNA-Ad5WHc-Ad35WHc immunization induces the same pattern of WHc-specific antibodies . However , no anti-WHs were detected after Ad35WHc immunization ( supplementary Fig . S1 ) . These data indicate that a potent immunization with WHcAg is able to boost a weak anti-WHs response by intermolecular help mechanism if WHsAg is present [34] . In the next step , we evaluated the impact of DNA only , heterologous DNA–Ad5WHc , or DNA–Ad5WHc-Ad35WHc regimens on induction of cellular immune response 2 weeks after the last immunization . In the first step , we assessed the presence of antigen-specific T-cells within the splenic lymphocytes ex vivo , using WHcAg-derived peptide c13-21-specific dimer . As Fig . 2A shows , we could detect WHcAg-specific CD8+ T-cells in the spleens of mice immunized with WHcAg-expressing vaccines but not in ‘empty’ pCG-Ad5GFP control group . The mean background values of the assay obtained for these controls was 0 . 16% and was comparable to 0 . 15% detected in the isotype controls ( data not shown ) . The mean values of WHcAg-specific CD8+ T-cells in the spleens of mice immunized with DNA-only was 0 . 47% , with DNA-Ad5WHc was 0 . 83% , and with DNA-Ad5WHc-Ad35WHc was 0 . 52% ( Fig . 2B ) ( P<0 . 05 , compared to controls ) . The magnitude of the WHcAg-specific CD8+ and CD4+ T-cell responses elicited by the various vaccination regimens was compared by the intracellular IFNγ staining of splenocytes . Splenocytes were isolated two weeks after the last immunization and were stimulated in vitro for 7 days with the CD8+ T-cell epitope c13-21 . The percentages of IFNγ+ CD8+ T-cells determined in the spleens of mice vaccinated in the DNA prime – Ad5WHc boost manner ( mean 5 . 25% ) were significantly higher in comparison to the only DNA-immunized group ( mean 1 . 18%; P<0 . 0005 ) ( Fig . 2C , E ) . Unexpectedly , the magnitude of IFNγ response did not increase in the group of mice that received the second boosting immunization with Ad35WHc . The mean percentages of IFNγ+ CD8+ T-cells directed against c13-21 in this group were 4 . 26% and were slightly lower than in the DNA-Ad5WHc group . All immunization protocols were able to induce significant IFNγ secretion by CD4+ T-cells in response to stimulation with peptide c131-145 ( P<0 . 05 ) ( Fig . 2D ) . Nevertheless , no statistically significant difference in percentages of IFNγ+ CD4+ T-cells between the groups immunized with WHcAg-expressing vaccines was detected ( Fig . 2F ) . Further on , we compared the effector functions of the CD8+ T-cells induced by the various immunization regimens , such as degranulation capacity and secretion of the other antiviral cytokines e . g . TNFα and IL-2 . The ability of the CD8+ T-cells to degranulate was measured by flow cytometric detection of the CD107a marker [35] , [36] on the surface of the lymphocytes expanded in vitro for 7 days with the c13-21 epitope . The results show that the percentages of CD107a+ CD8+ T-cells were significantly higher ( P<0 . 005 ) in the groups of mice boosted once with Ad5WHc or twice with Ad5WHc and Ad35WHc ( mean values: 6 . 3% and 5 . 2% , respectively ) , than in the pCGWHc only immunized mice ( mean value: 2 . 6% ) ( Fig . 3A ) . The production of TH1 type cytokines by CD8+ T-cells , such as IFNγ , TNFα and IL-2 , was evaluated in the splenocytes ex vivo , after 6 h stimulation with the peptide c13-21 . As presented in Fig . 3B , all vaccination protocols induced significant percentages of CD8+ T-cells positive for all of the tested cytokines compared to mean background values ( 0 . 06%–0 . 08% ) detected in mice immunized with ‘empty’ pCG plasmid and boosted with Ad5GFP ( P<0 . 05 ) . The percentages of IFNγ+ or TNFα+ CD8+ T-cells were detectable at similar levels and the percentages of IL-2+ CD8+ T-cells were at slightly lower levels . Group of mice primed with pGCWHc and once boosted with Ad5WHc showed the highest percentages of IFNγ+ , TNFα+ , and IL-2+ CD8+ T-cells ( the mean values: 0 . 48% , 0 . 49% , and 0 . 39% respectively ) . The frequencies of CD8+ T-cells positive for all of the tested cytokines in this group ( 2×pCGWHc-Ad5WHc ) were significantly higher compared to the DNA only - immunized group ( the mean values: 0 . 28% , 0 . 26% and 0 . 24% , respectively ) ( P<0 . 05 ) . Mice immunized four times with DNA-Ad5WHc-Ad35WHc regimen exhibited significantly higher frequencies of only IFNγ-positive and TNFα-positive CD8+ T-cells compared to the DNA only – immunized group ( the mean values: 0 . 45% and 0 . 34% , respectively ) . In addition , we analysed the effector functions of hepatic WHcAg-specific T-cells induced by 2×pCGWHc-Ad5WHc immunization , as the liver is the major compartment of WHV replication in WHV transgenic mice . First , we evaluated the presence of hepatic antigen-specific T-cells ex vivo , using WHcAg-derived peptide c13-21-specific dimer . As shown in Fig . 3C , we detected WHcAg-specific CD8+ T-cells in the liver of mice immunized with the DNA prime – Ad5WHc boost regimen , but not in naïve WHV transgenic mice ( P<0 . 005 ) . The mean frequencies of WHcAg-specific CD8+ T-cells in the liver of immunized mice was 1 , 9% . Moreover , the percentages of dimer+ CD8+ T-cells in the liver were significantly higher compared to these detected in the spleen ( 0 , 6%; P<0 . 05 ) ( Fig . 3C–D ) . The frequencies of CD8+ and CD4+ T-cells producing IFNγ , TNFα and IL-2 ( detected ex vivo after 6 h stimulation with the CD8+ or CD4+ T-cell epitopes ) , were also higher in the liver than in the spleen of mice immunized with 2×DNA-Ad5WHc regimen . As presented in Fig . 3E , the immunization induced significant percentages of IFNγ+ , TNFα+ or IL-2+ CD8+ T-cells ( 2 , 97% , 2 , 11% , 0 , 62% , respectively ) compared to mean background values detected in naïve WHV transgenic mice ( 0 , 20% , 0 , 24% , 0 , 08% , respectively ) ( P<0 . 005 ) . Similarly , the mean frequencies of IFNγ+ , TNFα+ or IL-2+ CD4+ T-cells in immunized mice ( 1 , 30% , 1 , 10% , 0 , 40% , respectively ) were significantly higher , compared to the corresponding values detected in the naïve mice ( 0 . 37% , 0 . 28% and 0 . 20% , respectively ) ( P<0 . 05 ) ( Fig . 3F ) . We examined the impact of the WHcAg-based immunizations on the WHV replication in 1217 WHV Tg mice . The viral loads were monitored in the serum of mice before the immunizations were performed ( week −1 ) and afterwards , at the time point of sacrifice ( week 8 for single boost groups and 12 for double boost , respectively ) . As expected , in the control group of mice no difference in the viral loads in serum at the beginning and at the end of the experiment was observed ( Fig . 4A ) . In the group immunized three times with plasmid DNA – pCGWHc only , four out of twelve mice ( 33% ) had undetectable viral loads at the end of the experiment ( Fig . 4B ) . Other mice except two , showed significant 1 to 2 log decrease in viral load after the immunizations ( P<0 . 05 ) . As showed in Fig . 3C–D , mice immunized in the heterologous prime – boost manner using recombinant adenoviral vectors demonstrated the most significant reduction in viral loads ( P<0 . 0005 ) . At the end time point , the WHV DNA was undetectable in 13 out of 17 mice from the group boosted once with Ad5WHc ( 77% ) ( Fig . 4C ) . In the group of mice that received the fourth immunization with Ad35WHc , nine out of twelve mice ( 75% ) exhibited the WHV viremia below the detection limit at the end of the experiment ( Fig . 4D ) . We evaluated the effectiveness of heterologous DNA prime – AdV boost immunization as the therapeutic vaccine in chronically WHV-infected woodchucks . To increase the effect of the vaccination we used antiviral pretreatment with entecavir to reduce the WHV replication . The drug was administered for 23 weeks . Starting from week 8 , four animals received in total 9 sequential intramuscular immunizations with DNA plasmids expressing WHcAg and WHsAg , Ad5WHc , and Ad35WHc as shown in Fig . 5A . Two animals treated only with ETV served as controls . We included WHsAg expressing plasmid into the vaccination schedule , as the results obtained in WHV transgenic mice demonstrated that a certain amount of WHsAg is necessary to stimulate B-cells to produce anti-WHs antibodies ( Fig . 1; Fig . S1 ) . Therefore , including WHsAg as a part of vaccine may increase the immunotherapeutic effect . Moreover , after the administration of adenoviral vectors , we performed two additional DNA immunizations to maintain the induced WHV-specific T-cell responses . The WHV-specific T helper ( TH ) response was evaluated by 2[3H]adenine-based proliferation assay of woodchuck PBMCs stimulated with the known TH epitopes The significant WHV-specific proliferative responses ( SI≥3 . 0 ) were detectable in PBMCs of all chronically WHV-infected woodchucks that received the vaccinations ( DNA/Ad5WHc/Ad35WHc ) and ETV ( Fig . 5B–C ) , but not in animals treated only with ETV ( SI values ranging from 0 . 3 to 2 . 3 for all examined peptides , at all tested time points; data not shown ) . Except of woodchuck 61793 , which showed a transient response against WHsAg-derived peptide 336–351 ( SI = 6 . 8 ) shortly after beginning of ETV treatment ( week 6 ) , the other woodchucks from the combination therapy demonstrated the proliferation of virus-specific T-cells during the immunizations . Two out of four vaccinated woodchucks ( number 61792 and 61793 ) showed significant proliferative responses already after two immunizations with plasmid DNA vaccine ( week 12 of therapy ) . The detected responses were directed against WHsAg-derived peptides: s224-239 , s252-263 , or s420-431 ( SI ranging from 3 . 2 to 22 . 4 ) ( Fig . 5C ) and one WHcAg-derived peptide c85-100 ( SI = 3 . 0; woodchuck 61792 ) ( Fig . 5B ) After three DNA injections ( week 14 of therapy ) , all four immunized woodchucks showed significant proliferative responses against WHsAg-derived peptides ( SI ranging from 3 . 2 to 3 . 8 ) . In addition , 2 out of 4 woodchucks demonstrated WHcAg-specific TH responses . The WHV-specific proliferative responses were present in most of the woodchucks until week 25 ( two weeks after the last ETV treatment and 3 weeks after the second Ad35WHc/pWHsIm immunization ) . We identified 5 WHsAg-specific and 5 WHcAg-specific T-cell epitopes ( see supplementary Table S1 ) . The WHsAg-specific proliferation was predominantly directed against peptides s224-239 and s252-267 . The most frequently recognized WHcAg-derived peptides were: c64-79 and c117-132 . The evaluation of cytotoxic T-cell response against previously identified epitopes c96-110 and s220-234 [37] was performed by CD107a degranulation assay . The population of CD3+ CD4− lymphocytes was considered to be the CD8+ T-cells as there is no specific anti-woodchuck CD8+ antibody . The WHV-specific degranulation responses were not detectable in most of the animals before week 22 of the treatment . At week 4 of ETV therapy , only a brief elevation in percentages of WHcAg- and WHsAg-specific CTLs were observed in three WHV chronic carriers ( woodchucks 61786 , 61791 , and 61795 ) , as shown in Fig . 6 . This result indicates that a decrease of WHV replication by entecavir treatment is accompanied by a transient restoration of T-cell functions . All woodchucks from the combination therapy group and control animals had comparable background percentages of WHcAg- and WHsAg-specific CTLs at the beginning of the immunization phase ( week 8 ) ( Fig . 6A–B ) . The WHcAg-specific T-cell responses appeared in two immunized woodchucks 61792 and 61793 at week 22 ( Fig . 6A , C ) . The percentages of 1 . 51% and 1 . 43% of WHcAg-specific CTLs detected for woodchuck 61792 and 61793 , were three fold higher than the mean background value of 0 . 43% calculated for the negative controls ( unstimulated cells and cells stimulated with unrelated CMV-derived peptide ) of all woodchucks at all time points . The WHcAg-specific degranulation response was present in all 4 woodchucks from combination therapy group until the last monitored time point week 29 . The peak of WHcAg-specific CTLs detected in peripheral blood of the immunized woodchucks was detected at week 27 of treatment; the percentages of CD107a+ CD3+ CD4− T-cells were ranging between 1 . 2%–2 . 1% ( mean: 1 . 7% ) . The ETV only treated controls did not show any significant WHcAg-specific T-cell response ( Fig . 6D ) . The mean percentages of WHcAg-specific CTLs detected from week 22 to 29 were ranging between 0 . 29% and 0 . 55% and were comparable with the mean background value ( 0 . 43% ) . The WHsAg-specific CTL responses were not as prominent as the responses directed against WHcAg and appeared only transiently ( Fig . 6C ) . Nevertheless , woodchucks that received the combination therapy demonstrated higher percentages of WHsAg-specific CTLs ( mean 0 . 9–1 . 0% , weeks 25–29 ) in comparison to background values detected for only ETV treated animals ( mean 0 . 3–0 . 4 , weeks 25–29 ) ( Fig . 6B ) . The peak of WHsAg-specific degranulation response detected in peripheral blood of WHV chronic carriers that received immunization differed in time . At week 22 the highest percentages of WHsAg-specific CTLs were detected for 61792 ( 1 . 38% ) , at week 25 for 61789 ( 1 . 49% ) , at week 29 for 61786 and 61793 ( 1 . 63 and 1 . 34% respectively ) . The representative dot-plots of WHcAg- and WHsAg-specific CD107a degranulation responses are shown in supplementary Fig . S2 . We investigated the impact of the combination therapy on WHV replication , WHsAg levels , development of anti-WHs antibodies , and liver inflammation by measurements of liver transaminase - GOT . The baseline values of the viral loads prior to ETV therapy in WHV chronic carriers enrolled in the experiment was ranging from 3 . 1×109 to 1 . 2×1011 WHV GE/ml of serum . The initial levels of WHsAg significantly varied between the individual animals and were ranging from 96 . 20 µg/ml ( woodchuck 61792 ) to 693 . 83 µg/ml serum ( woodchuck 61786 ) . Only two woodchucks 61786 and 61795 demonstrated elevated GOT level in the serum at the beginning of the experiment ETV treatment ( 67 , and 150 IU/l , respectively ) . The WHV load decreased for approximately 5-logs during the first 8 weeks of the ETV pre-treatment period in all examined woodchucks ( Fig . 7A–B ) . At the time of the first immunization ( week 8 ) , no significant difference in the viral loads between the woodchucks from combination therapy group and ETV only treated controls was observed ( Fig . 7C ) . Only woodchuck 61792 showed the viral load below the detection limit at this time point . Between weeks 12 to 19 of therapy all woodchucks remained WHV negative in the blood . Following the decrease in viral load , levels of WHsAg in the sera decreased significantly in all woodchucks during the ETV treatment . The serum GOT initially reached levels below 50 IU/l in the sera of most of the examined woodchucks at weeks 12–14 of the therapy , indicating the reduction of liver inflammation by ETV-mediated decrease of WHV replication ( Fig . 7A–B ) . From the combination therapy group , only woodchuck 61789 showed significant elevation in GOT levels between week 10 and 14 ( 73–122 IU/ml ) , indicating a massive cytotoxic T-cells activity in the liver . The GOT levels for other woodchucks from this group slightly increased and were fluctuating around the value of 50 IU/ml . This observation suggests that the therapy induced a gradual elimination of the virus from the liver by the WHV-specific T-cells , without “acute” hepatotoxic effect . Starting from week 16 , the constant elevation in serum GOT levels was observed in one of the only ETV-treated control woodchucks: 61795 ( 75–90 IU/ml ) . This woodchuck did not show any WHV-specific T-cell response , and these elevated GOT levels might be a symptoms of progressing liver disease . After the end of ETV treatment , the GOT levels rapidly increased in both control WHV carriers and reached the values 106 IU/l in woodchuck 61791 and 845 IU/l in woodchuck 61795 at the end of monitoring period ( weeks 31–33 ) . At the same time point , the GOT values in woodchucks that received immunizations ( 61792 , 61793 , 61786 , and 61789 ) were considerably lower ( 12 , 17 , 21 and 71 IU/l ) . As shown in Fig . 7A , the two woodchucks from the combination therapy group ( 61792 and 61793 ) were WHV negative until the end of the monitoring period ( week 62 and 31 , respectively ) . Moreover , these two woodchucks became anti-WHs positive ( week 22 and 19 , respectively ) . Woodchuck 61792 had one of the lowest serum WHsAg levels ( 96 . 20 µg/ml ) at the beginning of the experiment . The WHsAg became detectable but not quantifiable at week 12 of the therapy and was finally cleared from serum of this woodchuck at week 52 . The effect of the combination therapy on the WHsAg in the other woodchuck 61793 , which developed anti-WHs antibodies , is difficult to assess due to the short monitoring period . At week 31 the animal had to be euthanized due to serious health problems not related to WHV infection ( bacterial infection ) . Woodchuck 61793 showed over 4-times higher WHsAg levels ( 424 . 32 µg/ml ) at week 0 than woodchuck 61792 . In addition , the reduction in the WHsAg due to the antiviral treatment was not so prominent in woodchuck 61793 . After anti-WHs development at week 19 , the level of WHsAg dropped in the serum for about 30% within 3 weeks . At the end of the monitoring period woodchuck still showed decreasing tendency in WHsAg levels . The other woodchucks receiving the immunizations ( animals 61786 and 61789 ) showed rebound of viremia and increased WHsAg levels at the end of the experiment . These woodchucks did not seroconverted to anti-WHs . Nevertheless , our results clearly show that heterologous DNA prime – AdV boost regimen leads to prolonged suppression of WHV replication ( 5 to 7 weeks ) , and as a consequence higher decrease in WHsAg in woodchucks 61786 and 61789 , compared to only ETV-treated control animals ( Fig . 7C ) . Reappearance of WHV DNA in the serum of the control woodchucks was seen at the end of ETV therapy ( week 22 ) . The woodchuck 61786 and 61789 showed a rebound of the viremia at week 27 and 29 , respectively . As shown in Fig . 7D , these animals showed a more prominent decrease in the WHsAg levels at week 22 [81% and 91%] than control woodchucks 61791 and 61795 [74% and 45%] . At the end of monitoring period , woodchucks from combination therapy group showed 23% , 17% , and 40% of the baseline WHsAg levels ( 91793 , 61786 , 61789 , respectively; the very low amount of WHsAg was not quantifiable for woodchuck 61792 at that time ) , whereas controls had 63% and 151% of baseline WHsAg level . In addition , we evaluated the replication of WHV in the liver samples collected post-mortem or through liver biopsy . Figure 8 shows the Southern blotting of WHV replicative intermediates , corresponding to the single-stranded DNA ( ssDNA ) and relaxed circular DNA ( RC DNA ) . The two woodchucks which were WHV DNA negative in the serum and developed anti-WHs antibodies ( 61792 and 61793 ) showed no or very low WHV replication in the liver . The other WHV chronic carriers showed comparable levels of WHV replication at the time points of sacrifice .
Studies in preclinical models of HBV infection such as woodchucks and chimpanzees as well as patients underline the important role of HBV-specific T-cell response as a leading factor of viral clearance [3] , [21] , [23]–[26] . In the presented study , we demonstrated that the heterologous DNA prime – recombinant AdV boost immunization is able to induce an effective virus-specific T-cell response to WHV antigens and efficiently suppress the viral replication in WHV transgenic mice and chronically WHV-infected woodchucks . Consistent to our previous studies in mice [27] the heterologous DNA prime – AdV boost regimens proved to be superior to DNA-only regimen also in WHV Tg mice . Mice immunized in DNA prime - Ad5WHc boost manner developed significantly higher levels of anti-WHc antibodies , and significantly more WHcAg-specific CD8+ T-cells in comparison to the group of animals immunized with DNA only . Interestingly , the fourth immunization with Ad35WHc ( DNA-Ad5-Ad35WHc ) was correlated with the development of anti-WHs antibodies . This experiment shows , that core-specific T helper response is able to prime WHsAg-specific B cells , consistent with the concept of “intermolecular help” [34] . At the same time , the magnitude of CD8+ T-cell response did not increase after Ad35WHc boost and remained comparable to mice that received only Ad5WHc . As these mice show inbred tolerance to WHV proteins , the maximal possible level of WHV-specific T-cell responses that could be induced by vaccination was probably achieved already after single Ad5WHc boost . However , WHV-specific antibody could be further expanded , as our vaccines were specifically designed to prime vigorous T-cell responses , but were quite poor inducers of humoral immune response ( especially DNA vaccine ) . Even though the strength of induced CD8+ T-cell response in WHV Tg mice was approximately 10 times lower than that detected in naïve mice [27] , the immunization with optimized vaccines was able to significantly reduce the WHV load in these mice . This effect was clearly due to potent and functional WHcAg-specific T-cell responses in the liver , the major compartment of WHV replication . The pronounced suppression in WHV replication was observed in DNA-Ad5WHc or DNA-Ad5-Ad35WHc vaccination groups , in which more than 70% of mice had undetectable viral loads at the end of the experiment . It was previously demonstrated that combination of ETV treatment and prime –boost vaccination with DNA and recombinant fowlpoxvirus expressing core and surface antigens of duck hepatitis B virus ( DHBV ) prevented the development of persistent infection in ducks [38] . Here , we examined the efficacy of the new DNA prime – AdV boost vaccination in combination with entecavir for the treatment of already established chronic hepadnaviral infection ( more than 1 year ) in woodchucks . There is an open scientific debate suggesting that multiple and frequent administration of the vaccine may be advisable in treatment of chronic hepatitis [39] , [40] . Chronically infected individuals usually exhibit the immune tolerance to targeted antigens . Therefore , the effect of the immunization results in much weaker response that and rarely reaches the level that would be expected in naïve patients or animals . Our immunization regimen based on 2–3 week time intervals between the immunizations enhanced the antiviral effect of ETV monotherapy and induced improved WHV-specific immune responses , resulting in the long term suppression of viral replication , and subsequently viral clearance in some animals . Previous reports indicate that combination of antiviral treatment and therapeutic vaccination may partially restore WHV-specific T-cell responses in chronic WHV carriers [30] , [32] . In our study , WHV-specific helper and cytotoxic T-cell responses were detected in PBMCs of all four chronically WHV-infected woodchucks that received the therapeutic vaccine . This outcome was clearly an effect of the improved therapeutic DNA prime–recombinant adenovirus boost immunization strategy , since ETV-only treatment did not induce sustained and significant T-cell responses in the control animals . The WHsAg-specific proliferative responses were predominantly directed against the peptides s224-239 and s252-267 . The position of the epitope s224-239 overlapped with the peptide s226-245 , preferentially recognized in WHV chronic carriers after clevudine/WHsAg combination therapy [30] . Moreover , several WHcAg-derived TH epitopes identified in woodchucks with acute self-limited WHV infection [3] were recognized in WHV carriers . Following the appearance of T helper cells , WHcAg- and WHsAg-specific CTLs were detectable in all woodchucks that received combination therapy . Only brief CTL responses were detected shortly after the beginning of ETV treatment . Those findings are consistent with the data obtained from chronic HBV patients treated with nucleoside analogues [28] , [29] . The appearance of sustained WHcAg-specific CTL response after the vaccinations was observed after Ad5WHc and Ad35WHc administration . All four woodchucks receiving vaccinations demonstrated significant WHcAg-specific CTL responses , whereas the WHsAg-specific CTL responses were not as prominent and appeared only transiently . Nevertheless , the contribution of these WHsAg-specific T-cell responses to overall therapeutic effect cannot be decisively assessed . Due to the high costs of woodchucks and long experimental periods , we could not yet provide the detailed answer whether both components are required for a successful therapeutic vaccination . This question needs to be addressed in the future studies on the larger number of woodchucks . The crucial criteria for resolution of HBV infection in humans are reduction of HBV load below the detection limit , loss of HBsAg and seroconversion to anti-HBs [41] . Colonno et al . reported that long-term ETV treatment ( over 1 to 3 years ) may be associated with partial control of WHV replication post-treatment [33] . These woodchucks showed very low levels of viral DNA , however , none of them developed anti-WHs antibodies . In addition to this study , we currently performed two other independent studies ( that are to be published soon ) , including in total 8 chronic WHV carriers treated only with ETV . The results of these studies are consistent with the findings described here . All woodchucks showed the rebound of WHV replication shortly after the end of ETV treatment ( our unpublished results ) . Our results demonstrate that the two immunized woodchucks ( 61792 and 61793 ) were WHV negative until the end of the monitoring period and developed anti-WHs as a proof of a sustained antiviral response . Interestingly , these therapeutic effects seem to be associated with induction of WHV-specific T-cell response by vaccination . These two woodchucks showed the earliest , strongest and the most sustained WHV-specific T-cell responses of all tested animals . In addition , woodchuck 61792 was negative for WHV replicative intermediates in the liver and cleared WHsAg from the blood . The process of the WHsAg clearance in woodchuck 61792 took more than 52 weeks . Since the WHV genome is integrated within the woodchuck genome [42] and a distinct pathway of surface antigen secretion was recently demonstrated [43] we assume that WHsAg clearance was achieved through elimination of the WHV-infected hepatocytes by virus-specific cytotoxic T-cells as well as hepatocyte turnover process . Woodchuck 61793 had to be prematurely euthanized due to a bacterial infection , and therefore was monitored for a much shorter period ( 31 weeks ) than woodchuck 61792 ( 64 weeks ) . At the time point of sacrifice , WHV ssDNA replicative intermediate was still present in the liver of woodchuck 61793 , implying that the WHV replication could be not completely suppressed in this animal . Nevertheless , residual HBV replication in the liver can be also observed in the patients who recovered from acute hepatitis B , despite the presence of anti-HBs . These patients are not viremic due to the high neutralizing anti-HBs antibody concentration , which prevent the reinfection of the uninfected hepatocytes [7] , [44] . Woodchuck 61793 developed anti-WHs , and the WHsAg load was continuously decreasing at the end of the monitoring period in this woodchuck . As the clearance of WHsAg in woodchuck 61792 , with approximately 4-times lower baseline level took a considerate amount of time ( 52 weeks ) , this decreasing tendency may indicate that long-term control of WHV infection in woodchuck 61793 was achieved . Some studies in chronically HBV-infected patients suggest that a low baseline HBsAg levels are associated with control of replication post therapy with nucleos ( t ) ide analogues [45] , [46] . However , no information about the influence of HBsAg levels on the effects of immunotherapeutic approaches is available in patients yet . In our previous study , using lamivudine treatment together with WHsAg/anti-WHs immune complexes vaccination , we observed the induction of anti-WHs antibodies predominantly in the woodchucks with high baseline WHsAg loads [32] . Nevertheless , these antibody responses were not sustained . In this study , we could demonstrate that the baseline of WHsAg was a poor predictor of the overall therapeutic effect . The WHsAg seroconversion was achieved in two animals: one with of the lowest , and one with the second highest WHsAg baseline values . It can be concluded that by ( 1 ) the addition of a potent antiviral drug ( entecavir ) ( 2 ) using the improved vectors for therapeutic vaccination , ( 3 ) and the prime-boost vaccination protocol , a novel and effective strategy in treatment of chronic hepadnaviral infections was obtained in the preclinical model . These findings may imply to perform the new clinical trials of therapeutic vaccination , using prime-boost regimens with DNA vaccines and recombinant viral vectors ( AdV or modified vaccinia ankara ( MVA ) ) in chronically HBV-infected patients .
All animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the local Animal Care and Use Committee ( Animal Care Center , University of Duisburg-Essen , Essen , Germany and the district government of Düsseldorf , Germany; permission numbers G1109/10 and G1117/10 ) . The experiments were performed under isofluran or ketamine-xylazine anesthesia , and all efforts were made to minimize suffering . WHV transgenic ( Tg ) mice lineages carrying WHV wild-type ( strain 1217 ) and a mutated transgenome lacking the L- , M- and S-WHsAg ( strain 1281 ) were created on C57BL/6 background ( genotype H-2b/b ) and previously characterized [Meng et al . , manuscript attached] . Wild-trapped chronically WHV-infected woodchucks were purchased from North Eastern Wildlife ( Harrison , ID ) . Laboratory animals were maintained according to the guidelines of the animal facility at the University Hospital Essen . The construction of pCGWHc plasmid and recombinant adenoviral vectors serotype 5 ( Ad5WHc ) and chimeric Ad5 displaying Ad35 fiber ( Ad35WHc ) expressing WHcAg was described previously [27] . Briefly , the WHV strain 8 WHcAg gene was obtained from WHcAg-encoding plasmid pWHcIm [47] and introduced between a β-globin intron sequence and polyadenylation signal of pCG vector [48] . The adenoviral vectors expressing WHcAg ( Ad5WHc and Ad35WHc ) were constructed using the AdEasy system and vectors pShuttle , pAdEasy-1 and pAdEasy-1/F35 ( Qbiogene , Carlsbad , CA ) . Generation of WHsAg-expressing pWHsIm plasmid was described earlier [27] , [47] . Ten to twelve weeks old sex-matched groups of mice were pretreated by intramuscular injection of 50 µl of cardiotoxin ( 10 µM in PBS; Latoxan , Valence , France ) into Tibialis anterior muscle one week before the plasmid immunization . Animals were then intramuscularly vaccinated twice with 100 µg of pCGWHc ( 50 µg per muscle ) at two weeks interval . Four weeks after the second DNA immunization groups of mice were immunized with 2×109 PFU of Ad5WHc or 2×109 PFU of Ad35WHc or 100 µg pCGWHc as a reference , according to the protocol described previously [27] . The group of mice immunized twice with pCGWHc in combination with Ad5WHc was boosted for a second time with 2×109 PFU of Ad35WHc . The vaccination was performed four weeks after Ad5WHc immunization . Mice of the control group were immunized twice with 100 µg of ‘empty’ pCG and boosted with 2×109 PFU of Ad5 expressing green fluorescent protein ( GFP; kindly provided by Dr . W . Bayer , Institute of Virology , University Hospital of Essen ) . Mice were sacrificed two weeks after the last immunization . Six chronically WHV-infected woodchucks ( number: 61786 , 61789 , 61791 , 61792 , 61793 and 61795 ) were treated for 23 weeks with the nucleoside analogue entecavir ( ETV , Bristol-Myers Squibb , New York , NY ) . Initially , the drug was administered for 12 weeks at dosage 1 . 4 mg per week by osmotic pumps ( DURECT , Cupertino , CA ) implanted surgically under the skin of the animals . From week 8 to 23 of the therapy subcutaneous injections of 1 mg of ETV were performed twice a week . At week 7 , four of the six ETV-treated animals ( number: 61786 , 61789 , 61792 , and 61793 ) were pretreated by intramuscular injection of 250 µl of cardiotoxin ( 10 µM in PBS; Latoxan ) into Tibialis anterior muscle . Starting from week 8 the animals received subsequently 9 intramuscular immunizations with: DNA plasmids , expressing WHcAg ( pCGWHc ) and WHsAg ( pWHsIm ) at weeks 8 , 10 , 12 , 25 and 27; Ad5WHc+pWHsIm at weeks 14 and 19; Ad35WHc+pWHsIm at week 16 and 22 of the therapy , as shown in Fig . 4 . For the vaccination 0 , 5 mg of plasmids and 1×1011 PFU ( plague forming units ) of Ad5WHc or Ad35WHc was used . Two animals ( number: 61791 , and 61795 ) treated only with ETV served as controls . Murine lymphocytes and woodchuck PBMCs were cultured in RPMI medium ( Invitrogen/Gibco , Karlsruhe , Germany ) and AIM-V medium ( Invitrogen/Gibco ) , respectively . Cell culture media were supplemented with 10% heat-inactivated fetal bovine serum ( FBS; Biochrom AG , Berlin , Germany ) and 10 U/ml penicillin-streptomycin ( PAA Laboratories , Pasching , Austria ) . Cells were maintained in a humidified 5% CO2 atmosphere at 37°C . Preparation of a single-cell suspensions of murine splenocytes was performed according to the procedure described previously [37] . Up to 1×106 of isolated splenocytes per well were plated in 96-well plates in 200 µl of cell culture medium . Hepatic lymphocytes were isolated from the liver using published methods [49] , [50] with some modifications . Briefly , livers were perfused with prewarmed PBS ( to flush blood from the hepatic vasculature ) and were forced through a 70 µm nylon cell strainer ( BD Falcon , Franklin Lakes , NJ ) . After washing , cell pellets were suspended in 5 ml of prewarmed enzyme solution , containing 0 , 05% Collagenase type II ( Sigma ) and 500 U/ml DNAse type I ( Sigma ) in Ca2+/Mg2+-free HBSS supplemented with 10% FBS , and digested for 40 min at 37°C . Cells were then layered on 40% Percoll solution ( Sigma ) in RPMI 1640 supplemented with 10 U/ml penicillin-streptomycin for density separation , and centrifuged at 300× g for 17 minutes at 4°C without brakes . Cell pellets were washed and suspended in 2 ml of Buffer EL ( Qiagen , Hilden , Germany ) to lyse red blood cells . Cell yields and viabilities were determined by trypan blue exclusion microscopy . Murine lymphocytes were stimulated 6 hours or 7 days ( in the presence of 10 U/ml of recombinant murine IL-2; Roche ) with the previously identified WHcAg-derived CD8+ epitope c13-23 ( YQLLNFLPL ) and CD4+ epitope c131-145 ( PYRPPNAPILSTLPE ) [27] , added to a final concentration of 2 µg/ml . Unstimulated cells and cells stimulated with CMV-derived peptide ( YILEETSVM ) served as negative controls . Prior to intracellular cytokine staining , cells were cultured for 5–6 hours in the presence of 1 µg/ml of α-CD28 antibody ( clone 37 . 51; BD Pharmingen , Heidelberg , Germany ) and 5 µg/ml of Brefeldin A ( Sigma-Aldrich ) . Cell surface staining was performed using the anti-CD8 ( clone 56 . 6-7; BD Pharmingen ) and anti-CD4 ( clone L3T4; BD Pharmingen ) T-cell antibodies . Staining of CD107a molecule ( monoclonal anti-mouse CD107a antibody , clone GB12 , dilution 1∶200; BD Pharmingen ) was performed during 5 h restimulation of the splenocytes . Dead cells were excluded from analyses using 7-aminoactinomycin D ( 7AAD ) ( Beckton Dickinson , Heidelberg , Germany ) . Intracellular cytokine stainings were performed as described elsewhere [51] with the following antibodies: anti-IFN-γ ( clone XMG1 . 2; BD Pharmingen ) , anti-TNF-α ( clone MP6-XT22; eBioscience , Hatfield , United Kingdom ) and anti-IL-2 ( clone JES6-5H4 , eBioscience ) . Data were acquired on FACS-Calibur or LSR II flow cytometers ( Becton Dickinson , Heidelberg , Germany ) from 150 000–300 000 lymphocyte-gated events per sample . Analyses were performed using FlowJo software ( Tree Star , Ashland , OR ) . WHcAg-specific CD8+ T-cells were detected using soluble DimerX H-2Db:Ig fusion protein technology ( BD Pharmingen ) according to the manufacturer's instructions . The H-2Db dimer consists of two extracellular major histocompatibility complex class I ( MHC-I ) H-2Db domains that are fused to variable regions of mouse IgG1 . Briefly , 0 , 8 µg of dimer per sample was loaded with 2 , 4 µg of H-2Db-restricted WHcAg-derived CD8+ epitope c13-23 overnight in 37°C . Freshly isolated splenic lymphocytes were pre-treated with anti-CD16/anti-CD32 antibodies ( Fc-Block , clone 2 . 4G2; BD Pharmingen ) diluted 1∶200 for 30 min in 4°C . Next , cells were incubated with anti-CD8+ antibody and 4 µl of peptide-loaded dimer per sample for 1 h in 4°C . As control mouse IgG1 isotype control antibody ( clone MOPC-21/P3; eBioscience ) was used . The detection of dimer- CD8+ T-cell complexes was performed by staining with secondary anti-mouse IgG1 antibody ( clone: A85-1 , BD Pharmingen ) diluted 1∶200 ( 30 min , 4°C ) . Woodchuck PBMCs were separated by Ficoll density gradient centrifugation and cultivated as described previously [37] . For stimulation , the previously identified WHcAg-derived epitope c96-110 ( KVRQSLWFHLSCLTF ) and a WHsAg-derived epitope s220-234 ( AGLQVVYFLWTKILT ) [37] were added to a final concentration of 2 µg/ml per peptide . Unstimulated cells and cells stimulated with CMV-derived peptide ( YILEETSVM ) served as negative controls . After 3 days of in vitro stimulation , cells were re-stimulated and stained for CD107a molecule with anti-mouse CD107a FITC-conjugated antibody ( clone GB12 , dilution 1∶100; BD Pharmingen ) as described previously [37] . For CD4 detection anti-human CD4 allophycocyanin-conjugated antibody ( clone L200; BD Pharmingen ) was used . Dead cells were excluded from analyses using 7AAD . Antigen-specific proliferation of woodchuck PBMCs was determined by 2[3H]adenine-based assay as described previously [47] . Briefly , 5×104 PBMCs were stimulated with a synthetic peptides added to a final concentration of 5 µg/ml for 5 days . For stimulation a panel of 10 WHcAg-derived peptides ( EMC microcollections , Tübingen , Germany ) and 16 WHsAg-derived peptides containing the known woodchuck TH epitopes was used ( see supplementary Table S2 and S3 ) . Unstimulated cells and cells stimulated with CMV-derived peptide ( YILEETSVM ) served as a negative control . Afterwards , cells were labelled with 1 µCi of 2[3H]-adenine ( Hartmann Analytic , Braunschweig , Germany ) for 16 h and collected using a cell harvester ( Perkin Elmer , Waltham , MA ) . Results for triplicate cultures are presented as a mean stimulation index ( SI ) [ ( mean total absorption for stimulated PBMCs ) / ( mean total absorption for unstimulated control ) ] . A SI≥3 . 0 was considered significant . Murine WHcAg-specific IgG , IgG1 and IgG2a as well as woodchuck anti-WHc and anti-WHs antibodies were detected by enzyme-linked immunosorbent assay ( ELISA ) as described previously [47] , [52] . WHV DNA was quantified by real-time PCR using Platinum SYBR Green Kit ( Invitrogen ) as described previously [37] . Total DNA from liver samples of chronically WHV-infected woodchucks was extracted using the QIAamp Tissue Kit ( Qiagen , Hilden , Germany ) according to the manufacturer's instructions . Total amount of 10 µg of isolated DNA was electrophoresed into agarose gel , then transferred onto Amersham Hybond-N+ positively charged nylon membrane ( GE Healthcare , Little Chalfont , United Kingdom ) using Vaccum Blotter 785 ( Bio-Rad Laboratories , München , Germany ) . WHV replication intermediates were analyzed by Southern blot hybridization with a full length WHV8 genome as probe using standard procedure [53] , [54] . Briefly , the radioactive labelling of the probe was performed using DecaLabel DNA Labelling Kit ( Fermentas , St . Leon-Rot , Germany ) , 50 ng of plasmid and [32P]-dCTPs , according to the manufacturer's protocol . The hybridization was performed overnight at 65°C with the probe in RapidHyb Buffer ( GE Healthcare , Little Chalfont , United Kingdom ) . After washing and drying , the membranes were exposed overnight onto the Cyclon's screens ( Packard , Meriden , CT , USA ) . The quantitative analysis of the signals on the blots was performed using a Cyclon Phospho-Imager ( Packard , Meriden , CT , USA ) . Serum WHsAg concentration was determined by electroimmunodiffusion in a similar way as described for HBsAg [32] , [55] . Glass slides ( 3×2 inches ) were coated with 6 ml of 0 . 6% agarose , containing 70 µl of rabbit polyclonal anti-WHs anti-serum per ml . Woodchuck serum samples were diluted 1∶10 in fetal calf serum . The volume of 10 µl of the diluted samples was applied in 3 mm holes , and run for 12 hours at 4 mA per slide . The length of the precipitation arc was converted in µg WHsAg/ml using a calibration curve and highly purified WHsAg from chronic WHV-infected woodchucks as reference antigen . The concentration of purified WHsAg was measured by UV spectrophotometry at 280 nm assuming an OD of 5 . 1 for 1 mg/mL [56] . The coefficient of variation of the assay was approximately 10% . The glutamic oxaloacetic transaminase ( GOT; also known as aspartate transaminase , AST ) activity in the serum was quantified according to standard clinical diagnostic procedures at the Central Laboratory of University Hospital Essen . The values above 50 IU ( international units ) per millilitre were considered as elevated . Statistical analyses were performed using GraphPad Prism version 5 ( GraphPad Software Inc . , San Diego , CA ) . Statistical differences were analyzed by one-way analysis of variance test using Newman-Keuls multiple comparison post-test . The evaluation of the statistical differences between the viral loads in WHV transgenic mice was performed by The Wilcoxon signed rank test . The P-values<0 . 05 were considered significant . | Chronic hepatitis B virus ( HBV ) infection is one of the major causes of liver cirrhosis and liver cancer worldwide . Recommended treatment regimens of chronic hepatitis B based on interferon alpha and nucleot ( s ) ide analogues do not lead to the satisfactory results . Over the last 20 years , continuous efforts have been undertaken to develop new immunotherapeutic approaches for the treatment of chronic hepatitis B , however , without satisfactory results . We proposed here that the combination of potent antivirals with a prime-boost vaccination protocol that is inducing appropriate virus-specific T-cell responses may restore immune control over HBV . To test this hypothesis we performed a proof-of-principle experiment using woodchucks , a widely accepted animal model of chronic HBV infection . We pretreated animals with entecavir to suppress viral replication and immunized them by a prime-boost regimen with DNA vaccines expressing woodchuck hepatitis virus ( WHV ) surface and core antigens and adenoviral vectors expressing WHV core antigen . Consistent with our hypothesis , the combination therapy achieved a stronger antiviral effect than the monotherapy alone , leading to sustained immunological control of chronic WHV infection and viral clearance in some animals . These data are encouraging and implicate the feasibility and usefulness of the immunotherapeutic strategies for the treatment of chronically HBV-infected patients . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"infectious",
"diseases",
"clinical",
"immunology",
"immunology",
"biology",
"microbiology"
] | 2013 | Combination of DNA Prime – Adenovirus Boost Immunization with Entecavir Elicits Sustained Control of Chronic Hepatitis B in the Woodchuck Model |
Changes in gene expression play an important role in evolution , yet the molecular mechanisms underlying regulatory evolution are poorly understood . Here we compare genome-wide binding of the six transcription factors that initiate segmentation along the anterior-posterior axis in embryos of two closely related species: Drosophila melanogaster and Drosophila yakuba . Where we observe binding by a factor in one species , we almost always observe binding by that factor to the orthologous sequence in the other species . Levels of binding , however , vary considerably . The magnitude and direction of the interspecies differences in binding levels of all six factors are strongly correlated , suggesting a role for chromatin or other factor-independent forces in mediating the divergence of transcription factor binding . Nonetheless , factor-specific quantitative variation in binding is common , and we show that it is driven to a large extent by the gain and loss of cognate recognition sequences for the given factor . We find only a weak correlation between binding variation and regulatory function . These data provide the first genome-wide picture of how modest levels of sequence divergence between highly morphologically similar species affect a system of coordinately acting transcription factors during animal development , and highlight the dominant role of quantitative variation in transcription factor binding over short evolutionary distances .
Despite four decades of interest in the evolution of transcriptional regulation , we still have a poor understanding of the molecular bases for regulatory divergence and the constraints under which cis-regulatory sequences evolve . Most regulatory sequences appear to be under strong selection to maintain their transcriptional output , and as a result , binding sites for the sequence-specific transcription factors that regulate mRNA synthesis are preferentially conserved [1] , [2] . However , even in regulatory sequences with highly conserved function , transcription factor binding sites can be gained and lost over time at a high rate , leading to considerable differences in the composition and arrangement of binding sites between even closely related species [2]–[10] . Whether and how this binding site turnover affects transcription factor binding , and what the consequences of changes in binding on transcription might be , remains unknown . After years in which the study of regulatory evolution was primarily a computational exercise , a series of recent studies have compared genome-wide in vivo binding of transcription factors in the same conditions or tissues of related species [11]–[14] . Among yeasts of the genus Saccharomyces [11] , [12] and between human and mouse [13] , [14] , a substantial fraction of experimentally observed interactions between transcription factors and DNA are species-specific . While these differences could , in principle , be due to divergence of transcription factors and other trans-acting factors , binding differences appear to be driven primarily in cis [13] , suggesting that differences in the sequences , and not the factors binding to them , drive the divergence in binding . Species-specific binding is generally associated with the gain/loss of sequence motifs recognized by the relevant factor [11] , [14] , although the correlations are weak . Here we examine how the binding of a group of six factors that direct temporal and spatial patterns of gene expression along the anterior-posterior ( A-P ) axis during early development differs between Drosophila melanogaster and its sister species D . yakuba . These two species , whose genomes have been fully sequenced [15] , [16] , diverged only five million years ago [17] . They are separated by a molecular distance less than half that between mouse and human [18] , and D . yakuba orthologs of virtually all D . melanogaster genomic regions can be readily identified and aligned . Though there are some subtle changes in the levels of expression of key regulators between these species ( our unpublished data ) , there is little difference in either their spatial expression patterns or those of their targets , a product at least in part of strong selection to maintain them [10] . In our earlier work on the binding of these factors in D . melanogaster , we showed that they bind to an overlapping set of thousands of genomic regions in vivo [19] , [20] , as has subsequently been observed for many other animal transcription factors [21] . A wealth of evidence suggests that , at least in D . melanogaster , and probably generally , only the several hundred most highly bound regions are directly involved in transcriptional regulation , with the remainder having a different , or more likely no , function [19] , [20] . Thus these two fly species provide an ideal opportunity to study the effects of modest sequence divergence on transcription factor binding , its origins in changes in genomic sequence , and its functional consequences . We expected binding differences between D . melanogaster and D . yakuba to be more modest than those observed between mouse and human , or between Saccharomyces species . However , we hoped that the more modest differences in their genomes would improve our ability to associate sequence and binding divergence , and that our earlier work establishing the relationship for these factors between binding levels and regulatory function would provide an invaluable context for analyzing the functional consequences of the binding differences we observe .
Unlike in the yeast and mammalian studies described above , the gain or loss of bound regions between D . melanogaster and D . yakuba was rare , with fewer than 1% to 5% of peaks ( depending on the factor ) found in one species clearly absent or displaced in the other ( Table 2 ) . The rate of gain/loss near known targets of the A-P factors was similar to the genome-wide rate ( Table 2 ) . The measured binding at orthologous regions bound in both species varied considerably ( Figures 2 , S5 , and S6 ) both in the highly bound regions that our previous studies suggested are functional targets of these factors [19] , [20] and in the poorly bound regions that likely are not . The more highly bound regions showed a greater total variation in binding ( Figure S7 ) , with the normalized divergence ( difference in binding over average binding level ) roughly constant across binding levels ( Figures 3 and S8 ) and relative to annotations ( Figure S9 ) . The divergence was marginally lower within the 44 characterized D . melanogaster cis-regulatory modules ( CRMs ) known to be targeted by one or more of these factors ( correlation rA-P from 0 . 62 to 0 . 91 compared to 0 . 57 to 0 . 75 ) [27] and in peaks near genes ( within 10 Kb of the 5′ end ) known to be regulated by these A-P factors ( correlation rA-P from 0 . 59 to 0 . 92 , depending on the factor ) . We sought to determine the extent to which sequence changes in the bound regions drove quantitative differences in binding . We first examined overall measures of sequence divergence . Levels of single-nucleotide divergence ( sequence identity ) and frequency of insertions and deletions in the 100 base pairs centered on the inferred peak of binding exhibited only low to moderate correlations with binding divergence ( 0 . 07 to 0 . 24; Figures S10 and S11 ) , consistent with our expectation that changes to specific short sequences , rather than entire regions , would have a disproportionate effect on binding . We next sought to identify short sequences ( e . g . , transcription factor binding sites ) whose gain or loss was associated with changes in binding levels . We devised an unbiased statistical approach that assessed the impact on binding of changes to a short sequence ( word ) by comparing the distribution of binding intensities in all bound regions where the word was conserved to the distribution in all bound regions where the word was present in one species but not the other ( defining bound regions as the 100 bp centered on peaks of maximal binding intensity ) . If alterations to a word affect binding , then these distributions should be different . We identified such words ( which we call divergence-driving words , or DDWs ) by comparing the conserved and non-conserved distributions for all 16 , 384 words of length 7 bp and picking those that showed a statistically significant difference . We found DDWs for four of the six factors , and in each case , virtually all of these DDWs matched the known sequence specificities of the corresponding factor ( Figure 4 ) . To quantify the fraction of binding divergence that is explained by the DDWs , we developed a method that used the gain and loss of DDWs to predict binding divergence between the species . For each factor for which we had identified DDWs , we built a simple linear model relating the divergence of DDWs in a bound region to interspecies difference in binding at that bound region . In the model , each divergent DDW in a bound region contributed a fixed amount to the predicted binding difference , with the effect of multiple divergence DDWs adding independently . The contribution of each DDW was determined by a regression using the least angle regression method [28] with extensive cross-validation ( see Methods ) . The correlations between predicted and observed divergence in binding of single factors across all peaks with at least one DDW in the two genomes ranged from 0 . 3 for HB to 0 . 41 for BCD ( Figures S12–S27 ) . While far from perfect , these correlations demonstrate that changes in a highly restricted collection of sequences ( for example , BCD has only a single 7 bp DDW ) drive an appreciable fraction of binding divergence between species . We additionally performed the same predictions using words derived from the in vitro factor binding specificities described by [29] . The correlations between predictions and observations ranged from 0 . 18 for HB to 0 . 39 in BCD , similar to or lower than the correlations resulting from our DDWs ( unpublished data ) . We investigated whether the lack of a strong relationship between probable enhancer function and quantitative conservation of binding was associated with similar trends at the sequence level . For each factor for which we identified DDWs , we quantified motif enrichment and conservation as a function of the level of transcription factor occupancy in D . melanogaster . Motif enrichment and conservation were elevated within bound regions above background levels across the genome ( Figure 5 ) . The fraction of peaks with motifs showed a weak dependence on binding levels , with the most strongly bound regions exhibiting the greatest density of motifs . The level of conservation of these motifs was weakly correlated with overall binding levels , consistent with our observation that quantitative divergence in binding strength decreased slightly near genes regulated by these factors . In our initial comparison of binding between species , we noticed that increases in binding of a single factor were often correlated with increases in binding of many other factors ( Figures S28–S33 ) . For example , changes in the binding of KR correlated with changes in the binding of other factors with r = 0 . 36 ( KNI ) to 0 . 62 ( CAD ) , and such coordinated changes are recapitulated for all pairs of factors . This widespread correlated change suggests a factor-independent mode of binding divergence . To obtain an unbiased assessment of the extent of these correlated changes in binding , we quantified binding divergence for all six factors in all regions significantly bound by any factor and performed principal component analysis ( PCA ) , a method for analyzing variation between many factors simultaneously rather than only pairs of factors , on these data ( Figure 6A ) . The first principal component , which represents the most significant axis of variation in the dataset , has the same direction and similar magnitude for all six factors , demonstrating that a pan-factor coordinated binding shift is the dominant driver of A-P factor binding divergence ( this principle component explains 38% of the overall variation in binding between the species ) . A similar effect was observed when we performed PCA on the binding levels in each species independently ( Figure 6B and 6C ) , suggesting that a common effect is responsible for much of the variation in binding both between species and within a single genome . The single-genome PCA revealed several interesting factor-specific correlations: increases in binding of the repressor GT are associated with decreases in binding of the activator HB ( PC2 in Figure 6B ) , increases in HB are associated with decreases in BCD ( PC3 in Figure 6B ) , etc . As expected , given the overall similarity of binding between the species , the single-genome PCA analyses of D . melanogaster and D . yakuba yielded essentially identical results . To investigate whether the features captured by these different principal components are related to specific sequences , we applied the same motif discovery method described above to projections of the binding data along each of the principal components shown in Figure 6A . We discovered substantially more motifs in this analysis ( Figure 7 ) than in the single-factor analyses , likely because of the increased statistical power derived from considering all regions bound by any , as opposed to a single , factor . Interestingly , one of the words whose divergence is associated with the first principal component is the “TAGteam” motif , CAGGTAG [30] , the binding site for Zelda , an activator of the early zygotic genome [31] . Zelda's mechanism of action is unknown , but the strong correlation between gain and loss of its binding site with variation in changes in binding of all factors supports a direct or indirect role for Zelda in nucleosome positioning and chromatin remodeling .
Although D . melanogaster and D . yakuba are closely related , we were not always able to accurately identify orthologous sequences , largely due to ambiguities in the draft D . yakuba assembly . Even where the orthology of regions was unambiguous , and despite this close evolutionary distance , base-level alignments were frequently uncertain . Our analysis of sequence-specific effects required a precise alignment , and inevitable alignment errors will make nucleotide-level analysis of regulatory changes challenging for more distantly related species ( although the alignment accuracy estimates produced by FSA may help to identify reliably aligned loci ) . Several aspects of this experiment should help direct future efforts to use comparative ChIP-Seq to study the relationship between sequence and binding divergence . The widespread quantitative binding divergence between D . melanogaster and D . yakuba demonstrates that even relatively similar species can be used to study binding changes . Indeed , given the magnitude of the binding divergence that we observe , we expect there to be quantitative differences between D . melanogaster and more closely related species , such as D . simulans , as well as among D . melanogaster individuals . While comparisons with more distantly related species will likely reveal greater binding divergence , and will help explain how such divergence affects expression and phenotype , the difficulties with aligning genomes at this distance , and comparing embryonic stages , may render sequence-based analyses less powerful . Even though we were working with very similar organisms , with similar timing and structure of embryonic development , there were undoubtedly subtle differences in our sampling of developmental stages in the two species . Because transcription factor binding is dynamic , such sampling differences have the potential to manifest themselves as apparent interspecies differences in binding . We do not believe this effect was significant in our data , however , as it is unlikely that this type of false-positive binding divergence would be associated with the specific sequence changes that we repeatedly observed . Nonetheless , this will be a major difficulty in future studies , especially when developmentally and morphologically different organisms are compared , as precisely those changes that make such comparisons interesting also make them far more difficult .
Both D . melanogaster and D . yakuba embryos were collected from population cages for 1 h , and then allowed to develop to late stage 4 and early stage 5 before being harvested and fixed with formaldehyde . The embryos from the two species developed very similarly , and the aging times to reach the desired age were 2 h for D . melanogaster embryos and 1 h and 45 min for D . yakuba embryos . The staged embryos were harvested and cross-linked with formaldehyde , and the chromatin was isolated through CsCl gradient ultracentrifugation essentially as previously described [19] . The chromatin used for immunoprecipitation was fragmented through sonication using a Branson Sonifier 450 to an average fragment size of 225 to 250 bp , which is shorter than the average size of chromatin used in our previous ChIP-chip experiments [19] . ChIP was carried out using affinity purified rabbit polyclonal antibodies , and for two of the factors , HB and KR , two affinity purified antibodies that recognize non-overlapping parts of each factor were used . These antibodies and the ChIP procedure were identical to those described in [19] . The DNA libraries for sequencing were prepared from the ChIP reaction and from Input DNA following the Illumina protocol for preparing samples for ChIP sequencing of DNA using the reagents provided in the genomic-DNA or ChIP-DNA sample preparation kits , with some modifications . Briefly , the DNA fragments were converted to phosphorylated blunt ends using T4 DNA polymerase , Klenow DNA polymerase , and T4 polymerase kinase , a 3′ A base overhang was added using Klenow DNA polymerase exo- ( 3′ to 5′ exo minus ) , and Illumina adapters were ligated to the fragments . We carried out the PCR step for enrichment of adapter-modified DNA prior to the library size selection , and limited the amplification to 10–13 cycles to minimize the potential bias associated with PCR amplification . After the amplification step , we size-selected DNA fragments of 150–500 bp ( including the adapter sequence ) for BCD , HB , GT , and KNI samples , and 200–500 bp for KR and CAD . The DNA library was quantified by QPCR using ABI Power SYBR green PCR master mix and pair primers that match the adapter sequences . We used a Solexa DNA library , which we generated with known concentration as a standard . Due to the extreme sensitivity , the DNA used in the reactions ranged from 0 . 0001–0 . 01 ng . The sequencing of the library DNA was performed on the Solexa/Illumina platform according to the manufacturer's instruction . Each library was analyzed in two lanes on the flow cell . We used the Apr . 2006 assembly ( dm3 , BDGP Release 5 ) of the D . melanogaster genome , downloaded from http://hgdownload . cse . ucsc . edu/goldenPath/dm3/bigZips/chromFa . tar . gz , and the Nov . 2005 assembly ( droYak2 ) of the D . yakuba genome , downloaded from http://hgdownload . cse . ucsc . edu/goldenPath/droYak2/bigZips/chromFa . tar . gz . We trimmed all sequenced tags to 20 bp and mapped the tags to the genomes using Bowtie v0 . 9 . 9 . 1 [22] with command-line options ‘-v 1 -m 1’ , thereby keeping only tags that mapped uniquely to the genome with at most one mismatch . Table 1 gives statistics on the total numbers of sequenced and mapped tags for all experiments . Note that while we mapped tags to the entire genomes , we did not use the heterochromatic chromosomes or unassembled sequence for any analyses . We used annotations from FlyBase r5 . 15 [34] for analyses using genes in D . melanogaster . We called peaks for each experiment using MACS v1 . 3 . 5 [25] with the option ‘--pvalue 0 . 00001’ . We used total chromatin as background controls , and set the ‘--mfold’ option to the maximum value for which MACS could find a sufficient number of paired peaks . In order to only consider peaks for which we could reliably assign orthology and to control for potential assembly errors in the draft D . yakuba genome , we used exonerate [35] to search for peaks whose associated sequence was duplicated in either genome . For each peak , we ( 1 ) searched for duplicated sequence in the genome where the peak was called and ( 2 ) used the whole-genome alignment to pull out the orthologous sequence in the other genome and searched for duplicates of that sequence in the other genome , which frequently indicated a potential assembly error due to the unfinished nature of the D . yakuba assembly . We discarded any peaks whose associated sequence was duplicated in either genome . We used a large-scale orthology mapping created by Mercator [23] to identify syntenic regions of the genomes , which were each aligned with FSA v1 . 11 . 0 with the options ‘--exonerate --softmasked --refinement -1 --mercator cons seqs . fasta’ . The resulting whole-genome alignment can be downloaded here: http://www . biostat . wisc . edu/~cdewey/data/fsa_mercator_alignments/drosophila_melanogaster-5 . 0-drosophila_yakuba-2 . 0-1 . 0 . tar . gz . We first normalized the total number of sequenced tags to a fixed number for each experiment , the standard method of controlling for the variable success of amplification and sequencing . This normalization , however , is insufficient for our purposes , since it does not take into account differences in genome size and background between the species . We therefore performed an additional comparative normalization step . Assuming that the total amount of binding near known regulatory targets of the six factors studied here ( A-P and D-V genes , as identified in [19] and listed below ) is constant , we scaled the total number of sequenced tags in D . yakuba for each factor such that the total difference in inferred binding strength across the 50 most highly bound peaks in each genome ( for a total of 100 ) within 10 kb of A-P targets was minimized ( using a least-squares linear regression ) . This comparative normalization procedure assumes there are no differences in the total number of molecules bound to A-P targets in the two genomes . Although this may not always be the case , we do not expect to see such global differences between such closely related species . It is also possible that by using the 50 most highly bound peaks near known A-P target genes for normalization we would underestimate variation in these genes . However , the effect of any single peak on the normalization was minimal , and the inferred divergence for any of these peaks did not change significantly when they were not included in the normalization ( unpublished data ) . We assessed binding strength by estimating a fragment density by extending each sequenced tag to the average fragment length based on the selected size distribution . We modified the SynPlot program [36] to display quantitative data along an alignment in order to create the plot in Figure 1 . We compared binding between the two genomes as follows: Given a peak called in one genome , we used the whole-genome alignment to project the 100 bp containing the peak onto the other genome and computed the maximum binding strength within that homologous sequence in the other genome . Note that therefore our maximum spatial resolution when assessing binding divergence is 50 bp , implying that if , for example , a binding site is present in D . melanogaster , and lost in D . yakuba but replaced by another site 30 bp away , then we will not detect any binding divergence if the two sites are bound at similar levels . We labeled peaks that were within 10 Kb of a gene in D . melanogaster known to be regulated by A-P factors as A-P target loci . We used the following list of genes: Brk , D , Doc1 , Doc2 , E ( spl ) , Kr , Phm , SoxN , Vnd , bowl , btd , cad , croc , dpp , ems , eve , fkh , ftz , gt , h , hb , hkb , ind , kni , knil , noc , nub , oc , odd , opa , os , pdm2 , pnr , prd , pxb , rho , run , salm , shn , sim , slp1 , slp2 , sna , sob , sog , ths , tld , tll , tsh , tup , twi , vn , wntD , zen . We identified DDWs for each factor as follows . For each word of a fixed length k , we identified all ( non-softmasked ) instances of the word ( on both strands ) within a 100 bp window centered on the empirical maximum of peaks called in D . melanogaster for that factor . We then accumulated two distributions of binding strength divergence ( D . melanogaster − D . yakuba ) for the word , pcons and pdiv , with pcons consisting of instances where the word was exactly conserved in D . yakuba and pdiv consisting of instances where the word was diverged in D . yakuba . We used a non-parametric statistical test , Kolmogorov-Smirnov test , to test for equality of distribution pcons ∼ pdiv . If equality of distribution could be rejected with p<0 . 01 , then we called the word a candidate DDW . We then performed the identical procedure in the opposite direction , wherein we examined peaks called in D . yakuba and assessed the conservation of words in D . melanogaster , and identified a second set of candidate DDWs . We took the intersection of these two sets to obtain final lists of DDWs . We performed this procedure to identify words of length k = 6 and 7 . We assessed whether sequence motifs matched the known DNA-binding specificities of A-P factors with position weight matrices ( PWM ) from [29] . When creating Figures 4 and 7 , we said that a word matched the specificity for a factor if it matched a subsequence of the corresponding PWM with ln ( p value ) <−4 as reported by Patser [37] . We used the Least Angle Regression ( LARS ) algorithm [28] , implemented in the package lars for R [38] , to learn a linear model of binding divergence using DDWs of length k = 6 . We performed 5-fold cross-validation to estimate the mean-squared prediction error ( MSE ) associated with each value of the lasso regularization parameter β and then chose the model given by the β that yielded the lowest MSE . This cross-validation procedure helps to prevent the over-fitting characteristic of standard least-squares linear regression , making the correlations that we estimated robust to generalization error . In order to ensure that ( 1 ) the DDWs that we identified truly have predictive value and ( 2 ) that the correlations reported are not due solely to base-composition effects , we randomly shuffled the nucleotides of each DDW to create a set of shuffled words with unchanged base composition , and then built a predictive model using these shuffled words . Models constructed using these shuffled words had no predictive value , indicating that the correlations that we report for our DDWs are not statistical artifacts . Figures S12–S27 show lasso variable selection curves and cross-validation curves for all values of β , as well as scatterplots of predicted and observed binding divergences , for predictive models constructed using our DDWs as well as their shuffled counterparts . The cross-validation curves make clear that while the DDWs are correlated with binding strength , the shuffled words are not: MSE decreases as more DDWs are included into the model , indicating the gain and loss of these words correlates with changes in observed binding strength , whereas MSE increases as more shuffled words are included into the model , indicating that these words are uncorrelated with binding . This provides clear evidence that our cross-validation procedure correctly chooses the model with the minimum generalization error , for example , that the models are not over-fit to the data . We performed an identical analysis using words derived from the in vitro binding specificity data described in [29] . We enumerated all k-mers that matched a subsequence of the corresponding PWM with ln ( p value ) <−8 as reported by Patser [37] , identifying four 6-mers for BCD , HB , and GT and sixteen 6-mers for KR , and then used the learning procedure described above to learn models of binding divergence using these words . We calculated binding strengths of the six factors across all called peaks , subtracted the empirical means for each factor , and scaled the data for each factor such that it had unit variance . We used the singular value decomposition routine in IT++ , a linear algebra library for C++ , to perform PCA , and created heatmaps of the PCA results using a modified version of the aspectHeatmap function in the ClassDiscovery package . In order to confirm that the putative chromatin signal represented by the first principal component did reflect coherent increases and decreases in binding of all six factors in our data , we randomly interchanged the measured binding strengths for a single factor across called peaks while holding all others unchanged ( Figure S34 , panels A–F ) and similarly randomly interchanged the binding strengths of all factors ( Figure S34 , panel G ) , thereby removing spatial correlations between the binding of single factors and the other five ( Figure S34 , panels A–F ) and removing spatial correlations between the binding of any factors ( Figure S34 , panel G ) . As expected , the chromatin signal disappeared after performing any of these transformations on the data . We identified sequence motifs associated with interspecies divergence of each principal component using the same procedure described above , but with the data projected along the principal component of interest . For each principal component , we accumulated the distributions pcons and pdiv across all peaks called for any of the six factors . All sequence reads from the experiments described are available from the NCBI's GEO database with accession number GSE20369 . Processed datasets , including mapped reads , called regions and peaks , D . melanogaster − D . yakuba alignments , and all software described here , are available at http://rana . lbl . gov/data/melyak . | The differentiation of cells , tissues , and organs during animal development is established by a process in which genes that control cell identity and behavior are turned on and off at specific times and places . This process is choreographed , to a large extent , by a collection of proteins known as transcription factors that bind to specific sequences in DNA and thereby modulate the expression of neighboring genes . Because of the central role that transcription factors play in shaping organismal form and function , they have long been suggested to be major players in phenotypic evolution . However , we have a poor understanding of how changes to DNA affect transcription factor binding in living systems . Here , we use a combination of biochemical and genomic techniques to compare , between two closely related species of fruit flies in the genus Drosophila , the binding of six transcription factors that help establish the characteristic segments that form along the anterior-posterior ( head to tail ) axis in developing flies . We show that the patterns of transcription factor binding between these closely related species are broadly conserved , consistent with the nearly identical development and appearance of these species . However , we also show that , whereas the DNA changes that have accumulated between these species in the five million years since their divergence—roughly one difference per 10 basepairs—have not altered the locations where these factors bind , they have had a considerable effect on the amount of factor bound at each site across a population of embryos . We can trace these quantitative differences in binding to the gain and loss of the short sequences known to be preferentially recognized by these factors , giving us key insights into the effect that sequence changes have on the biochemical events that underlie animal development . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"genetics",
"and",
"genomics/comparative",
"genomics",
"genetics",
"and",
"genomics/functional",
"genomics",
"computational",
"biology/comparative",
"sequence",
"analysis",
"developmental",
"biology/developmental",
"evolution",
"computational",
"biology/evolutionary",
"modeling",
... | 2010 | Binding Site Turnover Produces Pervasive Quantitative Changes in Transcription Factor Binding between Closely Related Drosophila Species |
Staphylococcus aureus is a devastating mammalian pathogen for which the development of new therapeutic approaches is urgently needed due to the prevalence of antibiotic resistance . During infection pathogens must overcome the dual threats of host-imposed manganese starvation , termed nutritional immunity , and the oxidative burst of immune cells . These defenses function synergistically , as host-imposed manganese starvation reduces activity of the manganese-dependent enzyme superoxide dismutase ( SOD ) . S . aureus expresses two SODs , denoted SodA and SodM . While all staphylococci possess SodA , SodM is unique to S . aureus , but the advantage that S . aureus gains by expressing two apparently manganese-dependent SODs is unknown . Surprisingly , loss of both SODs renders S . aureus more sensitive to host-imposed manganese starvation , suggesting a role for these proteins in overcoming nutritional immunity . In this study , we have elucidated the respective contributions of SodA and SodM to resisting oxidative stress and nutritional immunity . These analyses revealed that SodA is important for resisting oxidative stress and for disease development when manganese is abundant , while SodM is important under manganese-deplete conditions . In vitro analysis demonstrated that SodA is strictly manganese-dependent whereas SodM is in fact cambialistic , possessing equal enzymatic activity when loaded with manganese or iron . Cumulatively , these studies provide a mechanistic rationale for the acquisition of a second superoxide dismutase by S . aureus and demonstrate an important contribution of cambialistic SODs to bacterial pathogenesis . Furthermore , they also suggest a new mechanism for resisting manganese starvation , namely populating manganese-utilizing enzymes with iron .
The spread of antibiotic resistance amongst bacteria has led both the Centers for Disease Control and Prevention and the World Health Organization to state that infections represent a serious threat to human health [1 , 2] . This threat is exemplified by Staphylococcus aureus , a Gram-positive bacterium that asymptomatically colonizes one third of the population and is a leading cause of antibiotic-resistant infections [3–5] . A promising area of investigation is elucidating how pathogens overcome host defenses such as the active withholding of essential nutrients and the oxidative burst of immune cells . During infection , pathogens must obtain all of their nutrients from the host , including the essential metal ions that are needed for the approximately one-third of all bacterial proteins that require a metal cofactor [6–8] . This requirement is exploited by the host , which restricts the availability of these essential nutrients , a defense termed nutritional immunity [9–13] . The canonical example of nutritional immunity is the iron ( Fe ) -withholding response [10 , 11] . In addition to Fe , the host also restricts the availability of manganese ( Mn ) and zinc ( Zn ) [9 , 12–14] . The prototypical example of Mn and Zn restriction is the staphylococcal abscess , which is rendered virtually free of these metals during infection [9 , 14] . A critical component of the Mn- and Zn-withholding response is the host protein calprotectin ( CP ) [9 , 12 , 14] . This innate immune effector is highly expressed in neutrophils in which it comprises 40–60% of the cytoplasmic protein , and at sites of infection it can be found in excess of 1 mg/ml [15 , 16] . Loss of CP results in host defects in metal sequestration and increased sensitivity to a number of bacterial and fungal pathogens , including S . aureus [9 , 14 , 17–19] . In culture , CP inhibits the growth of a similarly wide range of pathogens [16–20] . The antimicrobial activity of CP is dependent on binding of metal ions to its two transition metal-binding sites [14 , 20 , 21] . The first site or ‘Mn/Zn site’ is comprised of six histidines and is capable of binding either Mn or Zn with nanomolar and picomolar affinities ( Kd ) , respectively [14 , 20–22] . The second site or ‘Zn site’ is comprised of three histidines and an aspartic acid and binds Zn with picomolar or sub-picomolar affinity [14 , 20 , 22] . In order to cause disease , invading pathogens must not only overcome nutrient starvation but also simultaneously cope with other host defenses , such as the oxidative burst of neutrophils and other immune cells [23] . Bacteria defend themselves from the oxidative burst by numerous mechanisms , including enzymes such as superoxide dismutases ( SODs ) that detoxify the damaging reactive oxygen species with which they are bombarded [24–27] . The activating metal cofactor divides the SOD enzymes into several families , with the most common amongst bacteria belonging to a single protein superfamily , which utilizes either Mn or Fe as cofactor . It has proven exceptionally difficult to predict which metal is utilized by a given Mn/Fe-dependent SOD [24 , 28] , in part due to the fact that Fe- and Mn-SODs coordinate their metal cofactor using the same protein ligands within an identical protein fold [28 , 29] . This structural similarity also enables both Fe- and Mn-SODs to bind the other , non-cognate metal , but this usually results in an inactive enzyme; most members of this protein family are strictly dependent on their cognate metal for catalysis [28–31] . Notably , a subset of the Mn/Fe-dependent SOD family is active when loaded with either Fe or Mn [32–40] . While these ‘cambialistic’ SODs are present in a diverse group of microbes , analysis of their activity has largely been limited to in vitro studies , limiting our understanding of how cambialism benefits microbes and the contribution of these enzymes to colonization of the host . S . aureus possesses two SODs , SodA and SodM , both of which are cytoplasmic and are reported to be Mn-dependent [41–43] . While all staphylococci possess SodA , SodM is unique to S . aureus [44] . Highlighting their importance to virulence , loss of either SodA or SodM in a skin model of infection or loss of both SODs in a systemic mouse model , reduces the ability of S . aureus to cause disease [14 , 42] . However , a molecular explanation for the advantage that S . aureus gains by expressing two apparently Mn-dependent SODs is unknown . Host-imposed Mn starvation mediated by CP reduces total staphylococcal SOD activity , both in culture and during infection , which renders S . aureus more sensitive to oxidative stress and neutrophil-mediated killing [13 , 14 , 20] . Yet paradoxically , the simultaneous loss of both SodA and SodM renders S . aureus more sensitive to CP [14] , indicating that SodA and/or SodM somehow enhance the ability of S . aureus to resist metal starvation . Given the importance of the two staphylococcal SODs to infection , we have elucidated their respective contributions to resisting oxidative stress and nutritional immunity . This analysis revealed that SodA is important for resisting oxidative stress and infection when Mn is abundant , whereas SodM is important under Mn-deplete conditions . Our data demonstrate that SodM is in fact cambialistic , possessing equal enzymatic activity when loaded with Mn or Fe . We propose that the ability of SodM to utilize Fe enables S . aureus to retain SOD activity when starved of Mn by the host , thereby enhancing the ability of the bacterium to overcome nutritional immunity , resist oxidative stress , and ultimately cause infection .
Several prior studies have examined the impact of Mn availability on the expression of SodA and SodM; however , different conclusions were reached [41 , 42 , 45] . In light of this ambiguity , we initially assessed the impact that CP and oxidative stress have on sodA and sodM expression . When normalized to optical density in order to account for differences in growth , high levels of sodA transcription were observed regardless of whether CP was present ( Fig 1A ) , whereas the presence of CP enhanced sodM expression independent of the presence of oxidative stress ( Fig 1B ) . These results indicate that sodM , but not sodA , is induced in response to Mn or Zn limitation . To clarify which metal gave rise to this effect , we used mutant CP variants that lack either the Mn/Zn site ( ΔMn/Zn site mutant , which does not bind Mn ) or the Zn site ( ΔZn site mutant , which binds both Mn and Zn ) [20] . As expected , neither mutant induced the expression of sodA ( Fig 1C ) . The increased expression of sodM observed with wild type ( WT ) CP is lost in the presence of the ΔMn/Zn site mutant , but not the ΔZn site mutant , indicating that sodM is induced in response to Mn limitation ( Fig 1D ) . We also evaluated the impact of oxidative stress on the expression of the SODs using the superoxide-generating compound paraquat ( PQ ) . Similar to previous studies [42] , PQ induced the expression of sodA in metal-replete media ( Fig 1A ) . However , PQ did not alter the induction of expression of sodM observed with CP ( Fig 1B ) , nor did it change the sodA and sodM expression pattern observed with the ΔMn/Zn or ΔZn site mutants ( Fig 1E & 1F ) . Cumulatively , these observations suggest that the expression of SodM increases when S . aureus is Mn-limited regardless of level of oxidative stress experienced by S . aureus . The propensity of Mn/Fe-SODs to acquire the wrong metal can result in discordance between expression levels and enzymatic activity [46] . In order to determine if SodA and SodM activity correlated with expression , total and individual SOD activity were assessed in the presence and absence of CP and PQ . Consistent with prior results , CP significantly reduced total staphylococcal SOD activity [14] ( S1A Fig ) . In the absence of CP , the predominant activity comes from SodA ( Fig 2A & 2B ) . The presence of the SodA/SodM heterodimer [41 , 43] indicates that SodM is present in Mn-replete media , although the SodM homodimer’s activity is barely detectable in the gel . In the presence of CP , the relative contribution of SodA to total SOD activity decreased while that of SodM increased ( Fig 2A & 2B ) . Notably , not only did the fractional contribution of SodM change , the absolute level of SodM activity also increased ( Figs 2A & S1C ) . Consistent with prior studies , the addition of PQ increased total staphylococcal SOD activity [14] ( S1A Fig ) . However , the addition of PQ did not change the impact that Mn availability had on the relative contributions of SodA and SodM to total staphylococcal SOD activity ( Figs 2B , S1B & S1C ) . Together , these results indicate that in Mn-replete environments SodA is the primary source of SOD activity but SodM becomes the predominant SOD when S . aureus experiences Mn starvation . To test the respective contribution of each SOD to resisting Mn starvation , the ability of ΔsodA and ΔsodM single mutants , as well as a ΔsodAΔsodM double mutant , to grow in the presence of CP was assessed . Similar to previous results [14 , 42] , ΔsodAΔsodM was profoundly more sensitive to CP and PQ than wild type ( Fig 3A & 3B ) , while ectopic expression of either SodA or SodM reversed this sensitivity ( S2 Fig ) . In the absence of PQ , ΔsodA grew as well as wild type S . aureus in both the presence and absence of CP ( Fig 3A ) , whereas the ΔsodM mutant , although it did not reach significance , displayed consistent reduced growth relative to wild type at high levels of CP ( Fig 3A ) . Given the role of SodA and SodM in detoxifying superoxide , we also evaluated the impact of Mn availability on the ability of ΔsodA and ΔsodM to resist oxidative stress . Consistent with the activity analysis and its reported role as the primary SOD expressed by S . aureus [41 , 42] , loss of SodA resulted in increased sensitivity to PQ in the absence of CP ( Fig 3B ) . However , at high concentrations of CP ΔsodA was no more sensitive to PQ than wild type S . aureus ( Fig 3B ) . When compared to ΔsodA , the impact that CP had on the sensitivity of ΔsodM to oxidative stress was reversed; the ΔsodM mutant was no more sensitive to PQ than wild type bacteria in the absence and presence of low concentrations of CP , but the mutant was significantly more sensitive at high concentrations ( Fig 3B ) . Notably , in the presence of intermediate CP concentrations in which both SodA and SodM are active , neither ΔsodA nor ΔsodM is more sensitive than WT S . aureus to oxidative stress . Utilization of the CP metal-binding site mutants revealed that both in the presence and absence of PQ , the increased sensitivity of ΔsodM is due to Mn sequestration ( Figs 3C , 3D , S2C & S2D ) . Cumulatively , these results indicate that in Mn-replete environments SodA is primarily responsible for protecting S . aureus from oxidative stress , whereas SodM is critical for protecting S . aureus from oxidative stress in Mn-deplete environments . They also suggest that SodM promotes resistance to nutritional immunity by facilitating the retention of SOD activity and resistance to oxidative stress . Paradoxically , our results indicate that the reportedly Mn-dependent enzyme SodM promotes resistance to host-imposed Mn starvation . In light of these observations , we analyzed the metal specificities of recombinant SodA and SodM . To facilitate these studies , the SODs were expressed in and purified from E . coli grown in iron-replete media . Following expression in E . coli and consistent with negligible Mn accumulation by E . coli in the absence of oxidative stress [46] , inductively coupled plasma mass spectrometry ( ICP-MS ) analysis revealed that both of the purified recombinant staphylococcal SODs were loaded with Fe when recovered from the heterologous host ( S3 Fig ) . Substantial activity was observed with purified Fe-SodM ( 210 +/- 21 U/mg protein ) , but negligible activity was detected from Fe-SodA ( 4 +/- 1 U/mg ) . Each of the recombinant proteins were denatured in the presence of metal chelators and then refolded in the presence of Mn in vitro , with successful elimination of Fe and loading with Mn confirmed by ICP-MS ( S3 Fig ) . Enzymatic analysis revealed that the Mn-SodA form has substantial activity ( 1594 +/- 81 U/mg ) in contrast to the Fe form . Surprisingly , Mn-SodM was also active ( 215 +/- 21 U/mg ) and to a degree similar to that of the Fe-SodM , although both forms display activity substantially lower than that of Mn-SodA . The comparable activity of the Mn- and Fe-loaded forms of SodM indicate that it is not Mn-dependent , as previously suggested , but cambialistic [43] . The cambialistic properties of SodM raise the possibility that in Mn-deficient conditions , including those induced by the presence of CP , Fe-loaded SodM predominates in the cell . We took advantage of the fact that Fe-dependent SODs can be selectively inactivated by hydrogen peroxide to evaluate if both the Mn- and Fe-loaded forms of SodM are present in S . aureus [31] . For these experiments , SodM was expressed from a plasmid in ΔsodAΔsodM and SOD activity was assessed following growth in Fe- and Mn-replete media . Control experiments using purified protein confirmed that the Fe-loaded form of SodM , but not the Mn-loaded forms of SodA or SodM , is sensitive to peroxide poisoning ( Fig 2A ) . Consistent with prior studies [43] , following growth in Mn-replete media SodM activity was not affected by peroxide indicating that the protein is loaded with Mn . However , when grown in Fe-replete media SodM activity was sensitive to hydrogen peroxide , indicating that it was Fe-loaded ( Fig 4 ) . These results indicate that in S . aureus SodM can be active with either Mn or Fe . CP has been observed to bind Fe2+ , although it is unclear if this binding contributes to antimicrobial activity [13 , 47] . As such , it raises the possibility that when exposed to CP S . aureus may be incapable of populating SodM with Fe . To evaluate the metallation state of SodM , lysates from cells cultured in the presence of CP were treated with H2O2 and assayed for SOD activity . In the absence of CP there was no reduction in activity associated with the heterodimer following peroxide treatment indicating that SodM is loaded with Mn . However , in the presence of CP , peroxide treatment eliminated almost all SodM activity , indicating that it is loaded with Fe ( Fig 2A ) . Oxidative stress did not change the metal that was associated with SodM in both the presence and absence of CP ( S1B Fig ) . Cumulatively , these observations suggest that the cambialistic nature of SodM enables S . aureus to resist host-imposed Mn starvation by facilitating the retention of SOD activity . In order to evaluate if SodA or SodM differentially contribute to pathogenesis based on Mn abundance during infection , we took advantage of the difference in Mn availability in wild type and CP-deficient mice [9] . Initially , the respective contributions of SodA and SodM to systemic disease in wild type C57BL/6 mice , in which the staphylococcal abscess is devoid of Mn , was assessed . In wild type mice , infection with ΔsodA resulted in a modest , but not significant , reduction in bacterial burden when compared to wild type S . aureus . In contrast , in wild type mice infected with ΔsodM there was a significant reduction in bacterial burden relative to wild type S . aureus ( Fig 5 ) , indicating that in the absence of Mn , SodM is critical for staphylococcal infection . Next , we infected CP-deficient mice , which fail to sequester Mn from staphylococcal liver abscesses [9] . Consistent with prior results , higher bacterial burdens were recovered from CP-deficient mice ( C57BL/6 S100A9-/- ) than wild type C57BL/6 mice infected with wild type S . aureus [9] ( Fig 5 ) . CP-deficient mice infected with ΔsodA had significantly reduced bacterial burdens when compared to wild type S . aureus . This result contrasts with CP-deficient mice infected with ΔsodM , which had bacterial burdens comparable to that of those infected with wild type S . aureus . In total , these observations indicate that SodA , but not SodM , contributes to staphylococcal disease when Mn is abundant . They also support the hypothesis that SodM contributes to the ability of S . aureus to resist host-imposed Mn starvation during infection .
During infection the innate immune system combats invading microbes by restricting the availability of the essential nutrient Mn [9 , 12 , 14] . At the same time , S . aureus and other pathogens must also overcome other host defenses including the oxidative burst of immune cells [23] . Accomplishing this latter task is made more challenging , as host-imposed Mn starvation inactivates bacterial Mn-dependent SODs [14] . The current investigations revealed that the possession of a cambialistic SOD enables S . aureus to counter these dual host threats both in culture and during infection . This strategy represents an entirely new mechanism for resisting host-imposed Mn starvation and establishes that cambialistic SODs contribute to bacterial pathogenesis . The Fe/Mn superfamily of SODs is widely distributed in bacteria , archaea , and eukaryotes . Members of this family are generally thought to be reliant on either Mn or Fe for catalytic activity [28 , 30 , 48] . However , since the 1980s , predominantly in vitro analyses have suggested that a subset of these enzymes , termed cambialistic SODs , are capable of using both Fe and Mn , [32–34 , 38–40] . Cambialistic SODs have been reported in both Gram-positive and Gram-negative bacteria , including the human pathogens Porphyromonas gingivalis , Streptococcus pneumoniae , and Streptococcus mutans and suggested to be present in other microbes including Bacteroides fragilis , and Bacteroides thetaiotaomicron [32–40 , 49–52] . However , the lack of detailed in vivo studies and the fact that many cambialistic SODs have greater activity when loaded with one or the other cofactor in vitro has resulted in skepticism regarding the importance of cambialism [32 , 36 , 38–40] . As such , a false dichotomy exists that members of the Fe/Mn SOD family must use either Mn or Fe but not both . This dichotomy has led to confusion over the biologically relevant metal utilized by several bacterial SODs , especially given the difficulty in predicting the cofactor utilized by this family of enzymes using bioinformatics [28] . The observation that SodM has equal activity with either Mn or Fe in vitro , can be activated with both metals in vivo , and promotes resistance to nutritional immunity during infection establishes a clear and important role for cambialistic SODs in facilitating resistance to host defenses . Given the ubiquity of CP and host-imposed metal starvation during infection , it seems likely that expression of a cambialistic SOD would provide a benefit to other pathogens as well . Cambialistic SODs are also found in a diverse collection of environmental microbes [32 , 33 , 35 , 39] , suggesting cambialism may represent a generalized strategy used by organisms to maintain a defense against superoxide in niches where Fe and Mn availability can fluctuate . While metal-dependent mononuclear enzymes have historically been thought to utilize a specific cofactor , it has become apparent that there can be significant plasticity in the metal cofactor they can utilize , particularly in the case of Mn- and Fe-utilizing enzymes [53] . In response to peroxide stress E . coli and many other pathogens sequester intracellular Fe and increase the expression of Mn importers , which in turn leads to accumulation of this metal [53–55] . In E . coli , this action results in Fe-utilizing enzymes , such as ribulose-5-phosphate 3-epimerase , becoming populated with Mn [53 , 56 , 57] . This change in cofactors enables E . coli to both maintain enzymatic activity and prevent Fenton chemistry-induced damage , which can arise from the interaction of Fe2+ with oxidants [53 , 56 , 57] . Notably , in many Fe-centric organisms , including E . coli , Salmonella typhimurium , and Yersinia pestis , Fe starvation increases the expression of Mn uptake systems and the accumulation of Mn [54 , 58–60] . In addition to enabling bacteria to activate Mn-dependent isozymes [61] , the increased Mn levels may also allow them to replace Fe with Mn in non-redox enzymes . These observations in conjunction with our findings suggest that populating metalloenzymes with an alternative yet catalytically active metal may be a general strategy used by bacteria to survive when a specific metal is limiting . While specific examples for metals other than Fe and Mn are currently lacking , conceptually this cofactor plasticity may enable microbes to maintain critical metabolic processes when limited for other essential metals . Amongst the staphylococci , S . aureus is the most pathogenic species and the only one that expresses two SODs [41–44] , with SodM presumably being gained through duplication and subsequent divergence . However , the advantage that S . aureus gains by expressing two Mn-dependent SODs , which are 75% identical at the level of their primary sequence , had not been apparent . Our current studies found that SodM is induced by CP-imposed Mn-starvation . The observation that SodM is cambialistic and enables S . aureus to maintain SOD activity when Mn starved by the host provides a rationale for its acquisition . It also suggests a model in which the Mn-dependent SodA is important during the initial colonization of a tissue , while SodM becomes important later during infection following the imposition of Mn starvation by the host immune response . Notably , S . aureus is not the only pathogen to express multiple superoxide dismutases that initially appear to be functionally redundant in culture , but upon subsequent analysis possesses properties that enhance fitness in the context of pathogenesis [62 , 63] . For example , a second Cu/Zn SOD expressed by some Salmonella is protease-resistant and binds to peptidoglycan , which enables it to retain activity and promote survival within the phagolysosome [63] . Cumulatively , these observations and our results emphasize the importance of evaluating the contribution of apparently redundant SODs to resisting oxidative stress in the context of the other stressors that an organism encounters within its ecological niche . The antimicrobial activity of CP is generally thought to be mediated by the sequestration of Mn2+ and Zn2+ [9 , 20 , 21] . However , CP was recently shown to bind Fe2+ , resulting in the suggestion that Fe restriction is a primary driver of its antimicrobial activity [47] . Notably , CP does not bind Fe3+ , the ionic state that exists in oxidizing environments such as sites of infection [9 , 20 , 47] . Additionally , several experimental lines of evidence suggest that Mn limitation contributes to the antimicrobial activity of CP both in culture and during infection . In both Acinetobacter baumannii and S . aureus , CP reduces intracellular Mn but not Fe levels [17 , 64] . The current observation that S . aureus replaces Mn in SodM with Fe even in the presence of concentrations of CP approaching 1 mg/ml further supports the idea that in culture CP is not imposing Fe limitation on S . aureus . Furthermore , in wild type mice loss of MntABC and MntH , the two Mn importers expressed by S . aureus , results in a substantial reduction in virulence; however , this defect is completely reversed in CP-deficient mice [13] . The observation that SodM is critical for infection in wild type but not CP-deficient mice further supports the idea that Mn but not Fe sequestration by CP contributes to controlling infection . Perhaps not surprisingly , given the myriad of high affinity staphylococcal Fe acquisition systems [65] , it also suggests that during infection S . aureus more successfully competes with the host for Fe than Mn . Cumulatively , these findings strongly support , at least in the case of S . aureus and A . baumannii , that Mn and not Fe sequestration significantly contributes to the antimicrobial activity of CP . Antibiotic resistance is a serious and growing threat to human health , with multiple agencies calling for the development of new approaches to treat bacterial infections [1 , 2] . Understanding how pathogens overcome innate immune defenses has the potential to reveal new opportunities for therapeutic intervention . Our studies reveal a new mechanism by which bacteria can overcome a two-pronged attack by the host . They also clearly demonstrate a role for cambialism in resisting nutritional immunity and bacterial pathogenesis . Moreover , these results provide newfound importance for a neglected family of proteins that is widely distributed throughout the tree of life .
All animal work was approved by the Vanderbilt University Institutional Animal Care and Use Committee ( protocol #M1600123 ) and was performed in accordance with United States Public Health Service Policy on Humane Care and Use of Laboratory Animals and the US Animal Welfare Act . Staphylococcus aureus strain Newman was used unless otherwise indicated . All strains and plasmids used in this study are listed in Tables 1 and 2 . S . aureus was routinely grown in tryptic soy broth ( TSB ) and on tryptic soy agar plates ( TSA ) , while E . coli was routinely cultivated in Luria Broth ( LB ) and on Luria agar plates . Both species were grown at 37°C . As needed for plasmid maintenance or gene inductions , 10 μg/ml of chloramphenicol , 50 μg/ml of kanamycin , 100 μg/ml of ampicillin , or 10 ng/ml of anhydrotetracycline was included in the media used . All strains were stored at -80°C in media containing 30% glycerol . The ΔsodA and ΔsodM mutants were created via Phi85 transduction of the sodA::tet and sodM::erm alleles from RN6390 [43] . The staphylococcal SodA and SodM expression constructs were created by amplifying sodA and sodM using the primers indicated in Table 3 and then cloning these fragments into the anhydrotetracycline-inducible plasmid pRMC2 [66] . To generate the YFP reporter plasmids the promoters for sodA and sodM were amplified with the primers listed in Table 3 and then cloned using standard techniques into pAH5 [67] . To generate the promoterless YFP construct pAH5 was digested with PstI and KpnI to remove the existing promoter , blunted and then self-ligated . Calprotectin growth assays were performed largely as previously described [13 , 20] , with the exception that overnight cultures were performed in Chelex-treated RPMI + 1% casamino acids ( NRPMI ) supplemented with 1 mM MgCl2 , 100 μM CaCl2 , and 1 μM FeCl2 . These cultures were diluted 1:100 into 100 μl of culture medium in a 96-well round-bottom plate and incubated at 37°C and with shaking at 180 rpm . The culture medium consisted of 38% TSB and 62% CP buffer ( 3 mM CaCl2 , 20mM Tris base , and 100 mM NaCl , 10 mM β-mercaptoethanol , pH 7 . 5 ) supplemented with 1 μM MnCl2 and 1 μM ZnSO4 . Where indicated 0 . 1 mM PQ was added to the media . The same growth conditions were utilized for the expression studies . For both growth and expression assays , optical density ( OD600 ) and fluorescence was assessed after 8 hrs of growth . Total and individual superoxide dismutase ( SOD ) activity were assayed using a water-soluble tetrazolium salt assay and a gel-based nitro blue tetrazolium assay , respectively , as previously described [14 , 68] . For both assays , the bacteria were grown as for the CP assays and harvested in exponential phase ( OD600 of ~ 0 . 3–0 . 35 ) . The cells were collected and then resuspended in 0 . 5 mM KPO4 buffer at pH 7 . 8 with 0 . 1 mM EDTA [46] . The bacteria were then lysed via mechanical disruption and centrifuged to remove insoluble material . The protein concentration in the cell lysate was determined via BCA assay ( Pierce ) . Total SOD activity was assessed using the SOD Assay Kit ( Sigma-Aldrich ) , per the manufacturer’s instructions . To evaluate the individual activity of each SOD , the cell lysates were normalized to total protein concentration and resolved on 10% native polyacrylamide gel . The gels were incubated in buffer containing 0 . 05 M potassium phosphate pH 7 . 8 , 1 mM EDTA , 0 . 25 mM nitro blue tetrazolium chloride , and 0 . 05 mM riboflavin and then exposed to light , as previously described [68] . To evaluate if SOD activity was due to iron-loading , prior to assessing activity the gels were incubated with 20 mM H2O2 or water for 20 minutes . Gels were imaged using a BioRad imager Universal Hood II and the fractional distribution of SOD activity was determined using the BioRad Quantity One software . The sodA and sodM genes were amplified by PCR from S . aureus genomic DNA using Pfu polymerase ( NEB ) and the primer pairs sodA_for and sodA_rev and sodM_for and sodM_rev , respectively , which incorporated 5’ NdeI and 3’ BamHI restriction sites . PCR products were A-tailed with Taq polymerase ( NEB ) and cloned into the pGEM-T vector ( Promega ) to yield pGEM-T-sodA and pGEM-T-sodM , respectively . An internal NdeI site in the sodM sequence was silently mutated by site-directed mutagenesis using the primer pair sodMqc1_for and sodMqc1_rev to yield pGEM-T-sodMqc . The genes were sub-cloned through NdeI/BamHI ( NEB ) digestion of the pGEM-T constructs , purification of the gene inserts by agarose gel electrophoresis , and subsequent ligation into NdeI/BamHI-digested pET29a vector ( Novagen ) to yield pET29a-sodA and pET29a-sodM constructs . Both pET29a constructs were sequenced ( GATC Biotech , Germany ) . A sequence error detected in pET29a-sodM ( deletion of base A570 ) was subsequently corrected through site-directed mutagenesis using primers sodMqc2_for and sodMqc2_rev , and the final construct confirmed through sequencing . The pET29a-sodA and pET29a-sodM constructs were transformed into Escherichia coli BL21 ( λDE3 ) cells and selected on LB agar plates containing 50 μg/ml kanamycin . For each cell type , cells were inoculated into M9 medium containing 10 μM FeSO4 and 50 μg/ml kanamycin and cultured overnight at 37°C with 180 rpm orbital shaking . This overnight culture was used to inoculate 1 L M9 medium containing 10 μM FeSO4 and 50 μg/ml kanamycin , cultured at 37°C with 180 rpm orbital shaking . At OD600 ~0 . 5 , protein expression was induced by addition of 1 mM isopropyl β–D-1-thiogalactopyranoside ( IPTG ) plus a further 20 μM FeSO4 and incubation for 4 h under the same conditions . Cells were harvested by centrifugation ( 20 min , 4 , 000 g , 4°C ) , washed in 20 mM Tris ( hydroxymethyl ) aminomethane ( Tris ) , pH 7 . 5 , 10 mM ethylenediaminetetraacetic acid ( EDTA ) , followed by a further wash in 20 mM Tris , pH 7 . 5 , 150 mM NaCl and stored at -20°C . Cells were resuspended in 20 mM Tris , pH 7 . 5 and lysed by sonication ( 6 x 10 s , with 1 min intervals , on ice ) and the lysate clarified by centrifugation ( 20 min , 19 , 000 g , 4°C ) . The soluble lysate was loaded onto a 5 ml HiTrap Q HP anion exchange chromatography ( AEC ) column ( GE Healthcare ) , the column was washed with 5 column volumes ( CV ) of buffer ( 20 mM Tris , pH 7 . 5 ) , followed by elution with a 9 CV linear NaCl gradient ( 0–1 M NaCl ) in the same buffer , collecting 2 ml fractions , using an Äkta fast performance liquid chromatography ( FPLC ) system ( GE Healthcare ) . Fractions were analyzed for protein by sodium dodecyl sulfate polyacrylamide gel electrophoresis ( SDS-PAGE ) . SodA eluted from AEC at ~230 mM NaCl , whereas SodM eluted at ~248 mM NaCl . Aliquots ( 1 ml ) of the peak AEC fractions containing the recombinant protein were further purified using the Äkta FPLC by size exclusion chromatography ( SEC ) on a Superdex 200 16/60 column ( GE Healthcare ) , resolved in 20 mM Tris , 150 mM NaCl , 5 mM EDTA , pH 7 . 5 at 1 ml/min , collecting 2 ml fractions . Concentrations of purified recombinant SodA and SodM were determined from A280nm measurements , using the empirically determined extinction coefficients ( ε280nm ) of 62 , 681 M-1 cm-1 for SodA and 64 , 949 M-1 cm-1 for SodM , each derived from quantitative amino acid analysis ( Alta Bioscience , UK ) . The metal content of the purified proteins was assessed by ICP-MS , as described below . To reconstitute recombinant SodA and SodM , which contained primarily Fe when purified from E . coli , with exclusively Mn they were unfolded and refolded in excess Mn , as previously described [69] , with modifications . Unfolding was performed in 2 . 5 M guanidine hydrochloride , in the presence of 5 mM EDTA and 20 mM 8-hydroxyquinoline to remove bound metal ions , followed by refolding through several rounds of dialysis against 20 mM Tris , 100 mM NaCl , 10 mM MnCl2 , pH 7 . 5 , to yield protein containing exclusively manganese . To analyze the bound metal , aliquots of each purified protein ( ~2 mg in 0 . 5 ml ) were resolved on a Superdex 200 Increase 10/30 column ( GE Healthcare ) in 20 mM Tris , 150 mM NaCl , 5 mM EDTA , pH 7 . 5 , resolved at 0 . 75 ml/min and collecting 0 . 5 ml fractions using the Äkta FPLC . Eluant fractions were analyzed for protein by A280nm and by SDS-PAGE , for elemental composition by ICP-MS , and for enzyme activity by in-gel activity assay . For elemental analysis via ICP-MS , protein-containing samples were diluted 50-fold into a solution of 2 . 5% HNO3 ( Suprapur , Merck ) containing 20 μg/l Co and Pt as internal standards . Matrix-matched elemental standards ( containing analyte metal concentrations of 0–500 μg/L ) were prepared by serial dilution from individual metal standard stocks ( VWR ) with identical solution compositions , including the internal standard . All standards and samples were analyzed by ICP-MS using a Thermo x-series instrument operating in collision cell mode ( using 3 . 0 ml/min flow of 8% H2 in He as the collision gas ) . Isotopes 55Mn , 56Fe , 59Co , 66Zn , and 195Pt were monitored using the peak-jump method ( 100 sweeps , 20–30 ms dwell time on 3–5 channels per isotope , separated by 0 . 02 atomic mass units ) in triplicate , and metal concentrations determined from the standard curve . Nine-week-old female black C57BL/6 and CP -/- ( C57BL/6 S100A9-/- ) mice were infected using a retro-orbital infection model [9 , 13 , 14] . Livers were harvested and homogenized 96 hours post-infection . Serial dilutions were then plated and counted for colony forming units . | During infection , pathogens must overcome the restriction of essential nutrients such as manganese by the host , while simultaneously coping with other host defenses such as the oxidative burst . Using the host protein that limits manganese availability during infection and mice lacking this effector , we determined that acquisition of a second superoxide dismutase that is capable of using either manganese or iron enhances the ability of Staphylococcus aureus to cause infection . When manganese-starved by the host , this cambialistic enzyme enables S . aureus to maintain superoxide dismutase activity and survive when exposed to oxidative stress . These results reveal the important contribution of cambialistic superoxide dismutases to bacterial pathogenesis and represent a new mechanism for resisting manganese starvation during infection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"manganese",
"oxidative",
"stress",
"enzymes",
"pathogens",
"microbiology",
"enzymology",
"dismutases",
"staphylococcus",
"aureus",
"bacterial",
"diseases",
"nutrition",
"bacteria",
"ma... | 2017 | A Superoxide Dismutase Capable of Functioning with Iron or Manganese Promotes the Resistance of Staphylococcus aureus to Calprotectin and Nutritional Immunity |
Over two-thirds of the world's population lives in regions where rabies is endemic , resulting in over 15 million people receiving multi-dose post-exposure prophylaxis ( PEP ) and over 55 , 000 deaths per year globally . A major goal in rabies virus ( RABV ) research is to develop a single-dose PEP that would simplify vaccination protocols , reduce costs associated with RABV prevention , and save lives . Protection against RABV infections requires virus neutralizing antibodies; however , factors influencing the development of protective RABV-specific B cell responses remain to be elucidated . Here we used a mouse model of IL-21 receptor-deficiency ( IL-21R−/− ) to characterize the role for IL-21 in RABV vaccine-induced immunity . IL-21R−/− mice immunized with a low dose of a live recombinant RABV-based vaccine ( rRABV ) produced only low levels of primary or secondary anti-RABV antibody response while wild-type mice developed potent anti-RABV antibodies . Furthermore , IL-21R−/− mice immunized with low-dose rRABV were only minimally protected against pathogenic RABV challenge , while all wild-type mice survived challenge , indicating that IL-21R signaling is required for antibody production in response to low-dose RABV-based vaccination . IL-21R−/− mice immunized with a higher dose of vaccine produced suboptimal anti-RABV primary antibody responses , but showed potent secondary antibodies and protection similar to wild-type mice upon challenge with pathogenic RABV , indicating that IL-21 is dispensable for secondary antibody responses to live RABV-based vaccines when a primary response develops . Furthermore , we show that IL-21 is dispensable for the generation of Tfh cells and memory B cells in the draining lymph nodes of immunized mice but is required for the detection of optimal GC B cells or plasma cells in the lymph node or bone marrow , respectively , in a vaccine dose-dependent manner . Collectively , our preliminary data show that IL-21 is critical for the development of optimal vaccine-induced primary but not secondary antibody responses against RABV infections .
RABV is a single-stranded negative sense RNA virus of the genus lyssavirus in the Rhabdoviridae family that kills approximately 55 , 000 people annually . Up to 60% of rabies cases are in children , making rabies the seventh most important infectious disease in terms of years lost [1] . In Africa , a person dies of rabies every 20 minutes [2] . In China , rabies became the leading cause of infectious disease mortality in 2006 , which increased by more than 27% from 2005 [3] . In the United States , cases of rabies in wildlife are detected in virtually all states and Puerto Rico ( Hawaii is considered rabies-free ) . Except for cattle and foxes , the incidence of rabies in domesticated or wildlife remained unchanged or significantly increased in the US in 2011 compared to the five-year average for each species [4] , exemplifying the difficulty in containing zoonotic viral infections even in industrialized nations . The cost associated with rabies in the US , Africa and Asia is almost $1 billion annually [5] , [6] contributing to the financial burden of global health care costs . Furthermore , rabies is a NIAID Category C Priority Pathogen , indicating rabies is an emerging infectious disease with the potential for mass dissemination and harm to people [7] . Together , rabies is considered a neglected global zoonotic infectious disease that disproportionately affects children and , therefore , understanding how B cells develop in response to experimental RABV-based vaccination may help to support efforts to develop a single-dose human rabies vaccine for use in both developing and industrialized countries . A wide array of RABV variants exist , ranging from highly pathogenic strains to attenuated RABV vaccine strains such as the molecular clone SAD B19 [8] . Live attenuated RABV vaccine strains are highly immunogenic and potentially could serve as a single-dose human RABV vaccine to replace currently used multi-dose inactivated RABV-based vaccine regimens . Due to residual pathogenicity of these live virus strains , however , several “second-generation” RABV-based vaccines are under investigation in which entire genes are deleted from the RABV genome [9]–[12] , or multiple pathogenic markers are genetically modified [13] . Data from these studies indicate that very safe and effective live RABV-based vaccine vectors can be generated . Despite extensive efforts to attenuate live RABV-based vaccine vectors for safety , little information is available on factors that influence the generation of effective antibodies in response to live RABV-based vaccines . Virus neutralizing antibodies ( IgG but not IgM ) directed against the RABV glycoprotein ( G ) are protective against pathogenic RABV strains [14] , [15] . In the case of a replication-deficient RABV-based vaccine in which the matrix gene is deleted , VNAs are generated by T cell-independent and –dependent ( extrafollicular and germinal center ) mechanisms [16] , suggesting multiple pathways of B cell activation and differentiation could be exploited to rationally design a single-dose RABV vaccine for use in both pre- and post-exposure settings . With respect to typical vaccine-induced antibody responses , APC-primed T cells most likely display an intermediate Tfh phenotype ( i . e . , “pre-Tfh cell” ) characterized phenotypically as CD4+CXCR5hiPD1lo , which migrate the T and B cell border of secondary lymphoid organs and interact with their cognate antigen-primed B cells [17] . This T∶B cell interaction typically results in the Tfh cells producing optimal amounts of IL-21 , and in the B cells differentiating into early short-lived extrafollicular antibody secreting cells or migrating into the follicles and forming GCs . With additional signals provided by Tfh cells in GCs , B cells mature and differentiate into long-lived plasma cells ( PCs ) secreting high affinity antibodies or into memory B cells . Due to the importance for PCs secreting high affinity antibodies and memory B cells in vaccine-induced immunity , the development of Tfh cells and CG B cells is critical for vaccine-induced protection against future exposures . In the context of RABV-specific vaccination in post-exposure settings , the rapid induction of extrafollicular B cell responses may also be critical to prevent infection of the CNS , especially in cases where treatment is delayed after exposure to a potentially infected animal . As such , understanding factors that generate short- and long-term anti-viral B cell responses will help design more efficacious RABV vaccines for use in humans . Cytokines present at the time of antigen exposure influence T and B cell activation and GC formation and , therefore , also affect the outcome of vaccination . IL-21 [18] , [19] is a type 1 cytokine that is a member of the common γ-chain receptor family , which also includes IL-2 , IL-4 , IL-7 , IL-9 , and IL-15 . It is produced primarily by activated Tfh and Th17 cells and has pleiotropic effects throughout innate and adaptive immunity [reviewed in [20]] . The role for IL-21 in regulating Tfh and B cell functions was originally identified using model antigens [21]–[23] . In addition , the role for IL-21 in immunity and protection against helminth [24] , viral [25] , [26] and bacterial [22] infections has been studied . IL-21 is a key mediator for the control of persistent viral infections in mouse models of LCMV [27]–[29] and hepatitis B virus [30] , or in humans infected with HIV [31]–[33] or HIV in combination with the 2009 H1N1 influenza virus vaccine [34] . However , the complexity and diversity of persistent infections makes it is difficult to pinpoint the effects of IL-21 on anti-viral B cells versus CD8+ T cells , both of which can contribute the control of many chronic infections [35] . Conversely , B cells play a critical role for clearance of most acute viral infections and for the efficacy of vaccines against most vaccine-preventable diseases . Clearance of RABV infections relies strictly on B cell-mediated effector functions , but not CD8+ T cells , for protection , making RABV infection an excellent mouse model to pinpoint the role for IL-21 in vaccine-induced immunity against RABV infections and potentially for other pathogens that rely solely on B cells for protection . In this report , our preliminary data indicate that IL-21 is critical for the development of effective vaccine-induced primary antibody responses against RABV infections by influencing GC B cells or PC generation in a vaccine dose-dependent manner , while also showing IL-21 is dispensable for RABV-specific secondary antibody responses when a primary antibody response develops .
All animal work was reviewed and approved by the Institutional Animal Care and Use Committee ( IACUC ) of Jefferson Medical College , Thomas Jefferson University . Work was completed in accordance with international standards [Association for Assessment and Accreditation of Laboratory Animal Care ( AAALAC ) ] and in compliance with Public Health Service Policy on Humane Care and Use of Laboratory Animals , The Guide for the Care and Use of Laboratory Animals of the National Institutes of Health ( NIH ) . The construction of the live RABV-based vaccine ( rRABV ) used in this study was described elsewhere and was previously named SPBN [9] , [36] , [37] . This vaccine is a molecular clone derived from the attenuated SAD-B19 vaccine strain of RABV . Virus stocks were propagated on baby hamster kidney cells and then concentrated and purified over a 20% sucrose cushion . The challenge virus used was the pathogenic Challenge Virus Strain-N2c ( CVS-N2c ) , which is a mouse-adapted sublclone of CVS-21 RABV [38] . CVS-N2c was initially propagated in neonatal mouse brains and then passaged once in-vitro on a neuroblastoma cell line ( NA cells ) . The titer of CVS-N2c required to kill unvaccinated mice was determined experimentally by inoculating serial ten-fold dilutions into naïve immune-competent mice [39] i . m . and then observing mice daily for clinical neurological symptoms of rabies . The titer required to kill unvaccinated mice within 8 days post-infection , which is typical for CVS-N2c [9] , [10] , was determined to be 105 focus forming units ( ffu ) /mouse . Cryopreserved embryos of mice deficient in the IL-21 receptor ( B6;129-IL21r tm1Wjl/Mmucd ) ; #015505-UCD ) [39] were obtained from the Mutant Mouse Regional Resource Centers ( NIH ) and implanted and bred in-house at Thomas Jefferson University in a pathogen-free animal facility . Control C57BL/6 mice were obtained from the Frederick National Laboratory for Cancer Research ( NCI ) . The following antibodies where purchased from BD Biosciences , unless otherwise noted , and used for flow cytometry staining: APC-Cy7-B220 ( clone RA3-6B2 ) , PerCP-CY5 . 5-CXCR5 ( clone 2G8 ) , APC-CD138 ( clone 281-2 ) , PE-Cy7-CD95/Fas ( clone Jo2 ) , FITC-GL7 , eFluor 450-CD4 ( clone RM4–5 , eBioscience ) , PE-PD1 ( clone J43 , eBioscience ) , Alexa Fluor 700-CD38 ( clone 90 , eBioscience ) , rat anti-mouse CD16/32 ( FcBlock; Pharmingen ) . Groups of 6- to 10-week-old female IL-21R−/− or wild-type C57BL/6 mice were inoculated intramuscularly ( i . m . ) with 103 or 105 focus forming units ( ffu ) /mouse with rRABV or an equivalent volume of PBS as controls . Five weeks post-immunization , mice were challenged i . m . with 105 ffu/mouse with CVS-N2c and then observed for three weeks for clinical signs of rabies . Mice were euthanized at the onset of neurological symptoms . At various times post-immunization and challenge , blood was collected retro-orbitally . Three-fold serial dilutions of sera were tested by ELISA to determine RABV G-specific IgG antibodies as described [10] and reported as the reciprocal serial dilution . Data represented two independent experiments ( N = 9–11 mice per group ) . To measure virus neutralizing antibody titers , the Rapid Fluorescent Foci Inhibition Test ( RFFIT ) was completed on pooled sera from two independent experiments as described previously [16] , [40] , [41] . Groups of 6- to 10-week-old female IL-21R−/− or wild-type mice were inoculated i . m . with 103 or 105 ffu/mouse with rRABV or PBS/naive as controls . Draining lymph nodes and bone marrow cells were collected 7 and/or 14 days post-immunization . Single cell suspensions ( 106 cells/sample ) were incubated with rat anti-mouse CD16/32 ( 1 ug/106 cells ) in fluorescence-activated cell sorter ( FACS ) buffer ( PBS supplemented with 2% fetal bovine serum ) for 1 h on ice . Cells were washed twice with FACS buffer and incubated with fluorescently conjugated antibodies ( 0 . 2 ug/106 cells ) for 30 min at RT in the dark . Cells were subsequently washed 2 times with FACS buffer and fixed in 2% paraformaldehyde in PBS for 30 minutes . Flow cytometry was completed using FACScan ( BD LSRII ) and analyzed by FlowJo software . Data represents samples completed in duplicate ( N = 3–5 mice ) [16] . Kaplan-Meier survival curves were analyzed by the log rank test; *p = <0 . 05 indicates significant survivorship between two immunization groups [9] , [10] . Statistical difference between two groups of data was compared using an unpaired , two-tailed t test and data is presented at the mean ± SEM; *p<0 . 05 , **p = 0 . 01–0 . 001 , ***p≤0 . 001 [9] , [10] .
Cytokines present at the time of immunization have the ability to affect the outcome of vaccine-induced B cell responses . Due to the importance for IL-21 in promoting effective T-dependent B cell responses [21] , [22] , we examined the requirement for IL-21R signaling in the generation of antibodies in a mouse model of RABV immunogenicity and protection using a mouse model of IL-21R-deficiency . These mice , designated here as IL-21R−/− mice , lack the IL-21R extracellular and transmembrane domains but show normal lymphoid development [39] . T and B cells exhibit similar proliferative responses to CD3-speficic antibodies or LPS , respectively , when compared to wild-type controls [39] . As a group , IL-21R−/− mice immunized with a low dose of a live recombinant RABV-based vaccine ( rRABV ) ( 103 ffu/mouse ) showed only low levels of anti-RABV G antibodies that were not significantly different from PBS-immunized mice at all time points tested post-immunization ( Figure 1 , left panels , a–e ) , although at least three IL-21R−/− mice developed anti-RABV antibodies by day 28 post-immunization ( Figure 1B ) . The seroconversion of these three IL-21−/− mice may explain the limited protection observed in the pathogenic challenge experiments described later in this report . Wild-type mice immunized with the same dose of rRABV showed significant levels of anti-RABV G antibodies as early as 7 days post-inoculation compared to rRABV-immunized IL-21R−/− mice , and these antibody responses continued to increase through day 21 post-immunization . VNA titers detected 28 days post-immunization ( Figure 1C ) are consistent with the antibody titers detected by ELISA ( Figure 1B ) indicating anti-RV antibody titers detected by ELISA are representative of the ability for vaccine-induced antibodies to neutralize rabies virus . IL-21R−/− mice immunized with a higher dose of rRABV ( 105 ffu/mouse ) showed significantly reduced anti-RABV antibody responses when compared to wild-type mice also immunized with 105 ffu/mouse of rRABV ( Figure 1 , right panels , f–j ) . Together , this data indicates that IL-21 is critical for the induction of optimal primary anti-RABV antibody responses , especially when low doses of vaccine are used . We next wanted to evaluate the effect of IL-21 on vaccine-induced antibody recall responses and protection against pathogenic RABV challenge . Five weeks post-immunization with rRABV , mice from Figure 1 were challenged with 105 ffu/mouse of a highly pathogenic mouse-adapted RABV strain ( Challenge Virus Strain-N2c; CVS-N2c ) [38] , which typically kills naïve mice within 8 days post-infection [9] , [10] . Consistent with the low antibody titers detected during the primary antibody response , significantly less anti-RABV antibodies were detected three or five days post-challenge in IL21R−/− mice immunized with 103 ffu/mouse of rRABV compared to the antibody recall response detected in wild-type mice ( Figure 2 , panels a and b , and Figure 2B ) and antibody recall titers were not significantly different from PBS-immunized/CVS-N2c-challenged mice ( Figure 2 , e ) . Only 40% of IL-21R−/− mice immunized with 103 ffu/mouse of rRABV were protected against pathogenic RABV challenge , while all wild-type mice that were similarly immunized were protected against challenge ( Figure 3 , left panel ) . As expected , those mice with higher antibody titers showed protection compared to mice with lower antibody titers ( Figure 2B ) . Conversely , an antibody recall response was induced in IL-21R−/− mice immunized with 105 ffu/mouse of rRABV within three days post-challenge with CVS-N2c at levels equivalent to rRABV-immunized wild-type mice ( Figure 2 , panels c and d ) and all immunized IL-21R−/− mice were protected against pathogenic RABV challenge ( Figure 3 , right panel ) . Taken together , IL-21 is required for optimal primary ( Figure 1 ) but not secondary ( Figure 2 ) antibody responses to RABV vaccination . Furthermore , IL-21 is required for protection against pathogenic RABV in a vaccine dose-dependent manner ( Figure 3 ) . Next we wanted to determine whether the impaired primary anti-RABV antibody response in IL-21R−/− mice was due to an overall defect in the generation of Tfh and/or GC B cells . Lymph nodes from IL-21R−/− or wild-type mice were collected 7 or 14 days post-immunization with 103 or 105 ffu/mouse of rRABV or PBS alone to determine the influence of IL-21 on Tfh and B cell populations . Representative gating strategies [17] , [42] to identify CD4+ T cells from the total live lymph node cultures ( Figure 4A ) or Tfh ( CD4+CXCR5hiPD1hi ) cells from the CD4+ T cell populations ( Figure 4B ) are shown . A significant increase in the number of CD4+ T cells displaying a Tfh phenotype was detected in IL-21R−/− mice 14 days post-immunization with 103 or 105 ffu/mouse rRABV compared to similarly immunized wild-type mice ( Figure 4C and Figure 4D , respectively ) . However , the formation of optimal GC B cells appears to be dependent on the dose of vaccine administered ( Figure 5 ) . Representative gating strategies [22] , [43] , [44] to identify B220+ B cells from the total live lymph node cultures ( Figure 5A ) or GC B cells ( B220+GL7hiCD95/Fashi ) from the B220+ B cell population ( Figure 5B ) are shown . IL-21R−/− mice immunized with a low dose of vaccine failed to induce optimal GC B cell formation compared to wild-type mice 14 days post-immunization , as shown by a significant decrease in the number of GC B cells in IL-21R−/− mice compared to wild-type mice ( Figure 5C ) . However , a significant increase in the number of GC B cells was detected in IL-21R−/− mice immunized with 105 ffu/mouse of rRABV compared to wild-type mice 14 days post-immunization ( Figure 5D ) . The data indicates that the suboptimal primary antibody responses detected in IL-21R−/− mice immunized with 103 ffu/mouse of rRABV appears to be due to the lack of GC B cell formation , while the suboptimal primary antibody response detected in IL-21R−/− mice immunized with 105 ffu/mouse most likely does not result from a defect in Tfh or GC B cell development . Our analysis above shows that IL-21R signaling is dispensable for the formation of the Tfh and GC B cell populations in response to higher doses of rRABV , indicating that other B cell types are more likely responsible for the suboptimal primary antibody titers detected in IL-21R−/− mice immunized with 105 ffu/mouse . Since IL-21 can also influence the balance between the generation of memory B cells and PCs [24] , [45]–[47] , we investigated the role for IL-21R signaling to regulate memory B cell and PCs populations in response to RABV vaccination . Figure 6A and Figure 6B show representative gating strategies [48]–[50] to identify the memory B220+ B cells ( CD38+CD138− ) from the lymph node and PC ( B220loCD138+ ) populations from the bone marrow from mice immunized with 103 or 105 ffu/mouse of rRABV or PBS alone . The presence or absence of IL-21R does not appear to influence the development of memory B cells in mice immunized with 103 ffu/mouse of rRABV ( Figure 6C ) . However , the percentage of memory B cells was significantly increased in the lymph node cell cultures from IL-21R−/− mice immunized with 105 ffu/mouse of rRABV as early as 7 days post-immunization compared to immunized wild-type mice ( Figure 6C ) . By day 14 post-immunization , similar memory B cell populations were measured ( data not shown ) , indicating that IL-21 is not required for the formation of memory B cells in response to live RABV-based vaccination . This is consistent with the findings in Figure 1 and Figure 2 indicating that IL-21 is dispensable for secondary antibody responses against RABV infection when a primary antibody response develops . However , the percentage of PCs was reduced in the bone marrow of IL-21R−/− mice immunized with 105 ffu/mouse compared to rRABV- or PBS-immunized wild-type mice 14 days post-immunization ( Figure 6D ) , consistent with the suboptimal primary antibody titers detected in IL-21R−/− mice .
Current rabies PEP regimens are based on multiple doses of inactivated RABV-based vaccines administered intramuscularly or intradermally . In cases of severe exposure , rabies immune globulin ( RIG ) is administered [51]–[53] . The development of a single-dose vaccine would greatly benefit human rabies prevention by reducing the cost of vaccination and saving lives . Understanding immune parameters that influence the magnitude and/or quality of anti-RABV antibody responses may lead to more effective single-dose vaccines [54] . Correlates of protection against rabies infections are defined as virus neutralizing antibodies directed against the single viral transmembrane glycoprotein ( G ) [15] , [52] , [54] , [55] . CD8+ T cells do not appear to be important for the clearance of RABV infections [56] . Protection against RABV infection typically requires CD4+ T cell help [56]–[60] , although we recently showed that this requirement is not absolute and that protection against pathogenic RABV challenge can be afforded in mice devoid of all T cells ( TCRβδ−/− mice ) vaccinated with a matrix gene-deleted RABV-based vaccine ( rRABV-ΔM ) [16] . Furthermore , we show that mice immunized with rRABV-ΔM also induce antibodies by T cell-dependent extrafollicular B cell responses before GC-derived B cells are detected . Together , our previous work identified multiple pathways of B cell development that can be exploited to make more efficacious RABV-based vaccines for use in humans . Nonetheless , very little information is available on how effective B cells develop in response to live RABV-based vaccination . IL-21 is a pleotropic cytokine that is produced by NKT cells and CD4+ T cells , most notably Th17 and Tfh cells . IL-21 binds to the IL-21R on a wide variety of cells involved in innate immunity , including DCs , NK cells , NKT cells , and macrophages , as well as on cells involved in adaptive immunity , such as B cells and CD4+ or CD8+ T cells [reviewed in [20]] . Due to its multiple roles in innate and adaptive immunity , IL-21 has the potential to influence the quality and magnitude of vaccine-induced immunity to acute viral infections . Here we used a mouse model of IL-21R-deficiency to evaluate the role for IL-21R signaling in vaccine-induced protection against RABV; i . e . , an acute viral infection that relies on B cells for protection that has implications for global public health initiatives . In this report , we showed that IL-21R signaling is critical for the generation of optimal primary anti-RABV antibody responses to vaccination . Primary anti-RABV antibody titers were significantly reduced in immunized IL-21R−/− mice compared to wild-type mice at almost all time points tested post-immunization , suggesting IL-21R signaling plays important roles throughout RABV-specific primary B cell responses . Nonetheless , IL-21R signaling appears to influence immunity in a vaccine dose-dependent manner , which is consistent with findings by others suggesting the influence of IL-21 is dependent on the model studied [22] , [61] . Indeed , significantly less IL-21R−/− mice immunized with low-dose vaccination were protected against pathogenic challenge compared to wild-type mice while all IL-21R−/− and wild-type mice immunized with high-doses of vaccine survived challenge similarly . Despite the differences in protection elicited in IL-21R−/− mice immunized with different doses of vaccine , it appears that IL-21 is critical for the generation of optimal RABV-specific primary B cell responses . One potential explanation for the suboptimal primary antibody responses observed in immunized IL-21R−/− mice compared to immunized wild-type mice might be that GC B cells failed to form in IL-21R−/− mice , therefore , the GC B cell compartment was analyzed in mice immunized with different doses of vaccine . GC-derived B cells were reduced in IL-21R−/− mice immunized with a low dose of vaccine compared to similarly immunized wild-type mice , indicating that IL-21 is required for optimal GC B cell formation in response to low-dose RABV-based vaccination . On the other hand , GC-derived B cells expanded in IL-21R−/− mice immunized with a high dose of vaccine compared to similarly immunized wild-type mice , indicating that IL-21 is dispensable for GC B cell formation after high-dose vaccination with rRABV-based vaccines . Furthermore , the data indicates that factors other than IL-21 were responsible for GC B cell formation in IL21R−/− mice immunized with higher doses of vaccine . Multiple signals lead to B cell activation and functions . These signals can come from BCR or Toll-like receptor ( TLR ) ligation , TNF superfamily receptor engagement ( eg . , via BAFF and APRIL ) or cytokine signaling . Furthermore , B cell activation is contextual , meaning B cells are differentially activated in the presence of different signals at the time of antigen exposure . Due to the repetitive display of rabies antigen on the surface of infectious particles , the potential exists that cross-linking BCRs and/or TLRs on the surface of B cells overcame the requirement for IL-21R signaling in B cell activation when high doses of vaccine are administered . The influences of these and other B cell signaling events in the context of RABV-based vaccine-induced B cell activation were not directly measured in these studies and remain to be elucidated . Nonetheless , based on the results reported here , it appears that IL-21R signaling is important for optimal primary vaccine-induced antibody responses to RABV vaccination especially when low doses of vaccine are administered . As noted above , a rapid antibody response is critical for rabies PEP to neutralize virus before it reaches the CNS . We have recently shown that RABV-based vaccines are able to induce early and rapid T cell-dependent extrafollicular antibody responses before GC B cells are formed [16] . These early pre-GC B cell responses contributed to the protection against pathogenic RV challenge early post-immunization , which is an important attribute for PEP [16] . In the studies described in this report , we detected a significant reduction in antibody titers in IL-21R−/− mice as early as 5 days post-immunization with a high dose of rRABV compared to immunized wild-type mice , suggesting that IL-21 may be influencing the outcome of extrafollicular antibody responses in the context of RABV vaccination , although this was not directly studied in this report . Nonetheless , the frequency of Tfh and GC B cells was similar in IL-21R−/− mice compared to wild-type mice 7 days post-immunization and , therefore , it would appear that the early suboptimal antibody responses in IL-21R−/− mice may be due to impaired extrafollicular PCs directly and not through impaired Tfh or GC B cell formation . This is consistent with the findings that IL-21 can promote Blimp-1 expression and PC development [47] , IL-21- or IL-21R-deficiency decreases extrafollicular PCs in a model of NP-KLH immunity [62] , and that IL-21 acts on early stages of B cell differentiation before GC or PC B cells are formed [42] . Finally , IL-21 has been reported to be important for Tfh cell maintenance but not formation . Together , existing data suggests that IL-21R−/− signaling influences early events in pre-GC B cell development in the context of RABV vaccination [22] . IL-21 can also influence the balance of B cell differentiation into memory B cells or PCs [45]–[47] . The specific role for IL-21 in memory B cell responses is not completely clear and appears to rely on the type of antigen used and the model studied [61] , [62] . In the context of RABV vaccination , IL-21R signaling was not required for the generation of B cells displaying a memory B cell phenotype in IL-21R−/− mice immunized with either vaccine dose . This is consistent with our finding showing that IL-21R signaling is not required for optimal secondary anti-RABV G antibody titers after challenge with pathogenic RABV . However , we detected a decrease in the number of PCs in IL-21R−/− mice immunized with either dose of vaccine compared to wild-type mice . We cannot determine whether the slightly suboptimal PC subset detected in IL-21R−/− mice immunized with low doses of vaccine was indirectly a result of impaired GC B cell development or directly as a result of impaired PC formation itself . However , in mice immunized with a high dose of vaccine where we observed an expansion of GC-derived B cells in IL-21R−/− mice , we also observed a decrease in PCs in the bone marrow compared to wild-type mice , indicating that IL-21 acts directly on the formation of PCs . Together , the data shows that IL-21 influences the balance between memory and PC B cell formation in the context of RABV vaccination . Despite the impaired primary antibody response and PC B cell formation in immunized IL-21R−/− mice , we detected an expansion of CD4+ T cells displaying a pre-Tfh ( data not shown ) , Tfh cell and GC B cell phenotype in IL-21R−/− mice compared to wild-type mice at 14 days post-immunization . The expansion of GC B cells and Tfh cells in the absence of IL-21R signaling was also shown by King I . L . et al in a model of Heligmosomoides polygyrus immunity [24] , suggesting that IL-21R signaling may not be necessary for the generation of these cell types in response to a wide range of pathogens or vaccination . Furthermore , the increase in GC B cells and Tfh cells in H . polygyrus-infected or RABV-vaccinated IL-21R−/− mice suggests that IL-21R signaling may play an inhibitory role in the development of T and B cells in the context of some pathogens , which is consistent with the ability for IL-21 to activate or inhibit immune function depending on the antigen and available co-stimulatory signals [20] . The expansion of Tfh and GC B cells in IL-21R−/− mice compared to wild type mice is also consistent with the finding that IL-21 has the ability to mediate apoptosis in primary resting and activated murine B or to promote apoptosis or growth arrest for non-specifically activated B cells [63] . Alternatively , the elevated number of GC Tfh cells could be a result of the lack of PC that developed in the IL21R−/− mice . Pelletier et al described a negative regulatory feedback-loop in which antigen-specific PCs negatively regulate antigen-specific Tfh cell development and function [64] . In this report , they also observed a significant expansion of Tfh cells and GC B cells in the absence of PC development . Together , the role for IL-21 in the homeostatic balance of T and B cell development in the context of infectious diseases appears to be important and remains to be fully elucidated . Additional studies are needed to identify the exact cell type ( s ) responsible for the affects described in this report . While we speculate that B cell-intrinsic IL-21R signaling is responsible for the induction of optimal anti-RABV antibody responses , we cannot rule out the influence of other cell types that also express IL-21R . IL-21 has the ability to influence the function of macrophages , NK cells and NKT cells by affecting survival/apoptosis , antigen processing , and cytokine secretion [reviewed in [20]] . The function of these cells of the innate immune system may indirectly be affecting the outcome of B or T cell functions in the context of RV vaccination . Nonetheless , IL-21 has the potential to influence a wide range of B cell functions and pathways . Our preliminary data indicates that IL-21 is critical for the formation of optimal vaccine-induced primary antibody responses and demonstrates an important role for IL-21 in the generation of vaccine-induced immunity against RABV infection and perhaps other acute infections that rely on B cell-mediated effector functions for protection . | Over two-thirds of the world's population lives in regions where rabies is endemic , resulting in over 15 million people receiving post-exposure treatment . A person , disproportionately a child , dies of rabies every 20 minutes and the cost of rabies prevention exceeds $1 billion US dollars per year . The development of a single-dose human rabies vaccine would greatly reduce the burden of rabies globally by lowering the cost associated with rabies vaccination and saving lives . Understanding how B cells develop to produce protective virus neutralizing antibodies would greatly help to achieve the goal of developing a single-dose vaccine . In this report , we show that IL-21 is critical for the induction of primary vaccine-induced anti-RABV G antibody titers and that the effects of IL-21 are highly dependent on the dose of vaccine administered . In our model of rabies immunogenicity and protection , the lack of IL-21 receptor influenced the detection of B cells in germinal centers in lymph nodes or of plasma cells in bone marrow after immunization with low or high doses of vaccine , respectively . Overall , these preliminary results indicate that IL-21 has the potential to influence B cell development and functions in the context of rabies vaccine-induced immunity and protection . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"humoral",
"immunity",
"medicine",
"viral",
"vaccines",
"immune",
"cells",
"emerging",
"viral",
"diseases",
"immunity",
"to",
"infections",
"immunology",
"microbiology",
"host-pathogen",
"interaction",
"emerging",
"infectious",
"diseases",
"global",
"health",
"immunizatio... | 2013 | Investigating the Role for IL-21 in Rabies Virus Vaccine-induced Immunity |
We have performed a metabolite quantitative trait locus ( mQTL ) study of the 1H nuclear magnetic resonance spectroscopy ( 1H NMR ) metabolome in humans , building on recent targeted knowledge of genetic drivers of metabolic regulation . Urine and plasma samples were collected from two cohorts of individuals of European descent , with one cohort comprised of female twins donating samples longitudinally . Sample metabolite concentrations were quantified by 1H NMR and tested for association with genome-wide single-nucleotide polymorphisms ( SNPs ) . Four metabolites' concentrations exhibited significant , replicable association with SNP variation ( 8 . 6×10−11<p<2 . 8×10−23 ) . Three of these—trimethylamine , 3-amino-isobutyrate , and an N-acetylated compound—were measured in urine . The other—dimethylamine—was measured in plasma . Trimethylamine and dimethylamine mapped to a single genetic region ( hence we report a total of three implicated genomic regions ) . Two of the three hit regions lie within haplotype blocks ( at 2p13 . 1 and 10q24 . 2 ) that carry the genetic signature of strong , recent , positive selection in European populations . Genes NAT8 and PYROXD2 , both with relatively uncharacterized functional roles , are good candidates for mediating the corresponding mQTL associations . The study's longitudinal twin design allowed detailed variance-components analysis of the sources of population variation in metabolite levels . The mQTLs explained 40%–64% of biological population variation in the corresponding metabolites' concentrations . These effect sizes are stronger than those reported in a recent , targeted mQTL study of metabolites in serum using the targeted-metabolomics Biocrates platform . By re-analysing our plasma samples using the Biocrates platform , we replicated the mQTL findings of the previous study and discovered a previously uncharacterized yet substantial familial component of variation in metabolite levels in addition to the heritability contribution from the corresponding mQTL effects .
Expression quantitative trait loci ( eQTL ) studies have proved a powerful aid to functional genomics , with many thousand genetic loci now highlighted that affect RNA transcription levels or splicing in human tissues [1] . eQTL studies have accelerated the characterization of biological mechanisms governing gene regulation [2]–[5] , and genome-wide multi-tissue maps of known eQTLs have clarified the biological basis for a proportion of disease-associated [6]–[7] and positively selected [8] loci ( e . g . http://eqtl . uchicago . edu/cgi-bin/gbrowse/eqtl/ ) . Genetic variation at eQTLs can be incorporated into network models that help define dependence between genotypes , molecular traits , environment , and physiological states [9]–[10] . The success of eQTL studies points to the potential value in applying the eQTL paradigm to other molecular traits besides mRNA transcript levels [11]–[14] . In the current study , we associate genome-wide genetic variation with concentrations of metabolites , small molecules involved in biochemical processes in living systems , which can be measured in samples such as biofluids and tissue extracts using 1H nuclear magnetic resonance spectroscopy ( 1H NMR ) [15]–[17] , or by the Biocrates platform . ( For convenience , we use the term ‘Biocrates platform’ in the current paper to refer to the targeted-metabolomic platform using flow-injection tandem mass spectrometry—FIA-MS—developed by Biocrates Life Sciences [14] , [18] . ) Metabolites are mechanistically further removed from the genome than are mRNAs , creating an important qualitative distinction between metabolite QTL ( mQTL ) and eQTL studies . The mRNA-to-gene mapping is a useful property of eQTL studies , allowing the search for a cis eQTL of each mRNA to be focused on a relatively small , gene-centred region . Moreover , most known eQTLs are cis-acting single-nucleotide polymorphisms ( SNPs ) , lying usually within tens of kb of the genes whose expression they influence [1] , [5] . Whilst metabolite concentrations are influenced indirectly by mRNA and protein expression , there is not typically a one-to-one metabolite-to-gene correspondence known , or indeed expected , a priori . An mQTL study tests variation in each metabolite for association with genome-wide genetic variation . As such a large number of tests is performed , effect sizes must be substantially larger to be reach statistical significance . Thus , as well as being potentially rarer , mQTLs are typically more difficult to detect than eQTLs of equivalent effect size . A number of recent studies have reported mQTLs for serum metabolite concentrations in humans [14] , [19] . Illig et al . [14] genotyped 1 , 809 individuals of Northern European ancestry at genome-wide single-nucleotide polymorphisms ( SNPs ) , and determined concentrations of 163 metabolites in serum samples from the same individuals , using the Biocrates platform ( targeted metabolomics using FIA-MS ) [18] . They went on to quantify association between each SNP and a derived set of 26 , 569 metabolic traits ( including 163 raw metabolite concentrations and all pair-wise metabolite concentration ratios ) . They discovered nine significant , replicable associations between metabolite concentration ratios and SNPs . We demonstrate in the current paper that their study [14] was well powered to detect mQTLs explaining approximately 3% or more of population variation in those serum metabolites targeted by Biocrates . In the current paper the effect size of an mQTL is defined to be the proportion of population variation in metabolite concentration that is explained by genetic variation at the corresponding mQTL SNP . The primary question addressed by our study is: ‘Are there 1H NMR-detectable metabolites in urine or plasma that are strongly influenced by common single-locus genetic variation ? ’ To this end , we performed an mQTL-discovery study using 1H NMR to analyse plasma and urine samples from multiple cohorts ( see Results and Materials and Methods ) . 1H NMR is an untargeted , discovery-driven approach that covers many important substances involved in major biochemical functions and key intermediary processes [16] . Our study demonstrates the existence of mQTLs of larger effect size than those reported in [14] for the untargeted set of metabolites detectable by 1H NMR , in urine as well as plasma ( urine is previously unexplored for mQTLs ) . The current paper's secondary aim was to provide further support for the findings of [14] . We conducted replication of the findings of [14] , using the Biocrates platform to assay our set of plasma samples . We replicated additional mQTLs , and characterized the familial component of biological variation in mQTL-driven metabolite levels , augmenting the mQTL-derived heritability .
We collected plasma and urine samples from participants across two cohorts—MolTWIN and MolOBB—as part of the MolPAGE programme . The MolTWIN cohort comprised 142 female twins of Northern European descent , who donated samples longitudinally . The MolOBB cohort comprised 69 participants in the Oxford Biobank ( OBB ) [20] . For all participants across both cohorts we acquired: 1H NMR spectra on plasma and urine samples; Biocrates-platform metabolite concentration data on plasma samples; and genome-wide SNP data ( see Materials and Methods ) . Analysis of a biological sample by 1H NMR provides a spectrum , which is comprised of the superimposed spectral profiles of individual metabolites; a metabolite's profile is made up of peaks from each chemically distinct hydrogen atom in the corresponding molecule . The peak position of a given hydrogen on the horizontal ( frequency ) axis is known as a chemical shift and is quoted in parts per million ( ppm , often termed a value ) from that of a reference substance . The concentration of each detectable hydrogen-containing metabolite can be inferred from the area under its total specific profile , or under a specific peak if the number of protons contributing to it is known . We preprocessed spectra , and extracted a total of 526 metabolite peaks from each pair of samples , i . e . the two samples ( plasma and urine ) donated by a participant on a visit to the clinic . These peaks represent fewer than 526 metabolites with some redundancy ( see Materials and Methods ) . Using data from the MolTWIN cohort , each of the 526 metabolite peaks was tested for association with 2 , 541 , 644 autosomal SNPs ( of which 2 , 245 , 627 were imputed and 296 , 017 were typed ) . In order to address both multiple testing and the kinship of twin pairs , we used a permutation-based procedure , constraining the genome-wide false-discovery probability to be less than 0 . 001 for each metabolite peak's genome-wide scan . We detected , and then replicated , four metabolites driven largely by SNP variation ( ) , across three genomic regions , explaining between 40%–64% of biological population variation in these four metabolites' concentrations . Genetic details of the hit regions are shown in Table 1 . Note that there are only three hit regions for the four metabolites because two metabolites mapped to a single , shared region . One of the mQTLs is in strong linkage disequilibrium ( LD , reviewed in [21] ) with SNP variation associated with renal function [22]–[23] . We found that two of the three mQTL regions exhibited genetic evidence of having experienced strong , recent positive selection in European populations ( further details of these findings are presented later , in dedicated sections in Results and Discussion ) . We proceeded to identify as many of the mQTL-driven metabolites as possible using a combination of: the web-based human metabolome database [24] , our in-house developed database , statistical total correlation analysis [25] , and other literature [26] . We unambiguously identified three out of four metabolites , and partially identified the fourth . The mQTL at chromosome 10q24 . 2 had two associated metabolites , identified as trimethylamine in urine ( TMAu ) , and dimethylamine in plasma ( DMAp ) . The mQTL at 5p13 . 2 affects urine concentration of 3-amino-isobutyrate ( a . k . a . β-amino-isobutyrate , denoted by BAIBu ) . The mQTL at 2p13 . 1 associates with concentrations of one or more urine metabolites that we partially identified as N-acetylated compound ( s ) : X . NH . CO . CH3 , with X unknown; we denote this set of one or more metabolites as N-ACu . We were unable to annotate N-ACu unambiguously despite conducting a number of additional experiments , including: seven experiments in which we spiked candidate compounds into selected urine samples and then re-measured the 1H NMR spectra; solid phase extraction experiments on urine samples in which we attempted to separate out N-ACu and thus aid its identification; and 2-dimensional 1H-13C heteronuclear single quantum coherence NMR spectroscopy experiments on selected urine samples . Table 2 summarizes these metabolite annotations . Figure S1 displays the three mQTL-driven urine metabolite peaks on the same scale , allowing visual assessment of their relative size . We went on to characterize more accurately each metabolite's associations with SNPs within 200 kb of the hit regions . We used a linear mixed-effects model to account for: the sharing of genes and environment across twins , the collection of multiple samples longitudinally from some subjects , and the technical replication of each biological sample ( see Materials and Methods ) . Under this model , we calculated p-values for the test of no association between the metabolite and each regional SNP in turn . Figure 1 , Figure 2 , Figures S2 and S3 display the p-values for all regional tests of association superimposed on patterns of LD and the positions of genes . The details of association of each metabolite with its most strongly associated SNP are listed in Table 3 , while Table S1 contains association results for SNPs within 200 kb of hit regions . The relationship between metabolite concentration and genotype is presented graphically in Figure 3 . For 1H NMR mQTLs , we estimated the proportion of biological variation in the metabolite's concentration explained by the corresponding mQTL SNP , and decomposed the remaining variation into familial , individual-environmental , and longitudinally fluctuating ( visit ) effects ( Figure 4 , Table 4 , and Materials and Methods ) . The familial component of variation modelled the combined effects of genome-wide identity-by-descent genetic sharing , and common environment ( i . e . environmental influences shared by twins after their conception ) . The individual-visit and common-visit components of variation modelled the longitudinal fluctuations between sample-donation visits that were respectively non-shared and shared by twins in a pair ( the common-visit effect was included in the model because twins visited the clinic in pairs ) . The proportions shown in Figure 4 and Table 4 are proportions of phenotypic variance after the experimental variance has been removed . It was useful to extract the experimental variance prior to comparison across platforms , as the primary focus was on the variability properties of the metabolite concentrations , not on the experimental variation associated with the measurement process . The mQTLs explained 40%–64% of biological population variation in the corresponding 1H NMR metabolite levels . We also performed a variance decomposition of the metabolic traits , quantified on the Biocrates platform , for which mQTLs were identified in [14] ( Figure 4 , Table 4 , and Materials and Methods ) . The Biocrates-platform mQTLs explained up to 35% of biological variation in the corresponding metabolic traits ( smaller effect sizes than for the 1H NMR mQTLs ) . Our results qualitatively extended the findings of [14]: the current study's design allowed the decomposition of the component of variation in metabolite concentration that was not explained by the mQTL itself ( see Discussion ) . To investigate potential bias in effect-size estimates ( the “winner's curse” phenomenon [27] ) , we compared effect-size estimates across discovery and replication studies , both for the Biocrates-platform mQTLs ( Figure S4 ) , and for the 1H NMR mQTLs ( Figure S5 ) . We found there to be a good degree of consistency in effect-size estimates between discovery and replication studies . Figure 5 relates the detectable effect size ( the proportion of variance in concentration explained by the mQTL SNP , quantified by ) to the sample size for each study ( power calculations used the GeneticsDesign R package ) . Our study had power to detect associations with approximately , while [14] had power to discover much smaller effects ( approximately ) . Better powered studies such as [14] have the potential to offer further interesting insights into the mQTL basis of the 1H NMR metabolome . We searched within 200 kb of each metabolite's hit region for SNPs previously associated with phenotypes in GWASs [28] . SNP rs13538 is in strong LD with the N-ACu hit region at chromosome 2p13 . 1 ( between rs13538 and rs9309473 in the HapMap 3 individuals of Northern European ancestry , i . e . HapMap-CEU [29] ) . Variation at rs13538 has been shown to correlate with serum creatinine concentration and other measures of renal impairment , as well as with susceptibility to chronic kidney disease [22]–[23] . Upon surveying the literature related to genes in the region of the N-ACu mQTL , we realized that several papers had highlighted this particular region as carrying one of the strongest signatures of selection that has been discovered in the human genome ( see , e . g . , [8] , [30] ) . This led us to check all known mQTLs for coincidence with positively selected regions ( as identified by the genome-wide scan for such regions in [8]; see also [31] for a review of the detection and relevance of the genetic signature of natural selection ) . We compared the locations of all mQTLs discussed in the current paper to the positively selected loci identified in [8] . Two of our three replicated mQTL hits were within such regions ( the mQTL for N-ACu , and the mQTL that affects both TMAu and DMAp ) . We also examined the genomic locations of each of Illig et al . 's 13 replicated mQTLs ( see dedicated section below on replication ) , and found none to be within positively selected regions as identified in [8] . SNPs significantly associated with TMAu and DMAp fall within a haplotype block of approximately 40 kb at chromosome 10q24 . 2 , which contains the PYROXD2 gene , a probable pyridine nucleotide-disulphide oxidoreductase gene , previously named C10orf33 ( see Figure 2 and Figure S2 ) . The most strongly associated SNP , rs7072216 , has alleles C/T at frequency 0 . 25/0 . 75 in Europe ( HapMap-CEU [29] ) . Our data indicate that TMAu concentration and DMAp concentration both increase with the number of copies of the major ( T ) allele . TMAu displays non-additivity , with the T allele recessive , and the TT homozygote class showing a greater-than-additive increase ( on logarithmic scale ) on the levels of the other two genotypic classes ( Figure 3 ) . There is a non-synonymous SNP—rs2147896—in strong LD with rs7072216 ( ; see also Table S2A ) . Functional predictions ( SIFT [32] and PolyPhen [33] ) and the PhyloP conservation score [34]–[35] for rs2147896 did not point to a clear functional impact , or to it being significantly conserved ( Table S2B ) . SNP rs2147896 does not lie in a known protein domain , and web-based protein structure-modelling tools [36]–[37] did not predict that the rs2147896 polymorphism would have an effect on PYROXD2's ligand binding site . However , PYROXD2 ( C10orf33 ) eQTLs have been discovered in fibroblasts ( rs2147897 [2] ) and liver ( rs2147901 [38] ) , with these eQTLs in high LD ( up to and respectively ) with mQTL SNPs of TMAu and DMAp ( Table S3 ) . This raised the possibility that eQTL-driven population variation in mRNA transcription at PYROXD2 mediates the mQTL of TMAu and DMAp . In order to investigate this eQTL hypothesis further , we extracted estimates of PYROXD2 mRNA abundance from two separate gene-expression microarray data sets measured on abdominal subcutaneous adipose tissue and whole-blood samples from the MolTWIN cohort ( Materials and Methods ) . We found that PYROXD2 was expressed in whole blood , but found no evidence of rs7072216 being an eQTL of PYROXD2 in whole blood ( ) , a finding consistent with [2] , in which PYROXD2 eQTLs were neither discovered in T cells nor in lymphoblastoid cell lines . However , we did find rs7072216 to be an eQTL of PYROXD2 in subcutaneous abdominal adipose tissue ( ) , with gene expression decreasing in the number of copies of the T allele . We plotted the mutual dependence between rs7072216 genotype , PYROXD2 gene expression in adipose tissue , and TMAu concentration ( Figure 6 ) . TMAu concentration was strongly negatively correlated with PYROXD2 expression ( Pearson's , ) . We examined these particular gene expression data ( i . e . measured in fat and blood cells ) because they had been acquired already on MolTWIN cohort members . In performing this analysis , we were not suggesting that variation in gene expression in fat has a direct impact on the concentration of TMAu or DMAp . However , a substantive proportion of eQTLs modulate expression in a similar way in different tissues [39] . Thus , in identifying and characterizing the mutual dependence of TMAu concentration , rs7072216 genotype , and PYROXD2 expression in a mechanistically unrelated tissue ( i . e . fat ) , we have raised the possibility that a qualitatively similar relationship with PYROXD2 expression will be observed in the tissue that truly mediates the mQTL effect ( likely to be liver or kidney ) . The SNPs that are significantly associated with BAIBu map to chromosome 5p13 . 2 within AGXT2 ( alanine-glyoxylate aminotransferase 2 ) . AGXT2 is known to be expressed in human liver and kidney . An eQTL for AGXT2 was reported in liver ( [38] and Table S3 ) , but this eQTL is not in LD with the mQTL SNPs ( ) , and so does not explain the BAIBu mQTL . Two of the most significant mQTL SNPs for BAIBu were rs37369 ( T/C at 0 . 09/0 . 91 ) and rs37370 ( C/T at 0 . 08/0 . 92 ) , with ( HapMap-CEU [29] ) between the two SNPs ( Table S2A ) . At SNP rs37370 , one of the MZ twin pairs in the study was homozygous for the minor C allele; these subjects had higher BAIBu concentration than those in the other genotypic classes . Each of rs37369 and rs37370 is a non-synonymous , missense coding mutation in AGXT2 , leading to an amino acid substitution in AGXT2 . At rs37369 , the base change C619T leads to the valine-to-isoleucine substitution V140I . At rs37370 , T506C leads to the asparagine-to-serine substitution N102S . At each SNP , the concentration of BAIBu increased in the number of copies of the minor allele . Both SNPs lie in the pyridoxal phosphate-dependent transferase major domain ( IPR015424 ) with rs37369 in subdomain 1 , and rs37370 in subdomain 2 . We extracted functional predictions ( SIFT [32] and PolyPhen [33] ) and PhyloP conservation scores [34]–[35] for rs37369 and rs37370 , but discovered no substantive evidence in favour of functional impact or of either SNP being significantly conserved ( Table S2B ) . We used the web servers Phyre2 [36] and 3DLigandSite [37] to predict AGXT2 protein structure and to investigate whether rs37369 and rs37370 were likely to affect AGXT2's predicted ligand binding site , but neither SNP was identified in these analyses as having an impact on the binding site . We analysed the 15 mQTL associations reported in Illig et al . [14] using SNP genotypes and Biocrates-platform data from the MolOBB and MolTWIN cohorts ( having removed individuals overlapping with the TwinsUK cohort used in [14] ) . We replicated 12 of the 15 mQTLs ( Table 5 ) , with four additional mQTLs replicated beyond the nine replicated by Illig et al . themselves ( so that now a total of 13 of the 15 mQTLs identified in [14] have been replicated ) . The same significance level was used as in the replication section of [14] , specifically a level of 0 . 05 adjusted by the Bonferroni method to account for 15 tests being performed ( i . e . an adjusted significance level of 0 . 0033 ) .
Genetic variation at PYROXD2 has experienced recent positive selection in European populations [8] , with the T allele of rs7072216 at frequency 0 . 75 in Europe ( HapMap-CEU ) , and at 0 . 14 in Africa ( HapMap 3 individuals from Yoruba in Ibadan , Nigeria , Africa , i . e . HapMap-YRI [29] ) . The haplotype that was relatively advantageous in European populations is associated with decreased expression of PYROXD2 and increased concentration of TMAu and DMAp . Further work will be necessary to clarify the mechanisms linking: DMAp and TMAu levels; PYROXD2 gene expression; and genetic variation in LD with rs7072216 ( such as the non-synonymous SNP , rs2147896 ) . The signature of selection at PYROXD2 is indirectly suggestive of biomedical relevance; we also note that the set of genes showing evidence for positive selection is enriched for genes involved in oxidoreductase activity [8] . There have been a number of studies that have examined the sources of variation in physiological concentrations of methylamines and their derivatives , e . g . [45]–[46] . The current paper sheds light on this field from a new genetic angle , and it will be useful to integrate the mQTL effects into known pathways . Gut microbiota play an important role in the formation of methylamines from dietary sources in mammals—they create TMA from choline , and convert TMA into DMA [45]–[46] . Trimethylamine N-oxide ( TMAO ) is formed endogenously in the liver via the N-oxygenation of TMA by the flavin-containing monooxygenase ( FMO ) protein family , and particularly by FMO3 [47] . Gut microbial activity has been linked to disease through physiological levels of DMA , TMA and TMAO [48]–[49] . It may prove productive to relate the TMAu mQTL finding to the rare recessive genetic disorder trimethylaminuria , in which mutations at FMO3 disrupt conversion of TMA to TMAO , resulting in high physiological levels of TMA and an accompanying fish-odour phenotype [47] . Trimethylaminuria cases exhibit relatively low values of the ratio TMAOu/ ( TMAOu + TMAu ) , where TMAOu denotes urine TMAO concentration . Subjects in the current study have values of this ratio that are within the range typical of trimethylaminuria controls ( Figure S6 and [50] ) . It will be interesting to investigate the effect , if any , of genetic variation at the TMAu mQTL on TMA levels among trimethylaminuria cases . The N-ACu mQTL lies within a large 500 kb haplotype block ( Figure 1 ) , and there are a number of genes ( and eQTLs , Table S3 ) in LD with it . Of these genes , NAT8 is a likely candidate for mediating the association between SNP variation and N-ACu ( urine concentration of N-acetylated compound ( s ) ) , since NAT8's encoded enzyme specifically catalyzes N-acetylation—NAT8's enzyme is cysteinyl-conjugate N-acetyltransferase , CCNAT [51] . We relate our N-ACu mQTL finding to other research that has shown: ( i ) that the region harbours SNPs associated with renal function [22]–[23] , [52]; and ( ii ) that the region has been the site of positive selection on standing genetic variation [8] , [30] . Two recent renal-function GWASs identified rs13538 as a clinically associated SNP [22]–[23] , [52] with the minor G allele increasing susceptibility to renal dysfunction . Chambers et al . [22] proposed that a non-synonymous mutation in NAT8 ( the A595G change at rs13538 , producing a non-conservative amino acid change F143S in CCNAT ) reduces acetylation efficiency , thus leading to toxin-induced kidney injury . The N-ACu mQTL SNP rs9309473 is in strong LD ( ) with GWAS SNP rs13538 ( both SNPs with alleles A/G at frequency 0 . 79/0 . 21 in HapMap-CEU [29] ) . We found the non-synonymous mutant allele ( G ) at rs13538 to be associated with increased levels of N-ACu . Thus , whilst our findings provide evidence of differential acetylation efficiency driven by genetic variation in LD with rs13538 , their directionality is not consistent with the specific mode of action proposed in [22] . Furthermore , a recent functional study [51] found enzymatic activity of mutant ( F143S ) CCNAT to be comparable to that of the wild-type protein ( and so is also inconsistent with the mode of action proposed in [22] ) . Scheinfeldt et al . [30] studied the signature of selection in this region , specifically examining two complementary sets of haplotypes: the “ancestral” and “derived” haplogroups ( HapA/HapD respectively , at frequency 0 . 26/0 . 74 in the HapMap-CEU European population , but at 0 . 89/0 . 11 in the HapMap-YRI African population [29] ) . It has been proposed that positive selection drove up the frequency of HapD ( relative to HapA ) in Eurasian populations about 15 , 000 years ago [30] . The N-ACu mQTL SNP rs9309473 is in strong LD with HapA/HapD status ( in HapMap-CEU , with alleles G/A of rs9309473 highly predictive of HapA/HapD status respectively ) . An increasing number of copies of HapA is associated with increased urine concentration of N-acetylated compound ( s ) ( N-ACu ) , and with increased susceptibility to renal dysfunction [22]–[23] . The BAIBu mQTL was also identified by Suhre et al . [44] , where they noted the following . Elevated levels of BAIBu had been shown through family studies to be autosomal recessive [53] , but the causal locus had been previously unknown . The association of a SNP in AGXT2 with BAIBu levels is consistent with the role of AGXT2's encoded enzyme , mitochondrial aminotransferase , which is expressed primarily in the kidney and catalyzes the reaction of BAIB with pyruvate to form 2-methyl-3-oxopropanoate and alanine ( EC 2 . 6 . 1 . 40 ) . It had also been previously suggested that altered BAIB homeostasis might contribute to hyper-β-amino-isobutyric aciduria , a relatively common Mendelian metabolic disorder in humans [54] . Suhre et al . [44] proposed rs37369 as a likely candidate for the causative SNP driving both variation in BAIBu concentration and susceptibility to hyper-β-amino-isobutyric aciduria . We found the non-synonymous SNP rs37369 ( ) to be marginally more significantly associated with BAIBu concentration than the other non-synonymous SNP , rs37370 ( ) . This mildly supports rs37369 as the causal SNP driving BAIBu levels , relative to rs37370 , though the true causal genetic polymorphism may be neither of these SNPs , but instead variation in strong LD with them . We used existing tools to predict the effect of rs37369 and rs37370 polymorphism on AGXT2 function ( see Results ) , but this analysis did not reveal any clear functional consequences of these non-synonymous polymorphisms . Further work will be necessary to characterize with certainty the causal link between genetic variation at the AGXT2 locus and BAIBu concentration . In conclusion , we have designed and conducted an mQTL study of plasma and urine metabolites detectable by 1H NMR . We discovered and replicated four novel metabolite-SNP associations , with each SNP explaining 40% or more of biological variation in metabolite concentrations . The mQTLs that we discovered have interesting properties: two of the three mQTL regions have experienced recent positive selection in European populations; one mQTL is in strong LD with a SNP identified in a kidney-function GWAS . Our findings pave the way forward for investigating the potential biomedical relevance of these regions .
The MolTWIN study was approved by St . Thomas' Hospital Research Ethics Committee ( EC04/015 Twins UK ) . The MolOBB study received ethical approval from Oxfordshire REC C ( 08/H0606/107 ) . The 142 participants in the current study were recruited from the UK Adult Twin registry at St . Thomas' Hospital ( www . twinsUK . ac . uk ) : a longitudinal epidemiological study of 11 , 000 twins ( mostly female ) , for which extensive clinical , anthropometric , lifestyle , and demographic information , and a wide range of biological measurements have been collected [55] . Eligible volunteers were healthy , Caucasian , post-menopausal females of Northern European descent , between 45–76 years of age . Eligible twins were sent an information sheet containing details of the study , and two consent forms . After each twin had returned a completed consent form , she was contacted by letter and phone to book her appointment . The composition of the cohort was: 51 MZ pairs , 19 DZ pairs , and two unrelated individuals . In the MolTWIN cohort , 33 of the MZ twin pairs donated samples twice; the median inter-visit time across all such pairs was 118 days ( IQR: 96-134 ) . Both twins in a pair always visited on the same day , and each visit was scheduled at either 10:00 or 14:00 ( with repeated visits of each individual not necessarily scheduled at the same time of day ) . The 69 participants in the current study were selected from the Oxford Biobank [20] ( OBB ) . Specific OBB cohort members were selected on the basis of case/control status for metabolic syndrome according to International Diabetes Foundation Criteria [56] . The set of subjects comprised 42 controls ( 17 female , 25 male ) , and 27 cases ( 12 female , 15 male ) . Fasting blood and urine samples were collected at all clinic visits of each participant . Spot urine samples were centrifuged ( 16060 × g ) at 4°C for 10 min before being stored at −80°C . Fresh blood was collected in a 9 mL tube through venepuncture . Samples for 1H NMR analysis were collected in heparin tubes , whilst samples for Biocrates-platform analysis were collected in EDTA tubes . Blood samples were kept on ice for 20 min prior to centrifugation ( 16060 × g ) at 4°C for 10 min , and subsequent storage at −80°C . DNA was extracted from whole-blood samples using GeneCatcher ( Invitrogen Life Technologies , Carlsbad , USA ) according to manufacturer's protocol . Genome-wide SNP genotypes were measured on a total of 166 individuals: 70 from the MolOBB cohort , and 96 from the MolTWIN cohort ( one MZ twin from each MZ pair was genotyped , whilst both members of each DZ twin pair were genotyped ) . The genotyping assay used was the Illumina 317K BeadChip SNP array ( Illumina , San Diego , USA ) . Quality control on the genotyped subjects was performed in a way similar to those described previously by the Wellcome Trust Case Control Consortium [57] . Two MolTWIN samples were removed due to sample genotyping success rate < 95% and three samples ( two from MolTWIN , one from MolOBB ) were removed due to non-European ancestry ( note that the cohort compositions given in the Participant Recruitment sections are after quality control ) . SNPs were removed ( i ) if MAF < 1% , or ( ii ) if genotyping success rate <95% and MAF > 5% , or ( iii ) if genotyping success rate <99% and MAF < 5% . Hardy-Weinberg equilibrium ( HWE ) was calculated by combining all unrelateds of the MolOBB and MolTWIN data sets ( i . e . one twin per twin pair ) and the hypothesis of HWE was tested at a significance level of 10−4; SNPs at which HWE was rejected were omitted from the study . After quality control , the genotypes of ungenotyped MZ twins were copied from their corresponding genotyped twin . The final data set prior to imputation comprised 69 MolOBB members and 142 MolTWIN members genotyped at 296 , 017 autosomal SNPs . Measured genotypes were used to impute an additional 2 , 245 , 627 SNPs using the HapMap-CEU population ( release 22 ) as reference [29] . The imputations were performed using IMPUTE [58] . We included SNPs in our analysis only if the imputation quality score was greater than 0 . 4 . As output for a single SNP in an individual , IMPUTE provided probabilities of the individual having each of three possible genotypes ( zero , one , or two copies of the reference allele ) . Prior to incorporating imputed genotypes into the statistical models , we preprocessed them , estimating the true genotype by that which was allocated highest probability by IMPUTE . Including both typed and imputed SNPs , we used a total of 2 , 541 , 644 autosomal SNPs for association analysis . Total RNA was extracted from adipose tissue biopsies with TRIreagent ( SIGMA-ALDRICH , Gillingham , UK ) and quantified using a NanoDrop . For whole-blood samples , PAXgene tubes were used , and RNA was extracted according to the manufacturer's protocol ( PAXgene , QIAGEN ) . RNA was labelled using the MessageAmp II 96-well amplification kit ( Applied Biosystems , CA , USA ) . Labelled RNA was hybridized onto Affymetrix HGU133 Plus2 arrays , washed , stained , and scanned for fluorescence intensity according to manufacturers protocols ( Affymetrix , Inc . , USA ) . Data were preprocessed using the RMA method without background correction ( i . e . quantile normalization followed by robust probe-set summarization ) [59] . Whole-blood array data were preprocessed separately from adipose-tissue array data . Publicly available custom chip-definition files ( CDFs ) were downloaded ( version 11 ) ( http://brainarray . mbni . med . umich . edu/Brainarray/Database/CustomCDF/CDF_download . asp ) and used to group probes into sets , each set corresponding to an Ensembl-annotated gene , resulting in 18 , 394 such genes represented in the array data . See [60] for a description of how these CDFs were created , along with a comparison of their properties with the CDFs produced by Affymetrix . Expression data were extracted at the PYROXD2 gene , and used in the current paper's analysis of the mQTL for TMAu and DMAp . EDTA plasma samples were vortexed after thawing and centrifuged at 4°C for 5 min at 10 , 000 x g prior to loading of 10 µL of supernatants onto the 96-well kit plate . Processing of the AbsoluteIDQ kit followed the protocol specified by the manufacturer , including the following automated steps on the Hamilton ML Star robotics platform ( Hamilton Bonaduz AG , Bonaduz , Switzerland ) : ( i ) drying plasma samples under a nitrogen stream , ( ii ) derivatization of amino acids with 5% phenylisothiocyanate reagent ( 20 µL ) , ( iii ) drying of samples , ( iv ) extraction of metabolites and kit internal standards with mM ammonium acetate in methanol ( 300 µL ) , ( v ) centrifugation through filter plate ( 2 min , 500 x g ) , vi ) dilution with 600 µL MS running solvent . 20 µL of the final extracts were applied to flow injection analysis mass spectrometry . Samples were analyzed using an API 4000 triple quadrupole mass spectrometer ( ABSciex ) equipped with an Agilent 1200 Series HPLC and a HTC PAL auto sampler from CTC controlled by the software Analyst 1 . 5 . The standard flow injection method comprising two 20 µL injections ( one for positive and one for negative electrospray ionisation mode ) was applied for all measurements . Quantification was achieved by multiple reaction monitoring detection in combination with the use of stable isotope-labelled and other internal standards [61] . Data evaluation for quantification of metabolite concentrations was performed with the MetIQ software package ( integral part of the AbsoluteIDQ kit ) . Concentrations of all metabolites are initially calculated in µM . The method has been proven to conform to FDA-Guidelines [62] , which imply proof of reproducibility within a given error range . Analytical specifications for detection limit ( LOD ) and evaluated quantification ranges , further LOD for semi-quantitative measurements , identities of quantitative and semi-quantitative metabolites , specificity , potential interferences , linearity , precision and accuracy , reproducibility and stability were described in Biocrates manual AS-P150 . The LODs were set to three times the values of zero samples . The lower and upper limits of quantification were determined experimentally by Biocrates AG ( Innsbruck , Austria ) . In addition , the technical variability of the Biocrates platform had been quantified previously by Illig et al . [14] . Their Supplementary Table 4 displayed the coefficient of variation , CV , for each of 163 metabolite concentrations assayed in [14] , and measured under the same conditions on the same platform in the current study . The median CV across metabolites was 7 . 4% ( IQR: 6 . 1%-12 . 4% ) [14] , which demonstrated a useful degree of precision for the majority of metabolites . We performed quality-control checks , including boxplots and principal-component score plots , on the Biocrates-platform data to identify failed assays , where an assay refers to the measurement of 163 metabolite concentrations in a biological sample . Of a total of 356 assays across the MolOBB and MolTWIN cohorts , we identified two assays that exhibited anomalously low concentrations of all metabolites ( relative to the levels observed in the other assays ) ; we omitted those two assays from further analysis . Thawed samples were centrifuged at 16060 × g for 10 min . Samples were aliquotted into two technical replicates prior to sample preparation . Plasma was diluted 1 in 4 in physiological saline prepared in 20% D2O supplemented with 0 . 1% ( w/v ) sodium azide as a bacteriostatic agent and 1 . 5 mM sodium formate as a chemical-shift reference ( δ8 . 452 ) . Urine was diluted 2 in 1 in phosphate buffer ( 20% D2O , pH 7 . 4 ) supplemented with 1 mM trimethylsilyl-2 , 2 , 3 , 3-tetradeuteropropionic acid ( TSP; δ0 . 00 ) and 0 . 1% ( w/v ) sodium azide . Sample aliquots were allocated to 96-well plates ( and wells thereon ) in a randomized design . Each experiment was acquired on a Bruker DRX 600 MHz spectrometer ( Rheinstetten , Germany ) operating at 600 MHz ( for 1H ) using a 5 mm TXI flow-injection probe equipped with a z-gradient coil , at 300 K , at a spectral width of 12019 Hz , with 96 transients being collected with 8 dummy scans using 64k time domain data points . For both plasma and urine samples a standard 1D spectrum [RD−90°−3 µs−90°−tm−90°−acquire] with selective irradiation of the water resonance during the relaxation delay ( RD , 2 s ) and during the mixing time ( tm , 0 . 1 s ) was acquired . Additionally , for the plasma samples , a spin-echo ( Carr-Purcell-Meiboom-Gill ) spectrum [RD−90°− ( τ/2−180°−τ/2 ) n−acquire] with a total echo time of 608 ms ( n = 304 , τ = 2000 µs ) and a diffusion-edited spectrum made using a bipolar pulse-pair longitudinal eddy current delay pulse sequence with spoil gradients immediately following the 90° pulses after the bipolar gradient pulse pairs were acquired . Continuous wave irradiation was applied during the relaxation delay at the frequency of the water ( or HOD ) resonance . Eddy current recovery time ( Te ) was 5 ms , and the time interval between the bipolar gradients ( | ) was 0 . 5 ms . Further details may be found in [15] , [26] , [63] . Each of four data sets was passed independently through a semi-automated preprocessing pipeline: phasing , alignment , denoising , baseline correction , manual bin selection , normalization , quality control , peak extraction , and logarithmic transformation . Spectra were phased using in-house software ( NMRProc , T . M . D Ebbels and H . C . Keun , Imperial College London ) . All other data analysis was performed in R [64] . Spectra were zero-filled to 216 points . Urine spectra were aligned to TSP , set at δ0 . 00; plasma spectra were aligned to formate , set at δ8 . 452 ( peak centres were defined by the position of the local maximum ) . The spectra were denoised in the frequency domain using wavelet-based methodology ( a method similar to that described in [65] ) . For baseline correction , we initially fitted a constant baseline to each spectrum; however , visual inspection revealed that , for a number of spectra , the fit was better on one side of the water peak than on the other; natural variations in ionic strength resulting in altered phase of the residual water resonance may contribute to such an effect . Hence , a two-piece piecewise-constant baseline was fitted to and subtracted from each spectrum; specifically , the baseline on each side of the water peak was estimated by the 5th percentile of the spectral points in the corresponding interval ( a robust estimator of baseline location ) . We plotted each peak; for those peaks that visually displayed consistent presence across spectra , we manually created a bin and used the bin to extract the peak's data across all spectra . The datum extracted from a bin in a spectrum was the intensity of the highest local maximum ( i . e . we used peak height as a proxy for peak area ) , or was coded as a missing value if no local maximum was present . We chose peak height to be the estimator of concentration as , in addition to its simplicity , it had relatively good robustness properties in the context of spectral artefacts ( e . g . when a peak's location varied across spectra , or when neighbouring peaks overlapped within spectra ) . If the width ( at half height ) of a peak varies substantially across spectra , then peak height may be less precise than area at quantifying concentration . Plots of peaks did not reveal substantial peak-width variation in our data sets . Only common peaks—present in at least 80% of spectra in their corresponding data set—were included in downstream statistical analysis , and only a peak's non-missing data were included at the statistical modelling stage . A missing datum , corresponding to there being no local maximum in the peak's ppm interval , typically occurred for one of two reasons: ( a ) the corresponding metabolite's concentration was too low to create a local maximum , or ( b ) a relatively large neighbouring peak overlapped the peak of interest ( i . e . the missing concentration is censored , but not necessarily low ) . The omission of type ( a ) missing values from the analysis potentially decreased statistical power to detect mQTLs driving metabolite concentration variation at levels near or below the level of detection . The omission of type ( b ) missing values from the analysis increased the robustness of inference ( and conserved power ) in the face of artefactual effects of overlapping peaks . To illustrate , in Figure S7 we plotted the seven spectra ( out of 432 ) with missing values for DMAp , and the four spectra ( out of 432 ) with missing values for N-ACu . ( There were no missing values for TMAu and BAIBu . ) At the DMAp peak , missing data were representative of relatively low concentrations , approximately within the lowest quartile of observed concentrations ( so we may have lost a small amount of power through missing-data handling ) . For N-ACu's missing data , the relevant peak's size was obscured by signal from an overlapping peak ( missing values did not necessarily correspond to near-zero concentration ) . Prior to model fitting , we discarded any peaks that were annotated to exogenous metabolites ( of ibuprofen or acetaminophen ) , to a spike-in compound ( TSP in urine , formate in plasma ) , or to urea ( the area of which is affected by water peak saturation irradiation through chemical transfer of saturated protons ) . Across the three plasma data sets , 104 peaks were annotated to glucose; we discarded all but one representative glucose peak in each plasma data set . The spectra were normalized using probabilistic quotient normalization [66] . The normalization was performed using data from the retained peaks only; spectra were normalized to a reference spectrum comprising median peak heights; missing values were excluded from the calculation of medians . After quality control , urine spectra were available for 142 MolTWIN participants and 67 MolOBB participants; plasma spectra were available for 140 MolTWIN participants and 68 MolOBB participants . A logarithmic transformation was applied to make the peak height distributions more symmetric–the entire spectrum-wide set of peak heights were collectively shifted and scaled to lie between zero and 100 and then transformed . We tested each metabolite peak in turn for association with 2 , 541 , 644 autosomal SNPs . For this stage we averaged and transformed the peak data as follows: ( i ) we averaged each subject's metabolite peak data across all biological and technical replicates; ( ii ) for robustness , we mapped the quantiles of the resulting inter-subject distribution to the quantiles of a standard Gaussian distribution . We denote the resulting data vector by . We fitted the following additive genetic model by ordinary least-squares regression at each SNP:where indexed subject; was the number of copies of the reference allele possessed by individual ; and was the residual error term . At each SNP , we calculated the conventional t-statistic for the test of the null hypothesis . We then took the maximum absolute t-statistic observed across all SNPs tested , and this statistic , , was the test statistic used for testing the null hypothesis , : the metabolite's concentration was not associated with variation at any SNP in the genome-wide panel . We characterized the ( metabolite peak-specific ) null distribution of by permutation . For each of 5 , 000 permutations , we randomly reassigned the measured metabolite levels of each MZ pair to a different MZ pair , and randomly reassigned the measured metabolite levels of each DZ pair to a different DZ pair , yielding for the th permutation . Such a permutation crucially preserved the existing covariance structure on induced by polygenic genetic relatedness ( identity-by-descent sharing ) and common-environmental effects between twins , while breaking down any existing associations between and identity-by-state variation at SNPs . For the th permutation , we calculated t-statistics as before , quantifying the additive genetic association between and the genotypes at each SNP . We then calculated the maximum absolute t-statistic across SNPs , yielding the th draw from the null distribution , . For each metabolite , we rejected only if the observed test statistic exceeded all 5 , 000 draws from its null distribution , i . e . if . Such a procedure constrained ( to be small ) the family-wise error rate ( FWER ) for testing a single metabolite against genome-wide SNP variation . Specifically , ( 0 , 0 . 0007 ) was an exact 95% confidence interval for the FWER , based on the observation that none of the 5 , 000 draws from the null distribution of exceeded the observed statistic [67] . We concluded that our testing procedure controlled the false-positive probability for each metabolite's entire genome-wide scan to be less than 0 . 001 . was rejected for six of the 526 metabolite peaks tested . These six peaks redundantly represented four metabolites , listed in Table 2 . For each metabolite , we examined the subset of SNPs that reached genome-wide significance ( defined as those SNPs whose t-statistics exceeded , in absolute value , the metabolite's maximum null test statistic , ; shown in Table S1 ) . For each metabolite , the set of genome-wide significant SNPs co-localized to a single genomic region; we defined a metabolite's hit region to be the smallest contiguous region containing all genome-wide significant SNPs . Underlying the observed data at a metabolite peak ( i . e . across all spectra ) was a complex correlation structure , induced by the sharing of alleles , individuals , and samples by different sample aliquots . In the follow-up analysis of hit regions we explicitly modelled this covariance structure while quantifying the metabolite's association with each local SNP in turn ( i . e . with each SNP within 200 kb of the hit region and with MAF > 5% ) . To deal with potential deviations from the Gaussian distributional assumptions , we mapped the quantiles of the empirical data distribution at each peak to the quantiles of a standard Gaussian distribution , yielding the transformed data vector , . In contrast to the genome-wide analysis described in the previous section ( based on the averaged data , ) , technical and biological replicates were not averaged for this analysis ( instead , variation between replicates was retained in and modelled ) . We fitted the following mixed-effects model: where twin pairs were indexed by , the twins within a pair were indexed by , the visits of a twin pair were indexed by , and the aliquots of a sample were indexed by . The ‘fixed effects’ in the model were , the , and . The additive effect of the SNP under consideration was modelled by , with denoting the number of copies of the reference allele possessed by twin in pair . The parameters controlled for experimental inter-plate effects , with mapping spectra to plates . The parameter controlled for sampling time-related effects , with in the equation above mapping visits to sample-collection times ( in 24-hour format; times were mostly 10 or 14 ) . The other terms in the model were ‘random effects , ’ which modelled the covariance structure across observations induced by familial , individual-environmental , temporally dynamic , and non-biological effects . Similarly to [68] , there was one term for each MZ pair and two such terms , and , for each DZ pair ( i . e . if was an MZ pair , whilst if was a DZ pair ) . Each ‘random effect’ followed a zero-mean Gaussian distribution with its corresponding standard deviation from ( e . g . the independently followed ) . The current paper's model induced a covariance structure on that was identical to that which is used in the standard methodology for modelling twin data ( see , e . g . , [68]–[70] ) . In the parameterization above , modelled the familial variance ( i . e . the variance attributable to genetics and common environment ) , modelled the individual-environmental variance . Additionally , our model included variance parameters representing longitudinally unstable variation . These ( and ) were referred to as the ‘common-visit’ and ‘individual-visit’ effects respectively , because they measured the component of phenotypic variation that fluctuated between visits , and which was shared and non-shared respectively between twins; the common-visit parameterization was included in the model because twins visited the clinic in pairs . Finally , there was a parameter to model experimental variation . In the variance decompositions of the current paper ( Figure 4 and Table 4 ) , variances were expressed as proportions of the total biological variance , which was defined as , where was the phenotypic variance explained by the corresponding mQTL SNP . The biological variance did not include the experimental variance , , and was therefore appropriate for comparing the properties of molecular phenotypes across platforms when the level of experimental variation on the platforms was not of primary interest . For each SNP within 200 kb of the hit region , we fitted the mixed-effects model both with and without the term . From these fitted models , we calculated the p-value for the test of the null hypothesis that , using as a test statistic ( where denotes the likelihood ratio ) , and employing its asymptotic null distribution ( a chi-squared density with one degree of freedom ) . These p-values are displayed in the text , Figure 1 , Figure 2 , Figures S2 and S3 , and Table 3 , Tables S1 and S2A . At the most strongly associated SNP , we went on to fit the model in a Bayesian framework , quantifying the precision of parameter estimates using posterior credible intervals . For this analysis we used directly the log-transformed metabolite concentrations , denoted by ( see section on Data preprocessing and feature extraction ) . For priors , we specified Uniform densities on the standard deviation parameters in ( as discussed in [71] ) :where denotes the sample standard deviation of the data , . The prior on the ‘fixed effects’ vector , , was a diffuse multivariate Gaussian distribution , with mean at the least squares estimates , , and diagonal covariance matrix with entries . The results of fitting the model in a Bayesian framework are summarized in Table 4 and Figure 4 . For each of the mQTLs discovered in the MolTWIN cohort , we re-tested the association using only data from the MolOBB cohort . Specifically , we mapped the quantiles of the metabolite's concentration data to the corresponding quantiles of a standard Gaussian distribution; we then tested for an additive association with the corresponding SNP's genotype data , including age and gender as covariates in the linear model . Resulting p-values are shown in Table 3 . We used the concentration data directly as output from the Biocrates platform , and calculated metabolic traits from concentration ratios as in [14] . We removed individuals overlapping with the TwinsUK cohort used in [14] , after which a total of 202 individuals were included in our Biocrates replication analysis ( 133 MolTWIN participants and 69 MolOBB participants ) . We fitted similar models to those specified in the Materials and Methods subsection ‘Mixed-effects analysis of hit regions , ’ though now with fixed effects for genotype ( number of copies of reference allele ) , plate , age , and gender . For each metabolic trait , the genotype data in the model was from the single corresponding mQTL SNP as reported in [14] . The results of the non-Bayesian analysis are shown in Table 5 and the results of the Bayesian analysis are in Table 4 and Figure 4 . The data underlying the current paper's analyses are available for download from an FTP server ( host: svilpaste . mii . lu . lv; login: Moltwin_NMR; password: Moltwin_NMR1; path: /home/George/PLoS_Genetics_mQTL_data ) . | Physiological concentrations of metabolites—small molecules involved in biochemical processes in living systems—can be measured and used to diagnose and predict disease states . A common goal is to detect and clinically exploit statistical differences in metabolite concentrations between diseased and healthy individuals . As a basis for the design and interpretation of case-control studies , it is useful to have a characterization of metabolic diversity amongst healthy individuals , some of which stems from inter-individual genetic variation . When a single genetic locus has a sufficiently strong effect on metabolism , its genomic position can be determined by collecting metabolite concentration data and genome-wide genotype data on a set of individuals and searching for associations between the two data sets—a so-called metabolite quantitative trait locus ( mQTL ) study . By so tracing mQTLs , we can identify the genetic drivers of metabolism , characterize how the nature or quantity of the corresponding expressed protein ( s ) feeds forward to influence metabolite levels , and specify disease-predictive models that incorporate mutual dependence amongst genetics , environment , and metabolism . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"mathematics",
"epidemiology",
"statistics",
"genetics",
"chemistry",
"biology",
"genomics",
"biostatistics",
"genetics",
"and",
"genomics",
"genetic",
"epidemiology"
] | 2011 | A Genome-Wide Metabolic QTL Analysis in Europeans Implicates Two Loci Shaped by Recent Positive Selection |
Hybridization between species is an important mechanism for the origin of novel lineages and adaptation to new environments . Increased allelic variation and modification of the transcriptional network are the two recognized forces currently deemed to be responsible for the phenotypic properties seen in hybrids . However , since the majority of the biological functions in a cell are carried out by protein complexes , inter-specific protein assemblies therefore represent another important source of natural variation upon which evolutionary forces can act . Here we studied the composition of six protein complexes in two different Saccharomyces “sensu stricto” hybrids , to understand whether chimeric interactions can be freely formed in the cell in spite of species-specific co-evolutionary forces , and whether the different types of complexes cause a change in hybrid fitness . The protein assemblies were isolated from the hybrids via affinity chromatography and identified via mass spectrometry . We found evidence of spontaneous chimericity for four of the six protein assemblies tested and we showed that different types of complexes can cause a variety of phenotypes in selected environments . In the case of TRP2/TRP3 complex , the effect of such chimeric formation resulted in the fitness advantage of the hybrid in an environment lacking tryptophan , while only one type of parental combination of the MBF complex allowed the hybrid to grow under respiratory conditions . These phenotypes were dependent on both genetic and environmental backgrounds . This study provides empirical evidence that chimeric protein complexes can freely assemble in cells and reveals a new mechanism to generate phenotypic novelty and plasticity in hybrids to complement the genomic innovation resulting from gene duplication . The ability to exchange orthologous members has also important implications for the adaptation and subsequent genome evolution of the hybrids in terms of pattern of gene loss .
The Saccharomyces sensu stricto yeasts represent a diverse , monophyletic group of species that have the ability to produce viable and stable hybrids that can propagate mitotically . Hybrids among yeast species and strains seem to be common , especially amongst wine , and beer brewing yeasts [1] , [2] , but also within natural ecological niches [3] . When two parental genomes merge in yeast hybrids there is a potential for genetic novelty but also for a genetic conflict to occur . Dominant genetic incompatibilities do not seem to occur in the S . cerevisiae sensu stricto group [4] , however evidence of recessive allelic incompatibilities between nuclear and mitochondrial genomes have recently been uncovered [5] . Hybridization can play an important role in evolution since hybrids could occupy a different niche from both parental species and eventually establish a new lineage . The presence of naturally occurring yeast hybrids isolated from specific environments seem to confirms this hypothesis [6] , [7] . So far , many unique characteristics of the Saccharomyces “sensu stricto” species and hybrids have been attributed to changes in gene expression , including novel cis-trans interactions [8] and to divergence in regulatory regions [9] . Nevertheless , in the hybrid cellular environment , where two sets of homologous proteomes coexist , there is also the potential for the cell to form chimeric assemblies between homologus protein complexes . Analysis of large-scale proteomics data has shown that the majority of cellular processes are carried out by protein assemblies rather than single proteins and that over 60% of yeast proteins form obligate complexes [10] . Since the correct formation of a complex is essential to carry out the biological function , we would expect that any sub-optimal protein interaction would be detrimental to the cell and therefore discouraged by the cell . On the other hand , spontaneous chimeric assemblies may widen the adaptation potential of the cell , since several different combinations of the same protein complex can be used . Therefore , such situation can lead to new phenotypic variants that are beneficial to the hybrid in novel contexts . The primary aim of this work is to establish proof of principle that chimeric protein complexes can form freely in hybrids of Saccharomyces species despite the intra-specific co-evolutionary forces and to quantify the impact that such complexes can have on the overall fitness of the hybrids . In fact , chimericity in protein-protein interaction represents a potentially important mechanism for generating phenotypic diversity upon which evolutionary forces can act , and may constitute a molecular explanation of hybrid vigour .
To test for the existence of natural chimeric complexes in yeast hybrids , we analysed six physically stable ‘obligatory’ protein complexes ( Table S1 ) each of which have constitutively expressed members that were previously recovered by large-scale protein interaction studies and also by independent small-scale biochemical studies [11] , [12] . We created S cerevisiae/S . mikatae ( Sc/Sm ) and S . cerevisiae/S . uvarum ( Sc/Su ) hybrids by crossing either S . mikatae or S . uvarum with S . cerevisiae strains carrying a molecular tag ( TAP-tag ) at the C-terminus of a selected member of the protein complex ( Figure S1 ) . Tagged proteins , along with their interacting partners , were isolated via affinity chromatography and all the members of the protein complex were identified via mass spectrometry . If only species-specific parental complexes are established in the hybrid , just proteins from the species carrying the TAP-tag ( S . cerevisiae ) will be identified . However , if chimeric protein complexes are formed , proteins from the other parental species ( S . mikatae or S . uvarum ) will also be isolated and identified ( Figure 1 ) . The protein fractions were analyzed by mass spectrometry to identify tryptic peptides in a custom protein database of six Saccharomyces sensu stricto yeast proteomes . Species-specific peptides were distinguished from the shared peptides that are identical between the two parental species . As control experiment to test whether in vitro chimeric interactions were generated artefactually during the protein extraction procedure ( as opposed to in vivo within the hybrid cellular environment ) , a mixture of parental cells ( i . e . S . cerevisiae and S . mikatae or S . uvarum ) were grown separately and mixed together just prior to cell lysis . To establish that both parental genomes were present , all hybrids were screened for chromosomal content via PCR using species-specific primers ( Figure S2 ) . To check for genomic alterations after hybridisation , meiosis was induced and spore viability was assessed . Hybrids between yeast species are sterile ( <1% survival rate ) but they can present a higher rate of spore viability if the cells undergo aneuploidy incrementing their chromosomes number . After dissecting 128 tetrads per hybrid background , no viable cells were detected ( Figure S3 ) , suggesting that the hybrids were 2n . Transcription of the homologous members of the protein complexes in the hybrids was also confirmed via RT-PCR ( Figures S4 , S5 , S6 , S7 , S8 , S9 ) . The first complex we considered was the Sec 62/63 complex , a tetramer that is involved in the transport of proteins across the ER membrane , composed of two essential proteins , Sec62p and Sec63p and two non-essential proteins , Sec66p and Sec72p [13] . In both hybrids Sc/Sm and Sc/Su , the mass spectrometry analysis identified Sec63p and Sec72p from either S . mikatae or S . uvarum , respectively , demonstrating that in yeast hybrids the assembly of the Sec62/63 complex can be spontaneously chimeric ( Figure 2 , Figure S10 , S11 , S12 , S13 , S14 , S15 , S16 , S17 , Table S2 and S3 ) . Evidence of chimeric interactions were also detected between members of the TRP2/TRP3 complex , involved in the tryptophan biosynthesis [14] ( Figures S18 and S19 , Tables S4 and S5 ) and the CTK complex , involved in transcription and translation regulation [15] ( Ctk1p , Ctk2p , Ctk3p; see Figures S20 and S21 , Table S6 and S7 ) , in both Sc/Sm and Sc/Su hybrids . In the case of the MBF complex , a dimer composed of two proteins , Mbp1 ( a transcription factor responsible for DNA synthesis at the G1/S phase of the cell cycle ) and Swi6p ( a trans-activating component ) [16] , chimeric complexes were only identified in hybrids Sc/Su , while , surprisingly , no free interaction was detected in the hybrids of the more closely related species S . cerevisiae and S . mikatae ( Figures S22 and S23 , Tables S8 and S9 ) . Targeted mass-spectrometry was also performed on Sc/Sm hybrid to seek specifically S . mikatae Swi6 peptides , which constituted the majority of the tryptic digest ( ca 76% of all peptides ) . However , no specific Sm peptides were detected , indicating that this protein was not present in the complex at significant levels ( Table S10 ) . The level of expression of Sm Swi6p is higher than that that one of Sc Swi6p in Sc/Sm background , and is also higher than that one of Su Swi6p in Sc/Su hybrids , as showed by Real time PCR experiments ( Figure S24 ) , ruling out the lack of detection due to the insufficient expression of Swi6p in the Sc/Sm hybrid . This results indicates that , given the choice , Mbp1p from Sc prefer to form uni-specific complexes with Swi6p from Sc in Sc/Sm background . When considering protein-protein interactions the sequence identity of the binding interfaces is likely to be more important than the phylogenetic relationship . In fact , Swi6p shows greater gene sequence similarity between S . cerevisiae and S . uvarum than between S . cerevisiae and S . mikatae , despite their phylogeny ( Figure S25 ) . The remaining two complexes tested , the RAM ( Ram1p and Ram2p , farnesyltransferase complex involved in the prenylation of Ras proteins ) [17] and KU ( Yku70p and Yku80p ) , involved in double strand breaks repair and non-homologous end joining ) [18] , appeared unable to form chimeric complexes in any hybrid background ( Tables S11 , S12 , S13 , S14 ) . In fact , using Yku70p as TAP-bait , no specific Yku80p peptides from S . uvarum and S . mikatae parental species were ever found in any biological replica tested , while numerous S . cerevisiae specific Yku80p peptides were consistently isolated . Although the failure to detect such interactions in mass spectrometry is not a definite proof that chimeric complexes are not at all assembled , this data suggests that chimericity within RAM and Ku complexes may at least occur rarely , and that the proteins forming such complexes tend to assemble in uni-specific manner if given the option . Interestingly , an independent study of the KU complex in hybrids of two diverged strains of S . paradoxus showed that negative epistatic interactions occur between the different homologues of Yku70p and Yku80p , suggesting either lack of assembly or functionality of the heterodimer [19] . The inability to detect spontaneous chimeric complex formation in both Sc/Sm and Sc/Su hybrids observed in this work support the idea that the prevention of complex formation could be the possible mechanism for the negative epistasis identified between Yku70p and Yku80p in the S . paradoxus strains . We evaluated the impact that chimeric interactions have on fitness by forcing the hybrids to use only one specific type of complex to carry out the biological function . We chose to investigate the TRP2/TRP3 ad the MBF complex , since the relationship between the functional complexes and the resulting output fitness could be clearly measured under tryptophan starvation and respiratory growth condition , respectively . In fact , the TRP2/TRP3 complex is involved in the first step of the tryptophan biosynthesis [14] , and null mutants of Mbp1p and Swi6p display a range of fitness defects including decrease rate of respiratory growth and abnormal mitochondrial morphology [20] . We created different combinations of the TRP2/TRP3 and MBF complexes by deleting different protein members in both Sc/Sm and Sc/Su hybrid backgrounds ( Figures 3A and 4A ) , and then scored the growth rates of the hybrids carrying either uni-specific or chimeric complexes . For the TRP2/TRP3 complex in the Sc/Su background , a large range of fitness levels was detected for the different types of assemblies ( Figure 3B ) . The S . uvarum parent grows poorly compared to the S . cerevisiae parent , while the hybrid shows an intermediate fitness ( Figure 3B ) . When comparing the growth of the four strains bearing different combinations of TRP2/TRP3 protein complexes ( i . e . possessing the same TRP2/TRP3 copy number in the same hybrid genetic background ) , we found that the strain with the Trp2pSu/Trp3pSc chimeric complex grew much better than all the other strains in a medium lacking tryptophan ( Figure 3B ) . The uni-parental hemizygous controls Trp2pSu/Trp3pSu showed the lowest fitness , while the chimeric Trp2pSc/Trp3pSu and the hemizygote Trp2pSc/Trp3pSc showed an intermediate growth ( Figure 3B ) . When tryptophan was added to the SD medium the phenotypic difference between the hybrids carrying different protein complexes was minimised ( Figure S26 ) . The strain with the chimeric combination Trp2pSu/Trp3pSc seems to grow similarly to the S . cerevisiae parent and better than the original hybrid . It is possible that , in the parent hybrid , a higher percentage of uni-specific S . uvarum complexes are formed , which are the most unfit of all four combination ( Trp2pSu/Trp3pSu , Figure 3B and C ) , and could therefore partially compromise the fitness of the hybrid . In fact , although the quantitative expression of the two TRP2 orthologs is similar in the hybrid , the S . uvarum TRP3 copy is more expressed than the S . cerevisiae counterpart ( Figure S27 ) . To confirm the increased fitness of the strain expressing a Trp2pSu/Trp3pSc chimeric complex , competition experiments between the chimeric hybrids and a GFP reference strain was carried out using FACS analysis [21] . The results showed that strains with the chimeric Trp2pSu/Trp3pSc complex were more fit than those with the other chimeric complex ( Trp2pSc/Trp3pSu ) and those with both uni-specific protein-protein interaction combinations ( Figure 3C ) . Moreover , a competitive growth essay between the hybrid carrying the fittest chimeric complex Trp2pSu/Trp3pSc and the reference strain was carried out in SD medium with and without tryptophan . The fitness gain of the strain carrying Trp2pSu/Trp3pSc complex was lessened in the medium containing tryptophan ( Figure S28 ) . For the MBF complex in the Sc/Su background all the engineered hybrids carrying the different type of complexes were able to grow on glucose medium , however only the hybrid carrying the uni-specific combination Mbp1pSu and Swi6pSu derived from S . uvarum was able to grow in media containing glycerol , a carbon source that can only be respired ( Figure 4 ) . The other parental combination of Mbp1pSc/Swi6pSc could not be rescued by adding either Mbp1pSu or Swi6pSu to its genotype , showing that the presence of both S . uvarum members of the MBF complex is required for hybrid growth on glycerol ( Figure S29 ) . Interestingly , the restriction analysis of the mitochondrial genes COX2 and COX3 indicated that the Sc/Su hybrids harbour the Su mitochondrial DNA ( data not shown ) . Recently , incompatibilities between nuclear and mitochondrial genes have been proposed as general mechanism causing reproductive isolation between species . This is a type of Dobzhansky-Muller incompatibility involving lack of interaction or malfunctioning of interacting alleles derived from two different species . For example , the S . uvarum nuclear encoded mitochondrial protein Aep2p is unable to regulate the translation of the S . cerevisiae mitochondrial gene OLI [5] , and the S . cerevisiae Mrs1p is not able to splice either the S . paradoxus or the S . uvarum COX1 gene [22] . In the case of MBF complex , we have shown an example of phenotypic plasticity of different chimeric assemblies , and found a novel case of hybrid incompatibility between S . cerevisiae and S . uvarum when cells are grown on a non–fermentable medium and the mitochondria function become essential for cell viability . Fitness variation between the different types of protein assemblies was not otherwise observed in Sc/Sm hybrids either for the TRP2/TRP3 or for the MBF complex ( Figure S30 ) , underlying the dependency of these phenotypes on their genetic background ( manifesting in Sc/Su but not in Sc/Sm hybrids ) . This background dependency is not entirely surprising given the fact that , even between two strains belonging to the same S . cerevisiae species ( i . e . BY4743 and Sigma 1278b ) several conditional essential genes have been discovered [23] . Here we have shown that protein complexes in hybrids of S . cerevisiae/S . mikatae and S . cerevisiae/S . uvarum are able to spontaneously exchange components for inter-specific orthologs , and , while this manuscript was under review , a study on protein-protein interactions among members of the nuclear pore complex and the RNA polymerase II complex in other S . cerevisiae “sensu stricto” hybrids ( i . e . S . cerevisiae/S . kudriazvevii ) also concluded that chimeric protein complexes could assemble [24] . Out of the six complexes studied four were convincingly found to form natural chimeric protein assemblies in either one or both genetic hybrid background ( i . e . Sec62–63 , TRP2/TRP3 , MBF , and CTK complex ) . These results provide evidence that chimeric protein interactions in hybrids can arise to generate evolutionary novelty in protein-protein interaction networks , providing a new evolutionary mechanism to complement innovation by gene duplication [25] . We also found that some complexes prefer to form species-specific configurations in the natural hybrid cell environment ( i . e . Ku and RAM complex ) . The lack of spontaneous chimeric assembly in these cases could be due to less favourable changes in the binding interfaces of the proteins , or to stoichiometry imbalance between homologous proteins in the hybrid [26] . The inability to create chimeric interaction can be responsible for some negative epistatic effect seen in hybrids [19] . We showed that different type of complexes can cause a variety of phenotypes in selected environments . In the case of TRP2/TRP3 , we find that chimeric complex formation can lead to hybrid vigour , reinforcing the idea that the ability to form different types of protein assemblies could be advantageous to the hybrid in specific nutritional contexts . We can speculate that the advantage of the chimeric combination can be due to a more harmonious expression of some alleles leading to a better stoichiometry of that specific type of complex . Alternatively , the chimeric complex may be more efficient in its biological function in the hybrid background . In the case of MBF complex only one parental combination of protein-protein interaction was compatible with cell viability under respiratory condition , highlighting a new case of allelic incompatibilities in yeast hybrids . These phenotypes were proved to be dependent on both genetic and environmental backgrounds since we did not observe any fitness change in Sc/Sm hybrids and the advantages could be lost or gained in different media , such as in the case of the strains carrying different combination of the MBF complex grown in YPD or YP-glycerol ( Figure 4B ) . Ultimately , this study proposes a novel molecular mechanism for creating phenotypic variation within a hybrid cell , with important implications for understanding the evolutionary forces that govern the reshaping of hybrid genomes . The genomic fate of the homolog genes will in fact be influenced by the ability or not of the hybrid to create inter-specific protein assemblies ( Figure S31 ) . Moreover , chimeric complexes may be able to recruit new proteins and evolve new functions in the cell [27] . In the future , the genomic information of naturally occurring hybrids ( like S . pastorianus strains ) will provide insight into the nature of how the formation of chimeric interactions influences selective gene retention of members of protein complexes and networks .
All the TAP-tagged constructs , based on S . cerevisiae MGD353-13D strain , were obtained from the EUROSCARF strains collection ( http://web . uni-frankfurt . de/fb15/mikro/euroscarf/cellzome . html ) . Hybrids between S . cerevisiae strains ( bearing the TAP-tag in selected members of different protein complexes ) and wild-type S . mikatae 1815 and S . uvarum NCYC2669 species were generated using a Singer Instruments MSM micromanipulator as previously described [28] . To enable selection of hybrid colonies , we made the S . cerevisiae TAP strains geneticin-resistant by inserting a kanMX in the neutral AAD3 locus . Hybrid colonies were then selected on minimal media containing geneticin G418 ( see Figure S1 ) . The nature of the chromosomes were verified by chromosomal PCR using genomic DNA from the hybrid as template and species-specific primers designed to distinguish between S . cerevisiae , S . mikatae and S . uvarum alleles ( see Figure S2 and Table S15 , S16 , S17 ) . After the hybrid was created it took ca . 24 generations ( growing in two different selective plates ) to select the hybrids before the PCR was made to check the chromosomes , and another 16 generations before the TAP tagging experiment ( total of about 40 generations since the production of the hybrid ) . The hybrid was then maintained in glycerol stock at −80 C . Hybrid genomic DNA and RNA was isolated using the DNasy Blood & Tissue kit and the RNeasy mini kit ( Qiagen , Crawley , UK ) , respectively . The expression levels of S . cerevisiae , S . uvarum and S . mikatae SWI6 , TRP2 and TRP3 alleles in Sc/Su and Sc/Sm background were performed on the cDNA samples amplified using the Quantitect real time PCR kit from Qiagen . Optimized reactions were carried out using 10 ng/µl of cDNA , 5 pmoles of each primer and syber green according to the manufacturer instructions ( Table S18 ) . Actin ( ACT1 ) was used as a housekeeping reference gene . The expression of each gene was estimated using the Ct Values . Purification of the protein complexes was carried out using the standard TAP protocol [29] optimized for these specific classes of proteins . In particular , two affinity binding steps , the IgG Sepharose and Calmodulin Binding Protein ( CBP ) binding and TEV protease cleavage were carried out for 2 hours at 4°C instead of 16°C . The protein mixtures were resolved using 1D gel electrophoresis , stained with Coomassie Bio Safe ( Bio-Rad ) and digested with trypsin ( Promega ) . The trypsin digest was carried out overnight at 37°C according to Shevchenko , A . et al . [30] . The digested protein mixture was separated by the high performance liquid chromatography ( HPLC ) and analyzed by tandem mass spectrometry ( ESI MS/MS ) ( Micromass CapLC-Q-ToF , Waters , Manchester , UK ) . The system was either used in a discovery manner with the system selecting peptides automatically or in a targeted manner with the system selecting peptides directed from a list of peptides of interest . Spectra acquired for every protein complex member were compared against a custom database containing all proteins from S . cerevisiae “sensu stricto” species , using Mascot version 2 . 2 . 06 ( Matrix Science Inc . , Boston , MA ) . Scaffold ( Scaffold_2_01_00 , Proteome Software Inc . , Portland , OR ) was used to validate MS/MS based peptide identification . A peptide match was acknowledged if it could be established at greater than 50 . 0% probability as specified by the Peptide Prophet algorithm [31] . The peptide criteria were set to 50% as we were looking specifically at homologous proteins and shared peptides are generally given lower confidence scores because it cannot be determined which protein the peptides originate from . Significant peptides were checked manually to ensure all the major fragments were matched and a contiguous series of at least 4 y or b ions were present . Protein identifications were accepted if they could be established at greater than 95 . 0% probability by Protein Prophet and contained at least 2 identified peptides . The Liverpool Peptide Mapping Tool ( http://www . liv . ac . uk/pfg/Tools/Pmap/pmap . html ) was used to generate proteolytic peptide maps of protein complex members . The peptide maps were generated with one trypsin miscleavage per site after lysine and arginine ( K-X , R-X ) but not at lysine-proline and arginine-proline ( K-P , R-P ) sites . Chimeric and unispecific versions of the TRP2/TRP3 and MBF complexes in both Sc/Sm and Sc/Su hybrids were generated by PCR-mediated gene deletion strategy using hygromycin ( HPH ) and nourseothricin ( NAT ) as selectable markers [32] . The S . cerevisiae TRP2 and TRP3 copies were replaced with HPH while the S . uvarum ones were deleted using NAT ( see Figure 3 ) . Similarly for the MBF complex , the S . cerevisiae orthologs of Mbp1 and Swi6 were disrupted using HPH , while the S . uvarum copies of Mbp1 and Swi6 were deleted using NAT ( see Figure 4 ) . Yeast hybrids were grown in YPD , SD and minimal F1 media [33] at 30°C for 40 hours with continuous shaking . Growth rates were measured by absorbance at OD595 at 5 minutes intervals using Fluostar Optima bioscreen workstation ( BMG Labtech ) . Fitness competition assays were carried out by FACS analysis according to Lang et al . [21] . As reference strain we used the FY3 strains bearing the GFP tag at the C-terminus of CDC33p ( generated for the purpose of this experiment ) , and the competition was carried out in SD media lacking tryptophan . The hybrids strains were mixed with the reference strain in 4∶1 ratio , and a total of 1×105 cells , counted on a cellometer ( Auto M10 , Nexcelom ) , were inoculated into a 1 ml of fresh medium . The strains were allowed to grow for 12 hours and then the ratio of the number of hybrid cells over the fluorescent reference was determined using the Dako CyAn flow cytometer , with a total counting total 50 , 000 cells for each time point . Three biological and three technical replicates were performed for each fitness measurement . The sg fitness coefficient was calculated using the following equation:where , H and R are the cell number of the hybrid and reference strain and g0 and gf are the number of generations at the beginning and after a time interval ( 12 hours ) . | The Saccharomyces cerevisiae “sensu stricto” group represent an excellent example of closely related species which can readily hybridise to occupy new ecological niches . Hybrids harbour the DNA of both parents and can display diverse pattern of gene expression . Less is known about the protein interactions that occur in hybrids , where two diverged proteome co-exist and are responsible for the correct execution of the biological function . In fact , hybrids could potentially form different chimeric variants of the same protein complex by using all the different combinations of parental alleles available . Chimeric interactions are expected to be sub-optimal and therefore discouraged since the members forming the protein complex are from different parents and have a different evolutionary history . Interestingly , here , we show experimentally that chimeric protein assemblies are spontaneously established in different yeast hybrids , and that such chimericity produces different phenotypic variants displaying loss or gain of fitness according to their genetic background and to the environment that they are exposed . These findings imply that the formation of chimeric complexes offers a new source of natural variation , widens the adaptation potential of the hybrids towards new nutritional contexts , and may influence genome evolution through selective retention of optimal alleles . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Chimeric Protein Complexes in Hybrid Species Generate Novel Phenotypes |
Almost all attention and learning—in particular , most early learning—take place in social settings . But little is known of how our brains support dynamic social interactions . We recorded dual electroencephalography ( EEG ) from 12-month-old infants and parents during solo play and joint play . During solo play , fluctuations in infants’ theta power significantly forward-predicted their subsequent attentional behaviours . However , this forward-predictiveness was lower during joint play than solo play , suggesting that infants’ endogenous neural control over attention is greater during solo play . Overall , however , infants were more attentive to the objects during joint play . To understand why , we examined how adult brain activity related to infant attention . We found that parents’ theta power closely tracked and responded to changes in their infants’ attention . Further , instances in which parents showed greater neural responsivity were associated with longer sustained attention by infants . Our results offer new insights into how one partner influences another during social interaction .
Attention and learning are supported by endogenous oscillatory activity in the brain [1–4] . The nature of these oscillations and their relationship to behaviour develop and change from infancy into adulthood [5–9] . In infants , convergent research has suggested that theta band oscillations , which are particularly marked during early development [10] , are associated with attentional and encoding processes . Theta band activity increases in infants during periods of anticipatory and sustained attention [11]; in 11-month-old infants , differences in theta band oscillations during object exploration predict subsequent object recognition during preferential looking [12] . Theta activity also increases in infants in social compared to nonsocial settings [13] and is particularly marked in naturalistic settings [13] . Although considerable previous research has investigated how brain oscillations relate to an individual’s behaviour , only a smaller body of research has investigated the neural mechanisms through which interpersonal and social factors influence behaviour [14–16] . This is despite the fact that our brains have evolved for social living [17] , and most of our lives—particularly early life—are spent in social settings [18] . Understanding how social influences on attention and learning are substantiated across the brains of people engaging in social interaction , particularly during the crucial early stages of attention and learning , is an important goal for research [19 , 20] . Previous work has shown that social factors influence infant attention and behaviour over short time-frames ( seconds/minutes ) and long timeframes ( months/years ) . Over long timeframes , the children of parents who engage in more joint engagement during play show superior cognitive outcomes [21–23] . Over short timeframes , when an infant and social partner jointly attend to the same object during naturalistic play , infant attention is increased [24] . Recent research has contrasted two explanations for this finding: first , that social context may cause infants to be more attentive because they are more in control of their own attention behaviours . Second , that social context may offer increased opportunities for parents to scaffold their child’s attention using external attention cues—so infants are more attentive even though they are less in control of their own attention behaviours [25] . Time-series analyses conducted to evaluate these two hypotheses provided evidence more consistent with the latter hypothesis: first , infants’ rate of change of attentiveness was faster during joint play than solo play , suggesting that internal attention factors , such as attentional inertia , may influence looking behaviour less during joint play [26] . Second , adults’ attention forward-predicted infants’ subsequent attention more than vice versa [25] . These behavioural results suggest that infants’ increased attentiveness during social relative to solo play may be attributable to the presence of attention scaffolding from parents using exogenous attention cues [27] . However , to our knowledge , no previous work has examined this question from the neural perspective . Previous research has shown that ostensive social cues such as eye gaze and vocalisations can lead to increases in interpersonal neural synchrony between infants and adults [28] . Bidirectional Granger-causal influences between the brains of infants and adults engaged in social interaction were observed in the theta and alpha frequency bands , which were stronger during direct relative to indirect gaze [28; see also 29; 30] . Infants vocalised more frequently during direct gaze , and individual infants who vocalised longer elicited stronger synchronisation from the adult [28] . These findings raise the possibility that conversely , interpersonal influences between the brains of individuals engaged in social interaction may also actively drive their partners’ attentional processes and behaviour . However , in this previous research , the direct link to attention and behaviour was not examined . Here , we examined the neural and behavioural dynamics of infants’ and adults’ attention in two contexts ( see Fig 1 ) . During joint play , each dyad was presented consecutively with toy objects and asked to play together . During solo play , a 40-cm-high divider was placed between the infant and the parent , and two identical toys were presented concurrently to child and parent , who played separately ( see Fig 1 ) . Looking behaviour was videoed and coded post hoc , frame by frame , at a rate of 30 Hz . Time-lagged cross-correlations were used to assess how changes in one time series preceded or followed changes in another [31; cf . 32 , 33]—an approach similar , but not identical , to Granger causality [34] . Our analyses examined whether changes in one time series ‘forward-predicted’ changes in the other . The age of the infants was selected to be 12 months because this is considered the age at which the capacity for endogenous control of attention first starts to develop rapidly [35 , 36] . As is typical [e . g . , 24] , visual attention was coded as the presence or absence of looking behaviour towards the play object—albeit that previous research has shown the limitations of looking behaviour alone as an index of attention [37 , 38 , 39] . Based on previous research [10 , 13] , we expected that fluctuations in infant theta activity would associate with and forward-predict fluctuations in infant attentiveness . Based on our previous research [25] , we predicted that the forward-predictive relationship between infants’ own endogenous brain activity and infants’ attentiveness would be higher during solo play than joint play because of the increased prevalence of exogenous parental attention scaffolding ( and capture ) during joint play . Further , since previous research indicates that parental responsiveness is an influential factor for early developing cognition [40 , 41] , we also examined the short-term associations between infants’ attention and neural activity in the parent . We predicted , in the absence of prior investigations in this area , that a higher association between infant attention and neural activity in the parent would predict greater attentiveness from the infant .
Fig 2 shows time-lagged cross-correlations between EEG power and visual attention for solo play . Fig 2A and 2B show correlations across the frequency spectrum , with time-lag on the x-axis and EEG frequency on the y-axis . Fig 2C and 2D show results of the cluster-based permutation test . These suggested that the results for both infant solo play ( p = 0 . 002 ) and adult solo play ( p = 0 . 002 ) differed significantly from chance . For infants , the effect was most pronounced in the 3 Hz–7 Hz range ( Fig 2D ) ; for adults , in the 6 Hz–12 Hz range ( Fig 2E ) . In addition , to further confirm the results , a separate bootstrapping analysis was conducted as described in the S1 Text ( section 2 . vi ) , which yielded identical results . In order to examine at which time window the peak cross-correlation was observed between EEG power and visual attention , we excerpted the cross-correlation values just for those frequency bands identified from the cluster-based permutation test ( infants: 3 Hz–7 Hz; adults: 6 Hz–12 Hz; see Fig 2C ) . For infants , the peak cross-correlation was observed at t = –750 ms ( i . e . , between EEG power at time t and attention 750 ms after time t ) . For adults , the peak cross-correlation was observed at t = –1 , 000 ms . ( Of note , these numbers do not indicate the time lag of the EEG data relative to the onset of a look but rather the time lag of the largest cross-correlation between EEG power and attention when treated as two continuous variables . ) Fig 3 compares the mean time-lagged cross-correlations for infant solo play and infant joint Play . All data , including unpaired data , have been included ( see Participants ) . Fig 3A and 3B show cross-correlation plots across the frequency spectrum . ( Fig 3A is identical to 2a and included to allow comparison with Fig 3B ) Fig 3D shows the cluster-based permutation test for the infant joint play condition . This suggested that the infant joint play condition differed significantly from chance ( p = 0 . 008 ) . To directly compare the peak cross-correlation values obtained for infant solo play and infant joint play , we excerpted the cross-correlation values just for those frequencies that the cluster-based permutation test indicated as showing marked differences in both conditions ( 3 Hz–6 Hz; see Fig 2C ) . For solo play , the peak cross-correlation was at t = –1 , 500 ms ( EEG power at time t to attention 1 , 500 ms after time t ) ; for joint play , the peak cross-correlation was at t = +3 , 000 ms . In addition , separate unpaired t tests were conducted at each time window to compare the results across conditions and adjusted for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure [42] . Time windows showing significant differences are indicated using black dots above the plot in Fig 3C . Results indicate that larger cross-correlations were observed during solo play relative to joint play for all time lags between t = –10 , 000 ms and t = +1 , 250 ms . Fig 4A and 4B show the mean time-lagged cross-correlations for parent solo play and parent joint play . Fig 4E shows the cluster-based permutation test for parent joint play , which indicated significant differences from chance ( p = 0 . 001 ) . For parent solo play , the most marked associations between EEG power and attention were at 6 Hz–12 Hz ( Fig 2B ) ; for parent joint play , the most marked associations were at 2 Hz–8 Hz ( Fig 4E ) . To assess the significance of this difference , we measured the frequency of peak association between EEG power and attention for parents during solo play and joint play across all frequency bands under consideration ( 2 Hz–12 Hz ) during the ±1 , 000 ms time window . Results obtained from the two conditions were compared using a paired t test; a significant difference between the two conditions was observed ( t ( 44 ) = 3 . 42 , p = 0 . 001 ) . This suggests that the peak association between brain activity and attention in the parent was observed at lower frequencies during joint play than during solo play . Fig 5A and 5B show the mean time-lagged cross-correlations , and Fig 5D and 5E show the cluster-based permutation tests , for the relationship between parents’ EEG power and infants’ attention . For parent EEG and infant attention in the joint play condition , a significant relationship was identified ( p = 0 . 041 ) . The most marked associations were identified in the 4 Hz–6 Hz range ( Fig 5E ) . An identical analysis examining the relationship between parent EEG and infant attention in the ( concurrent but separate ) solo play condition identified no significant relationship . In addition , a further bootstrapping analysis was performed ( see S1 Text ) , which confirmed that the observed cross-correlation values significantly exceed chance for joint play but not solo play . For the within-participant analysis of solo play , the peak cross-correlation values observed were consistently negative ( ‘brain pre-look’ ) ( Figs 2C and 3C ) . In order to directly compare the peak cross-correlation values obtained between the solo play and joint play conditions , we excerpted the cross-correlation values just for those frequency bands identified from the cluster-based permutation test as showing marked differences during joint play ( 4 Hz–6 Hz ) ( see Fig 5C ) . For joint play , the peak cross-correlation value occurred at a t = +750 ms ( i . e . , between infant attention at time t and adult EEG 750 ms after time t , ‘adult brain post-infant look’ ) . In addition , we conducted a further analysis using separate procedures from those used in Analyses 1 and 2 . Whereas Analyses 1 and 2 examine the cross-correlation between EEG power and attention when treated as two continuous variables , Analysis 3 examines changes in EEG power relative to the onsets of individual looks . We examined all looks to the play objects that occurred during the session . For each look , we excerpted the power in the theta band for three time windows immediately prior to the onset of each look ( 3 , 000–2 , 000 , 2 , 000–1 , 000 , and 1 , 000–0 ms pre-look onset ) and three windows immediately after the onset of each look ( 0–1 , 000 , 1 , 000–2 , 000 , and 2 , 000–3 , 000 ms post look onset ) . Theta power was defined according to the frequency bands identified from the cluster-based permutation tests as showing the most marked differences from chance . These were infant solo play ( Fig 2D ) —3 Hz–7 Hz; infant joint play ( Fig 3D ) —4 Hz–7 Hz ) ; adult to infant ( Fig 5E ) —4 Hz–6 Hz . We then calculated separate linear mixed effects models for each of the six windows to examine the relationship between EEG power within that time window and look duration . Full results are shown in S1 Table , and key results are shown in Fig 6 . In the solo play condition ( Fig 6a ) , a relationship was observed between infants’ theta power and look duration , consistent with the results of Analysis 1 ( Fig 2A ) . Theta power in the time window –1 , 000 to 0 ms prior to look onset significantly predicted the subsequent duration of that look , consistent with the forward-predictive relationship noted in Fig 2C . The strength of this relationship increased for time windows after the onset of the look . Conversely , for joint play ( Fig 6B ) , there was no significant relationship between infants’ theta power and look duration . Again , this finding is consistent with the results of Analysis 1 ( Fig 3C ) . During joint play , parental theta power associated significantly with infant attention in the time windows after the onset of the look ( 0–1 , 000 ms and 1 , 000–2 , 000 ms; Fig 6C ) . However , there is no relationship in the time windows prior to look onset . This result is also consistent with the results of Analysis 2 ( Fig 5C ) .
It is well established that attention and learning are supported by the endogenous oscillatory neural activity of the person attending . However , relatively little is known about how interpersonal and social influences on attention are substantiated in the brain [16 , 43] . To investigate this , we examined how the oscillatory dynamics of attention are shared between infant–parent dyads and how these dynamics differ between noninteractive and interactive social play . We found that when infants were engaged in solo play , continuous fluctuations in theta power forward-predicted visual attention in infants ( Fig 2 ) . Consistent with this , a separate analysis identified a positive association between theta power in the 1 , 000 ms prior to look onset and the subsequent duration of that look ( Fig 6 ) . For adults , a similar functional relationship was observed but at a higher frequency ( 6 Hz–12 Hz ) in the alpha band , consistent with considerable previous research into the role of prestimulus alpha activity in anticipatory visual attention [44 , 45] . Our infant findings are also consistent with previous research suggesting that theta oscillations increase during anticipatory and sustained attention and encoding [10; 12 , 13] , but they are novel insofar as we demonstrated these effects during spontaneous attention in seminaturalistic settings . During interactive social play , however , we found that this forward-predictive relationship between infants’ endogenous theta activity and visual attention was still present but much reduced . Again , this result was observed consistently across two separate analyses ( Fig 3 and Fig 6 ) . Particularly of interest was Fig 3C , which suggested that negative-lag relationships ( attention forward-predicting EEG power ) were similar across the solo and joint play conditions but that positive-lag relationships ( EEG power forward-predicting attention ) were present only during solo play . These results are consistent with our previous research suggesting that endogenous factors , such as attentional inertia , influence infants’ attention more during solo ( noninteractive ) play than during joint play [25] . Taken together , our results suggest that infants’ endogenous neural control over attention is greater during solo play . These results appear unlikely to be attributable to oculomotor artefact associated with the onsets and offsets of looks for a number of reasons . First , during data preprocessing , we removed oculomotor artefacts via independent component analysis ( ICA ) ( see S1 Text ) ; second , we have only reported data in this paper from two channels near the vertex—C3 and C4 , which show the least contamination by muscle and motion artefacts . ( See S5 and S6 Figs for comparable plots of anterior and posterior midline groups . ) Third , the cross-correlation analysis across different frequencies ( Fig 2A ) indicated that relationships were specific to the theta band . Muscular artefacts generally produce the highest contamination in delta , beta , and gamma bands [46 , 47] . Fourth , effects were present around the onsets of looks in the solo play but not the joint play condition ( Fig 3A and 3B ) . Our findings are also unlikely to be attributable to differences in mean look duration between the two conditions ( see S1 Fig ) for two reasons . First , as in Analysis 1 , any artefactual effects would be random rather than directional ( i . e . , specifically affecting negative rather than positive lags ) . Second , Analysis 1 examined the relationship between attention and EEG power considered across continuous entire time series , whereas Analysis 3 examined power changes relative to the onsets of individual looks , and the results from the two analyses produced converging conclusions . Furthermore , this result is also not attributable to differences in relative power between the two conditions because the EEG power spectrum of infants did not differ across conditions ( S2 Fig ) . Overall , however , we found that despite the fact that infants’ endogenous attention control over their own behaviour patterns appeared to be lower , they were more attentive towards objects during joint play ( S1 Fig ) —a finding consistent with previous research [24] . To understand why , we examined how adult brain activity related to infant attention . First , we found that during joint play , the frequency of adults’ peak association between EEG power and attention was down-shifted to the theta range—similar to infants’ peak frequency of association ( Fig 4 ) . Second , we found that parent EEG theta power significantly tracked infant attention . Again , this result was observed across two separate analyses . Analysis 2 ( Fig 5D and 5E ) suggested that infant attention associated , over a time-frame of ±2 seconds , with increased parental theta power . Analysis 3 ( Fig 6C ) suggested that individual infant attention episodes accompanied by greater parental EEG power were longer lasting . Importantly , we found that the direction of the peak association differed between solo and interactive play . During solo play , the peak cross-correlation between infant theta power and infant attention was observed at negative lag ( ‘brain pre-look’ ) ( Figs 2C and 3C ) , and theta power 1 , 000 ms prior to look onset predicted look durations ( Fig 6C ) . During joint play , the peak cross-correlation between adult theta power and infant attention was observed at positive lag ( ‘brain post-look’ ) ( Fig 5C ) , and Analysis 3 identified backwards-predictive but not forward-predictive relationships between adult theta power and infant look duration ( Fig 6C ) . These findings appear to suggest that , during joint play , parents’ theta power tracks and responds to changes in infants’ attention . One possible account of our findings we considered is that infant attention may ( Granger- ) cause adult attention , which in turn causes increased theta activity in adults . This explanation appears unlikely , however , because in S1 Text , we report a control analysis in which instances in which an attention shift from the infant was immediately followed by an attention shift from the parent were excluded . The results obtained from this subset of the data were highly similar to those reported in the main text ( see S8 Fig ) . Furthermore , as we show in Fig 1D , adults’ gaze forward-predicted infants’ attention more than vice versa , which also appears inconsistent with this explanation . Overall , then , our results suggest that adults show neural responsivity to the behaviours of the child , and that increased parental neural responsivity associates , look by look , with increased infant attentiveness . Temporally fine-grained patterns of parental responsivity to infants have previously been shown using methods other than neuroimaging , such as microcoding of facial affect [48 , 49] , autonomic physiology [50] , visual attention [51] , and vocalisations [52; 53] . And , using neuroimaging , research with adults has provided evidence for common activation elicited when experiencing emotions such as disgust [54] , touch [55] , or pain [56] in oneself and when perceiving the same feelings in others . However , this is the first study , to our knowledge , to demonstrate temporal associations between infants’ attentiveness and parental neural correlates of attention and to show that moment-to-moment variability in adults’ neural activity associates with moment-to-moment variability in infants’ attentiveness . Although demonstrated here in the context of parent–child interaction , future research should explore whether our present findings extend to cover other aspects of social interaction [57] . They should also be extended to explore individual differences—whether some social partners show greater neural responsiveness to others and how this influences behaviour [49]—and to other aspects of interpersonal neural influences than shared attention during joint play . Finally , future work should examine the mechanisms through which the children of parents who show increased responsivity over shorter timeframes develop superior endogenous attention control over long timeframes [21–23 , 58 , 59] .
The study was conducted according to guidelines laid down in the Declaration of Helsinki , with written informed consent obtained from a parent or guardian for each child before any assessment or data collection . All procedures involving human subjects in this study were approved by the Psychology Research Ethics Committee at the University of Cambridge ( Number PRE . 2016 . 029 ) . No financial inducements were offered other than the reimbursement of travel expenses and the gift of a T-shirt for participating infants . Twenty-four and twenty-five parents contributed usable data for the joint play and solo play conditions , respectively; for infants , it was 21 and 25 for joint play and solo play , respectively . Paired parent–child data were available for 20 dyads for joint play ( 10 M and 10 F infants; mean [SE] infant age 345 . 1 [12 . 1] days; mother age 34 . 7 [0 . 8] years ) and for 22 dyads for solo play ( 12 M and 10 F infants; mean [SE] infant age 339 . 2 [10 . 3] days; mother age 34 . 1 [1 . 0] years ) . All participating parents were female . It should be noted that the recruitment area for this study , Cambridge , United Kingdom , is a wealthy university town , and the participants were predominantly Caucasian and from well-educated backgrounds and so do not represent an accurate demographic sample [60] . As previously reported [25] , infants were seated in a high chair , which was positioned immediately in front of a table . The toys on the table were within easy reach ( see Fig 1 ) . Parents were positioned on the opposite side of the 65-cm-wide table , facing the infant . In the solo play condition only , a 40-cm-high barrier was positioned across the middle of the table ( see Fig 1A ) . When the barrier was in place , parent and child had line of sight to one another ( to reduce the possibility of infant distress ) , but neither could see the objects with which the other was playing . Each infant–parent dyad took part in both the joint play and solo play conditions . Presentation order was randomised between participants , but the two conditions were presented consecutively , with a short break in between . Parents were informed that the aim of the study was to compare behaviour while they were attending to objects separately from each other and when they were attending to the same object . During the solo play condition , parents played silently with the toys alone . During the joint play condition , they played silently with the toys whilst involving their infant in the play . A research assistant was positioned on the floor out of the infant’s sight . The research assistant placed the toys onto the table one at a time . In the joint play condition , one toy was presented at a time . In the solo play condition , two identical toys were presented concurrently to the infant and parent , one on either side of the barrier . The toys were small ( <15 cm ) , engaging objects . Presentation order was randomised between conditions and between participants . Approximately every two minutes , or more frequently if the child threw the object to the floor , the current toy object was replaced with a new object . The mean ( SE ) duration for which each object was presented was 140 . 1 ( 17 . 9 ) seconds for joint play and 110 . 3 seconds ( 7 . 9 ) for solo play . Approximately 10 minutes of data was collected per condition from each dyad . The mean ( SE ) duration of play for each condition was 10 . 80 ( 0 . 46 ) minutes for joint play and 10 . 35 ( 0 . 33 ) minutes for solo play . When the infant became fussy during testing , data collection was stopped earlier; however , this occurred fairly rarely: the number of infants contributing sessions that lasted less than 8 minutes was 2/3 for the joint play/solo play conditions . Play sessions were videoed using two camcorders positioned next to the child and parent , respectively . Further details of video coding and synchronisation are given in S1 Text . The visual attentional patterns of parents and infants were manually coded by reviewing their respective video recordings on a frame-by-frame basis ( 30 frames per second , 33 . 3 ms temporal acuity ) using video editing software ( Windows Movie Maker ) ( see Fig 1 ) . This coding identified the exact start and end times of periods during which the participant was looking at the toy object . A previous report based on these data , which contained behavioural findings only , reported that infants showed longer look durations towards the object during joint play relative to solo play , together with shorter periods of inattention ( see S1 Fig ) [25] . EEG signals were obtained using a 32-channel wireless Biopac Mobita Acquisition System ( Biopac Systems , Goleta , CA , USA ) and 32-channel Easycap . Further details of EEG acquisition are given in S1 Text . Automatic artefact rejection followed by manual cleaning using ICAs was performed . Full descriptions are given in S1 Text . Because previous analyses have shown that movement and muscle artefacts can contaminate EEGs [46 , 47] , data from all channels other than the two channels close to the vertex , C3 and C4 , were excluded , and only frequencies between 2 and 14 Hz were examined . Analyses suggested that these frequencies show the least EEG signal distortion due to sweating , movement , or muscle artefact [46] . Prior literature [e . g . 11 , 61] suggests that these frequencies were also most likely to show associations with visual attention . In S5 and S6 Fig , we also include comparison plots based on alternative anterior and posterior midline electrode groupings , which are consistent with the results reported in the main text . For each electrode , we computed the Fourier transform of the activity averaged over artefact-free epochs , using the fast Fourier transform algorithm implemented in MATLAB ( The MathWorks , Natick , MA , USA ) ( see S1 Text for full description ) . The FFT was performed on data in 2 , 000 ms epochs , which were segmented with an 87 . 5% ( 1 , 750 ms ) overlap between adjacent epochs . Thus , power estimates of the EEG signal were obtained with a temporal resolution of 4 Hz and a frequency resolution of 1 Hz . S2 Fig compares EEG power for infants and parents between solo play and joint play; no significant between-condition differences were observed . The attention data used for the cross-correlation analysis were resampled as continuous and time-synchronised data streams at 4 Hz ( to match that of the EEG power estimate ) . Attention data were coded as 1 and 0 ( either attentive towards the play object or not ) . The cross-correlation calculations were performed separately for each frequency band ( in 1 Hz bands ) and for each member of the dyad ( infant brain–infant attention and parent brain–parent attention ) ( Analysis 1 ) . Then , they were calculated across the dyad ( parent brain–infant attention ) ( Analysis 2 ) . For each computation , the zero-lag correlation was first calculated across all pairs of time-locked ( i . e . , simultaneously occurring ) epochs , comparing the EEG power profile with the attention data using a nonparametric ( Spearman’s ) correlation . In S4 Fig , we also show the results of the same tests repeated using an alternative test , the Mann–Whitney U test , for which results were identical . ) The mean correlation value obtained was plotted as time ‘0’ ( t = 0 ) in the cross-correlation . Next , time-lagged cross-correlations were computed at all lags from –10 to +10 seconds in lags of ±250 ms ( corresponding to one data point at 4 Hz ) . For example , at lag time t = –250 ms , the EEG power profile was shifted one data point backwards relative to the attention data , and the mean correlation between all lagged pairs of data was calculated . Based on an average of 10 . 5 minutes of data per condition , sampled at 4 Hz and allowing for some attrition at artefact rejection due to the max-min thresholding criteria , the N of the cross-correlation was approximately 2 , 300 for the zero-lag correlation and up to 40 fewer for the most shifted correlation . In this way , we estimated how the association between two variables changed with increasing time lags . The individual cross-correlation series were then averaged across participants to obtain the group mean cross-correlation at each time interval and frequency band . To compare the distribution of time × frequency data between any single condition and a null distribution , a cluster-based permutation test was conducted across time × frequency data using the FieldTrip function ft_freqstatistics [62] . In comparison to other approaches to solving the family-wise error rate , this approach identifies clusters of neighbouring responses in time/frequency space [63] . In particular , corresponding time × frequency points were compared between contrast condition and null distribution with a t test , and t values of adjacent spatiotemporal points with p < 0 . 05 were clustered together with a weighted cluster mass statistic that combines cluster size and intensity . The largest obtained cluster was retained . Afterwards , the whole procedure , i . e . , calculation of t values at each spatiotemporal point followed by clustering of adjacent t values , was repeated 1 , 000 times , with recombination and randomised resampling before each repetition . This Monte Carlo method generated an estimate of the p value representing the statistical significance of the originally identified cluster compared to results obtained from a chance distribution . In addition , a supplementary analysis was conducted using bootstrapping in order to further verify our results ( see S1 Text ) . Analysis 3 examined whether individual looks accompanied by higher theta power are longer lasting . To calculate this , we examined all looks to the play objects that occurred during the play session . The onset times of these looks were calculated , as described above , at 30 Hz . Then , for each look , we excerpted the EEG power for three time windows immediately before and after the onset of each look ( 3 , 000–2 , 000 , 2 , 000–1 , 000 , and 1 , 000–0 ms pre-look onset; 0–1 , 000 , 1 , 000–2 , 000 , and 2 , 000–3 , 000 ms post-look onset ) . Separately , we calculated the duration of each look towards the object . Since these were heavily positively skewed , as is universal in looking time data [64] , they were log-transformed . Then , we calculated separate linear mixed effects models for each of the six windows using the fitlme function in MATLAB . For each model , we examined the relationship between EEG power within that time window and look duration , controlling for the random effect of participant . In this way , we examined whether , for example , theta power in the time window 1 , 000–0 ms prior to the onset of a look showed a significant relationship to the subsequent duration of that look . | We are a social species . Most infants and young children spend the majority of their early waking hours in the company of others . However , almost everything that we know about how the brain subserves early attention and learning comes from studies that examined brain function in one individual at a time just because it is easier to do experiments that way . Here , we examine the neural correlates of how attention is shared between two people engaged in social interaction . We recorded brain activity from infants and parents using scalp electroencephalogram during parallel solo play with toys and during joint play . We examined the associations between attention and brain activity in each member of the dyad independently ( infant attention–infant brain , parent attention–parent brain ) , and we also examined cross-dyad associations ( infant attention–parent brain ) . Our findings suggested that infants’ attention is more endogenously controlled during solo play than joint play . They also suggested that parents are neurally responsive to their infants during social play , and that , when the parent is more neurally responsive , the infant is more attentive . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"learning",
"children",
"medicine",
"and",
"health",
"sciences",
"sociology",
"brain",
"electrophysiology",
"social",
"sciences",
"electrophysiology",
"neuroscience",
"learning",
"and",
"memory",
"social",
"systems",
"age",
"groups",
"clinical",
"medicine",
"cognitive",
... | 2018 | Parental neural responsivity to infants’ visual attention: How mature brains influence immature brains during social interaction |
Predictions of interactions between target proteins and potential leads are of great benefit in the drug discovery process . We present a comprehensively applicable statistical prediction method for interactions between any proteins and chemical compounds , which requires only protein sequence data and chemical structure data and utilizes the statistical learning method of support vector machines . In order to realize reasonable comprehensive predictions which can involve many false positives , we propose two approaches for reduction of false positives: ( i ) efficient use of multiple statistical prediction models in the framework of two-layer SVM and ( ii ) reasonable design of the negative data to construct statistical prediction models . In two-layer SVM , outputs produced by the first-layer SVM models , which are constructed with different negative samples and reflect different aspects of classifications , are utilized as inputs to the second-layer SVM . In order to design negative data which produce fewer false positive predictions , we iteratively construct SVM models or classification boundaries from positive and tentative negative samples and select additional negative sample candidates according to pre-determined rules . Moreover , in order to fully utilize the advantages of statistical learning methods , we propose a strategy to effectively feedback experimental results to computational predictions with consideration of biological effects of interest . We show the usefulness of our approach in predicting potential ligands binding to human androgen receptors from more than 19 million chemical compounds and verifying these predictions by in vitro binding . Moreover , we utilize this experimental validation as feedback to enhance subsequent computational predictions , and experimentally validate these predictions again . This efficient procedure of the iteration of the in silico prediction and in vitro or in vivo experimental verifications with the sufficient feedback enabled us to identify novel ligand candidates which were distant from known ligands in the chemical space .
In the early stages of the drug discovery process , prediction of the binding of a chemical compound to a specific protein can be of great benefit in the identification of lead compounds ( candidates for a new drug ) . Moreover , the effective screening of potential drug candidates at an early stage generates large cost savings at a later stage of the overall drug discovery process . In the field of virtual screening for the drug discovery , docking analyses and molecular dynamics simulations have been the principal methods used for elucidating the interactions between proteins and small molecules [1]–[4] . Fast and accurate statistical prediction methods for binding affinities of any pair of a protein and a ligand have also been proposed for the case where information regarding 3D structures , binding pockets and binding affinities ( e . g . pKi ) for a sufficient number of pairs of proteins and chemical compounds is available [5] . However , the requirement of these programs for 3D structural information is a severe disadvantage , as the availability of these data is extremely limited . Although a number of structures in PDB [6] is increasing ( from 23 , 642 structures in 2003 to 48 , 091 structures in 2007 ) , not all proteins which have been derived from many genome-sequencing projects are suitable for experimental structure determination . Hence , the genome-wide application of these methods is in fact not feasible . For example , among the GPCRs ( G-protein coupled receptors ) , whose modulation underlies the actions of 30% of the best-known commercial drugs [7] , the full structure of only a few mammalian members , including bovine rhodopsin [8] and human beta 2 adrenoreceptor [9] , is known . To achieve more comprehensive and faster protein-chemical interaction predictions in the post-genome era producing a vast number of protein sequences whose structural information is not available , it is essential to be able to utilize more readily available biological data and more generally applicable methods which do not require 3D structural data [10]–[12] . In our previous study , we developed a comprehensively applicable statistical method for predicting the interactions between proteins and chemical compounds by exploiting very general biological data , including amino acid sequences , 2-dimensional chemical structures , and mass-spectrometry ( MS ) data [11] . These statistical approaches provided a novel framework where the input space consists of pairs of proteins and chemical compounds . These pairs are classified into binding and non-binding pairs , while most chemoinformatics approaches assess only chemical compounds and classify them according to their pharmacological effects . Our previous study [11] demonstrated that screening target proteins for a chemical compound could be performed on a genome-wide scale . This is due to the fact that our method can be applied to all proteins whose amino acid sequences have been determined even though the 3D structural data is not yet available . Genome-wide target protein predictions were conducted for MDMA , or ecstasy , which is one of the best known psychoactive drugs , from a pool of 13 , 487 human proteins , and known bindings of MDMA were correctly predicted [11] . Although the method yielded a relatively high prediction performance ( more than 80% accuracy ) in cross-validation and usefulness in the comprehensive prediction of target proteins for a given chemical compound with tens of thousands of prediction targets [11] , it suffered from the problem of predicting many false positives when comprehensive predictions were conducted . Although these false positives might include some unknown true positives , they were mainly due to the low quality of the negative data , which is one of the common problems in utilizing statistical classification methods such as Support Vector Machines ( SVMs ) and Artificial Neural Networks ( ANNs ) . In this paper , we describe two strategies , namely two-layer SVM and reasonable negative data design , which are used for the purpose of reducing the number of false positives and improving the applicability of our method for comprehensive prediction . In two-layer SVM , in which outputs produced by the first-layer SVM model are utilized as inputs to the second-layer SVM , in order to design negative data which produce fewer false positives , we iteratively constructed SVM models or classification boundaries and selected negative sample candidates according to pre-determined rules . By using these two strategies , the number of predicted candidates was reduced to around 100 ( Table 1 ) in experiments in which the potential ligands for some druggable proteins ( UniProt ID P10275 ( androgen receptor ) , P11229 ( muscarinic acetylcholine receptor M1 ) and P35367 ( histamine H1 receptor ) ) are predicted on the basis of more than 100 , 000 compounds in the PubChem Compound database ( http://pubchem . ncbi . nlm . nih . gov/ ) . With the aim of validating the usefulness of our method , our proposed prediction model with fewer false positives was applied to the PubChem Compound database in order to predict the potential ligands for the “androgen receptor” , which is one of the genes responsible for prostate cancer . We verified some of these predictions by measuring the IC50 values in an in vitro assay . Biological experiments , conducted to verify the computational predictions based on statistical methods , docking methods or molecular dynamics methods , typically involve success as well as failure . In addition to fast calculation and wide applicability , one of the merits of using statistical methods that involve training with known data is that results obtained by verification experiments can be efficiently utilized as feedback to produce new and more reliable predictions . Most previous work on virtual screening has focused on the computational prediction and listing of dozens or hundreds of candidates , followed by their experimental verification . However , only on rare occasions have these experimental results been utilized for the further improvement of computational predictions and experiments . Moreover , even without verification experiments , additional data acquired from , for example , relevant literature can be used for enhancing the prediction reliability . Therefore , we propose a strategy based on the effective combination of computational prediction and experimental verification . Our second computational prediction utilizing feedback from the first experimental verification successfully discovered novel ligands ( Figure 1 and 2 ) for the androgen receptor . Our approach suggests the significance of utilizing statistical learning methods and feedback from experimental results in drug lead discovery . In the following section , we first describe the real application of our method involving the computational prediction , the experimental verification and the feedback , and then explain the computational experiments conducted to verify the usefulness of our computational prediction method in comprehensive prediction .
In bioinformatics , statistical approaches extract rules from numerical data corresponding to biological properties . Here , it is not guaranteed that the extracted rules are biologically valid , and furthermore it is possible to utilize statistical methods to obtain general rules from any kind of numerical data which are meaningless and irrelevant to biological properties . The biological relevance of our approach can be verified as follows on the basis of supporting evidence which indicates that our method can extract significant rules only if biologically valid and relevant data is given . First , high prediction performances on diverse datasets might support the validity of our approach . In several datasets consisting of known pairs of proteins , including nuclear receptors , GPCRs , ion channels and enzymes , and drugs and random protein-drug pairs , our statistical approach with SVM showed high prediction performances ( details are provided in Text S1 , Table S1 and Figure S2 ) . The fact that more than 0 . 85 AUC and an accuracy of 80% were obtained for diverse datasets suggests that it is possible to extract some properties accountable for interactions between proteins and drugs by statistical approaches . This possibility can be further supported by the fact that integrating several datasets whose target proteins were not relevant to each other improved the prediction performances with respect to pairs of proteins and chemical compounds which had a specific binding mode ( details are provided in Text S1 and Table S2 ) . Second , we showed the biological relevance of these high prediction performances by calculating the prediction performances using biologically meaningless artificial datasets as positives . Several datasets which contained fractions of valid samples found in the DrugBank dataset , and which comprised artificial pseudo-positive samples of protein-chemical pairs produced by shuffling with the same frequency of chemical compounds and proteins as that in the DrugBank dataset , were generated . Our method was applied to these shuffled artificial datasets ( Figure 3 ) . Here , if our approach did not depend on the biological properties of the given dataset but only succeeded in classifying given pairs comprising a protein and a chemical compound and random pairs derived from them , the prediction accuracy for each shuffled dataset was assumed not to fluctuate . As shown in Figure 3 , the prediction accuracy was proportional to the content rate of the biologically valid samples . Therefore , the classification of our approach was shown to function only when a certain amount of biologically valid pairs comprising a protein and a chemical compound are given . This result suggests that our statistical approach succeeds in extracting the rules which are only relevant for the biological binding properties . It is often observed that although statistical learning approaches achieve very high prediction performances in given datasets , statistical prediction models suffer from the problem of generating vast prediction sets including many false positives when applied to a huge dataset , such as the PubChem database . In our approach , SVM models based on feature vectors directly representing amino acid sequences , chemical structures , and random protein-compound pairs as negatives also produced many predictions and inevitably yielded many false positives ( Table 1A random ) . Upon the introduction of the two-layer SVM and the negatives designed to overcome this drawback , the prediction precision , or the confidence of positive prediction , was significantly improved in computational experiments based on the DrugBank dataset ( Table 2 ) . In Table 2 , the external dataset consisted of 170 positives and 2 , 450 negatives that were randomly chosen from 1 , 731 positives and 24 , 500 designed negatives with the mlt rule ( details are provided in Materials and Methods ) and that were excluded in constructing first-layer and second-layer SVM models . The external dataset contained much more negatives than positives as it simulated the real application of virtual screening with vast databases where only a fraction of chemical compounds in the databases have the effect of interest . Tables 2A and 2B showed improvement of precision by introducing the designed negatives and the two-layer SVM respectively . Table 2B also indicated that the application of SVM to outputs of the first-layer SVM models was superior to other statistical learning methods [15] and naive combination of the first-layer SVM models , and that rational selection of the first-layer SVM models achieved significant higher precision ( P-value = 0 . 0081 by t test ) than randomly selected models ( other comparisons are provided in Text S1 , Table S3 and Table S4 ) . Particularly , the second-layer SVM utilizing the allpos first-layer SVM models achieved higher precision than use of higher thresholds in the other SVM models ( Table 2C ) . The high precision contributes to the selection of more reliable predictions and thus to the reduction of the number of false positives . Following these results on given datasets , our approaches were evaluated with respect to comprehensive binding ligand prediction . For three proteins ( UniProt ID P10275 ( androgen receptor ) , P11299 ( muscarinic acetylcholine receptor M1 ) and P35367 ( histamine H1 receptor ) ) , their binding ligands were predicted from PubChem Compound 0000001–00125000 which contains 109 , 841 compounds ( Table 1 ) . Here , P35367 and P11299 are the two most frequently targeted proteins in the DrugBank dataset , and P10275 is a protein of average occurrence in the DrugBank dataset . Among the 109 , 841 compounds , 47 , 45 , and 5 known ligands were included for P35367 , P11299 , and P10275 , respectively . As shown in Tables 1A , 1B and 1C , the use of carefully selected negatives , the introduction of the two-layer SVM , and the integration of these two approaches efficiently reduced the number of predictions and thus the number of false positives . For example , in comparison to Tables 1A and 1C , the number of candidates discovered by using the max dataset in the allpos two-layer SVM approach was about one fiftieth of the number of chemical compounds predicted by using the random negative dataset in the one-layer SVM . Furthermore , in comparison to other approaches based solely on the use of chemical compounds ( Tables 1D and 1E ) , our approaches gave a reasonable number of predictions ( other comparisons are described in Text S1 and Tables S5 , S6 , S7 ) . These results suggest that our prediction models select a reasonable number of ligand candidates from all chemical compounds in large databases and encourage the comprehensive binding ligand prediction for the target protein . The experimental verification of the computational predictions produces feedback data or samples which are not included in the given training datasets . The efficient utilization of these data can contribute to the fast identification of compounds with the desired properties and can be of advantage to statistical learning approaches . We compared several strategies for utilizing feedback data as follows . For three proteins ( UniProt ID P10275 ( androgen receptor ) , P11299 ( muscarinic acetylcholine receptor M1 ) and P353367 ( histamine H1 receptor ) ) , ligand data which were not included in the DrugBank dataset were collected from relevant literature [16]–[18] and public databases , PDSP Ki database [19] and GLIDA [20] , in February 2008 . Overall , 35 androgen receptor-ligand pairs , 49 muscarinic acetylcholine receptor M1-ligand pairs , and 1 , 060 histamine H1 receptor-ligand pairs were supplemented . Additional models were constructed by using these supplemental pairs as positives ( details are provided in Text S1 ) . As shown in Figure 4 , the use of the additional model with a sufficient weighting factor controlled the increase of the predictions with a slight decrease of the recall rate . The use of large weighting factors results in the relative decrease of the influence of other first-layer SVM models derived from the DrugBank dataset in classification . However , the low performance of “only additional model:st2” , shown in Figure 4A , where only one first-layer SVM model derived from additional data was used to construct the second-layer SVM model , indicates the need for first-layer SVM models derived from the DrugBank dataset as well as combinations of these first-layer SVM models with an additional first-layer SVM model . With this efficient strategy for utilizing feedback data , computational prediction and experimental verification improve each other to enable faster search toward the identification of useful small molecules .
We proposed a comprehensively applicable computational method for predicting the interactions between proteins and chemical compounds , in which the number of false positives was reduced in comparison to other methods . Furthermore , we proposed the strategy for the efficient utilization of experimental feedback and the integration of computational prediction and experimental verification . The application of our method to the androgen receptor resulted in 67% ( 4/6 ) prediction precision according to in vitro experimental verification in the first computational prediction and 60% ( 3/5 ) in the second prediction , which included the feedback of the first experimental verification . However , these relatively low precision values do not represent the true statistical significance of the method . This 60–70% precision can also be evaluated by using the following P-value . Here , N is the number of prediction targets , M the number of ligands potentially binding to the target proteins , t is the number of tested compounds , and p is the number of true positives . With N = 19171127 , which is the number of chemical compounds in the PubChem Compound database , and M = 19171127× ( 456/3000 ) × ( 7/964 ) ≒21160 , which is based on the optimistic assumption that all compounds can be regarded as potential drugs for some target protein , it is estimated that 3 , 000 druggable proteins exist [21] . Moreover , the distribution of target proteins and drugs in the DrugBank dataset , consisting of 456 target proteins and 964 drugs , including 7 known ligands for the human androgen receptor , and P-values of 2∶21×10−11 and 1∶34×10−8 are obtained for the prediction precision of the first and the second computational prediction , respectively . These extremely small P-values prove the significance of the virtual screening and its precision in the drug discovery process . These prediction performances are as good as or better than several previous virtual screening studies based mainly on docking analyses [22]–[24] . For example , at a threshold of 100 µM , 7% precision ( 3/39 ) for Mycobacterium tuberculosis adenosine 5′-phosphosulfate reductase [22] , 71% precision ( 22/31 ) for Staphylococcus aureus methyonyl-tRNA synthetase [23] and 8% precision ( 16/192 ) for human DNA ligase I [24] were obtained , respectively . In addition , 0 . 566 AUC was achieved in the docking analysis using AutoDock [3] ( Figure 5 ) for the 17 chemical compounds ( 12 chemical compounds verified in the first experimental verification , with the exception of 6 known drugs , and 5 chemical compounds verified in the second experimental verification ) . In contrast , 0 . 681 AUC was obtained with our method . Here , in the calculation of AUC , the threshold level of IC50 = 100 µM for experimental verification was used to define a label ( binding or non-binding ) for each chemical compound , and or the predicted probability was regarded as a value for each molecule . Note that the docking analysis with AutoDock was not applied to the 19 , 171 , 127 compounds in the PubChem Compound database for the screening purpose , but was applied only to 17 compounds , which were the results of virtual screening by our method . In terms of computational time , for binding prediction of one pair of a protein and a chemical compound , using one Opteron 275 2 . 2 GHz CPU , AutoDock took approximately 100 minutes on average with 100 genetic algorithm ( GA ) runs , while our method required less than 0 . 3 seconds . These computational time comparisons indicate that our method can perform a virtual screening of more than 19 million chemical compounds from the PubChem Compound database for any proteins in genome-wide scale and this immense screening task would be infeasible to accomplish with any of the existing docking methods . Therefore , our statistical approach can contribute as the first fast and rather accurate virtual screening tool for the drug discovery process . It can be followed by the application of more time-consuming but more informative approaches , such as docking analysis and molecular dynamics analysis , which can provide information regarding the binding affinities and the molecular binding mechanisms to outputs of the first screening . In another perspective , the re-evaluation of statistical prediction approaches by using 23 chemical compounds experimentally verified in this study showed that our proposed methods , which utilized information of both protein sequence and chemical structures , were superior to a conventional LBVS ( Ligand Based Virtual Screening ) method where only structures of specific chemical compounds were considered ( Figure 6 ) . As shown in Figure 6A , our proposed methods ( “one-layer SVM” , “two-layer SVM-subpos” and “two-layer SVM-allpos” ) achieved a higher recall rate at ranks higher than 500 compared to a conventional Ligand Based Virtual Screening method ( “only compound SVM” in Figure 6A ) . The fact that experimentally verified chemical compounds were identified at higher ranks in the pool by our proposed prediction models suggests that our proposed models were highly efficient with respect to the screening method . Figure 6B also shows that our proposed methods were more successful at discriminating between 15 experimentally verified binding and 8 non-binding ligands better than the LBVS method . These comparisons suggest that our proposed method utilizing information of protein sequences as well as chemical structures can be regarded as a more useful substitute for usual ligand-based virtual screening methods utilizing only chemical structures . Furthermore , the fact that the second computational prediction , or the use of feedback data , contributed to the discovery of novel ligands ( Figure 2B–D ) supports the utilization of statistical learning methods in virtual screening . Regarding the computational prediction method used in this paper , we made the method available to the public as a web-based service named COPICAT ( COmprehensive Predictor of Interactions between Chemical compounds And Target proteins; http://copicat . dna . bio . keio . ac . jp/ ) .
The DrugBank dataset was constructed from Approved DrugCards data , which were downloaded in February , 2007 from the DrugBank database [25] . These data consist of 964 approved drugs and their 456 associated target proteins , constituting 1 , 731 interacting pairs or positives . Given Np positive and Nn negative samples in known data and Mp positives and Mn negatives in additional or feedback data , a straightforward strategy for the integration of additional data into statistical training , such as SVM , is to train a statistical model based on a dataset consisting of Np+Mp positives and Nn+Mn negatives . When the two-layer SVM strategy is applied , another strategy of feedback and supplement involves the utilization of an additional model based on additional data . In this strategy , the second-layer SVM is trained on the basis of Np+Mp positives and Nn+Mn negatives , and a sample si in the second layer is represented as follows , Here , is an output of the additional model trained on the basis of Mp positives and Mn negatives . is an output of the first-layer SVM model j , and is a weighting factor . AutoDock 4 [3] was applied to the human androgen receptor ligand-binding domain ( PDB code; 2AM9 [31] ) and tested compounds whose 3D structure was generated by Obgen in the Open Babel package ver . 2 . 2 . 0 [32] or CORINA [33] . The conditions of AutoDock followed Jenwitheesuk and Samudrala , 2005 [34] . ARG752 of 2AM9 , which was considered important for the binding of androgens by the human androgen receptor [31] , was set to a flexible residue in AutoDock . | This work describes a statistical method that identifies chemical compounds binding to a target protein given the sequence of the target or distinguishes proteins to which a small molecule binds given the chemical structure of the molecule . As our method can be utilized for virtual screening that seeks for lead compounds in drug discovery , we showed the usefulness of our method in its application to the comprehensive prediction of ligands binding to human androgen receptors and in vitro experimental verification of its predictions . In contrast to most previous virtual screening studies which predict chemical compounds of interest mainly with 3D structure-based methods and experimentally verify them , we proposed a strategy to effectively feedback experimental results for subsequent predictions and applied the strategy to the second predictions followed by the second experimental verification . This feedback strategy makes full use of statistical learning methods and , in practical terms , gave a ligand candidate of interest that structurally differs from known drugs . We hope that this paper will encourage reevaluation of statistical learning methods in virtual screening and that the utilization of statistical methods with efficient feedback strategies will contribute to the acceleration of drug discovery . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"chemical",
"biology",
"mathematics/statistics",
"pharmacology/drug",
"development",
"computational",
"biology"
] | 2009 | Integrating Statistical Predictions and Experimental Verifications for Enhancing Protein-Chemical Interaction Predictions in Virtual Screening |
M . tuberculosis N-acetyl-glucosamine-1-phosphate uridyltransferase ( GlmUMtb ) is a bi-functional enzyme engaged in the synthesis of two metabolic intermediates N-acetylglucosamine-1-phosphate ( GlcNAc-1-P ) and UDP-GlcNAc , catalyzed by the C- and N-terminal domains respectively . UDP-GlcNAc is a key metabolite essential for the synthesis of peptidoglycan , disaccharide linker , arabinogalactan and mycothiols . While glmUMtb was predicted to be an essential gene , till date the role of GlmUMtb in modulating the in vitro growth of Mtb or its role in survival of pathogen ex vivo / in vivo have not been deciphered . Here we present the results of a comprehensive study dissecting the role of GlmUMtb in arbitrating the survival of the pathogen both in vitro and in vivo . We find that absence of GlmUMtb leads to extensive perturbation of bacterial morphology and substantial reduction in cell wall thickness under normoxic as well as hypoxic conditions . Complementation studies show that the acetyl- and uridyl- transferase activities of GlmUMtb are independently essential for bacterial survival in vitro , and GlmUMtb is also found to be essential for mycobacterial survival in THP-1 cells as well as in guinea pigs . Depletion of GlmUMtb from infected murine lungs , four weeks post infection , led to significant reduction in the bacillary load . The administration of Oxa33 , a novel oxazolidine derivative that specifically inhibits GlmUMtb , to infected mice resulted in significant decrease in the bacillary load . Thus our study establishes GlmUMtb as a strong candidate for intervention measures against established tuberculosis infections .
The cell wall , which contains a number of virulence determinants , is the first line of defence for survival of the pathogen in the hostile host environment [1] . The mycobacterial cell envelope includes three layers of cell membrane and a cell wall made up of peptidoglycan , mycolic acid , arabinogalactan and lipoarabinomannan ( LAM ) [2–4] . Most existing first line and second line drugs used to treat TB such as isoniazid , ethambutol , ethionamide and cycloserine , act on enzymes engaged in the synthesis of different cell wall components [5] . The current high mortality rates of infected individuals as well as increasing incidence of multidrug-resistant ( MDR ) and extensively drug-resistant ( XDR ) tuberculosis ( TB ) among patients underscore the importance of finding new targets for therapeutic intervention . GlmUMtb is a bi-functional enzyme , with acetyltransferase and uridyltransferase activities catalyzed by the C- and N- terminal domains respectively ( Fig 1A ) [6 , 7] . The carboxy-terminal domain of GlmUMtb transfers the acetyl moiety from acetyl CoA onto glucosamine-1-phosphate to generate N-acetylglucosamine-1-phosphate ( GlcNAc-1-P ) . The N-terminal uridyltransferase domain of GlmUMtb then catalyzes the transfer of UMP ( from UTP ) to GlcNAc-1-P to form UDP-GlcNAc ( Fig 1A ) [6] . The UDP-GlcNAc thus produced is among the central metabolites that is required for the synthesis of peptidoglycan , lipid A of LAM , arabinogalactan , Rha-GlcNAc linkers , mycothiol ( required for maintaining redox homeostasis ) [8–14] . The crystal structure of M . tuberculosis GlmU ( GlmUMtb ) displays two-domain architecture with an N-terminal α/β- like fold and a C-terminal left-handed parallel-β-helix structure [15 , 16] . Unlike its orthologs , GlmUMtb has a long carboxy-terminal tail which displays little secondary structure [17] . Results from transposon mutagenesis experiments have indicated glmUMtb to be an essential gene , supported by the fact that M . smegmatis is unable to grow in the absence of glmUsmeg [18–20] . However , no studies have addressed the question of whether both the activities of GlmUMtb are independently essential for the growth or survival of the bacterium . While the enzymes required for the synthesis of UDP-GlcNAc are well conserved among prokaryotes , they are very different from those found in eukaryotes , making GlmUMtb an attractive putative drug target [21 , 22] . Researchers have developed compounds that inhibit the activities of the orthologs of GlmUMtb ( GlmU from T . brucei , P . aeruginosa , E . coli , H . influenza and X . oryzae ) in vitro [23–30] . Bioinformatic analyses and kinetic modelling data advocate GlmUMtb to be a potential target for the development of suitable inhibitors [31] . In concurrence with these predictions , effective inhibitors have been developed against , the acetyltransferase and uridyltransferase domains of GlmUMtb [32 , 33] . However , the precise sites of inhibitor-protein interactions and the efficacy of the inhibitors ex vivo or in vivo have not been investigated . Subjecting Mtb cultures in vitro to gradual decrease of oxygen ( hypoxic stress ) results in reprogramming of metabolic pathways and up-regulation of stress response genes , and is considered to be an in vitro model for the dormancy [34 , 35] . The importance of GlmUMtb for growth under hypoxic conditions and in an in vivo infection model is yet to be investigated . In the present study we have generated a conditional gene replacement mutant of glmUMtb and used this mutant to investigate any role GlmUMtb may play in modulating the growth of the bacterium in vitro , ex vivo and in vivo . The data presented here demonstrate that GlmUMtb is a viable and promising target for therapeutic intervention against tuberculosis .
As the tetracycline-inducible system is an effective means to regulate gene expression [36] , we introduced the integration-proficient pST-KirT-glmU construct ( wherein glmUMtb gene was cloned under a promoter that shuts down upon ATc addition; S1A Fig ) into Mtb H37Rv ( Fig 1B ) . Whereas the expression of GlmUMtb from its native locus remained unaltered , the expression of FLAG-GlmUMtb in Rv::glmU strain was drastically compromised in the presence of ATc ( Western blot inset , Fig 1B ) . This merodiploid strain was transduced with temperature sensitive phage , and the fidelity of homologous recombination at the native locus was confirmed by amplification across the replacement junctions using appropriate primers ( Fig 1C ) . A comparison of GlmUMtb expression in the presence and absence of ATc revealed that the protein was not detectable by western blot analysis after 6 days of growth in the presence of ATc ( Fig 1D ) . While the growth of Rv∆glmU in the absence of ATc was similar to Rv , in the presence of ATc the growth was drastically compromised ( Fig 2A ) . A comparative analysis of growth by spotting of serially diluted cultures of Rv and Rv∆glmU grown in the presence versus absence of ATc showed that GlmUMtb depletion by addition of ATc led to complete inhibition of growth , with no growth detected after 6 days ( Fig 2B ) . Interestingly , analysis of GlmUMtb expression every 24 hours post-ATc addition uncovered significant reduction in GlmUMtb expression by the third day itself ( Fig 2C ) . These results indicate that GlmUMtb is required for the Mtb survival . To determine the impact of GlmUMtb depletion on cellular morphology we carried out SEM and TEM imaging analysis of Rv and Rv∆glmU cells grown for three days in the absence or presence of ATc . SEM analysis revealed severe morphological perturbations in the absence of GlmUMtb , with the bacilli showing wrinkled surface and fused cells ( Fig 2D ) . TEM analysis showed that whereas in Rv and Rv∆glmU cell wall structure and thickness are comparable , there was a marked decrease in cell wall thickness in Rv∆glmU cells where GlmUMtb was not expressed ( cells grown in the presence of ATc; Fig 2E and 2F ) . Next we used the Wayne model to investigate the consequence of GlmUMtb depletion on the dormant bacteria under hypoxic conditions [35] . Accordingly , hypoxia was established and maintained for 42 days with depletion of GlmUMtb or addition of INH for either 22 days , or for 2 days ( Fig 3A , line diagram ) . In agreement with previous reports , we observed that bacteria were tolerant to INH under hypoxic conditions [37] ( Fig 3C ) , with a thicker cell wall being observed under hypoxic conditions compared with the normoxic cultures ( Fig 3D and 3E ) . Depletion of GlmUMtb for 22 days resulted in complete clearance of growth ( Fig 3B ) , which was also reflected in severe morphological perturbations and drastic reduction in cell wall thickness ( Fig 3D and 3E ) . Significantly , GlmUMtb depletion for as less as 2 days decreased cell viability by three orders of magnitude ( Fig 3B ) and decrease in cell wall thickness ( ~18%; Fig 3D and 3E ) . Taken together , the data suggests that the absence of GlmUMtb in hypoxic condition leads to aberrant cell wall thickness and morphology , eventually leading to the death of the cell . Biochemical investigations have shown that the N-terminal fragment ( 1–352 amino acids ) and C-terminal fragment ( 150–495 amino acids ) of GlmUMtb can independently undertake uridyltransferase and acetyltransferase activities respectively ( Fig 4A and 4B ) [15 , 17] . The active site residues that are necessary for these activities have also been identified ( Fig 4A and 4B ) [17] . To investigate if both activities are essential for cell survival , we have generated previously reported truncation mutants GlmU-N and GlmU-C [38] . We also generated GlmUK26A and GlmUH374A , the uridyltransferase and acetyltransferase active site mutants , and GlmUDM wherein both the active site residues were concomitantly mutated . GlmUMtb wild type and mutant proteins were purified ( Fig 4C ) and their uridyltransferase and acetyltransferase activities were assayed . While GlmU-C and GlmUK26A mutants showed acetyltransferase activity , as expected they did not show any uridyltransferase activity ( Fig 4D ) . On the other hand GlmU-N and GlmUH374A had uridyltransferase activity but not the acetyltransferase activity ( Fig 4D ) . As expected the double mutant did not have either uridyl or acetyltransferase activity ( Fig 4D ) . Next complementation assays using one or other truncations / active site mutants were carried out . The FLAG-GlmUMtb and the complemented untagged wt-GlmUMtb proteins were found to be expressed at similar levels ( Fig 4E ) . The episomally expressed wt-GlmUMtb could rescue the Rv∆glmU phenotype in the presence of ATc ( Fig 4F ) . Contrastingly , while the various GlmUMtb mutant proteins were expressed at levels comparable to that of FLAG-GlmUMtb ( Fig 4E ) ; none of them rescued the growth defects of the Rv∆glmU strain in the presence of ATc ( Fig 4F ) . These results indicate that both uridyltransferase and acetyltransferase activities of GlmUMtb are essential for pathogen survival and imply that the only source of the metabolites GlcNAc-1-P and UDP-GlcNAc is through the GlmUMtb mediated synthesis pathway . Mtb cells devoid of an intact cell wall have been found to be capable of surviving inside the host [39 , 40] . Some pathogens have been reported to resort to cell wall “recycling” for the synthesis of UDP-GlcNAc , and others have been known to utilize GlcNAc from the host for this purpose [41–44] . However , such mechanisms have not yet been reported in Mtb . To investigate these possibilities we examined the impact of GlmUMtb depletion on survival of the pathogen in the host . Using an ex vivo THP-1 infection model we observed ~80% phagolysosome fusion in the absence of GlmUMtb ( Fig 5A and 5B; compare Rv∆glmU with Rv∆glmU +ATc ) . This was also reflected in the survival pattern of the pathogen upon depletion of GlmUMtb ( Fig 5C ) , with survival being strongly compromised in absence of GlmUMtb . The impact of GlmUMtb depletion was evident as early as 24 h post-infection , with a dramatic drop in survival by 48 hours post-infection ( Fig 5C ) . The consequences of GlmUMtb depletion on survival of the pathogen in vivo were evaluated using guinea pig infection model . CFUs obtained 24 h after infection suggested efficient and equivalent implantation of both wild type and mutant bacilli in the lungs of guinea pigs ( Fig 5D ) . Discrete bacilli were observed in the lungs of guinea pigs infected with Rv and Rv∆glmU 28 days post-infection ( Fig 5E and 5F ) . In contrast , the lungs of the guinea pigs infected with Rv∆glmU in the presence of doxycycline were clear ( Fig 5E and 5F ) . In addition splenomegaly was significantly reduced upon depletion of GlmUMtb ( Rv∆glmU + Dox; S2A and S2B Fig ) . Whereas the bacillary load in the lungs and spleen of guinea pigs infected with Rv and Rv∆glmU were comparable , we did not detect any bacilli when the Rv∆glmU infected guinea pigs were administered Dox ( Fig 5D ) . In accordance with these observations , while the gross pathology of lungs infected with Rv and Rv∆glmU displayed considerable granulomatous architecture , normal lung parenchyma was observed upon GlmUMtb depletion ( Fig 5E ) . These results suggest that the presence of GlmUMtb is obligatory for mycobacteria to survive in the host . It was apparent from the data presented above that the addition of ATc or Dox at the time of inoculation or at the time of infection does not allow mycobacterial cell growth or survival in the host . In the ideal candidate for therapeutic intervention , inhibiting the activity of/ depleting the enzyme at any stage of the infection should result in pathogen clearance . We assessed this parameter of GlmUMtb by providing ATc at different stages of bacterial growth ( early , log and stationary phases ) and investigating its influence on cell survival in liquid cultures . Addition of ATc to Rv∆glmU cultures on the 2nd , 4th or 6th day after inoculation significantly thwarted growth ( Fig 6A ) . A similar analysis of bacterial growth by serial dilution of cultures followed by spotting on solid medium also revealed that viability was compromised by ~2 log fold 48 h after the addition of ATc , indicating that GlmUMtb depletion negatively impacted cell survival regardless of which stage of cell growth it was depleted at ( S3A Fig ) . The influence of GlmUMtb depletion on an established ex vivo infection was estimated by providing ATc 24 h post-infection in a THP-1 infection model . As expected the bacillary load in THP-1 cells infected with Rv and Rv∆glmU were similar at 0 and 24 h after infection ( Fig 6B ) . In contrast , while at 96 h post-infection the bacillary load for Rv and Rv∆glmU- infected THP-1 cells remained the same , the addition of ATc to Rv∆glmU- infected THP-1 cells 24 h after infection decreased the pathogen load by ~2 . 5 log fold , indicating that the reduction of GlmUMtb levels impacts pathogen survival even in an established ex vivo infection ( Fig 6B ) . We extended this investigation to analyze the effect of GlmUMtb depletion from a fully-infected lung using murine infection model . As anticipated , the bacillary load in the lungs of mice infected with Rv and Rv∆glmU were comparable both on Day 1 and on Day 28 . Administration of Dox to Rv∆glmU infected mice for the next 56 days ( Day 28 to Day 84 ) drastically decreased the CFUs in the lungs ( Fig 6C ) and the pathogen was completely cleared from the spleen ( S3B Fig ) . Unlike the lungs of mice infected with Rv and Rv∆glmU , mice infected with Rv∆glmU to whom Dox was administered displayed a total absence of lesions and granulomas in the lungs ( Fig 6D and 6E ) . Collectively , these data suggest a fundamental role for UDP-GlcNAc , the end product of the GlmUMtb -mediated enzymatic reaction , in modulating the persistence of Mtb infection . In addition to the acetyltransferase and uridyltransferase active site pockets , GlmUMtb also contains an allosteric site . Binding of any suitable molecule/inhibitor to the allosteric site would prevent the conformational change essential for GlmUMtb uridyltransferase catalytic activity . To target the allosteric site on GlmUMtb we drew on crystal structure data of H . influenza GlmU ( GlmUHI ) bound to its allosteric small molecule inhibitor ( S4A and S4B Fig ) [27] . Alignment of the GlmUMtb and GlmUHI allosteric pocket residues suggested that the interacting residues were conserved between the two proteins ( S4C and S4D Fig ) . The Asinex database was screened against shape as described ( S5A Fig ) and the resulting 43 hits were biochemically characterized for their ability to inhibit GlmUMtb uridyltransferase activity . One of the promising molecules was used for further structural optimization ( S5A Fig ) . Of the 53 structurally optimized compounds one molecule , namely ( 4Z ) -4- ( 4-benzyloxybenzylidene ) -2- ( naphthalen-2-yl ) -1 , 3-oxazol-5 ( 4H ) -one ( Oxa33; Synthesis scheme provided in Figs 7A and S5B ) , was found to be an efficient inhibitor of GlmUMtb activity with an IC50 of 9 . 96±1 . 1 μM ( Fig 7B ) . Isothermal titration analysis suggested an adequately high affinity binding for the compound ( Ka = 2 . 35×106 M-1 ) , with a binding stoichiometry of 0 . 7 ( S6A Fig ) . We sought to identify the residues in GlmUMtb that are critical for interacting with Oxa33 . Docking and MD simulation studies revealed polar , non-polar and hydrophobic interactions between Oxa33 and the allosteric site residues ( Figs 7C , S6B and S6C ) . Based on the obtained data a panel of GlmUMtb proteins each carrying a single mutation was created , the mutant proteins were purified ( S6D Fig ) , and their uridyltransferase activity assayed . While all the mutants had similar levels of uridyltransferase activity there was a substantial increase in their IC50 values , suggesting a loss of interaction with Oxa33 ( Fig 7D and 7E ) . To decipher the mechanism of Oxa33 mediated inhibition of uridyltransferase activity , we superimposed the GlmUMtb-Oxa33 complex with the unbound GlmUMtb structure . Upon Oxa33 binding , the loop regions ( in the range of 3–6 Å ) at the uridyltransferase active site undergo significant conformational changes , decreasing the active site volume , which results in occlusion of the substrates ( Fig 7F ) . Differential scanning fluorimetry ( DSF ) analysis of GlmUMtb in the presence of Oxa33 showed a 3°C shift in protein melting temperature ( Tm ) validating the conformational changes ( S5E Fig ) . Interestingly we also observed much higher relative fluorescence units ( ~10000 vs 2500 ) in the presence of Oxa33 , which is likely due to the compound induced structural changes facilitating increased binding of the dye ( S6E Fig ) . Together , these data demonstrate that Oxa33 binds to the allosteric site at N-terminal domain of GlmUMtb and inhibits its uridyltransferase activity by causing structural changes . Subsequently we investigated the ability of Oxa33 to inhibit the in vitro growth of Mtb H37Rv . Oxa33 inhibited the in vitro growth of Mtb H37Rv with a minimum inhibitory concentration ( MIC ) of ~75 μM ( ~30 μg / ml ) and a maxium bacteriocidal concentration ( MBC ) of ~150 μM ( ~60 μg / ml ) . To ascertain if this inhibitory effect was due to inhibition of GlmUMtb activity we overexpressed GlmUMtb in the cells prior to drug treatment and determined the effect of this on the MIC value ( Figs 8A and S6A ) . Whereas the inhibition of growth in the presence of INH was similar with or without GlmUMtb overexpression in the cells ( Fig 8A , lower panel ) , Oxa33 failed to inhibit cell growth even at concentrations as high as 150 μM ( 60 μg/ ml ) ( Fig 8A , upper panel ) . Interestingly when sub lethal concentration of Oxa33 was provided , the MIC of INH decreased from 32 to 16 ng/ml ( S7B Fig ) . The impact of Oxa33 on THP1 cells 24 h after infection with either Rv or Rv::glmUtet-on was also investigated . In concurrence with the in vitro growth data , overexpression of GlmUMtb alleviated Oxa33-mediated clearance of Mtb from THP-1 cells ( Figs 8B , S6D and S6E ) . These results suggest that the inhibition of mycobacterial growth by Oxa33 is specifically due to inhibition of endogenous GlmUMtb . Finally , we analysed the efficacy of Oxa33 in clearing bacilli from infected lungs using a murine infection model . Oxa33 compound is highly hydrophobic in nature . After trying many solvents , we could successfully resuspend it in 2 . 5% Tween-80 . Prior to performing the experiments we examined the maximum dose tolerance and survival analysis to determine the toxicity ( S8A and S8B Fig ) . Based on the data obtained we chose 50 mg / kg as the appropriate dose . Since it was difficult to predict the fate of Oxa33 during the process of digestion , we avoided using the oral administration route . We chose intra peritoneal route for administering the compound as the intravenous ( I/V ) injection of Tween 80 ( solvent ) in the animals was known to cause hypersensitivity and anaphylactic shock [45 , 46] . Groups of mice were infected with Rv and were treated with vehicle , INH , or Oxa33 at 28 days post-infection , for a duration of 56 days ( Fig 8C , line diagram ) . Compared with the vehicle-treated group where we observed a marginal increase in bacillary load , a significant reduction in the bacillary load was observed in the lungs and spleen of both , INH- and Oxa33-treated groups ( ~4 and 2 . 5 log fold , respectively for lungs ) ( Figs 8C and S9A ) . This was also reflected in the gross pathology and histopathology of lungs ( Fig 8D and 8E ) . Although in vitro MBCvalues of Oxa33 was ~150 μM ( 60 μg/ml ) , it seems to be a relatively more efficacious in vivo , which could be due to its accumulation in the lungs of the infected mice . To investigate this possibility uninfected mice were treated with 50 mg/kg Oxa33 for a period of 3 weeks or 8 weeks . In order to estimate the concentration of Oxa33 in the lung , we first determined the absorbance spectra for Oxa33 , which gave a clear peak at 401 nm ( S10A Fig ) . We determined the A401 at different concentrations of Oxa33 and the standard curve was plotted ( S10B and S10C Fig ) . Oxa33 was extracted from the lungs and its concentration was determined . The concentrations of Oxa33 in the lungs were in the range of ~200–300 μg /lung at 3 weeks and ~800–1300 μg/lung at 8 weeks ( S10D Fig ) . The accumulation of Oxa33 in the lungs is ~13 to 18 fold higher than the MBC values , which may be the reason for higher potency of Oxa33 in vivo compared with the in vitro experiments . Taken together , the results presented in this study establish GlmUMtb to be an effective target against which new sets of inhibitors may be developed .
Cell wall provides the structural rigidity and protects bacteria from various environmental and physiological insults . Biosynthesis of the cell wall of bacteria is a complex process requiring enzymes localized to different cellular compartments [47] . Due to the essentiality of the enzymes involved they are considered attractive targets for anti-microbial therapies . The majority of the first line and second line anti-tuberculosis drugs from the existing regimen target enzymes involved in cell wall synthesis [5] . These include Isoniazid and Ethionamide targeting enoyl-[acyl-carrier-protein] reductase and inhibiting mycolic acid synthesis , Ethambutol targeting arabinosyl transferase and inhibiting arabinogalactan biosynthesis , and Cycloserine targeting D-alanine racemase and ligase , which inhibits peptidoglycan synthesis [5] . However most of these drugs are not very effective against dormant/ latent Mtb [48] . UDP-GlcNAc is a critical metabolite both in prokaryotes and eukaryotes . In eukaryotes it is mainly utilized for O- or N- GlcNAcylation , sialic acid biosynthesis and hylauronic acid biosynthesis [49–51] . In addition to the peptidoglycan synthesis [10] , in gram negative bacteria UDP-GlcNAc is required for the synthesis Lipid A subunit of lipopolysaccharide [52] and in gram positive bacteria it is required for Rha-GlcNAc linker [53] , arabinogalactan [54] , teichioc acid synthesis [55] . In few prokaryotes , UDP-GlcNAc has also been shown to be required for sialic acid [56] , N-GlcNAcylation [57] and poly ( -GlcNAc- ) n [58] . GlmUMtb , an enzyme with dual activity , synthesizes a core metabolite necessary for the synthesis of cell wall peptidoglycan , UDP-GlcNAc [6] . Interestingly , we found that depleting GlmUMtb during both normoxic and hypoxic growth resulted in substantial decrease in cell viability ( Fig 3 ) . This may be due to the requirement of UDP-GlcNAc , which in addition to participating in cell wall synthesis is also required for other cellular processes such as mycothiol biosynthesis ( to maintain redox homeostasis ) [14 , 59] . However , the TEM data clearly shows decreased cell envelop thickness even in hypoxic conditions ( Fig 3 ) . Although the CFUs do not change significantly the cells may be undergoing significant replication , which might be balanced by death [60] . Alternatively , new cell envelop may be required even if the bacteria are not replicating . Thus one can rule out the possibility that decreased viability may well be due to requirement of UDP-GlcNAc for the cell envelop synthesis . While UDP-GlcNAc is a critical metabolite for both prokaryotes and eukaryotes , the enzymes involved in its de novo synthesis are significantly different [10] . In addition , both prokaryotes and eukaryotes can utilize GlcNAc from different sources to synthesize UDP-GlcNAc through salvage pathways [61–63] ( S11 Fig ) . Capnocytophaga canimorsus , a member bacteria from Bacteroidetes phylum lacks endogenous GlmM and GlmU required for the synthesis of GlcNAc and it instead relies on GlcNAc obtained from forages glycans from the host mucin and N-linked glycoproteins [42] . Depending on the enzymes of the salvage pathway present in the bacterial system , it would require either both the activities or only the uridyltransferase activity of GlmUMtb for UDP-GlcNAc synthesis . Till date the presence of alternate salvage pathway in Mtb has not been demonstrated . However , even with an operating salvage pathway GlmUMtb is essential for the utilization of host GlcNAc to form UDP-GlcNAc ( S11 Fig ) . In line with this , we find that depletion of GlmUMtb during ex vivo or in vivo infection either at the start or after infection has been definitively established leads to clearance of pathogen . GlmUMtb and the acetyltransferase and uridyltransferase enzymes found in eukaryotes share very little sequence similarity . Although efforts have been made by different groups to target bacterial GlmU proteins , the specificity of these inhibitors for GlmU in vivo have not been established [23–30] . Most GlmU inhibitors characterized till date target either the acetyl- or uridyltransferase active sites . In contrast , inhibitors of GlmUHI target the allosteric site near the uridyltransferase active site [27] . The interaction of the inhibitor with the enzyme via this allosteric site perturbs the active site conformation of the protein , thus inhibiting uridyltransferase activity [27] . In the present study we have used shape based designing and developed a novel oxazolidine molecule , Oxa33 , and characterized its ability to bind to the GlmUMtb allosteric site . MD simulation and mutation of critical interacting residues to defined the possible allosteric site residues required for Oxa33 binding ( Fig 7 ) . DSF ( S6E Fig ) and structural superimposition ( Fig 7 ) supports that inhibition of uridyltransferase activity is due to structural changes in the N-terminal domain of GlmUMtb . Further in order to determine the specificity of Oxa33 , GlmUMtb over expressing strains of Rv was used to determine the MIC . Both in vitro and ex vivo results ( increased MIC or MBC ) validate that Oxa33 specifically binds to GlmUMtb inside the bacteria . Administrating the Oxa33 to fully infected ( 28 days ) mice resulted in partial ablation of pathogen load in the lungs . Taken together results presented here demonstrates that GlmUMtb is a viable and promising target for therapeutic intervention and Oxa33 can be pursued as a lead molecule , which needs to be developed further to improve its efficacy .
Restriction enzymes and Phu DNA polymerase were purchased from New England Biolabs . pENTR/directional TOPO cloning kit ( Invitrogen ) , pQE2 ( Qiagen ) , were procured from the respective sources . Analytical grade chemicals and oligonucleotide primers were procured from Sigma . Malachite green phosphate assay kit ( POMG-25H ) was purchased from BioAssay System ( Gentaur ) . Electron microscopy reagents were purchased from Electron Microscopy Sciences . Media components were purchased from BD Biosciences . Doxycycline hydrochloride was purchased from Biochem pharmaceutical . The hexa-His tag in the pST-KiT construct[15] was replaced with an N-terminal FLAG tag , and the tetracycline repressor gene ( tetR ) was replaced with a reverse tetR ( r-tetR ) from pTC28S15-OX [64] to create plasmid pST-KirT . To generate the integrating shuttle plasmid pST-KirT-glmUMtb , the glmUMtb gene was excised from pQE2-glmUMtb using NdeI-HindIII digestion and was subcloned into the corresponding sites on pST-KirT . The resulting pST-KirT-glmU construct expresses GlmUMtb in the absence of inducer ATc . Upon addition of ATc , ATc binds to the r-TetR repressor resulting in the conformational change that would allow it to bind to the operator seqeunces in PmyctetO ( S1 Fig ) [64] . The integration-proficient plasmid containing the inducible glmUMtb gene was electroporated into mycobacterial cells to create the merodiploid strain Rv::glmUMtb . 5’ and 3’ genomic flank sequences of glmUMtb ( approximately 1 kb on either side ) were amplified , the amplicons digested with PflMI , and ligated with the antibiotic resistance cassette along with the oriE and cosλ fragments generated from pYUB1474 construct , to generate the allelic exchange substrate ( AES ) [65] . The AES was linearized using the unique PacI site and then cloned into temperature sensitive shuttle phagemid phAE159 at the PacI site . A conditional gene replacement mutant of RvΔglmU was created from the merodiploids with the help of specialized transduction methodology ( S1A Fig ) [66] . RvΔglmU recombinants obtained were analyzed by PCR amplification to verify the fidelity of the recombination event . H37Rv ( Rv ) and RvΔglmU cultures were grown in Middlebrook 7H9 medium supplemented with 10% ADC ( albumin , dextrose and catalase ) , or in 7H11 medium supplemented with 10% OADC ( oleic acid , ADC ) . To analyze bacterial growth in vitro , Rv and Rv∆glmU mutant bacterial cultures were inoculated at A600 of 0 . 1 , in the presence or absence of anhydrotetracycline ( ATc ) , and A600 was measured every 24 h for 6 or 8 days . For spotting analysis , cells were harvested by centrifugation , washed twice with PBST ( 0 . 05% Tween 80 ) to remove ATc , resuspended in 7H9 medium , and serially diluted in the same medium , followed by spotting 10 μl aliquots of the various cell dilutions on 7H11 agar plates to assess cell viability . To determine the impact of GlmUMtb depletion during hypoxia in Rv and RvΔglmU strains , we established hypoxia in 1 . 5 ml HPLC tubes or 500 ml flasks with penetrable caps , following modified Wayne model [35] . ATc ( 2 μg/ml ) or isoniazid ( INH ) ( 50 ng/ml ) were injected into the cultures at different time points and the number of CFUs were determined after 42 days . Scanning and transmission electron microscopy ( SEM & TEM ) analysis of Rv and Rv∆glmU mutant grown in the presence or absence of ATc were performed as described earlier [67] . Transmission electron microscopy was performed using standard protocols . Briefly , bacteria was fixed in 2 . 5% gluteraldehyde and 4% paraformaldehyde , dehydrated in graded series of alcohol and embedded in Epon 812 resin . Ultrathin sections were cut and stained with uranyl acetate and lead citrate [68] . SEM images were procured using Carl Zeiss Evo LS scanning electron microscope , and TEM images were captured using Tecnai G2 20 twin ( FEI ) transmission electron microscope . Site directed mutations of glmUMtb were generated with the help of overlapping PCR and the amplicons were cloned into NdeI-HindIII sites of pQE-2 , pNit and pST-KT vectors [15 , 69] . The tetracylin repressor ( TetR ) expressed from the plasmids binds to the operator sequence in the promoter PmyctetO in the absence of ATc ( S1B Fig ) [70] . Addition of ATc to TetR alleviates the repression thus inducing the expression of GlmU . pST-KT-glmU was electroporated into Rv to generate Rv::glmUtet-on strain . pNit-glmU ( wild type and mutated ) constructs were electroporated into Rv∆glmU to generate Rv∆glmU::glmUwt/mutant strains . Rv and Rv∆glmU::glmUwt/mutant strains were grown in the presence or absence of ATc as described above . GlmUMtb was expressed and purified using plasmid pQE2-GlmUMtb , as described earlier [15] . Whole cell lysates ( WCL ) isolated from Rv , Rv∆glmU and Rv∆glmU::glmUwt/mutant strains that had been grown for 5 days in presence or absence of Atc , were resolved on 10% SDS-PAGE , transferred to nitrocellulose membrane , and probed with anti-GlmUMtb and anti-GroEL1 antibodies as described earlier [15 , 67] . THP1 infection experiments were carried out with either unlabelled or FITC-labelled Rv and Rv∆glmU strains at 1:10 MOI , as described earlier [71] . For examination of cells under a fluorescence microscope , infected cells ( 48 h post-infection ) were labelled with Lyso Tracker red DND 99 dye ( 50 nM ) and mounted with Antifade ( Invitrogen ) mounting agent . To determine CFUs per infected cell , the infected cells were lysed in PBS containing 0 . 1% TritonX-100 for 15 min and different dilutions were plated on OADC-containing 7H11 agar plates . For animal infection experiments , Rv and Rv∆glmU strains grown till A600 of 0 . 6 were used to infect 3 to 4 week old guinea pigs or ~ 2 month old mice as described previously [72 , 73] . We initially used guinea pig model system as it has robust immune response . However , for the remaining experiments we chose to use Balb/C mice model of infection as the cost associated with performing the experiments and the amount of Oxa33 required for guinea pig experiments was prohibitive . To determine the implantation dosage , the bacillary load in the lungs of guinea pigs or mice was determined 24 h post-infection . To investigate the impact of glmUMtb depletion on survival of the pathogen , doxycycline hydrochloride ( Dox , 1 mg/ kg with 5% dextrose in drinking water ) was provided to Rv and Rv∆glmU-infected animals as indicated , either from the time of the infection ( guinea pig experiment ) , or 4 weeks post-infection ( mice infection experiments ) . To assess the impact of INH or Oxa33 treatment on pathogen survival , Rv-infected mice ( 4 weeks post-infection ) were supplied with INH ( 25 mg/ kg body weight , with 5% dextrose in drinking water ) or Oxa33 ( 50 mg/ kg body weight , with 2 . 5% Tween 80 , injected intra peritoneally ) every third day for 8 weeks . Bacillary loads in the lungs and spleens of infected guinea pigs and mice were determined 4 weeks and 12 weeks post-infection . Histopathological evaluation of the harvested organs was performed as described earlier [67 , 72 , 73] . ROCS ( Rapid Overlay of Chemical Structures ) , a shape based technique for rapid similarity analysis was used to assess the compounds . Gaussians and shape tanimoto were used to assess the volume and shape overlaps of the compounds , respectively . As the chemical functionality is critical , the chemical feature based similarity was also considered using ROCS colour score whose force field was composed of SMARTS patterns of the chemical functions [74 , 75] . The shape tanimoto score and scaled color score were considered during the selection of the compounds for further virtual screening . The compounds selected were subjected to molecular docking studies using Glide v5 . 8 of Schrödinger molecular modelling suite 2012 ( Glide v5 . 8 , Schrödinger ) . The compounds were subjected to a series of docking protocols–high throughput virtual screening ( HTVS ) , standard precision ( SP ) and extra precision ( XP ) docking . As the docking progresses from HTVS to XP , the algorithm differs , which starts from a simple docking of compounds and ends with docking protocol with high precision and parameterization while cutting off the number of compounds . To the glycine solution ( 3 . 0 g , 39 . 89 mmol ) in water under constant stirring at 0°C , NaOH ( 3 . 19 g , 79 . 78 mmol ) was added . This was followed few minutes later by the addition of 1-naphthoyl chloride ( 7 . 20 mL , 47 . 86 mmol ) in 1 , 4-Dioxane ( 20 ml ) and the contents were stirred at room temperature for 6 h . The reaction mixture was concentrated to half the volume and 60 ml EtOAc was added . The EtOAc layer was washed with sat NaHCO3 ( 2 × 30 mL ) followed by H2O ( 2 × 20 mL ) . The separated organic layer was dried and concentrated over anhydro Na2SO4 to obtain solid compound , which was washed with hexanes to get 2- ( 1-naphthamido ) acetic acid ( 8 . 30 g , 90% ) as a white solid ( M . P . 152°C1 ) . 2- ( 1-naphthamido ) acetic acid ( 2 . 0 g , 8 . 73 mmol ) , NaOAc ( 0 . 21 g , 2 . 62 mmol ) and 4-benzyloxybenzaldehyde ( 1 . 85 g , 8 . 73 mmol ) were taken in acetic anhydride and heated at 106°C for 3 h . The solid formed were filtered and washed with water to remove traces of acetic anhydride , and ethanol to remove unreacted aldehyde and other organic impurities . Final compound 4- ( 4- ( benzyloxy ) benzylidene ) -2- ( naphthalen-1-yl ) oxazol-5 ( 4H ) -one ( Oxa33; 3 . 14 g , 88% ) , purified as a yellow solid , was confirmed with nuclear magnetic resonance ( NMR ) [76] . To investigate the binding of Oxa33 to GlmUMtb , we performed Isothermal Titration Calorimetry ( MicroCal 2000 VP-ITC , GE Healthcare ) [28] . Oxa33 was re-suspended in dialysis buffer ( 25 mM Tris pH 7 . 4 , NaCl 140 mM and 15% glycerol ) , 100 μM of MgCl2 containing 2% DMSO . 625 μM of Oxa33 was injected for titrations from syringe ( rotating at 307 rpm ) into ITC cell containing 25 μM of GlmU or blank buffer at 25°C . Each injection lasted for 20 sec with 300 sec interval between every step . The quantity of heat associated by every injection was calculated by combining the area beneath every heat burst curve ( microcalories/second vs . seconds ) . Data was corrected for the buffer signal and fitting was done by one-site binding model . Origin software ( version 7 . 0 ) was used to obtain different thermodynamic binding parameters . Oxa33 was evaluated for its cytotoxic activity in THP1 cells with the help of alamar blue assay . Serially diluted inhibitors ( in 2 . 5% DMSO ) incubated with 5 x 103 differentiated THP-1 cells in 96 well plates for 3 days . After 3 days cells were incubated for 5 h with 10 μl of alamar blue and color development was measured using micro-plate reader at 570 nm . Molecular dynamics ( MD ) simulation for the protein-ligand complex was carried out for a time scale of 20 ns so as to analyze the stability of molecular interactions between ligand and protein employing Newton’s Laws of Motions . Desmond molecular dynamics system v3 . 1 was used for carrying out the simulations employing OPLS-AA force field [77] . The protein-ligand complex was solvated using TIP3P water model which was setup as an orthorhombic solvent box , keeping a cut-off of 10 Å from any solute atom in all directions [78] . Na+ counter ions were added in order to neutralize the system . A cut-off of 14 Å was maintained for calculating the solvent-solvent and solute-solvent non-bonded interactions . Initially , the system was minimized keeping the convergence threshold criteria of 1 . 0 kcal . mol-1 . Å-1 so as to allow the adjustment of atoms to the system environment . A simulation for each system was performed using isothermal-isobaric ensemble ( NPT ) including a relaxation process . Under NPT , the system was simulated for 12 ps using a Berendsen thermostat and a Berendsen barostat with temperature of 10K and a pressure of 1 atm . The later step of relaxation protocol included the simulation of the system for 24 ps with a temperature of 300 K and 1 atm pressure with and without restraints on solute heavy atoms . M-SHAKE algorithm was used with an integration time step of 2 fs for rearranging the hydrogen bonds in the simulation [79] . The temperature and pressure of the system were maintained at 300 K and 1 . 013 atm respectively . The molecular dynamics simulation was run for 20 ns recording the trajectory frames at an interval of every 4 . 8 ps and the trajectory analysis was carried out using the Simulation Event Analysis of Desmond . Uridyltranferase assays were performed using malachite green phosphate detection kit as described previously [17] . Acetyltransferase activity of GlmUMtb was carried out in the presence of 500 μM each of GlcN-1-P and acetyl-CoA in a 30 μl reaction volume for 30 min at 30°C as described earlier [80] . To determine the percent inhibition by different compounds the enzyme was preincubated with either 5% DMSO or 100 μM compounds for 30 min prior to performing uridyltransferase activity assays . In order to determine the IC50 values , GlmUwt/mutant proteins were preincubated with different concentrations of Oxa33 compound for 30 min followed by the uridyltransferase assay . To determine minimum inhibitory concentration ( MIC ) , 5x105 bacteria of Rv or Rv::glmUtet-on ( overexpressing GlmUMtb ) cultures ( grown in the presence or absence of 2 μg/ml ATc ) were mixed with 100 μl of 2 . 5% DMSO or different concentrations of Oxa33/ INH in 96-well plates , and incubated at 37°C for 6 days . After 6 days , 40 μl of resazurin dye ( 0 . 02% in 5% Tween-80 ) was added to each well and the colour change was observed after 12 h . Experimental protocol for the animal experiments was approved by the Institutional Animal Ethics Committee of National Institute of Immunology , New Delhi , India ( the approval number is IAEC# 315/13 ) . The approval is as per the guidelines issued by Committee for the Purpose of Control and Supervision of Experiments on Animals ( CPCSEA ) , Govt . of India . Student’s t-test ( two tailed non parametric ) was used to analyze the significance of cell wall thickness , THP1 and animal infection experimental results . SigmaPlot version 10 . 0 and GraphPad Prism version 5 . 0 was used for the statistical analysis and for plotting the results . | The synthesis of the Mtb cell wall involves a cascade of reactions catalyzed by cytosolic and cell membrane-bound enzymes . The reaction catalyzed by GlmUMtb ( an enzyme with acetyltransferase and uridyltransferase activities ) generates UDP-GlcNAc , a central nucleotide-sugar building block of the cell wall . Apart from cell wall synthesis UDP-GlcNAc is an essential metabolite participating in other cellular processes including disaccharide linker and mycothiol biosynthesis . GlmUMtb shares very little sequence similarity with eukaryotic acetyltransferase and uridyltransferase enzymes . Many pathogens have alternative pathway ( s ) for foraging GlcNAc from the host . The present study was undertaken to see the effects of depleting GlmUMtb on pathogen survival in the host animal . We have generated a conditional gene replacement mutant of glmUMtb and find that depletion of GlmUMtb at any stage of bacterial growth or in mice infected with Mtb including a well-established infection , results in irreversible bacterial death due to perturbation of cell wall synthesis . We have developed a novel anti-GlmUMtb inhibitor ( Oxa33 ) , identified its binding site on GlmUMtb , and shown its specificity for GlmUMtb . The study demonstrates that GlmUMtb is a promising target for therapeutic intervention and Oxa33 can be pursued as a lead molecule . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Depletion of M. tuberculosis GlmU from Infected Murine Lungs Effects the Clearance of the Pathogen |
In budding yeast , the major regulator of the mitotic exit network ( MEN ) is Tem1 , a GTPase , which is inhibited by the GTPase-activating protein ( GAP ) , Bfa1/Bub2 . Asymmetric Bfa1 localization to the bud-directed spindle pole body ( SPB ) during metaphase also controls mitotic exit , but the molecular mechanism and function of this localization are not well understood , particularly in unperturbed cells . We identified four novel Cdc5 target residues within the Bfa1 C-terminus: 452S , 453S , 454S , and 559S . A Bfa1 mutant in which all of these residues had been changed to alanine ( Bfa14A ) persisted on both SPBs at anaphase and was hypo-phosphorylated , despite retaining its GAP activity for Tem1 . A Bfa1 phospho-mimetic mutant in which all of these residues were switched to aspartate ( Bfa14D ) always localized asymmetrically to the SPB . These observations demonstrate that asymmetric localization of Bfa1 is tightly linked to its Cdc5-dependent phosphorylation , but not to its GAP activity . Consistent with this , in kinase-defective cdc5-2 cells Bfa1 was not phosphorylated and localized to both SPBs , whereas Bfa14D was asymmetrically localized . BFA14A cells progressed through anaphase normally but displayed delayed mitotic exit in unperturbed cell cycles , while BFA14D cells underwent mitotic exit with the same kinetics as wild-type cells . We suggest that Cdc5 induces the asymmetric distribution of Bfa1 to the bud-directed SPB independently of Bfa1 GAP activity at anaphase and that Bfa1 asymmetry fine-tunes the timing of MEN activation in unperturbed cell cycles .
In eukaryotes , mitotic entry is driven by a rise in cyclin-dependent kinase ( Cdk ) activity , which is required for the formation of a bipolar spindle and chromosome segregation ( For a review , see [1] ) . For cells to subsequently undergo cytokinesis and enter the G1 phase of the next cell cycle , Cdk-mediated phosphorylation events are reversed and Cdk activity declines ( For reviews , see [2] , [3] ) . In budding yeast , this transition , called mitotic exit , is triggered by a signaling cascade known as the mitotic exit network ( MEN ) . The MEN activates and releases the Cdc14 phosphatase from the nucleolus , and this phosphatase reverses the phosphorylation of Cdk substrates and inactivates the mitotic Cdks ( For a review , see [4] ) . The MEN must be tightly regulated for each daughter cell to receive a complete set of chromosomes . When the MEN is prematurely activated in cells undergoing mitosis , genomic instability results [5] . Therefore , the MEN is a crucial target of various checkpoints that keep mitotic Cdk activity high until the daughter chromosomes have segregated properly . The MEN coordinates spindle position and mitotic progression in asymmetrically dividing cells such as budding yeast , where the division plane is predetermined . A pathway called the spindle position checkpoint ( SPOC ) ensures that the MEN is activated only if the extended mitotic spindle is correctly positioned . When spindles misalign relative to the division plane , mitotic exit is delayed by preventing the MEN activation [6] . The Tem1 GTPase functions to activate the MEN [7] . The MEN signaling cascade is triggered when the two-component GTPase-activating protein ( GAP ) for Tem1 , composed of Bfa1 and Bub2 , becomes inactivated . The polo kinase Cdc5 also contributes to MEN activation by directly phosphorylating and inhibiting the GAP activity of Bfa1/Bub2 and/or disrupting its interaction with Tem1 [8] , [9] . Impaired Bfa1/Bub2 GAP activity allows mitotic exit in cells that have either mitotic exit defects or activated checkpoints [10] . Consistent with this , Bfa1 remains unphosphorylated when the SPOC prevents mitotic exit [8] . Lte1 , which was once suggested to promote mitotic exit as a putative guanine nucleotide exchange factor ( GEF ) for Tem1 , has been reported to contribute to mitotic exit by controlling asymmetric Bfa1 localization and cell polarization [11] , [12] . A recent study demonstrated that loading of Tem1 onto the spindle pole bodies ( SPBs ) is required for activation of the MEN [13] . Thus a misaligned spindle markedly delays mitotic exit in cells with low GAP activity for Tem1 [10] . These recent studies have suggested more complex ways by which MEN is regulated , including localization of MEN components to the SPB , together with the GTPase-switch model for Tem1 . The SPB acts as a scaffold for many MEN components ( For a review , see [4] ) . The association of Tem1 with SPBs depends on Bfa1 and Bub2 , which are mutually required for the other's localization to the SPB [14] . The Bfa1/Bub2 complex localizes to SPBs in an asymmetric manner: as the spindle aligns along the mother-bud axis and the dividing nucleus migrates to the bud neck , the complex is exclusively found on the bud-oriented SPB [14] , [15] . Conversely , on misoriented spindles that lead to delayed mitotic exit , Bfa1/Bub2 is present on both SPBs . This suggests that the spatial distribution of Bfa1/Bub2 is directly connected to the control of mitotic exit [15] , [16] . Consistent with this hypothesis , a Bub2 variant that localizes to both SPBs throughout the cell cycle prevented mitotic exit in certain MEN-impaired mutants [17] . Also a recent quantitative analysis showed that Bfa1 dynamics at the SPBs establishes asymmetry in MEN signaling and regulates MEN activity . Bfa1 associates with both SPBs in a transient fashion , but its binding to the daughter SPB ( SPBd ) is stabilized by cell polarity determinants and their interactions with microtubules [18] . As a consequence , Bfa1 accumulates on the SPBd during metaphase , whereas it disappears from the mother SPB ( SPBm ) , thereby establishing Bfa1 asymmetry [18] . When the spindles are improperly positioned , Bfa1 association becomes highly dynamic on both SPBs , which is required for proper SPOC function [19] . Despite the role in the fidelity of mitosis , the molecular details governing the asymmetry of Bfa1/Bub2 positioning have yet to be fully elucidated . Furthermore , the importance of the asymmetry in the unperturbed cell cycle remains unclear . Bfa1 asymmetry is required for recruiting MEN components exclusively to the SPBd during metaphase [18] . Bfa1 reaches its maximum phosphorylation state when it associates preferentially with the SPBd , whereas Bfa1 is unphosphorylated and localizes to both SPBs in SPOC-activated cells [8] , [16] . We have identified new phosphorylation sites on Bfa1 that function in directing its asymmetric distribution to SPBs . We present evidence that the phosphorylation of these sites by Cdc5 does not inhibit Bfa1 GAP activity , but induces Bfa1 asymmetry and achieves timely MEN activation during unperturbed mitotic progression .
Bfa1 is a cell cycle-regulated phosphoprotein [8] that forms a complex with Bub2 and negatively controls the activation of Tem1 , a key upstream regulator of mitotic exit [8] , [9] . Cdc5 polo kinase phosphorylates Bfa1 during mitosis to down-regulate Bfa1/Bub2 , and thus activates mitotic exit [8] , [9] . In cdc15-2 cells , Bfa1 becomes phosphorylated by Cdc5 and Tem1 is activated , but mitotic exit is not permitted since Cdc15 acts downstream of Tem1 in the MEN [7] . Thus , we used the cdc15-2 mutant , which contains wild-type CDC5 , as a control for Bfa1 phosphorylation and localization in Figure 1 . In this experiment , we compared wild-type CDC5 in the cdc15-2 strain to the cdc5-1 and cdc5-2 mutants to characterize the effects of these mutations on late mitosis at the restrictive temperature [20] , [21] . When α-factor-synchronized cdc15-2 cells were released at restrictive temperature ( 35°C ) , Bfa1 was detected as a sharp band in G1 phase; the band accumulated as slower-migrating forms during mitosis , and finally attained its maximal phosphorylation states ( Figure 1A left ) . When treated with phosphatase , the slower-migrating forms collapsed into a sharp band ( Figure 1A right ) . The two cdc5 mutations , cdc5-1 and cdc5-2 , affected Bfa1 differently . Bfa1 was phosphorylated as usual in the cdc5-1 mutant , whereas the cdc5-2 mutant was severely defective in phosphorylating Bfa1 at the restrictive temperature ( Figure 1A; [8] ) . We hypothesized that phosphorylation of Bfa1 by Cdc5 might influence its subcellular location . Before examining this possibility we investigated cell cycle progression and the arrest points in cdc5-1 and cdc5-2 cells . We shifted G1-synchronized cells to 35°C , and counted every 10 min the number of large budded cells with elongated nuclei stretching along the neck or two segregated nuclei ( Figure 1B ) . Compared to the cdc15-2 mutant , both cdc5 mutants exhibited a delay during nuclear elongation ( more markedly in cdc5-2 than in cdc5-1 ) , but eventually arrested as large budded cells with separated nuclei , a phenotype similar to that of the cdc15-2 mutant ( Figure 1B ) . Quantitative analysis of pole-to-pole distances in cells with segregated nuclei revealed that the spindle length relative to cell length was shorter in cdc5-1 and cdc5-2 cells than in cdc15-2 cells; mean spindle length was approximately 83% of that of cdc15-2 for cdc5-1 , and 53% for the cdc5-2 mutant ( Figure 1C ) . These results demonstrate that the cdc5-1 and cdc5-2 mutants have defects in nuclear spindle elongation but eventually undergo nuclear division with spindles that are not fully elongated; thereafter they arrest as large budded cells with separate nuclei , as does the cdc15-2 mutant . Hereafter , for simplicity , we refer to the arrest point of cdc5-1 , cdc5-2 , and cdc15-2 as late anaphase . We then examined the localization of Bfa1 in cdc5-1 , cdc5-2 and cdc15-2 cells , where Bfa1 and Spc42 were fused to GFP and RFP , respectively . Bfa1-GFP was found on both SPBs immediately after their separation and before the nucleus moved to the bud neck [14] , [18] . During metaphase of the cdc15-2 cells , when the spindle was oriented along the division axis and the nucleus was positioned at the bud neck , Bfa1-GFP was predominantly localized to the SPB closest to the bud neck ( Figure 1D; [18] ) . Bfa1 continued to be selectively localized in cdc15-2 cells , with elongated dividing nuclei or segregated nuclei ( Figure 1D ) . The cdc5-1 mutant displayed a Bfa1-GFP localization pattern similar to the cdc15-2 mutant , whereas Bfa1-GFP remained associated with both SPBs in the cdc5-2 mutant , even after nuclear segregation ( Figure 1D ) . In anaphase-arrested cells , Bfa1-GFP was present on both SPBs in 90 . 9±3 . 0% of cdc5-2 cells , whereas it was asymmetrically localized on the SPBd in 88 . 0±1 . 2% of cdc15-2 and 92 . 3±1 . 5% of cdc5-1 cells ( Figure 1E ) . We also observed that the fluorescence intensity of Bfa1-GFP on the SPBs in cdc5-2 cells was only about 25–28% of that in cdc15-2 cells ( Figure 1F ) . Recently , Monje-Casas and Amon [18] showed that the intensity of Bfa1-GFP fluorescence is a good measure of the affinity of Bfa1 for the SPB , and its dynamics . We therefore suggest that the phosphorylation of Bfa1 by Cdc5 regulates the dynamics of its behavior , and leads to its asymmetric distribution on the two SPBs . Since Cdc5 inhibits Bfa1/Bub2 GAP activity toward Tem1 by phosphorylating Bfa1 [9] , the presence of Bfa1 on both SPBs in cdc5-2 cells ( Figure 1 ) could be due to uninhibited Bfa1/Bub2 GAP activity or to the absence of Bfa1 phosphorylation . To distinguish between these possibilities , we examined the localization of GAP activity-defective variants of Bfa1 . In in vitro Tem1 GTPase assays with Bub2 , GAP activity was almost completely absent in Bfa1W422A , markedly decreased in both Bfa1M413I and Bfa1D416A , and slightly decreased in Bfa1G411E [10] . The cdc15-2 mutant was used to analyze the localization of each Bfa1 variant at anaphase . We integrated GFP-fused BFA1 , BFA1G411E , BFA1M413I , BFA1D416A , and BFA1W422A into cdc15-2SPC42-RFPΔbfa1 cells , and released these cells from G1 arrest at 35°C . Southern and Western blots verified that the Bfa1 mutants were integrated as single copies and were expressed at similar levels to wild-type Bfa1 ( Figure S1 ) . We reasoned that if Bfa1 asymmetry was promoted by inhibition of its GAP activity , the mutant forms of Bfa1 would establish Bfa1 asymmetry prematurely and localize to only one SPB throughout the cell cycle . In fact , however , like wild-type Bfa1 , they associated with both SPBs immediately after SPB separation and before nuclear migration to the bud neck ( data not shown ) . In addition , most of the Bfa1M413I , Bfa1D416A , and Bfa1W422A forms were present on both SPBs even after chromosome segregation , despite their low GAP activities ( Figure 2A and 2B ) . Bfa1M413I , Bfa1D416A , and Bfa1W422A bound to both SPBs in 78 . 4±1 . 7 , 68 . 7±0 . 6 , and 70 . 5±10 . 1% of anaphase-arrested cells , respectively , while most wild-type Bfa1 ( 88 . 0±1 . 2% ) and Bfa1G411E ( 83 . 2±4 . 5% ) was asymmetrically localized to the SPBd . To exclude the possibility that residual GAP activity of Bfa1M413I , Bfa1D416A , and Bfa1W422A was responsible for their association with both SPBs at anaphase , we constructed another mutant , Bfa1DDR2 ( Deletion of Direct Repeat 2; Bfa1G411E M413I D416A W422A ) , which was predicted to have no GAP activity; indeed , in vitro assays revealed that the Bfa1DDR2/Bub2 complex completely failed to stimulate Tem1 GTPase activity ( Figure 2C ) . Consistent with this , Bfa1DDR2 was utterly unable to prevent mitotic exit in vivo ( Figure S2 ) . After confirming single copy integration and normal expression levels ( Figure S1 ) , we observed that Bfa1DDR2 also persisted at both SPBs in anaphase ( 76 . 0±5 . 3%; Figure 2A and 2B ) , clearly demonstrating that inhibition of GAP activity does not induce Bfa1 asymmetry . These results suggest that the persistence of Bfa1 on both SPBs in the cdc5-2 mutant is not due to failure to inhibit GAP activity . We then asked if phosphorylation by Cdc5 is required for the asymmetric distribution of Bfa1 on SPBs . We examined phosphorylation of the Bfa1 mutants shown in Figure 2 using SDS-PAGE mobility shift assays in the cdc15-2 background . When α-factor- synchronized cells were released at 35°C , Bfa1G411E became phosphorylated with similar kinetics to wild-type Bfa1 ( Figure 3A ) . In contrast , slower-migrating forms of Bfa1M413I , Bfa1D416A , Bfa1W422A , and Bfa1DDR2 were not detected after the release from G1 arrest ( Figure 3A ) . We did not observe any mobility shift of these Bfa1 mutants even when Cdc5 was overexpressed ( Figure S3 ) . Using yeast two-hybrid assays we showed that these Bfa1 mutants interacted with Cdc5 like wild-type Bfa1 , demonstrating that the lack of Bfa1 phosphorylation by Cdc5 in these mutants is not due to reduced interaction with Cdc5 ( Figure S6B ) . To examine the extents of phosphorylation of the Bfa1 mutants , we purified GST-Cdc5 and GST-Cdc5KD ( a kinase-dead control ) from S . cerevisiae and incubated them with MBP-Bfa1 proteins ( MBP-Bfa1M413I , -Bfa1D416A , -Bfa1W422A , and -Bfa1DDR2 ) expressed in E . coli and purified . As the amount of GST-Cdc5 was increased , wild-type Bfa1 began to appear as multiple , slower migrating forms and eventually appeared as the slowest migrating form , while the Bfa1 mutants remained as multiple , less slowly-migrating forms ( Figure 3B ) . The results of these in vitro kinase assays differed slightly from the in vivo results in which slower migrating forms of the Bfa1 mutants were rarely seen ( Figure 3A; Figure S3 ) . This difference is probably due to either non-specific phosphorylation by the excessive Cdc5 activity used , or the presence of Cdc5 sites that are not easily phosphorylated in vivo . In either case , these Bfa1 mutants were obviously resistant to phosphorylation by Cdc5 . Note that Bfa1G411E localized asymmetrically to SPBd and was phosphorylated by Cdc5 like wild-type Bfa1 , whereas the other Bfa1 variants ( Bfa1M413I , Bfa1D416A , Bfa1W422A , and Bfa1DDR2 ) were distributed to both SPBs and were not phosphorylated as efficiently as wild-type Bfa1 by Cdc5 . We thus conclude that the asymmetric distribution of Bfa1 is probably linked to its phosphorylation by Cdc5 . Eleven Cdc5 phosphosites have been mapped previously , and substituted with Ala in the Bfa1-11A mutant [8] . To confirm the relationship between Bfa1 localization and phosphorylation , we examined the distribution of Bfa1-11A on SPBs . Since Bfa1-11A mobility is not greatly retarded under conditions that normally produce hyperphosphorylated wild-type Bfa1 ( Figure 4A ) , we expected this mutant to be present on both SPBs in anaphase cells . However , most of the Bfa1-11A ( 90 . 1±3 . 4% ) was only associated with SPBd , like wild-type Bfa1 ( 88 . 0±1 . 25% ) ( Figure 4B ) . We therefore examined whether Bfa1-11A was further phosphorylated by Cdc5 . When MBP-fused Bfa1-11A was incubated with GST-Cdc5 and γ-[32P] ATP , 32P incorporation was observed , whereas no 32P incorporation was detected in reactions with an equivalent amount of GST-Cdc5KD ( Figure 4C ) . These observations are consistent with a previous report that mutation of these 11 residues reduces in vitro phosphorylation of Bfa1 by only 75% [8] , and demonstrate that not all Cdc5 phosphorylation sites are mutated in Bfa1-11A . Based on these results , we reasoned that there are unidentified Cdc5 phosphosites and that these could be responsible for the asymmetric distribution of Bfa1-11A . We further hypothesized that these novel sites are not efficiently phosphorylated on Bfa1M413I , Bfa1D416A , and Bfa1W422A , thereby causing these Bfa1 variants to persist at both SPBs in anaphase . If that were the case , the introduction of the M413I , D416A , or W422A mutation into Bfa1-11A should impair Bfa1-11A asymmetry and reduce its phosphorylation . Indeed , 32P incorporation into Bfa1M413I-11A , Bfa1D416A-11A , and Bfa1W422A-11A was less efficient than into Bfa1-11A ( Figure 4C ) , and Bfa1M413I-11A , Bfa1D416A-11A , and Bfa1W422A-11A were each symmetrically distributed to both SPBs ( Figure 4D ) . Consistent with this , when Cdc5 was overexpressed to force phosphorylation of Bfa1 by Cdc5 , we detected a mobility shift in Bfa1-11A but not in Bfa1M413I-11A , Bfa1D416A-11A , or Bfa1W422A-11A ( Figure S4A ) . However , in the anaphase-arrested cdc15-2 background , the mobilities of Bfa1M413I-11A , Bfa1D416A-11A , and Bfa1W422A-11A were nearly the same as that of Bfa1-11A ( Figure 4A ) , suggesting that phosphorylation of the residue ( s ) responsible for asymmetry is not efficient in cells expressing endogenous Cdc5 levels . Together , these results support the presence of unidentified Cdc5 target residues required for establishing Bfa1 asymmetry . We previously showed that the C-terminal 184 residues of Bfa1 ( Bfa1-D8391–574 ) sufficiently inhibit the MEN and localize predominantly to the SPBd ( Figure 5A and 5B; [22] ) . As expected , GFP-fused Bfa1-D8M413I , Bfa1-D8D416A , and Bfa1-D8W422A were found at both SPBs ( Figure 5B ) . In addition , the Bfa1-D8 mutants were less efficiently phosphorylated by Cdc5; Bfa1-D8 was more intensely labeled by 32P and detected in a slower-migrating form than Bfa1-D8M413I , Bfa1-D8D416A , and Bfa1-D8W422A ( Figure 5C ) . This indicated that the putative Cdc5 target site ( s ) responsible for Bfa1 asymmetry was probably located within the C-terminal 184 residues of Bfa1 . We next searched for possible kinase targets within the C-terminal 184 residues of Bfa1 . Cdc5 is a Ser/Thr protein kinase , and Bfa1-D8 contains 23 Ser and 12 Thr residues ( Figure 5A ) . We first systematically mutated 21 of the 23 Ser residues to Ala in Bfa1-D8 and constructed 16 different Ser mutants as GFP fusion proteins: Bfa1S395A , Bfa1S404A , Bfa1S426A , Bfa1SSS452AAA , Bfa1SKS459AAA , Bfa1S469A , Bfa1S478A , Bfa1S490A , Bfa1SS504AA , Bfa1SS510AA , Bfa1S530A , Bfa1S541A , Bfa1S551A , Bfa1S555A , Bfa1S559A , and Bfa1S571A ( Figure 5A , Table 1 ) . Both 424S and 447S were excluded because they were included in Bfa1-11A . We integrated each of the Ser mutants into cdc15-2Δbfa1 cells , and examined their localization at anaphase . The results are summarized in Table 1 . Most of the GFP-fused Ser mutants exhibited the localization pattern of wild-type Bfa1-GFP ( Table 1; and data not shown ) . However , the percentage of anaphase-arrested cells with Bfa1-GFP on both SPBs was significantly increased in cdc15-2BFA1SSS452AAA ( for simplicity , cdc15-2BFA13A ) and cdc15-2BFA1S559A cells; Bfa1-GFP was present on both SPBs in 12 . 0±1 . 2% of cdc15-2BFA1 cells , 67 . 3±6 . 5% of cdc15-2BFA13A cells , and 29±2 . 6% of cdc15-2BFA1S559A cells ( Figure 5D and 5E , Table 1 ) . Cells expressing the substitutions of each Ser in Bfa13A , GFP-tagged Bfa1S452A , Bfa1S453A , or Bfa1S454A , also had larger numbers of anaphase cells with Bfa1-GFP on both SPBs ( Table 1 ) , demonstrating that all three Ser residues contribute to Bfa1 asymmetry . In contrast , mutation of some of the Thr residues to Ala around the newly identified 452S , 453S , 454S , and 559S sites ( 465T , 497TT , 500T , 537T , 546T , 552T , and 572T; Figure 5A ) had little effect on Bfa1 asymmetry ( Table 1; data not shown ) . To confirm these results , we constructed Bfa14A , in which all four Ser residues were substituted with Ala ( SSS452AAA S559A referred to as 4A ) . Bfa14A-GFP was detected on both SPBs in 79 . 2±4 . 8% of anaphase cells , compared with 67 . 3±6 . 5% for cdc15-2BFA13A and 29±2 . 6% for cdc15-2BFA1S559A cells , again showing that all four Ser residues play a role in establishing Bfa1 asymmetry ( Figure 5D and 5E ) . Quantification of the GFP signal in anaphase-arrested cells showed that the fluorescence intensity of Bfa14A-GFP was nearly the same at both SPBs , weaker than that of wild-type Bfa1-GFP at the SPBd ( Figure 5F ) and similar to that of Bfa1-GFP in the cdc5-2 mutant ( Figure 1 ) . In order to verify that the symmetric localization of Bfa14A was not caused by defective interaction with Cdc5 , we compared the physical interactions of Bfa14A and wild-type Bfa1 with Cdc5 using yeast two-hybrid assays . As shown in Figure S6A , Bfa14A interacted as strongly with Cdc5 as wild-type Bfa1 . The asymmetric localization of Tem1 to the daughter SPB in anaphase depends on Bfa1 and Bub2 [14] . Therefore we asked whether the presence of the Bfa14A mutant on both SPBs affected the asymmetric localization of Tem1 . To answer this question we integrated GFP-fused BFA14A or BFA1 into cdc15-2TEM1-RFPΔbfa1 cells , arrested these cells in G1 , and released them at 35°C . As expected , Tem1-RFP followed the localization pattern of Bfa1-GFP in anaphase: it was present on both SPBs in the Bfa14A background and distributed asymmetrically to the SPBd in the wild-type background ( Figure 5G ) . We then asked if 452S , 453S , 454S , and 559S of Bfa1 are phosphorylated by Cdc5 . Measuring phosphorylation of these four Ser without phosphorylation of the 11 known targets was not feasible . Therefore , since the N-terminus of Bfa1 contains nine of the Cdc5 targets of Bfa1-11A , we used the C-terminal 184 residues of Bfa1 in in vitro kinase assays . MBP-tagged Bfa1-D8 , Bfa1-D8-11A , Bfa1-D84A , and Bfa1-D84A-11A were incubated with GST-Cdc5 or GST-Cdc5KD . The extents of phosphorylation were determined from the resulting mobility shifts and phospho-Bfa1-D8 bands by Phos-tag SDS-PAGE ( Figure 6A ) . The band representing Bfa1-D84A-11A was tighter in mobility shifts and the phospho-Bfa1-D84A-11A forms were not detected in Phos-tag SDS-PAGE ( Figure 6A ) . The extents of phosphorylation were further confirmed by measuring γ-[32P] incorporation and Bfa1-D84A-11A less intensely labeled with 32P than that of Bfa1-D8-11A ( Figure 6B ) , demonstrating that the 4A mutations reduce Bfa1 phosphorylation by Cdc5 . In addition , we speculated that the slower-migrating forms of Bfa1-11A observed in CDC5-overexpressing cells ( Figure S4A ) resulted , at least in part , from the phosphorylation of residues , such as 452S , 453S , 454S , and S559 , required for Bfa1 asymmetry . Indeed , the 4A mutations abolished the slower migrating forms of Bfa1-11A in cells overexpressing CDC5 ( Figure S4A ) , consistent with the results in Figure 6A and 6B . Thus , we suggest that Cdc5 phosphorylates 452S , 453S , 454S , and 559S of Bfa1 . To confirm this we used mass spectrometry ( MS ) to map Bfa1 phosphorylation sites ( Figure 6C ) . Recombinant MBP-Bfa1 was phosphorylated in vitro with GST-Cdc5 or with GST-Cdc5KD as a negative control ( Figure 6C ) . Subsequently , MBP-Bfa1 was purified by SDS-PAGE , excised , digested with trypsin , and analyzed by phosphopeptide-selective precursor ion-scanning liquid chromatography ( LC ) MS . The success of this approach was assessed by seeing if we could detect the 11 previously identified Cdc5 phosphorylation sites . Although we were unable to identify eight of the 11 phosphorylation residues of Bfa1 , likely due to their presence on extremely small ( 547S-549K; three amino acids ) or large ( 28F-75K; 48 amino acids ) trypsin peptides , we did detect phosphorylated forms of 17S , 24T , and 447S with high frequency . The same tandem LC-MS/MS analysis identified two phosphopeptide species , 452SSpSPFLR458 ( pS , phosphorylated S; Figure 6C ) and 452SpSpSPFLR458 ( data not shown ) , containing 453S and 454S , two of the four Bfa1 phosphorylation residues responsible for asymmetry identified above . Phosphorylation of 452S and 559S was not detected by this approach . To ask if the GAP activity of Bfa14A affects its association with both SPBs at anaphase , we directly measured the GAP activity of Bfa14A/Bub2 . When MBP-Bfa14A was added to the reaction together with Tem1 and GST-Bub2 , as in the experiment of Figure 2C , γ-Pi increased rapidly with kinetics similar to those obtained with wild-type Bfa1 , indicating that the 4A substitutions had no effect on Bfa1 GAP activity ( Figure 6D ) . We also examined the control of mitotic exit by Bfa14A in vivo: BFA14A cells arrested as large budded cells , as did wild-type BFA1 cells , in the presence of various checkpoint-activating signals ( Figure S5A–S5C ) . Together , these results show that the Bfa14A has functional GAP activity and demonstrate that the presence of Bfa14A at both SPBs is independent of GAP activity . Since Bfa1 is found on both SPBs in cells with misaligned spindles [16] , it has been proposed that symmetrically localized Bfa1 , and in particular , the Bfa1 associated with the SPB in mother cells , contributes to the arrest of mitotic exit [17] . However , the BFA1M413I , BFA1D416A , BFA1W422A , and BFA14A cells grew well and did not show any apparent cell cycle delay in unperturbed conditions , despite having Bfa1 on both SPBs throughout the cell cycle . To better understand the function of Bfa1 asymmetry in mitotic exit , we examined cell cycle progression in these asymmetry-defective BFA1 mutants . Following G1 synchronization and release at room temperature , we monitored large budded cells with two divided nuclei ( for simplicity , anaphase cells ) . In both wild-type and mutant BFA1 cultures , anaphase cells began to accumulate approximately 70 min after release ( Figure 7A ) . In BFA1 cells , numbers of anaphase cells began to decrease about 110 min after release . Interestingly , in the BFA1M413I , BFA1D416A , BFA1W422A , and BFA14A cultures , the decrease in anaphase cells was delayed by about 10 min ( Figure 7A ) . To further examine the delay in cell cycle progression in the Bfa1 mutant cells , we measured Pds1 and Sic1 levels . Pds1 is an anaphase inhibitor that is degraded upon sister chromatid separation [23] and Sic1 is a negative regulator of mitotic CDKs that accumulates following activation of the MEN ( For a review , see [24] ) . Consistent with Figure 7A , the wild-type and all the BFA1 mutants exhibited a drop in Pds1 levels approximately 60 min after release ( Figure 7B and 7C ) , indicating that Bfa1 asymmetry and its persistent association with the SPBm does not alter the timing of anaphase onset . Importantly , Sic1 accumulation in the wild-type began about 80 min after release , whereas it began about 90 min after release in BFA1M413I , BFA1D416A , BFA1W422A , and BFA14A cells ( Figure 7B and 7C ) . BFA1G411E cells , with a normal Bfa1 distribution , displayed the Pds1 and Sic1 kinetics of the wild-type ( data not shown ) . These results show that asymmetry-defective BFA1 cells activate the MEN approximately 10 min later than cells with normal Bfa1 localization . We confirmed the 10 min delay of mitotic exit in BFA14A mutant cells by examining the dynamics of Cdc14 release in BFA1 and BFA14A mutant cells by live cell analysis using time-lapse confocal microscopy . Mitotic exit requires full activation of Cdc14 by releasing it from the nucleolus in late anaphase [25] . As shown in the captured images of Figure 7D , Cdc14 was fully released out of the nucleolus 95 and 105 min after BFA14A mutant cells were released from G1 arrest , while wild-type BFA1 cells released Cdc14 approximately 10 min earlier . When we examined time lapse images of several BFA14A and wild-type cells , the average time of Cdc14 release was 95 . 8±9 . 1 min after release from G1 arrest in the BFA14A mutant and 86 . 7±13 . 5 min in BFA1 wild-type ( Figure 7D ) . These observations clearly demonstrate that in the asymmetry-defective BFA14A cells the activation of MEN is delayed by approximately 10 min compared with wild-type BFA1 cells . We then investigated the physical interaction of Bfa14A with Tem1 and Bub2 using yeast two-hybrid assays to verify that the 10 min delay of mitotic exit was due to its symmetric localization resulting from lack of phosphorylation and not to any reduction in its interaction with Tem1 and Bub2 . As shown in Figure S6C and S6D , Bfa14A interacted with Tem1 and Bub2 like wild-type Bfa1 . Together , these observations demonstrate that Cdc5-dependent phosphorylation of the four identified serine residues in Bfa1 controls its displacement from the SPBm for timely mitotic exit in unperturbed mitosis . In order to confirm that Cdc5-dependent phosphorylation of 452S , 453S , 454S , and 559S in Bfa1 controls the asymmetric localization of Bfa1 for timely mitotic exit , we constructed Bfa14D ( SSS452DDD S559D referred to as 4D ) that mimics the negative charges due to phosphorylation . GFP-fused BFA14D was integrated into cdc15-2SPC42-RFPΔbfa1 cells , and as shown in Figure 8A , Bfa14D-GFP remained localized asymmetrically on the SPBs throughout the cell cycle , even in G2/M when wild-type Bfa1 is present on both SPBs . At G2/M , the cells with asymmetrically localized Bfa1 on the SPBs were significantly increased in cdc15-2BFA14D cells than in cdc15-2BFA1 cells; 73 . 9±6 . 3% of cdc15-2BFA14D and 19 . 7±5 . 8% of cdc15-2BFA1 cells ( Figure 8A ) . When cells were arrested in late anaphase , Bfa14D-GFP was asymmetrically localized on the daughter SPB in 98 . 6±1 . 5% of the cells , compared to 20 . 8±4 . 8% in cdc15-2BFA14A cells ( Figure 8A and Figure 5E ) . These observations demonstrate that symmetric localization of Bfa14A is caused by loss of phosphorylation , and suggest that phosphorylation of Bfa1 by Cdc5 on 452S , 453S , 454S , and 559S is necessary for establishing Bfa1 asymmetry . In kinase-defective cdc5-2 cells , Bfa1 was unphosphorylated and localized to both SPBs ( Figure 1D ) . The finding that Cdc5-dependent Bfa1 phosphorylation on 452S , 453S , 454S , and 559S residues regulates its asymmetric localization prompted us to ask whether the phospho-mimicking Bfa14D is asymmetrically located in cdc5-2 cells . For this , we integrated pRS304-BFA14D-GFP into the TRP1 locus of cdc5-2Δbfa1 cells . In late anaphase-arrested cdc5-2 cells , Bfa14D was asymmetrically localized in 72 . 7±0 . 1% of the cells , while Bfa1-GFP was present on both SPBs in 90 . 9±3 . 0% of cells ( Figure 1E and Figure 8B ) . These observations further support the notion that Cdc5-dependent phosphorylation of 452S , 453S , 454S , and 559S in Bfa1 is important for its asymmetric localization . Since Bfa1 is a target of Cdc5 phosphorylation for triggering mitotic exit , its deletion has been reported to rescue kinase-defective cdc5-2 cells arrested in late anaphase at restrictive temperatures [8] . To see whether Bfa14D can inhibit the MEN , we tested whether the viability of cdc5-2 could be restored by BFA14D . As shown in Figure S7 , BFA14D as well as wild-type BFA1 suppressed the growth of cdc5-2Δbfa1 cells , while knock-out of BFA1 rescued the viability of cdc5-2 cells . These results demonstrate that Bfa14D is able to inhibit MEN like wild-type Bfa1 . Since Bfa14D-GFP is exclusively localized on one of the SPBs during mitosis ( Figure 8A ) but functions as a negative regulator of the MEN , we asked whether the lack of dynamic localization of phospho-mimetic Bfa14D affects cell cycle progression . cdc15-2 cells expressing wild-type BFA1 or phospho-mimetic BFA14D were synchronized in G1 and released at room temperature , and their cell cycle progression was monitored by counting metaphase and anaphase cells . As shown in Figure 8C , BFA14D cells exhibited the same kinetics of cell cycle progression as wild-type BFA1 cells . In both wild-type and BFA14D cells , metaphase cells began to accumulate at 60 min and peaked at 100 min , while anaphase cells appeared at 100 min and reached a peak at 120 min after release ( Figure 8C ) . To further analyze the cell cycle progression , we measured the mitotic cyclin Clb2 , which is degraded upon activation of the MEN [26] . Consistent with the above result , Clb2 began to accumulate at approximately 60 min after release in both wild-type and BFA14D mutant cells , peaked at 90 min , and then declined ( Figure 8D and 8E ) . These results showed that BFA14D cells allow timely mitotic exit like wild-type BFA1 . Together they confirm that phosphorylation of 452S , 453S , 454S , and 559S regulates the asymmetric localization of Bfa1 and timely mitotic exit in unperturbed cell cycle . In addition , the asymmetric presence of Bfa14D protein on SPBs in G2/M did not affect early mitotic progression ( Figure 8 ) . Therefore we suggest that Bfa1 asymmetry is required for timely activation of the MEN but is not necessary for mitotic progression before anaphase in the unperturbed cell cycle . Previous studies have suggested that the symmetrical distribution of Bfa1/Bub2 is directly related to the delay of mitotic exit when the spindle is not properly aligned [15] , [16] . We therefore asked whether Bfa14D is able to function in the spindle position checkpoint and whether it is symmetrically localized in cells with misaligned spindles . Proper positioning of the mitotic spindle relies on two independent pathways , one involving the minus-end microtubule motor dynein , the other Bim1 , a plus-end microtubule-binding protein [27] , [28] . The absence of DYN1 induces anaphase spindle misalignment in the mother cell and thus triggers the spindle position checkpoint [6] , [27] . We first examined the localization of Bfa14D in Δdyn1 cells by integrating pRS304-BFA14D-GFP into the TRP1 locus of Δdyn1mCherry-TUB1Δbfa1 cells , as described in Materials and Methods . Surprisingly , Bfa14D was present on both SPBs in cells with misaligned spindles like wild-type Bfa1 ( Figure 9A ) . When the anaphase spindle is misaligned in the parent of Δdyn1 cells , BFA1 deletion induces improper mitotic exit , as a result of which both multinucleate and anucleate cells accumulate [14] , [29] . To assess the spindle position checkpoint functioning of Bfa14D , we monitored multinucleate and anucleate cells in Δdyn1BFA14D and compared them with Δdyn1 cells with the wild-type BFA1 ( Δdyn1BFA1 ) after arrest with α-factor and release at 16°C , when the spindle orientation defect is most pronounced . As shown in Figure 9B , the improper mitotic exit seen in Δdyn1Δbfa1 cells was significantly decreased in both Δdyn1BFA1 and Δdyn1BFA14D cells; 5 . 7±0 . 1% in Δdyn1BFA1 and 4 . 7±0 . 3% in Δdyn1BFA14D cells . We also examined the spindle position checkpoint function of phospho-mimetic Bfa14D in Δbim1Δbfa1 cells . Consistent with the above result , BFA14D rescued the viability of Δbim1Δbfa1 cells like wild-type BFA1 ( Figure 9C ) . These results demonstrate that BFA14D cells contain SPOC activity like wild-type BFA1 cells . The symmetric localization of Bfa14D is consistent with the SPOC activity of Bfa14D as well as previous evidence that the symmetrical localization of Bfa1 in cells with misaligned spindles is directly connected to the activation of SPOC [15] , [16] . Based on these observations , we suggest that the newly identified Cdc5-dependent phosphorylation residues in Bfa1 , 452S , 453S , 454S , and 559S , are only important for its asymmetrical localization and the timing of mitotic exit in unperturbed cells .
In budding yeast , Kar9 and dynein preferentially associate with the bud-directed SPB , from which astral microtubules emanate [30] , [31] . If these proteins distribute symmetrically to both SPBs , the mitotic spindle does not align properly , showing that SPB asymmetry is essential for mitosis [30] , [31] . Konig et al . showed that cyclin-dependent kinase 1 ( Cdk1 ) is asymmetrically recruited to the SPBm in early anaphase and negatively regulates MEN activity at the SPBm [32] . Caydasi and Pereira [19] reported that forced targeting of Bfa1 and Bub2 to both SPBs compromised SPOC function and Valerio-Santiago et al . demonstrated that control of Tem1 localization is essential for the proper functioning of the MEN and SPOC [13] . Recently , Bertazzi et al . showed that Lte1-promoted exclusion of Kin4 from the SPBd is essential for proper mitotic exit [33] . These previous studies mainly focused on the biological function of SPB asymmetry in cells with misaligned spindles . Here , we have demonstrated that Cdc5-dependent phosphorylation of Bfa1 contributes to its asymmetric distribution at SPBd , which is required for timely mitotic exit and , therefore , is required for the fidelity of cell division in unperturbed cells without misaligned spindles . Previously , GAP activity-defective Bub2-Myc was reported to lead to localization of the Bfa1/Bub2 complex to both SPBs throughout the cell cycle . This complex inhibits mitotic exit , but only in mutant backgrounds in which the MEN is partially impaired [17] . However , it was not clear whether the inhibition of mitotic exit was due to lack of asymmetry or to the absence of GAP activity . Therefore , the importance of Bfa1 asymmetry and its specific function in normal cell cycle progression were not well understood . In this study , we identified various asymmetry-defective Bfa1 mutants that persist on both SPBs throughout the cell cycle . In particular , unlike Bub2-Myc , the Bfa14A mutant stimulated the Tem1 GTPase ( Figure 6D ) , and activated checkpoints for mitotic exit control ( Figure S5A–S5C ) . These results demonstrate that wild-type Bfa1 and Bfa14A differ only in their localization patterns . Bfa14A delayed Sic1 accumulation and Cdc14 release by approximately 10 min relative to cells with normal localization of Bfa1 ( Figure 7B–7D ) . On the other hand , the phospho-mimetic Bfa14D mutant allowed timely mitotic exit like wild-type Bfa1 ( Figure 8C–8E ) . Therefore , we suggest that Bfa1 asymmetry and its disappearance from the SPBm regulate the timing of MEN activation in unperturbed cell division cycles . However in cells with misaligned spindles , Bfa14D was located on both SPBs and there was full SPOC activity ( Figure 9 ) . We therefore consider that the newly identified Cdc5-dependent phosphorylation residues in Bfa1 , 452S , 453S , 454S , and 559S , are only important for its asymmetrical localization and the timing of mitotic exit in unperturbed cells . The symmetric localization as well as the SPOC activity of Bfa14D is consistent with previous studies that showed that the symmetrical localization of Bfa1 in cells with misaligned spindles is directly connected with activation of the SPOC [15] , [16] . We speculate that cells override the Cdc5-depedent asymmetric localization of Bfa1 in the presence of a spindle orientation defect . Thus , the previously reported mechanisms that account for the symmetric localization of Bfa1 and the arrest of mitotic exit in response to misaligned spindles may apply to Bfa14D . It has been suggested that a 10 min delay in the cell division cycle is not biologically significant in controlling the cell division cycle . However , considering that the entire cell cycle of budding yeast is about 90 min and mitosis takes approximately 30 min [34] , a 10 min delay is not negligible . In fact , mitotic exit is only delayed by 15 minutes in the presence of constant peak levels of Clb2 , which blocks spindle disassembly [35] . Valerio-Santiago et al . recently showed that localization of Tem1 to the SPBs is essential for activation of the MEN [13] . As we showed in Figure 5G , Tem1-RFP localized to both SPBs in BFA14A cells . We also showed that Bfa14A binds to Tem1 like wild-type Bfa1 ( Figure S6C ) . These observations suggest that the delay of mitotic exit in BFA14A cells is a consequence of disrupting the asymmetric localization of Tem1 . What molecular details underlie Bfa1 asymmetry in unperturbed mitosis ? One significant contribution may come from cell polarity determinants [18] . Monje-Casas and Amon reported that the correct interaction of astral microtubules with the bud cortex alters the affinity of Bfa1 for SPBs and affects its asymmetry [18] . Consistent with this observation , Geymonat et al . showed that the activity of Lte1 in mitotic regulation depends on its localization to the bud cortex and contributes to the asymmetric localization of Bfa1 to the daughter SPB [12] . How can information in the cortex control the distribution of Bfa1 at SPBs , and how is Bfa1 able to bind to the SPBs with different affinities ? When spindles misalign , Kin4 kinase activity and its localization to SPBs are reported to regulate the residence time of Bfa1 at SPBs , as well as SPOC activity [19] . However , if the spindle is correctly aligned , Kin4 begins to associate with the SPBm in mid-anaphase at a time when Bfa1 asymmetry has already been established [36] . In addition , in Δkin4 cells with proper spindle positioning , Bfa1 has a normal localization pattern [19] . Furthermore , symmetric localization of Bfa14A is not caused by its defective interaction with Kin4 , since Bfa14A interacted with Kin4 like wild-type Bfa1 ( Figure S5D ) . Thus , other factors must regulate Bfa1 asymmetry , particularly in unperturbed cells . Fraschini et al . [17] proposed that the disappearance of Bfa1/Bub2 from the mother-directed SPB requires Bfa1/Bub2 GAP activity . We also observed that various GAP activity-defective Bfa1 mutants persisted at the SPBm during anaphase . However , we showed that Bfa1 asymmetry was not dependent on GAP activity . Because MEN activation requires inhibition of Bfa1 GAP activity and Bfa1 asymmetry , if Bfa1 asymmetry is regulated by its GAP activity , only a decline in GAP activity could promote Bfa1 loss from the SPBm . Nevertheless , this is probably not the case , as is shown by the symmetric localization of the GAP-defective Bub2-Myc and Bfa1DDR2 proteins . Bfa1 also localized on both SPBs in BFA14A cells with normal GAP activity , and in cdc5-2 cells where Bfa1 GAP activity is expected to be high due to lack of phosphorylation . Although we have shown in this study that Bfa1 asymmetry is not dependent on its GAP activity , we should consider the possibility that its GAP activity influences its localization indirectly , by affecting its phosphorylation . However , Bfa1D416A , which retains approximately 50% of the GAP activity of wild-type Bfa1 , is also defective in phosphorylation by Cdc5 , like Bfa1 mutants completely lacking GAP activity ( Figure 3 ) . Thus , it is unlikely that the GAP activity of Bfa1 influences its localization indirectly by affecting its phosphorylation . It may still be possible that asymmetric localization requires a certain threshold level of Bfa1 GAP activity ( which must be higher than the level in Bfa1D416A ) as a prerequisite for phosphorylation of the four serine residues that we have identified . Our observations that mutation of the four Cdc5-depedent phosphorylation residues , 452S , 453S , 454S , and 559S to Ala in Bfa14A significantly reduced its phosphorylation by Cdc5 , as well as affecting its localization , strongly support the role of these residues in directing asymmetric localization in unperturbed mitosis . This notion was further supported by phospho-mimetic Bfa14D , which was asymmetrically localized to the SPBs in cdc15-2-dependent arrested cells ( Figure 8A ) and even in kinase-defective cdc5-2 ( Figure 8B ) . However , Bfa14D were not asymmetrically localized in 100% of the cdc5-2 cells ( Figure 8B ) and phosphorylation of Bfa1-D811A+4A by Cdc5 was only reduced by approximately 25% compared with Bfa1-D8 ( Figure 6B ) . Therefore , we cannot exclude the possibility that Bfa1 contains some additional residue ( s ) that is/are also phosphorylated by Cdc5 and is/are involved in the asymmetric localization of Bfa1 . Bfa14A bound to both SPBs of anaphase-arrested cdc15-2 cells with properly segregated nuclei , whereas the 11 previously identified Cdc5 target sites ( Bfa1-11A ) had little or no effect on Bfa1 localization . However , the association of Bfa14A with the two SPBs at anaphase was not as stable as that of wild-type Bfa1 for the SPBd . We also found that the Bfa1DDR2 mutant formed a stronger association with the SPBs than wild-type Bfa1 ( Figure S8 ) . Thus , further characterization of this mutant may help uncover the molecular mechanisms underlying Bfa1 dynamics . One possibility is that all Bfa1 becomes phosphorylated , and phospho-Bfa1 has different affinities for the two SPBs . Alternatively , Cdc5 may differentially phosphorylate Bfa1 at one of the two SPBs . Another possibility raised by Monje-Casas and Amon [18] , is that some proteins mediating the association of Bfa1 with SPBs may control the affinity of Bfa1 for the SPBs by introducing various modifications . When we mapped the phosphosites of Bfa1 by mass spectrometry , phosphorylation of several characterized sites , such as 7T and 424S , was detected with higher efficiency than phosphorylation of the novel residues we identified as required for Bfa1 asymmetry ( Figure S9 ) . It is tempting to speculate that the Cdc5 target sites regulating Bfa1 asymmetry in unperturbed mitosis are phosphorylated with higher fidelity , and/or that other factor ( s ) are involved in modulating the efficiency of phosphorylation to control Bfa1 dynamics . Due to low phosphorylation efficiency , only 453S and 454S were detected as phosphor forms by MS . The low phosphorylation efficiency of these residues is consistent with the proposed biological role of their phosphorylation in controlling the timing of mitotic exit during unperturbed cell division cycles . Although phosphorylation of the newly identified phosphosites of Bfa1 by Cdc5 was not very efficient , a similar SSS934FL sequence in Claspin had been identified as the target of phosphorylation by Plx1 , a frog ortholog of budding yeast Cdc5 [37] . While mapping the Cdc5 target sites , we identified p150S and p180S , which were reported to be phosphorylated by Kin4 and to prevent further modification of Bfa1 by Cdc5 ( 10; [38] ) . Although we cannot exclude the possibility that p150S and p180S were phosphorylated nonspecifically due to the extremely high Cdc5 kinase activity , the phosphorylation efficiency at these sites was comparable to that of other Cdc5 targets responsible for GAP activity , such as 7T and 424S ( Figure S9 ) . Furthermore , despite the phosphorylation of these sites , other sites were phosphorylated by Cdc5 . Thus , we suspect that Bfa1 phosphorylation by Cdc5 and Kin4 , and the biological functions of these modifications are far more complicated than we currently understand , and will require further study . We found that the 452S , 453S , 454S in budding yeast Bfa1 are conserved as 589S , 590T , 591S in its fission yeast homologue byr4 ( Figure S11 ) , which is also localized asymmetrically to SPBs in anaphase [39] . It would be interesting to examine whether these conserved residues of byr4 are also phosphorylated by a polo-like kinase , contribute to the asymmetric localization of byr4 and regulate the timing of SIN in fission yeast . In summary , we have shown that Cdc5-mediated phosphorylation of the newly identified residues on Bfa1 modulates the affinity of Bfa1 for SPBs , and as a consequence contributes to the asymmetric distribution of Bfa1 at anaphase . The asymmetric Bfa1 distribution is required for timely mitotic exit , thus probably ensuring tight coupling of MEN activation and chromosome segregation during normal cell cycle progression . We have also uncovered a novel function of the polo kinase , Cdc5 , in the control of mitotic exit . Further studies are needed to identify factors that control the Cdc5-dependent Bfa1 phosphorylation responsible for asymmetric localization in unperturbed mitosis , and how it is overridden in the presence of a misaligned spindle . These studies promise to provide crucial insights into how centrosome asymmetry is generated , and its biological importance in the asymmetric division of eukaryotic cells .
All yeast cultures and genetic techniques were carried out as described by Kim et al . [22] . The S . cerevisiae strains used in this study are described in Table S1 . Strains were generated by PCR-based methods and verified by PCR , and Southern and western blot analysis [40] , [41] . The integrating plasmid , pRS304-BFA1-GFP , was linearized with EcoRV and integrated into the TRP1 locus , as described by Kim et al . [10] . Bfa1 mutants were constructed by PCR-based site-directed mutagenesis , as described by Kim et al . [10] . Cells were synchronized in G1 by adding 10 µg/ml or 50 ng/ml α-factor ( Sigma-Aldrich ) to BAR1 or bar1 cells , respectively , and at S phase with 0 . 2 M hydroxyurea ( Sigma-Aldrich ) for 2–3 h . CDC5 expression was driven by the GAL promoter . Fluorescence microscopy was performed essentially as described by Kim et al . [22] . Cellular labeling was visualized on an Axioplan2 ( Zeiss ) microscope with a Zeiss 100× Plan Neofluar oil immersion objective . Images were acquired using an Axiocam CCD ( Zeiss ) camera and AxioVision software ( Zeiss ) . The fluorescence intensity of GFP and RFP-fused proteins was quantitatively analyzed by confocal microscopy . For confocal images , we used a Nipkov disk-based UltraVIEW RS confocal system ( PerkinElmer ) equipped with a Nikon microscope ( TE2000-PFS ) . The 100× NA 1 . 4 oil immersion objective lens was controlled by a piezoelectric z stepper . In each experiment , 10 to 15 z sections were acquired at 0 . 5 µm steps with 2×2 binning , the same laser power and exposure time , and projected in UltraVIEW RS software ( PerkinElmer ) . Fluorescence intensity of selected regions of interest was quantified using UltraVIEW RS and Image J software ( Version 1 . 38u , NIH ) , and the background fluorescence was subtracted by placing the same measurement circle in nearby intracellular regions without a Bfa1-GFP or Tem1-RFP signal . Since the fluorescence intensity of Bfa1-GFP generally increased with Spc42-RFP intensity , we normalized the intensity of Bfa1-GFP to Spc42-RFP to precisely measure the amount of Bfa1 associated with the SPB . For time-lapse experiment , a Nipkov disk-based UltraVIEW RS confocal system ( PerkinElmer ) equipped with a Nikon microscope ( TE2000-PFS ) was used . Images for cells on agar plugs were taken every 5 min and processed with Adobe Photoshop 7 . 0 . No manipulations were added other than adjustments in brightness and contrast . For phosphatase treatment , TAP-tagged Bfa1 was precipitated with IgG Sepharose beads ( Amersham ) from total cellular lysates ( 1 mg in 700 µl modified H-buffer containing 1% NP-40 ) as described by Kim et al . [10] and treated with Calf Intestinal Alkaline Phosphatase ( CIP , New England Biolabs ) for 30 min at 37°C . For co-immunoprecipitation of TAP-tagged Bfa14A and 3XHA-tagged Kin4 , Kin4 was purified with anti-HA ( Roche ) followed by protein A-agarose ( sigma ) and co-precipitates were blotted with peroxidase anti-peroxidase ( PAP , Sigma ) for Bfa1 or Bfa14A . For western blot analysis , peroxidase anti-peroxidase ( PAP , Sigma ) , monoclonal anti-HA ( Roche ) , monoclonal anti-Myc ( Roche ) , polyclonal anti-GFP ( Santa Cruz Biotechnology ) , Clb2 ( produced in the lab ) , monoclonal anti-α-Tubulin ( Sigma ) , and polyclonal anti-Actin ( Santa Cruz Biotechnology ) were used . Band intensity was quantified and analyzed using the LAS-3000 image analyzer ( Fujifilm ) and Image J software ( Version 1 . 38u , NIH ) . GST-Cdc5 ( Δ70N-Cdc5 harboring S165D and T238D ) and GST-Cdc5KD ( Δ70N-Cdc5 harboring an N209A ) were expressed in S . cerevisiae , as described by Geymonat et al . [9] . Tem1 , GST-Bub2 , MBP-Bfa1 , and MBP-D8-Bfa1 were prepared from E . coli , as described by Kim et al . [10] . The intrinsic or GAP-stimulated GTPase activity of Tem1 was quantified , as described by Kim et al . [10] with an EnzCheck Phosphate Assay Kit ( E-6646; Molecular Probes ) . The amount of γ-Pi released from Tem1-GTP was monitored by measuring the absorbance at 360 nm . Yeast two-hybrid assays were performed as previously described Kim et al . [22] . In vitro kinase assays were performed , as described by Geymonat et al . [9] . For radioactive kinase assays , 100 ng of substrate was mixed with 10–50 ng of either GST-Cdc5 or GST-Cdc5KD in 15 µl kinase buffer ( 50 mM Tris-Cl , pH 7 . 5 , 10 mM MgCl2 , 1 mM dithiothreitol ) with 50 µM ATP and 0 . 1 µl γ-[32P]ATP ( Amersham Biosciences , 370 MBq/ml , 3000 Ci/mmol ) . After incubation at 30°C for either 15 min for full-length Bfa1 or 30 min for Bfa1-D8 , Laemli buffer was added to stop the phosphorylation reaction . γ-[32P]-labeling was visualized by autoradiography . To better detect the phosphorylation of Bfa1-D8 by Cdc5 , 1–10 µg substrates were incubated with 1–3 µg of either GST-Cdc5 or GST-Cdc5KD in the same kinase buffer with 0 . 5 mM non-radioactive ATP , separated on 7 . 5% SDS-PAGE containing 100 µM Phos-tag acrylamide ( MANAC Incorporated ) [42] and 200 µM MnCl2 , and stained with Coomassie brilliant blue . After in vitro phosphorylation of Bfa1 was performed with purified Cdc5 kinase and the product was digested with trypsin , liquid chromatography was carried out on a Dionex LC Packings nano HPLC system ( LC-Packings ) coupled to the QSTAR Pulsar ESI-hybrid Q-TOF tandem mass spectrometer ( Applied Biosystems ) , as described in Lee et al [43] . The column outlet was coupled directly to the high voltage ESI source ( typically 2 . 3 kV ) and peptides eluting from the column were sprayed directly into the orifice of the mass spectrometer . Information-Dependent Acquisition ( IDA ) mode was performed to acquire MS/MS spectra based on an inclusion mass list and dynamic assessment of relative ion intensity . For MS/MS , a full mass scan range mode was m/z = 100–2000 Da . After determining the charge states of an ion on zoom scans , product ion spectra were acquired in MS/MS mode with relative collision energy of 55% . The individual spectra from MS/MS were processed using the Analyst QS software ( v1 . 1 , Applied Biosystems ) and searched against a limited database containing only the protein of interest , Bfa1 , which was performed with mass tolerance 0 . 1 Da and with a confidence value no less . | During mitosis the replicated chromosomes are distributed equally to the daughter cells . Once the chromosomes have segregated properly , a pathway called the mitotic exit network ( MEN ) becomes activated to complete mitosis . How MEN activation is coordinated with segregation of the chromosomes is currently a focus of interest . In budding yeast , Tem1 initiates MEN activation and Bfa1 negatively regulates Tem1 with Bub2 . The polo kinase Cdc5 also activates MEN by directly phosphorylating and inhibiting Bfa1 . The spindle pole body ( SPB ) , which corresponds to the mammalian centrosome , acts as a platform for these MEN components . The Bfa1/Bub2 complex localizes to SPBs and regulates the association of Tem1 with the SPBs . When the spindle aligns correctly along the mother-bud axis , Bfa1/Bub2 is restricted to the bud-oriented SPB . Conversely , when the spindle is misaligned , Bfa1/Bub2 is present on both SPBs and mitotic exit is delayed , suggesting that the spatial distribution of Bfa1/Bub2 controls the timing of mitotic exit . In this study , we identified new Cdc5 target phosphorylation residues in Bfa1 that function in its asymmetric distribution on SPBs and showed that the asymmetric Bfa1 distribution was required for timely mitotic exit during unperturbed cell cycle of the budding yeast . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"mitosis",
"model",
"organisms",
"molecular",
"cell",
"biology",
"cell",
"division",
"cell",
"biology",
"gene",
"regulation",
"genetics",
"molecular",
"genetics",
"biology",
"yeast",
"and",
"fungal",
"models",
"saccharomyces",
"cerevisiae",
"microbiology",
"genetics",
... | 2012 | Cdc5-Dependent Asymmetric Localization of Bfa1 Fine-Tunes Timely Mitotic Exit |
Inhibition of human papillomavirus ( HPV ) replication is a promising therapeutic approach for intervening with HPV-related pathologies . Primary targets for interference are two viral proteins , E1 and E2 , which are required for HPV replication . Both E1 and E2 are phosphoproteins; thus , the protein kinases that phosphorylate them might represent secondary targets to achieve inhibition of HPV replication . In the present study , we show that CX4945 , an ATP-competitive small molecule inhibitor of casein kinase 2 ( CK2 ) catalytic activity , suppresses replication of different HPV types , including novel HPV5NLuc , HPV11NLuc and HPV18NLuc marker genomes , but enhances the replication of HPV16 and HPV31 . We further corroborate our findings using short interfering RNA ( siRNA ) -mediated knockdown of CK2 α and α’ subunits in U2OS and CIN612 cells; we show that while both subunits are expressed in these cell lines , CK2α is required for HPV replication , but CK2α’ is not . Furthermore , we demonstrate that CK2α acts in a kinase activity-dependent manner and regulates the stability and nuclear retention of endogenous E1 proteins of HPV11 and HPV18 . This unique feature of CK2α makes it an attractive target for developing antiviral agents .
Human papillomaviruses ( HPVs ) infect cutaneous and mucosal basal epithelial cells retaining an important position among sexually transmitted infections worldwide . Although the majority of HPV infections are transient , persistent infection with high-risk ( HR ) HPVs may instigate cellular transformation and cancer , whereas infection with low-risk ( LR ) HPVs induces benign tissue changes and warts [1] . At least 15 different virus types belong to HR HPVs contributing to the pathogenesis of virtually all cases of cervical cancer and a subset of other epithelial tumors such as head and neck cancers [2][3] . Although vaccination against the most prevalent HPV types is available , cervical cancer is also associated with HPV subtypes not covered by the vaccines [4] . Additionally , therapeutic strategies targeting already existing HPV infections harbored by up to 20% of the population are required to prevent and alleviate HPV-derived pathologies . The design and applications of antiviral drugs depend on thorough understanding of the HPV infection cycle and its relationship with host cells . The HPV infection cycle is strictly dependent on the epithelial differentiation program . Infection begins with intrusion of HPV into mitotically active basal epithelial stem cells , followed by virus genome uncoating , transport to the nucleus and expression of viral early genes E1 and E2 . E1 and E2 encode the only viral proteins required for HPV genome replication throughout all phases of the virus life cycle . E1 is an ATP-dependent DNA helicase that forms a double-hexamer in its enzymatically active form and unwinds DNA , whereas transcription factor E2 assists in the loading of E1 onto the viral origin , thus participating in the initiation of DNA replication ( reviewed in [5] ) . Due to their crucial functions , E1 and E2 proteins are in particular focus to intervene with HPV-related pathologies . Only a few chemical inhibitors targeting either ATP-ase activity of E1 proteins or E1/E2 interaction are available [6] . However , these inhibitors are either inefficient in cell-based assays or highly specific for particular HPV types . In addition , two small molecule inhibitors targeting cellular topoisomerase function have been shown to inhibit the replication of multiple HPV types [7] . On the other hand , E1 and E2 are phosphoproteins , and adjustment of their activities to support replication of the viral genome relies on recruitment of host cell protein kinases that might represent secondary targets to achieve inhibition of HPV replication . Targeting protein kinases has been proven to be a very effective way to modulate cellular physiology and thereby treat a number of disorders . Imanitib ( BCR-Abl antagonist ) and gefitinib ( EGF receptor antagonist ) have been identified as excellent examples of protein kinase inhibitor drugs ( reviewed in [8] ) . Protein kinases that have been shown or suspected to be involved in the regulation of E1 and/or E2 activities and cellular localization include protein kinases A and C , MAP kinases , casein kinases 1 and 2 , cyclin dependent kinase ( cdk ) and FGF receptor 3 [9][10][11][12][13][14][15][16] . Casein kinase 2 ( CK2 ) is an ubiquitously expressed dual specificity protein kinase involved in almost all aspects of cellular life . Additionally , CK2 is an anti-apoptotic kinase upregulated in multiple cancers ( reviewed in [17] ) . It has been shown that CK2-dependent phosphorylation of bovine papillomavirus ( BPV ) E1 and E2 leads to loss of their DNA binding activity and attenuation of viral DNA replication [12] . Phosphorylation of putative CK2 sites in the hinge region of BPV E2 protein leads to its proteasomal degradation and reduction of BPV replication [18] . In contrast , CK2-mediated phosphorylation of the cellular protein BRD4 facilitates its interaction with HR HPV E2 proteins , which is required for proper segregation of the viral genome during mitosis , thereby rendering a positive effect on virus infection [19][20] . Although CK2-mediated inhibition of E1 DNA binding activity has been shown to be conserved in vitro in HPV11 and HPV31 , the impact of CK2 on the replication of HPV genomes has remained obscure . In cells , CK2 is mainly present as a heterotetrameric holoenzyme consisting of two catalytic ( α and/or α´ , are encoded by the CSNK2A1 and CSNK2A2 genes , respectively ) and two regulatory ( ββ , is encoded by the CSNK2B gene ) subunits . Additionally , CK2 subunits may function independently of the holoenzyme ( reviewed in [21] ) . Despite the overall similarity and overlapping biochemical characteristics , CK2α and CK2α´ isoforms exhibit a number of functional specializations , such as affinity to the β subunit , cellular localization , phenotypic differences of knock-out mice , involvement of CK2α´ in cellular proliferation , and specific interactions of the unique C-terminus of CK2α [22][23][24][25] . There is also evidence for differential substrate specificity of CK2 subunits: for instance , caspase-3 is phosphorylated exclusively by CK2α´ kinase and only in the absence of CK2β [26] . It is not yet clear whether this phenomenon is an exception to a rule , since most studies have explored the activity of only one CK2 protein , mainly CK2α , further extrapolating the obtained data to both catalytic subunits . In the present study , we analyzed the impact of CK2 on the replication of different HPV genomes using CK2 RNAi and CX4945-mediated inhibition of CK2 catalytic activity . We have taken advantage of the U2OS cells that efficiently support the replication of various HPV types and used adult pooled normal human epithelial keratinocytes ( NHEKs ) as well as HPV31b-positive CIN612 cells [27] . Our study reveals that efficient replication of various HPV subtypes requires the CK2α subunit , which regulates the stability and nuclear retention of endogenous E1 protein in a kinase activity-dependent manner .
Several lines of evidence suggest the involvement of CK2 in the regulation of the PV life cycle . However , the direct impact of CK2 on the replication of different HPV types has not been analyzed . Additionally , it is not clear whether the catalytic activity of CK2 is involved in the regulation of the HPV life cycle . To analyze the overall impact of CK2 catalytic activity on HPV replication , we tested the HPV5 ( β genus , cutaneous ) , HPV11 ( α genus , LR , mucosal ) , HPV18 ( α genus , HR , mucosal ) and HPV31 ( α genus , HR , mucosal ) genomes in a transient replication assay in U2OS challenged with different concentrations of ATP-competitive small molecule CK2 inhibitor CX4945 ( Fig 1A ) . Our data revealed that CX4945 suppresses the replication of the HPV5 , HPV11 and HPV18 genomes in a concentration-dependent manner . In contrast , replication of HPV31 was not downregulated in the presence of CX4945 . Level of HPV31 increased in response to 6 μM CX4945 suggesting that CK2 kinase activity may have a variable impact on different HPV types . Efficiency of CX4945 was tested by analyzing the status of CK2β subunit known to be phosphorylated by CK2 catalytic subunits [28] ( S1A Fig ) . CK2β was over-expressed in U2OS cells treated with 3 and 6 μM CX4945 for 24 h . Whole cell extracts ( WCEs ) were analyzed using Western blot ( WB ) . Two bands , with the phosphorylated form of CK2β migrated above the non-phosphorylated protein were detected in the control cells , whereas the level of the phosphorylated form of CK2β was severely reduced in the cells treated with 3 and 6 μM CX4945 . To rule out the possibility that inhibition of the HPV replication by CX4945 is a result of alterations in the cell cycle progression , we analyzed the cell cycle profiles of the HPV18-transfected U2OS cells incubated for 2 , 3 or 6 days in the presence of dimethyl sulfoxide ( DMSO ) or different concentrations of CX4945 ( S1B Fig , Fig 1B , and S1C Fig , respectively ) . During the course of the experiment , the cell cycle profile shifted towards G0/G1 phase that is a common feature of a cell culture with continuously increasing cell density ( 62% , 66% and 78% of the cells were detected in the G0/G1 phase at 2 , 3 and 6 days , respectively ) . In contrast , CX4945 induced a slight shift of the cell cycle towards G2/M and S phases at the expense of cells in the G0/G1 phase . We propose that these small changes are not significant and therefore are not likely to be responsible for the changes in virus replication observed . The viability of U2OS cells challenged with different concentrations of CX4945 was examined using methylthiazolyldiphenyl-tetrazolium bromide ( MTT ) assay ( Fig 1C ) . Our results showed that the viability of U2OS cells remained similar to the control cells in all samples except the cells treated with the highest concentration of CX4945 ( 6 μM ) for 6 days that resulted in approximately 40% loss of viable cells . Nevertheless , the negative effect of CX4945 on HPV replication is clearly detectable on the 3rd day of incubation when the viability of the cells was similar to the control ( Fig 1A and 1C ) . Additionally , in contrast to HPV5 , HPV11 and HPV18 , replication of HPV16 and HPV31 increased in response to 6 μM CX4945 during prolonged incubation ( Fig 1A and 1D ) . Therefore , the reduced viability of the U2OS cells in the presence of 6 μM CX4945 could not lead to the inhibition of replication of the particular HPV types . To test the involvement of CK2α and CK2α’ catalytic subunits in the regulation of HPV replication , we applied RNAi using siRNAs specific to either of the CK2 catalytic subunits . U2OS cells were transfected with HPV5 , HPV11 , HPV18 , HPV31 and HPV16 genomes and CK2α- and/or CK2α’-specific siRNAs . Negative siRNA ( Neg . siRNA ) was used as a control . One set of cells transfected with Neg . siRNA was treated with 6 μM CX4945 to assess the impact of CK2 catalytic activity on HPV replication during prolonged incubation . Levels of replicated HPV genomes were analyzed using Southern blot ( SB ) after 3 , 4 and 5 days of incubation ( Fig 1D ) . The levels of all tested HPVs increased in time in the samples treated with Neg . siRNA confirming that HPV genomes replicate in the proliferating U2OS host cells and that their copy number per cell increases over time . Similar to our previous results ( compare Fig 1A and 1D ) , CX4945 effectively inhibited the replication of the HPV5 , HPV11 and HPV18 , but upregulated the replication of the HPV31 and HPV16 genomes . Interestingly , silencing of the CK2α subunit led to strong inhibition of replication of all tested HPVs , including HPV31 and HPV16 . A similar effect was observed in cells treated with a mix of siCK2α and siCK2α’ . In contrast , knockdown of CK2α’ had no negative effect on replication of all tested HPVs except HPV16 . Nevertheless , CK2α RNAi had more profound negative effect even in the case of HPV16 . To express the effects of CK2 RNAi and CX4945 challenges in quantitative terms , all SB panels from at least 3 independent experiments obtained for each HPV type were quantified ( S2 Fig ) . The obtained results confirmed the conclusions drawn . Taken together , our data indicate that only the CK2α subunit acts as a positive regulator of HPV5 , HPV11 , HPV16 , HPV18 and HPV31 transient replication in U2OS cells , whereas CK2α’ is not required for replication of these HPV types . The efficiency of siCK2α , siCK2α’ and the mixture of the two was tested in U2OS cells using WB ( Fig 1E ) . Levels of CK2 catalytic subunits remained below the detection limit in the samples treated with the respective siRNAs . To test the long-term effect of CX4945 on the replication of the HPV18 genome and perform a rescue experiment , U2OS cells were transfected with HPV18 genome , the next day treated with CX4945 and incubated for 3 , 6 , 9 or 12 days . One set of the cells challenged with CX4945 for 6 days was switched to treatment with DMSO and incubated for additional 3 or 6 days ( 9 and 12 days post transfection in total , respectively ) . The level of HPV18 replication was assayed using SB ( Fig 1F left panel ) , and SB signals from three independent experiments were quantified ( Fig 1F right panel ) . After the 6th day of incubation , HPV18 copy number remained constant ( Fig 1F lanes 3 , 7 and 9 ) . The level of HPV18 was similar in all samples treated continuously with CX4945 ( Fig 1F lanes 2 , 4 , 8 , 10 ) . However , termination of treatment with CX4945 led to increase of HPV18 copy number after 6 days of rescue ( Fig 1F lane 6 vs 5 and 10 ) . Depletion of the CK2α subunit in cells had a profound negative effect on HPV replication . Next , we performed an opposite experiment and analyzed the impact of CK2 overexpression on HPV replication . We generated constructs encoding N-terminally Flag-tagged wild-type ( wt ) CK2 subunits and their mutants CK2α ( K68R ) and CK2α’ ( K69R ) bearing point mutations in their ATP binding pockets . Catalytic activity of the overexpressed and immuno-purified proteins was tested in vitro using casein as a substrate ( Fig 2A ) . Flag-tagged CK2α and CK2α’ proteins were confirmed to be active kinases able to phosphorylate casein in vitro . In contrast , CK2α ( K68R ) and CK2α’ ( K69R ) mutants demonstrated severely reduced catalytic activity . However , compared to an empty vector , some residual activity was observed suggesting that the indicated point mutation does not completely destroy kinase activity of CK2 subunits . Of the four CK2 proteins overexpressed , only wt CK2α was able to enhance HPV11 and HPV18 replication in U2OS cells ( ~1 . 5–1 . 6 times ) ( Fig 2B and 2C ) . These results are consistent with the results obtained in RNAi and CX4945 inhibitor experiments , and indicate that it is the presence and activity of CK2α that modulate HPV11 and HPV18 DNA replication . Replication of HPV genomes is generally analyzed using either SB or qPCR . These time consuming and expensive methods require purification of total or extrachromosomal DNA prior analysis . Recently , luciferase HPV DNA replication readout assays have been developed [29]; and an HPV18RLuc construct was generated and used for screening for small molecule inhibitors of the HPV life cycle [7] . To make the test system more sensitive , we generated a novel HPV18 genome containing NanoLuc ( NLuc ) encoding sequence in frame with E2 introduced immediately after the E1 stop codon , followed by foot-and-mouth disease virus ( FMDV ) 2A sequence and full-length E2 ORF . The resulting HPV18NLuc construct expresses NLuc fused to the first 24 E2 amino acids N-terminally and self-processed FMDV-2A peptide sequence C-terminally . The inserted NLuc/FMDV-2A encoding sequence enlarges the genome size only by 657 bp , which is much smaller , as compared to the previously reported HPV18RLuc genome ( the inserted sequence was 1077 nt ) . Therefore , this construct can also be a valuable tool to test HPV cell entry inhibitors in a high-throughput format . Additionally , HPV5NLuc and HPV11NLuc constructs were generated using the same approach . The maps of the HPV5NLuc , HPV11NLuc and HPV18NLuc constructs are depicted in S3A Fig . Replication of HPV5NLuc , HPV11NLuc and HPV18NLuc was tested in U2OS cells using luciferase assay ( Fig 3A ) . All of these genomes replicated in U2OS cells as were assessed by increase of normalized NLuc activity at 2 , 3 and 4 days after transfection . We analyzed the behavior of HPV18NLuc genome in more details . First , replication efficiency of HPV18NLuc construct was compared to that of the wt HPV18 genome in U2OS cells . Our data revealed that the mutant and wt genomes replicated similarly ( S3B Fig ) . Next , we transfected U2OS cells with 250 or 400 ng of HPV18NLuc genome in combination with firefly luciferase ( FFLuc ) encoding plasmid and analyzed the replication efficiency of the HPV18NLuc using SB after 3 , 4 and 5 days of incubation ( Fig 3B ) . The data from these experiments confirmed that the HPV18NLuc genome replicated similarly to the wt HPV18 genome ( Figs 3B and 1D , respectively ) . SB signals corresponding to the replicated HPV18NLuc genome were quantified using ImageQuant software ( Fig 3C left panel ) . Alternatively , cells were seeded on a 96-well plate , and NLuc and FFLuc activities were measured at the respective time points ( Fig 3B right panel ) . In both cases , the signals obtained in the cells transfected with 250 ng of the HPV18NLuc genome and incubated for 3 days were set as 1 . Normalized NLuc activity correlated well with quantified SB signals . Linear regression analysis revealed a statistically significant correlation ( R2 = 0 . 96 , p<0 . 0005 ) ( S3C Fig ) indicating that normalized NLuc activity may be used for the analysis of HPV18NLuc genome replication . Replication of the HPV18NLuc genome was inhibited by CX4945 and CK2 RNAi in both luciferase and SB assays ( Fig 3D and 3E , respectively ) . We tested 3 different concentrations of CX4945: 3 , 6 and 9 μM . The luciferase assay showed that 6 and 9 μM CX4945 had similar effects , therefore usage of concentrations higher than 6 μM was not necessary . Replication of HPV18NLuc in cells lacking CK2α was also severely reduced regardless of the initial amount of the genome used for transfection ( Fig 3E ) . Efficiency of the CK2 siRNAs in U2OS cells harboring the HPV18NLuc genome was controlled using WB ( S3D Fig ) . The obtained results were confirmed using HPV5NLuc , HPV11NLuc and HPV18NLuc in the NHEKs ( Fig 3F ) . The cells were transfected with the respective genomes , incubated for 2 days , treated with 1 μM CX4945 or vehicle for additional 2 or 3 days and subjected to the luciferase assay . All tested genomes replicated in NHEKs , and CX4945 effectively inhibited their replication . Taken together , our data introduce a novel fast , sensitive and reliable test system suitable for studying HPV replication in quantitative terms in both U2OS cell line and normal human keratinocytes . Our results clearly show that CX4945 inhibits replication of these genomes , although in principle addition of NLuc sequence into HPV genomes might also affect viral transcription , splicing , mRNA stability and translation . To confirm the obtained results in another cell culture model system supporting HPV replication , we chose HPV31b+ CIN612 keratinocytes . First , we treated CIN612 cells with CK2-specific siRNAs one or two times and tested CK2α and CK2α’ protein levels using immunoblotting ( Fig 4A ) . After the first transfection , both proteins were fairly detected , whereas the double transfection resulted in complete loss of expression of both proteins indicating that CK2 RNAi is functional in CIN612 cells . To be convinced that both antibodies recognize the indicated CK2 proteins , we also monitored the levels of CK2α and CK2α’ mRNAs ( Fig 4B ) in CIN612 cells subjected to single or double transfection with CK2 siRNAs ( Fig 4B left and right panels , respectively ) . Levels of CK2α and CK2α’ mRNAs were reduced more than 90 and 95% after the 1st and 2nd transfection , respectively . Next , CIN612 cells were transfected with CK2-specific siRNAs; total DNA from the transfected cells was treated with the restriction endonuclease linearizing the HPV31b genome , and subjected to SB analysis ( Fig 4C left panel ) . CK2α RNAi induced clear downregulation of HPV31b stable replication in CIN612 keratinocytes , similar to the HPV31 transient replication in U2OS cells . These data were confirmed by quantification of the SB signal obtained from 3 independent experiments ( Fig 4C right panel ) . To test the effect of CX4945 on stable replication of HPV31b , CIN612 cells were treated with different concentrations of CX4945 , incubated for 3 or 6 days and subjected to total DNA isolation and SB analysis ( Fig 4D ) . We could not detect any CX4945-mediated negative effect on HPV31b replication ( Fig 4D ) . Moreover , similarly to transient replication of HPV31 in U2OS cells ( Fig 1A and 1D ) , we observed some upregulation of HPV31b copy number in the presence of CX4945 . However , even the lowest concentration of CX4945 used ( 1 . 5 μM ) was toxic to the cells , as was assessed by MTT assay ( Fig 4E ) . Therefore , we could not analyze the effects of 3 or 6 μM CX4945 on replication of HPV31b during prolonged incubation . One likely explanation for the failure of CX4945 to suppress HPV31b replication in CIN612 keratinocytes relies on its putative ability to induce differentiation of these cells , although it does not explain the CX4945-mediated positive effect on transient replication of HPV31 in U2OS cells . This ability could also explain the slight increase in the HPV31b copy number in CX4945-treated cells , since differentiation of CIN612 cells results in an increase in viral genomes per cell [30] . Support for this hypothesis comes from the study showing that a CK2 inhibitor similar to CX4945 induces differentiation of normal human keratinocytes at least partially in a CK2-independent manner [31] . To rule out this possibility , we analyzed the keratinocyte-specific gene mRNA expression levels in CIN612 cells treated either with DMSO and 0 . 5 μM CX4945 or transfected with Neg . siRNA and siCK2α once or twice and incubated for 3 and 6 days , respectively ( Fig 4F and S4A Fig , respectively ) . Expression levels of keratin 10 ( KRT10 ) , keratin 14 ( KRT14 ) , involucrin ( IVL ) and loricrin ( LOR ) mRNAs were analyzed . Neither CX4945 nor CK2α RNAi could induce expression of these genes indicating that differentiation of CIN612 cells was not induced by these treatments . Also , we analyzed the cell cycle profile in the respectively treated CIN612 cells ( S4B Fig ) . Cell cycle profiles were similar in the cells treated with CX4945 and DMSO for 3 days . Similar to the results obtained using U2OS cells , CX4945 induced a shift towards G2/M phase in the case of prolonged incubation ( Fig 1B and S4B Fig ) . However , a number of cells in the S-phase did not change substantially in time in contrast to the control DMSO-treated cells , which demonstrated strong increase of the cells in the S-phase ( approximately 2 . 4 times ) . Compared to the cells transfected with Neg . siRNA , knockdown of CK2α led to a decrease in cell number in the S-phase ( approximately 40 and 45% after 3 and 6 days of incubation , respectively ) . Therefore , we cannot formally rule out the possible association between the changes in the cell cycle progression and inhibition of HPV31b replication induced by CK2α RNAi . Nevertheless , all our SB analyses were normalized to the amount of the genomic DNA , thereby indicating that we are measuring the HPV31b copy number per cell . Finally , we were interested to examine the ability of CX4945 to inhibit the replication of other HPV types in CIN612 cells . CIN612 cells were transfected with HPV5NLuc , HPV11NLuc and HPV18NLuc genomes , incubated for 2 days , treated with 0 . 5 μM CX4945 for additional 2 and 3 days and subjected to luciferase assay ( Fig 4G left panels ) . Also , total DNA was isolated and subjected to SB analysis to detect HPV31b replication at the same conditions ( Fig 4G right panel ) . As it was expected , replication of HPV31b was not suppressed by 0 . 5 μM CX4945 in time . However , similar to the results obtained in NHEKs and U2OS cells , CX4945 inhibited the replication of HPV5NLuc , HPV11NLuc and HPV18NLuc in CIN612 cells . Helicase E1 and transcription factor E2 are the only viral proteins required for PV replication . It has been shown that CK2 is able to directly phosphorylate the BPV1 E1 and E2 protein , modulating their activities [12] . Since our data show that CK2α is required for efficient replication of different HPV types , we hypothesized that CK2α may mediate its effects via E1 and/or E2 proteins . Transient replication assays of HPV11 and HPV18 revealed that CX4945 mimics the effect of siCK2α and may be used as an advantageous alternative to CK2α RNAi . First , pharmacological intervention with CK2 kinase activity influences all cells in the population . Second , it is much faster than RNAi-mediated knockdown of protein expression , which effectiveness depends on transfection efficiency . To gain insight into the molecular mechanisms of CK2-dependent downregulation of HPV replication , we generated HPV11 and HPV18 genomes containing HA epitope encoding sequence in their E1 ORFs after the 15th nucleotide ( HPV11E1HA and HPV18E1HA , respectively ) . Since the HA epitope could potentially influence the properties of the E1 protein , the replication of these constructs was compared with that of the respective wt genomes in U2OS cells challenged with 6 μM CX4945 or vehicle ( Fig 5A and S5A Fig ) . Our data revealed that the mutant and wt genomes replicated similarly in both cases indicating that the functioning of wt and HA-tagged E1 is similar . Next , we analyzed the level of HA-tagged E1 protein in U2OS cells treated with 6 μM CX4945 on the 2nd day after transfection and incubated for the additional 1 , 3 or 4 days . Concurrently , the level of linearized HPV DNA was analyzed in the same samples using SB . Data from the representative experiments are shown in Fig 5B and S5B Fig for HPV18E1HA and HPV11E1HA , respectively . In contrast to CX4945-treated cells , the levels of HPV18E1HA and HPV11E1HA DNA increased in time in the control cells . We failed to detect the HA-tagged E1 proteins by WB directly from WCEs , but immunoprecipitated E1 protein was detected as a single band migrating at approximately 90 kDa . The levels of E1 protein immunoprecipitated from cells challenged with CX4945 were reduced already after 24 h of treatment , whereas decrease in HPV DNA copy number was not detected at this time point ( compare lanes 1 and 3 in SB and WB panels of Fig 5B and S5B Fig ) . Even further , HPV18 and HPV11 genomes were readily detectable after 72 h of treatment with CX4945 on day 5 , whereas we were unable to detect any E1 protein at this time point ( lanes 5 in SB and WB panels of Fig 5B and S5B Fig ) . These data suggest that the CK2 inhibitor induces the degradation of E1 protein . Next , we analyzed the levels of nuclear and cytoplasmic HPV18 E1 protein in U2OS cells either transfected with siCK2α or treated with 6 μM CX4945 for 24 h . The nuclear and cytoplasmic extracts were isolated; the E1 protein of HPV18E1HA was immunoprecipitated and analyzed using WB ( Fig 5C ) . Compared to the respective control cells , the levels of nuclear and cytoplasmic E1 protein in the cells treated with siCK2α or CX4945 were lower . To exclude the possibility of transcriptional downregulation of E1 expression , we analyzed the levels of HPV18-derived transcripts in U2OS cells either challenged with CX4945 for 24 h or transfected with siCK2α . E1 is encoded by the longest viral pre-mRNA that also includes ORFs of several other genes generated via alternative splicing [32] . The levels of mRNAs corresponding to HPV18 E1 , E2 , E1^E4 and E8^E2 transcripts were analyzed using RT-PCR and qPCR 48 h after transfection ( Figs 5D and S5C , respectively ) . Our data showed that neither CX4945 nor siCK2α inhibited the transcription of the analyzed HPV18 genes . To confirm our observations and exclude the possibility of a CK2 independent “side” effect of CX4945 or a simple decrease in E1 levels in cells possessing fewer copies of HPV18E1HA or HPV11E1HA genomes , we analyzed the levels of the HA-tagged E1 protein of HPV18 or HPV11 in U2OS cells treated with CX4945 and proteasome inhibitor MG132 . The HA-tagged E1 protein was immunoprecipitated and analyzed using immunoblotting and two different antibodies–m-a-HA ( clone HA-7 ) or rat-a-HA-HRP ( clone 3F10 ) ( Fig 6A and S5D Fig for HPV18E1HA and HPV11E1HA , respectively ) . CX4945 induced decrease in E1 protein level was rescued to some extent in the presence of the proteasome inhibitor MG132 , suggesting that E1 is subjected to proteasomal degradation in response to CK2 inhibitor . Also , the WB signals corresponding to HPV18 E1 protein were quantified and set as 100% in the cells treated with DMSO . The data from other samples were calculated relative to 100% . Statistical analysis was performed using T-test assuming equal variations ( Excel ) , and two-tail p values were calculated ( Fig 6A right panel ) . MG132 alone had no significant effect on HPV18 E1 protein levels . However , level of the E1 protein decreased approximately 50% in response to treatment with CX4945 for 6 h , which was rescued in the presence of MG132 . To study this phenomenon in further detail , we isolated the nuclear and cytoplasmic extracts of U2OS cells transfected with HPV18E1HA or HPV11E1HA genomes and treated with 6 μM CX4945 for 2 , 4 , 8 and 12 h . Along with CX4945 treatment , the cells transfected with the HPV18E1HA genome were treated with MG132 for 4 h . E1 proteins of HPV18E1HA and HPV11E1HA were immunoprecipitated and analyzed using immunoblotting ( Fig 6B and S5D Fig , respectively ) . WB data from at least three independent experiments performed using HPV18E1HA genome were quantified . E1 level was set as 100% in the nuclear extracts of cells treated with DMSO . The analysis revealed significantly lower E1 levels in nuclear and cytoplasmic extracts within 4 h of treatment with CX4945 ( Fig 6B right panel ) . The decrease of HPV18 E1 protein level was rescued by MG132 to the levels similar to the ones observed in the control cells . Although we could not fully exclude the possibility of N-terminal proteolysis of full-length E1 proteins , our data suggest that CK2α catalytic activity is required for posttranslational stabilization and nuclear retention of E1 proteins . In contrast to the E1 protein , the levels of nuclear and cytoplasmic CK2α and CK2α’ proteins were not markedly reduced in response to CX4945 in the HPV11E1HA transfected U2OS cells ( S5D Fig ) .
First , we show that pharmacological inhibition of CK2 catalytic activity using the ATP-competitive inhibitor CX4945 leads to suppression of transient replication of HPV5 , HPV11 and HPV18 in U2OS cells . Also , transient replication of HPV5NLuc , HPV11NLuc and HPV18NLuc genomes was inhibited by CX4945 in NHEKs and CIN612 cells . Nevertheless , CX4945 does not inhibit the transient replication of HPV16 and HPV31 genomes in U2OS cells as well as the stable replication of the episomal HPV31b genome in CIN612 cells . Instead , the copy number of HPV16 , HPV31 and HPV31b genomes increased in response to CX4945 . The fact that the replication of these genomes is not suppressed in response to CX4945 is puzzling . Several reasons might explain this phenomenon . First , although CX4945 is currently the best inhibitor for CK2 ( IC50 1 nM in kinase assay in vitro ) , it still targets a number of other kinases at nanomolar range [34] . It may be possible that an “off-target” kinase has a positive effect on HPV16 and HPV31 replication . Second , it has been shown that CX4945 has CK2 independent function as an inhibitor of alternative splicing [35] . Because the splicing patterns and translation efficiency of different polycistronic transcripts varies between HPV types , it is plausible to speculate that CX4945 may alter the translational outcome in a HPV-type-dependent manner . Alternatively , CK2 might possess additional , kinase activity-independent positive roles in the replication of a subset of HPV types , as exemplified by HPV16 and HPV31 in this paper . Such regulatory CK2 activity has been shown in melanoma cells , where CK2 acts as a scaffold to keep ERK kinase active [36] . The last hypothesis is supported by the fact that CK2α RNAi leading to loss of CK2α protein results in inhibition of HPV16 and HPV31 replication in U2OS and HPV31b replication in CIN612 cells . Further studies are needed to comprehensively explain the positive effect of CX4945 on HPV16 , HPV31 and HPV31b replication . The effects of small molecule modulators are often pleiotropic and might be indirect . It might be envisaged that CX4945 induces cell cycle alterations or loss of cell viability in U2OS cells and the effect on HPV replication is non-specific . We ruled out this possibility , by showing that the minor changes in cell cycle and viability of U2OS cells detected in the presence of CX4945 at the concentrations used could not be a reason of such dramatic inhibition of HPV5 , HPV11 and HPV18 replication . Besides , replication of HPV16 and HPV31 was not inhibited at the same experimental conditions . Furthermore , the signals of our replication assays are normalized against the amount of genomic DNA , and therefore , the signal represents copies of the HPV genome per cell . To obtain more precise insight into the function of CK2 in HPV replication , we used siRNA-mediated knockdown of both CK2 catalytic subunits , CK2α and CK2α’ . Our results show that surprisingly only CK2α is required for efficient transient and stable replication of HPV genomes , although both subunits are present in the cells analyzed . Inactivation of CK2α’ had no or little effect on the replication efficiency of the tested HPV types . This result was corroborated by performing the opposite experiment and showing that overexpression of CK2α , and not CK2α’ , stimulates transient replication of HPV11 and HPV18 genomes in U2OS cells . Additionally , we show that the catalytic activity of CK2α is required for this stimulation , since the mutant with impaired kinase activity was not able to enhance HPV replication . CK2 generally functions as a heterotetrameric enzyme , consisting of two catalytic subunits ( CK2α or CK2α’ ) and two regulatory CK2β subunits . Catalytic subunits are highly homologous ( almost 90% ) in their kinase domain and more divergent in C-termini . The identified consensus phosphorylation site is shared by both catalytic subunits ( reviewed in [17] ) . Despite this fact , many interactors have been identified as specific to either of the CK2 catalytic subunits , indicating certain functional divergence [37] . Additionally , a number of substrates are phosphorylated by one or another of the subunits ( [26] and references therein ) , although the sequence specificity underlying this phenomenon has not been discovered . In addition , some CK2 substrates are phosphorylated by the catalytic subunit only and not by the holoenzyme [38][26] . Our results showing that CK2α , and not CK2α’ , regulates HPV replication leave two possible explanations . First , it is conceivable that one or more proteins participating in HPV replication harbor phosphorylation sites unique to CK2α . Second , it is possible that the levels of different CK2 subunits in the cells analyzed are very strongly in favor of CK2α’ , and enough catalytically active CK2α’ remains in the cells even after siRNA-mediated knockdown of CK2α’ subunit . We believe the first possibility to be correct , since the residual expression of CK2α and CK2α’ mRNAs was similar ( approximately 3–5% ) , and the levels of both proteins remained under their detection limits as assessed by immunoblotting in our RNAi experiments . Furthermore , we do not observe a cooperative negative effect on HPV replication if knock-down of both subunits is performed . A positive role of CK2 in HPV replication has been observed previously [19] . This report shows that CK2-mediated phosphorylation of the cellular protein BRD4 is required for efficient replication of E1- and E2-dependent replication of the origin-containing plasmid in C33A cells . The role of phosphorylated BRD4 in replication most likely relies on correct targeting of the E2 protein to the origin of replication [20] . Our present paper confirms the requirement of CK2 in HPV replication and proves that this also holds true in the HPV genome context . Furthermore , our data show that in addition to the role of CK2 in transient replication , it also regulates stable replication of the HPV31b genome . Another phenomenon we observed in the RNAi experiments was that inhibition of stable HPV31b replication in CIN612 cells was milder than inhibition of transient HPV31 replication in U2OS cells . This is expected since the HPV genome is duplicated in synchrony with cellular DNA during stable replication , whereas viral genome copy number is increasing during every cell cycle during initial transient replication . Therefore , the effect on stable replication should be smaller as the number of HPV genome replication initiation events is much less in a given period of time . A number of studies have shown that BPV1 E1 and E2 proteins are phosphorylated by CK2 [39][12][13] . These studies were performed using recombinant viral proteins purified from E . coli or Sf9 cells followed by an in vitro kinase assay . Most of these identified sites are conserved in HPV E1 and E2 proteins . It has been shown that the consequence of these phosphorylations is the inactivation of DNA binding of both E1 and E2 [12] . In addition , BPV1 E1 is phosphorylated in putative CK2 sites in vivo while overexpressed in Sf9 cells , although it cannot be completely ruled out that phosphorylation takes place during purification steps and not in cells [13][40] . The in vivo function of these CK2 sites in BPV1 E1 and E2 proteins has been studied by mutating the sites in the context of BPV genomes and analyzing the consequences in C127 cells [12] [41] . The conclusions drawn from these experiments , however , differ between laboratories . One group has shown that alanine substitutions in putative CK2 sites in E1 have no effect on transient or stable replication of BPV , while mutations in E2 phosphosites have a positive effect on BPV replication [12] . Another group has found , however , that a single glycine substitution in BPV1 E1 serine 48 , which is phosphorylated by CK2 , renders it a genome that is not able to replicate in C127 cells [41] . Taken together , conflicting evidence has been proposed for CK2 in regulating the activity of BPV replication proteins . Our data show that the CK2 inhibitor CX4945 has a negative effect on E1 proteins encoded by HPV11 and HPV18 genomes . We demonstrate that the levels of the E1 proteins decrease markedly upon CK2α RNAi or CX4945 challenge . Moreover , the reduced level of E1 protein is not caused by transcriptional silencing or decreased copy number of the viral genome , since the level of the E1 protein was significantly reduced already after four hours of challenge with CX4945 and proteasome inhibitor MG132 alleviated CX4945-induced degradation of E1 protein . Thus far , the effect of CK2 on the stability of E1 protein has not been shown . Instead , cdk2-mediated phosphorylation has been shown to stabilize BPV E1 protein in vitro [42][43] . Analysis of the in vivo effect of cdk has shown the requirement of the kinase for nuclear retention of HPV E1 [14][15] . Therefore , our data provide for the first time a glimpse of how the stability of HPV E1 protein is regulated in vivo by CK2 .
Parental plasmids encoding HPV5 , HPV11 and HPV18 genomes on the basis of the pMC . BESPX minicircle production vector have been described previously [44][45][46] . The hemagglutinin ( HA ) epitope encoding sequence was inserted into the HPV11pMC . BESPX and HPV18pMC . BESPX parental plasmids in their E1 open reading frames ( ORFs ) after the 15th nucleotide starting from the E1 1st AUG . The resulting constructs were named HPV11E1HApMC . BESPX and HPV18E1HApMC . BESPX . The HPV18NLucpMC . BESPX construct was generated by cloning the codon optimized NLuc encoding sequence followed by the 2A region of the FMDV and the wt E2 encoding sequence after the 72nd nucleotide of the HPV18 E2 ORF that corresponds to the E1 stop codon . Along with wt HPV18 proteins , the resulting construct contains NLuc fused with 24 amino acids of E2 protein N-terminally and self-processed FMDV-2A sequence C-terminally . The HPV11NLucpMC . BESPX and HPV5NLucpMC . BESPX constructs were generated using the same approach . All of the abovementioned HPV genomes were generated as minicircle plasmids in E . coli strain ZYCY10P3S2T using minicircle DNA technology as previously described [46] . The HPV16 and HPV31 genomes have been previously described [27] . Plasmids encoding N-terminally Flag-tagged CK2α and CK2α’ subunits were generated by cloning CK2α and CK2α’ ORFs lacking the 1st ATG into the pCMV-Flag-4 vector ( Sigma-Aldrich ) between the HindIII and BamHI sites . Flag-tagged CK2α and CK2α’ kinase activity-deficient mutants CK2α ( K68R ) and CK2α’ ( K69R ) bearing a point mutation in their ATP binding pockets at positions 68 and 69 , respectively , were generated by PCR mutagenesis using CK2αpCMV-Flag-4 and CK2α’pCMV-Flag-4 templates and the following primers: CK2α ( K68R ) sense ( s ) GAAAAAGTTGTTGTTAGAATTCTCAAGCCAG , CK2α ( K68R ) antisense ( as ) CTGGCTTGAGAATTCTAACAACAACTTTTTC , CK2α’ ( K69R ) s GAGAGAGTGGTTGTAAGAATCCTGAAGCC , CK2α’ ( K69R ) as CTTCACTGGCTTCAGGATTCTTACAACCACTC . All new constructs were verified by DNA sequencing . All plasmid DNAs were purified from bacteria using the NucleoBond Xtra Midi EF Kit ( Macherey-Nagel ) . The human osteosarcoma cell line U2OS ( ATCC No HTB-96 ) was propagated in normal growth medium ( NGM ) containing Iscove's Modified Dulbecco's Medium ( IMDM , Pan Biotech ) , 10% fetal calf serum ( FCS ) and 1% penicillin/streptomycin ( PEST , Sigma-Aldrich ) at 37°C and 5% CO2 . Cells were transfected by electroporation ( 220 V and 975 μF ) using a Gene Pulser XCell system ( Bio-Rad Laboratories ) . The following amounts of wt or modified HPV minicircle genomes were used for transfection of 106 U2OS cells: HPV5 1 . 5 μg , HPV11 0 . 3 μg , HPV18 1 . 2 μg . The HPV16 and HPV31 genomes were religated and transfected as previously described [27] . We transfected 250 and 125 ng of CK2αpCMV-Flag-4 and CK2α’pCMV-Flag-4 constructs , respectively , per 106 U2OS cells . Transfected U2OS cells were seeded at an approximate density of 3x104 per 1 cm2 . CIN612 cells and NHEKs ( PromoCell ) ( kind gift from Dr . Frank Stubenrauch and Dr . Ana Rebane , respectively ) were grown in Defined Keratinocyte-SFM Medium ( DKSM ) ( Gibco , Thermo Fisher Scientific ) . Cells were detached using 0 . 25% Trypsin-EDTA solution , immediately transferred into DKSM containing 25% NGM , centrifuged at room temperature ( RT ) and 1400 rpm for 3 min and plated onto a new culture dish and fresh DKS after every 3–4 days . CIN612 cells and NHEKs were transfected with 1 . 5 μg of HPVNLuc minicircle genomes using 1 . 5 μl Plus reagent and 2 . 5 μl Lipofectamine LTX ( Invitrogen , kind gift of Dr . Eva Žusinaite ) per 12 wells of 96-well plate . 293T cells ( ATCC no CRL-3216 ) were propagated in DMEM-high glycose ( Sigma-Aldrich ) supplemented with 10% FCS and 1% PEST . CX4945 was purchased from Santa Cruz Biotechnology and MG132 ( kind gift of Dr . Reet Kurg ) was purchased from Sigma-Aldrich . Both chemicals were diluted in DMSO . Total and extrachromosomal DNA was extracted from cells as described previously [27] . DNA isolated from transfected U2OS cells was treated with DpnI restriction enzyme to digest bacterially methylated input DNA . The following restriction endonucleases were used to linearize HPV genomes: SacI ( HPV5 ) , HindIII ( HPV11 and HPV31b ) , BglI and BstXI ( HPV18 and its mutants ) , EcoRI ( HPV31 ) , and BamHI ( HPV16 ) . Southern transfer and hybridization were performed as described [27] . The following amounts of total DNA were used for the detection of different HPV genomes: HPV5–5 μg; HPV11–2 μg; HPV18–5 μg; HPV31–10 μg; and HPV31b ( CIN612 ) – 1 . 5 μg . HPV16 genome was detected using 30 μg of Hirt extract . All SB assays were performed at least three times , and representative images are shown . SB signals corresponding to the replicated HPV genomes were quantified using ImageQuant software . The following siRNAs were used: CK2α UGUCCGAGUUGCUUCCCGA 20 nM , CK2α’ GCUGCGACUGAUAGAUUGG 30 nM , scrambled Neg . UAGCGACUAAACACAUCAA ( Sigma-Aldrich ) . All siRNAs contained dTdT overhang . Initially , U2OS cells were transfected with siRNAs together with HPV genomes by electroporation . In the case of prolonged incubation , siRNAs were additionally delivered on the 3rd day after the 1st transfection using Lipofectamine RNAiMax reagent ( Thermo Fisher Scientific ) according to the manufacturer’s instructions . Total RNA was isolated using the Quick RNA MiniPrep Kit ( Zymo Research ) . Approximately 10 μg of total RNA was treated with 8 U of Turbo DNase ( Thermo Fisher Scientific ) for at least 4 h and subsequently precipitated with 7 . 5 M LiCl . Complementary DNA was synthesized using 1 . 2 μg of total RNA , oligo ( dT ) and RevertAid First Strand cDNA Synthesis Kit ( Thermo Fisher Scientific ) in the presence or absence of Reverse Transcriptase , if indicated . RT-PCR and quantitative RT-PCR ( qPCR ) were performed using HOT FIREPol PCR Mix and HOT FIREPol EvaGreen qPCR Mix ( Solis Biodyne ) , respectively . The primers used are listed in S1 Table . The data of qPCR analyses are expressed as normalized with ACTB or GAPDH average means ± SD of three measurements obtained from at least three independent experiments . NLuc and FFLuc activities were measured using a Nano-Glo Dual-Luciferase Reporter Assay System ( Promega ) according to the manufacturer’s instructions for 96-well plates . Alkaline phosphatase activity and total protein concentrations were measured as previously described [47] and used for normalization of NLuc activity . U2OS cells were transfected with HPV11E1HA or HPV18E1HA minicircle genomes , incubated for the indicated periods of time , washed with PBS and lysed . WCEs of approximately 107 cells were used for each immunoprecipitation ( IP ) of endogenous HA-tagged E1 . WCEs were prepared in RIPA buffer ( 50 mM Tris pH 7 . 5 , 150 mM NaCl , 2 mM EDTA , 0 . 1% SDS and 0 . 1% TRITON-X100 ) supplemented with protease inhibitor cocktail ( PIC , Roche ) . Approximately 2x107 cells were fractioned to cytoplasmic and nuclear lysates using NE-PER Nuclear and Cytoplasmic Extraction Reagents ( Thermo Fisher Scientific ) . Cytoplasmic and nuclear extracts were diluted in 5 ml of RIPA buffer containing PIC . Lysates were incubated with m-anti-HA ( Sigma-Aldrich ) or r-a-HA ( Labas Ltd . ) antibodies ( 2 or 3 μg per sample , respectively ) and protein A Sepharose at 4°C at slow rotation overnight . Immuno-complexes were washed 3 times with 5 ml of RIPA buffer at 4°C for 15 min , lysed in Laemmli sample buffer , denatured at 100°C for 5 min and subjected to SDS-PAGE analysis . WB was performed as previously described [48] . The following antibodies were used: m-a-HA ( clone HA-7 , Sigma-Aldrich ) 1:3000 , rat-a-HA-HRP ( clone 3F10 , Sigma-Aldrich ) 1:1500 , r-a-HA ( Labas Ltd . ) 1:2000 , m-a-CK2α ( Santa Cruz Biotechnologies ) 1:300 , m-a-CK2α’ ( Santa Cruz Biotechnologies ) 1:300 , m-a-GAPDH ( Sigma-Aldrich ) 1:10000 , m-a-lamin B ( Santa Cruz Biotechnologies ) 1:500 , a-Flag-HRP M2 ( Sigma-Aldrich ) 1:5000 , m-a-tubulin α ( Sigma-Aldrich ) 1:5000 , m-a-myc ( Sigma-Aldrich ) 1:1000 . The SuperSignal West Dura Extended Duration Substrate ECL Kit was used for the detection of E1-HA and CK2α proteins; other proteins were detected using the SuperSignal West Pico Kit ( both Thermo Fisher Scientific ) . All immunoblotting assays were performed at least two times , and representative images are shown . Flag-tagged CK2 catalytic subunits or their kinase activity-deficient mutants were overexpressed in 293T cells for 48 h and immuno-purified using Flag-M2 affinity binding resin ( Sigma-Aldrich ) according to the manufacturer’s instructions . Immuno-complexes were washed two times with ice-cold kinase buffer ( 50 mM Tris-HCl pH 7 . 4 , 100 mM MgCl2 , 20 μM ATP ) . Kinase reactions were carried out in the kinase buffer in the presence of 1 μCi [γ-32P]-ATP and 1 μg of casein at RT for 30 min , stopped by addition of Laemmli sample buffer , incubated at 100°C for 5 min and subjected to SDS-PAGE . Cell cycle profiles of the HPV18-transfected U2OS cells or CIN612 cells treated with DMSO or different concentrations of CX4945 were analyzed as previously described [7] . Viability of the cells in the presence of different concentrations of CX4945 was measured using MTT assay . The cells were grown on 96-well plates . MTT 0 . 5% solution was added directly to 100 μl of growth media , and the cells were incubated at 37°C and 5% CO2 for 2–3 h until purple precipitate was formed . The incubation media was aspirated; the cells were treated with 100 μl of DMSO and incubated on a shaker at RT for 30 min . Optical density was measured at 540 nm . | Human papillomaviruses ( HPVs ) are small DNA viruses that infect epithelial cells and can cause a variety of hyperplastic changes in the infected tissue . Replication of the HPV genomes relies largely on the host cell factors . The only viral proteins that are necessary for the virus replication are E1 and E2 . Relatively little is known about the host cell factors that modulate the activities of E1 and E2 proteins and thereby regulate replication of the HPV genome . We have discovered that the activity of an ubiquitously expressed protein kinase CK2 is needed for efficient replication of a number of different HPV types . We further show that only one of the two catalytic subunits of CK2 existing in the cell , CK2α , is required for the replication of HPV genomes , whereas a closely related CK2α’ has no or little effect . Finally , we demonstrate that inhibition of CK2 activity results in rapid degradation of the E1 helicase and thereby inefficient replication of the HPV genomes . Taken together our results delineate the function of a host cell protein kinase CK2 in the HPV life cycle and describe a surprising differential effect of two closely related catalytic subunits of CK2 in the replication of HPV genomes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"phosphorylation",
"urology",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"gene",
"regulation",
"pathogens",
"hpv-11",
"microbiology",
"viruses",
"dna",
"replication",
"dna",
"viruses",
"sexually",
"transmitted",
"diseases",
"h... | 2019 | Activity of CK2α protein kinase is required for efficient replication of some HPV types |
The acceleration of the control of soil-transmitted helminth ( STH ) infections in Nigeria , emphasizing preventive chemotherapy , has become imperative in light of the global fight against neglected tropical diseases . Predictive risk maps are an important tool to guide and support control activities . STH infection prevalence data were obtained from surveys carried out in 2011 using standard protocols . Data were geo-referenced and collated in a nationwide , geographic information system database . Bayesian geostatistical models with remotely sensed environmental covariates and variable selection procedures were utilized to predict the spatial distribution of STH infections in Nigeria . We found that hookworm , Ascaris lumbricoides , and Trichuris trichiura infections are endemic in 482 ( 86 . 8% ) , 305 ( 55 . 0% ) , and 55 ( 9 . 9% ) locations , respectively . Hookworm and A . lumbricoides infection co-exist in 16 states , while the three species are co-endemic in 12 states . Overall , STHs are endemic in 20 of the 36 states of Nigeria , including the Federal Capital Territory of Abuja . The observed prevalence at endemic locations ranged from 1 . 7% to 51 . 7% for hookworm , from 1 . 6% to 77 . 8% for A . lumbricoides , and from 1 . 0% to 25 . 5% for T . trichiura . Model-based predictions ranged from 0 . 7% to 51 . 0% for hookworm , from 0 . 1% to 82 . 6% for A . lumbricoides , and from 0 . 0% to 18 . 5% for T . trichiura . Our models suggest that day land surface temperature and dense vegetation are important predictors of the spatial distribution of STH infection in Nigeria . In 2011 , a total of 5 . 7 million ( 13 . 8% ) school-aged children were predicted to be infected with STHs in Nigeria . Mass treatment at the local government area level for annual or bi-annual treatment of the school-aged population in Nigeria in 2011 , based on World Health Organization prevalence thresholds , were estimated at 10 . 2 million tablets . The predictive risk maps and estimated deworming needs presented here will be helpful for escalating the control and spatial targeting of interventions against STH infections in Nigeria .
Soil-transmitted helminth ( STH ) infections belong to the neglected tropical diseases ( NTDs ) . In terms of at-risk population and number of people infected , the STHs are the most frequent NTDs worldwide . The three common STHs are the roundworm ( Ascaris lumbricoides ) , the whipworm ( Trichuris trichiura ) , and the hookworms ( Ancylostoma duodenale and Necator americanus ) [1–3] . The most recent estimates suggest that 819 million people worldwide are infected with A . lumbricoides , 465 million with T . trichiura , and 439 million with hookworm [4] . STH infections thrive where there are poor hygiene practices , including limited environmental sanitation , unsafe water sources , inadequate toilet facilities , and poor fecal disposal methods , coupled with poverty and low household income [5–7] . School-aged children ( 5–14 years ) , in particular , are at high risk of infection and morbidity due to STHs , and hence , are the main target of preventive chemotherapy [8 , 9] . Nigeria has the highest total number of people infected with STHs in sub-Saharan Africa [10–12] . However , there is a paucity of empirical data on the spatial distribution of STH infections and this has hindered control . The planning , implementation , and rigorous monitoring of a national control program targeting STH infection can be enhanced with detailed knowledge of the spatial and temporal distribution of infection and morbidity [13] . In light of the global commitment to escalate the control of NTDs [14–17] , knowledge of the spatial distribution of STH infections is a necessary prerequisite for the implementation of control and elimination measures , such as large-scale administration of anthelmintic drugs . Thus far , two NTD-specific risk maps have been published for Nigeria; onchocerciasis [18] and schistosomiasis [19] . As the country prepares to implement large-scale preventive chemotherapy campaigns against STH infection , a nationwide map of the spatial distribution of STH using available survey data can help in advocacy , resource sourcing for funds , and implementation of control/elimination activities . The purpose of the current study was to produce high-resolution STH infection risk maps , including estimated number of school-aged children infected with A . lumbricoides , T . trichiura , and hookworm in Nigeria . We used recently obtained survey data and employed Bayesian geostatistical models to predict STH infection risk across Nigeria . Additionally , we computed annualized treatment requirements with the anthelmintic drugs albendazole and mebendazole . An important aspect of this study is to provide STH program managers with information for effective implementation of STH control activities .
The work presented here is derived from an in-depth analysis of STH infection survey data obtained from the Federal Ministry of Health ( FMoH ) of Nigeria in 2011 . Ethical clearance , informed consent procedures , and treatment were according to FMoH guidelines and recommendations . The data are aggregated and do not contain identifiable individual or household level information . Thus , no specific ethical approval was required for the secondary analysis presented here . In 2011 , a national survey was carried out in Nigeria , pertaining to STH infection among children aged 5–14 years . The overreaching goal of the survey was to prepare the country for mass drug administration with albendazole or mebendazole , provide evidence-based data for advocacy , funding , and support of preventive chemotherapy . The survey was conducted by FMoH , in collaboration with State Ministries of Health and non-governmental organizations using trained field workers . The survey used standard protocols put forth by the World Health Organization ( WHO ) for rapid mapping of STH infection in schools , collection of stool samples , and laboratory work-up . The diagnostic method used was the Kato-Katz technique , with duplicate Kato-Katz thick smears prepared from fresh stool samples; one per participant [20] . In each community/school , 60 school-aged children were examined . Data were collected at 555 locations across Nigeria , with the exception of areas in the north-eastern part of the country , where the state of security at the time of the survey did not allow doing so . Study locations were geo-referenced , using a hand-held global positioning system ( GPS ) device ( Garmin Etex; Garmin Corp , Kansas , United States of America ) . Quality checks were performed to authenticate that the coordinates indeed corresponded to specific locations using readily available Google Map and Google Earth tools . Environmental data were obtained from open-access remote sensing data sources , as detailed in Table 1 . These include normalized difference vegetation index ( NDVI ) as vegetation proxy , day and night land surface temperature ( LST ) , altitude , soil acidity , and soil moisture . Environmental data were processed as described elsewhere [21] . Annual averages for the year 2011 were calculated and used in all subsequent analyses . Maps showing the variation of the covariates across the country are shown in the Supporting Information ( S2 Fig ) . Data on rural and urban extents for Nigeria were downloaded from the Center for International Earth Science Information Network [22] . Population data for 2010 , at 100 x 100 m spatial resolution , was downloaded from the Afripop population database hosted by the World Population . The data were adjusted to 2011 by multiplying each pixel value with the Nigerian annual growth rate of 2 . 8% ( http://data . worldbank . org/indicator/SP . POP . GROW ) . The total population for 2011 was 154 , 731 , 365 of which 26 . 8% correspond to school-aged children ( http://www . census . gov/population/international/data/idb/region . php ) . We applied Bayesian binomial geostatistical models to relate STH infection risk with environmental and socioeconomic predictors . We used integrated nested Laplace approximations ( INLA ) [23] and a stochastic partial differential equations approach [24] for fast approximate Bayesian inference . Analysis was carried out in R [25] and the INLA package ( www . r-inla . org ) . Details of how models were implemented are provided in Supporting Information ( S1 Text ) [24 , 26 , 27] . We followed an approach detailed by Karagiannis-Voules et al . [28] , which has also been used for STH geostatistical modeling in Cambodia [29] , to select the best predictive model . In brief , we fitted Bayesian bivariate geostatistical models to select the functional form of the effect of each predictor based on the cross-validated logarithmic score [30 , 31] . We considered linear and categorical functional forms of effects . The categorical functional form of the covariates was generated using 25th , 50th , and 75th percentile to group each covariate into specific categories . Non-linearity was addressed through random walk processes of order 1 and 2 [32] . The form of each predictor giving the lowest mean logarithmic score was chosen . To identify the set of the most important predictors , we fitted geostatistical models with all possible combination of covariates ( i . e . , 256 models for each STH species-specific infection ) and selected the one , for each of the three STH species , with the best logarithmic score . The final models were used to predict infection risk at a grid of 3 x 3 km including areas where infection data were not available . The form of the covariate that was included in the final model used in the prediction of each species of STH is shown in Table 2 . The posterior estimates and Bayesian credible intervals for the effects of the predictors are presented in odds ratios . Additional details are provided in Supporting Information ( S1 Text ) . Due to the large number of observed zero prevalence data , we additionally fitted zero-inflated binomial models with invariant probability of zero-inflation . These models have shown better predictive ability in geostatistical modeling of malaria . [33] . In the present study , the zero-inflated models did not improve predictions ( based on the cross-validated logarithmic score ) . Hence , we report results from the binomial models . The school-aged population infected with STHs was estimated by combining the predictive posterior distribution of the infection prevalence at the pixel level with the school-aged population size at each pixel . The number of infected school-aged children was calculated by summing the respective values for each pixel , as described by Schur et al . [34] . The amount of anthelmintic treatment ( i . e . , albendazole or mebendazole ) that would be required to treat infected school-aged children at the unit of the state in Nigeria was computed from the pixel level risk estimates . Following recommendations by WHO , school-aged children should be treated twice a year in areas where the infection prevalence is ≥50% , while annual treatment is recommended in areas where infection prevalence ranges between 20% and 50% [9] . Hence , we computed the total number of anthelmintic drugs needed by multiplying the number of school-aged children , per pixel , by a factor of 2 ( prevalence ≥50% , biannual treatment ) or 1 ( prevalence 20–50% , annual treatment ) . We considered the estimated prevalence of STH at pixel-level , calculated under the assumption that the species-specific prevalences are independent . Treatments were aggregated over all pixels within individual states [19 , 33] . We compared treatment needs calculated from both pixel and population-adjusted district level prevalences . The estimation of the country-wide number of treatments was based on the sum of the treatment distributions of all local government areas ( LGAs ) and was conducted using both the approaches described above .
STH infections were diagnosed in the stool of school-aged children surveyed in 20 of the 36 states , including the Federal Capital Territory , Abuja . A . lumbricoides was present in 305 ( 55 . 0% ) locations in 16 states , and prevalence at the unit of the state varied from 1 . 6% to 77 . 8% ( Fig 1A ) . Hookworm infection showed the widest geographic distribution , as it was found in 482 ( 86 . 8% ) locations in all 20 states , with prevalence at the unit of the state ranging from 1 . 7% to 51 . 7% denoted with the varying colours in Fig 1B . T . trichiura was found in 55 ( 9 . 9% ) locations in 12 states with state-prevalence ranging from 1 . 0% to 25 . 5% ( Fig 1C ) . A . lumbricoides and hookworm HHhHhinfections were co-endemic in 16 states , while co-occurrence of all three STH species was observed in 12 states . Areas with high infection risk ( ≥50% ) of A . lumbricoides were predicted for the south-western part of Nigeria . For most areas in the northern and southern parts of Nigeria , the predicted prevalence was below 5% ( Fig 1D ) . Predicted pixel level prevalence revealed that high risks areas for A . lumbricoides infection occur within the states of Ogun , Ondo , Kwara , and Kogi , and some areas in Anambra and Taraba states . Our Bayesian geostatistical model for A . lumbricoides risk suggests that extreme high LST ( ≥34°C ) is negatively associated with A . lumbricoides , while a positive association was found between A . lumbricoides infection and high NDVI value ( Table 2 ) . It was observed that most south-eastern , western , and middle belt parts of Nigeria fell either in the high-risk ( ≥50% pixel level prevalence ) or moderate-risk areas ( 20–50% pixel level prevalence ) of hookworm infection ( Fig 1E ) . The predicted pixel level risk of hookworm in the northern states of Nigeria ranged from 5% to 10% ( Fig 1E ) . However , there are some high-risk ( ≥50% pixel level prevalence ) communities in the states of Katsina , Zamfara , and Sokoto in the north-western part of the country . Only few areas in Nigeria showed a pixel level predicted prevalence of hookworm below 5% . Areas with predicted hookworm pixel level prevalence greater than ≥50% , which are considered as high-risk areas , were observed in the states of Taraba , Benue , Oyo , Kwara , Katsina , Zamfara , Sokoto , and Kebbi ( Fig 1E ) . The risk of hookworm infection at pixel level in Jigawa , Ogun , Osun , and parts of Zamfara and Sokoto states were predicted to be below 5% . Our Bayesian-based geostatistical model for hookworm showed that high NDVI values and low day LST values are positively associated with hookworm infection ( Table 2 ) . The prevalence of hookworm infection was lower in urban compared to rural areas ( Table 2 ) . Infection risk of T . trichiura , ranging between 10% and 20% ( pixel level prevalence ) , was predicted for areas in the south-west ( Ondo state ) , while all other parts of the country showed pixel level prevalence risks below 10% . The predicted risk of T . trichiura was considerably higher in the southern part of Nigeria compared to the north ( Fig 1F ) . Pixel level prevalence revealed that areas within Ogun , Ondo , Anambra , and Enugu states and some areas of Taraba state are at high risk of T . trichiura . Our Bayesian geostatistical model for T . trichiura suggests a random walk process for night LST , indicating that extreme high temperatures ( ≥34°C ) are associated with the absence of T . trichiura in Nigeria . A negative association was found between altitude ( increase in altitude ) and risk of T . trichiura infection . Out of the 41 . 5 million school-aged children in Nigeria , an estimated 5 . 7million are predicted to be infected with any STH , an overall predicted prevalence of 13 . 8% . Kano state has the highest number of infected school-aged children , while the Federal Capital Territory , Abuja has the lowest prevalence ( Table 3 ) . Following WHO recommended cut-offs of 20% and 50% for annual and bi-annual preventive chemotherapy with either albendazole or mebendazole , the estimates aggregated at state level showed that only 3 out of the 37 states had a population-adjusted prevalence between 20% and 50% . These states are Cross River , Kwara , and Ondo ( Table 3 ) . The LGA is the third administrative level in Nigeria and the preferred unit for health intervention . According to the aforementioned prevalence cut-offs , we computed that the number of albendazole or mebendazole tablets needed for treatments using pixel-level prevalence is 10 , 222 , 409 , tablets , whereas using population-adjusted LGA-level prevalence , it is 9 , 025 , 229 tablets . These numbers correspond to the median of the country-wide distributions of treatment needs rather than the sum of the median LGA predicted requirements ( S1 Table ) .
We provide spatially explicit model-based risk estimates of the three main species of STHs in Nigeria . We used Bayesian geostatistical methods which have become essential tools in infectious disease risk profiling [35] . Our estimates are based on a large ensemble of recent survey data that were obtained using standard protocols . Hence , our estimates are more robust than those obtained from previous mapping exercises that collated historic survey data employing different collection methods and diagnostic approaches [28 , 36 , 37] . Our predictive risk maps are important and useful for planning , implementation , and evaluation of STH control programs [21] . Indeed , as a first step , the maps will help prioritize the implementation of intervention programs for the control of STH infections , particularly the spatial targeting of preventive chemotherapy . This is important in light of the current global moves toward control and elimination of NTDs [14 , 38] . Additionally , the model-based risk map of STH presented here complements a recent model-based risk map of schistosomiasis in Nigeria [19] for concurrent control of STH and schistosomiasis [39 , 40] . An integrated approach for the control of multiple helminthiases would reduce operational costs in the planning and implementation of control programs , as the primary target risk group for preventive chemotherapy are school-aged children , and hence the education system is the most convenient platform for drug administration [38 , 41 , 42] . It should be noted , however , that recent mathematical modeling work revealed that adults should also be targeted by preventive chemotherapy if substantial gains of morbidity control and interruption of transmission are aimed for [43] . A similar result was supported by a sub-continental geostatistical analysis of STH in sub-Saharan Africa [28] . Our predictions show that most areas in Nigeria are characterized by STH infection prevalence below 20% . This estimate is in line with the distribution pattern of STH infections in most endemic populations . Infections are usually aggregated where most infected individuals in a community will have infections of light or moderate intensity , while a few will be heavily infected [44] . The heavily infected individuals are at highest risk of clinical consequences of STH infection and serve as the reservoir host for the continuous transmission to the rest of the community [41] . Although WHO does not recommend large-scale preventive chemotherapy in areas where prevalence is below 20% [9] , detailed information of the number of infected individuals for lower risk areas is important from operational and programmatic points of view [45] . The overall relatively low prevalence of STH infection across Nigeria could be due to the periodic deworming of school-aged children by health officials and non-governmental health organizations working in the country . Currently ongoing in Nigeria are deworming programs targeting onchocerciasis and lymphatic filariasis , which include ivermectin treatment given to school-aged children 5 years and above for onchocerciasis and/or ivermectin plus albendazole against lymphatic filariasis . Another reason may be attributed to good access to cheap sachet drinking water popularly called “pure water” in many rural communities in Nigeria . This 500 ml nylon-bagged potable water is basically available everywhere in Nigeria and is sold at US$ 0 . 03 per sachet . The availability of this product may be a factor in reducing the fecal-oral transmission of A . lumbricoides and T . trichiura . On the other hand , the comparatively higher prevalence and distribution of hookworm infection in Nigeria is associated with the transmission of this parasite through the skin . Hence , barefoot walking by school-aged children is a risk factor and is likely to be driven by low socioeconomic status [44] . It should also be noted that Nigeria in the equatorial zone is suitable for hookworm larval development [46] . Our predictions revealed that less than 15% of school-aged children were infected with STHs in 2011 . Thus , the acceleration of STH control is important to maintain this relatively low level of prevalence in the most populous country in Africa [47] . Our data are useful in reviewing the current STH control program in Nigeria in light of the findings presented here . Based on our predictions , the estimated annualized needs for anthelmintic drugs have been determined to be 10 . 2 million tablets . This amount should be further reviewed when the security issue in north-eastern Nigeria is resolved and prevalence data for this region become available to update model-based estimates . The fact that infection prevalence of STHs are considerably lower when compared to past estimates and projections [11] points to progress made , thanks to efforts by various governmental and non-governmental health development agencies implementing deworming programs across the country . Hence , these efforts should be sustained with adequate funding [12] . We fitted Bayesian geostatistical models to identify environmental and socioeconomic predictors that influence the distribution of each of the three STH species . Our results show that NDVI is a major environmental predictor for hookworm infection , while day LST is negatively associated with the distribution of A . lumbricoides . The results of our predictions are supported by the biology , ecology , and epidemiology of STHs . In fact , low humidity , associated with high temperature , leads to cessation of embryonation of A . lumbricoides , while high humidity promotes quick embryonation of A . lumbricoides eggs [48 , 49] . Our results are in line with earlier reports on the influence of temperature in the distribution and transmission of STH infections in Bolivia and the People’s Republic of China [37 , 50] . The observed prevalence data show that hookworm infection is the most widespread of the three common STH infections and has a higher predicted prevalence than the other two species . This finding is setting-dependent since other studies carried out in Bolivia , the People’s Republic of China , and Kenya found that A . lumbricoides is the predominant STH species [37 , 50 , 51] . The only socioeconomic predictor used ( i . e . , urban-rural classification ) did not show any relationship with A . lumbricoides and T . trichiura infections . Other socioeconomic proxies , such as sanitation level and access to clean water , may be able to better explain the spatial distribution of infection risk with STHs [52] . However , unless individual information on both infection and , for instance , sanitation become available , such socioeconomic proxies might not improve predictions [29] . The predictors identified indicate that high night LST , which is often observed in the desert part of Nigeria as well as high altitude , can prohibit the survival of T . trichiura . This result may explain the quasi-absence of this STH species in the northern part of Nigeria ( where there is extreme heat and a short wet season ) . The strength of this study is that our analysis is based on recent survey data obtained by the FMoH Nigeria in 2011 , adhering to standard and uniform diagnostic methods , focussing on school-aged children across all surveyed locations . This helps to avoid prediction bias associated with heterogeneities ( e . g . , due to diagnostic error and different age groups across surveys ) arising from historically compiled data [52] . More importantly , an accurate and up-to-date map of STH infections is more reliable in making decisions for helminthiasis control , as relying on historic data alone may not give a true picture of the current status of the disease [53] . A limitation of this analysis is the lack of data from most of the north-eastern part of the country due to security issues and therefore prediction uncertainties are high in that part of the country . There are two important points to consider in the calculation of treatments as well as number of people infected that are offered for discussion . First , the geographic level ( such as district , state , and country ) of aggregating pixel-level predictions has an impact on the overall result , as discussed by Schur et al . [34] . The treatments over an administrative area can be calculated using either pixel-level prevalence estimates ( aggregated over the area ) or population-adjusted prevalence over the area . In the present study we compared treatment needs calculated from both approaches using LGA as the level of aggregation . The total number of treatments for the whole country differs by 1 , 197 , 180 tablets between the two approaches . The main reason of this difference is that population-adjustment over an area ( e . g . , LGA ) might be dominated by largely populated pixels that usually correspond to urban settlements . Second , the pixel resolution affects the number of estimated treatments . Low resolution would lead to few pixels covering large surfaces and crossing boundaries of administrative levels . Therefore , calculations do not take into account the variation in the population density and can wrongly assign treatments to areas . Our estimates are lower than those recently reported in a geostatistical analysis of STH infection across sub-Saharan Africa [28] . The previous analysis used historic data over the past 50 years stemming from 33 states in Nigeria , some of which had high prevalence before 2000 . It also considered a common temporal trend across all countries and showed a prevalence decrease after 2000 , probably due to socioeconomic development as well as preventive chemotherapy that have been scaled up recently . In Nigeria , according to the preventive chemotherapy database of WHO , from 2003 to 2010 , there have been more than 10 million school-aged children and almost 7 million preschool-aged children treated for STHs . From 2010 onwards , more than 22 million school-aged children have received preventive chemotherapy . In our study , using recent survey data , we predicted lower prevalence , indicating a continuous decline in the prevalence of STH . In conclusion , we have produced spatially explicit model-based risk estimates of the geographic distribution of the three main species of STHs in Nigeria and determined underlying environmental risk factors . This is useful for planning the control of STH . We have further estimated the number school-aged children infected and at risk of infection , and provided annualized deworming requirement for Nigeria . With these data , the national STH control program can mobilize resources and attract local , national , and international support to escalate the implementation of preventive chemotherapy and other control measures nationwide . | Infections with three kinds of parasitic worms—hookworm , roundworm , and whipworm—are collectively known as soil-transmitted helminths ( STHs ) . These parasitic worm infections are widespread in Nigeria , but the exact distribution is poorly understood . In view of the global commitment to control STH infections , there is a need to accelerate the mapping of STH infections to guide control interventions , such as large-scale administration of deworming drugs . In this study , we collated survey data from the year 2011 for Nigeria . The data were utilized to predict the distribution of STH infection based on environmental and socioeconomic covariates , and employing a Bayesian geostatistical modeling approach . Our results indicated that STH infections are widely distributed across Nigeria with prevalence estimates as high as 83% for roundworm , 50% for hookworm , and 19% for whipworm infections at specific survey locations . We predict that 5 . 7 million school-aged children were infected with STHs . The numbers of deworming tablets for annual or bi-annual treatment of the school-aged population at local government areas level in Nigeria for 2011 were estimated to be 10 . 2 million . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Bayesian Geostatistical Model-Based Estimates of Soil-Transmitted Helminth Infection in Nigeria, Including Annual Deworming Requirements |
Metrics of phylogenetic tree reliability , such as parametric bootstrap percentages or Bayesian posterior probabilities , represent internal measures of the topological reproducibility of a phylogenetic tree , while the recently introduced aLRT ( approximate likelihood ratio test ) assesses the likelihood that a branch exists on a maximum-likelihood tree . Although those values are often equated with phylogenetic tree accuracy , they do not necessarily estimate how well a reconstructed phylogeny represents cladistic relationships that actually exist in nature . The authors have therefore attempted to quantify how well bootstrap percentages , posterior probabilities , and aLRT measures reflect the probability that a deduced phylogenetic clade is present in a known phylogeny . The authors simulated the evolution of bacterial genes of varying lengths under biologically realistic conditions , and reconstructed those known phylogenies using both maximum likelihood and Bayesian methods . Then , they measured how frequently clades in the reconstructed trees exhibiting particular bootstrap percentages , aLRT values , or posterior probabilities were found in the true trees . The authors have observed that none of these values correlate with the probability that a given clade is present in the known phylogeny . The major conclusion is that none of the measures provide any information about the likelihood that an individual clade actually exists . It is also found that the mean of all clade support values on a tree closely reflects the average proportion of all clades that have been assigned correctly , and is thus a good representation of the overall accuracy of a phylogenetic tree .
Phylogenetic analysis , once the province of systematists and evolutionary biologists , has become a fundamental tool of computational biology and biological disciplines as diverse as biochemistry , epidemiology , and developmental biology . While systematists use phylogenetic analysis of molecular sequences to elucidate the historical relationships among species , other disciplines tend to focus more on the historical relationships of the sequences themselves . The results of phylogenetic analyses are typically presented as phylogenetic trees , diagrams that graphically illustrate those historical relationships . Phylogenetic trees are just estimates of those historical relationships , and it is therefore important to have some way to evaluate the quality and reliability of phylogenetic reconstructions . The most widely used method of estimating the reliability of trees is the nonparametric bootstrap [1] . The bootstrap method addresses the reliability of the tree topology ( the branching order ) by calculating the bootstrap percentage ( BP ) for each interior node , or clade , in a tree . In the bootstrap method , the sites in a set of aligned sequences are randomly sampled with replacement to create a pseudo-alignment , and that pseudo-alignment is used to produce an estimated “bootstrap tree . ” Typically , 100–2 , 000 bootstrap trees are estimated , and the BP for a clade on the original phylogenetic tree is the percentage of the bootstrap trees that also include that clade . Thus , confidence in the groupings of taxa can be estimated . A drawback to the bootstrap method is that it can potentially be very time-consuming . For example , maximum likelihood is at present the most widely used statistical phylogenetic method , but because it is computationally intensive , performing a bootstrap analysis on maximum likelihood trees can require prohibitive amounts of time . Recently , a new approach to estimating branch ( or clade ) support , the approximate likelihood ratio test ( aLRT ) , has been introduced [2] . The aLRT is a fast and accurate method for assessing branch support for maximum likelihood trees . Under conventional LRT , the null hypothesis is that the branch has a length of zero ( i . e . , it does not exist ) , and the test statistic is 2 ( l1 − l0 ) , where l1 is the likelihood of the most likely tree and l0 is the likelihood of the tree in which the branch does not exist . In aLRT , the test statistic is approximated by 2 ( l1 − l2 ) , where l2 is the likelihood of the second most likely tree , an approximation that enormously decreases computational time and results in a practical and slightly conservative test statistic . The significance of the aLRT test statistic is calculated from a mixed χ2 distribution , with half drawn from zero and half drawn from one degree of freedom . The aLRT approach is implemented in the beta version of PHYML 2 . 4 . 5 [3] ( http://atgc . lirmm . fr/alrt ) . The most recent release of the beta version of PHYML also implements an alternative nonparametric Shimodaira–Hasegawa-like ( SH-like ) procedure that is typically more conservative than the χ2 approach , so PHYML now offers the option of assigning support as the smaller of the values calculated by the two methods . In the last decade , a new method of estimating phylogenetic trees , the Bayesian method , has gained increasing popularity [4–7] . The Bayesian method , as implemented by the program MrBayes [8 , 9] , estimates the posterior probabilities ( PPs ) of clades by calculating , among the trees with the highest posterior probabilities , the fraction of the time that each clade appears as those trees are visited in proportion to their probabilities . The Bayesian method has the advantage that it calculates PPs during the process of estimating the consensus tree . It is therefore much faster to obtain PP estimates of clade reliability by the Bayesian method than to obtain BPs of clade reliability by maximum likelihood . BP , aLRT , and PP are measures of clade support , but they are often presented as measures of the accuracy of the tree [5 , 10] . None , however , is a metric of accuracy . aLRT assesses the likelihood that a branch exists on a maximum likelihood tree . BP and PP are simply measures of repeatability; BP measures the repeatability with which a clade occurs among subsamples of the data used to create the original tree , and PP measures the repeatability with which a clade occurs among the set of nearly equally likely trees after the Bayesian process has converged on a set of trees with nearly identical likelihoods . Because of discrepancies between Bayesian posterior probabilities and bootstrapped maximum likelihood percentages , there has been considerable controversy about BP versus PP as measures of clade reliability [11–15] . For real , empirical data , we cannot know the accuracy of a tree because we have no way of knowing the true branching order of the taxa or sequences that are being considered . Simulated datasets , in which the true tree is known , have been used to compare BP and PP with the accuracies of estimated trees . Several such studies have shown that BP underestimates clade reliability ( i . e . , clades in the estimated tree are more likely to exist in the true tree than is indicated by BP ) [10 , 11 , 13–16] . There have been conflicting reports about the relationship between PP and accuracy . In general , PP has been found to be less conservative than BP . Some studies [11 , 12] have concluded that PP is too liberal ( i . e . , overestimates accuracy ) , while others [13 , 14] conclude that PP better reflects accuracy . Another study concluded that BP and PP can be taken as potential upper and lower estimates of accuracy , but that they are not interchangeable and cannot be directly compared [15] . Similarly , Anisimova and Gascuel reported that aLRT using the χ2 approach is similar to posterior probabilities , and their unpublished data suggest that the SH-like approach is more conservative than the χ2 approach ( http://atgc . lirmm . fr/alrt ) . The conclusions of the above studies are only as reliable as the extent to which the simulations mimic real evolutionary processes that generate the empirical data to which we actually apply phylogenetic methods . In all cases , the simulations incorporate specific evolutionary models , the most common being the K2P model [17] , to guide the simulation process . The results will be no more realistic than the assumptions and biases of that model . Modeling evolution as a process of substitution confounds two distinct processes , mutation and selection , the outcome of which is the real substitution process [18] . The number of taxa used in the simulations reported in [10–16] ranges from four to 28 . Many , and probably most , phylogenetic studies involve many more taxa . Typically , branch lengths are uniform , although some studies included a specific pattern of length variation . Importantly , the simulations only consider base substitutions , not insertions or deletions ( indels ) . The resulting sequences thus need not be aligned . In reality , historical indels necessitate using multiple alignment programs to estimate the homologous characters within . The alignment process strongly affects the reliability of the resulting trees . For coding sequences , the accuracy of a tree is significantly increased by aligning the corresponding protein sequences and using that alignment to place the corresponding gaps into the DNA coding sequences [18] . However , when the average percentage identity of the amino acids is within the “twilight zone” of 20%–30% , only 80% of residues are correctly aligned [19] , and when identity is below 10% , less than 50% of residues are correctly aligned [20] . The failure to include indels in the simulation process therefore reduces considerably the confidence we can place in applying conclusions drawn from those simulations to real data . The EvolveAGene simulation program [18 , 21] was designed to mimic the evolution of sequences in a more biologically realistic fashion . A real sequence is used for the root node of the tree , and a strictly bifurcating bilaterally symmetrical tree is evolved . Branch lengths are randomly varied from zero to a value chosen by the user . Mutation and selection are treated as separate processes . The mutation process is simulated by introducing random mutations , including base substitutions , insertions , and deletions , into the sequences according to the spontaneous mutation spectrum of Escherichia coli . ( The mutational spectrum is the experimentally determined relative frequencies with which the various base substitutions and indels of different lengths occur , before selection or drift act on those mutations . ) The selection process is simulated by ( 1 ) assuming that all frameshift and nonsense mutations are strongly deleterious and thus not accepting those mutations , and ( 2 ) accepting nonsynonymous base substitutions with a probability that corresponds to a user-specified nonsynonymous substitution per nonsynonymous site to synonymous substitution per synonymous site ( dN/dS ) ratio , which can be set to biologically realistic values . The EvolveAGene program has been used to compare accuracies of various phylogenetic methods [18] and to explore the accuracies with which parsimony and Bayesian methods can reconstruct ancestral protein sequences [21] . In this study we are not particularly interested in comparing PP , BP , and aLRT per se . Instead , we are interested in asking two questions: ( 1 ) for all measures of clade credibility , how well does the credibility of a clade reflect the probability that that clade really exists on the true tree; and ( 2 ) how well does the average clade support reflect the topological accuracy of the tree ? Topological accuracy is defined as the fraction of clades on the estimated tree that actually exist on the true tree . In this study , both the true tree and the estimated trees are strictly bifurcating , so the number of interior clades is the same . Thus , the number of false positive errors ( clades found on the estimated tree that do not exist on the true tree ) is identical to the number of false negative errors ( clades on the true tree that are not on the estimated tree ) . We simulate the evolution of several genes under biologically realistic conditions and find that none of the estimates of clade support correlates with topological accuracy; in other words , clade supports tell us nothing about the likelihood that an inferred clade actually exists . However , we find that the average clade support does correlate well with the topological accuracy of the tree .
Ten simulations were initiated from each of five E . coli K12 coding sequences to assess how well BPs from maximum likelihood trees , aLRT support by the χ2 approach , aLRT support by the minimum of SH-like and χ2 approaches , and posterior probabilities from Bayesian trees corresponded to clade accuracies . The simulation conditions were chosen to generate datasets that were at the practical limits for reliable alignments . Indeed , the typical average Jukes–Cantor [22] distances among the sequences for those datasets was 1 . 39 ± . 02 substitutions per site , well above the limit of 1 . 0 , above which Nei and Kumar [23] state that neighbor-joining trees are unreliable . We define “true clades” as clades in the estimated tree that exist in the true tree , and we define “accuracy” as the percent of the total clades in the estimated tree that are true clades . Tables 1–4 show , respectively , the results for Bayesian trees and for maximum likelihood trees by the bootstrap and the two aLRT approaches . In each case , rows are ranges of clade credibility values . For all methods , accuracy increases as the lengths of the sequences increase . Mean BPs are conservative estimates of mean topological accuracy , and in keeping with [13] and [14] we find that average posterior probabilities are a better estimate of topological accuracy than are BPs . BPs underestimate accuracy , particularly for trees based on the shorter sequences , more than do posterior probabilities or aLRT supports . We do not interpret this finding to mean that BPs should be the “gold standard” measure of reliability; indeed , we find the notion that the less-accurate estimate should be the gold standard to be slightly ludicrous . Our results , however , differ strikingly from those of [13] and [14] with respect to the correspondence between individual clade confidences and the accuracy of those clades . Alfaro et al . [13] found that for most topologies accuracy was higher than either BP or PP when clade confidences were greater than ∼40% , but lower than clade confidences when those measures were less than ∼40% . Hillis and Bull [10] obtained similar results for BP , and Wilcox et al . [14] obtained similar results for PP , but a low crossover point at about 20% for BP . In contrast , we find no significant correlation between individual clade supports and the probability that a clade is correct , whichever method of clade support is used . We regressed the fraction of clades that actually exist within each decile against the midpoint for each decile of clade support; thus , a slope approaching one would be expected for a perfect correlation between those values , whereas a slope of zero would indicate no correlation . Only two out of the 28 plots have slopes that are significantly different from zero ( genes nuoK and rplF for bootstrap support of maximum likelihood trees , with p = 0 . 03 and 0 . 02 , respectively ) . One of these slopes is slightly positive , and the other is slightly negative; thus , both likely represent outliers . The absence of correlation between clade support and the likelihood that a clade exists means that , whatever the method , clade support values provide no information about , and have no predictive power as to , the likelihood that the clade exists . We attribute the differences between our results and those of others [10 , 13 , 14] to our use of a more biologically realistic simulation . It is conceivable that this startling finding is the result of reconstructing relatively large ( 64 taxon ) trees under some false assumption that is unknowingly incorporated into the simulation: perhaps with so many taxa , resulting in an enormous number of possible trees , any clade that has even mild support is likely to be a true clade . We think this possibility is unlikely , because the fraction of false clades that have 81%–100% support is roughly the same as the fraction of false clades that have low ( 0%–40% ) support . Nevertheless , to test the idea that our findings are an artifact of considering large trees , we tested all four methods with 16-taxon datasets of nuoK , replicated ten times each . ( The nuoK gene was chosen because it is the shortest gene and the gene for which all methods were the least accurate . ) As before , the taxa were a random sample from a 128-taxon dataset . When the number of taxa was reduced from 64 to 16 , the quality of the alignments was reduced so that the amino acid identity was <20% , below the zone in which alignments are reliable . The average branch length was therefore reduced from 0 . 18 to 0 . 15 substitutions per site to produce alignments that exhibited an average of 24% amino acid identity , well within the “twilight zone” [19] . Table 5 shows that the results are essentially the same as in Tables 1–4: there is no significant correlation between clade support and the fraction of true clades . Both mean accuracy and mean clade support are generally lower than in Tables 1–4 , but it remains the case that clade support provides no information about the likelihood that a clade actually exists . Random sampling of taxa typically results in well-balanced trees ( Figure 1 ) , and it is conceivable that our findings apply only to trees with similar topology . To test that possibility , we nonrandomly sampled ten nuoK datasets to generate highly pectinate , unbalanced 16-taxon trees ( Figure 2 ) . For the unbalanced trees , the topological accuracies were higher than for the 16-taxon balanced trees , and all methods of clade support again underestimated that accuracy . Again , there was no significant correlation between clade support and the likelihood that a clade existed on the true tree ( Table 6 ) . We conclude that our results are general , and not simply attributable to large trees or to balanced trees . This is one of those good news–bad news stories beloved by comedians . The bad news is that none of the methods of assessing clade support provides any reliable estimation that the clade has been correctly assigned . The good news is that , even with data that are near the practical limits for phylogenetic tree reconstruction , both maximum likelihood and the Bayesian method estimate topologies so well that even when clade support is very low there is a better than 80% chance that the clade is correctly assigned . In addition , for each method , averaged over the 70 datasets , there is a significant correlation between the average branch support and the accuracy of the tree ( Table 7 ) , and for all methods except nonparametric bootstrap , the average clade support value is a good , if slightly conservative , estimator of the overall fraction of clades that actually exist . It might be argued that having a good estimate of overall accuracy is not very useful , and that we are generally interested in identifying unreliable branches . Our results show that we simply cannot identify what particular branches are unreliable based on measures of clade support . Thus , with current methods of determining clade credibility we cannot have what we might generally want , and it is important to acknowledge the limitations of those metrics . On the other hand , methods of determining clade confidence do provide a good estimation of the overall reliability of a phylogenetic tree , and permit us to infer how many untrustworthy branches may be present . Just as we can make good predictions about the diffusion of a mass of molecules over time , but not about the motions of individual molecules in that mass , we can make good estimations of the overall topological accuracy of a tree , but not about the accuracy of individual branches .
Simulations were performed by EvolveAGene [18 , 21] . Five coding sequences from E . coli K12 were selected from the E . coli genome entirely on the basis of length , and used to initiate the simulations: nuoK , 300 bp , encodes the NADH dehydrogenase subunit K; rplF , 530 bp , encodes 50S ribosomal protein L6; tauB , 763 bp , encodes a taurine transport ATP-binding protein; add , 999 bp , encodes adenosine deaminase; and araB , 1 , 698 bp , encodes ribulokinase . The genes are not functionally related to each other , and none exhibits detectable homology to another by pairwise BLAST comparisons . For the simulations in Tables 1–4 , the average branch length was 0 . 18 substitutions per site , with lengths ranging from 0 to 0 . 36 substitutions per site; for Tables 5 and 6 the average branch length was 0 . 15 substitutions per site , ranging from 0 to 0 . 30 substitutions per site . The tree was evolved for seven “generations” to give 128 terminal taxa; thus , the average length from the root to the tip was 1 . 26 substitutions per site . The probability of accepting an indel was 0 . 02 , and the probability of accepting a nonsynonymous base substitution was 0 . 2 . Ten independent simulations were carried out from each of the five root sequences . When all of the terminal sequences descended from the root node were included in the dataset , both Bayesian and maximum likelihood trees included very few nodes with clade confidences <80% ( unpublished data ) . When trees were based on a random subsample of the sequences , both methods produced trees with more low-confidence clades . Datasets used to estimate trees were therefore based on a random sample of 16 or 64 of the 128 evolved sequences . Sequences were aligned by ClustalW [24] as implemented by MEGA 3 . 1 [25] . Sequences were translated to their corresponding protein sequences by MEGA 3 . 1 , aligned with a gap-opening penalty of 3 . 0 and a gap-extension penalty of 1 . 8 . The average pairwise amino acid identities in the resulting alignment were typically 21%–22% , near the lower boundary of the “twilight zone” below which alignments are not sufficiently reliable to produce valid phylogenetic trees [19 , 20] . Triplet gaps corresponding to the gaps in the protein alignment were introduced back into the DNA sequences by MEGA 3 . 1 . The resulting DNA sequence alignments were saved in the FASTA format and converted to the PHYLIP format ( for input to PHYML ) and to the Nexus format ( for input to MrBayes ) by a Perl script . Trees were estimated by two methods: maximum likelihood as implemented by PHYML 2 . 4 . 4 [3] , and the Bayesian method as implemented by MrBayes 3 . 1 . 2 [9] . In both cases , trees were estimated using the GTR + invariants + gamma model . For maximum likelihood trees , clade confidences were estimated from 100 bootstrap replicates . Bayesian trees were estimated from 600 , 000 generations , sampling every 100 generations , with a heating parameter of 0 . 15 , in two parallel runs . The consensus trees were calculated using the allcompat option ( strict consensus ) from the final 4 , 501 trees of each run . Convergence , as judged by the diagnostic average standard deviation of the split frequencies between two parallel runs falling below 0 . 02 , typically occurred before generation 120 , 000 ( 1 , 200 trees ) . A typical true tree , in this case initiated with the nuoK sequence , is shown in Figure 1A . Note that the true tree includes one near trichotomy , and that the distance from the root to the tips ranges from 0 . 52 substitutions per site for taxon ZZZZZZZ to 1 . 55 substitutions per site for taxon PPPPPPPPPP . The corresponding Bayesian estimated tree is shown in Figure 2 . Posterior probabilities <90% are indicated . This Bayesian tree is typical in that only 12 of the 61 interior nodes have posterior probabilities <90% , but only four of those low-PP clades ( indicated by arrows ) are not present in the true tree . A Perl script , InferAcc , was used to compare the estimated trees with the true trees . Clades were sorted into bins as indicated in Tables 1–6 , and the clade was scored as existing if it was present in the true tree . Mean accuracy is the fraction of clades in the estimated tree that exist in the true tree averaged over the ten trees in the set . | The construction of phylogenetic trees , which depict past relationships between groups of DNA or protein sequences , has valuable application in many fields of study , most commonly evolutionary and population biology . Before drawing conclusions from phylogenetic trees , it is important to assess how accurate those reconstructions are . This is typically accomplished by examining measures of “clade credibility” ( such as bootstrap or posterior probability values ) , which represent how reproducible relationships are within the tree based on the parameters of the phylogenetic analysis . However , such measures do not necessarily reflect how likely inferred relationships are to have actually occurred in nature . Therefore , using simulated data where relationships are known , we have determined how well several measures of clade credibility correlate with the likelihood that a deduced phylogenetic grouping actually exists in reality . Surprisingly , we found no such correlation , and that the inferred relationships were correctly assigned about as often in cases where clade credibility values were very low as where they were high . This finding suggests that current measures of phylogenetic tree reliability are not useful in predicting whether specific inferred relationships have actually occurred . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"evolutionary",
"biology",
"none",
"computational",
"biology"
] | 2007 | Measures of Clade Confidence Do Not Correlate with Accuracy of Phylogenetic Trees |
The hepatic circadian clock plays a key role in the daily regulation of glucose metabolism , but the precise molecular mechanisms that coordinate these two biological processes are not fully understood . In this study , we identify a novel connection between the regulation of RORγ by the clock machinery and the diurnal regulation of glucose metabolic networks . We demonstrate that particularly at daytime , mice deficient in RORγ exhibit improved insulin sensitivity and glucose tolerance due to reduced hepatic gluconeogenesis . This is associated with a reduced peak expression of several glucose metabolic genes critical in the control of gluconeogenesis and glycolysis . Genome-wide cistromic profiling , promoter and mutation analysis support the concept that RORγ regulates the transcription of several glucose metabolic genes directly by binding ROREs in their promoter regulatory region . Similar observations were made in liver-specific RORγ-deficient mice suggesting that the changes in glucose homeostasis were directly related to the loss of hepatic RORγ expression . Altogether , our study shows that RORγ regulates several glucose metabolic genes downstream of the hepatic clock and identifies a novel metabolic function for RORγ in the diurnal regulation of hepatic gluconeogenesis and insulin sensitivity . The inhibition of the activation of several metabolic gene promoters by an RORγ antagonist suggests that antagonists may provide a novel strategy in the management of metabolic diseases , including type 2 diabetes .
RORγ constitutes with RORα and RORβ , the retinoic acid-related orphan receptor ( ROR; NR1F1–3 ) subfamily of the nuclear receptors , which regulate transcription by binding as monomers to ROR-responsive elements ( ROREs ) in the regulatory region of target genes [1] , [2] . Through alternative promoter usage , the RORγ gene generates 2 isoforms , RORγ1 and RORγ2 ( RORγt ) , that regulate different physiological functions . RORγt is restricted to several distinct immune cells and is essential for thymopoiesis , lymph node development , and Th17 cell differentiation [1] , [3]–[5] . RORγ antagonists inhibit Th17 cell differentiation and may provide a novel therapeutic strategy in the management of several autoimmune diseases [4] , [6] . In contrast to RORγt , relatively little is known about the physiological functions of RORγ1 . The expression of RORγ1 is highly restricted to tissues that have major functions in metabolism and energy homeostasis , including liver and adipose tissue , and in contrast to RORα and RORβ , RORγ is not expressed in the central nervous system , including the hypothalamus and suprachiasmatic nucleus [1] , [6]–[13] . In several peripheral tissues RORγ1 exhibits a robust rhythmic pattern of expression with a peak at zeitgeber time ( ZT ) 16–20 that is directly regulated by the clock proteins , brain and muscle ARNT-like ( Bmal1 ) and circadian locomotor output cycles kaput ( Clock ) , and the Rev-Erb nuclear receptors [1] , [8]–[12] , [14] , [15] . Although RORγ is recruited to ROREs in the regulatory regions of several clock genes , including Bmal1 , Clock , Rev-Erbα , and cryptochrome 1 ( Cry1 ) ; the loss of RORγ has little influence on the expression of Bmal1 and Clock , and only modestly reduces the expression of Rev-Erbα and Cry1 [10] , [12]; The robust oscillatory regulation of RORγ1 expression by the clock machinery raised the possibility that RORγ might regulate the expression of certain target genes in a ZT-dependent manner . Because the clock machinery plays a critical role in the circadian regulation of many metabolic pathways , including glucose metabolism [13] , [16]–[19] , RORγ may function as an intermediary between the clock machinery and the regulation of metabolic genes . Since recent studies indicated an association between the level of RORγ expression and obesity-associated insulin resistance in mice and humans [20] , [21] , these observations led us to propose that RORγ1 might be an important participant in the diurnal regulation of glucose metabolic pathways [10] , [16] , [18] , [22] . To study this hypothesis further , we examined the effect of the loss of RORγ on the diurnal regulation of glucose metabolism in ubiquitous and the hepatocyte-specific RORγ knockout mice . This analysis showed that loss of RORγ enhances glucose tolerance and insulin sensitivity particularly during early daytime ( ZT4–6 ) and reduces the peak expression of several glucose metabolic genes . RORγ cistrome and promoter analysis indicated that several of these metabolic genes were regulated directly by RORγ and involved ZT-dependent recruitment of RORγ to ROREs in their regulatory region . Together , our observations are consistent with the concept that RORγ directly regulates the diurnal expression of a number of glucose metabolic genes in the liver downstream of the hepatic clock machinery , thereby enhancing gluconeogenesis and decreasing insulin sensitivity and glucose tolerance . The inhibition of the activation of several glucose metabolic gene promoters by an RORγ antagonist suggests that such antagonists might provide a novel therapeutic strategy in the management of insulin resistance and type 2 diabetes .
Glucose tolerance and insulin sensitivity , as RORγ1 expression , have been reported to be under endogenous circadian control [23] , [24] . Recently , we proposed that RORγ1 might be an important participant in the diurnal regulation of several glucose metabolic pathways downstream of the circadian clock [10] , [22] . To study the potential role of RORγ in glucose homeostasis , we examined the effect of the loss of RORγ on insulin sensitivity , glucose tolerance and the rhythmic expression pattern of glucose metabolic genes in ubiquitous and hepatocyte-specific RORγ knockout mice . Our data revealed that the loss of RORγ expression had a significant effect on insulin tolerance ( ITT ) and glucose tolerance ( GTT ) in mice fed with a high-fat diet ( HFD ) . Comparison of the insulin responsiveness at two different time periods , ZT4–6 ( daytime ) and ZT18–20 ( nighttime ) showed that in wild type mice fed a HFD ( WT ( HFD ) ) insulin was more effective in controlling glucose levels at ZT18–20 than at ZT4–6 indicating that insulin sensitivity was ZT dependent [23] , [24] ( Figure 1A ) . Interestingly , this ZT-dependent difference in insulin responsiveness was greatly diminished in RORγ−/− ( HFD ) mice . ITT analysis showed that at ZT4–6 blood glucose levels remained significantly lower in RORγ−/− ( HFD ) mice after insulin injection than in WT ( HFD ) mice particularly after reaching a trough at 60 min ( Figure 1A and Table S1 ) . ITT performed at CT4–6 under constant darkness similarly showed improved insulin sensitivity in RORγ−/− ( HFD ) mice ( Figure S1A ) , suggesting that RORγ significantly affects insulin sensitivity also under a Zeitgeber-free condition . At ZT18–20 the difference in ITT response between WT ( HFD ) and RORγ−/− ( HFD ) mice was significantly smaller than at ZT4–6 . Consistent with the improved insulin sensitivity , GTT analysis showed that RORγ−/− ( HFD ) mice were more glucose tolerant than WT ( HFD ) particularly at ZT4–6 ( Figure 1C ) . Although the difference was smaller than in mice fed with a HFD , RORγ−/− ( ND ) mice fed with a normal diet ( ND ) were also significantly more insulin sensitive and glucose tolerant at ZT4–6 than WT ( ND ) mice ( Figure S1C and S1D ) . Because of the larger difference in mice fed a HFD , we focused much of our further analysis particularly on these mice . Altogether our observations indicate that the loss of RORγ enhanced glucose tolerance and insulin sensitivity particularly at ZT4–6 and CT4–6 . Analysis of the areas under the curves ( AUC ) for ITT and GTT was consistent with this conclusion ( Figure 1B and 1D ) . To obtain further insights into the improved insulin sensitivity in RORγ−/− mice , we compared the level of insulin-induced activation of Akt phosphorylation ( P-Akt ) , one of the most sensitive phosphorylation targets in the insulin signaling pathway , in liver and several other metabolic tissues ( Figure 1E ) . No significant difference in P-Akt was observed at ZT4–6 in liver , brown and white adipose tissue ( BAT , WAT ) , skeletal muscle between WT ( HFD ) and RORγ−/− ( HFD ) mice after insulin stimulation . Moreover , no significant difference in P-Akt was observed between insulin-treated WT and RORγ−/− primary hepatocytes ( Figure 1F ) . These results suggest that loss of RORγ does not alter insulin-dependent phosphorylation of Akt in several metabolic tissues . Next , we examined insulin sensitivity and glucose fluxes at daytime by the hyperinsulinemic-euglycemic clamp test . Consistent with the results of ITT , the glucose infusion rate ( GIR ) required to maintain blood glucose level under constant insulin infusion was significantly higher in RORγ−/− ( HFD ) mice than in WT ( HFD ) mice at daytime ( ZT2–9 ) , while their glucose absorption rate estimated by whole-body glucose disappearance ( Rd ) was almost equal during the clamp ( Figure 2A , S2A , S2B ) . Importantly , basal hepatic glucose production ( HGP ) and clamp HGP were significantly lowered in RORγ−/− mice . Insulin equally suppressed the HGP about 70% in both WT and RORγ−/− mice ( Figure 2B ) , indicating that the insulin responsiveness was not changed in RORγ−/− mice , consistent with the observation in Figures 1E and 1F . Glucose turnover estimated from the steady-state infusion of 3H-glucose ( Basal HGP and Rd ) [25] was lower in RORγ−/− mice , indicating that the glucose absorption rate might also be reduced . These results suggest that the increased GIR required to maintain blood glucose level in RORγ−/− mice was due to reduced hepatic glucose production and not due to improved insulin responsiveness . The clamp test suggested that the output of hepatic glucose produced by gluconeogenesis and glycogenolysis was reduced in RORγ−/− mice . Because hepatic gluconeogenesis is under close control of the circadian clock [18] , [23] , [26] , we analyzed gluconeogenesis efficiency at 2 different ZTs in WT and RORγ−/− mice fed with either a ND or HFD . The pyruvate tolerance test ( PTT ) indicated that gluconeogenesis was significantly higher at ZT4–6 than at ZT18–20 in both WT mice RORγ−/− mice with fed either a HFD or ND ( Figure S1E ) . However , gluconeogenesis was greatly reduced at ZT4–6 in RORγ−/− mice compared to WT mice independent of whether the mice were fed a ND or HFD , while little difference in pyruvate tolerance was observed at ZT18–20 between the two genotypes ( Figure 2C , S1E ) . Analysis of the AUC for PTT supported this conclusion ( Figure 2D , S1E ) . RORγ−/− ( HFD ) mice also showed a reduced gluconeogenesis at CT4–6 , a subjective daytime , under constant darkness ( Figure S1B ) . Together , these observations indicate that loss of RORγ affects pyruvate tolerance particularly at ZT4–6 and support a regulatory role for RORγ in the circadian control of hepatic gluconeogenesis . To obtain additional evidence that RORγ enhances hepatic gluconeogenesis , we analyzed PTT in RORγ−/− mice in which RORγ was over-expressed in liver by adenovirus administration . As shown in Figure 2E , gluconeogenesis was significantly increased in mice injected with RORγ-expressing adenovirus compared to mice injected with empty adenovirus . Further support for a role of RORγ in gluconeogenesis was provided by data showing that over-expression of RORγ in RORγ−/− primary hepatocytes increased glucose production ( Figure S2C ) . Together these results suggested that RORγ modulates insulin resistance and glucose tolerance by regulating hepatic gluconeogenesis . Food intake during daytime and nighttime was not significantly changed in RORγ−/− ( HFD ) mice ( Figure 3A ) and although glucose levels tended to be somewhat lower during daytime , a period in which gluconeogenesis was reduced , serum glucose levels were largely maintained in RORγ−/− ( HFD ) mice ( Figure 3B ) . Serum insulin levels in WT mice exhibited a circadian pattern reaching peak levels at ZT16 , while insulin levels were significantly lower in both RORγ−/− ( HFD ) and RORγ−/− ( ND ) mice particularly during ZT12–20 ( Figure 3B , S3A ) . Glucose-stimulated insulin secretion ( GSIS ) experiments indicated no difference in insulin secretion between WT and RORγ−/− mice fed with either a ND or HFD ( Figure 3C ) . In addition , little difference was observed in the level of pancreatic insulin at ZT16 , the time at which the difference in serum insulin levels was the greatest ( Figure 3D ) . These results suggested that lower serum insulin levels in RORγ−/− mice were not due to impaired insulin secretion or reduced pancreatic β-cell mass . Moreover , the amount of insulin secretion in response to the same quantity of glucose injected was not changed , suggesting that the reduced insulin level in RORγ−/− mice is likely due to reduced glucose production . Glyconeogenesis and glycogenolysis play an important part in glucose homeostasis; 10–20% of hepatic glucose production in mice fasting for 4 h depends on glycogenolysis [27] . Hepatic glycogen reached its highest level at ZT0 and its lowest between ZT8–12 in both WT ( HFD ) and RORγ−/− ( HFD ) mice; however , its peak level was significantly lower in RORγ−/− ( HFD ) mice ( Figure 3E ) . After 16 h fasting , the level of hepatic glycogen was dramatically reduced in both WT ( HFD ) and RORγ−/− ( HFD ) mice , but levels remained significantly lower in RORγ−/− ( HFD ) mice ( Figure 3F ) . The level of hepatic glycogen was also reduced in RORγ−/− mice fed with a ND ( Figure S3B ) . Glycogen accumulation was increased in RORγ−/− ( HFD ) mice injected with RORγ-expressing adenovirus ( Figure 3G ) , indicating that RORγ positively contributes to hepatic glycogen accumulation . Altogether , these results indicate that RORγ−/− mice are able to maintain blood glucose levels at lower insulin levels due to reduced hepatic glucose production and possibly reduced glucose uptake by the liver . The latter is consistent with the reduced glycogen accumulation and clamp test data showing that basal HGP/Rd was reduced in RORγ−/− mice ( Figure 2A ) . We next examined the behavior activity and energy homeostasis in WT ( ND ) and RORγ−/− ( ND ) mice in relationship to the effect of RORγ on circadian rhythm and hepatic glucose metabolism . No significant difference in total body weight was observed between WT and RORγ−/− mice fed a ND ( Figure S3C ) . The wheel running test showed that the circadian phase of behavioral activity was not changed in RORγ−/− ( ND ) mice consistent with a previous report [12] , but peak activity was lower than in WT mice ( Figure S3D ) . Indirect calorimetry showed that oxygen consumption ( VO2 ) , CO2 production ( VCO2 ) , respiratory exchange ratio ( RER ) , and heat production were significantly lower in RORγ−/− ( ND ) mice compared to WT ( ND ) mice particularly at nighttime ( Figure 3H and Figure S3E ) . Lower RER particularly at nighttime might indicate a preference for fatty acid consumption over glucose for energy production . Plotting of these parameters as a ratio between RORγ−/− ( ND ) and WT ( ND ) mice showed that the largest difference between WT and RORγ−/− mice occurred around ZT20 ( Figure 3I ) , which corresponds closely to the peak expression of RORγ [10] . These results indicate that the change in glucose metabolism in RORγ−/− mice is associated with reduced energy expenditure . To obtain further insights into the mechanism underlying the regulation of hepatic glucose metabolism by RORγ , we performed ChIP-Seq analysis to determine the genome-wide map of cis-acting targets ( cistrome ) of RORγ in murine liver at ZT22 , a few hours after the peak expression of RORγ ( Figure S4A ) [10] . This analysis identified 3 , 061 RORγ binding sites ( P<0 . 001 ) that were localized within intergenic regions ( 40 . 5% ) , introns ( 34 . 5% ) , within a 5 kb region upstream of the transcription start site ( TSS ) ( 11 . 5% ) , and the 5′UTR ( 10 . 8% ) ( Figure 4A , 4B ) . Notably , RORγ-binding sites were enriched near the transcription start sites ( Figure 4C ) . De novo motif analysis using MEME program identified a classic RORE motif , AGGTCA preceded by an AT-rich region ( Figure 4D and 4E ) as well as direct repeat 1 ( DR1 ) -like nuclear receptor binding motif and a RORE variant motif . Interestingly , a similar DR1 and variant RORE motifs were recently found within the binding sites of Rev-Erbs [14] , [28] . Gene ontology analysis of 1 , 443 RORγ candidate target genes , defined as those that have one or more detected RORγ binding site within 5 Kb upstream of the TSS and/or within the gene body , indicated that the RORγ cistrome was enriched for genes involved in fatty acid , amino acid , and carbohydrate metabolism ( Table 1 and Table S2 ) . Comparison of the ChIP-Seq data with those obtained from our previous microarray analysis [29] indicated that about 23% of the RORγ candidate target genes were differentially expressed between WT and RORγ−/− liver . CircaDB ( http://bioinf . itmat . upenn . edu/circa/ ) database analysis indicated that about 25% of the RORγ target genes exhibited a rhythmic expression pattern . Because RORα and RORγ bind similar DNA response elements , we examined the degree of functional redundancy between RORγ and RORα in regulating hepatic gene expression by comparing the RORα and RORγ binding sites identified by ChIP-Seq analyses . The specificity of each anti-ROR antibody was confirmed by WB and ChIP assays using chromatin of ROR-deficient mice as a negative control ( Figure S4B and S4C ) . ChIP-Seq analysis identified 1 , 319 RORα binding sites ( P<0 . 001 ) and 957 candidate target genes ( Figure 4F ) . Comparison of the RORα and RORγ cistromes revealed that 288 sites , including the ROREs within several clock genes reported previously [10] , recruited both RORα and RORγ ( Figure 4G and Table S3 ) . Thus , the relatively small overlap indicates that in liver RORα and RORγ exhibit a limited functional redundancy . Our ChIP-Seq analysis indicated that RORγ is recruited to regulatory regions of several genes implicated in hepatic glucose metabolism , including G6pase , Pepck , Glut2 , Pklr , Gck , Gckr , Gys2 , Pparδ , Pcx and Klf15 ( Figure 4G , S5 ) . Loss of RORγ resulted in a ZT-dependent decrease in the hepatic expression of most of these genes ( Figure 5A–5D ) and are consistent with our ChIP-Seq data indicating that their transcription is directly regulated by RORγ . The expression of G6pase was repressed in RORγ−/− liver during most of the circadian cycle , while Pepck expression was reduced during ZT4–12; both genes play a key role in gluconeogenesis ( Figure 5A ) . Peak expression of Gys2 , encoding a rate-limiting enzyme for glycogenesis , and Pparδ , which regulates several genes involved in glucose and lipid metabolism [30] , was decreased between ZT4–16 and ZT16-4 , respectively . The expression of several other gluconeogenic genes , including Pcx and Klf15 , the glucose transporter Glut2 , and several genes important in the glycolysis pathway , including Plkr , Gck , and Gckr , was also diminished in RORγ−/− liver ( Figure 5A–5D ) . Decreased expression of these genes was also observed in liver of RORγ−/− mice fed with a HFD ( Figure 5C ) . Importantly , the loss of RORγ had very little effect on the expression of Bmal1 and Clock , and a limited influence on the expression of Cry1 and Rev-Erbα [10] , which all play a critical role in the circadian regulation of lipid/glucose metabolic genes ( Figure S6 ) [10] , [12] . These results are consistent with the conclusion that the changes in the circadian pattern of expression of glucose metabolic genes are directly related the loss of RORγ rather than changes in the regulation of clock genes by RORγ . We further showed that exogenous expression of RORγ in RORγ−/− liver tissue by adenovirus significantly increased the expression of G6pase , Pepck , Gck , Gckr , Pparδ , Pcx , and Klf15 as well as the RORγ-target gene , Avpr1a ( Figure 5E ) [10] . Similarly , exogenous expression of RORγ in RORγ−/− primary hepatocytes significantly activated the expression of several of these genes ( Figure 5F ) . These data are consistent with the conclusion that these genes are positively regulated by RORγ . To examine whether any of these changes in gene expression translated into alterations in corresponding protein , we analyzed the expression of Pklr , which plays a key role in catalyzing the formation of pyruvate from phosphoenolpyruvate . As shown in Figure 5A and 5B , the level of Pklr protein in WT and RORγ−/− liver correlated rather well with the level of RNA expression . The levels of Pklr protein and RNA were higher at ZT16 than at ZT4 and clearly repressed in RORγ−/− liver . Our ChIP-Seq analysis indicated that in liver both RORα and RORγ are recruited to the proximal promoter of G6pase and to intron 2 of Pparδ ( Figure 4G and Figure S5A ) . ChIP-QPCR analysis showed higher association of RORγ with these regulatory regions at ZT22 compared to ZT10 , whereas relatively little recruitment was observed in RORγ−/− liver at either ZT10 or ZT22 ( Figure S5D , S5E ) . Analysis of the G6pase proximal promoter ( −500/+58 ) identified , in addition to a classical RORE ( RORE1 ) [31] , a RORE variant motif ( RORE2 ) , and a PPAR responsive-element ( PPRE ) ( Figure 6A ) , which has been reported to mediate the transactivation of G6pase by PPARα [32] . Reporter gene analysis showed that both RORγ and RORα were able to highly activate the G6pase promoter ( Figure 6A ) , while the RORγ-selective antagonist “A” [10] inhibited the activation by RORγ at concentrations as low as 100 nM ( Figure 6B ) . Mutation of either the RORE1 or RORE2 greatly reduced the activation by RORs . Interestingly , these RORE mutations also inhibited the transcriptional activation of the G6pase promoter by PPARα . Inversely , a PPRE mutation significantly reduced the activation by RORs as well as by PPARα , while mutation of both ROREs and PPRE almost totally abolished G6pase transactivation ( Figure 6A ) . These observations suggested that RORs and PPARα collectively regulate G6pase expression . The ROR binding region in intron 2 of Pparδ contains three putative ROREs , including a variant sequence ( Figure 6C ) . Reporter analysis showed that RORγ and RORα activated the Luc reporter gene driven by this regulatory region about 45- and 140-fold , respectively . Mutation of any of these 3 ROREs strongly reduced the activation of the reporter by RORγ , while the triple mutation almost totally abolished activation . The RORγ antagonist inhibited this activation in a dose-responsive manner ( Figure 6D ) . These results support the conclusion that Pparδ transcription is directly regulated by RORγ through these response elements and suggest that the circadian regulation of certain metabolic outputs by RORγ may be in part due to its regulation of Pparδ expression . Although RORα was recruited to the RORE-containing regions of G6pase and Pparδ ( Figure S5D , S5E ) and activated the G6pase and the Pparδ regulatory region in reporter assays , loss of RORα had little effect on the circadian expression of G6pase and Pparδ ( Figure 6E ) . The expression of these genes in double knockout RORαsg/sgRORγ−/− liver was reduced to a similar degree as in RORγ−/− liver ( Figure 6F ) . These results suggest that under the conditions tested RORγ rather than RORα , plays a significant role in the hepatic regulation of G6pase and Pparδ in vivo . In addition to G6pase and Pparδ , RORγ was recruited to several other genes important in glucose homeostasis , including intron 1 of Gck , the proximal promoter ( −685/+42 ) of Gckr ( Figure 6G and 6H , Figure S5B ) , intron 2 of Glut2 , the promoter of Gys2 , and the promoter of Dlat ( Figure S7A ) . RORγ was able to activate the Luc reporter gene driven by these regulatory regions . Mutation or deletion of the RORE ( s ) in the Gck and Gckr regulatory region as well as addition of the RORγ antagonist significantly reduced the activation by RORγ ( Figure 6G , 6H , S7B ) . ChIP-Seq analysis showed that RORα was not associated with these genes , and except for Gys2 , RORα-deficiency had little effect on the expression of these genes in vivo ( Figure S7C , S7D ) . Together , these results support the conclusion that RORγ directly regulates the transcription of a series of genes important in glucose metabolism and homeostasis . To determine whether the effects on hepatic glucose metabolism were based on the hepatocyte-specific loss of RORγ function rather than loss of RORγ in other metabolic tissues or immune cells , we analyzed liver-specific RORγ-deficient ( RORγfx/fxAlb-Cre+ ) mice . Our data confirmed that RORγ expression was completely lost in the liver of RORγfx/fxAlb-Cre+ mice and was not changed in the kidney ( Figure 7A ) . ITT , GTT , and PTT analysis showed that , as demonstrated for the RORγ ubiquitous knockout mice , RORγfx/fxAlb-Cre+ ( HFD ) mice exhibited a greater glucose tolerance , were more responsive to insulin , and showed reduced gluconeogenesis , respectively ( Figure 7B–7D ) . Moreover , as in RORγ−/− mice , the blood insulin concentration at ZT16 was significantly reduced in RORγfx/fxAlb-Cre+ ( HFD ) mice and so was the peak accumulation of hepatic glycogen at ZT0 ( Figure 7E ) . Moreover , gene expression analysis showed that the hepatic expression of a series of RORγ target genes important in glucose metabolism , including G6pase and Pparδ , were also decreased in RORγfx/fxAlb-Cre+ mice as seen in RORγ−/− mice ( Figure 7F ) . Together , these observations suggest that the changes in hepatic glucose metabolism are related directly to the loss of RORγ function in the liver and support the conclusion that RORγ directly contributes to the regulation of hepatic gluconeogenesis and glucose metabolism .
In this study , we identify a novel function for RORγ in the daily regulation of hepatic glucose metabolism and insulin sensitivity . Our results demonstrate that at ZT4–6 RORγ−/− mice are significantly more insulin sensitive and glucose tolerant than WT mice , while there was a smaller difference between the two strains at ZT18–20 . The euglycemic clamp test revealed that hepatic glucose production was considerably reduced in RORγ−/− mice ( Figure 2A ) . This was supported by PTT data showing that the conversion of exogenously administered pyruvate to glucose was significantly lower in RORγ−/− mice particularly at ZT4–6 ( Figure 2C ) . Inversely , ectopic RORγ expression in RORγ−/− liver tissue or primary hepatocytes increased glucose production ( Figure 2E , S2C ) . Our ITT and PTT data indicate that the regulation of glucose metabolism by RORγ is also functional at subjective daytime , CT4–6 , under constant darkness ( Figure S1A , S1B ) . Together , these observations demonstrate that gluconeogenesis is less efficient in RORγ−/− liver and support the conclusion that RORγ is an important positive regulator of hepatic gluconeogenesis and insulin sensitivity particularly during early daytime . The regulation of glucose metabolism is complex and not only depends on hepatic metabolism , but also involves control of metabolic pathways in other tissues in which RORγ is expressed , such as adipose and skeletal muscle . It also involves certain regions of the brain , including the SCN and the hypothalamus , which are implicated in the regulation of the central circadian clock and appetite , respectively [16]–[18] . However , in contrast to RORα and RORβ , RORγ is not or very poorly expressed in the SCN , hypothalamus or other parts of the brain [11] , [33] . Therefore , it appears unlikely that the brain plays a major role in the phenotypic changes observed in RORγ−/− mice . In addition , many of the changes in RORγ−/− mice , including the reduction in glucose metabolic gene expression , were also observed in liver-specific RORγ-deficient mice , indicating that these effects are directly related to the loss of RORγ in hepatocytes and separate from the loss of RORγ in other metabolic tissues ( Figure 7F ) . Since RORγ functions as a transcription factor , the reduced gluconeogenesis in RORγ-deficient mice must involve alterations in the transcription of RORγ target genes . De novo motif analysis of the RORγ cistrome identified , in addition to the classic RORE , two variant RORE-like motifs . The variant ROREs appear to allow a greater diversity in ROR binding than expected from the in vitro binding assays [34] , [35] . A greater promiscuity in in vivo DNA binding has also been observed for other nuclear receptors , and might be due to promoter context , chromatin structure , and histone modifications . Gene ontology analysis showed that many of the potential RORγ-target genes are linked to metabolic pathways ( Table 1 and Table S2 ) , including glucose homeostasis ( e . g . , G6pase , Pepck , Pklr , Pparδ , Gck , Gckr , Glut2 , Gys2 , Dlat , Pcx , and Klf15 ) . Analysis of their rhythmic pattern of expression demonstrated that RORγ deletion reduced peak expression of most of these genes , without affecting their phase . Regulation of these genes by RORγ was supported by data showing that exogenous expression of RORγ in RORγ−/− liver and primary hepatocytes significantly enhanced their level of expression ( Figure 5E , 5F ) . Promoter and mutation analysis demonstrated that RORγ was able to activate several of the RORE-containing promoters , while mutation of either the classic or variant ROREs significantly reduced this activation by RORγ indicating that these motifs are functional . The RORγ-mediated promoter activation was further supported by data showing that treatment with a RORγ-selective antagonist considerably inhibited this activation ( Figure 6B , 6D , S7B ) . Our RORγ cistrome data together with the mRNA expression and promoter analysis support the model that in murine liver , RORγ positively regulates the expression of a series of glucose metabolic genes directly through RORE binding . The reduced peak expression of several key metabolic genes , including G6pase and Pepck , which are critical for gluconeogenesis , the glucose transporter Glut2 , and several genes important in the glycolysis pathway , including Plkr , Gck , and Gckr , likely contribute to the reduced glucose uptake , the less efficient gluconeogenesis and the lower glycogen accumulation observed in RORγ deficient liver . In addition to RORγ , glucose metabolism is under the control of a number of other transcription factors . Although loss of RORγ reduced peak expression of several glucose metabolic genes , most of these genes still exhibited a substantial rhythmic pattern of expression , indicating that additional factors are involved . For example , analysis of the G6pase promoter showed that in addition to the classic and variant RORE proximal promoter , it contained a PPRE ( Figure 6A ) , which has been reported to mediate the transactivation of G6pase by PPARα [32] . Mutation of either the ROREs or PPRE caused a significant reduction in the activation of this promoter suggesting that RORγ and PPARα cooperatively regulate G6pase . Although comparison of the RORα and RORγ cistromes indicated that RORα and RORγ have largely distinct functions , there was a 10% overlap in target genes that included several glucose metabolic genes , such as G6pase and Pparδ ( Figure S5 ) . However , in contrast to RORγ−/− mice , loss of RORα did not affect the expression of G6pase or Pparδ ( Figure 6E , 6F ) suggesting that under the conditions tested these genes are regulated by RORγ rather than RORα . Although several studies have demonstrated a role for Bmal1 and Clock in the regulation of several metabolic genes and shown that RORγ is recruited to ROREs in Clock and Bmal1 , the loss of RORγ had little effect on the hepatic expression of Bmal1 and Clock ( Figure S6 ) [8] , [10] . These observations suggest that changes in glucose metabolic genes in RORγ−/− liver are not due to changes in Clock or Bmal1 expression and are consistent with the hypothesis that RORγ regulates these genes downstream of the clock machinery . However , cistrome analysis has shown that Bmal1 can also be recruited to certain glucose metabolic genes , such as G6pase , suggesting that Bmal1 in conjunction with RORγ positively regulates the expression of these genes . In addition , RORγ might cause changes in chromatin structure and as such influences the recruitment of Bmal1 or Clock to common target genes . The Rev-Erb nuclear receptors also play a critical regulatory role in the robust oscillation of circadian expression of a number genes [14] . RORs and Rev-Erb receptors can interfere with each other's activity by competing for RORE binding [10] . Despite the modest reduction in peak expression of Rev-Erbα in RORγ−/− liver ( Figure S6 ) , which should result in increased target gene expression , the loss of RORγ may reduce the competition with Rev-Erbα for RORE binding and as a consequence increase the repression of gene transcription by Rev-Erbα . A more comprehensive comparison between the cistrome of RORs , clock proteins , and Rev-Erbs is needed to provide further insights into the crosstalk between these transcription factors . Although insulin levels were significantly lower in RORγ−/− mice , blood glucose levels were largely maintained ( Figure 3B ) . At daytime , hepatic glucose production is less efficient in the knockout mice and consistent with this , blood insulin level was significantly reduced at ZT4 . We hypothesize that insulin sensitivity in RORγ−/− mice is also improved during nighttime due to reduced hepatic glucose production , which as a consequence would require less insulin to maintain blood glucose level and explain the lower level of blood insulin in RORγ−/− mice . This is supported by AUC analysis for ITT , which indicates that also at nighttime insulin sensitivity was significantly better in RORγ−/− mice ( Figure 1B ) . When mice eat during nighttime , more insulin is required to maintain blood glucose levels and this may account for the greater difference in blood insulin level compared to the difference at daytime . The observation that the PTT indicated little changed in gluconeogenesis efficiency at nighttime may be related to the fact that the PPT determines the efficiency of the gluconeogenesis pathway by measuring the formation of glucose from pyruvate after exogenous pyruvate injection , which is not a total reflection of all the pathways involved in the regulation of hepatic gluconeogenesis in vivo because pyruvate for gluconeogenesis can be supplied by other metabolic pathways . A lower RER is considered to indicate that fat is increasingly preferred as a fuel source , whereas a higher RER is indicative for an increased use of carbohydrates . Thus , the lower RER observed at daytime in both WT and knockout mice indicates a greater preference for fatty acid consumption over glucose compared to nighttime ( Figure 3H ) , while the lower nighttime RER levels in RORγ−/− mice compared to WT mice indicate a greater preference for fatty acid consumption over glucose . The latter is likely related to reduced glucose production and reduced glucose uptake in RORγ knockout liver . Our data show that hepatic glycogen accumulation was reduced in RORγ knockout mice during ZT16-0 clearly indicating that loss of RORγ also affects glucose homeostasis at nighttime . This reduction in glycogen is likely due a reduced glucose uptake , which correlate with the lower levels of blood insulin in RORγ knockout mice ( Figure 3B and 3E ) . Further analyses will be needed to precisely understand the precise interrelationships between various transcription factors , their diurnal regulation of various metabolic pathways and glucose and energy homeostasis . In summary , our study identifies a novel function for RORγ in the regulation of gluconeogenesis and insulin resistance . Our data are consistent with the model in which RORγ directly regulates the expression of glucose metabolic genes in the liver downstream of the hepatic circadian clock , thereby enhancing gluconeogenesis , and decreasing insulin sensitivity and glucose tolerance ( Figure 7G ) . The temporal organization of tissue metabolism is coordinated by reciprocal crosstalk between the core clock machinery and key metabolic enzymes and transcription factors . Our study indicates that RORγ is a novel important participant in this crosstalk . The improved insulin sensitivity and glucose tolerance observed in RORγ-deficient mice suggest that the loss of RORγ might be beneficial in controlling glucose homeostasis and in the management of metabolic diseases . This is supported by recent studies showing that in human patients the level of RORγ expression positively correlates with insulin resistance [20] , [21] . The inhibition of the activation of several glucose metabolic gene promoters by an RORγ-selective antagonist , thereby mimicking the effects in RORγ−/− liver , suggests that such antagonists might provide a novel therapeutic strategy in the management of insulin resistance and type 2 diabetes .
Heterozygous C57BL/6 staggerer ( RORα+/sg ) were obtained from the Jackson Laboratories ( Bar Harbor , ME ) . RORγ−/− and RORαsg/sgRORγ−/− double knockout ( DKO ) mice were described previously [10] , [36] . Liver-specific RORγ knockout mice , referred to as RORγfx/fxAlb-Cre+ , were generated by crossing B6 ( Cg ) -Rorctm3Litt/J ( RORγfx/fx ) with B6 . Cg-Tg ( Alb-cre ) 21Mgn/J transgenic mice ( Jackson Laboratories ) . Mice were supplied ad libitum with NIH-A31 formula ( normal diet , ND ) and water , and maintained at 23°C on a constant 12 h light∶12 h dark cycle . Two month-old male mice were fed with a high fat diet ( 40% kcal fat ) ( HFD: D12079B Research Diets Inc . , New Brunswick , NJ ) for 6 weeks . Littermate wild type ( WT ) mice were used as controls . All animal protocols followed the guidelines outlined by the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the NIEHS . After 16 h fasting , WT and RORγ−/− mice ( n = 8–10 ) fed a ND or HFD for 6 weeks were injected intraperitoneally with glucose ( 2 g/kg ) , insulin ( 0 . 75 U/kg ) ( Eli Lilly , Indianapolis , IN ) or sodium pyruvate ( 2 g/kg ) ( Sigma-Aldrich ) at ZT4 or ZT18 . The blood glucose was measured every 20 min for up to 140 min with glucose test strips ( Nova Biomedical , Waltham , MA ) . These tests were performed in the same way using RORγfx/fxAlb-Cre+ and RORγfx/fxAlb-Cre− mice ( n = 11 ) fed a HFD . ITT and PTT were also performed under red light at CT4 after WT ( HFD ) and RORγ−/− ( HFD ) mice ( n = 12 ) were kept for 1 day under constant darkness . Total AUC ( Area under the curve ) was calculated by the trapezoid rule . Two-way ANOVA was performed using GraphPad PRISM software . To evaluate insulin signaling , liver , BAT , WAT , and skeletal muscle were isolated from fasting WT ( HFD ) and mice RORγ−/− ( HFD ) mice 30 min after injection with either 0 . 75 U/kg insulin or PBS . Protein from these tissues was extracted with lysis buffer ( 25 mM Tris-HCl pH 7 . 6 , 150 mM NaCl , 1% Nonidet P-40 , 1% sodium deoxycholate , 0 . 1% SDS ) . In a separate experiment , primary hepatocytes isolated from WT and RORγ−/− mice were treated with 20 nM insulin in serum-free 199 medium ( Sigma-Aldrich ) for 10 min . Phosphorylated Akt ( Ser473 ) and whole Akt proteins were detected by Western blot analysis with antibodies 7408 and 7102 from Cell Signaling Technology . Pklr and Gapdh were detected in liver lysates from WT and RORγ−/− mice ( n = 3 ) at ZT4 and ZT16 by Western blot analysis with anti-Pklr ( 22456-1-AP , Proteintech Group Inc . , Chicago , IL , USA ) and anti-Gapdh ( Cell Signaling Technology ) antibodies . WT and RORγ−/− mice ( n = 5 ) fed a HFD for 6 weeks underwent surgery under anesthesia to attach catheters to the jugular vein and carotid artery . Mice were left at least 2 days to recover . After a 3 . 5 h fasting , the basal rates of glucose turnover were measured by continuous infusion of HPLC-purified D-[3-3H] glucose ( 0 . 05 µCi/min ) ( Perkin Elmer , Boston , MA ) for 90 minutes following a bolus of 1 µCi . Blood samples ( about 40 µl ) were taken from the carotid artery catheter at 75 and 85 min after the infusion to determine the plasma [3-3H] glucose concentration . Subsequently the hyperinsulinemic euglycemic clamp test was performed for 120 min in conscious , restrained mice . Human insulin ( HumulinR , Eli Lilly ) was infused at a constant rate ( 30 mU/kg/min ) through the end of the experiment following a bolus of 90 mU/kg/min for 3 min . Glucose was measured every 10 min in blood from tail vein with glucose test strips . The glucose concentration was maintained at 110–130 mg/dl by a variable rate of 20% glucose infusion under a continuous infusion of [3-3H] glucose ( 0 . 1 µCi/min ) . Blood samples ( about 40 µl ) were taken from the carotid artery catheter every 10 min during the last 40 min . [3H]-glucose was used to trace hepatic glucose production and glucose turnover . The experiment was performed during daytime at ZT2–9 . For the determination of the plasma 3H-glucose concentration , plasma samples were deproteinized with 0 . 3 N Ba ( OH ) 2 and ZnSO4 and dried to remove 3H2O before the radioactivity was measured in a liquid scintillation counter . Basal hepatic glucose production ( Basal HGP ) was calculated as the ratio of the preclamp [3H]-glucose infusion rate ( GIR ) ( dpm/min ) to the specific activity of plasma glucose . Clamp whole-body glucose disappearance ( Rd ) was calculated as the ratio of the clamp [3-3H] GIR ( dpm/min ) to the specific activity of plasma glucose . Clamp glucose production ( Clamp HGP ) was determined by subtracting the average GIR in the last 40 min from the Rd . Recombinant adenoviruses were generated using the AdEasy adenoviral system ( Agilent Technologies , Palo Alto , CA ) . Full-length RORγ1 cDNA was inserted to pShuttle-IRES-hrGFP-1 vector , and co-transformed with pAdEasy-1 in BJ5183-AD-1 bacteria by electroporation . The recombinant adenovirus plasmid was then transfected in AD-293 cells . The amplified adenoviruses were purified and concentrated by cesium chloride density gradient centrifugation . The empty control and RORγ expressing adenoviruses were injected into the retro-orbital sinus of RORγ−/− ( HFD ) mice ( n = 6–7 ) . Pyruvate tolerance test was performed 4 days later and after an additional four days , liver was collected at ZT8 to analyze glycogen accumulation and gene expression . Hepatocytes from 2 month-old WT and RORγ−/− mouse were isolated with a Hepatocyte Isolation System ( Worthington Biochemical Corporation , New Jersey , USA ) according to the manufacturer's instructions . Primary hepatocytes were cultured in collagen-coated dishes with Medium 199 supplemented with 100 nM dexamethasone , 1 nM insulin , 10 nM triiodothyronine , 5% fetal bovine serum , and penicillin/streptomycin . After 8–12 h , cells were infected with empty lentivirus pLVX-mCherry-N1 or RORγ1-expressing lentivirus . 24 h later cells were washed twice in PBS and then incubated in serum-free medium 199 in the presence or absence of 100 nM insulin or 100 nM glucagon ( Sigma-Aldrich ) for 6 h before RNA was isolated . Glucose production was measured with a glucose production buffer ( glucose/phenol red-free DMEM ( Sigma-Aldrich ) , 1 mM lactose , 2 mM sodium pyruvate ) in RORγ−/− hepatocyte infected with lentivirus for each empty and RORγ expression ( n = 3 ) . Glucose in the medium was measured with a Glucose assay kit ( Sigma-Aldrich ) . Serum and liver samples were collected from WT and RORγ−/− mice on a HFD ( n≥5 ) every 4 h over a period of 24 h . Serum insulin was measured by a sandwich ELISA with a Rat/Mouse Insulin ELISA kit ( EZRMI-13K , Millipore ) . Glucose stimulated insulin secretion ( GSIS ) was measured at ZT4 in WT and RORγ−/− mice on a HFD ( n = 5–6 ) or ND ( n = 2–3 ) . Serum was collected at 2 . 5 , 5 , 15 , and 30 min after intraperitoneal injection of glucose ( 2 g/kg ) . Pancreatic insulin was determined by rapidly removing the pancreas from WT and RORγ−/− mice ( n = 10–14 ) on a HFD . Pancreas was then homogenized and extracted overnight with acid-ethanol at −20°C . Insulin in the extracts was measured with the insulin ELISA kit . Insulin was normalized by total pancreatic protein . Glycogen extracted from liver with 30% KOH at 100°C for 2 h followed by precipitation by ethanol , was measured with a Glycogen Assay Kit ( BioVison Inc . , Mountain View , CA ) . To analyze metabolic parameters including oxygen consumption , CO2 production , respiratory exchange ratio , heat production , and food/water consumptions were measured in WT and RORγ−/− mice ( n = 8 ) with a LabMaster system ( TSE systems Inc . , Chesterfield , MO ) during 4 successive days . The ChIP assay was performed using a ChIP assay kit from Millipore ( Billerica , MA ) according to the manufacturer's protocol with minor modifications as described previously [10] . Briefly , livers collected from WT , RORαsg/sg , and RORγ−/− mice at ZT10 and ZT22 were homogenized with a polytron PT 3000 ( Brinkmann Instruments ) and crosslinked by 1% formaldehyde for 10 min at room temperature . After a wash in PBS , an aliquot of the crosslinked chromatin was sonicated and incubated overnight with an anti-RORα or anti-RORγ antibody [10] generated against amino acids 129–231 and 121–213 in mouse RORγ1 and RORα4 , respectively . After incubation with protein G agarose beads for 2 h , DNA-protein complexes were eluted . The crosslinks were reversed by overnight incubation at 65°C in the presence of 25 mM NaCl , digested with RNase A and proteinase K , and then the ChIPed-DNA was purified . The amount of ChIPed-DNA relative to each input DNA was determined by QPCR . All QPCR reactions were carried out in triplicate . Sequences of primers for ChIP-QPCR are listed in Table S4 . ChIPed-DNA and input DNA as a control were prepared using RORγ- and RORα-specific antibodies as described previously [10] . ChIP-Seq analysis was performed by the NIH Intramural Sequencing Center and data were analyzed as reported previously [37] . The sequencing reads were obtained from base-calling of Illumina Genome Analyzer . The wiggle-formatted alignment results were visualized on UCSC Genome Browser using mouse mm9 reference genome . SISSRs ( Site Identification from Short Sequence Reads ) were used for identification of significant RORγ and RORα binding sites ( P<0 . 001 ) that have enriched reads in each ChIPed-DNA versus input control across the whole genome [38] . The distance from each ROR peak to the nearest transcriptional start sites was determined using custom scripts . De novo consensus motif search within ROR binding sites was performed using MEME . ChIP-Seq data was compared with gene expression data using Kolmogorov-Smirnov ( KS ) plot . Gene ontology analysis was performed using the NIH Database for Annotation , Visualization , and Integrated Discovery ( DAVID ) online web-server , and based on PANTHER Biological process definitions . To quantify gene expression during circadian time , liver tissues were collected from WT , RORγ−/− , and RORαsg/sg mice every 4 h over a period of 24 h , processed overnight in RNAlater solution ( Ambion , Austin , TX ) at 4°C , and then stored at −80°C until use . Tissues were then homogenized with a Polytron PT-3000 ( Brinkmann Instruments , Westbury , NY ) . Liver tissues were also collected from RORαsg/sgRORγ−/− DKO mice and littermate control WT mice , and RORγfx/fxAlb-Cre+ and RORγfx/fxAlb-Cre− mice at zeitgeber time ( ZT ) 8 and ZT20 . RNA was then extracted using a QIAshredder column and RNeasy Mini kit ( Qiagen , Valencia , CA ) according to the manufacturer's instructions . The RNA was reverse-transcribed using a High-Capacity cDNA Archive Kit ( Applied Biosystems ) . QPCR analysis was performed using SYBR Green I ( Applied Biosystems , Foster City , CA ) . The reactions were carried out in triplicate using 20 ng of cDNA and the following conditions: 10 min at 95°C , followed by 40 cycles of 15 sec at 95°C and 60 sec at 60°C . The results were normalized by the amount of Gapdh mRNA . Primer sequences are listed in Table S4 . The promoter or intron region of mouse G6Pase ( promoter; −500/+58 ) , Pparδ ( intron 2; +46417/+46987 ) , Gck ( intron 1; +29709/+30121 ) , Gckr ( promoter; −685/+42 ) , Glut2 ( intron 2; +16294/+16805 ) , Gys2 ( promoter; −256/+345 ) , and Dlat ( promoter; −1151/+22 ) genes was amplified using mouse genomic DNA ( Promega , Madison , WI ) and cloned into either the promoter-less reporter plasmid pGL4 . 10 or pGL4 . 27 containing a minimal promoter ( Promega , Madison , WI ) . Point mutations in ROREs and PPREs were generated using a Quickchange Site-Directed Mutagenesis Kit ( Stratagene , La Jolla , CA ) . Human hepatoma Huh-7 cells were co-transfected with the indicated pGL4 reporter plasmid , pCMV-β-Gal , and p3xFlag-CMV10-RORγ , –RORα , -Rev-Erbα , or -PPARα expression plasmids using lipofectamine 2000 ( Invitrogen , Carlsbad , CA ) . After 24 h incubation , the luciferase and β-galactosidase activities were measured with a Luciferase Assay Substrate kit ( Promega ) and Luminescent β-galactosidase Detection Kit II ( Clontech ) . All transfections were performed in triplicate and repeated at least twice . In certain experiments cells were treated for 24 h with a RORγ-selective antagonist “A” , ( R ) -N- ( 1- ( ( 4-methoxy-phenyl ) sulfonyl ) -4-methyl-1 , 2 , 3 , 4-tetrahydroquinolin-7-yl ) -2 , 4 , 6-trimethylbenzene-sulfonamide provided by Dr . Veronique Birault ( GlaxoSmithKline ) [10] or with the selective PPARα antagonist , Wy14 , 643 ( 10 µM; Sigma-Aldrich ) as indicated . | The circadian clock plays a critical role in the regulation of many physiological processes , including metabolism and energy homeostasis . The retinoic acid-related orphan receptor γ ( RORγ ) functions as a ligand-dependent transcription factor that regulates transcription by binding as a monomer to ROR-responsive elements . In liver , RORγ exhibits a robust circadian pattern of expression that is under direct control of the hepatic circadian clock . However , the connection between the circadian regulation of RORγ and its control of downstream metabolic processes is not well understood . In this study , by using ubiquitous and liver-specific RORγ-deficient mice as models , we demonstrate that hepatic RORγ modulates daily insulin sensitivity and glucose tolerance by regulating hepatic gluconeogenesis . Genome-wide cistromic profiling , gene expression , and promoter analysis revealed that RORγ is targeting and regulating a number of novel metabolic genes critical in the control of glycolysis and gluconeogenesis pathways . We provide evidence for a model in which RORγ regulates the circadian expression of glucose metabolic genes in the liver downstream of the hepatic circadian clock , thereby enhancing gluconeogenesis and decreasing insulin sensitivity and glucose tolerance . This study suggests that attenuating RORγ activity by antagonists might be beneficial for the management of glucose metabolic diseases including type 2 diabetes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"carbohydrate",
"metabolism",
"body",
"weight",
"medicine",
"and",
"health",
"sciences",
"functional",
"genomics",
"diabetic",
"endocrinology",
"gene",
"regulation",
"regulatory",
"proteins",
"dna",
"transcription",
"diabetes",
"mellitus",
"physiological",
"parameters",
"... | 2014 | Retinoic Acid-Related Orphan Receptor γ (RORγ): A Novel Participant in the Diurnal Regulation of Hepatic Gluconeogenesis and Insulin Sensitivity |
The mismatch negativity ( MMN ) is an event related potential evoked by violations of regularity . Here , we present a model of the underlying neuronal dynamics based upon the idea that auditory cortex continuously updates a generative model to predict its sensory inputs . The MMN is then modelled as the superposition of the electric fields evoked by neuronal activity reporting prediction errors . The process by which auditory cortex generates predictions and resolves prediction errors was simulated using generalised ( Bayesian ) filtering – a biologically plausible scheme for probabilistic inference on the hidden states of hierarchical dynamical models . The resulting scheme generates realistic MMN waveforms , explains the qualitative effects of deviant probability and magnitude on the MMN – in terms of latency and amplitude – and makes quantitative predictions about the interactions between deviant probability and magnitude . This work advances a formal understanding of the MMN and – more generally – illustrates the potential for developing computationally informed dynamic causal models of empirical electromagnetic responses .
Recent advances in computational neuroimaging [1] have enabled inferences about the neurophysiological mechanisms that generate non-invasive measures of task or stimulus-evoked neuronal responses; as measured by functional magnetic resonance imaging ( fMRI ) or electroencephalography ( EEG ) . One such approach is dynamic causal modelling [2] that tries to explain EEG data in terms of synaptic coupling within a network of interacting neuronal populations or sources . However , this description is at the level of physiological processes that do not have a direct interpretation in terms of information processing . Cognitive scientists have been using formal models of cognitive processes to infer on information processing from behaviour for decades [3] , but it has remained largely unclear how such inferences should be informed by neurophysiological data . We argue that one may overcome the limitations of both approaches by integrating normative models of information processing ( e . g . , [4] , [5] ) with physiologically grounded models of neuroimaging data [4] , [5] . This approach may produce computationally informed neuronal models – or neurocomputational models – enabling one to test hypotheses about how the brain processes information to generate adaptive behaviour . Here , we provide a proof-of-concept for this approach by jointly modelling a cognitive process – perceptual inference – and the event related potential ( ERP ) that it may generate – the mismatch negativity ( MMN ) . Specifically , we ask whether the MMN can be modelled by a neuronal system performing perceptual inference , as prescribed by predictive coding [4] , [5] . The MMN is an event-related potential that is evoked by the violation of a regular stream of sensory events . By convention , the MMN is estimated by subtracting the ERP elicited by standards , i . e . events that established the regularity , from the ERP elicited by deviants , i . e . events violating this regularity . Depending on the specific type of regularity , the MMN is usually expressed most strongly at fronto-central electrodes , with a peak latency between 100 and 250 milliseconds after deviant onset [1] . More precisely , the MMN has been shown to depend upon deviant probability and magnitude . Deviant probability is the relative frequency of tones that violate an established regularity . In studies of the MMN evoked by changes in sound frequency , deviance magnitude is the ( proportional ) difference between the deviant frequency and the standard frequency . The effects of these factors are usually summarized in terms of changes in the MMN peak amplitude and its latency ( see Table 1 ) . While increasing the deviance magnitude makes the MMN peak earlier and with a larger amplitude [4] , [6] , [7] , decreasing deviant probability only increases the MMN peak amplitude [8] but does not change its latency [9] . The question as to which neurophysiological mechanisms generate the MMN remains controversial ( cf . [10] vs . [11] ) , even though this issue has been addressed by a large number of studies over the last thirty years [12] . One reason for an enduring controversy could be that the MMN's latency and amplitude contain insufficient information to disambiguate between competing hypotheses ( but see [13] ) . While the MMN is the sum of overlapping subcomponents that are generated in temporal and frontal brain areas [12] , [14] – and are differentially affected by experimental manipulations [15] – it is a continuous function of time . This means that the underlying ERP waveforms may contain valuable information about MMN subcomponents , the physiological mechanisms that generate them and , critically , their functional correlates ( see e . g . [16] ) . Predictive coding offers a unique and unified explanation of the MMN's neurophysiological features . In brief , predictive coding is a computational mechanism that formally links perception and learning processes to neural activity and synaptic plasticity , respectively [17] . More precisely , event-related electrophysiological responses are thought to arise from the brain's attempt to minimize prediction errors ( i . e . differences between actual and predicted sensory input ) through hierarchical Bayesian inference . In this context , the MMN simply reflects neuronal activity reporting these prediction errors in hierarchically organized network of auditory cortical sources . If this is true , then the rise and fall of the MMN may reflect the appearance of a discrepancy between sensory input and top-down predictions – and its resolution through perceptual inference . These ideas have been used to interpret the results of experimental studies of the MMN [8] , [18] and computational treatments of trial-wise changes in amplitude [6] . However , no attempt has been made to quantitatively relate predictive coding models to empirical MMN waveforms . Here , we extend these efforts by explicitly modelling the physiological mechanisms underlying the MMN in terms of a computational mechanism: predictive coding . In other words , our model is both an extension to dynamic causal models of observed electrophysiological responses [18] , [19] to information processing , and a neurophysiological view on meta-Bayesian approaches to cognitive process [15] . We establish the face validity of this neurocomputational model in terms of its ability to explain the observed MMN and its dependence on deviant frequency and deviance magnitude . This paper comprises two sections . In the first section , we summarize mathematical models of predictive coding ( as derived from the free energy principle ) , and describe the particular perceptual model that we assume the brain uses in the context of a predictable stream of auditory stimuli . The resulting scheme provides a model of neuronal responses in auditory oddball paradigms . In line with the DCM framework , we then augment this model with a mapping from ( hidden ) neuronal dynamics to ( observed ) scalp electrophysiological data . In the second section , we use empirical ERPs acquired during an oddball paradigm to tune the parameters of the observation model . Equipped with these parameters , we then simulate MMN waveforms under different levels of deviant probability and deviance magnitude – and compare the resulting latency and amplitude changes with findings reported in the literature . This serves to provide a proof of principle that dynamic causal models can have a computational form – and establish the face validity of predictive coding theories brain function .
Perception estimates the causes ( ) of the sensory inputs ( ) that the brain receives . In other words , to recognise causal structure in the world , the brain has to invert the process by which its sensory consequences are generated from causes in the environment . This view of perception as unconscious inference was introduced by Helmholtz [2] in the 19th century . More recently , it has been formalized as the inversion of a generative model of sensory inputs [20] . In the language of probability theory , this means that the percept corresponds to the posterior belief about the putative causes of sensory input and any hidden states that mediate their effect . This means that any perceptual experience depends on the model of how sensory input is generated . To capture the rich structure of natural sounds , the model has to be dynamic , hierarchical , and nonlinear . Hierarchical dynamic models ( HDMs ) [21] accommodate these attributes and can be used to model sounds as complex as birdsong [22] . HDMs generate time-continuous data as noisy observations of a nonlinear transformation of hidden states and hidden causes : ( 1 ) where the temporal evolution of hidden states is given by the differential equation: ( 2 ) This equation models the change in as a nonlinear function of the hidden states and hidden causes plus state noise . The hidden causes of the change in are modelled as the outputs of a hidden process at the second level . This second process is modelled in the same way as the hidden process at the first level , but with new nonlinear functions and : ( 3 ) As in the first level , the hidden dynamics of the second level are driven by hidden causes that are modelled as the output of a hidden process at the next higher level , and so forth . This composition can be repeated as often as necessary to model the system under consideration – up to the last level , whose input is usually modelled as a known function of time plus noise: ( 4 ) The ( Bayesian ) inversion of HDMs is a difficult issue , which calls for appropriate approximation schemes . To explain how the brain is nevertheless able to recognise the causes of natural sounds , we assume that it performs approximate Bayesian inference by minimising variational free energy [23] . More generally , the free-energy principle is a mathematical framework for modelling how organisms perceive , learn , and make decisions in a parsimonious and biologically plausible fashion . In brief , it assumes that biological systems like the brain solve complex inference problems by adopting a parametric approximation to a posterior belief over hidden causes and states . It then optimises this approximation by minimizing the variational free-energy: ( 5 ) One can think of this free-energy as an information theoretic measure of the discrepancy between the brain's approximate belief about the causes of sensory input and the true posterior density . According to the free-energy principle , cognitive processes and their neurophysiological mechanisms serve to minimize free-energy [24] – generally by a gradient descent with respect to the sufficient statistics of the brain's approximate posterior [5]: ( 6 ) This idea that the brain implements perceptual inference by free-energy minimization is supported by a substantial amount of anatomical , physiological , and neuroimaging evidence [4] . Algorithms that invert HDMs by minimizing free-energy , such as dynamic expectation maximization [25] , [26] and generalized filtering ( GF ) [4] , [5] , [23] , [27] , [28] , are therefore attractive candidates for simulating and understanding perceptual inference in the brain . Importantly , algorithmic implementations of this gradient descent are formally equivalent to predictive coding schemes . In brief , representations ( sufficient statistics encoding approximate posterior expectations ) generate top-down predictions to produce prediction errors . These prediction errors are then passed up the hierarchy in the reverse direction , to update posterior expectations . This ensures an accurate prediction of sensory input and all its intermediate representations . This hierarchal message passing can be expressed mathematically as a gradient descent on the ( sum of squared ) prediction errors which are weighted by their precisions ( inverse variances ) : ( 6b ) where are prediction errors and are their precisions ( inverse variances ) . Here and below , the ∼ notation denotes generalised variables ( state , velocity , acceleration and so on ) . The first pair of equalities just says that posterior expectations about hidden causes and states change according to a mixture of prior prediction– the first term – and an update term in the direction of the gradient of ( precision-weighted ) prediction error . The second pair of equations expresses precision weighted prediction error as the difference between posterior expectations about hidden causes and ( the changes in ) hidden states and their predicted values ( , ) , weighed by their precisions . The predictions are nonlinear functions of expectations at each level of the hierarchy and the level above . In what follows , this predictive coding formulation will serve to simulate perceptual recognition . We will then use prediction errors as a proxy for neuronal activity producing ERPs . To simulate neuronal processing using Equation 6 , we need to specify the form of the functions that constitute the generative model: To model auditory cortical responses , we assume that cortical sources embody a hierarchical model of repeated stimuli . In other words , the hierarchical structure of the auditory cortex recapitulates the hierarchical structure of sound generation ( cf . [25] ) . This hierarchical structure was modelled using the HDM illustrated in Figure 2 . Note that this model was used to both generate stimuli and simulate predictive coding – assuming the brain is using the same model . The model's sensory prediction took the form of a vector of loudness modulated frequency channels ( spectrogram ) at the lowest level . The level above models temporal fluctuations in instantaneous loudness ( ) and frequency ( ) . The hidden causes and of these fluctuations are produced by the highest level . These three levels of representation can be mapped onto three hierarchically organized areas of auditory cortex: primary auditory cortex ( A1 ) , lateral Heschl's gyrus , and inferior frontal gyrus . A1 and lateral Heschl's gyrus contain neuronal units encoding posterior expectations and prediction errors , respectively . The activity of the expectation units encodes the time course of for A1 and expectations about hidden states for Heschl's gyrus . Error units encode prediction error , i . e . the difference between posterior expectations and top-down predictions . Top-down connections therefore convey predictions , whereas bottom-up connections convey prediction errors . The hidden causes are the expectations of , providing top-down projections from units in inferior frontal gyrus . Our model respects the tonotopic organization of primary auditory cortex ( see e . g . [26] ) by considering 50 frequency channels . It also captures the fact that , while most neurons in A1 have a preferred frequency , their response also increases with loudness [29]–[31] . Specifically , we assume that the activity of neurons selective for frequency is given by: ( 7 ) We can rewrite this equation in terms of the loudness and a tuning function that measures how close the log-frequency is to the neuron's preferred log-frequency : ( 8 ) This is our ( perceptual ) model of how the frequency and loudness is encoded by frequency-selective neurons in primary auditory cortex . We use it to simulate the activity of A1 neurons . Note that a neuronal representation of depends only on frequency . In the brain , frequency representations that are invariant to the sound level ( and other sound attributes ) are found in higher auditory areas; for instance in marmoset auditory cortex [32] . Neuroimaging in humans suggests that periodicity is represented in lateral Heschl's gyrus and planum temporale [33] , and LFP recordings from patients again implicate lateral Heschl's gyrus [34] . We therefore assume that is represented in lateral Heschl's gyrus . The dynamics of the instantaneous frequency is given by ( 9 ) This equation says that the instantaneous frequency converges towards the current target frequency at a rate of . In the context of communication , one can think of the target frequency as the frequency that an agent intends to generate , where the instantaneous frequency is the frequency that is currently being produced . The motivation for this is that deviations from the target frequency will be corrected dynamically over time . The agent's belief about reflects its expectation about the frequency of the perceived tone and its subjective certainty or confidence about that expectation . Therefore , the effect of the deviant probability – in an oddball paradigm – can be modelled via the precision of this prior belief . The temporal evolution of the hidden states and ( encoding loudness ) was modelled with the following linear dynamical system: ( 10 ) In this equation the first hidden cause drives the drives the dynamics of hidden states , which spiral ( decay ) towards zero in its absence . Finally , our model makes the realistic assumption that the stochastic perturbations are smooth functions of time . This is achieved by assuming that the derivatives of the stochastic perturbations are drawn from a multivariate Gaussian with zero mean: ( 11 ) The parameters of this model were chosen according to the biological and psychological considerations explained in Supplementary Text S1 . Having posited the relevant part of the generative model embodied by auditory cortex , one can now proceed to its inversion by the Bayesian generalized filtering scheme described in Equation 6 . This is the focus of the next section , which recapitulates how auditory cortex might perceive sound frequency and amplitude using predictive coding mechanisms , given the above hierarchal dynamic model . The production of the MMN from prediction errors was modelled as a two stage process: the generation of scalp potentials from neuronal responses and subsequent data processing ( see Figure 1 ) . We modelled the scalp potentials ( at one fronto-central electrode ) as the linear superposition of electromagnetic fields caused by the activity of prediction error units in the three simulated cortical sources – plus background activity . Specifically , prediction error units in the A1 source are assumed to encode – the precision weighted sensory error; error units in lateral Heschl's gyrus were assumed to encode – the precision weighted errors in the motion of hidden ( log-frequency and amplitude ) states; and prediction error units in the inferior frontal gyrus were assumed to encode – the precision weighted errors in their inferred causes . The prediction errors were transformed into event related potentials by three transformations . First , the time axis was shifted ( to accommodate conduction delays from the ear ) and scaled so that the simulated stimulus duration was 70 ms . Second , a sigmoidal transformation was applied to capture the presumably non-linear mapping from signed precision-weighted prediction error to neural activity ( i . e . the firing rate cannot be negative and saturates for high prediction error ) and in the mapping from neuronal activity to equivalent current dipole activity; these first two steps are summarized by ( 12 ) Finally , the scalp potential is simulated with a linear combination of the three local field potentials plus a constant: ( 13 ) Data processing was simulated by the application of down-sampling to 200 Hz and a 3rd order Butterworth low-pass filtering with a cut-off frequency of 40 Hz , cf . [6] , [23] , [28] , [39] . We performed two simulations for each condition . In the first simulation the subject expected stimulus A but was presented with stimulus B ( deviant ) . In the second simulation , the subject expected stimulus B and was presented with stimulus B ( standard ) . The MMN was estimated by the difference wave ( deviant ERP – standard ERP ) . This procedure reproduces the analysis used in electrophysiology [7] , [40] . This completes the specification of our computationally informed dynamic causal model of the MMN . To explore the predictions of this model under different levels of deviant probability and magnitude , we first estimated the biophysical parameters ( i . e . the slope parameters in ( 12 ) and the lead field in ( 13 ) ) from the empirical ERPs described in [19] , using standard nonlinear least-squares techniques ( i . e . the GlobalSearch algorithm [41] from the Matlab Global Optimization toolbox ) . We then used the estimated parameters to predict the MMN under different combinations of deviant probability and magnitude . In particular , the simulated MMN waveforms were used to reproduce the descriptive statistics typically reported in MMN experiments , i . e . MMN amplitude and latency . MMN latency was estimated by the fractional area technique [19] , because it is regarded as one of the most robust methods for measuring ERP latencies [42] . Specifically , we estimated the MMN latency as the time point at which 50% of the area of the MMN trough lies on either side . This analysis was performed on the difference wave between the first and last point at which the amplitude was at least half the MMN amplitude . This analysis was performed on the unfiltered MMN waveforms as recommended by [43] . MMN amplitude was estimated by the average voltage of the low-pass filtered MMN difference wave within a ±10 ms window around the estimated latency .
Figure 4 shows that the waveforms generated by our model reproduce the characteristic shape of the MMN , the positivity evoked by the standard and the negativity evoked by the deviant . The latency of the simulated MMN ( 164 ms ) was almost identical to the latency of the empirical MMN ( 166 ms ) . Its peak amplitude ( −2 . 71 µV ) was slightly higher than for the empirically measured MMN ( ) , and its width at half-maximum amplitude ( 106 ms ) was also very similar to the width of the empirical MMN waveform ( 96 ms ) . In short , having optimised the parameters mapping from the simulated neuronal activity to empirically observed responses , we were able to reproduce empirical MMNs remarkably accurately . This is nontrivial because the underlying neuronal dynamics are effectively solving a very difficult Bayesian model inversion or filtering problem . Using these optimised parameters , we proceeded to quantify how the MMN waveform would change with deviance magnitude and probability . To simulate the effect of deviant probability , we simulated the responses to a deviant under different degrees of prior certainty . To simulate the effect of deviance magnitude , we varied the discrepancy between the expected and observed frequency , while keeping the deviant probability constant . Finally , we investigated potential interactions between deviance magnitude and deviant probability by simulating the effect of magnitude under different prior certainties and vice versa .
Our MMN simulations predict a nonlinear interaction between the effects of deviant probability and magnitude . The upper plot in Figure 6 suggests that the effect of deviant probability on MMN peak amplitude increases with increasing deviance magnitude . Conversely , the effect of deviance magnitude increases with decreasing deviant probability . Furthermore , the lower plot in Figure 6 suggests , that the effect of deviant probability on MMN latency depends on deviance magnitude: If deviance magnitude is at most 12 . 7% , the MMN latency does not depend on deviant probability , but when deviance magnitude is as large as 32% , the MMN latency increases with deviant probability . Conversely , the size of the effect of deviance magnitude on MMN latency depends on deviant probability . Hence , our simulations predict a number of interaction effects that can be tested empirically . Although the physiological mechanisms generating the MMN have been modelled previously [9] , the model presented here is the first to bridge the gap between the computations implicit in perceptual inference and the neurophysiology of ERP waveforms . In terms of Marr's levels of analysis [53] , our model provides an explanation at both the algorithmic and implementational levels of analysis – and represents a step towards full meta-Bayesian inference – namely inferring from measurements of brain activity on how the brain computes ( cf . [13] , [19] , [51]–[55] ) . Our model builds upon the proposal that the brain inverts hierarchical dynamic models of its sensory inputs by minimizing free-energy in a hierarchy of predictive coding circuits [56] . Specifically , we asked whether the computational principles proposed in [15] , [20] are sufficient to generate realistic MMN waveforms and account for their dependence on deviant probability and deviance magnitude . In doing so , we have provided a more realistic account of the algorithmic nature of the brain's implementation of these computational principles: While previous simulations have explored the dynamics of perceptual inference prescribed by the free-energy principle using dynamic expectation maximization ( DEM ) [23] , [39] , the simulations presented here are based on GF [23] , [39] . Arguably , GF provides a more realistic model of learning and inference in the brain than DEM , because it is an online algorithm that can be run in real-time to simultaneously infer hidden states and learn the model; i . e . , as sensory inputs arrive . In contrast to DEM it does not have to iterate between inferring hidden states , learning parameters , and learning hyperparameters . This is possible , because GF dispenses the mean-field assumption made by DEM . Another difference to previous work is that we have modelled the neural representation of precision weighted prediction error by sigmoidal activation functions , whereas previous simulations ignored potential nonlinear effects by assuming that the activity of prediction error units is a linear function of precision weighted prediction error [6] , [24] , [27] , [39] . Most importantly , the model presented here connects the theory of free-energy minimisation and predictive coding to empirical measurements of the MMN in human subjects . To our knowledge , our model is the first to provide a computational explanation of the MMN's dependence on deviance magnitude , deviant probability , and their interaction . While [26] modelled the effect of deviance magnitude , they did not consider the effect of deviant probability . Although [6] , [24] modelled the effect of deviant probability , they did not simulate the effect of deviance magnitude , nor did they make quantitative predictions of MMN latency or amplitude . Mill et al . [13] , [55] simulated the effects of deviance magnitude and deviant probability on the firing rate of single auditory neurons in anaesthetized rats . While their simulations captured the qualitative effects of deviance magnitude and deviant probability on response amplitude , they did not capture the shortening of the MMN latency with decreasing deviant probability . By contrast , our model generates realistic MMN waveforms and explains the qualitative effects of deviant probability and magnitude on the amplitude and latency of the MMN . Beyond this , our model makes remarkably accurate quantitative predictions of the MMN amplitude across two experiments [53] examining several combinations of deviance magnitude and deviant probability . The simulations reported in this paper demonstrate that predictive coding can explain the MMN and certain aspects of its dependence on the deviant stimulus and its context . However , they do not imply that the assumptions of predictive coding are necessary to explain the MMN . Instead , the simulations are a proof-of-concept that it is possible to relate the MMN to a process model of how prediction errors are encoded dynamically by superficial pyramidal cells during perceptual inference . For parsimony , our model includes only those three intermediate levels of the auditory hierarchy that are assumed to be the primary sources of the MMN . In particular , we do not model the subcortical levels of the auditory system . However , our model does not assume that predictive coding starts in primary auditory cortex . To the contrary , the input to A1 is assumed to be the prediction error from auditory thalamus . This is consistent with the recent discovery of subcortical precursors of the MMN [52] . Since MMN waveforms were simulated using the parameters estimated from the average ERPs reported in [9] , [10] , the waveforms shown in Figure 4 are merely a demonstration that our model can fit empirical data . However , the model's ability to predict how the MMN waveform changes as a function of deviance magnitude and deviant probability speaks to its face validity . Our model's most severe failure was that while our model correctly predicted that MMN latency shortens with deviance magnitude , it failed to predict that this shortening occurs gradually for deviance magnitudes between 2 . 5% and 7 . 5% . In principle , the model predicts that the latency shortens gradually within a certain range of deviance magnitudes , but this range did not coincide with the one observed empirically . There are clearly many explanations for this failure – for example , an inappropriate generative model or incorrect forms for the mapping between prediction errors and local field potentials . Perhaps the more important point here is that these failures generally represent opportunities . This is because one can revise or extend the model and compare the evidence for an alternative model with the evidence for the original model using Bayesian model comparison of dynamic causal models in the usual way [57]–[59] . Indeed , this is one of the primary motivations for developing dynamic causal models that are computationally informed or constrained . In other words , one can test competing hypotheses or models about both the computational ( and biophysical ) processes underlying observed brain responses . This work is a proof-of-principle that important aspects of evoked responses in general – and the MMN in particular – can be explained by formal ( Bayesian ) models of the predictive coding mechanism [19] . Our model explains the dynamics of the MMN in continuous time and some of its phenomenology at a precision level that has not been attempted before . By placing normative models of computation within the framework of dynamic causal models one has the opportunity to use Bayesian model comparison to adjudicate between competing computational theories . Future studies might compare predictive coding to competing accounts such as the fresh-afferent theory [60]–[62] . In addition , the approach presented here could be extended to a range of potentials evoked by sensory stimuli , including the N1 and the P300 , in order to generalise the explanatory scope of predictive coding or free energy formulations . This sort of modelling approach might be used to infer how perceptual inference changes with learning , attention , and context . This is an attractive prospect , given that the MMN is elicited not only in simple oddball paradigms , but also in more complex paradigms involving the processing of speech , language , music , and abstract features [7] , [53] , [63] . Furthermore , a computational anatomy of the MMN might be useful for probing disturbances of perceptual inference and learning in psychiatric conditions , such as schizophrenia [13] , [55] . Similarly , extensions of this model could also be used to better understand the effects of drugs , such as ketamine [12] , [64]–[66] , or neuromodulators , such as acetylcholine [67]–[69] , on the MMN . We hope to pursue this avenue of research in future work . | Computational neuroimaging enables quantitative inferences from non-invasive measures of brain activity on the underlying mechanisms . Ultimately , we would like to understand these mechanisms not only in terms of physiology but also in terms of computation . So far , this has not been addressed by mathematical models of neuroimaging data ( e . g . , dynamic causal models ) , which have rather focused on ever more detailed inferences about physiology . Here we present the first instance of a dynamic causal model that explains electrophysiological data in terms of computation rather than physiology . Concretely , we predict the mismatch negativity – an event-related potential elicited by regularity violation – from the dynamics of perceptual inference as prescribed by the free energy principle . The resulting model explains the waveform of the mismatch negativity and some of its phenomenological properties at a level of precision that has not been attempted before . This highlights the potential of neurocomputational dynamic causal models to enable inferences from neuroimaging data on neurocomputational mechanisms . | [
"Abstract",
"Introduction",
"Models",
"Results",
"Discussion"
] | [] | 2013 | A Neurocomputational Model of the Mismatch Negativity |
Beyond Mycobacterium ulcerans—specific therapy , sound general wound management is required for successful management of Buruli ulcer ( BU ) patients which places them among the large and diverse group of patients in poor countries with a broken skin barrier . Clinically BU suspicious patients were enrolled between October 2013 and August 2015 at a primary health care ( PHC ) center and a municipal hospital , secondary health care ( SHC ) center in Ghana . All patients were IS2404 PCR tested and divided into IS2404 PCR positive and negative groups . The course of wound healing was prospectively investigated including predictors of wound closure and assessment of infrastructure , supply and health staff performance . 53 IS2404 PCR positive patients—31 at the PHC center and 22 at the SHC center were enrolled—and additionally , 80 clinically BU suspicious , IS2404 PCR negative patients at the PHC center . The majority of the skin ulcers at the PHC center closed , without the need for surgical intervention ( 86 . 7% ) compared to 40% at the SHC center , where the majority required split-skin grafting ( 75% ) or excision ( 12 . 5% ) . Only 9% of wounds at the PHC center , but 50% at the SHC center were complicated by bacterial infection . The majority of patients , 54 . 8% at the PHC center and 68 . 4% at the SHC center , experienced wound pain , mostly severe and associated with wound dressing . Failure of ulcers to heal was reliably predicted by wound area reduction between week 2 and 4 after initiation of treatment in 75% at the PHC center , and 90% at the SHC center . Obvious reasons for arrested wound healing or deterioration of wound were missed additional severe pathology; at the PHC center ( chronic osteomyelitis , chronic lymphedema , squamous cell carcinoma ) and at the SHC center ( malignant ulceration , chronic lymphedema ) in addition to hygiene and wound care deficiencies . When clinically suspicious , but IS2404 PCR negative patients were recaptured in the community , 76/77 ( 98 . 7% ) of analyzed wounds were either completely closed ( 85 . 7% ) or almost closed ( 13% ) . Five percent were found to have important missed severe pathology ( chronic osteomyelitis , ossified fibroma and suspected malignancy ) . The wounds of most BU patients attending the primary health care level can be adequately managed . Additionally , the patients are closer to their families and means of livelihood . Non-healing wounds can be predicted by wound area reduction between 2 to 4 weeks after initiation of treatment . Patients with clinically BU suspicious , but PCR negative ulcers need to be followed up to capture missed diagnoses .
Buruli ulcer ( BU ) is a chronic necrotising disease of the skin and subcutaneous soft tissue . It is currently endemic in more than 30 countries , particularly in West and Central Africa , where it is predominantly found in children [1] . The causative organism , Mycobacterium ulcerans , produces a macrolide toxin , mycolactone , which causes tissue destruction and inhibits local immune responses [2] . The disease commonly starts as a papule , nodule or a plaque and progresses to ulceration and often permanent disability in advanced disease [3 , 4] . Current World Health Organization ( WHO ) guidelines for BU recommend antimycobacterial drug treatment with rifampicin in combination with streptomycin ( “standard antibiotic treatment” ) and refer to”growing evidence of the efficacy of some rifampicin-based oral therapies” ( e . g . clarithromycin ) for 8 weeks [4] . Thermotherapy shows promise , as recently demonstrated in a proof-of-principle study and a clinical trial [5 , 6] . Specific antimycobacterial therapy and general wound management are equally challenging tasks to cure BU patients . Irrespective of wound etiology , healing requires favorable systemic conditions such as balanced nutrition , protection from trauma and infection , moist wound environment , and peri-lesional edema and pain control [7 , 8] . Of concern in the treatment of BU wounds is that the time point of transition from mycobacterial cure to the phase where only general wound management is needed , can currently not be determined with certainty . Persistent M . ulcerans infection , relapse , immune reconstitution-associated so called paradoxical reactions and secondary infections by other pathogens are all possible differential diagnoses of failing wound healing [9–11] . Mycolactone-based point-of care tests may solve this problem in the future . Awareness for this problem comes from clinical monitoring of wounds in the context of clinical trials [6 , 12–14] . We are not aware of large prospective wound management studies conducted outside clinical BU trials capturing real life condition in the health care system . In most health care centers of countries with limited resources , wound management guidelines are not strictly implemented . Additionally , vertical programs select disease-specifically leaving wounds which are not in the focus unattended . In the present study , patients with IS2404 PCR positive and negative ulcers were prospectively observed , comparing wound management at a center of the primary and the secondary health care levels . Additionally , patients with ulcers clinically diagnosed as BU at the PHC center but not confirmed by IS2404 PCR were recaptured in the community to verify the correctness of the classification as non-BU cases , to assess the course of wound healing , to make a final diagnosis and to provide treatment , if wounds persisted .
The study was conducted in accordance with the ethical principles that have their origin in the Declaration of Helsinki and in compliance with ICH-GCP , ISO 14155–1 and -2 , and the applicable laws and regulations of the participating country . Approval was obtained from the Institutional Review Board of the Noguchi Memorial Institute for Medical Research , Legon , Ghana and the Ghana Health Service Ethical Review Committee , reference number , GHS-ERC: 07/07/13 ( see S1 Ethical Clearance ) .
A total of 133 patients , 111 at the PHC center and 22 at the SHC center , were enrolled in the study . At the PHC center , 31 patients were IS2404 PCR positive , and at the SHC center 22 . At the PHC center 25 of the IS2404 PCR positive and 77 of the negative patients , and at the SHC center 15 of the IS2404 PCR positive patients entered the analysis of wound healing . For further details see Fig 1 . As expected for an African BU patient population , more than half of the IS2404 PCR positive patients at the PHC center ( 53%; 16/31 ) were younger than 16 years ( Fig 2A ) . Dominance of children was more pronounced among the patients with PCR negative ulcers , where 82 . 5% ( 66/80 ) were younger than 16 years ( Fig 2B ) . In contrast , the majority of PCR-confirmed BU patients at the SHC center ( 77 . 3%; 17/22 ) were older than 15 years ( Fig 2A ) . Among the PCR-confirmed BU patients , the male to female ratio was less balanced ( 63 . 6% males at the SHC center and 58 . 1% at the PHC center ) , than among the patients with PCR negative ulcers ( 52 . 5% males ) . Six percent ( 2/31 ) of patients at the PHC center and 36% ( 8/22 ) at the SHC center were underweight . The general health status of patients at the PHC center was good; comorbid conditions identified were HIV infection ( n = 1 ) , sickle cell disease ( n = 1 ) and arterial hypertension ( n = 1 ) . At the SHC center , comorbid conditions were more abundant and included arterial hypertension ( n = 7 ) , diabetes mellitus ( n = 2 ) , sickle cell disease ( n = 1 ) , HIV infection ( n = 1 ) , asthma ( n = 1 ) , peptic ulcer disease ( n = 1 ) and epilepsy ( n = 1 ) . All IS2404 PCR positive BU patients presented with ulcerative lesions ( Fig 3 ) . At the PHC center 30 ulcers were assessed and all were located on the lower ( 93 . 3% ) and upper ( 6 . 7% ) extremities with 63 . 3% of ulcers on the right side . Among the 20 ulcers assessed at the SHC center , 87 . 5% , 4 . 2% and 8 . 3% were located on the lower extremities , upper extremities ( 63 . 6% on the right side ) and the face , respectively . Of the 30 PCR positive skin ulcers assessed at the PHC center , 26 ( 86 . 7% ) healed completely without the need for surgical or other adjunctive therapy and 4 ( 13 . 3% ) did not heal during the observation period ( Fig 4A and 4B ) . Sixty-five percent ( 17/26 ) of healed skin ulcers healed in less than 3 months , 26 . 9% ( 7/26 ) between 3 and 6 months and 7 . 7% ( 2/26 ) after 6 months . Three out of four non-healing skin ulcers were reliably predicted to not respond to treatment , and fulfilled Flanagan’s criteria ( Fig 5 ) . For the 4 PCR positive skin ulcers that failed to heal , underlying problems identified were exposed bone ( n = 1 ) , wound infection ( n = 1 ) , wound location at a joint ( n = 1 ) and poor adherence to treatment ( n = 1 ) . Additional severe pathologies identified amongst the IS2404 positive skin ulcers were chronic osteomyelitis ( n = 2 ) , chronic lymphedema ( n = 2 ) ( Fig 6D ) , and squamous cell carcinoma ( n = 1 ) . These ulcers were excluded from the wound healing analysis because the additional severe pathologies found preclude healing ( see Fig 1 ) . Of the twenty IS2404 PCR positive skin ulcers assessed at the SHC center , 8 ( 40% ) healed at the endpoint after split-skin grafting ( 75% ) , excision ( 12 . 5% ) and without adjunct treatment ( 12 . 5% ) . Twelve IS2404 PCR positive ulcers ( 60% ) did not heal during the observation period of which 10 ( 83 . 3% ) persisted in a chronic state and 2 ( 16 . 7% ) in a sub-chronic wound state despite regular wound care . Underlying pathologies identified for delayed wound healing were: wound infection ( n = 10 ) , venous and arterial insufficiency ( n = 4 ) and nutritional deficiency ( n = 7 ) . Some of the patients had more than one possible cause of delayed wound healing . Ninety percent ( 9/10 ) of chronic wounds were reliably predicted to not respond to treatment , and fulfilled Flanagan´s criteria . At enrolment the mean wound area for sub-chronic and healed wounds was 22 . 2 cm2 ( 1–132 cm2 ) and for chronic wounds 78 . 1 cm2 ( 1–297 cm2 ) ( Fig 7A and 7B ) . Additional severe pathologies identified amongst the IS2404 positive skin ulcers were osteomyelitis ( n = 2 ) and malignant ulcer ( n = 2 ) ( see Fig 6A ) . These ulcers were excluded from the wound healing analysis because the additional severe pathologies found preclude healing ( see Fig 1 ) . Other complications are shown in Fig 6B and 6C . Fig 8 contrasts an uncomplicated IS2404 PCR positive ulcer of the left ankle in a PHC center patient with a non-healing IS2404 PCR positive ulcer of the right leg in an SHC center patient . Figs 5 and 9 illustrate the history and evolution of IS2404 PCR positive ulcers during the observation period . At the PHC center , 66 . 7% ( 16/24 ) of IS2404 PCR positive ulcers with a pre-enrolment duration of ≤6 months closed within a post-enrolment observation period of ≤3 months . In contrast , only 16 . 6% ( 1/6 ) of IS2404 PCR positive ulcers with a pre-enrolment duration of >6 months closed within a post-enrolment observation period of ≤3 months . Two of four patients with chronic IS2404 PCR positive ulcers which did not close whilst under treatment at the PHC center during the study period were lost to follow-up ( Fig 5 ) . The SHC center patients had observed their IS2404 PCR positive ulcers on average 33 ( range 3–156 ) months before visiting a health facility . At enrolment the patients were already hospitalized on the BU ward or had been treated as outpatients on average for 7 months ( range 1–49 ) . Fifty percent ( 5/10 ) of the IS2404 PCR positive ulcers with a pre-enrolment duration of ≤16 months closed within a post-enrolment observation period of ≤3 months . In contrast , only 20% ( 2/10 ) of IS2404 PCR positive ulcers with a pre-enrolment duration of >16 months closed within a post-enrolment observation period of ≤3 months . Three patients with chronic IS2404 PCR positive ulcers were lost to follow-up after an average observation time of 12 days . ( Fig 9 ) . At the SHC center 50% of all IS2404 PCR positive ulcers showed evidence of infection at least once during the observation period as compared to 9% at the PHC center . Fig 10 shows a typical case of wound infection with unhealthy , pale-looking granulation tissue , creamish discharge and slightly swollen adjacent skin . At the PHC center 54 . 8% ( 17/31 ) of patients with IS2404 positive ulcers experienced pain related to their wounds at least once during the observation period . In 52 . 9% ( 9/17 ) of these patients , the pain was intermittent , usually during wound dressing in 55 . 6% ( 5/9 ) while 47 . 1% ( 8/17 ) had constant pain . Of all patients with wound pain , 23 . 5% ( 4/17 ) described it as mild , 52 . 9% ( 9/17 ) as moderate and 23 . 5% ( 4/17 ) as severe . Of the pain associated with wound dressing , 40% ( 2/5 ) was mild , 20% ( 1/5 ) moderate and 40% ( 2/5 ) severe . None of the patients with wound dressing associated severe pain received analgesics , while 52 . 9% ( 9/17 ) of all patients who experienced pain used analgesics at some point during treatment ( see Fig 11 ) . The analgesics used were paracetamol , diclofenac and ibuprofen , all of which were not prescribed . At the SHC center , 59 . 1% ( 13/22 ) of all patients with IS2404 positive ulcers ( including those with chronic ulcers ) enrolled in the study complained about pain related to their wounds . The pain was localized at the wound itself and at the surrounding tissues in ( 53 . 8% , 7/13 ) or on the entire affected limb . Thirty-eight percent ( 5/13 ) of patients had wound dressing associated pain but none of the patients received analgesics prior to wound dressing . 53 . 8% ( 7/13 ) of all patients who complained about pain received analgesic therapy at some point ( see Fig 11 ) . The most frequently prescribed analgesics were paracetamol , diclofenac or a combination of the two . At the PHC center the facility had one treatment room where all wounds were dressed . These wounds included minor burns , traumatic ulcers , minor surgical wounds and Buruli ulcers . As a measure to prevent possible transmission of M . ulcerans to other wounds , BU wounds were treated only after all the non-BU wounds had been dressed . Large and complicated wounds were referred to the closest secondary health-care facility in the municipality . Availability and quality of dressing materials was limited and the observation of wound dressing techniques revealed some shortcomings: At the SHC center , separation of out- and in-patients , patients with contaminated , bacterially infected and non-contaminated wounds and wound management following standardised protocols were not fully installed and inconsistent . In the expert interviews , all experts identified hygiene and wound care deficiencies as a major cause of deterioration of wounds , in addition to a lack of identifying complicating underlying conditions . The wound management recommendations by the local experts were oriented closely at wound care principles of WHO [7] . The international expert highlighted additionally the problem of recognition of arrested wound healing and the lack of progression to active wound management , such as refreshing wound margins . Of the 81 IS2404 PCR negative skin ulcers of patients recaptured at the PHC center , 4 had additional severe pathologies , chronic osteomyelitis ( n = 2 ) ( see Fig 12a1 and 12a2 ) , suspected malignant ulcer ( n = 1 ) ( see Fig 12B ) , and suspected ossified fibroma ( n = 1 ) . They were excluded from the wound healing analysis because the additional severe pathologies found preclude healing ( see Fig 1 ) . Of 77 ulcers included in the analysis , 76 ( 98 . 7% ) were either completely closed ( 86 . 8% ) or almost closed ( 13 . 2% ) . The remaining 1 non-healing wound was infected ( confirmed microbiologically as Staphylococcus aureus ) ( see Fig 12C ) . All open IS2404 PCR negative wounds remained negative on repeat PCR ( recapture wounds–see S1 Wound documentation Recapture Study ) .
This is the first comprehensive prospective observational study on wound healing in BU patients with IS2404 PCR positive ulcers which compares a primary and secondary health care center . Additionally , patients with ulcers clinically diagnosed as BU at the PHC level but not confirmed by IS2404 PCR were recaptured in the community to verify the correctness of the classification as non-BU cases , to assess the course of wound healing , to make a final diagnosis and to provide treatment , if wounds persisted . At both the PHC and SHC levels , there were two main categories of IS2404 PCR positive BU ulcers . One group healed promptly , progressively reducing in size and closing completely within a maximum of 12 weeks after enrolment , whereas the other group was largely unresponsive to the wound management applied . The latter were chronic ulcers already persistent over various lengths of time before enrolment . The proportion of early versus longstanding , often chronic ulcers differed substantially between the PHC and SHC levels . At the PHC level most ( 52% ) of the IS2404 PCR positive ulcers were WHO Category I ulcers , whereas at the SHC level the majority ( 82% ) were Category III ulcers . As a consequence , the majority of the ulcers at the PHC level closed without the need for surgical intervention ( 86 . 7% ) compared to only 40% at the SHC level where the majority required split-skin grafting ( 75% ) or excision ( 12 . 5% ) . This shows clearly that ulcers which have become large or chronic run a very high risk of requiring invasive and expensive therapy to achieve wound closure . Assessment of healing progress is important to prevent overlooking the transition of ulcers into chronic forms which need active interventions to prevent longstanding arrest and further deterioration . Monitoring of wound surface area is a simple and reliable method . Flanagan has suggested on the basis of an extensive literature review that a reduction of less than 20 to 40% within 2 to 4 weeks after initiation of treatment is a strong indication of failing wound healing [16] . This criterion has been incorporated into the WHO guidelines [7] and we have applied it here to our patients ( see Figs 5 and 9 ) . Three out of four of non-healing wounds in the PHC level cohort , and 90% ( 9/10 ) at the SHC level were reliably predicted to fail to respond to treatment at the recommended assessment time points , and therefore fulfilled Flanagan’s criteria . Monitoring reduction in wound area as a sign of healing , faces difficulties during anti-mycobacterial therapy . Ulcers may enlarge during therapy , because the skin covering areas of necrotic subcutaneous tissue may break in and tissue debris may slough off [18] . Furthermore , immune reconstitution-associated paradoxical reactions [10 , 19] may contribute to an enlargement of the wounds , leading to an inflammatory phase preceding the wound healing phase . These causes of enlargement of wounds or healing delay often cannot easily be distinguished from persisting M . ulcerans infection or deterioration caused by secondary infection with other pathogens [9 , 10 , 19] . In this study , persistence of M . ulcerans infection after completion of antibiotic treatment was not observed . All IS2404 positive wounds which failed to close had underlying pathologies or causes in addition to BU . At the PHC level these were chronic osteomyelitis , chronic lymphedema , squamous cell carcinoma and wound infection and at the SHC level , venous and arterial insufficiency , malignant deterioration , nutritional deficiency and lymphedema . Other reasons for impaired wound healing at the SHC level were wound care and hygiene insufficiencies and repeated wound infections . Only 9% of wounds at the PHC level and as much as 50% at the SHC level were clinically infected . This significant difference may be attributable to differences in the wound spectrum cared for as well as the greater risk of acquiring nosocomial infections at the secondary level of the health care system ( with in-patient facilities ) compared to the primary level . Since all but two of the patients at the SHC level where managed as in-patients , the majority of them had intense daily contact with the hospital environment , in contrast to the PHC level patients who were all managed on an out-patient basis . At the SHC level there was no adequate spatial separation of patients with clean post-surgical wounds from those yet to have surgery and new patients with active ulcers . Thus , patients were under high risk of cross-contamination of wounds resulting in difficult-to-treat wound infections . When critical deficiencies were observed , the responsible persons of the BU control program and the health services were informed and corrective measures taken , e . g . supply of dressing materials and instruments and discussion with the authorities involved on the need to ensure regular supply of these materials; continuous training of nurses on wound bed preparation techniques and the need to be more sensitive to patient-centered concerns and its impact on wound healing . In summary , many factors account for failed wound healing . Patient-dependent reasons include underlying conditions interfering with wound healing , such as diabetes mellitus and arterial hypertension , malnutrition , and lack of compliance . On the health provider side it includes infrastructural problems , availability of dressing material , clean water for cleansing of wounds and lack of training and motivation of health staff , which put patients at risk of impaired wound healing and deterioration of wounds , with repeated secondary bacterial infections as the main driving force . We observed various factors facilitating better wound healing and early wound closure at the community level . For obvious reasons , patients tend to present early to community health centers and wounds are thus smaller on presentation compared to patients at the secondary health care level . Successful wound management is achieved with less effort for both the health care system and the patients and at much lower cost . Nosocomial infections can be more easily avoided . Additionally , patients continue to live in their communities which positively impacts on their nutritional status . We observed only 6% underweight patients at the PHC level as compared to 36% at the SHC level . BU wounds are traditionally regarded to be typically painless unless there is superimposed bacterial infection [3 , 4] . In our study , however , the majority of patients ( 54 . 8% at the PHC level and 59 . 1% at the SHC level ) experienced wound pain at various points in time , mostly severe and associated with wound dressing ( 55 . 6% at the PHC level and 40% at the SHC level ) . Alferink et al . and de Zeeuw et al . made a similar observation with nearly 30% of BU patients experiencing severe procedural pain [20 , 21] . Adequate preparation and planning of procedures such as wound dressing are key to pain prevention [7] . Pain negatively affects wound healing and severely impairs daily life [22] . However , only around half of the patients at both health care levels had access to analgesics . None of the patients received analgesics prior to wound dressing at both health facilities , and the majority of analgesics used by patients were not prescribed . This is avoidable sufferance and a significant deviation from the WHO wound management guidelines [23] . In vertical disease-specific control programs and clinical trials , patients who do not belong to the target group receive little , if any attention . BU is a particularly difficult disease in this respect since many more patients with wounds are clinically suspected to be BU than finally laboratory confirmed . This carries risks in two directions . False negatives do not get the indicated anti-mycobacterial treatment and true negatives may have other significant pathologies in need of specific treatment or require general wound management to prevent transition into chronic ulcers and systemic complications . We therefore re-captured at the PHC level all patients who went back into the community during the study period after clinically suspected BU had not been confirmed by IS2404 PCR . The aim was to assess the course of wound healing in these patients , to identify patients who have been misdiagnosed and to treat patients with persisting ulcers . Ninety-nine percent ( 76/77 ) of the IS2404 PCR negative ulcers were either completely closed ( 86 . 8% ) or almost closed ( 13 . 2% ) . This can be explained by the fact that the majority of these patients ( 60/80; 75% ) were children in the 6–15 year group . This is a physically very active age group , and their wounds were likely of traumatic origin . Very importantly , however , 5% of the recaptured patients were found to have missed diagnoses of therapeutic significance ( chronic osteomyelitis , wound infection , ossified fibroma and suspected malignant ulcer ) . All ulcers which were still open at the time of recapture remained negative in repeated IS2404 PCR analysis .
After completion of specific treatment , Buruli ulcers often require adequate general wound management over long periods to achieve wound closure . Even where attempts are made to follow WHO recommended standards of care , deficits of health care services and delays in the recognition of complications are continuing bottlenecks to satisfactory outcomes . This study indicates that with basic infrastructure , equipment and supplies at appropriate quality standards , well-trained health staff and adherence to wound management guidelines , most wounds can be adequately treated at the PHC level , where patients tend to report earlier , stay closer to their families , can maintain their means of livelihood and are less prone to nosocomial wound infections compared to in-patient facilities . Determining reduction in wound area after 2 to 4 weeks ( Flanagan’s criteria ) of treatment is useful to predict non-healing wounds . Recapture of patients with clinically BU suspicious , PCR negative wounds , clearly showed that even though the majority had healed without sequelae , significant pathology remains unattended if not carefully followed-up . Patient centered care needs to change from vertical to horizontal wound management and facility and training related issues need to be urgently addressed . | Buruli ulcer ( BU ) , currently endemic in more than 30 countries , particularly in West and Central Africa , causes chronic necrotising disease of the skin and subcutaneous soft tissue . The task is twofold , specific therapy directed against the causative agent Mycobacterium ulcerans and general wound management to get the often very large skin defects closed . Irrespective of wound etiology , healing requires favorable systemic conditions such as balanced nutrition , protection from trauma and infection , moist wound environment , and pain control . The authors compared wound healing in BU patients and the infrastructure and wound care practices at two levels of the healthcare system in Ghana . Our results indicate that with the basic infrastructure at the primary health care level , equipment and supplies at appropriate quality standards , well-trained health staff and adherence to established Buruli ulcer treatment and wound management guidelines , most wounds can be adequately managed . We further determined the outcome of clinically BU suspicious but not laboratory confirmed wounds and found that careful follow-up is needed to not miss important diagnoses requiring specific therapy . Patient centered care needs a horizontal approach to wound management . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion",
"Conclusion"
] | [
"rheumatology",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"drugs",
"tropical",
"diseases",
"osteomyelitis",
"physiological",
"processes",
"bacterial",
"diseases",
"signs",
"and",
"symptoms",
"ulcers",
"analgesics",
"neglected",... | 2017 | Assessing and managing wounds of Buruli ulcer patients at the primary and secondary health care levels in Ghana |
Wild mammals serve as reservoirs for a variety of microbes and play an important role in the enzootic cycles of these microbes . Some of them are vector-borne bacteria in the genera Anaplasma , Ehrlichia and Rickettsia of the order Rickettsiales , which can cause febrile illnesses in human beings as well as animals . Anaplasma spp . , Ehrlichia spp . and many spotted fever group ( SFG ) Rickettsia spp . are transmitted to mammalian hosts by tick vectors during blood meals . As a powerful sequencing method , the next generation sequencing can reveal the complexity of bacterial communities in humans and animals . Compared with limited studies on blood microbiota , however , much fewer studies have been carried out on spleen microbiota , which is very scarce in wild mammals . Chongming Island is the third biggest island in China . It was unclear whether there were any vector-borne bacteria in Chongming Island . In the present study , we explored the bacterial microbiota in the spleens of wild mice and shrews from the rural areas of Chongming Island and investigated the prevalence of vector-borne bacteria . Genomic DNAs were extracted from the spleen samples of 35 mice and shrews . The 16S rDNA V3-V4 regions of the DNA extracts were amplified by PCR and subjected to the 16S rDNA-targeted metagenomic sequencing on an Illumina MiSeq platform . All the 35 spleen samples obtained data with sufficient coverage ( 99 . 7–99 . 9% ) for analysis . More than 1 , 300 , 000 sequences were obtained after quality control and classified into a total of 1 , 967 operational taxonomic units ( OTUs ) clustered at 97% similarity . The two most abundant bacterial phyla were Firmicutes and Proteobacteria according to the analysis of rarefied sequences . Among the bacterial communities detected in this study , Anaplasma , Rickettsia and Coxiella were adjacently clustered by hierarchical analysis . Significant differences in many bacterial features between Anaplasma-positive and Anaplasma-negative samples were identified by LEfSe analysis and Wilcoxon rank-sum test , suggesting that the Anaplasma-infection of small wild mammals was associated with a specific pattern of spleen microbiota . Our study has comprehensively characterized the complex bacterial profiles in the spleens of wild mice and shrews from Chongming Island , Shanghai city . This work has revealed distinct spleen bacterial communities associated with tick-borne bacteria in wild animals . The detection of tick-borne bacteria highlights the risk of contracting pathogens with public health importance upon tick-exposure in the studied areas .
Wild mammals serve as reservoirs for a variety of microbes and play an important role in the enzootic cycles of these microbes . Some of them are vector-borne bacteria in the genera Anaplasma , Ehrlichia and Rickettsia of the order Rickettsiales . Anaplasma spp . , Ehrlichia spp . and many spotted fever group ( SFG ) Rickettsia spp . are transmitted to mammalian hosts by tick vectors . They are obligate intracellular bacteria , and their main target cells are white blood cells , erythrocytes , platelets and/or vascular endothelia [1–3] . These bacteria have evolved adapted strategies to evade and/or suppress host protective immune responses and can cause febrile illnesses in animals and/or humans [1–3] . They have been gradually recognized as emerging pathogens of public health importance around the world [3–7] . The prevalence of these tick-borne bacteria has been increasingly reported in China . For instance , Anplasma phagocytophilum , Anaplasma bovis , Anaplasma ovis , Anaplasma central , Anaplasma marginale , Anaplasma platys , Anaplasma capra , Ehrlichia chaffeensis , Ehrlichia canis , Candidatus Neoehrlichia mikurensis , Rickettsia heilongjiangiensis , Rickettsia sibirica , Rickettsia raoultii and Rickettsia conorii have been detected in ticks , animals or humans in many provinces of China [4 , 8–19] . However , the existence of these bacteria in Shanghai city , China is still unknown . Mammals are ecosystems that are inhabited by niche-specific microbiota including bacteria , viruses and fungi etc . The commensal microbiota plays essential roles in the development of immune system , modulation of metabolism and maintenance of health [20] . The perturbation of symbiotic microbiota has been shown to be associated with various diseases such as infection , immunological disorders , metabolic diseases and cancer etc . [20–22] . It had been thought that the circulatory system was sterile in healthy organisms , and that bacteria were present in the circulation only due to sepsis . Nevertheless , the presence of bacteria in the blood of healthy humans began to be documented several decades ago [23 , 24] . With the advance in sequencing technology , blood microbiota has been gradually uncovered in healthy organisms in the past decade [25 , 26] . The spleen , a peripheral lymphoid organ in vertebrates , acts as a blood filter . It plays an important role in the modulation of immune responses and hematopoiesis [27] . The spleen can be infected by the tick-borne bacteria from the order Rickettsiales [28–30] . During the establishment of intracellular infection in the spleen , these bacteria may have impacts on their host cells and alter the spleen niche . Therefore , we hypothesize that the changed spleen niche due to the infection with tick-borne bacteria would lead to the formation of specific spleen microbiota . As important reservoirs for the tick-borne bacteria , mice also serve as model animals for human infection . The present study explores the spleen microbiota in wild mice and shrews from Chongming Island , Shanghai city , China . The blood microbiota of wild mice from Israel and the spleen microbiota of wild voles from France have been recently reported , respectively , [31 , 32] . Compared with limited studies on blood microbiota , however , much fewer studies have been carried out on spleen microbiota , which is very scarce in wild mammals . Chongming Island is the third biggest island in China . Its major part belongs to an administrative county of Shanghai city . It locates at the mouth of the Yangtze River . There has been no description on tick species present in Chongming Island yet . Rhipicephalus sanguineus and Haemaphysalis longicornis , however , have been reported to be the ticks infesting on pet dogs in other areas of Shanghai [33] . It is unclear whether there are any tick-borne bacteria in Chongming Island . The aim of the present study is to explore the bacterial microbiota in the spleens of wild animals from the rural areas of Chongming Island and investigate the presence of tick-borne bacteria .
Animals were handled in accordance with National Guidelines for Ethic Review of Laboratory Animal Welfare . Animal treatment protocols were approved by the institutional animal ethics committee ( the Animal Ethics Committee of Tongji University School of Medicine , Shanghai , China ) . The Chongming Island has a humid subtropical monsoonal climate and a woodland habitat , which is suitable for tick infestation . Chongming Island has an area of around 1267 km2 . The mouse collection sites mainly covered the middle and west parts of Chongming Island where there was more green coverage . Spring-loaded bar mousetraps with bait were used to trap mice . Traps were strategically placed in the environment such as crop fields , residential houses , bank of rivers and forests where wild mice were seen or expected living or traveling . These traps were set in the evening and checked in the next morning . The latitude and longitude of locations where mice and shrews were trapped were recorded through the Global Positioning System ( GPS ) . Trapped live animals were transported in cardboard containers to the institutional Animal Biosafety Laboratory where they were euthanized by CO2 . Necropsies were conducted after euthanasia . The spleen samples of trapped animals were collected and stored at –80°C . Genomic DNAs were extracted from the spleen samples using the E . Z . N . A . tissue DNA extraction kit ( Omega Bio-tek , Norcross , GA , US ) according to the manufacturer’s protocol . The quality and quantity of extracted DNAs were examined by 1% agarose gel electrophoresis and NanoDrop 2000 spectrophotometer ( Thermo Scientific , MA , US ) . The V3-V4 regions of the 16S rDNA were amplified by PCR in a thermal cycler GeneAmp 9700 ( Applied Biosystems Inc , Foster City , CA , US ) . The PCR condition was 95°C for 3 min , followed by 30 cycles at 95°C for 30 s , 55°C for 30 s , and 72°C for 45 s and a final extension at 72°C for 10 min . The primers used were 338F ( 5’-barcode-ACTCCTACGGGAGGCAGCAG-3’ ) and 806R ( 5’-GGACTACHVGGGTWTCTAAT-3’ ) . The barcode is an eight-base sequence unique to each sample . The PCR reactions were performed in triplicate in 20 μL mixture containing 2 μL of 10 × PCR Buffer , 2 μL of 2 . 5 mM dNTPs , 0 . 8 μL of each primer ( 5 μM ) , 0 . 2 μL of rTaq DNA Polymerase ( TaKaRa Bio , Dalian , China ) , and 10 ng of template DNA . The PCR amplicons were extracted from 2% agarose gels and purified using the AxyPrep DNA gel extraction kit ( Axygen Biosciences , Union City , CA , US ) according to the manufacturer’s instructions . After being quantified using QuantiFluor-ST ( Promega , Madison , WI , US ) , the purified DNAs were pooled in equimolar and paired-end sequenced ( 2 × 300 ) on an Illumina MiSeq platform according to the standard protocols ( Majorbio , Shanghai , China ) . The raw reads were deposited into the NCBI Sequence Read Archive ( SRA ) database ( Accession Number: SRP118742 ) . Raw fastq files were demultiplexed and quality-filtered using QIIME ( version 1 . 9 . 1 ) . The following criteria were met: ( i ) The 300 bp reads were truncated at any site with an average quality score < 20 over a 50 bp sliding window , discarding the truncated reads shorter than 50 bps . ( ii ) Exact barcode matching; maximal 2 nucleotide mismatches in primer matching . ( iii ) Only sequences overlapping longer than 10 bps were assembled according to their overlapped sequence . Reads containing ambiguous characters were removed . Reads that could not be assembled were discarded . Operational taxonomic units ( OTUs ) were clustered with 97% similarity cutoff using UPARSE ( version 7 . 1 http://drive5 . com/uparse/ ) and chimeric sequences were identified and removed using UCHIME . The taxonomy of each 16S rRNA gene sequence was analyzed by RDP Classifier ( http://rdp . cme . msu . edu/ ) against the SILVA ( SSU123 ) 16S rRNA database using confidence threshold of 70% as previously described [34] .
The small wild mammals trapped in the present study included mice and shrews . Thirty five samples submitted for 16S rDNA-targeted metagenomic sequencing were listed in Table 1 . Except the four samples CS6 , CS7 , CS34 and CS88 from shrews , all the other 31 samples were from mice including Apodemus agrarius , Mus musculus and Rattus flavipectus . Shrews belong to the order Eulipotyphla . A . phago: A . phagocytophilum . Abbreviated environmental types: A , agricultural area; F , forest; LB , lake bank; R , residential area; RB , river bank near residential area . The latitudes and longitudes of locations where animals were trapped were provided . In this study , all the 35 spleen samples obtained data with sufficient coverage ( 99 . 7–99 . 9% ) for analysis . A total of 1 , 323 , 308 16S rRNA gene sequences with a read length of 469 bps were identified with an average of 37 , 808 reads per sample . And a total of 1 , 967 OTUs were clustered at 97% similarity across all samples . The number of OTUs per sample ranged from 127–528 . The inverse Simpson’s diversity indices were from 0 . 072 to 0 . 6551 , which indicated a broad variation in the bacterial diversity between samples . Rarefaction ( to 17 , 808 ) resulted in 85–520 OTUs per sample . Firmicutes was the most abundant phylum , and Proteobacteria was the second among the total taxa of 35 samples tested in this study except that the scenario for sample CS97 was inverse ( Fig 1 ) . Firmicutes and Proteobacteria had mean abundances of 71 . 53% ( SD 13 . 48% ) and 22 . 45% ( SD 11 . 85% ) , respectively , in the total taxa of 35 samples ( Fig 1 ) . The sum of their mean abundances accounted for greater than 90% of the total taxa . The next three following bacterial phyla were Bacteroidetes , Actinobacteria and Choloriflexi with a mean abundance of 2 . 33% ( SD 1 . 88% ) , 1 . 91% ( SD 1 . 94% ) and 0 . 47% ( SD 0 . 73% ) , respectively ( Fig 1 ) . At genus level , there were 11 major bacterial taxa with a mean abundance greater than 1% , which included Bacillus , Lactococcus , Peptoclostridium , Pseudomonas , Oceanobacillus , Clostridium_sensu_stricto_1 , Acinetobacter , Psychrobacter , Brochothrix , Bartonella and Anaplasma ( Fig 2 ) . In some samples e . g . , CS39 , CS55 , CS56 , CS57 and CS63 , the relative abundance of Peptoclostridium was exceptionally high , whereas the relative abundances of Bacillus and Lactococcus were quite low . The prevalence of vector-borne bacteria in the tested samples was summarized in Table 1 . Anaplasma ( Anaplasma ovis and/or Anaplasma phagocytophilum ) , Ehrlichia , Rickettsia , Coxiella and Bartonella were detected in 11 , 7 , 21 , 11 and 16 of 35 ( 31 . 43% , 20% , 60% , 31 . 43% and 45 . 7% ) samples , respectively . Anaplasma detected in the present study included two species , A . ovis and A . phagocytophilum . The relative abundance of A . ovis was much higher than that of A . phagocytophilum ( S1 Table ) . Ehrlichia detected in the present study was much less abundant than Anaplasma , Rickettsia , Coxiella or Bartonella ( S1 Table ) . Coxiella detected in the present study consisted of only one member , Coxiella_endosymbiont_of_Rhipicephalus_turanicus ( S1 Table ) . All Anaplasma-positive samples were co-infected with Coxiella and vice versa . All Ehrlichia-positive samples were co-infected with Anaplasma and Coxiella , and all Anaplasma/Coxiella-positive samples were co-infected with Rickettsia but not Bartonella . Anaplasma ( A . ovis and A . phagocytophilum ) , Ehrlichia , Rickettsia , Coxiella and Bartonella were detected in both the male and female animals . And they were all detected in the mouse samples . Except A . phagocytophilum and Ehrlichia , A . ovis , Rickettsia , Coxiella and Bartonella were detected in the shrew samples . The 35 samples were hierarchically clustered against the genera with top 50 relative abundances including Anaplasma , Rickettsia , Coxiella and Bartonella but not Ehrlichia ( Fig 3 ) . Notably , Anaplasma , Rickettsia and Coxiella were adjacently clustered to each other , whereas Bartonella were not adjacent to these three genera . This further reflected the closely related occurrences of Anaplasma , Rickettsia and Coxiella but not Bartonella in the wild mice and shrews . Principal coordinate analyses ( PCoA ) based on the Bray-Curtis metrics ( S2 Table ) were performed to look at the overall differences in the spleen microbiota of the 35 samples considering the factors of animal genders , types , locations or infection with Anaplasma . Anaplasma had a relatively high mean abundance among the tick-borne bacteria detected in the present study . As shown in Fig 4A , two relatively dense groups and one relatively scattered group were observed and circled . One of the two relatively dense groups consisted of 22 samples including 10 male and 12 female animals ( Fig 4A ) . Three of the 22 samples were shrews ( Fig 4B ) . And the geographic sites of these 22 samples covered all the 8 sites in the present study ( Fig 4C ) . All of the 22 samples were Anaplasma-negative ( Fig 4D ) . The other relatively dense group consisted of 8 samples including 7 male and 1 female animals ( Fig 4A ) . One of these 8 samples was shrew ( Fig 4B ) . The geographic sites of the 8 samples were from the 3 sites , JZ , BH and DP ( Fig 4C ) . And all of the 8 samples were Anaplasma-positive ( Fig 4D ) . The 5 samples from the relatively scattered group were all mice and from the 2 sites , BH and HX ( Fig 4B and 4C ) . Three of these 5 samples , CS39 , CS55 and CS57 , were close to each other , and they were all Anaplasma-positive ( Fig 4D ) . One of these three samples was male , and 2 were female ( Fig 4A ) . The remaining 2 samples from the relatively scattered group were more scattered , which were female animals and Anaplasma-negative ( Fig 4A and 4D ) . Compared with the samples in the other relatively dense groups , these 5 samples in the relatively scattered group had exceptionally high percentages of Peptoclostridium and low percentages of Bacillus and Lactococcus ( Fig 2 ) , which contributed greatly to their straying away from the other two groups in the PCoA plots ( Fig 4 ) . To further analyze the microbiota considering the factor of infection with Anaplasma , a hierarchical clustering using unweighted pair group method with arithmetic mean ( UPGMA ) was conducted to compare the microbiota similarities between Anaplasma-positive and Anaplasma-negative samples . The overall 35 samples were divided into two major clusters as shown in Fig 5 . Eight of the 11 Anaplasma-positive samples were in the bigger cluster , and they were sub-clustered into an independent group . The other 3 of the 11 Anaplasma-positive samples were in the smaller cluster , and they were clustered in a consecutively order . Among the spleen microbiota of tested samples , the overall similarities indicated by hierarchical clustering ( Fig 5 ) was consistent with the diversities revealed by the PCoA plot ( Fig 4D ) . A number of differentially abundant bacterial taxa between Anaplasma-positive and Anaplasma-negative samples were identified in the spleen microbiota by the linear discriminant analysis effect size ( LEfSe ) analysis as shown in Fig 6 . The differentially enriched taxa in Anaplasma-positive samples were mainly from the phyla of Proteobacteria and Fusobacteria , whereas the differentially enriched taxa in Anaplasma-negative samples were mainly from the phyla of Actinobacteria , Acidobacteria , Choloroflexi , Nitrospirae and Proteobacteria . Although there was no significant difference in the abundance of overall Proteobacteria phylum , there were significant differences in the class α-Proteobacteria and in some genera from the classes of β and γ- Proteobacteria between Anaplasma-positive and Anaplasma-negative samples . A comparison of the spleen microbiota between Anaplasma-positive and Anaplasma-negative samples to genus level revealed a list of differentially abundant bacterial features with absolute linear discriminant analysis ( LDA ) scores > 2 ( Fig 7 ) , suggesting that the Anaplasma-infection was associated with specific patterns of spleen microbiota from mice and shrews . The features with top 25 absolute LDA scores in Anaplasma-positive samples were Peptostreptococcaceae , Alphaproteobacteria , Rickettsiales , Clostridiaceae_1 , Clostridium_sensu_stricto_1 , Acinetobacter , Anaplasmataceae , Anaplasma , Stenotrophomonas , Dyadobacter , Xanthomonadaceae , Xanthomonadales , Rickettsia , Rickettsiae , Helcococcus , Escherichia_Shigella , Enterococcus , Enterococcaceae , Coxiella , Coxiellaceae , Leginellales , Ehrlichia , Roseateles , Moraxella and Arcobacter ( Fig 7 ) . The tick-borne bacteria , Anaplasma , Rickettsia , Coxiella and Ehrlichia , were all recognized within the top 25 differentially represented features of Anaplasma-positive samples . In contrast , the features with top 25 absolute LDA scores in Anaplasma-negative samples were Bacilli , Bacillales , Bacillaceae , Bacillus , Pseudomonodaceae , Pseudomonoas , Oceanobacillus , Actinobacteria , Actinobacteria , Psychrobacter , Micrococcales , Micrococcaceae , Arthrobacter , Brochothrix , Listericeae , Planococcaceae , Paenirhodobacter , Ruminococcaceae , Lysinibacillus , Aerococcaceae , Myroides , Chloroflexi , Bacteroidales_S24_7_group , Bacteroidales_S24_7_group_g_norank and Flavobacteriia . Bartonella were neither included in the differentially abundant bacterial taxa of Anaplasma-positive samples nor in those of Anaplasma-negative samples ( Fig 7 ) . In addition , a few differentially abundant taxa in the spleen microbiota between male and female animals , mice and shrews or animals from multiple geographic sites at the ranks below phylum were observed by LEfSe analysis , respectively ( S1 Fig ) . Nevertheless , no vector-borne bacteria , Anaplasma , Ehrlichia , Rickettsia , Coxiella or Bartonella , were identified among these differentially enriched taxa . Wilcoxon rank-sum test was used to further compare the relative abundances of taxa between Anaplasma-positive and Anaplasma-negative samples at phylum level . Consistent with LEfSe analysis ( Fig 7 ) , Actinobacteria , Chloroflexi , Acidobacteria and Nitrospirae were significantly more abundant in Anaplasma-negative samples than in Anaplasma-positive samples , whereas Fusobacteria was significantly more abundant in Anaplasma-positive samples than in Anaplasma-negative samples based on the analysis of Wilcoxon rank-sum test ( Fig 8 ) . At genus level , 44 significantly different genera were identified between Anaplasma-positive and Anaplasma-negative samples ( S2 Fig ) by the analysis of Wilcoxon rank-sum test , which was consistent with the result from LEfSe analysis too ( Fig 7 ) . Anaplasma , Rickettsia , Coxiella and Ehrlichia were all significantly more abundant in Anaplasma-positive samples than in Anaplasma-negative samples by the analysis of Wilcoxon rank-sum test ( S2 Fig ) .
In the present study , the two major bacterial phyla among the taxa of tested mouse and shrew samples were Firmicutes and Proteobacteria . Previous reports have shown that microbiota of blood and the spleen from humans or mice were mainly consisted of Proteobacteria and sometimes Firmicutes as well . For instance , the major prevalent bacterial genus in both gerbil rodent blood samples from Israel and mouse spleen samples from France was Bartonella , belonging to the phylum Proteobacteria [31 , 32] . Firmicutes and Proteobacteria were the two major phyla with comparable relative abundances in the blood microbiota of healthy human samples [26] . The predominant phylum , Proteobacteria , represented 90% of the overall microbiota in human blood samples [35] . Greater than 80% of the blood microbiota in 30 healthy blood donors was from the phylum Proteobacteria followed by the phyla of Actinobacteria , Firmicutes and Bacteroidetes [25] . The blood microbiota in nonalcoholic fatty liver disease ( NAFLD ) patients mainly consisted of Proteobacteria ( 87 . 9% ) , which was followed by Actinobacteria ( 7 . 3% ) , Firmicutes ( 3 . 7% ) and Bacteroidetes ( 1 . 1% ) [36] . Although there was variation in the proportions of Proteobacteria and Firmicutes between the spleen microbiota of wild mice and shrews in the present study and the blood and spleen microbiota of humans or mice from the aforementioned reports , the overall blood or spleen microbiota in humans or mice is different from the gut microbiota , which is dominated by Firmicutes and Bacteroidetes [37] . Intriguingly , the present study has shown that the infection of wild mice and shrews with Anaplasma has been associated with a specific spleen microbiota . A number of significantly differentially abundant bacterial taxa between Anaplasma-positive and Anaplasma-negative samples were revealed by both LEfSe analysis and Wilcoxon rank-sum test , respectively . As shown in Fig 5 , the 35 tested samples fell into two major clusters based on the analysis using unweighted pair group method with UPGMA . Eight Anaplasma-positive samples were independently sub-clustered within the bigger major cluster , and 3 Anaplasma-positive samples were adjacently clustered within the smaller major cluster . It seemed that the formation of these two major clusters were resulted from some unknown factors , which were different from the factors analyzed in the present study i . e . , animal genders , types , geographic sites or infection with Anaplasma ( Fig 4 ) . Additionally , it was interesting to notice that there were a few differentially enriched taxa identified in the spleen microbiota from animals with different genders , types or geographic locations at the rank levels below phylum by LEfSe analysis in the present study , suggesting that the factors of animal genders , types or geographic locations had impacts on the spleen microbiota . The present study is the first report on the detection of vector-borne bacteria , Anaplasma , Ehrlichia , Rickettsia , Coxiella and Bartonella , in Chongming Island , which suggests that the wild mice and shrews serve as important animal reservoirs for these vector-borne bacteria in the studied areas . Anaplasma , Ehrlichia , Rickettsia , Coxiella and Bartonella were detected from multiple foci in Chongming Island in the present study , which reflected their relative wide distributions in this island . Among these bacteria , only Ehrlichia was not detected in shrews , which was probably due to the small quantities of shrew samples in the present study . Eight of 11 Anaplasma were male animals , which could be resulted from differences in the ecological behaviors between male and female wild mice and shrews tested in this study . This was less likely due to any potentially intrinsic differences in the spleen niche between male and female animals since Anaplasma , Ehrlichia , Rickettsia , Coxiella or Bartonella were not among the differentially enriched taxa identified in the spleen microbiota of animals with different genders . Furthermore , these bacteria were not among the differentially enriched taxa in the spleen microbiota of mouse versus shrew groups or different geographic site groups either , suggesting that infection with these bacteria was neither specific to mice or shrew nor to a single geographic site in Chongming Island . The closely related occurrences of Anaplasma , Ehrlichia , Rickettsia and Coxiella in the present study suggested that Anaplasma , Ehrlichia , Rickettsia and Coxiella shared the transmission routes in the studied areas . Both of Ehrlichia and Anaplasma belong to the family Anaplasmaceae and are tick-borne bacteria . Many members in the SFG Rickettsia are transmitted by ticks . Coxiella can be transmitted by ticks too . It was very likely that Anaplasma , Ehrlichia , Rickettsia and Coxiella in the co-infected animals in the present study were transmitted by ticks instead of other vectors . Therefore , there was a high chance to get infected with multiple of them upon a tick exposure . A . ovis was the major prevalent Anaplasma sp . in the present study . A . ovis infects ruminants and causes ovine anaplasmosis . A . phagocytophilum , a zoonotic pathogen , can cause anaplasmosis in both humans and animals . Infection with Ehrlichia causes febrile diseases in mammalian hosts . Coxiella and Rickettsia were usually considered as vector-borne pathogens . However , with the advance of molecular biology , some members of Coxiella and Rickettsia are gradually recognized as non-pathogenic intracellular bacteria , which are actually endosymbionts to their hosts [38] . Coxiella_endosymbiont_of_Rhipicephalus_turanicus , also called Coxiella-like endosymbiont ( Coxiella-LE ) , was the only member of Coxiella detected in this study . Coxiella-LE distributes in ticks worldwide [38] . Coxiella and Rickettsia were among the ten maternally inherited bacteria found in ticks summarized by Bonnet , i . e . , Coxiella-LE , Rickettsiella , Arsenophonus , Francisella-LE , Cardinium , Spiroplasma , Lariskella , Midichloria , Rickettsia and Wolbachia [38] . Besides Coxiella and Rickettsia , Rickettsiella were detected in our study , too . Rickettsiella was transferred from the order Rickettsiales to the family Coxiellaceae in the order Legionellales based on the phylogenetic analysis of 16S rRNA sequences [39] . Nevertheless , Rickettsiella were only detected in sample CS57 and had a much less relative abundance in the present study . There may be new tick borne-bacteria present in the differentially abundant bacterial taxa of Anaplasma-positive samples revealed in the present study . In the present study , the infection rate of Rickettsia was 60% , which was the highest among the vector-borne bacteria detected . The Rickettsia-positive samples covered all Anaplasma , Ehrlichia or Coxiella-positive samples but not all Bartonella-positive samples . Rickettsia was prevalent in mice and shrews from all types of environment investigated in the present study , i . e . , forests , agricultural fields , residential areas and banks . In contrast , Anaplasma , Ehrlichia or Coxiella were detected in mice and shrews from the forests and agricultural fields rather than residential areas or banks . Compared with residential areas and banks , the forests and agricultural fields in the studied areas had more green coverage and were more suitable for tick survival . It was unclear whether the Rickettsia spp . from the animals co-infected with Anaplasma , Ehrlichia or Coxiella in the forests and agricultural fields were same as those prevalent in residential areas and banks in the present study . Furthermore , it was unclear whether the vectors transmitting Anaplasma , Ehrlichia , Coxiella or Rickettsia in the forests and agricultural fields were same as those transmitting Rickettsia in residential areas and banks in this study either . Bartonella in the present study were probably transmitted by vectors other than ticks . Bartonella can be transmitted by several arthropod vectors such as fleas , keds , lice , sand flies and ticks , or direct bites by infected animals and often establish persistent infection in asymptomatic mammalian hosts [40] . Bartonella were the most frequently identified bacteria in the fleas collected from southern Indiana , USA [41] . Bartonella together with Mycoplasma were the dominant flea-borne bacteria detected in gerbil rodent blood samples from Israel [31] . Haemotrophic Mycoplasma has different subgroups and been detected in the spleen or blood samples of rodents [42] . In the present study , however , the relative abundance of Mycoplasma was less than 1% . Both Bartonella and Mycoplasma were among the 23 features detected in the rodent blood and/or flea samples summarized by Cohen et . al . [31] . Besides Bartonella , Mycoplasma and Rickettsia , other 14 of these 23 features i . e . , Aquabacterium , Bifidobacterium , Bradyrhizobium , Cyanobacterium ( phylum ) , Halomonas , Lactobacillus , Massilia , Methylobacterium , Neisseria , Ralstonia , Rhizobiales_unclassified ( order ) , Staphylococcus , Streptococcus and Sphingobacteria ( class ) , were detected in the present study too . The remaining 6 of the 23 features , Azovibrio , Catenuloplanes , Diaphorobacter , Saccharothrix , Spirosoma and Wolbachia , were not detected in the present study . Bartonella were also the most prevalent genus in vector-borne bacteria detected in the spleen microbiota of wild voles from France [32] . There were 45 potential zoonotic bacterial genera in total detected by Razzauti M et al [32] . Eleven of the 45 genera , Anaplasma , Bacillus , Bartonella , Clostridium , Coxiella , Escherichia/ Shigella , Moraxella , Rickettsia , Staphylococcus , Stenotrophomonas and Streptococcus , were among the relatively abundant bacterial genera listed in Fig 2 in the present study . Twenty one of the 45 genera , Aeromonas , Burkholderia , Campylobacter , Corynebacterium , Ehrlichia , Enterococcus , Eubacterium , Granulicatella , Haemophilus , Helicobacter , Leptospira , Mannheimia , Micrococcus , Mycobacterium , Mycoplasma , Neisseria , Neochlamydia , Pasteurella , Rhodococcus , Treponema and Vibrio , were detected with relatively low abundance in the present study ( S1 Table ) and hence not listed in Fig 2 . The left 13 genera , Bordetella , Borrelia , Brucella , Francisella , Klebsiella , Legionella , Listeria , Orientia , Salmonella , Ureaplasma and Yersinia were not detected in the present study . The prevalence of Rhodococcus , Legionella , Staphylococcus , Corynebacterium , Streptococcus and Stenotrophomonas , reported to contaminate laboratory reagents [43] , were high in the spleen microbiota of wild voles from France , and the authors suspected that the presence of bacteria in the samples were due to contamination instead of real infection of the animals [32] . In the present study , however , Rhodococcus and Corynebacterium were the genera with relatively low abundance , and Legionella was not detected . And Staphylococcus , Stenotrophomonas and Streptococcus were the 37th , 14th and 20th abundant genera in the present study , respectively . Blank controls were set throughout the 16S metagenomics sequencing in our study , and the detection of these bacteria was very likely due to real infection of wild animals rather than contamination of samples . However , as emphasized by Razzauti et al . [32] , caution should be taken when DNA-based techniques are used to detect microbes . In future , it would be important to investigate the molecular characteristics of vector-borne bacteria prevalent in the studied areas . At the same time , it would be also important to characterize the vectors . These will contribute to the prevention and control of vector-borne bacterial infection in the studied areas . Furthermore , it would be interesting to study the interaction between tick-borne bacteria and their host cells in the spleen . Studies on the mechanism underlying the alteration of the spleen microbiota due to infection with tick-borne bacteria would not only advance the knowledge of the pathogenesis of tick-borne bacteria but also shed light on the function of spleen microbiota from the perspective of infection . | In this study , the 16S rDNA-targeted metagenomic sequencing was used to determine the bacterial community and diversity in the spleens of small wild mammals from China . The 16S rDNAs were amplified from the spleen genomic DNAs of 35 small wild mice and shrews and sequenced by Illumina MiSeq technology . More than 1 , 300 , 000 sequences were obtained after quality control and classified into a total of 1 , 967 operational taxonomic units ( OTUs ) clustered at 97% similarity . The two most abundant bacterial phyla were Firmicutes and Proteobacteria according to the analysis of rarefied sequences . Within the bacterial communities detected in this study , vector-borne bacteria , Anaplasma , Rickettsia and Coxiella , were adjacently clustered by hierarchical analysis . Significant differences in many bacterial features between Anaplasma-positive and Anaplasma-negative samples were observed , suggesting that the infection of small wild mammals with Anaplasma is associated with a distinct pattern of spleen microbiota . This study has revealed the complex bacterial profiles in the spleens of wild mice and shrews . The detection of vector-borne bacteria highlights the role of wild mice and shrews as animal reservoirs with potential public health importance in the studied areas . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"geomorphology",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"microbiome",
"pathology",
"and",
"laboratory",
"medicine",
"landforms",
"shrews",
"spleen",
"pathogens",
"topography",
"microbiology",
"vertebrates",
"rickettsia",
"animals",
"mammals",
"ri... | 2018 | The spleen microbiota of small wild mammals reveals distinct patterns with tick-borne bacteria |
The Epstein-Barr virus ( EBV ) nuclear antigen leader protein ( EBNA-LP ) is the first viral latency-associated protein produced after EBV infection of resting B cells . Its role in B cell transformation is poorly defined , but it has been reported to enhance gene activation by the EBV protein EBNA2 in vitro . We generated EBNA-LP knockout ( LPKO ) EBVs containing a STOP codon within each repeat unit of internal repeat 1 ( IR1 ) . EBNA-LP-mutant EBVs established lymphoblastoid cell lines ( LCLs ) from adult B cells at reduced efficiency , but not from umbilical cord B cells , which died approximately two weeks after infection . Adult B cells only established EBNA-LP-null LCLs with a memory ( CD27+ ) phenotype . Quantitative PCR analysis of virus gene expression after infection identified both an altered ratio of the EBNA genes , and a dramatic reduction in transcript levels of both EBNA2-regulated virus genes ( LMP1 and LMP2 ) and the EBNA2-independent EBER genes in the first 2 weeks . By 30 days post infection , LPKO transcription was the same as wild-type EBV . In contrast , EBNA2-regulated cellular genes were induced efficiently by LPKO viruses . Chromatin immunoprecipitation revealed that EBNA2 and the host transcription factors EBF1 and RBPJ were delayed in their recruitment to all viral latency promoters tested , whereas these same factors were recruited efficiently to several host genes , which exhibited increased EBNA2 recruitment . We conclude that EBNA-LP does not simply co-operate with EBNA2 in activating gene transcription , but rather facilitates the recruitment of several transcription factors to the viral genome , to enable transcription of virus latency genes . Additionally , our findings suggest that EBNA-LP is essential for the survival of EBV-infected naïve B cells .
Epstein-Barr virus is a ubiquitous human herpesvirus that asymptomatically infects the vast majority of the human population , particularly in the developing world , where primary infection typically occurs during the first few years of life , leading to lifelong EBV latency . Where primary infection is delayed into adolescence or adulthood , it can result in the temporarily debilitating but relatively benign condition , infectious mononucleosis . The major disease burden caused by EBV is the range of malignancies with which it has been associated . In particular EBV contributes to high levels of Burkitt lymphoma in sub-Saharan Africa and of nasopharyngeal carcinoma in southeast Asia , as well as around a third of Hodgkin lymphoma cases , approximately one in ten gastric cancers , a range of B cell lymphomas in the immunosuppressed and more rarely with T and NK cell malignancies . Taken together EBV is implicated in around 1-1 . 5% of worldwide cancer incidence [1] . These diverse malignancies likely arise due to defects at different stages of the virus life cycle , or perhaps infection of cell types not involved in the virus’s natural life cycle [2] . The core of the EBV lifecycle occurs with the B cell compartment . EBV infection activates B cells , transforming them into proliferating lymphoblasts . In vitro these continue to proliferate into lymphoblastoid cell lines ( LCLs ) whereas in vivo they can differentiate – probably via a germinal center – into resting memory B cells where the virus is quiescent , producing RNAs but no viral proteins [3 , 4] . LCLs express the ‘growth’ program of EBV genes ( latency state III ) , comprising six EBV nuclear antigens ( EBNAs ) , the latency membrane proteins ( LMP1 , LMP2A and LMP2B ) and a number of EBV encoded RNAs , including the abundant nuclear RNAs EBER1 and EBER2 . The latency III transcriptional program takes over 2 weeks to reach this state [5] . The first EBV proteins detectable after primary infection are EBNA2 and the EBNA leader protein ( EBNA-LP ) [6 , 7] . Shortly after this the EBNA3 proteins also become detectable [6] , and the EBNAs rapidly reach the levels found in LCLs . In contrast , the LMP proteins take up to three weeks to reach LCL-like levels [5] [8] . EBNA transcription is initiated at multiple copies of Wp , the promoter in the major internal repeat ( IR1 ) of EBV . Soon after infection , the burden of EBNA transcription shifts to Cp , a promoter upstream of IR1 . Transcripts from both Cp and Wp are alternatively spliced , and translated in both cap- and IRES-dependent manners to produce the six EBNAs . The functions of most of the EBNAs are reasonably well understood: EBNA1 is important for the replication and segregation of the viral genome during the cell cycle , by binding to oriP . The EBNA1/oriP complex is also important in the switch from Wp to Cp [9] . EBNA2 is essential for the initial transformation of B cells , rapidly activating both host and viral genes through its recruitment to promoters or enhancers alongside cellular transcription factors such as Pu . 1 [10] , RBPJ ( also called CBF1 ) [11] , [12] , IRF4 and EBF1 [13 , 14] . The EBNA3 proteins are slow-acting transcriptional repressors that are important for suppressing senescence and apoptosis around 3 weeks after infection [15 , 16] . Notably , the EBNA3s and EBNA2 appear to have a close relationship with each other , co-regulating genes and being bound at many of the same chromosomal locations [13 , 17–19] . In contrast to the other EBNAs , the role played by EBNA-LP in B cell transformation is not clear . Through initiation at different Wp promoters and exon skipping in Cp transcripts , the EBNA-LP protein comprises a variable copy number of a 66 amino-acid N terminal repeat domain ( encoded by exons W1 and W2 within IR1 ) and a C terminal domain encoded by exons Y1 and Y2 . In LCLs , EBNA-LP mainly localizes to PML nuclear bodies [20] – nuclear foci implicated in repression of virus infection , also called nuclear domain 10 ( ND10 ) [21] – although it takes several days after infection to accumulate there [22] . Functionally , EBNA-LP has been shown to enhance the activation of host and viral genes by EBNA2 after transfection , although not all studies agree on which genes are affected [7 , 23–27] . The complex repetitive nature of the EBNA-LP gene makes its analysis in the viral context challenging . Previous genetic analyses of EBNA-LP have been restricted to mutation of the C-terminal Y exons [28 , 29] . These Y domain knockout viruses establish LCLs at a much reduced efficiency , and then only when the early outgrowth of the cell lines was supported by growth on irradiated fibroblast feeder cells . Deleting IR1 repeat units below five progressively reduced transformation efficiency [30] , but as well as reducing EBNA-LP size , this reduced transcript dosage of EBNA-LP and EBNA2 – and probably of the recently identified stable intronic sequence RNAs ( sisRNA1 and sisRNA2 ) [31] – due to the reduced Wp number producing fewer EBNA transcripts . These prior studies of EBNA-LP function have been conducted in the context of transfecting isolated genes , and/or in the presence of the truncated EBNA-LP protein produced by the P3HR1 virus , and not in the context of virus infection . Therefore , the aim of this project was to produce a complete EBNA-LP knockout virus , and use it to establish the importance of ( and a role for ) EBNA-LP in the transformation of B cells . While our first EBNA-LP knockout was additionally defective due to unintended mutations in the introns between the EBNA-LP exons in IR1 , a second , cleaner knockout showed that EBNA-LP is important but dispensable for the transformation of adult memory B cells , but is essential for the transformation of naïve B cells . Furthermore , both knockouts demonstrated that EBNA-LP is crucial for establishing and stabilizing the viral transcription program after infection , and for facilitating the recruitment of EBNA2 and the host protein EBF1 to the incoming virus genome . However , EBNA-LP did not enhance the induction of host genes by EBNA2 during infection , which had been expected based on prior understanding of its function .
EBNA-LP is transcribed across IR1 ( Fig 1A ) . Each repeat unit contains the Wp promoter from which its transcription can initiate , and repeat units can be skipped by splicing ( Fig 1B ) . Therefore the only approach to reliably knockout EBNA-LP is to introduce a nonsense mutation into the EBNA-LP coding region in each of the IR1 repeat units of EBV . To do this , we generated an array of 6 . 6 IR1 repeat units ( typical for circulating viruses [32] , and matching the size in the parental B95-8 BAC , WTHB9 ) containing a STOP codon mutation ( and restriction site for screening – Fig 1C ) . In order to separate the role of EBNA2 from that of EBNA-LP , an EBNA2 knockout ( E2KO ) EBV – and its revertant , E2rev – were also generated . E2KO retains the entire Y3 exon and its 3’ splice site . This allows qPCR detection of Y2-YH EBNA2 transcripts in the E2KO , despite being deleted for the rest of the EBNA2 ORF ( Fig 1D ) . In order to facilitate comparison with the previous genetic studies of EBNA-LP function in a P3HR1 strain backbone [28 , 29] , we also generated a pair of recombinant viruses ( designated YKO ) that deleted a genome region from within exon Y1 to within exon Y2 . This deleted the Y exon encoded region of EBNA-LP , but retained exon Y1 splice acceptor and exon Y2 splice donor ( Fig 1E ) . A revertant ( Yrev ) was generated for one of these knockouts . Two distinct approaches were used to generate the IR1 repeat arrays used to generate LPKO viruses . The first used type IIS restriction enzymes , a method that necessitated the inclusion of a point mutation in the small intron between exons W1 and W2 ( S1 Fig ) . Duplicate knockouts were made by this method , and a revertant was made of each ( Fig 1F ) . Because of the intronic mutation , these viruses were designated LPKOi and LPrevi respectively [the ‘i' indicating intronic mutation] . The recombinant viruses were checked for unwanted deletions by restriction digest and pulsed field gel electrophoresis ( S2 Fig ) . It was subsequently discovered that the IR1 repeat unit used to clone LPKOi and LPrevi viruses contained three minority variants ( in the BWRF1 putative open reading frame ) , normally found in only one IR1 repeat unit of B95-8 [33] . Thus LPKOi and LPrevi have four nucleotide changes to the intronic sequences in IR1 , relative to the B95-8 consensus sequence . Recombinant virus genomes were transfected into 293 cells and clones selected from which virus could be produced . EBV-BAC DNA was rescued into bacteria from these producer cell lines to ensure that the virus genome was unchanged . The recombinant viruses were initially assessed in BL31 cells , which can be used to assess EBVs regardless of whether they are capable of transforming primary B cells into LCLs [34 , 35] . The mutations did not alter the splicing of EBNA transcripts initiated at either Cp or Wp ( S3A Fig ) , other than the expected shortening of transcripts in YKO cell lines caused by the deletion in the Y1 and Y2 exons . Nor did they reduce the levels of any latency proteins other than those that had been mutated ( S3B Fig , S4 Fig ) . EBNA-LP protein levels were higher in E2KO-infected BL31 , while LPrevi viruses tended to have both larger and more abundant EBNA-LP isoforms . In contrast EBNA-LP levels were very low in YKO-infected cells . These changes in EBNA-LP levels were also seen by immunofluorescence in newly infected primary B cells , which also showed that most of the truncated EBNA-LP in YKO cells was restricted to a nuclear subregion , possibly the nucleolus ( S5 Fig ) . We have recently shown that the B95-8 strain of EBV , and the WTHB9 BAC contain a stop codon at the end of exon W1 in one of its IR1 repeat units , and that this results in the production of less EBNA-LP [33] . This explains the apparent increase in EBNA-LP size and quantity in LPrevi , and may contribute to the reduced protein level in YKO , but does not explain the substantially increased EBNA-LP production by E2KO . Resting adult B cells were infected with the recombinant viruses at equal titres to assess the abilities of viruses to transform B cells into LCLs . Three days post infection , E2KO-infected B cells were indistinguishable from uninfected cells , whereas all other viruses induced enlargement and aggregation of the cells ( S6 Fig ) . After that time , LPKOi showed very poor replication efficiency and usually failed to establish LCLs , whereas YKO consistently established LCLs . In the case where an LPKOi LCL was established on feeder cells and maintained the LPKOi genome ( identified by episome rescue ) , EBNA-LP originating from an alternative EBV strain was detected ( S7 Fig ) , suggesting that LPKOi could be complemented by the virus strain endogenous to the B cell donor , but could not establish an LCL alone . Analysis of cell proliferation in the first week post infection confirmed a previous report that proliferation of EBV-infected cells begins over three days post infection [36] , and showed that LPKOi was very inefficient at driving cellular proliferation , while YKO was better ( S8 Fig ) . However , LPrevi was also poor at driving proliferation , being similar to YKO rather than being like WTHB9 and the other revertants ( S8 Fig ) . Since two independent pairs of LPKOi and LPrevi showed the same phenotype – LPKOi was substantially more impaired than YKO ( which itself produced so little truncated EBNA-LP protein , that it may be functionally a knockout ) and LPrevi exhibited defective transformation of B cells – we suggest that at least one of the intronic mutations in these viruses ( ie the designed point mutation in the short intron , and/or the unintended minor variants in the BWRF1 ORF [33] ) causes this defect . In order to assess the function of EBNA-LP without these confounding intronic IR1 mutations , a second method – Gibson assembly [37] – was used to produce a new recombinants of IR1 , assembling repeat units that matched the B95-8 consensus between the flanking sequences from the B95-8 BAC ( S9A Fig ) . One repeat array was assembled to produce a new wild-type IR1 , and another containing the EBNA-LP mutation shown in Fig 1A . Thus the whole IR1 sequence ( other than the defined EBNA-LP mutation ) in these IR1 constructs matches the published B95-8 sequence ( NC_007605 ) . These repeat arrays were recombined into the IR1-knock-out to make two independent LPKOw BACs ( where ‘w’ indicates wild-type IR1 backbone ) and a repaired wild-type EBV with no IR1 heterogeneity ( WTw ) ( Fig 1F ) . These were validated by pulsed field gel electrophoresis ( S9B Fig ) and used to generate virus-producing cell lines . Thus these new viruses represent a clean EBNA-LP knockout ( LPKOw ) , and a wild-type virus ( WTw ) with a fully intact set of EBNA-LP exons . LPKOw and WTw were used to infect CD19-purified adult B cells alongside E2KO and YKO . Their proliferation after infection was assessed by dilution of the cell trace violet stain applied to the B cells prior to infection . As previously reported [36] EBV-infected cells did not divide until after day 3 post infection ( S10 Fig ) . At 8 days post infection LPKOw is superior to LPKOi in driving infected B cells to undergo proliferation , approaching the level seen for YKO , suggesting that many of the important functions lost in LPKOw are also missing in the YKO . WTw matches ( and perhaps exceeds ) the transforming capability of the parental wild-type BAC and revertants ( Fig 2A ) . LPKOw and WTw were both able to establish LCLs . Latency protein levels were largely similar between the LCLs , with LPKOw LCLs clearly lacking EBNA-LP ( Fig 2B ) , and WTw showing an elevated level of EBNA-LP relative to the parental wild-type ( WTHB9 ) , due to the absence of a W1 mutation in a single IR1 repeat unit ( Fig 2B and [33] ) . LCLs could be reliably established from YKO , LPKOw and LPrevi ( with weekly media changes that did not disturb cell clusters ) : robust expansion of B cell cultures infected by these mutant viruses was usually obvious from around 2 weeks after infection . To quantify this , 5x105 adult B cells from peripheral blood of two donors were infected at an MOI of 1 ( based on Raji green unit titre ) , and twice weekly the clumping cells were dispersed and cells counted ( Fig 2C ) . While donors differed in how rapidly LCLs were established , EBNA-LP-deficient B cells grew out more slowly than the wild-type viruses . There was little or no difference between LPKOw and YKO outgrowth , nor between WTw and WTHB9 or Yrev ( Fig 2C ) , even though WTw expressed considerably more EBNA-LP ( Fig 2D ) . LPrevi outgrowth was more delayed than in previous experiments , and proliferated poorly when at low density , suggesting that its outgrowth was negatively affected by the repeated disruption of cells for counting . Western blotting confirmed the lack of EBNA-LP in LPKOw , the very low level of truncated EBNA-LP produced by YKO and the wider range of EBNA-LP isoforms produced by LPrevi ( Fig 2D ) . In addition to these infections of adult B cells , we also infected mononuclear cells from umbilical cord blood to try to establish LCLs . However , we were reproducibly unable to establish cord blood LCLs with the LPKOw virus , regardless of whether CD19-purified B cells or total PBMCs were used , or if outgrowth was supported with irradiated MRC5 feeder cells . WTHB9 , WTw , and importantly , LPrevi – the severity of its transformation defect being similar to LPKOw – were all consistently able to establish LCLs from cord blood . To investigate what was happening to the infected cells , we compared the cell cycle profile of B cells from umbilical cord blood with adult blood after LPKOw and WTw infection ( Fig 3A ) . While more cell death is evident in LPKOw than WTw-infected cells for both adult and cord blood ( indicated by the subG1 population of cells ) the difference is much more stark in cord cells , with both more dead cells and – by day 11 – far fewer cells in S or G2 phases of the cell cycle . By approximately 14 days post infection ( precise timing varied with different donors ) , there were not enough live LPKOw-infected cord cells to analyse the cell cycle , and any remaining clumps of cells disintegrated and never recovered . In order to quantify this transformation defect , we conducted a dilution cloning experiment comparing transformation of blood from the umbilical cord with blood taken at the same time from the baby’s mother . This was performed for three donor pairs , and LPKOw and YKO viruses consistently failed to transform cord blood into LCLs , despite successfully transforming the maternal cells into LCLs ( Fig 3B ) . In contrast , both the wild type viruses ( WTw , Yrev and WTHB9 ) and LPrevi showed no difference in transformation efficiency between cord and maternal lymphocytes . We then used cell counting to further characterize the outgrowth efficiency of the viruses in cord blood . Counting only cells that were larger than resting B cells ( ie the activated , LCL-like cells ) it was clear that while wild-type and LPrevi-infected cell numbers increased consistently after infection , YKO and LPKOw infected cell numbers fell from around 10 days post infection ( Fig 3C ) , consistent with the death of EBNA-LP-deficient cells at this time . In addition , WTw-infected cells expanded faster than the other wild-types ( which make less EBNA-LP: Fig 2 and [33] ) . This faster expansion was also seen in the transformation assay . Since EBNA-LP-knockout EBV fails to transform cord blood cells , whereas the similarly deficient LPrevi can , this failure does not represent a technical inability to establish cord LCLs with defective viruses in general . Therefore , we can conclude that EBV-transformed cord blood cells require EBNA-LP for their survival at a specific phase approximately 10–14 days post infection . One key difference between adult B cells and cord-derived B cells is that the former are a heterogeneous mixture of different B cell subsets , whereas cord blood contains exclusively naïve B cells , as in utero the baby’s immune system has not yet encountered any pathogens . We therefore tested whether different adult B cell subsets differed in their susceptibility to transformation by LPKOw . B cells were sorted into subsets according to their CD27 and IgD status ( S11 Fig ) : CD27-positive B cells are either class-switched ( IgD- ) or non-switched ( IgD+ ) memory B cells; naïve B cells are CD27-/IgD+; and the double-negative population ( CD27-/IgD- ) is undefined , but is reported to have memory B cell-like properties [38] . All populations were isolated at 97–100% purity ( S12 Fig ) . These populations ( from six different donors ) were infected with LPKOw and WTw EBV , and cultured to establish LCLs . Both WTw and LPKOw LCLs were established from the two memory B cell subsets of all donors tested , and ( despite small cell numbers ) the double negative population also established LPKOw LCLs . Despite using the same MOIs and culture conditions , adult naïve B cells established WTw LCLs more slowly than memory B cells . Nine attempts ( from six donors ) were made to transform naïve B cells with LPKOw and only one ( LC2 ) did so , and during outgrowth this culture exhibited a substantial loss of cells two weeks post infection . To better quantify the outgrowth , two additional donors were sorted into subsets and infected with different virus mutants at an MOI of approximately 1 ( although cell numbers varied according to the yield of cells of each population–S1 Data ) and counted twice weekly ( Fig 4A ) . For these two donors , LPKOw and YKO were able to establish LCLs from the sorted naïve B cells , although they took considerably longer to grow out than from memory B cells , and exhibited a fall in the number of infected cells in weeks 2–3 before they grew out ( S1 Data ) . We confirmed that these cell lines were not expressing EBNA-LP from a complementing virus ( Fig 4B ) . It has been reported that the IgD status of an LCL matches that of the initially infected cell population [39] . We therefore measured the IgD and CD27 status of LCLs from the outgrowth experiments of total B cells and the different B cell subsets ( Figs 5A and S13 ) . The majority of the LCL cells established from mixed adult B cells were CD27-positive , reflecting the greater speed of transformation of memory B cells observed in Fig 4A . Wild-type LCLs established from naïve and unswitched memory cell subsets–expected to express IgD–were typically only two thirds IgD-positive suggesting that this mark is only modestly preserved in LCLs . In contrast , over 90% of cells from memory B cell-derived LCLs were CD27-positive . The profiles of switched and unswitched memory B cell-derived LCLs were not different between EBNA-LP mutant and wild-type viruses , retaining the cell surface phenotype of the parental subset . In contrast all of the LPKOw and YKO LCLs derived from naïve B cells had memory B cell profiles , having a consistently higher CD27 status than wild-type naïve LCLs ( Fig 5B ) , although all LCLs have a much higher rate of CD27+ve cells than would be predicted by their starting profile . Similarly , YKO and LPKOw LCLs from total B cells have a fewer IgD-positive cells than wild-type LCLs ( Fig 5C ) , suggesting a preferential transformation of memory B cells by these mutants . Overall , the low outgrowth efficiency of naïve B cells infected with EBNA-LP mutants , and the memory-skewed profiles of EBNA-LP-deficient LCLs suggest that EBNA-LP is important for the survival of naïve B cells , a phenotype most robustly observable in cord blood LCLs . It has been widely reported that cotransfection of EBNA-LP is able to enhance the transcription of viral and host genes induced by EBNA2 [7 , 23–27] . We therefore used qPCR to analyse host and viral transcript levels in two independent time courses across the first 30 days after infection of CD19 isolated mixed adult B cells at an MOI of 2 . Comparing the EBNA-LP mutant viruses ( LPKOi , LPKOw and YKO ) with the wild-type viruses ( WTHB9 , WTw , E2rev , Yrev and LPrevi ) highlighted a number of differences between the groups that were consistent between the two time courses ( S3 Data ) . EBNA2 transcript levels were very similar across all infections ( Fig 6A ) , except for E2KO , which had a 10-fold higher level transcript in both the EBNA2 and Wp assays ( S3 Data ) . Since EBNA2 levels were transcriptionally constant between EBNA-LP mutant and wild-type viruses ( and by immunofluorescence EBNA2 protein levels were also no different–Fig 3C ) , we assessed the levels of host genes that are regulated by EBNA2 . Two days post infection , known EBNA2-induced genes CD21 and HES1 [27 , 40] were not induced by E2KO , but were induced normally by the EBNA-LP mutants ( Fig 6B and 6C ) , suggesting that EBNA-LP does not enhance the transcription of these genes by EBNA2 in the context of EBV infection . By day 9 , transcription levels were higher in the EBNA-LP mutants than the wild-types – the opposite of what would be predicted from the transfection-based studies – for CD21 and HES1 , and also for IL7 ( Fig 6D ) , which is not induced by EBNA2 upon EBV infection , but does bind EBNA2 at its promoter [14] , suggesting that in this context , EBNA-LP may restrain EBNA2’s transactivation ability . The cyclin D2 gene ( CCND2 ) , was the first EBNA2 target reported to be enhanced by EBNA-LP [7] , but unlike the previous examples is indirectly activated by EBNA2 [41] . Its transcript levels may be higher in wild-type infections than EBNA-LP mutants ( Fig 6E ) , but this effect is inconsistent between replicates ( S2 Data ) . Any apparent effect of EBNA-LP on CCND2 could alternatively be explained by the differences in proliferation induced by EBNA-LP mutants and wild-type EBVs , which could independently affect cyclin transcription . MYC activation by EBNA2 ( which occurs both directly and indirectly [41] ) is clearly unaffected by the absence of EBNA-LP ( Fig 6F ) . Having failed to support the idea that EBNA-LP enhances the transactivation potential of EBNA2 on host genes , we then investigated the expression of virus genes in these time courses . In keeping with the EBNA2 enhancement hypothesis , transcription of the LMP genes ( LMP1 , LMP2A and LMP2B ) was delayed in EBNA-LP mutant EBV infections , reaching wild-type expression levels in the second week post infection ( Fig 7A–7C ) . However , EBNA3 transcripts were also lower in abundance from EBNA-LP mutant EBVs ( Fig 7D–7F ) . Since EBNA3s and EBNA2 ( which was unaffected by EBNA-LP ) are transcribed from the same Wp and Cp promoters ( see Fig 1B ) , we measured promoter usage , and observed an inconsistent tendency to have slightly more Wp and less Cp transcription in EBNA-LP mutants ( S14 Fig and S3 Data ) . Most surprising was the failure of EBNA-LP mutants to promptly induce transcription of the EBER genes , which are not EBNA2 dependent ( Fig 7G and 7H ) . Overall , we found that mutation of EBNA-LP led to a delayed transcription of all viral latency genes ( other than EBNA2 ) , regardless of whether they were EBNA2-dependent , but did not reduce the induction of host genes by EBNA2 . EBNA-LP has reportedly been detected by chromatin immunoprecipitation ( ChIP ) at various genomic loci , often in the presence of EBNA2 [42] . We have attempted to perform EBNA-LP ChIP , but have been unable to detect any difference in EBNA-LP ChIP-qPCR signal at either of the LMP promoters between wild-type and EBNA-LP knockout viruses in LCLs or during primary infections ( S4 Data ) . Since EBNA2 has been repeatedly shown to bind to and regulate these genes , the binding of EBNA2 to both known binding sites and negative control sites ( Fig 8A ) was assessed across three 30 day infection time courses . The EBNA2 binding signal was low on day 2 , so differences in EBNA2 binding were only sometimes detectable ( S4 Data ) . By day 5 there is very little recruitment of EBNA2 to known transcription factor binding sites at both LMP promoters on the LPKOw genome compared to WTw ( Fig 8B ) . EBNA2 recruitment to Cp was modestly reduced in LPKOw infections , but still showed a considerable binding signal . In contrast , recruitment of EBNA2 to host genes IL7 and HES1 was more efficient after the LPKOw infection , consistently showing elevated EBNA2 binding on days 2 and 5 , but not at later time points ( Fig 8C ) . The reduced binding of EBNA2 to the LPKOw promoters was seen consistently for the first 2 weeks after infection ( Fig 8D ) . The reduced detection of viral DNA in LPKOw ChIPs is not due to there being less viral DNA in the LPKOw cells , as the ChIP measures recovery as a percentage of input DNA , thereby controlling for variable input DNA levels . Thus these changes in EBNA2 binding are consistent with the lower expression of viral genes during the same period ( Fig 7 ) , and the elevated levels of host gene transcripts on day 9 ( Fig 6B–6D ) . EBNA2 does not bind directly to DNA , but rather it is tethered to DNA by host proteins , in particular the transcription factor RBPJ ( also called RBP-Jκ and CBF1 ) . We therefore also assessed RBPJ binding to the viral and host loci . RBPJ binding was slightly ( but consistently ) lower in LPKOw at day 5 post infection at the LMP promoters but identical at Cp and host locus IL7 ( Fig 9A ) . No differences were observed on host genes at any time point ( Fig 9B ) , or from 9 days on for virus genes ( Fig 9C ) , suggesting that the apparent delay in RBPJ recruitment to the LPKOw genome is slighter than EBNA2 . However , it is notable that the peak of RBPJ recruitment to the viral genome ( 9 days ) is later than the peak of EBNA2 recruitment ( 5 days ) at all loci tested ( Fig 9C compared to Fig 8D ) . This is consistent with a previous report that shows EBNA2 arriving at loci ahead of RBPJ [14] . Recent genome-wide analyses have shown that EBNA2 and RBPJ are often located with early B cell factor ( EBF1 ) on the genome [42] , and that the three proteins can bind together to chromatin [14] . In addition , EBF1 binds to two RBPJ-independent sites on the EBV genome , one near to the EBERs and the other near oriP . On day 5 post infection , EBF1 showed lower occupancy of all of its sites on the LPKOw genome ( Fig 9D ) , while the virus did not alter EBF1 binding to host genome loci at any time ( Fig 9E ) . In wild-type infections , EBF1 binding reached maximal levels at all EBNA2-bound viral sites between 5 and 9 days post infection ( Fig 9F ) , similar to EBNA2 . Recruitment of EBF1 both to EBNA2-dependent LMP and Cp promoters ( Fig 9F ) and to the EBNA2-independent EBER/oriP loci was delayed in LPKOw infection . It took at least two weeks to approach wild-type levels at all of the viral locations tested , the same dynamics as EBNA2 recruitment ( Fig 8D ) . Overall these gene regulatory observations show a widespread failure of the LPKOw virus to support the activation of viral genes , and the recruitment of both viral and host transcription factors to the EBV genome , including to EBNA2-independent genes . In contrast , EBNA2-bound host genes exhibited a transient elevation of EBNA2 recruitment that was matched by a transient transcriptional activation . Together this suggests that EBNA-LP is required to facilitate the recruitment of transcriptional activators to the EBV genome , particularly between LMP2A and oriP , but is not required for the activation of host genes by EBNA2 .
Our observations ( Figs 3–5 ) suggest that EBNA-LP is far more important for the extended survival and proliferation of naïve B cells than memory B cells . This phenotype is clearest in cord blood , which invariably fail to survive more than two weeks after LPKO infection . For adult cells it is less clear . We have used CD27-negative , IgD-positive as a definition for naïve B cells in adult blood . The EBNA-LP-deficient LCLs established from this naïve population were all CD27+ , and indeed many of the naïve B cell-derived wild-type LCLs contained substantial CD27+ populations . This may simply reflect a technical failure of the sorting process to remove sufficient memory B cells ( although purity was over 98% ) combined with a strong selective growth advantage for memory B cells after transformation . Alternatively , it is possible that in vitro EBV drives adult ( but not cord ) B cells from a naïve to a more memory-like phenotype . Whatever the characteristic of cord blood B cells that makes EBNA-LP essential , this characteristic is shared by most but perhaps not all adult CD27-/IgD+ cells . It may also be that CD27/IgD status does not accurately reflect a cord-like naïve phenotype . Indeed , CD27 is a somewhat flawed marker of B cell memory , since 5–25% of cord blood B cells are typically CD27-positive [44 , 45] despite the neonatal immune system never having met antigen , while CD27-/IgD- cells are reportedly a memory B cell subset [38] . Despite these uncertainties over a robust definition of naïve B cells , it is clear that EBNA-LP is essential for the survival of a naïve or cord-like B cell subset after transformation , but not for the survival of memory B cells . Since LPKO-infected umbilical cord B cells died two weeks after infection of both mixed lymphocyte and CD19-isolated B cells , the difference must be intrinsic to the B cell subsets rather than a bystander cell type . Another example of an EBV mutant exhibiting different transformation phenotypes in different B cell subsets , is shown by a BZLF1-knockout EBV being better able to transform germinal center B cells than memory or naïve B cells [46] . Nevertheless , the difference between naïve and memory cells is still surprising , as ( at the transcriptome level ) they are much more similar to each other than they are to germinal center cells [47] . Biologically , some differences have been reported that separate the behaviour of memory and naïve B cells . We noted a slower outgrowth of LCLs from naïve than from memory B cells , although this was not seen in a previous study [39] . Interestingly , adult naïve B cells enter cell cycle later than memory cells after CD40L stimulation [48] , and produce fewer cells from such cultures [49] . In cord cells the defect is more profound , with CD40 agonism barely inducing any activation markers , whereas equivalent naïve B cell subsets from adults did respond [50] . Additionally , IgM crosslinking in cord cells failed to induce ERK1 phosphorylation , in contrast with adult cells [50] , while BCR crosslinking on adult cells induced a larger response in memory than naïve cells [51] . These observations might be relevant , since the signals from antibody crosslinking and CD40 activation are also invoked by the LMP proteins [52] . LPKO viruses show delayed LMP gene activation , but the inference that the deregulation of LMPs in LPKO may be responsible for the death of naïve B cells is countered by the observation that this death mainly manifests around 10–14 days post infection , by which time LMP transcripts in LPKO infections are approaching wild-type levels ( Fig 7 ) . Other phenotypic differences between naïve and memory cells may also contribute . For instance , IL2 stimulation enhanced the production of memory cells by CD40L , but not naïve cells [48] , while 95% of cord B cells are negative for the IL2 receptor [45 , 50] . Naïve B cells also have a much lower level of the anti-apoptotic BCL2 family members MCL-1 and BCL-XL ( but not BCL2 or Bim ) than memory cells [53 , 54] , which may contribute to the apoptotic phenotype of the LPKOw naïve cells . Together these reports suggest that naïve and memory B cells are phenotypically different , both in their response to pro-proliferative signaling and their resistance to apoptosis . What is less clear is how EBNA-LP overcomes these differences in naïve cells . EBNA-LP can interact with a complex of the tumor suppressors MDM2 , p53 and the cyclin-dependent kinase inhibitor p14ARF [55] , which could influence both proliferation and apoptosis responses . It can also interact with the mitochondrial protein HAX1 [56] , which can be either pro- or anti-apoptotic [57] . Alternatively , metabolic stress has been reported to be an important limitation to B cell transformation , and has been linked to an elevated EBNA-LP:EBNA3 ratio [58] . Furthermore , both EBNA-LP and EBNA3A have been shown to bind the prolyl-hydroxylase proteins that influence HIF1a stability , with the suggestion that this alters the metabolic state of the infected cells [59] . However , further study is required to understand the biology underlying the difference in transformation of naïve and memory cells , and to understand whether these differences are important for the in vivo biology and pathogenesis of EBV . The function of EBNA-LP has long been linked to EBNA2 because of their co-expression immediately after infection , and from a series of co-transfection experiments that appeared to show an enhancement of EBNA2’s transactivation function in the presence of EBNA-LP [7 , 23 , 24 , 26 , 27 , 40 , 60] . These studies demonstrated an ability to enhance transcription from reporter constructs [24 , 60] , from host genes—most notably HES1 and CCND2 ( Cyclin D2 ) [7 , 40]—and from EBV promoters repressed in the latency I transcriptional profile , including LMP1 [23 , 27] , Cp [60] and LMP2A [40] . Some studies have failed to replicate the regulation of some of these genes [27] , but activation of the LMP1/LMP2B bidirectional promoter has been observed consistently . By taking a genetic approach to controlling the presence of EBNA-LP , and by analyzing gene expression in the context of viral infection rather than transfecting isolated EBV proteins , our approach differs considerably from previous studies . In doing so , we have made substantially different observations . First we have not observed any conclusive examples of EBNA2-regulated host genes being enhanced by EBNA-LP . Outside transfection-based reporter assays , the only evidence of a direct relationship between EBNA2 and EBNA-LP comes from the observation that EBNA-LP lacking its acid-rich C-terminus ( but not full-length EBNA-LP ) can bind to EBNA2 [61] , and that ChIP-seq experiments imply overlap between EBNA2 and EBNA-LP binding sites on the genome [42] . Our data are unclear as to whether ( as previously reported [7] ) EBNA-LP enhances the activation of CCND2 ( indirectly regulated by EBNA2 [41] ) , as the modest reduction in its transcription early after infection of EBNA-LP deficient cells could also be due to their reduced proliferation . In contrast , the transient increase in EBNA2 binding to CD21 , IL7 and HES1 in EBNA-LP-deficient infections ( Fig 8B and 8C ) followed by a transient surge in their transcript levels ( Fig 6B–6D ) , could be explained by a direct relationship between EBNA2 and EBNA-LP , but one where EBNA-LP restrains or modulates the transactivation of those genes , rather than enhances it . It is also possible that the increased levels of EBNA2 at the genes may be a consequence of the wider deregulation of the virus genome , either by modestly increasing the EBNA2 protein levels in the cell , or freeing up EBNA2 that would normally bind to the virus , leaving it available to bind to the host genome . Either way , these data suggest that whatever the relationship between EBNA2 and EBNA-LP , it is not simply one of enhancer and co-activator . In contrast to host genes , transcription of both EBNA2-dependent ( LMPs ) and EBNA2-independent ( EBER ) genes is profoundly delayed in the absence of EBNA-LP , and there is also a widespread delay in the recruitment of transcription factors–both viral ( EBNA2 ) and cellular ( EBF1 , and a lesser extent , RBPJ ) –to the LPKOw genome . EBNA transcription is also affected , and while EBNA2 transcripts were not affected by the loss of EBNA-LP , the reduced levels of the EBNA3 transcripts downstream suggest that the processing of the transcripts is different in the LPKO infection , since–like EBNA2 –they are also initiated at the Cp and Wp promoters . A similar elevated ratio of upstream to downstream EBNAs ( as a ratio of EBNA-LP to EBNA3C protein levels ) was observed in cells that have only proliferated 1–3 times after EBV infection [36] , and in cells that arrest after an initial period of proliferation [58] . This could result from either an increase in polyadenylation after EBNA2 , a change in splice site usage , or reduced elongation of transcripts . Indeed , there is evidence that the elongation complex pTEFb is important for transcriptional elongation from Cp , but is predicted to be less important for Wp [62] , leading to speculation that elongation of Wp transcripts is less efficient , which would lead to lower yields of downstream EBNAs . Our data only weakly suggest a greater Wp and lesser Cp usage . Either delayed EBNA2 recruitment to Cp or reduced EBNA1 production ( which might be similar to the EBNA3s , as they are similarly spliced ) and binding to oriP could explain this difference . A more profound effect was seen on the EBV latency genes between LMP2A and oriP ( see schematic in Fig 8A ) . Activation of this whole genome region was severely delayed in EBNA-LP mutant infections , and this correlated with the delayed recruitment of EBF1 , EBNA2 and – albeit less dramatically – RBPJK to the viral genome . The failure to induce transcription of the EBERs demonstrates that EBNA-LP is not simply working through EBNA2 , or indeed through RNA polymerase II , as the EBERs are transcribed by RNA polymerase III , albeit regulated by transcription factors more typically associated with RNA polymerase II [63] . The region of latency genes from the LMP2A promoter to oriP represents a coordinately regulated genomic locus . It is flanked by CTCF binding sites [64] , and these loop together to form a transcriptional unit . Disruption of the CTCF binding site near the LMP2A promoter can disrupt this loop , consequently reducing LMP gene transcription and increasing repressive histone and DNA methylation in LCLs [65] . The simplest interpretation is that EBNA-LP is important for the proper establishment of this transcriptional unit . There is a profound delay in the recruitment of transcription factors , but by 4 weeks post infection , the LPKO LCLs have reached normal levels of LMP and EBER transcripts , so there does not appear to be a defect in the maintenance of the locus once it is established in LPKO cells . EBF1 and RBPJ have been described as a pioneer factors: transcription factors that are able to access chromatinised DNA and establish new enhancer regions [14] , so it is surprising that they would require EBNA-LP it to efficiently access the incoming EBV genome . Notably , this region also includes the terminal repeats , and the virus – linear in the virion – needs to recircularize before LMP2 can be transcribed , and perhaps before this whole region is properly regulated . In addition the terminal repeats contain a binding site for PAX5 , which is directed to the viral genome by EBER2 [66] . Two of the factors reported to bind to the EBER2/PAX5 complex ( NONO and SPFQ ) have also been found to bind to EBNA-LP in a tandem affinity mass spectrometry experiment [67] , although both frequently exhibit non-specific interactions according to the CRAPome repository [68] . Thus it is possible that EBNA-LP is important for the efficient recircularisation of the terminal repeats , or for assembling the EBER2/PAX5 complex on them . Rather than just the LMP/oriP locus , EBNA-LP could be relieving the repression of the viral genome as a whole . EBNA-LP binds to Sp100 , a component of PML nuclear bodies ( or ND10 ) . In so doing , EBNA-LP transiently disperses Sp100 from ND10 [22] , and subsequently localises to ND10 in LCLs [20] . ND10 are known inhibitors of herpesvirus early gene expression and lytic replication [21] . EBNA-LP – along with the EBV tegument protein BNRF1 , which disrupts the ND10 component DAXX – is able to complement an ICP0-null herpes simplex virus , which otherwise is repressed by ND10 [69] . Thus it is reasonable to suggest that EBNA-LP is important for preventing the suppression of latency genes of EBV by ND10 via its interaction with Sp100 , although if this is the case , it is unclear how EBNA2 transcription has evaded this effect . Sp100 has also been implicated in the ability of EBNA-LP to enhance EBNA2-dependent gene transcription in transfection assays , so these phenomena may have a common origin . However , EBNA2 coactivation has also been attributed to EBNA-LP’s ability to bind to HDACs 4 and 5 [60] , or NCOR [40] . EBNA-LP binding to NCOR and HDACs are both reported to sequester these repressive proteins away from EBNA2-inducible genes , thereby improving transactivation [40 , 60] . Any ( or a combination ) of these remain reasonable hypotheses as to how EBNA-LP facilitates viral transcription after infection . In summary , we have undertaken a genetic analysis of EBNA-LP function and shown that EBNA-LP is important for B cell transformation , and essential for the transformation of naïve B cells , and that the role of EBNA-LP is far more complex than the previously proposed cofactor for EBNA2 , being particularly important for establishing the viral transcription program . We also suggest that future analyses of EBV mutants would be better performed in distinct B cell subsets , as it is clear that phenotypes can vary considerably according the differentiation state of the infected B cells , and perhaps also the age of the B cell donor . The observations and genetic manipulation strategies described herein also extend approaches to study EBNA-LP , the EBV-sisRNAs and the wider functions of IR1 in the future .
In order to introduce mutations into IR1 , we have devised a strategy for introducing a constructed IR1 repeat into EBV . This entails first deleting the virus’s endogenous IR1 ( to prevent the constructed repeat from recombining with the original one ) and then inserting the rebuilt repeat . To achieve this , we used RecA-mediated recombineering as previously described [70] . The viral IR1 was deleted by joining together homology regions from the unique ( non-repetitive ) sequences flanking IR1: The upstream region ( NC_007605 positions 11413–12008 ) , which contains exon C2 , was cloned SfiI/PciI from the B95-8 BAC ( clone WTHB9 ) ; the downstream region ( position 35239–35869 ) was cloned XhoI/MluI . This region was introduced by recombineering in place of IR1 . The same homology regions were used as flanks for newly assembled IR1 repeats containing EBNA-LP mutations . We have used two distinct methods to generate the synthetic IR1 . Both approaches generate an IR1 with 6 . 6 copies , which is a typical size for circulating EBV strains [32] and is the size of IR1 in the parental EBV-BAC clone , WTHB9 . In both cases , the IR1 was assembled in a pBR322-based plasmid in DH5alpha bacteria grown at 30°C to reduce unwanted recombination . The first approach used to assemble a modified IR1 adapted a strategy that used type IIs restriction endonucleases to assemble tandem repeats [71–73] . A BamW fragment was subcloned from the B95-8-BAC clone WTHB9 into a vector that contains binding sites for the type IIb restriction endonucleases BsmBI and BtgZI . These restriction sites were engineered to both cut at the site of the BamHI restriction site ( S1B Fig ) . A DNA fragment ( between the MfeI and AgeI restriction sites in BamW ) was synthesized , containing a point mutation of the BsmBI restriction site in the intron between exons W1 and W2 ( S1A Fig ) , and also containing mutations that introduced STOP codons and a PvuI restriction site ( Fig 1C ) , for making the EBNA-LP knockout virus , LPKOi . A second synthesized fragment containing the BsmBI mutation but not the EBNA-LP mutation was also synthesized for producing the revertant virus , LPrevi . These fragments were cloned into the BamW repeat unit , and then both the LPKOi and LPrevi repeat units were assembled into an array using the method described in S1C Fig . The array was then incorporated into independent IR1 knockouts [WKO] according to the scheme shown in Fig 1F , generating two independent LPKOi viruses , and their revertants . Unintentionally , the BamW fragment used to make these viruses contained three sequence differences from the B95-8 consensus that are present as minor variants in B95-8 , found in one of its IR1 repeat units [33] . Subsequent recombinant viruses that contained changes in IR1 were made without the need to mutate the BsmBI restriction site in the W1-W2 intron , and using a BamW fragment that matched the B95-8 consensus sequence . A BamW repeat unit was cloned into a pBR322-based plasmid that contained BtgZI restriction sites that cut the BamHI sites flanking the repeat unit . This was then modified with oligonucleotide linkers on either ( or both ) sides of the BamW fragment , such that the BamW sequence was extended approximately 20bp from the BamHI restriction site ( S9A Fig ) . To generate the wild-type IR1 , the constructs were cut and assembled as shown in S9A Fig , and the assembly was cloned into pKovKan and recombined into the WKO . 4 that had been used to produce LPKOi . 2 , thereby generating WTw . 1 ( see Fig 1F ) . To generate the new EBNA-LP knockout ( LPKOw ) the BsmBI point mutation in the synthesized LPKO region was reverted to the wild-type sequence by InFusion mutagenesis , and subcloned into the new wild-type BamW fragment . The IR1 synthetic array was then assembled in the same way as the wild-type array , and used to independently generate LPKOw . 2 and LPKOw . 4 viruses by recombineering into WKO . 4 . E2KO , E2rev , YKO and Yrev BACs were generated by RecA-mediated recombineering essentially as described elsewhere . The precise sequences of the E2KO and YKO deletions are shown in Fig 1 . Revertants were made by reintroducing wild-type sequence into the knockouts by the same method ( Fig 1F ) . BACs were screened for integrity using EcoRI , AgeI , HindIII , NotI and BamHI restriction digests and run on a CHEF DRII chiller pulsed field gel electrophoresis system ( Bio-Rad ) . We noted that the family of repeats ( FR ) region of oriP is smaller in WTHB9 than predicted by sequence . This reflects a previous observation that the family of repeats region ( FR ) of oriP is unstable , even in BACs , and that the FR in the p2089 BAC ( of which WTHB9 is a subclone ) is 300 bp smaller than the authentic sequence of B95-8 [74] . Therefore , in addition to restriction digests , all recombinant BACs were screened by PCR , using the KA2 and KA3 primers [75] with Q5 DNA polymerase ( NEB ) to ensure that the FR region was the same size in all recombinants . Recombinant EBV BAC DNA was purified from bacteria by alkaline lysis followed by cesium chloride density gradient centrifugation . DNA was assessed by pulsed field gel electrophoresis to ensure a predominance of intact supercoiled BAC DNA , as DNA integrity appears to influence the number and quality of producer cell lines . The BAC DNA was transfected into 293-SL cells ( a culture of the HEK-293 cell line provided by Claire Shannon-Lowe; University of Birmingham ) using a peptide 6 and lipofectin transfection reagent described previously [76] . Cells were selected with hygromycin and colonies isolated by ring cloning . Individual hygromycin-resistant colonies were screened for GFP expression , for their ability to produce virus . The integrity of episomes from the producer lines was assessed by recovery into bacteria [77] and analyzed by restriction digest and pulsed field gel electrophoresis . Cell lines were only used if at least 80% of recovered episomes were indistinguishable from the parental BAC . To generate virus stocks , 293-EBV producer cell lines were seeded in 10 cm dishes and after 1–2 days these were transfected at approximately 25% confluency with equal quantities of BALF4 and BZLF1-expressing plasmids: 12 μg total DNA per 10 cm plate when transfecting with peptide6+lipofectin or 6 μg per plate using GeneJuice reagent ( Merck-Millipore ) . Supernatant was harvested after 5 days and filtered through a 400 nm syringe filter . Virus titer was assayed by infecting 2x105 Raji cells in 1 . 5 ml with 10-fold dilutions of virus . After two days , 0 . 5 ml media supplemented with 20 ng/ml TPA and 5 mM sodium butyrate was added to the Raji cells , and left overnight , to enhance GFP expression in the infected cells . Cell clumps were dispersed by pipetting and total number of green cells per well were counted under a fluorescence microscope . This gave a Raji green units ( rgu ) titer , which was typically in the range of 0 . 5-10x105 rgu/ml in the cell culture supernatant . LCLs , BL31 cells ( provided by Alan Rickinson , University of Birmingham ) , and 293-SL cells were grown in RPMI media supplemented with L-glutamine ( Life Technologies ) and 10% fetal calf serum . Sera were batch tested for the ability to establish 293-SL-EBV-BAC colonies after BAC transfection , and to support outgrowth of LCLs under limiting dilution . MRC5 foreskin fibroblasts ( ATCC CCL-171 ) , also grown in RPMI , were irradiated with 50 Gy and seeded as a confluent monolayer to support outgrowth in some experiments . Adult primary lymphocytes were isolated mainly from buffy-coat residues provided by NHS Blood and Transplant . Cells concentrated from a 500 ml original blood volume were diluted to 200 ml with PBS . Lymphocytes were isolated by layering blood-derivative on ficoll followed by centrifugation . The isolated peripheral blood leukocytes ( PBLs ) were washed twice in RPMI/1%FCS . B cells were purified from PBLs by hybridizing to anti-CD19 microbeads ( Miltenyi ) , using 0 . 5ml beads per 109 PBLs , followed by positive selection ( possel program ) on an autoMACS separator ( Miltenyi ) . Either purified B cells or PBLs were resuspended at 1-2x106 cells/ml in RPMI/15% FCS . B cell purity was measured by FACS for CD20 positivity , and was typically over 90% . For isolation of different adult B cell subsets , the CD19-sorted B cells were rested overnight in a cell culture incubator , and then stained with fluorescent antibodies ( from Biolegend ) against IgD ( PE-CF594 , clone IA6-2 ) and CD27 ( PE-Cy7 , clone M-T271 ) . The cells were sorted using a BD FACSAria III ( BD Biosciences ) into naïve and memory populations based on IgD and CD27 status as indicated in S11 Fig , and purity was assessed by counting a few hundred of the isolated cells on the FACSAria III ( S12 Fig ) . Cell populations were counted and resuspended in RPMI/15% FCS at 2x106 cells/ml for infection . CD27 and IgD status of LCLs was assessed using the same antibodies and staining protocol , measured on an LSRFortessa flow cytometer and analysed using FlowJo software . Isolated PBLs or B cells were infected within a few hours of isolation/purification , by adding virus at an MOI of 1–2 rgu/B cell , and shaking at 37°C for 3 hours , after which cells were centrifuged at 200g for 10 minutes and seeded at a density of 1-2x106 cells/ml in RPMI supplemented with L-glutamine and 15% FCS ( batch tested for LCL outgrowth–GE healthcare or Life Technologies ) and either 50 ng/ml ( for purified B cells ) or 500 ng/ml ( for mixed lymphocytes ) of cyclosporin A . During outgrowth , approximately half of the media volume was replaced every 5–7 days ( cyclosporin A was omitted after two weeks ) , harvesting up to half of the cells , depending on experiment . Blood from the umbilical cord and maternal blood was drawn from healthy full-term pregnancies with written informed consent of the mother ( an adult ) prior to the onset of labour , overseen by the UK National Health Service Research Ethics Committee ( approval REC 13/LP/1712 ) . Mononuclear cells were isolated from paired 0 . 5–2 ml blood samples of maternal and cord blood by ficoll gradient centrifugation . Variations in the yields of mononuclear cells meant that different infections were performed with different numbers of cells: two of the three donors used equal cell numbers for maternal and cord blood infections ( 3 . 4x105 and 1x105 cells per infection ) . The third pair used 5x104 maternal cells , and triplicate infections of 3x105 cord cells for each virus . For most viruses ( LPKOw . 4; WTHB9; WTw; LPrevi; YKO . 4 and Yrev . 4 ) 105 Raji infectious units were used for each dilution series . LPKOw . 2 was used at 106 rgu per dilution series , but this higher titer showed the same transformation efficiency as LPKOw . 4 . Each infection ( and an uninfected control well ) was placed in a well of a 96 well plate , and then serially diluted 2-fold ten times in RPMI/15% FCS/Cyclosporine A ( 100ng/ml ) . Media was changed weekly and after 6 weeks the number ( n ) of wells containing LCLs was counted , and number of transforming events per infection calculated as 2 ( n-1 ) . Quantitation by cell counting was undertaken by infecting B cells ( typically 5x10^5 cells–see S1 Data and S2 Data for exceptions ) at an MOI of 1 in falcon tubes for three hours , and centrifuging and resuspending the cells in in 1ml RPMI containing 15% FCS in a 48 well plate . Twice weekly , cells were dispersed by vigorous pipetting , and 50 ul was taken for counting . Once per week ( immediately prior to a count ) , 400 μl of media was removed , and replaced with 500 μl of fresh media , to feed the cells and ensure that the volume of culture was maintained . For counting , 2 . 5 ul of solution 13 ( Chemometec ) –which contains Acridine Orange to mark all cells , and DAPI to stain dead cells–was added to the 50 μl of cells . Ten μl of cells were added to an NC-slide A8 and cells automatically counted by fluorescence on a nucleocounter NC3000 ( Chemometec ) . Using Nucleoview NC-3000 software ( Chemometec ) , counted events were gated for size to exclude debris smaller than resting cells , and counts plotted both for total cells , and for activated cells only ( defined as those with a larger Acridine Orange area than resting B cells–S1 Data and S2 Data ) . For the time courses after infection of primary B cells , the cells were supplemented with an equal volume of fresh media 24 hours prior to harvesting . Then , half of the culture was taken ( typically 5x105 to 2x106 cells ) and RNA was extracted using RNeasy mini columns ( Qiagen ) . For all samples in a time course , the same quantity of RNA ( ~300ng ) was reverse transcribed using either Superscript III First-Strand Synthesis SuperMix for qRT-PCR ( Life Technologies ) 3 μl cDNA was mixed with TaqMan gene expression mastermix ( Life Technologies ) applied to a custom TaqMan low density array ( TLDA ) card containing duplicate assays ( S1 Table ) , which used ALAS1 , RPLP0 , GNB2L1 and 18S RNA as endogenous control genes . EBV TaqMan assays were designed by Applied Biosystems/Life Technologies using proprietary software , and validated using B95-8 cDNA . Sequence information is proprietary . The assay IDs in S1 Table can be used to obtain these assays . The EBNA2 TaqMan assay extends from exon Y2 to downstream of the exon Y3 splice donor . It therefore also detects EBNA2 transcripts in the E2KO virus , which retains these sequences . The EBNA3 TaqMan assays were designed spanning the exon junction between the U exon and the first exon of each EBNA3 . LMP exon junctions detected by LMP assays are shown in S1 Table . Additional assays ( primers in S2 Table ) were conducted using Kapa qPCR SYBR kit ( low ROX ) , and the IL7 TaqMan assay used Takyon low ROX Probe 2X MasterMix dTTP ( Eurogentec ) and normalised against ALAS1 and RPLP0 . Quantitation of qPCR data was performed using the delta-delta-Ct method , using DataAssist Software v3 . 01 ( Thermo Fisher Scientific ) . All quantitation is expressed relative to the level for WTHB9 on day 2 post infection . Bulk PCR of transcripts across IR1 was performed using Q5 polymerase ( NEB ) and Cp-forward or Wp-forward primers with U-reverse or Y2end-reverse primers ( S2 Table ) . ChIP was carried out using the Chromatin Immunoprecipitation ( ChIP ) Assay Kit ( Millipore ) according to manufacturer’s instructions . Briefly , 2x106 infected B cells were fixed for 10 minutes in 1% formaldehyde and neutralised with glycine . After two PBS washes , cells were lysed with SDS Lysis buffer on ice for 10 minutes and sonicated using the Diagenode UCD-200 Bioruptor for 15 minutes . Precleared chromatin , using 45μl protein A agarose beads was diluted with ChIP dilution buffer and incubated overnight with primary antibodies against EBNA 2 ( Abcam ab90543 ) , EBF1 ( Millipore AB10523 ) , RBPJk ( Abcam ab25949 ) or an IgG control ( Sigma ) . Protein A agarose beads collected the immune complexes , which were subsequently washed in low salt , high salt , lithium chloride and twice in TE buffers . The immune complexes were eluted from the beads using elution buffer and left overnight at 65 degrees . After proteinase K treatment for 2 hours at 50 degrees , DNA was then purified using the Qiagen QIAQuick gel extraction kit , and eluted in 120 μl water . Chromatin was quantified by qPCR using the Kapa qPCR SYBR kit ( low Rox ) on a QuantStudio7 real time PCR machine ( Applied Biosystems ) . Primers used for ChIP have been described previously [78] [79] [14] [80] [62] , and are listed in S3 Table . Absolute quantity ( relative to input ) was calculated from standard curves generated from input DNA that was serially diluted 1:4 , four times . 2 μl of ChIP sample was amplified in triplicate for each qPCR assay . Prior to infection , primary cells were resuspended at 106 cells/ml in PBS containing 5 μM CellTrace Violet ( Life Technologies ) and incubated for 20 min at 37°C in dark . This was then diluted 5 times in complete B cell media and incubated for 5 min at room temperature in the dark . Cells were washed by centrifugation and resuspended in fresh pre-warmed complete B cell media for infection . We noted that CellTrace violet staining had a variable propensity to kill primary B cells , so individual tubes were tested for toxicity by staining PBLs and comparing B cell percentage with and without staining . Tubes exhibiting less than 50% loss of B cells were used in experiments . For assay , cells ( a volume equivalent to 106 cells in the initial infection ) were harvested on ice and stained for CD20-PEVio770 ( Life Technologies ) , and resuspended in PBS/1%BSA containing DRAQ7 live/dead cell stain ( BioStatus ) . Cells were analysed on a FACS machine ( BD LSR II or LSRFortessa ) and cell proliferation visualised for live CD20+ singlet cells using FlowJo software . Approximately 106 infected B cells were resuspended in 50 μl PBS and added to 450 μl of ice cold 70% ethanol and stored until all samples had been harvested . After least 12 hours in this ethanol , cells were pelleted by centrifugation at 500g for 5 minutes , resuspended in 1 ml PBS , incubated for 1 minute , pelleted , and resuspended in 100 μl PBS containing 1% triton X-100 and 1μg/ml DAPI . 30 μl of cell suspension was transferred to a NC-Slide A2 and imaged in a nucleocounter NC-3000 ( Chemometec ) Western blotting was performed as described previously , using RIPA lysates and run and blotted onto nitrocellulose using the mini-Protein systems ( Bio-Rad ) . Antibody clones used were: EBNA-LP ( clones JF186 [81] or 4D3 [82];gift from Yasushi Kawaguchi] ) ; EBNA2 ( Clone PE2 ) ; EBNA3A ( Ab16126 , Abcam ) ; EBNA3B ( Rat monoclonal 6C9 [17] ) ; EBNA3C ( mouse monoclonal A10 ) ; LMP1 ( monoclonal CS1-4 , Dako ) . For immunofluorescence , cells were grown on a 12 chamber slide ( Ibidi ) . Cells were gently washed with PBS and then fixed with 4% paraformaldehyde for 15 minutes . Cells were washed twice with PBS and covered with blocking buffer ( PBS/10% FCS/100mM glycine/0 . 2% Triton X-100 ) for 30 minutes . Cells were stained with primary antibody in 50 μl blocking buffer for one hour , washed thrice in PBS and stained with fluorophore-conjugated secondary antibody ( Cheshire Bioscience ) for an hour . Chambers were washed three times with PBS and then the chamber removed , the slide briefly dipped in deionized water , and a coverslip mounted on the slide with Prolong Gold Antifade mount with DAPI ( Life Technologies ) . Slides were imaged on a Zeiss LSM5 Pascal confocal microscope: 63x objective , 4x digital zoom and shown as a projection of z-stacks of 1μm sections . Adult blood cells were purchased from UK National Blood and Transplant as waste products of platelet isolation . As they are waste products from anonymous volunteer donors , no ethics approval is required . Umbilical cord blood ( and the maternal blood ) were obtained with written informed consent of the mother ( an adult ) prior to the onset of labour , under the MatImms study , approved by the UK National Health Service Research Ethics Committee ( approval REC 13/LP/1712 ) . Anonymized blood samples surplus to the requirements of the MatImms study were used in this project , distributed by the Imperial College Healthcare NHS Trust Tissue Bank ( REC 12/WA/0196 ) and approved by the tissue bank’s Tissue Management Committee ( project R15029 ) . Other investigators may also have received these same samples . | Epstein-Barr virus ( EBV ) infects almost everyone . Once infected , people harbor the virus for life , shedding it in saliva . Infection of children is asymptomatic , but a first infection during adolescence or adulthood can cause glandular fever ( infectious mononucleosis ) . EBV is also implicated in several different cancers . EBV infection of B cells ( antibody-producing immune cells ) can drive them to replicate almost indefinitely ( ‘transformation’ ) , generating cell lines . We have investigated the role of an EBV protein ( EBNA-LP ) which is thought to support gene activation by the essential virus protein EBNA2 . We have made an EBV in which the EBNA-LP gene has been disrupted . This virus ( LPKO ) shows several properties . 1 . It is reduced in its ability to transform B cells; 2 . ‘Naïve’ B cells ( those whose antibodies have not adapted to fight infections ) die two weeks after LPKO infection; 3 . Some virus genes fail to turn on immediately after LPKO infection . 4 . Binding of EBNA2 and various cellular factors to these genes is delayed . 5 . EBNA-LP does not affect EBNA2-targeted cellular genes in the same way . This shows that EBNA-LP is more important in naïve B cells , and that it helps to turn on virus genes , but not cell genes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"gene",
"regulation",
"pathogens",
"regulatory",
"proteins",
"immunology",
"microbiology",
"dna-binding",
"proteins",
"dna",
"... | 2018 | Epstein-Barr virus nuclear antigen EBNA-LP is essential for transforming naïve B cells, and facilitates recruitment of transcription factors to the viral genome |
Gastrointestinal infection is often associated with hypophagia and weight loss; however , the precise mechanisms governing these responses remain poorly defined . Furthermore , the possibility that alterations in feeding during infection may be beneficial to the host requires further study . We used the nematode Trichinella spiralis , which transiently inhabits the small intestine before migrating to skeletal muscle , as a biphasic model of infection to determine the cellular and molecular pathways controlling feeding during enteric and peripheral inflammation . Through the infection of genetically modified mice lacking cholecystokinin , Tumor necrosis factor α receptors and T and B-cells , we observed a biphasic hypophagic response to infection resulting from two separate immune-driven mechanisms . The enteroendocrine I-cell derived hormone cholecystokinin is an essential mediator of initial hypophagia and is induced by CD4+ T-cells during enteritis . In contrast , the second hypophagic response is extra-intestinal and due to the anorectic effects of TNFα during peripheral infection of the muscle . Moreover , via maintaining naive levels of the adipose secreted hormone leptin throughout infection we demonstrate a novel feedback loop in the immunoendocrine axis . Immune driven I-cell hyperplasia and resultant weight loss leads to a reduction in the inflammatory adipokine leptin , which in turn heightens protective immunity during infection . These results characterize specific immune mediated mechanisms which reduce feeding during intestinal or peripheral inflammation . Importantly , the molecular mediators of each phase are entirely separate . The data also introduce the first evidence that I-cell hyperplasia is an adaptively driven immune response that directly impinges on the outcome to infection .
Intestinal inflammation is commonly associated with reduced feeding ( hypophagia ) and weight loss [1] , [2] , yet the mechanisms and underlying principles of these responses is unknown . Infection with the intestinal parasites Ascaris suum and Trichostrongylus colubriformis results in hypophagia that is coupled with an increase in cholecystokinin ( CCK ) released from I-cells [3] , [4]; a subset of intestinal epithelial enteroendocrine cells ( EECs ) . Despite only comprising 1% of the epithelium , EECs collectively form the largest mammalian endocrine system . Regulatory peptides and amines are released from EECs in response to luminal nutrients [5] and these peptides signal via vagal afferent fibers to feeding control centers in the brain . These EEC signals in concert with leptin , produced from adipose tissues indicating levels of fat deposits , ultimately control our daily short-term feeding patterns . However , the true biological function and molecular mechanisms that orchestrate the pathways driving hypophagia and weight loss during inflammation have not been addressed . The nematode Trichinella spiralis produces a well characterized CD4+ T-cell , Th2 driven transient inflammation in the small intestine culminating in worm expulsion via a mast cell dependent process [6] . Recently we have observed a hypophagic response during the Th2 driven enteritis induced by T . spiralis infection [7] . However , the full mechanisms controlling hypophagia during enteritis and the precise effects reduced feeding have on immunity to intestinal infection require further elucidation . T . spiralis is experimentally highly attractive since the enteritis fully resolves , but is closely followed by a peripheral inflammatory phase characterized by skeletal muscle invasion and myositis as part of the parasite's life cycle . Here , we demonstrate that this two step inflammatory process following T . spiralis infection is mirrored by a biphasic hypophagic response , and mediated by two separate adaptive immune driven mechanisms . We have characterized these two phases using genetically modified mice lacking functional CCK or adaptive immunity and demonstrated that CD4+ T-cells drive I-cell hyperplasia and the resulting CCK is an essential mediator of the initial hypophagia observed during enteritis . Conversely the second phase of hypophagia during skeletal peripheral myositis is CCK-independent but mediated by the anorectic actions of TNFα signaling . Furthermore , we demonstrate for the first time that this immune-EEC driven alteration in feeding also contributes a protective role during gastrointestinal infection . The hypophagia and resulting weight loss causes a reduction in fat secreted leptin , and the reduction in this hormone , which also acts as an inflammatory adipokine , augments the protective Th2 immune response aiding parasite expulsion . These results highlight the importance of the immunoendocrine axis in the gut during infection induced immunity and provide a biological function and associated mechanism for commonly associated infection induced weight loss . These data have wide-acting implications for the biology of gut infection and inflammation , and may inform new leptin-derived therapeutic strategies . Furthermore , the T . spiralis infected mouse presents a novel preclinical platform to study the biological mechanisms affecting food intake in inflammatory disorders , and has the unique potential to experimentally dissociate gastrointestinal from peripheral signals in an individual model .
Proximal enteritis induced by T . spiralis has been associated with a period of hypophagia and an increase in CCK and serotonin secreting EECs [7] . Here , mice were examined for alterations in feeding during T . spiralis induced inflammation . Interestingly , a biphasic response in feeding was observed following infection ( Fig . 1A ) . Mice became hypophagic from days 6–10 post infection ( p . i . ) , during the transient period of T . spiralis induced enteritis , and we observed a significant increase in CCK positive I-cells in wild-types ( Fig . 1C and E ) mirroring hypophagia at days 6 and 9 p . i . Feeding then returned to baseline levels until undergoing a second period of hypophagia from day 18–19 p . i . The secondary period of hypophagia occurs during the period of muscle invasion and peripheral myositis , caused when larvae form the “nurse cell” in which the parasite resides . To further investigate the biological mediators of the hypophagic responses the two phases were mechanistically explored using a panel of genetically modified mouse strains . CCKlacZ mice , which do not express or secrete CCK peptide due to a knock in of a LacZ cassette [8] , were infected with T . spiralis and their food intake monitored . Strikingly , the initial period of hypophagia was completely absent in infected CCKlacZ mice ( Fig . 1B ) . LacZ positive I-cell hyperplasia was indistinguishable from that of the “natural” I-cell response in wild-type mice ( Fig . 1D and E ) . However , the absence of CCK during this I-cell hyperplasia resulted in the complete absence of initial hypophagia in CCKlacZ mice , despite comparable enteritis ( Fig . S1 ) . Even with the absence of CCK and initial hypophagia , CCKlacZ mice still exhibited secondary hypophagia from day 18–19 p . i . and this was comparable to infected wild-type mice ( Fig . 1B ) . The second period of hypophagia occurred in wild-type and CCKlacZ mice despite the resolution of enteritis ( Fig . S1 ) and transpires during the period of larvae encysting within skeletal muscle , representing an extraintestinal inflammatory response to the same biological agent . The lack of I-cell hyperplasia in wild-type mice at day 20 p . i . ( Fig . 1E ) and the presence of the second phase of hypophagia in CCKlacZ mice ( Fig . 1B ) confirm this extraintestinal period of hypophagia during myositis as CCK independent . Taken together , these data demonstrate a biphasic hypophagia correlating to T . spiralis induced enteritis and peripheral myositis , respectively . Furthermore , increased I-cell function , through the release of CCK , is essential for the initial hypophagia during enteritis , but not the secondary episode during peripheral myositis . As EEC hyperplasia during inflammation has been previously linked to T-lymphocytes [7] , [9] , the biphasic hypophagia generated by T . spiralis was examined in severe combined immunodeficient ( SCID ) mice , which lack B and T-cells . SCID mice demonstrated a complete absence of initial hypophagia during T . spiralis induced enteritis and secondary hypophagia during peripheral myositis ( Fig . 2A and B ) . The lack of hypophagia was mirrored by a complete lack of I-cell hyperplasia in parasitized SCID mice ( Fig . 2C ) . This absence of hyperplasia was not seen in all epithelial secretory cells as , concurrent with previous findings [10] , indistinguishable goblet cell hyperplasia occurred in infected SCID , adoptively transferred SCID and wild-type animals ( Fig . 2D ) . CD4+ T-cells play a key role in the resolution of T . spiralis infection [11] , so to assess if CD4+ T-cells could restore I-cell hyperplasia and hypophagia in SCID mice , CD4+ T-cells ( >90% purity; Fig . S2A ) , were adoptively transferred into SCID recipients before infection . Successful reconstitution was evident from CD4+ splenocytes present post-transfer and via successful worm expulsion kinetics ( Fig . S2B and C ) . The adoptive transfer of CD4+ T-cells into SCID mice restored I-cell hyperplasia ( Fig . 2C ) and initial hypophagia during T . spiralis induced enteritis . Recipient mice began to eat less from day 4 p . i . , with significant hypophagia at days 6 and 7 ( Fig . 2E ) . This hypophagia was not a direct result of cell transfer alone , as uninfected reconstituted mice displayed no hypophagia ( Fig . 2E ) . Interestingly , the adoptive transfer did not restore the secondary period of hypophagia during the peripheral inflammation induced by T . spiralis ( Fig . 2A ) . Collectively , these data confirm that the biphasic alterations in feeding behavior during T . spiralis induced gastrointestinal and peripheral inflammation is mediated by the adaptive immune system . Furthermore , CD4+ T-cells are identified as the key initiator in I-cell hyperplasia and resulting CCK driven hypophagia during T . spiralis induced enteritis . However , the adoptive transfer of functional CD4+ T-cells did not restore the second phase of hypophagia occurring during nurse cell formation-induced myositis . Therefore CD4+ T-cells are not sufficient for this secondary hypophagic period , during T . spiralis induced myositis . We next sought to investigate which factors of the adaptive immune response were responsible for the second phase of hypophagia seen during peripheral inflammation induced during the period of nurse cell formation . Both CD4+ and CD8+ T-cells are present during parasite encystation [12] and many apoptotic factors , including TNFα , are detected during nurse cell formation [13] . Interestingly , TNFα is associated with cachexia in parasite infections [14] , [15] . Consequently , we examined serum cytokine levels throughout T . spiralis infection . Indeed , TNFα was significantly increased in the serum of infected mice at the time of secondary hypophagia ( Fig . 3A ) , comparable to levels known to directly cause cachexia in mouse infection models [16] . To test the function of increased TNFα during myositis we infected p55/p75−/− mice , which lack functional TNFα receptors , and assessed if TNFα was responsible for hypophagia during T . spiralis infection . Although initial I-cell hyperplasia and hypophagia during enteritis were present in infected p55/p75−/− mice ( Fig . 3B , C and D ) , remarkably , infected p55/p75−/− mice displayed no period of secondary hypophagia ( Fig . 3B and C ) . Therefore , although the initial CD4+ T-cell and CCK driven hypophagia during enteritis is independent of TNFα , a peripheral peak in TNFα during myositis is functionally responsible for the second phase of hypophagia , via the receptors p55 and/or p75 , during T . spiralis induced peripheral inflammation . Secretory cell hyperplasia during intestinal infection is known to be advantageous during infection . Various goblet and Paneth cell products have been show to have anti-parasitic affects [17] , [18] . We therefore tested whether I-cell hyperplasia and hypophagia are simply by-products of a parallel switch towards this secretory lineage , or whether I-cell hyperplasia is in itself advantageous during infection . A link between hypophagia , weight loss and immunity is the adipokine leptin; produced mainly by adipose tissue it is a peripheral signal to the body of fat mass deposits but also acts as a pro-inflammatory Th1 cytokine [19] . Therefore reductions in leptin would be anticipated to occur following CCK induced hypophagia and consequent weight loss in this experimental model . Loss of leptin may consequently enhance Th2 immune responses which are protective during nematode infection . CD4+ T-cell mediated I-cell driven hypophagia during enteritis was seen to result in significant weight loss at days 8 and 12 p . i . , accompanied by a visible reduction in abdominal fat pads , whereas the brief TNFα driven secondary hypophagia produced no significant alteration in weight at day 20 p . i . ( Fig . 4a ) . This weight loss was correlated with a reduction in serum leptin levels from day 6 p . i . ( Fig . 4B ) . To determine whether alterations in leptin could influence a protective Th2 driven intestinal I immune response , mesenteric lymph node ( mLN ) cells were polarized towards a Th2 phenotype in the presence or absence of leptin . The addition of leptin resulted in a significant increase in the amount of intracellular pro-inflammatory IFN-y detectable in CD4+ T-cells , as well as a significant reduction in the protective Th2 cytokine IL-4 ( Fig . 4C ) . To assess if the reduction in leptin during T . spiralis induced enteritis enhances immunity to infection , leptin levels were maintained at basal levels during hypophagia via recombinant leptin injection ( Fig . 5A ) . Strikingly , the restoration of basal leptin levels resulted in delayed expulsion of adult worms and a corresponding increase in nurse cell encystation ( Fig . 5B ) . Although no increase in IFN-y levels was seen in re-stimulated mLNs of leptin treated mice , a significant decrease in both Th2 cytokines IL-4 and IL-13 was seen at day 8 p . i ( Fig . 5C ) . A key Th2 driven expulsion mechanism of T . spiralis is mastocytosis and this was seen to be significantly reduced upon the restoration of basal leptin levels analogous to delayed adult worm expulsion ( Fig . 5D ) . Taken together this suggests that CD4+ T-cells drive a cascade in which I-cell hyperplasia produces hypophagia and weight loss , lowering pro-inflammatory leptin levels which feed back to influence the protective Th2 immune response , augmenting mastocytosis and allowing parasite expulsion .
During enteritis food intake is often significantly reduced , perhaps serving to inhibit consumption of contaminated food or to prevent further gut injury . EECs are implicated in this response: elevated I-cell produced CCK and hypophagia have been demonstrated during Ascaris suum and Trichostrongylus colubriformis infection , leading to the hypothesis that CCK was responsible for inflammation induced alterations in feeding [3] , [4] However , the true biological function and molecular mechanisms that orchestrate the pathways driving hypophagia and weight loss during inflammation have not been addressed . The life cycle of T . spiralis has uniquely allowed us to investigate these questions during both intestinal and peripheral inflammation . Our data demonstrate that these separate inflammatory episodes are mirrored by a biphasic hypophagia driven by two independent immune mediated mechanisms . I-cell hyperplasia and CCK are essential for the initial hypophagia during enteritis , which is orchestrated by CD4+ T-cells . Importantly , we have also identified the second phase of hypophagia during the period of nurse cell formation to be mediated by a separate myositis-induced immune mechanism , fully dependent on the actions of TNFα . Furthermore we show for the first time that immune driven weight loss during enteritis results in reduced levels of the Th1 adipokine leptin augmenting a protective Th2 response during infection . Intestinal inflammation is often associated with hypophagia and weight loss [1] , [2] and we have now determined an important immune driven mechanism to explain why this is biologically functional . We have therefore identified a novel molecular pathway and can include I-cell hyperplasia and weight loss as an adaptively driven immune response which , through alterations in leptin , is beneficial during intestinal infection . Our finding that T . spiralis infected mice lacking CCK ( CCKlacZ ) do not undergo initial hypophagia correlates with our own and other previous findings that a single treatment with the CCK1 receptor antagonist loxiglumide partially restores food intake in parasitized animals [7] , [20] . This partial restoration now seems likely a result of the short half-life of loxiglumide as opposed to alternative satiety factors playing an essential role . The previous characterization of CCKlacZ mice demonstrated normal food intake , fat absorption and mass compared to wild-types [21] , [22] . This coupled with our own observations rules out any underlying defect in feeding of CCKlacZ mice being responsible for the absence of hypophagia during infection . An alternative possibility is that the loss of CCK in brain neurons , rather than gut EECs , underpins the complete absence of hypophagia . However , the persistence of the second phase of hypophagia in CCKlacZ mice and our previous findings involving loxiglumide [7] which does not cross the blood brain barrier , make I-cell hyperplasia and CCK-vagal interactions most likely to mediate gut-induced hypophagia . T . spiralis infected SCID mice were seen to display I-cell hyperplasia and hypophagia only upon reconstitution of CD4+ T-cells . These data clearly indicate that CCK induced satiety , via vagal afferent fibers signaling to feeding control centres in the brain [5] , is an inherent pathway that is utilized by the adaptive immune system to bring about hypophagia and weight loss . This finding corresponds to recent studies showing that CD4+ T-cells restore 5-HT cell hyperplasia in Trichuris muris infected SCID mice [23] . The precise mechanism by which CD4+ T-cells cause EEC hyperplasia during infection remains to be elucidated . EECs have been shown to possess functional TLRs [24] and IL-13 receptors are present on 5-HT cells [9] . However , it has previously been established that during T . spiralis infection in SCID mice NK cells produce ample levels of IL-13 to induce goblet cell hyperplasia [10] yet this IL-13 appears not sufficient to cause EEC hyperplasia . We also detected mRNA for both TNFα receptors p55 and p75 on EECs , yet mice genetically deficient in TNF receptor signaling demonstrated initial hypophagia and enteritis ( Fig . S3A–C ) arose independently of its actions . As all gut epithelial subtypes are derived from pluripotent crypt stem cells [25] , immune cell mediators may alter the transcription factors at the stem cell level leading to altered EEC hyperplasia during infection . Alternatively , analogous alterations in the post-stem cell , neurogenin+ , EEC specific progenitor cells could alter EEC proliferation . BrdU labeling studies in experimentally-induced inflammation demonstrated EEC hyperplasia does occur at the level of the stem/progenitor cell rather than fully differentiated epithelial cells [26] . Indeed , the uncoupling of goblet and I cell hyperplasia seen here in infected SCID mice supports the hypothesis that neurogenin+ EEC precursors , rather than the stem cell itself , is targeted by an unknown CD4+ dependent mechanism during T . spiralis driven I-cell hyperplasia . Identifying which factors drive this CD4+ T-cell-stem/progenitor cell interaction is an exciting area for further study . We have identified a novel and specific role for TNFα in hypophagia during T . spiralis induced peripheral inflammation . The absence of the second phase of hypophagia from p55/p75 −/− mice suggests that the drop in food intake during this extra-enteric inflammatory period is due to the anorexic effects of systemic TNFα or downstream targets . The significant peak of serum TNFα seen in mice correlated with the second period of hypophagia , occurring during nurse cell development and myositis , and strongly supports this notion . Indeed , we observed serum levels comparable to levels known to directly cause cachexia in mouse infection models [16] . Furthermore , TNFα is associated with cachexia in Trypanosoma cruzi infection [14] and during schistosomiasis [15] . There are numerous modes of action by which TNFα could cause anorexia [27] . Although central TNFα levels were not directly monitored , the 20 pg/ml serum TNFα measured during secondary hypophagia is below the levels required to induce anorexia by central administration [28] . It is therefore most likely that TNFα is acting on peripheral afferent nerves , as low level localized cytokine production can trigger afferent nerves without causing an increase in circulating cytokine levels [29] . It is also possible that myalgia and malaise may have contributed to reduced food intake: appetite per se cannot be measured in mice . The observed systemic peak in TNFα occurs during the period of encystation of T . spiralis new born larvae in cells of the striated muscle . Encystations are likely to arise from day 4–10 post-infection with rapid growth of the parasite occurring over the following 20 days as terminally differentiated muscle cells re-enter the cell cycle and establish a niche for the parasite [12] . Early non-significant increases in systemic TNFα were seen as early as day 8 post-infection; day 18–21 post-infection may therefore indicate a “tipping point” in peripheral TNFα levels , where significant myositis breaches the threshold required to produce anorexia . TNFα has been shown to be involved in nurse cell formation [13] , yet we observed no alteration in nurse cell development in p55/p75−/− mice that could alternatively explain the observations seen ( Fig . S3C–E ) . The cellular source of TNFα remains to be elucidated . CD4+ and CD8+ T-cells are reported to be present during parasite encystation [12] , as are macrophages interestingly peaking during hypophagia [30] . However , as we illustrate here , given the absence of secondary hypophagia in SCID mice reconstituted with CD4+ T-cells , where macrophages are present , the likelihood is that CD8+ T-cells may be the source of cachectic TNFα . Further studies are therefore required to ascertain the cellular source of TNFα which drives the hypophagia during T . spiralis induced myositis . Immune mediated secretory cell hyperplasia during intestinal infection is advantageous as goblet and Paneth cell products have been show to have anti-parasitic affects [17] , [18] . We therefore postulated whether I-cell hyperplasia and hypophagia are simply by-products of a parallel switch towards these beneficial secretory lineages or whether I-cell hyperplasia is in itself advantageous in nematode expulsion . Stimulation of the vagus nerve via nutritional release of CCK has also been shown to protect against hemorrhagic shock [31] . Therefore I-cell hyperplasia during nematode infection may represent a previously unidentified anti-inflammatory response . We therefore hypothesized that a reduction in weight as a result of I-cell induced hypophagia may alter the levels of the Th1 adipokine leptin [19] . A reduction in leptin could enhance the protective Th2 immune response to nematode infection . Indeed significant weight loss and reduced leptin levels did occur during T . spiralis induced hypophagia . Recent data on splenocytes demonstrated that leptin alters polarized CD4+ T-cells towards a Th1 phenotype via alterations in proliferation in vitro [32] and we demonstrated parallel results in mLN cells for the first time . Unfortunately CCKlacZ mice have overall reduced basal levels of leptin [33] and were hence unsuitable to study the affect of reduced leptin on intestinal inflammation . We therefore maintained basal leptin levels in infected hypophagic mice and strikingly saw a significant reduction in Th2 cytokines and mastocytosis culminating in delayed worm expulsion . Interestingly mastocytosis was similar in both leptin reconstituted and wild-type mice at day 8 p . i . demonstrating that initially mastocytosis can establish , but without the I-cell driven reduction in Th1 polarizing leptin it is blunted later in infection . These results complement other recent studies in identifying the adipokine leptin as a molecule which can greatly influence the response to infection . Mice lacking the leptin receptor are highly susceptible to infection from protozoa [34] , pneumonia [35] and Listeria [36] demonstrating how malnutrition can compromise Th1 driven immunity . However , our data demonstrate that brief alterations in leptin can benefit immunity in terms of Th2 driven resistance to infection . Indeed , a recent study has demonstrated that leptin receptor deficient mice are resistant to experimentally induced Th2-mediated colitis [32] . The precise action of leptin in our studies may be as a direct result of effects on CD4+ T-cell IL-4 production altering mast cell differentiation , proliferation and migration [37] or due to direct effects on mast cells which have recently been shown to express leptin receptors [38] . Leptin may also directly act on Th2 cytokine production itself as opposed to indirect alterations on Th1 cytokine production [32] . Further study is therefore required to address the leptin-mast cell axis which alters parasite expulsion in our model . In conclusion , we have identified two separate immune mediated mechanisms of hypophagia during infection induced gastrointestinal and peripheral inflammation , which act via the distinct pathways of I-cell hyperplasia and TNFα cachexia . Furthermore , we demonstrate for the first time an immunoendocrine feedback loop , in which CD4+ T-cell driven weight loss via CCK reduces leptin levels which impinge on CD4+ T-cell driven effector mechanisms for gastrointestinal infection resolution . Our data elucidate inflammation and weight loss , not just as commonly associated phenomena , but highlights them as a novel immune driven mechanism in parasite expulsion . These data offer potential specific treatment targets to modulate feeding and immune function during inflammatory diseases of the intestine .
Mice were housed in specific pathogen free conditions and experiments were carried out in accordance with the United Kingdom Home Office Scientific Procedures Act ( 1986 ) under Department for Environment , Food and Rural Affairs license . Male C57BL/6 and BALB/c mice were obtained from Harlan-Olac Ltd . CCKlacZ mice have a LacZ cassette knocked into the CCK locus on a C57BL/6 background , so homozygote animals are CCK null but faithfully express LacZ in the I cell population [8] . TNFα receptor null p55/p75 −/− ( C57BL/6 background ) and severe combined immunodeficient mice ( SCID , BALB/c background ) were generated as previously described [8] , [39] . The maintenance , infection and recovery of T . spiralis were carried out as previously described [40] . Mice were individually weighed on a daily basis . Food intake per mouse was derived by weighing the chow ( B and K , Hull , UK ) daily . Proximal small intestine was fixed and stained and I-cells were enumerated using , CCK specific , L421 anti-proCCK as previously described [7] . For CCKlacZ detection , transverse 12 µm sections of tissue were cut and fixed in 0 . 2% glutaraldehyde and stained with X-gal as previously described [8] . Mast or goblet cell sections were stained in toludine blue or Schiff's reagent , respectively . After mounting , positive cells were enumerated in 20 randomly selected villus crypt units ( VCU ) and results presented as mean number of positive cells/20 VCU ( ± s . e . ) . Mesenteric lymph node ( mLN ) cells were prepared from day 7 p . i . BALB/c mice , in RPMI-1640 , supplemented with 10% fetal calf serum , 100 µg/ml penicillin/streptomycin and 1 mM L-glutamine ( complete media ) . CD4+ T-cells were isolated via negative selection using an isolation kit ( Miltenyi Biotec ) . Evaluation of CD4+ purity was via flow cytometry . SCID mice received 4×106 cells in P . B . S . via intraperitoneal ( i . p . ) injection 2 days before infection . mLN cells at 5×106 cells/ml in complete media received 50 µg/ml of T . spiralis antigen ( Ag ) . Supernatants were collected after 24 hrs and cytokines measured using a cytometric bead array kit ( BD ) . Serum was obtained from blood at the time of sacrifice via centrifugation at 15000×g and cytokines measured using a cytometric bead array kit ( BD ) . Mouse leptin ELISA ( Linco ) was used to detect mouse serum leptin according to manufacturer's instructions . During the period of significant hypophagia , mice were treated at 10 a . m . and 6 p . m . via an i . p . injection of recombinant leptin ( R and D ) at 0 . 5 µg/g of initial body weight or control vehicle PBS [41] . 2×106/ml mLN cells were stimulated via 5 µg/ml αCD28 , 3 µg/ml αCD3 ( BD ) and polarized via 50 ng/ml IL-4 ( Peprotech ) , 50 µg/ml anti-IFN-γ with/without 500 ng/ml recombinant leptin . At 120 hrs 1 µg/ml Brefeldin A/1 µg/ml monensin ( Sigma-Aldrich ) for IFN-y/IL-4 staining was added for 4 hrs before blocking with anti-FcγR ( BD ) . Cells were stained for CD4 ( BD ) for 30 mins at 4°C before fixing in FACS fix buffer ( 1% formaldehyde , 0 . 1% BSA and 0 . 05% NaN3 in PBS ) . Cells were permeabilised in 0 . 1% saponin ( Sigma-Aldrich ) and stained with biotinylated anti-IFN-γ/anti-IL-4 ( BD ) for 25 mins at RT . Controls were stained with isotype controls ( BD ) . Biotinylated antibodies were detected by streptavidin APC conjugate ( Caltag ) at 1/200 in saponin for 25 minutes RT . Cells were analyzed on a FACScalibur using Flowjo . Two experimental groups were compared using Student's t-test . Three or more groups were compared using the Kruskal-Wallis test , Dunn's multiple comparison post-test . A p value of ≤0 . 05 was considered statistically significant . * , P< . 05; ** , P< . 01; or *** , P< . 005 for indicated comparisons , error bars represent SE of means . | Infection with intestinal parasites often results in a period of reduced appetite which can result in weight loss; however the factors which control these feeding alterations and the reason why they occur is unknown . We used the nematode parasite Trichinella spiralis , which during its life cycle causes intestinal and muscular inflammation , as a mouse infection model to study the factors which alter feeding during infection . We found that the mouse immune response to the parasite was driving two periods of reduced feeding by two distinct immune mediators during the intestinal and muscular periods of infection . Interestingly , the immune system was utilizing a hormone which usually terminates feeding during our daily meals to cause a reduction in weight and fat deposits . Furthermore , we found that a reduction in these fat deposits and their associated hormones actually helped the mouse expel the parasite from the intestine . Hence the immune driven weight loss was actually beneficial to the mouse's ability to resolve an infection . Our study provides novel insights into how the immune system interacts with feeding pathways during intestinal inflammation and may help us design new strategies for helping people with parasitic infections of the gut . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"small",
"intestine",
"immune",
"physiology",
"immune",
"cells",
"cytokines",
"immunologic",
"subspecialties",
"immunity",
"to",
"infections",
"immunology",
"anatomy",
"and",
"physiology",
"parasitic",
"diseases",
"helminth",
"infection",
"adaptive",
"immunity... | 2013 | Adaptive Immunity Alters Distinct Host Feeding Pathways during Nematode Induced Inflammation, a Novel Mechanism in Parasite Expulsion |
Yeast Npl3 is a highly abundant , nuclear-cytoplasmic shuttling , RNA-binding protein , related to metazoan SR proteins . Reported functions of Npl3 include transcription elongation , splicing and RNA 3’ end processing . We used UV crosslinking and analysis of cDNA ( CRAC ) to map precise RNA binding sites , and strand-specific tiling arrays to look at the effects of loss of Npl3 on all transcripts across the genome . We found that Npl3 binds diverse RNA species , both coding and non-coding , at sites indicative of roles in both early pre-mRNA processing and 3’ end formation . Tiling arrays and RNAPII mapping data revealed 3’ extended RNAPII-transcribed RNAs in the absence of Npl3 , suggesting that defects in pre-mRNA packaging events result in termination readthrough . Transcription readthrough was widespread and frequently resulted in down-regulation of neighboring genes . We conclude that the absence of Npl3 results in widespread 3' extension of transcripts with pervasive effects on gene expression .
Budding yeast Npl3 comprises two RNA binding domains ( RBDs ) and a C-terminal domain that is rich is Arg , Gly , Ser and Tyr residues . This structure shows similarities to the SR ( Ser-Arg rich ) class of metazoan pre-mRNA binding proteins [1 , 2] . Genetic and biochemical analyses have implicated Npl3 in many processes , including pre-mRNA splicing , polyadenylation , mRNA export and cytoplasmic translation [3–7] , as well as R-loop prevention and chromatin modification [6 , 8] . Transcription termination of RNA polymerase II ( RNAPII ) occurs by polyadenylation-dependent and polyadenylation-independent pathways , correlated with whether the transcript is coding or non-coding ( reviewed in [9 , 10] ) . Termination of mRNAs , requires two complexes termed cleavage and polyadenylation factor ( CPF ) and cleavage factor ( CF ) . Together , the CPF and CF complexes facilitate cleavage of the nascent RNA strand and removal of the elongating polymerase , resulting in a polyadenylated RNA product . Two mechanisms have been reported for these processes , which are likely to occur in combination . In the ‘torpedo’ pathway , the nascent RNA molecule is cleaved at the polyA site and the released 3’ fragment of the transcript still bound by RNAPII is degraded by the 5’-3’ exonuclease Rat1 . This is proposed to then destabilize the polymerase complex . A second “allosteric” mechanism leads to the elongating polymerase being disengaged from the nascent transcript downstream of the polyA site due to , poorly understood , conformational changes concomitant with assembly of the CPF-CF complex . Notably , analyses on reporter constructs indicated that Npl3 can act as an anti-terminator , by antagonizing cleavage factor 1 ( CF1 ) binding and thus restricting the use of cryptic poly ( A ) sites [4 , 7 , 11] . In addition to mRNAs , RNAPII also transcribes several classes of non-protein coding RNAs ( ncRNAs ) and the majority of these terminate by polyadenylation-independent pathways . These ncRNAs include the small nucleolar RNAs ( snoRNAs ) , 73 of which function in yeast ribosome synthesis , four small nuclear RNAs ( snRNAs ) that form the core of the pre-mRNA spliceosome , as well as diverse long ncRNAs ( lncRNAs ) such as the cryptic unstable trancripts ( CUTs ) . The snoRNAs are processed from pre-snoRNAs that can be independently transcribed , cleaved from polycistronic transcripts , or excised from pre-mRNA introns . Independently transcribed snoRNAs , snRNAs and CUTs are all thought to predominately terminate via a pathway that requires RNA-binding by Nrd1-Nab3 complex and the Sen1 helicase ( together termed the NNS complex ) [12–19] . Termination of snoRNAs and CUTs by the NNS complex is associated with recruitment of the TRAMP and exosome complexes to the nascent RNA [14 , 20–22] . The TRAMP complex tags RNAs by the addition of a short 3' oligo ( A ) tail , and directs target RNAs to the nuclear exosome for degradation [23–26] . This can result in either complete degradation of the RNA , in the case of CUTs , or the processing of long precursor snoRNAs to the shorter , mature form [27] . However , some snoRNAs can also be terminated by mRNA 3’ cleavage factors , with [20 , 28] or without [29] subsequent polyadenylation . In addition , surveillance factors can influence termination , since loss of exosome activity leads to defects in NNS termination [30–33] . Moreover , gene-length correlates with the termination pathway used , probably via changes in the phosphorylation state of RNAPII [34 , 35] and/or histone H3 , lysine 4 trimethylation [36] , both of which can promote NSS termination . Prior data indicate that a proportion of RNAPII transcription events terminate early on protein-coding genes [37–40] . These promoter proximal ncRNAs or “sCUTs” [39] are oligoadenylated , presumably by the TRAMP complex [37] , and targeted for turnover by the nuclear surveillance machinery . To better understand the in vivo functions of Npl3 , we determined its RNA binding profile , and identified changes in RNA abundance and RNAPII association when the NPL3 gene is deleted . The absence of Npl3 resulted in transcriptional termination defects at diverse RNAs , with readthrough observed on large subsets of both mRNAs and ncRNAs . These termination defects appear to cause widespread changes in gene expression , both through inappropriate termination and through transcriptional interference at neighboring genes .
To identify direct RNA targets of Npl3 binding , we performed in vivo UV cross-linking and analysis of cDNAs ( CRAC ) [41] . The endogenous NPL3 gene was tagged with an N-terminal ProteinA-TEV-His6 ( PTH ) tag , retaining the intact , endogenous NPL3 promoter . This construct supported wild-type growth as the sole source of Npl3 ( S1A Fig ) , indicating that the fusion protein is functional . Yeast cells were UV irradiated while actively growing and PTH-Npl3 was isolated , Npl3-bound RNA fragments were purified , converted to a cDNA library and sequenced by next generation sequencing ( all sequence data are available from GEO under accession number GSE70191 ) . S1B Fig shows expression of the tagged protein and an autoradiogram of labeled , associated RNAs . Npl3 binding sites were most frequent on mRNAs , consistent with previous studies [40 , 42 , 43] , but were also identified on several classes of ncRNA , including rRNAs , tRNAs , snRNAs and snoRNAs , as well as lncRNAs , including CUTs , stable unannotated transcripts ( SUTs ) and other unannotated transcripts apparently derived from intergenic regions or antisense transcription . The distribution of Npl3 across RNA classes is shown for two independent CRAC experiments in Fig 1A . In both datasets , Npl3 binding was predominately on RNAPII transcripts . The distribution of Npl3 along transcripts showed distinct patterns for different classes of RNA . On mRNAs , Npl3 binding was highest in the 5’ end region ( Fig 1B ) , consistent with other recent RNA-crosslinking data [40] . A previous ChIP analysis , in which Npl3 is crosslinked to chromatin , found Npl3 enriched at 3' ends [6] . This apparent discrepancy may reflect differences in Npl3 binding at the 5' and 3' ends of genes , with direct RNA binding occurring predominantly at the 5' end , and stronger association with the transcription and processing complexes at the 3' end . Similar 5’ enrichment was reported for nuclear surveillance factors including Nrd1 , Nab3 and Mtr4 as well as for RNAPII , and has been proposed to reflect a substantial level of premature transcription termination [37 , 38 , 40] . As described above , these promoter proximal lncRNAs are oligoadenylated by the TRAMP complex , and we therefore mapped the association of Npl3 with RNAs carrying non-encoded oligo ( A ) tails [37] . Among RNA fragments recovered in association with Npl3 , 24–28% carried oligo ( A ) tails , depending on the individual CRAC experiment , indicating that Npl3 frequently binds across the junction between truncated mRNAs and oligoA tails . Note that the total fraction of Npl3 target RNAs that are oligoadenylated is likely to be higher , as only a small region of each transcript is sequenced . In contrast , only 4–4 . 7% of RNAs bound by RNAPII were oligoadenylated ( see below ) . Fig 1C shows the distribution of Npl3 bound hits containing oligo ( A ) tails across different RNA classes for two independent CRAC experiments . The distribution of oligo ( A ) tails in Npl3 target sequences was similar to the overall distribution of hits on mRNAs ( Fig 1D ) . This indicated that Npl3 is frequently bound to degradation substrates or intermediates , including prematurely terminated mRNAs , and suggests that it may function with surveillance factors to mediate early transcription termination and/or RNA degradation . Npl3 is known to be required for efficient splicing of ribosomal protein gene ( RPG ) pre-mRNAs [3] . Consistent with this , we found that Npl3 strongly accumulated on introns of these pre-mRNAs relative to other intron-containing pre-mRNAs ( S1C Fig ) . On non-RPG , intron-containing pre-mRNAs , the binding of Npl3 dropped sharply at the 5’ end of the intron . The lower recovery of introns relative to mature message indicates that Npl3 remains bound to mRNAs after splicing ( S1D Fig ) . The distribution of Npl3 over the CUT class of lncRNAs was similar to that observed for mRNAs with strong enrichment towards the 5' end ( Fig 1E ) , consistent with the proposal that initial cotranscriptional packaging of pre-mRNAs and lncRNAs is similar [37] . In marked contrast , Npl3 binding was enriched towards the 3' end of snoRNAs ( Fig 1F ) , suggesting a role in transcription termination and/or 3' end processing of these ncRNAs . Overall , our RNA binding site data suggest that Npl3 is involved in surveillance and/or transcription termination of both mRNAs and ncRNAs . Motif analysis did not identify a specific Npl3 binding site . We note , however , that the four most overrepresented 4-nucleotide motifs each contain a U-G sequence ( S1E Fig ) . Npl3 , and particularly RRM2 , was reported to show strong in vitro binding to U+G rich sequences including U-G dinucleotides [46] . To identify functional targets of Npl3 , we assessed the transcriptome-wide effects of the loss of the protein on steady-state RNA levels , using strand-specific tiling arrays . Npl3 was reported to be highly abundant ( 78 , 700 copies per cell ) [47] and has many different targets , which might show differential binding to residual Npl3 following depletion or relocation . We therefore analyzed the effects of deletion of the NPL3 gene . Tiling array analyses and RNAPII crosslinking were determined using two independent strains in which NPL3 was deleted immediately prior to the commencement of the experiments . Wild-type ( WT ) and npl3Δ strains were grown to logarithmic phase , RNA was extracted and reverse transcribed to make cDNA , which was then hybridized to tiling arrays . Normalized probe intensity data for all detected transcripts can be found in S1 Table . Total RNA was extracted from WT and npl3Δ yeast strains , and equal amounts of cDNA were hybridized to strand-specific tiling arrays . Differential expression analysis identified 1391 mRNAs with significantly altered expression ( adjusted p-value <0 . 05 ) , of which 1229 were decreased and 162 were increased ( Fig 2A and 2B ) . S2 Table shows differential expression analysis for all mRNAs , snoRNAs , CUTs and SUTs . The opposite effect was observed for CUTs , with 410 showing significantly increased expression , and only 8 showing significantly decreased expression ( Fig 2C and 2D and S2 Table ) . Increased expression was also observed for snoRNAs; 33 showed significantly altered expression , 31 of which were increased in the mutant strain ( Fig 2E and 2F and S2 Table ) . To gain an understanding of how lack of Npl3 might lead to a global decrease in mRNA abundance , we ranked all mRNAs by log2 fold change in the mutant compared to the WT strain , according to the differential expression analysis ( S2 Table ) . We then focused our analyses on the 30 most down-regulated genes in the npl3Δ strain , and examined their genomic environment ( Table 1 ) . As expected , the most down-regulated gene was NPL3 , which is absent from the genome and was discounted from the analysis . We found that 15/30 ( 50% ) of down-regulated genes reside in a convergent orientation with an expressed protein coding gene . A previous analysis found that only 6% of all yeast genes reside in convergent orientations in which both genes are expressed [48] . The proportion of convergent mRNAs with reduced expression in npl3Δ strains was therefore unexpectedly high . At 11 of the 15 convergent mRNA loci ( 73% ) , the down-regulated gene is adjacent to a gene that showed clear transcription readthrough , suggesting that their expression is blocked by transcriptional interference . An additional nine down-regulated mRNAs are convergent with an ncRNA that showed transcription readthrough . A further four down-regulated mRNAs are located in tandem with an upstream gene that shows readthrough , while seven mRNAs are apparently down-regulated by both tandem and convergent readthrough . Three of the 30 most down-regulated genes do not appear to be inhibited by convergent or tandem readthrough , or by intergenic transcription . Of these , YJR015W is seemingly down-regulated due to transcription changes over a local chromosome domain , since both upstream tandem genes are also down-regulated , while FMP48 and TPO4 are down-regulated by unknown mechanisms . Although mRNA expression was most frequently decreased in npl3Δ strains , several mRNAs were up-regulated . We examined the genomic environment for the top 30 up-regulated genes ( S3 Table ) . Eleven of these correspond to spliced ribosomal protein genes , and increased intron signal in the npl3Δ strain accounts for the differential expression . A further eleven up-regulated genes showed increased readthrough from upstream mRNAs or ncRNAs , suggesting that apparent increased expression is due to readthrough signal from the neighboring gene rather than specific up-regulation . The remaining eight genes ( HSP12 , DDR2 , HES1 , YDR124W , YML007C-A , ALP1 , PUG1 and YCL049C ) are apparently specifically up-regulated in npl3Δ strains . To investigate whether the gene expression changes observed are indeed due to transcriptional interference , we more closely analyzed two strongly down-regulated mRNAs: THO1 and PTC7 ( Figs 3 and S2 , respectively ) . In Fig 3A , panels II and III show tiling array expression data for two biological replicates of the npl3Δ strain ( upper ) and WT ( lower ) strains in a genome viewer format . Panels I and IV show corresponding data for the association of RNAPII with the nascent transcript as determined by UV-crosslinking and analysis of cDNAs ( CRAC; see below ) . Features on the Watson strand are shown above the chromosomal nucleotide numbers and features on the Crick strand are shown below . Apparent readthrough from the VHR2 gene is associated with strong down-regulation of THO1 , which encodes a nuclear pre-mRNA binding protein ( Fig 3A ) . Strand-specific reverse transcription ( RT ) , followed by qPCR confirmed that the VHR2 gene was indeed extended , and that the increased downstream expression was not a distinct transcription product ( Fig 3B ) . Quantification by RT-qPCR indicated that 3’ extended VHR2 is elevated ~5 fold , whereas THO1 expression is reduced ~5 fold . The approximate positions of RT primers and qPCR amplicons are shown by green arrows and red lines , respectively , in Fig 3A . Similar analysis of the UPF2-PTC7 region revealed that apparent readthrough from the UPF2 gene is associated with strongly reduced expression of PTC7 , encoding a Type 2C serine/threonine protein phosphatase ( PP2C ) ( S2A Fig ) . In this case , RT-qPCR quantification revealed ~10 fold elevated readthrough from UPF2 , associated with ~6 fold suppression of PTC7 expression ( S2B Fig ) . This suggests that transcription termination defects in the npl3Δ strain lead to changes in expression of surrounding genes . It remained possible that changes in RNA abundance for the npl3Δ strain observed in tiling array and RT-qPCR data might reflect reduced pre-mRNA surveillance and degradation rather than altered transcription . To discriminate between increased readthrough and RNA stabilization , we assessed changes in RNAPII occupancy following loss of Npl3 . To do this , we used CRAC to crosslink RNAPII to the nascent transcript , which provides genome-wide , strand-specific , nucleotide resolution mapping data in vivo in growing cells . The CRAC technique was applied using strains in which the largest subunit of RNAPII , Rpo21 , carried a C-terminal , His6-TEV-Protein A ( HTP ) tag , as recently described ( Milligan et al . , submitted ) . Tagged Rpo21 was well expressed in WT and npl3Δ strains , and was shown to crosslink efficiently to RNA ( S3A Fig ) . Total RNAPII occupancy across different classes of RNA was largely unchanged between the WT and npl3Δ strains ( S3B Fig ) . However , significant differences in the location of RNAPII were observed for individual genes . In Figs 3A and S2A , blue plots show Rpo21 occupancy in WT yeast and red plots show occupancy in npl3Δ . The density of RNAPII was highest at the 5’ ends of most protein-coding genes , consistent with published NET-seq data that maps the transcribing polymerase by sequencing 3' ends of associated nascent transcripts [38] , and with the distribution of pre-mRNA binding factors , including Npl3 ( [37 , 40] and Fig 1 ) . Differences in RNAPII occupancy at the two convergent loci are summarized in Figs 3C and S2C . RNAPII occupancy within the VHR2 ORF was comparable between the two strains , and RNA accumulation was very similar in the mutant and WT strains ( Fig 3A and 3C ) . However , in the CUT557 region immediately downstream , RNA accumulation was increased 4 . 7 fold while polymerase occupancy was increased 1 . 8 fold in npl3Δ . RNAPII crosslinking in the region between CUT557 and the downstream gene HOR2 was also elevated by 2 . 1 fold in the mutant , indicating that transcriptional readthrough extends into this region . HOR2 itself appears to be inhibited by transcriptional interference acting in tandem , as shown by decreased RNA accumulation ( to 30% of WT ) , and polymerase occupancy ( decreased to 50% of WT ) . The THO1 transcript is greatly reduced in npl3Δ ( to 10% of WT ) , with polymerase occupancy reduced to 20% of WT . Analysis of expression and RNAPII occupancy over the UPF2-PTC7 locus also confirmed UPF2 readthrough and PTC7 down-regulation ( S2A and S2C Fig ) . In addition , RNAPII density was decreased over the downstream PPE1 gene . This indicates that the transcriptional readthrough from UPF2 also inhibits expression of this tandem , flanking gene . Down-regulation of PPE1 can only be determined from the RNAPII occupancy data and is not evident from tiling array data as the PPE1 signal is obscured by the UPF2 readthrough signal . This demonstrates the difficulty in discriminating down-regulation due to readthrough in tandem . We conclude that transcriptional readthrough of multiple mRNA genes results in down-regulation of downstream convergent and tandem genes . To determine whether correctly processed and polyadenylated mRNAs are also produced from genes showing transcriptional readthrough , we analyzed the 3' end of UPF2 in WT and npl3Δ by cleavage with RNase H using an oligo hybridizing ~250 nt upstream of the UPF2 annotated 3’ end . Cleavage reactions were performed with the gene-specific oligo , with and without the addition of oligo ( dT ) to deadenylate the cleavage product ( S2D Fig ) . We observed substantially less mature polyadenylated UPF2 mRNA in the mutant ( lanes 1 and 2 , compared to 4 and 5 ) , but the adenylation pattern was apparently the same ( lane 2 compared to 5 ) . This indicates that cleavage and polyadenylation of UPF2 mRNA is reduced in the npl3Δ strain , but the location of the residual activity is unaltered . The tiling array data indicate that expression of the CYC1 gene is down-regulated in npl3Δ due to transcriptional readthrough from the convergent gene UTR1 ( Table 1 ) . CYC1 encodes cytochrome C and transcription is up-regulated on glycerol medium . WT and npl3Δ strains were grown in either glucose or glycerol medium and the level of CYC1 mRNA was quantified by RT-qPCR ( Fig 3D ) . On glucose medium CYC1 was reduced ~5 . 9 fold in npl3Δ relative to WT , validating the findings of the tiling array . However , CYC1 abundance was increased 4 . 1 fold when the npl3Δ strains were transferred to glycerol medium , resulting in an expression level close to WT . In contrast , the level of THO1 was not increased by transfer of the npl3Δ strain to glycerol medium ( Fig 3D ) . This demonstrates that CYC1 expression remains subject to specific transcription regulation in the absence of Npl3 . Npl3 was crosslinked to ncRNAs ( Fig 1 ) and the npl3Δ mutation altered the expression of ncRNAs including CUTs and snoRNAs ( Fig 2 ) , suggesting that the loss of Npl3 might also affect transcription termination on ncRNA genes . Previous work identified genes that are regulated by upstream CUTs , which inhibit transcription of the downstream mRNA , including the nucleotide biosynthesis factors ADE12 and URA2 [49] . In npl3Δ strains , CUT680 upstream of URA2 and CUT324/325 upstream of ADE12 were accumulated , accompanied by reduced expression of the downstream protein-coding gene ( Fig 4A–4C ) . Metagene analyses show increased polymerase density at the 3' ends of CUTs , and immediately downstream , in npl3Δ compared to WT ( Fig 4D ) . These data suggest that that Npl3 is required for normal termination of CUTs , and that without proper termination these normally unstable transcripts are not efficiently turned over by the nuclear RNA surveillance machinery . Inspection of microarray data revealed 3’ extensions for many snoRNAs in npl3Δ strains . All H/ACA and C/D box snoRNAs were included in the analysis and , strikingly , we observed extended 3’ ends for 46 of the 51 RNAPII transcribed , monocistronic snoRNA genes , and for all five polycistronic pre-snoRNA transcripts . One gene ( SNR13 ) could not be interpreted due to missing probes ( Tables 2 and S4 ) . Another , SNR52 , is the sole snoRNA transcribed by polymerase III , and is therefore terminated through a different pathway . This leaves just three RNAPII transcribed snoRNAs that do not show readthrough: U3B ( SNR17B ) , SNR63 and SNR85 . Metagene analyses of the Rpo21 CRAC data showed increased RNAPII association towards the 3' ends of all snoRNAs in npl3Δ strains ( Fig 5A ) . Examples of extended snoRNAs are shown in Figs 5 and S4 . The box C/D snoRNA snR60 is extended approximately 500 nt in npl3Δ and appears to terminate about 100 nt into the downstream UBX6 gene ( Fig 5B ) . The presence of extended snR60 was confirmed by northern blot ( Fig 5C ) . S4 Fig shows extension of the box H/ACA snoRNA snR3 , determined by tiling array , and RNAPII occupancy data ( S4A Fig ) and confirmed by RT-qPCR ( S4B Fig ) . Comparison of expression and RNAPII occupancy at this locus is shown in S4C Fig . The snR3 transcript appears to be extended greater than 1000 nt downstream with transcription proceeding through downstream , annotated CUT genes ( CUT221/222/223 ) . In some cases , extension of snoRNA genes was associated with strongly reduced expression of neighboring genes . As an example , SNR3 readthrough correlates with reduced expression of EFM3 ( S4A–S4C Fig ) . Some snoRNAs appear to be extended many kilobases , apparently utilizing the termination site of the next downstream protein gene . To confirm that snoRNA 3’ extensions result from transcriptional readthrough , we calculated “readthrough scores” for three snoRNAs ( SNR11 , SNR30 and SNR60 ) that appeared to be extended based on tiling array data , as well as SNR17B that did not appear to be extended . We calculated the sum of all RNAPII hits in the 500 nt 3’ flanking region , relative to the sum of all hits within the snoRNA sequence , and compared this ratio for the WT and npl3Δ strains . For the extended snoRNAs , Rpo21 hits in the 3’ flanking region hits were elevated 1 . 16 to 2 . 17 fold in npl3Δ , but reduced to 0 . 84 fold of the WT for SNR17B ( Fig 5D ) . Overall , the magnitude of RNAPII occupancy changes downstream of snoRNAs in npl3Δ relative to WT is much less than changes in expression . We suggest that the extended snoRNA transcripts predominately reflect defects in RNA surveillance rather than processing/maturation , as we found the abundance of mature snoRNAs to be comparable in the npl3Δ mutant and WT strains ( S4D Fig ) . Many snoRNAs harbor a cleavage site for the endonuclease Rnt1 ( RNase III ) positioned downstream of the mature 3’ end ( reviewed in [50] ) . Cotranscriptional cleavage by Rnt1 provides an entry site for 3’-exonuclease processing back to the mature 3’ end of the snoRNA , and also allows the 5’ exonuclease Rat1 to degrade the nascent transcript and terminate the transcribing polymerase [51–58] . We therefore predicted that snoRNAs possessing 3' Rnt1 cleavage sites would not exhibit readthrough in npl3Δ strains . Unexpectedly , however , there was no apparent correlation between readthrough transcription in the npl3Δ strain and the presence or absence of reported Rnt1 cleavage ( S4 Table ) . No extension was seen on any of the RNAPII transcribed snRNAs ( U1 , U2 , U4 or U5 ) in the npl3Δ strain ( Table 2 ) . It had appeared that snRNAs and snoRNAs utilize related termination pathways [59] and a recent study found extended forms of both snoRNAs and snRNAs in strains lacking Rrp6 [31] . Furthermore , as for snoRNAs , Rnt1 cleavage sites flank the U1 , U2 , U4 and U5 genes [50] . However , despite these apparent similarities , there are clear differences in their requirement for Npl3 . Strains lacking Npl3 show transcription readthrough on protein coding genes , on which termination generally requires the cleavage and polyadenylation machinery , and on ncRNA genes that are terminated by the Nrd1-Nab3-Sen1 ( NNS ) complex . The NNS complex is implicated in termination of CUTs , snoRNAs and some mRNAs and physical interactions have been reported between Npl3 and the NNS components [60 , 61] . We therefore investigated whether this complex is properly recruited in npl3Δ . RNA crosslinking by Nab3 was more efficient than by Nrd1 , so we focused our analyses on this protein . To assess recruitment of the NNS complex we applied the CRAC approach to Nab3-HTP . The npl3Δ strain expressing tagged Nab3 grows very slowly ( doubling time 6h ) , indicating a negative genetic interaction . However , Nab3-HTP was well expressed in npl3Δ and crosslinked to RNA with even greater efficiency than in the WT ( S5A Fig ) . Crosslinking of Nab3 to different RNA classes was similar in npl3Δ and WT strains ( S5B Fig ) . Nab3 , like Npl3 , binds strongly at the 5' ends of mRNA transcripts ( S5C Fig ) and showed a substantial frequency of non-templated oligo ( A ) tails ( 36% in two experiments ) consistent with active surveillance in this region . Inspection of the VHR2-THO1 convergent gene locus ( Figs 3 and 6A ) revealed strong peaks of Nab3 binding at the 5’ ends of VHR2 and THO1 , reflecting the role of NNS in early termination on protein coding genes . In the npl3Δ strain the peak at the 5’ end of VHR2 was unaltered , whereas the peak on THO1 was lost due to transcription interference . A peak of Nab3 towards the 3’ end of CUT557 presumably reflects the known role of NNS in CUT termination . Notably , this peak was increased when Npl3 is absent , corresponding with the increased CUT557 expression . We conclude that the VHR2-CUT557 readthrough transcripts are likely to be terminated by the NNS pathway rather than by the CPF-CF pathway . Nab3 binding across CUTs was strongly increased in npl3Δ , particularly around the 3' ends of these transcripts and at downstream sites ( Fig 6B ) . The increased binding of CUTs by Nab3 in npl3Δ was greater than the increased RNAPII association we observe in the mutant strain ( Fig 4D ) suggesting that it reflects not only increased expression of these ncRNAs , but additional non-productive recruitment of this surveillance factor to normal degradation substrates . On snoRNAs we observe a contrasting phenotype , with reduced Nab3 binding across the length of the transcript in npl3Δ strains ( Fig 6C ) . Decreased Nab3 association with snoRNAs may be related to the apparent processing defect , since the NNS complex helps promote 3’ maturation by recruitment of the exosome [27] . Overall our Nab3 binding data suggest that readthrough transcripts are targets of the NNS complex , demonstrated by increased binding of Nab3 in the extended region in npl3Δ compared to WT . In the mutant strain we see a shift in Nab3 binding away from processing targets ( snoRNAs ) onto surveillance targets ( CUTs and extended mRNAs ) . This might explain why the npl3Δ/Nab3-HTP strain displays a synergistic growth defect . Efficient recruitment of Nab3 is likely to be more critical in an npl3Δ strain , in which many surveillance targets are produced . Mild interference with recruitment due to the tag might therefore have a negative effect on growth in the npl3Δ background , despite giving no clear phenotype in the WT . Widespread termination defects result in genome-wide expression changes . We next used individual probe intensity data from the tiling arrays to calculate the level of readthrough genome-wide . Three windows were defined for each transcript: DN100 ( 100 nt immediately downstream of the transcript 3’ end ) , DN200 ( 200 nt , starting immediately downstream of DN100 ) , and TRAN ( spanning the entire transcript , except for the first and last 50 nt ) . Median expression values ( normalized probe intensities ) were calculated for each and a “readthrough score” equal to DN200 / TRAN was obtained for each gene in WT and npl3Δ strains . The readthrough scores obtained for the two strains were then used to calculate readthrough ratios , comparing readthrough in the npl3Δ mutant strain to that in WT yeast ( S5 Table ) . A ratio greater than 1 indicates higher readthrough in the npl3Δ mutant strain . All mRNAs , snoRNAs , CUTs and SUTs were considered , with the exclusion of transcripts less than 200 nt in length , or closer than 400 nt to an annotated Ensembl feature on the same strand . S6A Fig shows the distribution of readthrough across all genes in the npl3Δ strain . The dark and light blue lines show the distribution of readthrough ratios for two replicate experiments , alongside the null ratio where WT is compared to WT ( red ) . Strikingly , most genes show some level of readthrough in the npl3Δ strain . The number of genes showing significant readthrough ( false discovery rate = 0 . 05 ) ranged from 29% ( 1165/3961 ) to 37% ( 1468/3961 ) , depending on the experiment . We applied stringent filters ( see Bioinformatics section in Experimental Procedures ) and plotted readthrough ratios for genes passing all filters ( 2234 ) against gene expression ( Fig 7A ) ; 32% of genes showed significant readthrough ( marked red; FDR = 0 . 05 ) , demonstrating a requirement for Npl3 in the termination of a substantial proportion of all RNAPII genes . We observed no clear correlation between readthrough ratio and expression level . We ranked all 2234 genes by readthrough ratio ( S5 Table ) and compared polymerase occupancy around the 3' ends of genes with the highest readthrough rank ( top 200 ) and the control group with a low readthrough rank ( 1200 genes ) . We found that polymerase occupancy downstream of the 3' end is higher in high readthrough genes than low readthrough genes in WT yeast ( Fig 7B ) . This suggests that these genes show a tendency towards readthrough , even in the presence of Npl3 . This effect is more pronounced in the absence of Npl3 ( Fig 7C ) , with a greater accumulation of polymerase downstream of the 3' end of high readthrough genes . We next sought to identify factors that might discriminate high readthrough genes from low readthrough genes . We found that readthrough correlated weakly with gene length . Longer genes were more likely to show readthrough ( S6B Fig ) , consistent with a report showing preferential binding of Npl3 to longer genes [6] . To identify potential motifs , we compared the 3’ regions from all genes in the top and bottom groups based on the readthrough ranking . This identified UAUAUA and UAAAUA motif as strongly over-represented in low readthrough genes ( Fig 7D ) . UAUAUA is the binding site for the pre-mRNA 3’-end processing factor Hrp1 [62] and comparison of the locations of the UAUAUA motifs showed enrichment at the expected location upstream of the pA site in low readthrough genes ( Fig 7E ) . The enrichment of Hrp1 binding sites in genes that do not show readthrough in the absence of Npl3 strongly suggests that direct , efficient recruitment of Hrp1 can bypass the requirement for Npl3 in termination . Gene ontology analysis showed that genes with higher readthrough were enriched for plasma membrane proteins and functions in localization and/or transmembrane transport ( S6 Table ) . This suggests these genes are potentially co-regulated through transcription termination .
Npl3 is bound to all classes of RNAPII transcripts , with enrichment for oligoadenylated RNAs characteristic of nuclear surveillance targets . Deletion of NPL3 revealed its involvement in termination on diverse transcripts that had not appeared to share termination systems . These included many mRNAs and ncRNAs including the CUT class of lncRNAs and most snoRNAs . In contrast , no defects were seen for snRNAs , which have 3’ processing and termination pathways that appeared to closely resemble snoRNAs . Significant transcription termination defects were seen on approximately 30% of protein coding genes in npl3Δ strains . Readthrough was associated with widespread gene expression changes due to transcriptional interference at downstream genes . This likely reflects the disruption of nucleosome positioning and/or transcription factor binding caused by passage of RNAPII through the nucleosome free regions characteristic of yeast promoters . The precise number of genes that are inhibited by this mechanism is difficult to determine accurately . In the cases of the convergent genes highlighted in the text , the phenotype is clear because the transcripts lie on opposite strands . However , actively transcribed , convergent genes are quite rare in yeast , and transcriptional interference on tandem genes may be less evident . Downstream gene expression may appear unaffected on microarrays , despite generating little functional mRNA , with downstream signal representing extended upstream gene products . From the RNAPII CRAC data it appears that sense-orientated genes some distance from a site of readthrough can display the hallmarks of decreased expression . This was shown , for example , by the decreased RNAPII peak at the 5’ end of the HOR2 gene , located downstream of the extended VHR2 transcript ( Fig 3 ) . The widespread interference seen in the absence of Npl3 highlights the necessity for very efficient release of RNAPII at the 3’ ends of genes . In general , fold changes in RNAPII occupancy were less marked than changes in downstream transcript levels . This indicates that readthrough by a small number of polymerases can drastically alter the regulation of gene-expression . In the case of snoRNAs , it appears that low levels of transcription readthrough , as determined by accumulation of downstream RNAPII , result in high levels of extended transcripts . Normal snoRNA termination and processing require the NSS complex , which stimulates exosome recruitment [27] , and Nab3 association with snoRNAs was reduced in npl3Δ strains . These observations strongly indicate that loss of Npl3 also leads to defects in snoRNA 3’ processing and/or surveillance of 3’ extended species . The relative contributions of impaired snoRNA processing versus impaired surveillance in npl3Δ mutants is difficult to assess—as is the case for many substrates for nuclear surveillance/processing factors . Distinguishing the contributions of processing and surveillance is not generally feasible when the phenotype is accumulation of extended species at steady state , and will require the development of very fast , in vivo kinetic analyses . Termination defects seen in the absence of Npl3 were restricted to RNAPII . However , while diverse classes of RNAPII transcripts are affected , this was not the case for all transcripts of any class . To try to understand what determines this apparent variability in the requirement for Npl3 , we ranked protein-coding genes by their degree of readthrough ( readthrough ratio ) in the absence of Npl3 , and sought correlated features in protein coding genes . A notable correlation was with the elevated presence of consensus , UAUAUA binding sites for the mRNA 3’ cleavage factor Hrp1 in the 3’ regions of transcripts with low readthrough scores ( i . e . with low dependence on Npl3 for termination ) . We postulate that association of Hrp1 and/or other cleavage factors with the pre-mRNA is normally promoted by Npl3-mediated packaging , but this requirement can be alleviated by the presence of high-affinity RNA-binding sites . In contrast , competition between binding of Npl3 and pre-mRNA cleavage and polyadenylation factors including Hrp1 was previously reported for GAL reporter constructs [7 , 11] . This apparent anti-termination activity of Npl3 is the opposite of our general findings . However , it could readily be envisaged that on individual genes , Npl3 binding sites conflict with the association of specific factors . The GAL genes are not expressed under the conditions used in our analyses , making it difficult to determine whether these effects are also seen on the endogenous genes . Readthrough ratio was weakly correlated with gene length , with longer genes more likely to exhibit termination defects when Npl3 was absent . Preferential association of Npl3 with longer transcripts as been reported [6] , suggesting that these may show greater changes in pre-mRNA packaging in its absence . However , we saw no clear length dependence for Npl3 in termination on ncRNAs , which are generally shorter than mRNAs . Several distinct , but overlapping pathways for RNAPII termination are normally used by transcripts that are extended in the absence of Npl3 . On pre-mRNAs , recognition of the cleavage and polyadenylation site is linked to changes in the transcribing polymerase that make it prone to termination at downstream pause sites . This may involve Tyr1 dephosphorylation in the CTD by the Glc7 phosphatase that associates with the CPF-CF [63] . Loss of Tyr1P promotes binding of the cleavage factor Pcf11 , as well as Rtt103 , which in turn recruits the Rai1/Rat1 complex for the “torpedo” termination pathway . In contrast , termination of a wide range of ncRNA transcripts involves the Nrd1/Nab3/Sen1 ( NNS ) complex , which binds to the nascent transcript and to the RNAPII CTD with Ser5P modification , as well as the TRAMP nuclear surveillance complex and promoter proximal nucleosomes with H3K4 trimethylation ( [18 , 22 , 34 , 36 , 59] reviewed in [64] ) . Other termination mechanisms are initiated by co-transcriptional cleavage by the RNase III homologue Rnt1 [52 , 65] and by formation of a transcription elongation “roadblock” due to Reb1 binding on the DNA [66] . We found no correlation between known Rnt1 or Reb1 targets and transcription readthrough in npl3Δ strains . Binding of Nab3 to the CUT lncRNAs was increased in npl3Δ strains . A simple , potential explanation might be that the absence of the , normally very abundant , Npl3 protein frees binding sites that can be occupied by other factors , including Nrd1-Nab3 . However , the abundance and readthrough of CUTs were also increased in the absence of Npl3 , and this may contribute to the apparent changes in Nab3 association . We propose that loss of Npl3 results in aberrant RNP formation that still permits Nab3 recruitment , but binding may be non-productive . Npl3 was reported to directly stimulate RNAPII elongation and a mutant that disrupts this function , npl3-120 , resulted in improved termination . The slower RNAPII elongation rate in npl3-120 strains may enhance termination by increasing the time available for recruitment of 3' end processing factors such as Hrp1 . In contrast , an Npl3 mutant ( S411A ) that blocks a phosphorylation site was associated with impaired transcription termination [67] . This defect was proposed to arise from retention of the mutant Npl3 in association with the RNAPII CTD and the mRNA . However , the list of genes showing 3’ extension in Npl3S411A strains overlaps substantially with the genes showing RT in npl3Δ , indicating that Npl3 retention is not solely responsible for this phenotype . Of the 818 genes showing 3’ extension in Npl3S411A , 143 overlap with the 614 genes showing significant readthrough in npl3Δ ( p-value 7 . 1e-16 , Fisher’s exact test ) . Npl3 is a highly abundant RNA binding protein that participates in many processing events and associates with all nascent RNAPII transcripts . It seems probable that its absence will result in substantial changes in the nascent RNP structure . We speculate that such inappropriately packaged RNA is associated with downstream defects in transcription termination , reflected by changes in binding by the termination factor Nab3 , consequently impairing a remodeling event that promotes removal of the polymerase from the nascent transcript .
Yeast were grown in standard SD medium at 30°C unless otherwise stated . Strains and plasmids used are listed in S7 Table . All oligonucleotides used are listed in S8 Table . All yeast analyses were performed in strains derived from BY4741 ( MATa; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0 ) or , in the case of N-PTH-NPL3 , BY4727 ( MATalpha; his3Δ200; leu2Δ0; lys2Δ0; met15Δ0; trp1Δ63; ura3Δ0 ) . N-PTH-NPL3 is a strain in which a sequence encoding a PTH ( proteinA-TEV-His ) tag was integrated at the 5' end of NPL3 , resulting in the formation of an N-terminally tagged protein utilizing the endogenous NPL3 promoter . As the protein is N-terminally tagged in this strain , the orientation of the tag is reversed , allowing the order of protein purification steps to be retained . Generation of this strain involved inserting a URA3 marker between the NPL3 promoter and the NPL3 ORF , and then replacing the URA3 marker with a sequence encoding the PTH tag . The second PCR , amplifying the PTH tag , was performed on a plasmid expressing N-PTH-NPL3 ( pRS415-NPL3-PTH ) , and amplified a region running from the start of the PTH tag to ~600 nt into the NPL3 ORF to increase integration efficiency . The CRAC procedure involves purifying protein/RNA complexes , where the RNA has been covalently UV crosslinked to the protein [41] . RNA-protein complexes are purified , and RNAs are partially digested to leave only the 'footprint' bound by the protein . Linkers are then ligated to both ends and the protein is removed by proteinase K digestion . RNAs are reverse transcribed and resulting cDNAs subjected to next generation sequencing using the Illumina platform ( Edinburgh Genomics ) . WT and npl3Δ yeast were grown to mid-log phase ( OD600 ~0 . 5 ) and cells were collected by brief centrifugation ( 3000 xg , for 5 min ) . Total RNA was isolated by a standard acidic hot phenol method and DNA was removed by treating with RNase-free DNaseI ( Turbo DNA-free kit; Ambion ) . Reverse transcription and array hybridizations were carried out as previously described [68] . Yeast cultures were grown to mid-log phase ( OD600 ~0 . 5 ) and cells were collected by brief centrifugation ( 3000 xg , for 5 minutes ) . Total RNA was isolated by a standard acidic hot phenol method and DNA was removed by treating with RNase-free DNaseI ( Turbo DNA-free kit; Ambion ) . Single stranded cDNA was generated using gene specific primers , designed to prime from the 3' end of the transcript ( to measure expression ) or from ~500 nt downstream ( to measure transcriptional readthrough ) . Reverse transcription reactions were performed using Superscript III ( Invitrogen ) . The expression level of individual transcripts was determined by quantitative PCR using SYBR green fluorescence for detection . Relative quantities were calculated using a standard curve made with known concentrations of genomic DNA , and were normalized to levels of ACT1 in each RNA sample . Total RNA was isolated by a standard acidic hot phenol method . For SNR60 readthrough analysis , equal amounts of RNA ( 10 μg ) were resolved on a 1 . 2% agarose gel in TBE buffer and transferred onto Hybond N+ nitrocellulose membrane overnight in 6x SSC . For detection of mature snoRNAs and RNase H cleavage assay products , samples ( 4 μg total RNA for snoRNA detection ) were resolved on an 8% acrylamide gel containing 8 . 3 M urea , in TBE buffer and transferred onto Hybond N+ nitrocellulose overnight in 0 . 5x TBE . Oligo probes were end labeled with [γ-32P] ATP and hybridized to the membrane overnight at 37°C in ULTRAhyb-Oligo ( Ambion ) . Signals were detected using a Fuji FLA-5100 . Samples ( 30 μg ) of RNA were annealed with 750 ng oligo-dT and/or 10 pMoles gene-specific oligo , heated to 65°C and allowed to cool slowly to 30°C . Samples were then incubated with 1 unit RNase H ( Roche ) at 30°C for 1 hour . Total extract from crosslinked CRAC samples were loaded onto 4–12% NuPAGE gels and Transferred onto Hybond C nitrocellulose membrane . Following blocking in 5% milk , the membrane was incubated first inn anti-TAP primary antibody ( 1:5000 overnight ) and then anti-rabbit secondary ( 1:10000 for 1 hour ) . Signal was visualized using the Licor Odyssey system . All sequence data are available from GEO under accession number GSE70191 . http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? token=gdivgqmivxcpzkp&acc=GSE70191 Sequencing data were processed and quality filtered using the fastx toolkit as previously described [37] . Processed reads were mapped to the Saccharomyces cerevisiae genome ( SGD v64 ) using Novoalign ( Novocraft ) with genome annotation from Ensembl ( EF4 . 74 ) , supplemented with non-coding sequences as previously described [37] . Reads mapping to different transcript RNA classes were determined using the pyCRAC package [17] ( Figs 1A , 1C , S3B and S5B ) . All analyses were performed using genome SGD v64 unless otherwise stated . The distribution of hits across transcripts of different classes was determined in several ways . Firstly , to examine the distribution of proteins at the 5' and 3' ends of mRNAs , hits within 300–900 nt windows aligned to the start ( TSS ) and end ( pA ) were plotted using published scripts [37] . The top 2000 bound mRNAs for each protein were included in the analysis and average distribution was plotted ( Fig 1B and 1D ) . A similar analysis was performed to assess binding at snoRNAs and CUTs ( Figs 1E , 1F , 4D , 5A , 6B and 6C ) . In this instance smaller windows were used and hits per million mapped reads were plotted , rather than average distribution . Reads were aligned to the TSS or 3' ends , with flanking regions included as shown . We included all snoRNAs in the analysis , but only included CUTs > 150 nt in length . Hits at introns were also plotted using this approach ( S1C and S1D Fig ) . As an alternative way to assess binding across transcripts , we used pyBinCollector from the pyCRAC package , which normalizes transcripts by length , dividing hits into a given number of bins ( S1E , S1F , S3C , S3D , S5C and S5D Figs ) . Rpo21 occupancy was calculated to determine transcriptional readthrough ( Figs 3C , 4C , 5D , S2C and S4C ) using pyPileup from the pyCRAC package , with default settings . Hits containing unencoded 3' oligoA tails of 2 of more were determined using a reported pipeline [37 , 69] . These hits were then mapped to transcript groups and plotted across RNA classes as described above ( Fig 1C and 1D ) . All microarray data are available in the ArrayExpress database ( http://www . ebi . ac . uk/arrayexpress ) , under accession number E-MTAB-3642 . Array data can also be visualized in a genome browser heat map format ( http://steinmetzlab . embl . de/tollerveyLabArray ) . Microarray data were aligned to SGD S . cerevisiae genome version ( SGD v57 ) . Normalization of microarray hybridizations was performed as previously described [70] and transcript boundaries shown are as published [71] . Differential expression analyses were carried out using the R-package , Limma [72] , controlling for the false discovery rate arising from multiple testing [73] . Five snoRNAs were not included in the differential expression analyses due to lack of transcript boundary information ( Fig 2 and S5 Table ) . These can , however , be viewed in the genome browser heat map . CRAC hit data were aligned to SGD S . cerevisiae genome version ( SGD v57 ) alongside tiling array expression data at individual loci ( Figs 3A , 4A , 4B , 5B , S2A and S4A ) . Hits were normalized for library size by plotting hits per million mapped reads at each nucleotide . Fig 6A shows CRAC data aligned to SGD v57 without array data . Readthrough scores were calculated for mono-exonic snoRNAs , mRNAs , CUTs and SUTs with coordinates previously defined [71] . Transcripts that are < 200 nt were excluded , as were transcripts < 400 bp upstream of another annotated transcript [71] or in Ensembl release 68 . The exception to this is when an mRNA has an annotated CUT or SUT immediately downstream of the 3' end . In some instances , these annotated ncRNAs appear to correspond to upstream mRNA readthrough , and therefore these mRNAs were not filtered out . Three windows were defined for each transcript: DN100 ( 100 nt immediately downstream of the transcript 3’ end ) , DN200 ( 200 nt , starting immediately downstream of DN100 ) , and TRAN ( spanning the entire transcript , except for the first and last 50 nt ) . The median normalized probe intensities ( in log2 space ) for each microarray sample were calculated for each window , although windows with < 8 probes were excluded . The readthrough score was then defined for each gene and each sample ( wild-type replicate 1 , wild-type replicate 2 , npl3-delta replicate 1 , and npl3-delta replicate 2 ) as the median intensity for DN200 , minus the median intensity for TRAN . The difference in npl3-delta and wild-type transcriptional readthrough was determined by calculating a readthrough ratio for each gene , defined as the readthrough score for npl3-delta minus the readthrough score for wild-type . Readthrough ratios were also calculated for wild-type replicate 2 versus wild-type replicate 1 , to provide an empirical null distribution and enable transcripts with a significant increase in readthrough for npl3-delta versus wild-type to be identified . The Benjamini–Hochberg procedure was used to control the false discovery rate at 0 . 05 . For this step , the two replicate experiments were treated separately , then a stringent list of genes with elevated readthrough obtained by intersecting the results from both replicates . A series of filters was used to exclude transcripts for which readthrough ratios may be inaccurate , either due to low expression or because of evidence of independent transcription initiation downstream . The following criteria were used: ( i ) there must be < 10 Cbc1 ( cap-binding complex protein 1 ) CRAC reads in the DN100 window , ( ii ) TRAN median probe intensity must be > -4 . 88 for wild-type and npl3-delta , ( iii ) for npl3-delta , the median probe intensity in the DN100 window must be > 70% that of the TRAN window , and ( iv ) the median probe intensity in the TRAN window for npl3-delta must be at least 70% that of the same window in the wild-type sample . For filters ( ii ) - ( iv ) , the mean of the two replicates was used . Plots of Pol II distribution in regions centered on transcript 3’ ends were obtained by taking the individual Pol II CRAC read distributions for each gene , linearly transforming each gene so that its maximum value was equal to 1 , and then summing at each nucleotide for the indicated set of genes ( either high or low readthrough groups ) . We observed that genes with the very lowest readthrough ranks had a spurious negative readthrough ratio due to having increased expression in the npl3Δ strain relative to WT . To limit the contribution of these genes , we took a larger number of genes for the low readthrough group ( 1200 compared to 200 ) . Npl3 binding sites were analyzed for enriched motifs by first filtering total reads to exclude low complexity sequences , as previously described [37] . The pyCRAC package [17] was used to calculate statistical overrepresentation scores for every possible k-mer ( S1D Fig ) using a previously described algorithm [69] . We used pyCRAC to calculate False Discovery rates ( FDRs ) and selected only reads forming clusters of 5 reads or more with an FDR < 0 . 05 for further analysis . Reads were further filtered to include only those with one or more T-C substitution , representing a site of crosslinking , and therefore predicted to indicate genuine binding sites with greater stringency . To identify sequence motifs that differentiate high- and low-readthrough genes , we considered the 1822 genes for which reliable readthrough scores could be established , and separated these genes into quartiles by their readthrough scores . 2234 genes were included in the genome-wide readthrough analysis , but the bottom 250 were excluded from the motif analyses as these were found to have spuriously low readthrough ratios resulting from increased expression in the npl3Δ mutant . Of the remaining 1984 genes , only those with well-defined polyA sites ( 1822 ) were included in the motif analysis . For each 6-mer nucleotide motif , we calculated the numbers of genes in each quartile that contained the motif within the region ( -80 to -20 nucleotides ) from the polyadenylation site ( polyA site ) . The polyA site was defined from Pab1 CRAC data as described [37] . We then identified the motifs that were significantly enriched in the low-readthrough genes , relative to high-readthrough genes , by calculating Z-scores as described [69] . To illustrate the localization of motifs relative the polyA site , we plotted the total coverage of UAUAUA motifs as a function of distance from the polyA site , separately for the top and bottom quartile of genes ranked by readthrough scores . | Npl3 is a yeast mRNA binding protein with many reported functions in RNA processing . We wanted to identify direct targets and therefore combined analyses of the transcriptome-wide effects of the loss of Npl3 on gene expression with UV crosslinking and bioinformatics to identify RNA-binding sites for Npl3 . We found that Npl3 binds diverse sites on large numbers of transcripts , and that the loss of Npl3 results in transcriptional readthrough on many genes . One effect of this transcription readthrough is that the expression of numerous flanking genes is strongly down regulated . This underlines the importance of faithful termination for the correct regulation of gene expression . The effects of the loss of Npl3 are seen on both mRNAs and non-protein coding RNAs . These have distinct but overlapping termination mechanisms , with both classes requiring Npl3 for correct RNA packaging . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Loss of the Yeast SR Protein Npl3 Alters Gene Expression Due to Transcription Readthrough |
The epistatic interactions that underlie evolutionary constraint have mainly been studied for constant external conditions . However , environmental changes may modulate epistasis and hence affect genetic constraints . Here we investigate genetic constraints in the adaptive evolution of a novel regulatory function in variable environments , using the lac repressor , LacI , as a model system . We have systematically reconstructed mutational trajectories from wild type LacI to three different variants that each exhibit an inverse response to the inducing ligand IPTG , and analyzed the higher-order interactions between genetic and environmental changes . We find epistasis to depend strongly on the environment . As a result , mutational steps essential to inversion but inaccessible by positive selection in one environment , become accessible in another . We present a graphical method to analyze the observed complex higher-order interactions between multiple mutations and environmental change , and show how the interactions can be explained by a combination of mutational effects on allostery and thermodynamic stability . This dependency of genetic constraint on the environment should fundamentally affect evolutionary dynamics and affects the interpretation of phylogenetic data .
As pointed out by Sewall Wright in the 1930's , the genetic makeup of a biological system should determine not only current functionality but also affect future evolutionary change [1] . How the present genetic architecture constrains future adaptive evolution is now starting to be addressed experimentally [2]–[4] . By systematically constructing single-mutant neighbors and assaying their function or fitness , proteins ranging from TEM β-lactamase [3] to steroid receptors [5] have been shown to exhibit sign epistasis , in which one mutation can be beneficial or deleterious depending on the presence of another mutation . Sign epistasis by itself does not imply evolutionary constraint , as the interacting mutations may simply not play a role in adaptation . However , when mutations essential for functional innovation exhibit sign-epistasis , constraints emerge for evolutionary trajectories that depend on fixing one adaptive mutation after another by positive selection [6] . For sign-epistatic interactions , the number of such adaptive trajectories is reduced . Two mutations may also be deleterious individually but jointly beneficial , as observed for mutations in the regulator MTH1 and glucose transporters HXT6/HXT7 in Saccharomyces cerevisiae [7] and between argH12 and pyrA5 mutants leading to arginine and pyrimidine deficiency in Aspergillus niger [8] . Such reciprocal sign epistasis is a necessary condition for multiple peaks in the fitness landscape [9] , which can completely block evolutionary trajectories in which mutations are fixed one-by-one by positive selection . Because of this ability to arrest , delay , and divert evolution , genetic interactions have been speculated to play a central role [10] in speciation [11] , [12] , the maintenance of biodiversity [13] , and developmental evolution [14] , [15] . So far , epistastic interactions have been studied predominantly for environments that are constant in time and favor a single function or phenotype . However , natural environments are characterized by irregular temporal changes , which in turn impose temporally changing demands on the expressed phenotypes . Indeed , the complexity of regulatory systems is considered to have evolved in response to environmental heterogeneity [16] , [17] . Experimentally , mutations are commonly observed to have different effects in different environments [18]–[20] . For example , in Escherichia coli the fitness effects of single Tn10 transposon insertion mutations [21]and mutations conferring resistance to bacteriophages λ and T4 have been shown to depend on the genetic background and the environment [22] . Correlations exist between epistatic interactions in plant viruses and their hosts [23] , and trade-offs have been observed between the effect of mutations in the presence of certain types or concentrations of antibiotics in Escherichia coli [24] , [25] and Pseudomonas aeruginosa [26] . These observations raise the question to which extent constraints themselves change when the environment changes . If mutations essential to functional innovation exhibit sign-epistatic interactions that are modulated by environmental change , adaptive trajectories will be drastically affected . For instance , evolutionary change hampered by adaptive valleys in one environment could be opened up to positive selection in another . Conversely , trajectories that can be positively selected for in constant environments [2] , [3] could be blocked by environment-induced sign epistasis , which could slow down overall evolutionary progress or drive adaptation to dead ends in genotype space . This environmental control over the accessibility of adaptive trajectories goes beyond merely defining a variable selective environment , and would invalidate commonly held assumptions in analyzing the historical evolutionary record by phylogenetic reconstruction ( 23 ) . These elementary issues can be readily investigated using a simple phenotype that responds to the environment . We focused on one of the most well-understood model systems for environmentally controlled gene expression , the Escherichia coli lac repressor LacI [27] . We considered the evolutionary transition to a variant that exhibits an altered regulatory response [28] . In the presence of the wild-type repressor , LacIwt , the lac operon is induced by the ligand IPTG , whereas in the presence of the variant LacIinv , expression is suppressed by IPTG . We have previously isolated LacI variants with such inverse phenotypes in evolutionary experiments [28] ( Text S1 ) , which serve as a basis to systematically assess how the environment affects epistasis between the mutations required for inversion . We find that the epistasis is highly environment-dependent , which implies that epistasis perceived in a constant environment does not properly inform on the evolutionary constraints in a variable environment . We can explain the generic pattern of higher-order genotype x genotype x environment interactions that is observed in all three variants using a simple model of changes in the allosteric transition and in protein stability .
To investigate the interplay between the environment and epistasis we focused on three inverse LacI variants [28] ( Text S1 ) . The three inverse variants each contained three to six point mutations relative to LacIwt . For all variants , three mutations appeared essential for the inverse function , as was determined by engineering lacI variants that contained sub-sets of these mutations . We denote these three inverse variants as LacIinv1 ( S97P , R207L , T258A ) , LacIinv2 ( S97P , L307H , L349P ) and LacIinv3 ( S97P , G315D , P339H ) . Note that all share the mutation S97P . Next , we constructed all the single and double mutants , and assayed the operon expression phenotypes in the absence of IPTG ( Env0 ) and in the presence of 1 mM IPTG ( Env1 ) ( Table S1 ) using a fluorogenic reporter assay ( materials and methods ) ( Figure 1A ) . Given the evolutionary objective of inversion , a high operon expression level is favored in Env0 , whereas a low expression level is favored in Env1 [28] ( Figure 1B ) . To compare the epistasis in each environment , we classified the epistatic ( genotype x genotype ) interactions for all pairs of mutations for each of the three inverse LacI variants . We distinguished three categories: magnitude epistasis ( M ) - both mutations are either beneficial or deleterious , irrespective of the genetic background , sign epistasis ( S ) – the effect of one mutation changes sign depending on the genetic background , or reciprocal sign epistasis ( R ) - both mutations are individually deleterious , but beneficial in combination [4] . Neutral mutations are not positively selected and are thus grouped under deleterious . We find that nine out of the eighteen mutation pairs display the same category in environments Env0 and Env1 ( Table 1 ) . For instance , in the P349 background , L307H and S97P exhibit sign epistasis in both environments ( Table 1 , LacIinv2 ) . Note that for all these nine pairs , the magnitude of the mutational effect does depend on the environment , but the sign does not . For the other nine mutation pairs , the category of epistasis differs between the two environments ( Table 1 ) . Some sign epistatic interactions are switched ‘off’ by the addition of IPTG . In the P97 background for instance , IPTG induces a sign change in the effect of R207L; it transforms the sign-epistasis between R207L and T258A in Env0 to magnitude epistasis in Env1 ( Table 1 , LacIinv1 ) . Sign epistasis is turned ‘on’ between other mutations . For instance , in a P97 background , L349P and L307H exhibit sign epistasis in an environment without IPTG , and reciprocal sign epistasis with IPTG ( Table 1 , LacIinv2 ) . Thus , environmental signals modulate sign-epistatic interactions between residues involved in the functional inversion of LacI . The above classification of genetic interactions into categories reveals a dependence on the environment , but it does not offer intuitive insights into their causes . These dependencies may also be viewed as three-way interactions between two genetic changes and one environmental change . Hence , they can be denoted as genotype x genotype x environment interactions , or briefly GxGxE; analogous to two-way GxG interactions between two genetic changes in a single environment , or two-way GxE interactions between one genetic change and one environmental change [17] . To analyze these higher-order interactions , we introduced a graphical method ( Figure 2A ) . Mutations are represented as vectors in a two-dimensional coordinate system , where the axes indicate the corresponding changes in expression phenotype in both environments . A vector pointing to quadrant I signifies functional improvements in both environments , whereas quadrants II and IV denote improvement in one environment and deterioration in the other , and quadrant III denotes deterioration in both . The probability of fixing neutral mutations is low compared to positively selected mutations that confer functional improvements [6] , [29] . Mutations that are neutral in both environments therefore correspond to quadrant III , while mutations that are neutral in one environment and beneficial in the other correspond to quadrants II or IV . Thus , mutations in quadrants II and IV indicate sign-changing GxE interactions . Higher-order interactions between two or more mutations and the environment can be visualized by sets of paths composed of two or more mutational vectors ( Figure 2 ) . The two mutational paths from genotype ab to AB ( via Ab or via aB ) form a four-sided polygon . The polygon is a simple parallelogram in the absence of any genetic interactions , which may occur either without ( Figure 2B ) or with GxE interactions ( Figure 2C ) . Deviations from the parallelogram indicate genetic interactions , or epistasis . Vectors at opposing sides of the polygon that have different angles but point in the same quadrant indicate magnitude epistasis . Opposing vectors pointing in different quadrants indicate sign-epistatic interactions ( GxG , Figure 2D ) , and when the sign change of opposing vectors is conditional on the environment higher-order GxGxE interactions can be observed ( GxGxE , Figure 2E ) . Thus , higher-order interactions between mutations and the environment can be graphically recognized and classified using the mutational vector plots . We analyzed the interactions for the three LacI variants by displaying the expression data as mutational vectors in Figure 3A , B and C . Because the transition to inversion is characterized by a decreasing operon expression in the presence of IPTG ( Env1 ) and an increasing operon expression in the absence of IPTG ( Env0 ) , we plotted 1/expression in Env1 against the expression in Env0 , such that the closer the phenotype comes to the objective of inversion , the more it moves towards the upper-right corner of Figure 3 . Inspection of the polygon shapes shows that half ( 50% ) lack the signatures of sign-changing higher-order interactions involving mutation pairs and the environment . For instance in Figure 3C , the opposing red and green vectors in the P97 background point in the same quadrant . The polygon is tilted , with both red vectors pointing in quadrant IV , indicating GxE interactions . However , the other half of the opposing mutational vector pairs in the polygons do not point in the same quadrant , indicating the pervasive presence of higher-order GxGxE interactions . For instance , in the P97 background , the addition of T258A turns the green vector ( R207L ) from quadrant III to IV , which is caused by the fact that R207L is neutral in the presence of IPTG and the absence of T258A , but increases expression by 20-fold in T258A's presence ( Fig . 3A ) . Another example is the addition of L307H , which rotates the red vector ( L349P ) from quadrant IV to II in the P97 background , which indicates that the effect of L349P on expression changes sign in both environments due to L307H ( Figure 3B ) . Overall , the pattern displayed by the three variants in the vector plots ( Figure 3A , B and C ) is strikingly similar , in contrast to the diverse environmental dependence of epistasis seen in Table 1 . The blue vectors initially point predominantly up along the Env1 axis ( the expression level decreases with IPTG ) , as the expression level in Env1 is strongly decreased , but turn diagonally to the upper-right corner when the red and green mutations are added ( the expression level increases simultaneously in the absence of IPTG ) ( Figure 3 ) . On the other hand , the green and red vectors either point downward along the Env1-axis , ( expression mainly increases in the presence of IPTG ) , or to the right along the Env0-axis ( expression increases in the absence of IPTG ) . Mutation S97P appears responsible for this rotation of the red and green vectors: in the LacIwt background they point along Env1 , while in the P97 background they point along Env0 . In other words , S97P represents a ‘switch’ that changes the interaction of the red and green mutations with the environment . This pattern is identical for all three inverse genotypes; all show a roughly similar rotation for the blue as well as for the red and green vectors . Thus , while the genetic solutions to the phenotypic inversion are different in the three variants , the main features of the underlying map of the interactions between genotypes and the environment are general . Note that one may also consider the presence of higher-order interactions that are purely genetic . Specifically , such GxGxG interactions arise when the addition of a third mutation changes the category of the two-way epistatic motif . For instance , in the wild type background , both green ( L307H ) and red ( L349P ) vectors point downward or are neutral along the Env1 axis ( Figure 3B ) , and hence point to magnitude epistasis . However , upon the application of S97P ( Figure 3B , blue vectors ) , one green and one red vector still points down , but one green and one red vector is rotated upwards . Thus , L307H and L349P display reciprocal sign epistasis in the presence of P97 , and hence their three-way interaction in Env1 cannot be captured by two-way epistasis alone . Note that this GxGxG interaction itself may in turn be dependent on the environment , indicating GxGxGxE interactions . Among other things , the presence of higher-order genetic interactions illustrates that conclusions on the accessibility of a genotype must be carefully considered . This is particularly relevant when it is unclear to what extent the mapped genotype space fully determines the considered function , as an untested mutation could open up mutational pathways to selection , which otherwise may have been considered blocked [30] . The principle of such effects of higher-order genetic interactions have previously been captured [3] , [4] , [7] , [15] , [31] when mapping a larger landscape and assessing the mutational pathways within it . Nonetheless , the explicit presence of GxGxG interactions underscores the care that must be taken when formulating conclusions about selection and constraint from fitness landscapes . The results also underscore that mechanisms that are comparatively simple on the molecular level , can give rise to GxE interactions . For instance , in the P97 background , L307H has the simple mechanistic effect of generally increasing expression both in the presence and absence of IPTG . In terms of selection , this change is beneficial in one environment ( in the absence of IPTG ) , and deleterious in the other ( in the presence of IPTG ) . Hence , L307H gives rise to a GxE interaction , a trade-off . Given the generic purpose of regulatory functions to modulate biological functions in response to input signals , one can expect such trade-offs that originate from simple molecular mechanisms to be rather generally present . The observed generality of the genotype-environment interaction maps ( Figure 3 ) suggests that they result from a generic structural cause . However , the positions of the mutated residues within the LacI crystal structure do not directly reveal generic features , as they appear scattered throughout the structure , with different locations for the different variants ( Figure S2 ) . Also , the mutations are not positioned at obvious functional sites such as the DNA or ligand binding regions . Alternatively , the origin of the interactions may be rooted in the mechanism of inversion , which has been speculated to be based on two effects [28] , [32] . First , the allosteric transition from high to low operator affinity is thought to be impeded by S97P , as P97 cannot form the transient bond with K84′ and V94′ [33] , which in turn locks the structure in the DNA-bound confirmation [34] , [35] . Second , the response to inducer is assumed to be inverted through changes in the thermodynamic stability of the protein: the additional two mutations in each variant would lower the stability in the absence of IPTG , which would confer an increased expression level in Env0 , while the binding of the ligand IPTG to LacI would confer a stabilizing effect that conserves a low expression level in Env1 . Our experiments showed that in a LacIwt background , S79P lowers expression in Env1 to repressed levels while maintaining a relatively low expression level in Env0 . Thus , these data are indeed consistent with the proposed locking of LacI in the DNA-bound confirmation . The data further show that expression in Env1 varies along the mutational trajectories from LacIwt to LacIinv ( Figure 4A ) . In contrast , in Env0 , the trajectories to inversion show a generic increasing trend in the expression level; all first mutations yield little to no changes , while second and third show increasingly large expression increases ( Figure 4B ) . The pattern of changes in expression level in both environments is consistent with stability-decreasing mutations , as: 1 ) correlation between the stability and the expression level should be stronger in Env0 , as the ability to tightly bind DNA in that environment is dependent on structural stability , in contrast with the ability to efficiently release from the DNA in Env1 , and 2 ) it has been argued that protein function is robust against initial stability decreases , but can be expected to deteriorate when accumulated mutations drive the system across their so-called stability threshold [36]–[38] . We investigated the destabilizing effect of the mutations by analyzing the stability changes due to amino acid substitutions in silico with FoldX [39] , [40] . In the absence of IPTG ( Env0 ) , FoldX indeed showed significant stability decreases for most ( 8 out of 11 , Table S2 ) of the studied mutants , including S97P . The expression measurements suggest that in particular S97P brings LacI to the edge of the stability threshold , as subsequent mutations strongly increase expression ( Figure 3 , Env0 ) . Thus the S97P substitution acts as a switch that systematically alters the phenotypic effect of the other mutations . While we have addressed the central features of the interaction map , various more detailed interactions between mutations and the environment remain to be explained mechanistically . However , overall the analysis indicates that the combined effects of two independent and simple molecular mechanisms can explain complex higher-order GxGxE interactions between multiple mutations and the environment .
Recent systematic reconstructions of evolutionary intermediates have provided a first view on adaptive landscapes and the causes of evolutionary constraint [4] . Sign epistatic interactions between mutations have been shown to limit the number of mutational trajectories that can be followed under positive selection in constant environments [2] , [3] . Directed evolution experiments revealed evolutionary constraints that delay or prevent adaptation [15] , [28] , and measured trade-offs between environments indicated how such constraints affect selection in variable environments [28] , [41]–[43] . Here we investigated how the environment affected the adaptive landscape describing a specific functional innovation , by reconstructing the evolutionary intermediates on route to three different inverse LacI genotypes . The three evolved genotypes indicated a redundancy within the LacI genetic architecture to develop regulatory functions that respond to the environment , mirroring similar results obtained for microbial populations evolving in constant environments [44]–[46] . We found that a mechanistic model of inversion provided an explanation for the origin of this parallelism . First , a mutation ( S97P ) blocks the IPTG-induced allosteric transition , and thus affects expression only in the presence of IPTG . Second , the initial mutations have little effect on the ability to repress in the absence of IPTG , while later mutations have a large effect . Third , binding to the ligand IPTG increases the protein stability and hence the ability to repress . Thus , a combination of simple molecular mechanisms can explain the observed complex higher-order interactions between multiple genetic changes and an environmental change . The data showed that the genetic epistasis in LacI was pervasively dependent on the environment . As the studied genetic changes were not chosen randomly but jointly confer a novel regulatory response , these results inform on constraints in the evolution of a novel biological function . They indicate that limitations in the selective accessibility of trajectories , as detected in a constant environment , not properly inform on evolutionary limitations in the natural variable environment . Due to the environmental dependence of epistasis , some trajectories are closed-off by environmental change while others are opened-up to positive selection . Intriguingly , a consequence of environmental dependence of epistasis is that few mutations are blocked in all environments , and many are positively selected in at least one environment . This suggests that genetic constraints may be more readily overcome in certain variable environments than expected from epistasis detected in constant environments [47] , [48] . More generally , the results underscore the complex and diverse roles of the environment in evolutionary dynamics . The environment does not only define a selective pressure on a phenotypic trait or induce a phenotypic change , but also modulates the underlying genetic constraint . This interdependence has a number of consequences . For instance , it affects our ability to understand the evolutionary record as interpreted from extant genetic sequence data . By modulating evolutionary constraint in time , environmental variations can change substitution rates across evolutionary trees [49] , [50] , referred to as heterotachy , even if selection on a phenotypic trait is constant . It can result in topological inaccuracies in phylogenetic trees [51] such as long-branch biases [52] , [53] and a lack of phylogenetic resolution [52] , [54] if the underlying adaptive landscapes are shaped differently in each of the environments . This can ultimately affect the predictive power of phylogenetic reconstruction techniques in their use for the prognosis of the emergence and the spread of diseases , such as the spread of the influenza virus [55] , where the host can be viewed as a biotic environment [56] . And lastly , it renders a walk on evolutionary branches of life unpredictable and unrepeatable [3] , [57] , as some adaptive trajectories are constrained in some environments , but not in others . It will be intriguing to explore the prevalence of the higher-order genotype x genotype x environment interactions in other biological systems . It is not obvious that all biological functions will show such interactions; in particular those specialized to a single environmental factor . On the other hand , the ability to respond to environmental stimuli is one of the defining properties of living systems . Given the inherent interdependency between regulatory systems and the environment , we expect that such insights into the interplay between genetic architecture and the environment will be crucial for a mechanistic understanding of the evolution of biological functions .
Escherichia coli K12 strain MC1061 [58] , which carries a deletion of the lac operon was used in all experiments . This strain was obtained from Avidity LLC , Denver CO , USA , as electrocompetent strain EVB100 ( containing an additional chromosomal birA ) . Plasmid pRD007 was constructed based on the pZ vector system [59] and contains LacI , driven by the PLO1-Tet promoter . The reporter plasmid pReplacZ , used for the quantification of LacZ expression , was created by deletion of lacI and Ptrc in pTrc99A [60] followed by insertion of the Plac-lacZ fragment of MG1655 [61] . In all experiments EZ defined rich medium ( Teknova , Hollister , CA , USA ) with 0 . 2% glucose and 1 mM thiamine HCL ( Sigma ) was used . Isopropyl β-D-1-thiogalactopyranoside ( IPTG ) was purchased from Sigma , and was added to the medium , if applicable , in a 1 mM quantity . Mutations were introduced into the coding region of lacI by site-directed mutagenesis with the QuickChange II–E Site–Directed Mutagenesis Kit ( Stratagene , USA ) according to the manufacturer's protocol [28] . Constructs are available upon request . Cultures were grown at 37°C in a Perkin & Elmer Victor3 plate reader , at 200 µl per well in a black clear-bottom 96 well plate ( NUNC 165305 ) . Expression measurements were performed in EZ Rich Defined medium with added 0 . 2% glucose ( Teknova , Hollister , CA , USA , cat . nr . M2105 ) supplemented with 1 mM thiamine HCl and the appropriate antibiotics for the selective maintenance of plasmid pRD007 and pRepLacZ . Optical density at 600 nm was recorded every 4 min , and every 29 min 9 µl sterile water was added to each well to counteract evaporation . When not measuring , the plate reader was shaking the plate at double orbit with a diameter of 2 mm . Cells were fixed after the cultures had reached an optical density of at least 0 . 015 and at most 0 . 07 , by adding 20 µl FDG-fixation solution ( 109 µM fluorescein di-β-D-galactopyranoside ( FDG , Enzo Life sciences , NL ) , 0 . 15% formaldehyde , and 0 . 04% DMSO in water ) . Fluorescence development was measured every 8 min ( exc . 480 nm , em . 535 nm ) , as well as the OD600 . Shaking and dispensing conditions were as mentioned above . When cells are not induced with IPTG , directly before or after fixation an appropriate amount of inhibitive IPTG was added . Analysis of the fluorescence trace is as described in [28] . Significance of the phenotypic effect of mutations in LacI was tested with a t-test with Bonferroni correction for multiple comparisons ( P<0 . 05 ) . While the phenotypic effect of S97P in the wild type background in Env0 , was not significant in the data set of one inverse Lac variant ( LacIinv3 ) , it was significant for the two other variants , and hence S97P was considered significant for the wild type background and Env0 . A FoldX plugin [40] ( version 1 . 4 . 22 ) in the Yasara software package [62] ( version 11 . 11 . 4 ) was used for the stability analysis of the single , double and triple ( only LacIinv1 ) mutants on basis of the DNA bound dimeric LacI crystal structure ( 1EFA ) [63] , which lacks the tetramerization domain . The structure was minimized without ONPF before addition of the mutations , and the calculation of the stability changes . The stability calculation was performed three times for each mutation , with standard deviations among the calculations smaller than ΔΔG = 0 . 5 kcal/mol . | Epistatic interactions limit the number of adaptive trajectories to peaks on evolutionary fitness landscapes , and may therefore hamper the progress of evolution . Recent research has focused on adaptive landscapes in one constant environment . However , adaptive evolution is generally known to occur in variable , heterogeneous environments . Here , we have constructed fitness landscapes of three inverse lac repressor variants in two contrasting environments . We find that the epistatic interactions between the pairs of mutations are profoundly altered upon an environmental change . We develop a new graphical method to analyze the underlying higher-order interactions between genetic changes and the environment , and explain the complex environmental dependencies in terms of simple molecular mechanisms . Our results show that the information about epistatic interactions acquired in one environment does not inform on the true limitations of adaptive evolution . We argue that this dependency of genetic constraints on the environment will have important effects on the progress of adaptation in heterogeneous environments , and will affect our ability to establish realistic genealogies from the phylogenic record . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"physics",
"systems",
"biology",
"biochemistry",
"ecology",
"genetics",
"biology",
"computational",
"biology",
"evolutionary",
"biology",
"microbiology",
"biophysics"
] | 2013 | Environmental Dependence of Genetic Constraint |
The parasite Leishmania often relies on gene rearrangements to survive stressful environments . However , safeguarding a minimum level of genome integrity is important for cell survival . We hypothesized that maintenance of genomic integrity in Leishmania would imply a leading role of the MRE11 and RAD50 proteins considering their role in DNA repair , chromosomal organization and protection of chromosomes ends in other organisms . Attempts to generate RAD50 null mutants in a wild-type background failed and we provide evidence that this gene is essential . Remarkably , inactivation of RAD50 was possible in a MRE11 null mutant that we had previously generated , providing good evidence that RAD50 may be dispensable in the absence of MRE11 . Inactivation of the MRE11 and RAD50 genes led to a decreased frequency of homologous recombination and analysis of the null mutants by whole genome sequencing revealed several chromosomal translocations . Sequencing of the junction between translocated chromosomes highlighted microhomology sequences at the level of breakpoint regions . Sequencing data also showed a decreased coverage at subtelomeric locations in many chromosomes in the MRE11-/-RAD50-/- parasites . This study demonstrates an MRE11-independent microhomology-mediated end-joining mechanism and a prominent role for MRE11 and RAD50 in the maintenance of genomic integrity . Moreover , we suggest the possible involvement of RAD50 in subtelomeric regions stability .
Genomic integrity maintenance is essential for cellular development and viability [1–3] . Failure to repair DNA will lead to genomic instability ( reviewed in [4–7] ) . DNA structural changes can manifest as inversion , deletion , duplication , translocation , chromosome end-to-end fusion , aneuploidy [8–10] and some of these events such as gene amplification have been associated in Leishmania with response to drug and oxidative stress [11–14] . Increased numbers of DNA rearrangements have been reported in many inherited cancer susceptibility human syndromes [9] . Specific DNA repair genes are mutated in these genomic disorders such as ATM in the Ataxia telangiectasia syndrome , MRE11 in the Ataxia telangiectasia-like disorder , NBS1 in the Nijmegen breakage syndrome and BLM in the Bloom’s syndrome [15–18] . It has been suggested that errors occurring during DNA replication such as stalled or broken replication forks can lead , if left unrepaired , to DNA double strand breaks ( DSBs ) that are precursors of DNA rearrangements [10 , 19] . DSBs can also occur during replication or result from exposure to DNA-damaging agents such as ionizing radiation or chemotherapeutic drugs [1 , 20] . The two main strategies to cope with DSBs are non-homologous end joining ( NHEJ ) and homologous recombination ( HR ) [20] . However , only a few NHEJ factors are present in Leishmania ( MRE11 , Ku70/Ku80 and APTX ) while Artemis , XRCC4 and the DNA ligase IV are absent , suggesting that this pathway is not functional in the parasite [21–23] . Another pathway that is normally suppressed when NHEJ is present is called microhomology-mediated end joining ( MMEJ ) or alternative end joining and has been reported in the related parasite Trypanosoma brucei [22–24] . In MMEJ , small regions of homology ( 2 to 20 nucleotides ) are used for ligation after resection of each DNA ends in a Ku-independent manner [20] . In this process , DNA ends created from DSBs are recognized by PARP-1 and resected by the MRN ( MRE11-RAD50-NBS1 ) complex followed by annealing and ligation of the two ends by XRCC1/DNA ligase III [25 , 26] . The HR pathway has been shown to be important for the recovery of stalled replication forks , genomic integrity and telomere maintenance [27–29] . In HR , DSBs are first recognized by the MRN complex and resected by EXO1 and MRE11 . Therefore , MMEJ and HR share the same initial step of resection which involves the MRN complex in order to produce regions of homology . Nevertheless , the length of DNA resection as well as the length of the homologous sequence differ between the two processes [23 , 30] . The tripartite MRN complex has been shown to act as DSBs sensor and DSBs repair effector and is also associated with telomere maintenance [31–33] , displaying a major role in the maintenance of genomic stability [34–37] . The complex is composed of MRE11 and RAD50 , highly conserved between species , and NBS1 ( also represented by XRS2 in yeast ) is less conserved and only present in eukaryotes . We previously demonstrated that LiMRE11 displays the same DNA binding and exonuclease activity as human MRE11 , but the protein is not essential in Leishmania infantum [38] . In addition , we showed the importance of MRE11 and its nuclease domain in extrachromosomal linear amplicons formation under drug pressure . The RAD50 protein is a DNA binding ATPase that displays sequence and structural homology to structural maintenance of chromosome ( SMC ) family members . An anti-parallel coiled-coil domain contains a central zinc hook ( CXXC ) motif and might contribute in holding together separate DNA ends [8 , 34] . The third member of the complex is NBS1 and possess a MRE11 binding domain . NBS1 is thought to stimulate the MRE11-RAD50 complex DNA binding and nuclease activities but biochemical activities of the NBS1 protein itself are not yet elucidated [8 , 34] . Disruption of the MRE11 and RAD50 genes have been shown to increase gene rearrangements rate up to 1000 fold [2 , 39] . Null mutations in any of the MRN proteins lead to embryonic lethality in mice and have been associated in yeast with DNA rearrangements and chromosome loss events as well as defect in both HR and NHEJ [39–44] . In this manuscript , we present the conditional inactivation of L . infantum RAD50 orthologue , a gene essential in the MRE11 proficient wild-type background but apparently dispensable in the MRE11-/- background . We also demonstrate chromosomal translocations in the MRE11 and RAD50 deficient cells . These translocations happened through a MRE11-independent MMEJ mechanism where sequence microhomology were found at the translocations breakpoints .
It is standard practice to generate null mutant parasites by replacing the entire ORF with resistance markers . The genes coding for the blasticidin-S deaminase ( BLAST ) and puromycin acetyltransferase ( PURO ) were cloned between the 5’- and 3’- L . infantum RAD50 flanking regions and the BLAST or PURO constructs were transfected independently by electroporation . Hybridization with a 5’UTR probe should lead to 3 . 2 , 1 . 7 and 1 . 5 kb SacI-SacI bands in the wild-type ( WT ) , BLAST RAD50-/+ , PURO RAD50-/+ cells respectively ( Fig 1A ) . We generated BLAST RAD50-/+ and PURO RAD50-/+ heterozygous lines ( S1B Fig , lanes 2–3 ) , but surprisingly , in the BLAST/PURO/WT RAD50-/-/+ line we observed a remaining intact RAD50 allele ( S1B Fig , lane 4 ) . Despite many attempts , the generation of a RAD50 null mutant failed . This generation of polyploidy at specific locus is frequently observed in Leishmania [45–47] and is thought to occur at locus reputed to be essential . To provide further support for the essentiality of RAD50 , we first introduced a RAD50 rescue plasmid ( Psp-NEO-RAD50WT , Fig 1A ) in the BLAST RAD50-/+ cells . Upon the transfection of the PURO cassette we could generate a chromosomal BLAST/PURO RAD50-/- cell with no more intact RAD50 chromosomal copy ( Fig 1B , lane 4 ) but with the presence of the extrachromosomal rescue RAD50 copy with its diagnostic 4 . 9 kb SacI-SacI band ( Fig 1A ) hybridizing with the RAD50 ORF probe ( Fig 1C , lane 4 ) . Removing the drug NEO pressure ( the marker of the rescuing episome ) for several passages would lead to either maintenance or loss of the plasmid depending on whether RAD50 is essential or not . Cells grown in absence of selection for the NEO marker maintained the episome ( up to 55 passages ) ( Fig 1C , lane 5 ) . This was not due to an unusual stability of the plasmid since introduction of the same NEO plasmid in WT cells and then growth in absence of selection pressure led to the loss of the rescuing episome after 35 passages ( Fig 1C , lanes 2 , 3 ) . To further investigate the essentiality of the RAD50 gene , we generated a mutated version of the Psp-NEO-RAD50 after introduction of the K42A mutation in the RAD50 ATPase domain ( Psp-NEO-RAD50K42A ) ( S2A Fig ) . The wild-type and mutated recombinant proteins were purified and the K42A mutation indeed impaired the ATPase activity of RAD50 ( S2D Fig ) . The Psp-NEO-RAD50K42A construct was transfected in the BLAST RAD50-/+ cells . In contrast to cells complemented with Psp-NEO-RAD50WT ( Fig 1B , lane 4 , S2B Fig , lane 4 ) , cells complemented with Psp-NEO-RAD50K42A showed a remaining RAD50 allele after integration of the PURO marker ( S2B Fig , lane 5 ) . The presence of the Psp-NEO-RAD50K42A plasmid was confirmed by hybridization with a RAD50 probe ( S2C Fig ) . This result indicates that the ATPase activity of the rescue RAD50 copy is necessary to allow inactivation of both RAD50 chromosomal alleles . While inactivation of RAD50 was not possible in a WT background , it was easily achieved in a MRE11-/- mutant . Indeed an MRE11-/- mutant was already available [38] and its RAD50 locus was shown here to be intact , as a 3 . 2 kb fragment was present after SacI digestion of the genomic DNA ( Fig 1B and 1C , lane 6 ) . In the MRE11-/- background we could inactivate both RAD50 alleles with the BLAST and PURO markers without the need for a rescuing plasmid ( Fig 1B and 1C , lane 7 ) . To confirm the absence of both MRE11 and RAD50 genes in the MRE11-/-RAD50-/- strain , we performed PCR amplification using two sets of primers ( Fig 1D ) . The use of primers sets aa’ and bb’ should only amplify a PCR fragment if the MRE11 and RAD50 genes are present respectively . As expected , no PCR amplification was detected for both MRE11 and RAD50 genes in the MRE11-/-RAD50-/- strain ( Fig 1D , lane 7 ) . We also carried out qRT-PCR for RAD50 mRNA levels in a number of lines and inactivation of one RAD50 allele by either PURO or BLAST reduced the mRNA by half compared to WT ( S1C Fig , lanes 2 , 3 ) . A similar fold decrease was observed in the PURO/BLAST RAD50-/- cells that have an extra copy of the gene ( S1B Fig , lane 4 ) , indicating that this new allele is actively expressed ( S1C Fig , lane 4 ) . The level of RAD50 mRNA in the MRE11-/- cells was similar to the WT strain but was undetectable in the MRE11-/-RAD50-/- cells ( S1C Fig , lanes 5 , 6 ) . Overall , a nice correlation between RAD50 copy number and mRNA expression was observed . We also attempted inactivating the RAD50 gene in cells with only one MRE11 allele but mutated for its nuclease activity ( HYG/PUR-MRE11H210Y [38] ) . We failed to generate a RAD50 null mutant in this HYG/PUR-MRE11H210Y cells since we observed the maintenance of a third RAD50 chromosomal allele after integration of both BLAST and NEO resistant markers in the RAD50 locus ( S3B Fig , lane 3 ) . RAD50 thus appears to be essential in MRE11H210Y nuclease dead cells . The availability of a RAD50 null mutant ( in the MRE11-/- background ) has allowed to test for a number of phenotypes . The MRE11-/- and MRE11-/-RAD50-/- mutants had similar growth properties ( S4A Fig ) , susceptibility to the DSBs inducing alkylating damaging agent methyl methanesulphonate ( MMS ) ( S4B Fig ) ; and displayed a reduced ability to carry out homologous recombination ( HR ) ( S4E Fig ) . The RAD50-/- mutant with its episomal rescue had similar growth phenotype and recombination proficiency as the WT cells ( S4 Fig ) . To ensure that the phenotypes observed in the MRE11-/- null mutant were not due to a reduction in RAD50 protein levels , we overexpressed RAD50 as part of an episomal construct ( Psp-RAD50 ) in the MRE11-/- cells . The MRE11-/- and MRE11-/- Psp-RAD50 cells had similar growth properties and susceptibility to MMS ( S4C and S4D Fig ) . In response to drug pressure , Leishmania amplifies specific portion of its genome either as part of extrachromosomal circular or linear amplicons . Circles are dependent on RAD51 and RAD51-4 [14 , 48] while linear amplicons depend on MRE11 [38] . Selection of Leishmania WT cells for resistance to the antifolate methotrexate ( MTX ) often leads to the extrachromosomal amplification of the pteridine reductase gene PTR1 ( usually as part of linear amplicons ) or of the dihydrofolate reductase-thymidylate synthase gene DHFR-TS ( usually as part of circular amplicons ) [49] . An example of a PTR1 containing linear amplicon ( at 450 kb ) is provided in a WT cell that was selected fro MTX resistance ( Fig 2A ) . The 770 kb PTR1 hybridizing band corresponds to the chromosomal alleles . We showed previously that in contrast to WT cells , the MRE11-/- mutant selected for MTX resistance did not have PTR1 amplified as part of linear amplicons ( Fig 2B ) [38] . We investigated the ability of the MRE11-/-RAD50-/- null mutants to perform extrachromosomal amplification by selecting clones for MTX resistance in a stepwise manner ( up to 1600 nM , a 16-fold increase in resistance compared to parent cells ) . Leishmania chromosomes extracted from ten MTX resistant clones derived from MRE11-/-RAD50-/- parasites were separated by pulse field gel electrophoresis ( PFGE ) ( S5 Fig ) and hybridized with the PTR1 gene . Hybridization data revealed the 770 kb PTR1 containing chromosome but no hybridizing bands diagnostic for PTR1 linear amplicons ( Fig 2C and S5 Fig ) . However , clones 1 , 4 , 7 and 8 had a PTR1 circular amplification , as deduced from the characteristic hybridization profiles of circles in PFGEs , including the hybridization in the slots ( corresponding to open circles ) and the hybridizing smears ( corresponding to topoisomers of the circles ) [50] ( Fig 2C lanes 1 , 4 , 7 , 8 ) . Amplification of the DHFR-TS gene is rarely observed in L . infantum selected for MTX resistance and this was further confirmed , where only the 520 kb chromosomal copies hybridized to a DHFR probe ( Fig 2D ) . However since several resistant mutants had no PTR1 amplification ( Fig 2C ) we hybridized the same resistant clones with a DHFR-TS probe . We observed the 520 kb band corresponding to the DHFR-TS containing chromosome but no sign for either circular or linear amplicons ( Fig 2F ) . However , we detected in all clones a 795 kb band that surprisingly was also present in the parent MRE11-/-RAD50-/- cells before MTX exposure ( Fig 2F , lane 0 ) . A similar , but clearly not identical ( 950 kb vs 795 kb ) chromosomal rearrangement was previously observed in the MRE11-/- strain after MTX pressure ( Fig 2E , lane + ) [38] , suggesting that this locus is prone to chromosomal rearrangement . The chromosomal rearrangement involving the DHFR-TS chromosome in the MRE11-/-RAD50-/- strain was further studied by genome sequencing of the nuclease deficient strains . The genomes of the WT , MRE11-/- and MRE11-/-RAD50-/- lines were subjected to Illumina next-generation paired-ends sequencing ( NGS ) . Sequencing reads were first aligned to the genome of L . infantum JPCM5 using bwa-mem alignment [51] . The alignments were then screened for discordant read pairs and split reads alignments using the Lumpy-sv and the Delly software [52 , 53] . This provided a list of chromosomal translocations present in the MRE11 and MRE11/RAD50 null mutants . A total of five translocations were observed in the MRE11 and MRE11/RAD50 deficient cells ( Table 1 ) . The analysis of the genomic sequences allowed the detection of the translocation of part of the DHFR-TS chromosome observed in Fig 2F . It involved 433 kb of chromosome 12 and 362 kb of chromosome 06 ( encoding DHFR-TS ) , giving a hybrid chromosome T 12–06 of 795 kb ( Fig 3A ) . To further characterize experimentally this translocation , we hybridized the chromosomes of the WT , MRE11-/- and MRE11-/-RAD50-/- cells with probes spanning the translocation breakpoints ( filled and open squares and circles in Fig 3A ) . Hybridization with the gene LinJ . 12 . 0671 ( ■ ) revealed a band corresponding to chromosome 12 ( 568 kb ) and an additional 795 kb band corresponding to the T 12–06 translocation in the MRE11-/-RAD50-/- strain ( Fig 3B ) . The same 795 kb band hybridized to LinJ . 06 . 0480 ( ● ) , a gene derived from chromosome 06 and part of the T 12–06 translocation ( Fig 3B ) . The genes LinJ . 12 . 0690 ( □ ) and LinJ . 06 . 0470 ( ○ ) should not be part of the hybrid chromosome ( Fig 3A ) and indeed when these genes were used as probes they only hybridized to bands corresponding to chromosomes 12 and 06 respectively ( Fig 3B , Table 1 ) . Two additional translocations also implicated chromosome 12 ( T 12–17 , T 12–18 , Table 1 ) as described below . This is possible because several chromosomes of Leishmania are polyploids [13 , 54–56] and read counts indicate that chromosome 12 is tetraploid in our L . infantum WT strain ( Fig 3C ) . One translocation led to a hybrid composed of 386 kb of chromosome 12 fused to 159 kb of chromosome 17 ( Fig 4A ) . The size of chromosome 12 and the hybrid chromosome are too similar for their discrimination by PFGE but hybridization with LinJ . 17 . 1180 ( ■ ) showed a band corresponding to chromosome 17 ( 667 kb ) and the 545 kb band corresponding to T 12–17 ( Fig 4B ) . The third translocation implicating chromosome 12 involved chromosome 18 ( 408 kb of chromosome 12 and 167 kb of chromosome 18 leading to an hybrid chromosome of 575 kb , ( Fig 4C ) ) . The size of chromosome 12 and the hybrid were again too similar for discrimination by PFGE but hybridization with LinJ . 18 . 1520 ( ● ) revealed the 575 kb band corresponding to T 12–18 ( Fig 4D lanes 2 , 3 ) . The hybridization patterns in the two nuclease mutants are more complex with no band exactly migrating with the intact chromosome 18 band ( Fig 4D ) . In the case of the MRE11-/-RAD50-/- mutant , this can be explained in part by an additional translocation of chromosome 18 with chromosome 20 ( Fig 4E ) where the new hybrid chromosome ( T 18–20 ) had 674 kb of chromosome 18 and 69 kb of chromosome 20 for an estimated length of 743 kb ( Fig 4D and 4F lane 3 ) . In the MRE11-/- mutant we observed three bands hybridizing with the LinJ . 18 . 1520 ( ● ) probe ( Fig 4D , lane 2 ) . The band of 575 kb corresponds to the T 12–18 hybrid chromosome , the highest band at 778 kb corresponds to one of two versions of T 18–20 ( with an internal duplication that was highlighted by reads depth analysis , see below ) . This band also hybridized with LinJ . 20 . 1570 ( ▲ ) ( Fig 4F , lane 2 ) . The middle band hybridizing to LinJ . 18 . 1520 ( ● ) appears slightly smaller than the 720 kb WT chromosomal copy ( Fig 4D , lanes 1 and 2 ) and may correspond to a truncated form of chromosome 18 . The final translocation highlighted by NGS involved chromosome 08 and 17 ( Table 1 ) and T 08–17 consists of 175 kb of chromosome 17 and 395 kb of chromosome 08 leading to an hybrid chromosome of 570 kb ( Fig 5A ) . Hybridization with LinJ . 08 . 0290 ( ■ ) revealed a 570 kb band corresponding to T 08–17 and a 465 kb band slightly smaller than the expected size of chromosome 08 ( 495 kb ) ( Fig 5B , lane 3 ) . This smaller 465 kb band also hybridized with LinJ . 08 . 0280 ( □ ) ( not part of T 08–17 ) , and may , similarly to one copy of chromosome 18 discussed above , correspond to an internal deletion or truncation of the original chromosome but the bioinformatics analysis did not provide support for these potential scenarios . Hybridization with probes derived from chromosome 17 further supported the formation of the hybrid chromosome T08-17 ( Fig 5B , lane 3 ) . The Lumpy-sv and Delly software also revealed a fusion between chromosome 27 and chromosome 02 that was already present in the WT cells , highlighting a difference between the L . infantum 263 WT strain compared to the reference L . infantum JPCM5 WT ( S6 Fig ) . Most of chromosome 27 ( 1044 kb ) is fused with the last two genes on chromosome 02 ( 4 kb ) ( S6A Fig ) . Sequence homology between the end of chromosomes 2 and 27 has already been described for L . major [57] with subtelomeric repeats and this rearrangement occurring in the WT may correspond to telomere exchange rather than translocation . In the past , we have used normalized reads depth coverage over the 36 chromosomes to predict copy number variations [11 , 12] . Sequenced reads of the 36 chromosomes of the L . infantum 263 strain indicated that while the majority of chromosomes were mostly diploid , chromosomes 12 , 13 and 31 were polyploid . There were no changes in ploidy in the nuclease mutants except when translocation occurred . Normalized log2-transformed read counts for chromosome 06 in WT cells , MRE11-/- , and MRE11-/-RAD50-/- revealed a shift at the T 12–06 breakpoint in the MRE11-/-RAD50-/- null mutant , leading to an increased number of reads for part of chromosome 6 starting with gene LinJ . 06 . 0480 ( Fig 3D ) . At one telomere end of chromosome 06 in WT and MRE11-/- strains we observed increased number of reads for a region of 60 kb ( Fig 3D ) that corresponds to a linear extrachromosomal amplicon that we have previously characterized in L . infantum 263 WT [14] . This linear amplicon was lost in the MRE11-/-RAD50-/- strain . Reads depth coverage indicated that chromosome 12 is tetraploid in L . infantum 263 WT and that MRE11-/- and MRE11-/-RAD50-/- parasites probably contained only one intact copy of chromosome 12 ( Fig 3C ) but all the hybrid chromosomes T 12–06 , T 12–17 and T 12–18 contribute to a higher ploidy for most sequences of chromosome 12 ( Fig 3C ) . The normalized reads depth of chromosome 12 also highlighted the T 12–06 , T 12–17 and T 12–18 breakpoints ( Fig 3C ) . Similarly , the breakpoints for T 12–18 and T 18–20 also fitted with a change in reads depth on chromosome 18 and chromosome 20 ( S7 Fig ) . In the case of chromosome 18 , reads depth showed a shift at the T 12–18 and T 18–20 breakpoints in the two mutants and overlapping regions between T 12–18 and T 18–20 ( from LinJ . 18 . 1330 to LinJ . 18 . 1530 ) were present in three copies ( S7A Fig ) . In the MRE11-/- strain , normalized read counts for chromosome 18 also highlighted internal duplication of 15 kb close to the T 18–20 breakpoint , increasing the size of the T 18–20 ( S7A Fig ) . In the same translocation , part of chromosome 20 also showed a duplication of 20 kb ( S7B Fig ) in the MRE11-/- cells , increasing the size of the translocation T 18–20 from 743 kb to 778 kb in that strain ( Fig 4E ) . Reads mapping to chromosome 20 also showed a decreased number of reads at one telomere end in the MRE11-/-RAD50-/- cells ( S7B Fig ) , a phenomenon that we observed for several other chromosomes ( see below ) . Finally , normalized reads depth coverage over chromosomes 08 and 17 highlighted the T 08–17 , T 12–17 and T 08–17 breakpoints ( Fig 5C and 5D ) . We used PCR to validate the new junctions created by the fusion of portions of chromosomes in all translocations . Oligonucleotide primers located on each side of the translocation points were designed to target the new junctions . The ORFs identities ( or position of intergenic regions ) closest to the breakpoints can be found in Table 1 . We were able to precisely define the junction of 4 of the 5 translocations ( T 12–06 , T 12–18 , T 18–20 and T 08–17 ) . In all cases the sequencing of the fusion points in the hybrid chromosomes revealed that the rearrangements occurred at the level of microhomologies between 7 and 17 bp ( Fig 6A–6D ) . There were no common sequence features between the various repeats . We were unsuccessful to map precisely the translocation breakpoints for T 12–17 . Indeed the breakpoint is located inside a region containing repeated DNA sequences along 60 kb and specific PCR amplification of the junction has not been possible . The genome of Leishmania is constituted of large polycistronic clusters of genes that are co-expressed [58 , 59] . Interestingly some of the translocation would create new regions where co-directional gene clusters diverge or converge ( T 12–18 , T 18–20 , T 08–17 ) that may impact on gene expression . PCR amplification of the junction between chromosomes 27 and 02 revealed an insertion of 21 bp at the junction of chromosome 27 and chromosome 02 in the 27–02 hybrid ( S6B Fig ) . This rearrangement is clearly not similar to the translocation events characterized in this study and may correspond to exchange of telomeric sequences between chromosomes 27 and 02 . The MRE11-/-RAD50-/- parasites displayed a decreased number of reads mapping to chromosomes ends , suggesting sequences near telomeres were impaired in that strain . This phenomenon occurred for eleven chromosomes ( Fig 7 and S8 Fig ) and three were experimentally verified by Southern blot ( Fig 7 ) . Genomic DNAs from the WT , MRE11-/- and MRE11-/-RAD50-/- cells were hybridized with a chromosome 05 probe close to the telomeres ( LinJ . 05 . 0060 ) and hybridization intensities were compared with probe LinJ . 05 . 0560 used as an internal control for DNA loading ( Fig 7A ) . Hybridization intensities yielded a 0 . 7 fold-decrease for the MRE11-/-RAD50-/- strain compared to the WT or MRE11-/- cells . Similar analyses were performed with probes derived from chromosome 28 ( Fig 7B ) and chromosome 34 ( Fig 7C ) and when telomeric proximal probes were used the signal was consistently lower in the MRE11-/-RAD50-/- parasites , compared to either WT cells or the MRE11-/- mutant . Genomic DNAs from WT , MRE11-/- and MRE11-/-RAD50-/- cells were also digested with Sau3aI , AluI and RsaI and hybridized with a telomeric probe [60 , 61] . After hybridization , discrete bands were present in both the WT and MRE11-/- but in the MRE11-/-RAD50-/- cells we observed a smear ( S9A Fig ) . When an internal probe far from telomere was used ( PTR1 gene ) , a single band was observed in all three lines ( S9B Fig ) . The results suggest that within the MRE11-/-RAD50-/- population there is considerable heterogeneity at the end of chromosomes in individual cells ( explaining the smear when hybridized with a telomeric probe ) , in line with the decreased number of reads mapping chromosomes ends ( Fig 7 ) .
Gene rearrangement in Leishmania is genome wide [14 , 38] and can lead to extrachromosomal elements [13 , 55] , to chromosomes in multiple copies and to mosaic aneuploidy [56] . It is thought that these events can lead to selective advantage [11–14] and our recent work has shed some light on the enzymes involved in these processes . RAD51 and at least one of its paralog ( RAD51-4 ) are involved in the formation of extrachromosomal circles [14 , 48] , while MRE11 is involved in the formation of linear amplicons [38] . MRE11 is partnering with RAD50 and NBS1 as part of the MRN complex [34 , 35] . MRE11 is one of the main sensor of DNA DSBs while RAD50 modulate the activity of the complex [8] . It was shown in Saccharomyces cerevisiae that the RAD50 coiled-coil domain is indispensable for MRE11 functions since truncation of this domain in RAD50 impaired telomere maintenance , meiotic DSB formation , HR and NHEJ , indicating its need for MRN activities [62] . We initiated this work to test whether MRE11 and RAD50 functions would overlap and whether these proteins are involved in the maintenance of genomic integrity . To test this we used gene inactivation and while we were able to obtain a MRE11-/- null mutant [38] , it has been impossible to generate a RAD50-/- null mutant in a WT background ( Fig 1 and S1 Fig ) . We could only inactivate both alleles if a rescue episomal copy of RAD50 was present but despite prolonged passages in absence of the selecting drug , we could not lose the episomal RAD50 copies , a strong suggestion that RAD50 is essential in Leishmania . A chromosomal copy of RAD50 was maintained upon gene inactivation if cells were complemented with a mutated RAD50K42A rescue plasmid indicating that a fully functional RAD50 is essential for cell survival ( S2 Fig ) . In mammals , RAD50 and MRE11 are essential [41 , 42] but in yeast both proteins are dispensable [33 , 63 , 64] , thus diverse organisms have different requirements for proteins part of the MRN complex . In both human cells and S . cerevisiae , introduction of mutations in the RAD50 ATPase domain impaired DNA binding and DNA unwinding [65] suggesting RAD50 is required for the stability of the DNA-MRN complex interaction [66 , 67] . We also tried to inactivate the RAD50 gene in MRE11H210Y nuclease-deficient cells but this did not lead to a RAD50 null mutant ( S3 Fig ) . The MRE11H210Y protein is deficient in nuclease activity but still capable of DNA binding [38] . Our results provide good evidence that inactivation of RAD50 may only be possible in the absence of MRE11 . There may possibly be a need for RAD50 when MRE11 is present , even if its nuclease domain is inactivated such as in MRE11H210Y . The inactivation of MRE11 nuclease activity in murine cells did not change the MRN complex formation [36] , and possibly the presence of MRE11 forces the presence of RAD50 and the formation of MRE11 H210Y/RAD50 interactions . We infer that , in addition to its interaction with MRE11 , the Leishmania RAD50 protein might also have important functions which would only occur in the presence of MRE11 . Indeed inactivation of RAD50 was easily achieved in a MRE11-/- background ( Fig 1B and 1C lanes 7 ) . One hypothesis is that MRE11 inactivation leads to genetic compensation in Leishmania and this compensation makes RAD50 dispensable to any other putative important function that RAD50 may have . It is also possible that MRE11 is detrimental in the absence of a RAD50-mediated regulation that might happened through maintenance of the MRE11/RAD50 complex stoichiometry [68] . This hypothesis was plausible with previous observations showing that overexpression of MRE11 in Leishmania was detrimental for cell growth [38] , possibly because of stoichiometry disruption . Absence of the MRE11/RAD50 complex led to a growth defect , a sensitivity to MMS and an altered capacity for HR ( S4A and S4B Fig ) . Although knockdown of individual components of MRN in human cells led to a decrease in the other two MRN members [69] , our results suggest that RAD50 is normally expressed in the MRE11-/- strain at the RNA levels ( S1C Fig ) and functionally ( S4C and S4D Fig ) . The MRE11 and RAD50 deficient cells had an incapacity of generating PTR1 linear amplicons upon MTX selection ( Fig 2 and [38] ) . Whole genome sequencing indicated that translocations were observed in mutants lacking a fully functional MRE11/RAD50 complex ( Table 1 ) and the only clear difference between the MRE11-/- and MRE11-/-RAD50-/- mutants was at the level of subtelomeric sequences where the number of sequenced reads was much lower in several subtelomeric loci for the MRE11-/-RAD50-/- mutant ( Fig 7 and S8 Fig ) . One new aspect of this work is the discovery of chromosomal translocations which have not been observed before in old world Leishmania species [70–72] . Studies in yeast have also indicated an increase of translocations events and chromosomal rearrangements when either MRE11 or RAD50 are mutated [2] . Translocations are likely to have occurred after DSBs which are usually repaired by either HR or NHEJ . Several components of NHEJ are absent in Leishmania and the parasite thus relies mostly on HR [23] . Since HR is diminished in the MRE11 and RAD50 mutants ( S4E Fig ) , the cells may use alternative strategies to repair DNA . One of these alternative pathways used for repair of DSBs is based on MMEJ . Indeed , MMEJ has been implicated in chromosomal translocation in yeast [73] , mammals [74] , and in the related parasite T . brucei [75] . Mice defective in NHEJ have exhibited an increased level of translocations mediated by an alternative NHEJ that relied on microhomology [76] . The PCR reactions of the junctions created following translocations revealed that these events occurred via a mechanism of MMEJ where the microhomology is between 7 and 17 bp ( Fig 6 and Table 1 ) . There is no sequence specificity between the various translocation breakpoint sequences . We have shown previously that the Leishmania genome is filled with large repeated sequences [14] and we found that several of the microhomology sequences are either part of large repeated sequences ( for T 12–17 ) or close to repeated sequences ( for T 12–06 and T 12–18 ) . It is well known that repeated sequences can be fragile sites and therefore more prone to DSBs and could explain translocation mediated by MMEJ [77–79] . A recent study done in Leishmania donovani has described the use of MMEJ for repair of Cas9-induced DSBs using the CRISPR-Cas9 system , even though the parasites mostly relied on HR for DSBs repair [80] . The Cas9/gRNA complex is , however , continuously generating DNA breaks in every chromosomal allele of the targeted region , complicating the search for intact homology by the HR machinery , hence favoring alternative end-joining mechanism such as MMEJ . When a template with sequence homology was provided to the parasites , the HR mechanism largely dominated the DSBs repair [80] . In our study , the homologous chromosomal allele is thought to be intact but the defect in HR probably led the cells to a MMEJ mechanism for DNA repair . Several studies done in yeast have shown the importance of resection by MRE11 for the first step of MMEJ [30 , 69 , 81–83] but we suggest MMEJ can be MRE11-independent in Leishmania . It is possible that upon genetic compensation in the knock-out strains , the expression of other nucleases is increased and could perform some of the activities usually carried out by MRE11 . However , the other nucleases encoded by Leishmania ( e . g EXO1 , DNA2 [23] ) are reputed for extensive DNA resection which is unfavorable for MMEJ that usually favors short length resections ( performed by MRE11 ) [83 , 84] . Further experiments could help in deciphering the MRE11-independent MMEJ in Leishmania . Few chromosomes have been implicated in translocation and some of these chromosomes ( e . g chromosomes 12 , 17 and 18 ) have been implicated in more than one translocation . It is not clear whether the initial state of ploidy has a role to play seeing as chromosome 12 is tetraploid , chromosomes 17 and 18 are diploid but other chromosomes not involved in translocation are also polyploid like chromosomes 13 , 31 and 32 ( in L . infantum 263 WT ) . The rearrangements observed were also stable since re-sequencing of 3 clones from each null mutants after six months of continuous growth highlighted the same translocations and no additional one . Thus either Leishmania can support only few translocations or those are relatively rare events that can be maintained . Translocations and the formation of hybrid chromosomes change ploidy of specific regions ( Figs 3C , 3D , 5C and 5D ) but in general , a minimum of diploid state is conserved because of the overlap with the different translocations . Overall , none of the translocation breakpoints correspond to transcription initiation or termination sites and one aspect that was not studied is whether the formation of hybrids has consequences on the expression of genes in this novel context . This is particularly relevant as some of the rearrangements created regions where co-directional gene clusters diverge or converge ( e . g T 12–18 , T 18–20 , T 08–17 ) . Those divergent and convergent regions could represent regions where RNA polymerase can enter or exit , but transcription could also initiate or terminate within directional gene clusters [59 , 85 , 86] . Furthermore , when we compared our results to a recent study revealing that Leishmania chromosomes are replicated by a single origin ( instead of multiple sites of replication origins as other eukaryotes ) [87] , we observed that the translocation events generated hybrid chromosomes that also contained a single origin of replication ( coming from either one of the chromosome involved in the translocation ) . This observation suggest that even though genomic integrity was altered by the formation of hybrid chromosomes , the parasites were consistent in maintaining a single-origin of replication per chromosome . In the MRE11-/-RAD50-/- mutant , chromosome 8 is smaller than in the WT cells and in the MRE11-/-parasites , chromosome 18 is also smaller than in the WT , suggesting that some rearrangements events ( e . g deletions ) happened in the mutants . While analysis of sequence reads did not allow us to confirm this deletion on chromosome 8 , sequence reads has been highly useful in the past to detect changes in copy number [11 , 88] . We nonetheless conducted a reads depth analysis to detect either deletions or duplications and the bioinformatics analysis revealed the potential presence of several of them in the null mutants . However experimental validation by Southern blot only allowed confirming 1 out of 5 deletions and 1 out of 5 duplications deduced from the bioinformatics analyses . This suggests that the 5 kb window given by our bioinformatics pipeline might not be optimal for the detection of deletion and duplication events . Nevertheless our results suggest that there are more than translocation events as part of gene rearrangements in the nuclease null mutants . One key change that reads depth analysis detected and that we could confirm experimentally is a reduction of reads of several subtelomeric sequences exclusively in the MRE11-/-RAD50-/- mutant ( Fig 7 and S8 Fig ) , suggesting that the absence of RAD50 altered chromosome end stability as already observed in human cells [89] . The log2-transformed read counts would suggest that populations are not clonal but that several cells within the population have various levels of subtelomeres shortening including coding sequences ( Fig 7 ) . Decrease of sequence reads extends up to 100 kb from the telomeres in some of the cells although in general , the shortenings are smaller . The T . brucei subtelomeres harbor fragile sites [90] and subtelomeric regions are known to be more sensitive to DSBs that are processed differently than internal DSBs [91 , 92] . The decreased in sequence reads observed here is possibly due to an altered repair of DSBs in the MRE11-/-RAD50-/- mutant that lead to shortening of subtelomeric sequences . Indeed , a Southern blot of DNAs derived from MRE11-/-RAD50-/- hybridized with a telomeric probe revealed and hybridization smear suggesting considerable heterogeneity at chromosomes ends for individual cells within the population . RAD50 seems to have an important role in this since this phenomenon is not observed in the MRE11-/- mutant . This study provides strong evidence that MRE11 gene knock-out is a prerequisite for RAD50 inactivation in Leishmania . Chromosomal translocations are observed in the cells lacking a fully functional MRE11/RAD50 complex , and subtelomeric regions stability is altered in the absence of RAD50 . Moreover , we report for the first time in Leishmania a MRE11-independent alternative end-joining mechanism that relies on microhomology sequences . Overall , these results show a predominant role of the two DNA repair proteins MRE11 and RAD50 in chromosomal organization . Deciphering DNA repair mechanisms and maintenance of genomic integrity in Leishmania parasites may allow novel strategies for their control as they seem to rely on gene amplification and rearrangement for surviving the changing environment in which they grow .
Promastigotes of Leishmania infantum ( MHOM/MA/67/ITMAP-263 ) and all recombinants were grown in SDM-79 medium at 25°C supplemented with 10% fetal bovine serum , 5μg/ml of hemin at pH7 . 0 . Independent clones generated in this study were selected for methotrexate ( MTX ) resistance in M199 medium , using a stepwise selection starting from an EC50 of 100nM up to 1600nM of MTX . All chemical reagents were purchased from Sigma-Aldrich unless specified . The L . infantum RAD50 null mutant ( RAD50-/- ) was obtained by targeted gene replacement . RAD50 flanking regions were amplified from L . infantum 263 wild-type ( WT ) genomic DNA and fused to blasticidin-S deaminase ( BLAST ) , puromycin acetyltransferase ( PURO ) and neomycin phosphotransferase ( NEO ) genes using a PCR fusion based-method as described previously [93] . Briefly , 5’UTR of RAD50 was amplified using primers C and D for the BLAST cassette , primers C and E for the PURO cassette and primers C and F for the NEO cassette . The BLAST , PURO and NEO genes were amplified with primers G and H , I and J and K and L respectively . 3’UTR of RAD50 was amplified using primers M and N for all inactivation cassettes ( see primer sequences in S1 Table ) . At least 3μg of the 5’UTR-BLAST-3’UTR , 5’UTR-PURO-3’UTR or 5’UTR-NEO-3’UTR linear fragments were transfected by electroporation ( as described in [94] ) in L . infantum WT , L . infantum MRE11-/- or L . infantum HYG/PUR-MRE11H210Y cells [38] to replace both RAD50 alleles . Recombinants were selected in the presence of 80μg/ml of blasticidin-S hydrochloride , 100μg/ml of puromycin dihydrochloride ( Wisent ) and 40μg/ml G418 ( Geneticin; Sigma-Aldrich ) . After 4–5 passages , cells resistant to the drug selection were cloned in SDM-Agar plates ( 1% ) in the presence of the same concentrations of drugs . PCR analysis of the recombinants was done using forward primer located in the MRE11 5’ flanking region with reverse primer inside the MRE11 gene ( primers set aa’ ) , and forward primer in the RAD50 5’ flanking region with reverse primers located inside the RAD50 gene ( primers set bb’ ) ( see primers sequences in S1 Table ) . An episomal construct , Psp72-α-NEO-α-RAD50WT was designed to express RAD50 in the cells before inactivation of the second RAD50 genomic allele . Briefly , the RAD50 gene was amplified by PCR using primers O and P from L . infantum WT genomic DNA . The amplified product was first cloned in pGEM-TEasy vector and then subcloned in Psp72-α-NEO-α [95] in the HindIII and NdeI sites of the vector . Site-directed mutagenesis ( Stratagene , Quickchange ) was used to introduce the K42A mutation in the RAD50 ORF and generate the Psp72-α-NEO-α-RAD50K42A using primers Q and R ( S1 Table ) . Both Psp72-α-NEO-α-RAD50WT and Psp72-α-NEO-α-RAD50K42A plasmids were then transfected by electroporation in the L . infantum BLAST RAD50-/+ mutants and cells were selected with 40μg/ml of G418 ( Geneticin; Sigma-Aldrich ) . After inactivation of the second RAD50 genomic allele with the PURO cassette , attempts to lose the Psp72-α-NEO-α-RAD50 construct were performed by removing the G418 drug pressure up to 55 passages . Genomic DNAs from clones were isolated using DNAzol as recommended by the manufacturer ( Invitrogen ) . SacI or Sau3aI/AluI/RsaI digested genomic DNAs or separated chromosomes were subjected to Southern blot hybridization with [α-32P] dCTP-labeled DNA probes according to standard protocols [96] . All probes were obtained by PCR from L . infantum genomic DNAs except the telomeric probe obtained from a Psp72-PT4 [97] . Intact chromosomes were prepared from L . infantum promastigotes harvested from log phase cultures , washed once in 1X Hepes-NaCl buffer then lysed in situ in 1% low melting agarose plugs as described in [38] . Leishmania intact chromosomes were separated in 1X TBE buffer ( from 10X TBE: 1M Tris , 1M Acid boric , 0 , 02M EDTA ) by Pulsed-Field Gel Electrophoresis ( PFGE ) using a Bio-Rad CHEF-DRIII apparatus at 5V/cm and a 120° separation angle as described previously [47] . The range of chromosome separation was between 150 and 1500 kb . Late log phase promastigotes ( 30ml ) were pelleted at 3000 rpm for 5 minutes and pellets were washed once with 1X HEPES-NaCl , resuspended in suspension buffer ( 100mM EDTA , 100mM NaCl , 10mM Tris pH 8 . 0 ) , then lysed in 1% SDS and 50μg/ml proteinase K at 37°C for 2 hours . Genomic DNA was extracted with 1 volume phenol , precipitated with 2 volumes 99% ethanol , washed with 70% ethanol twice and dissolved in 1ml 1X TE buffer . RNAse A ( Qiagen ) was added at 20μg/ml and DNA was incubated at 37°C for 30 minutes , followed by the addition of 50μg/ml of proteinase K and 0 . 1% SDS at 37°C for 30 minutes . DNA was extracted with 1 volume of phenol , precipitated and washed in ethanol , and dissolved in DNase free-water ( Millipore ) at 37°C overnight . Sequencing libraries were produced with the Nextera DNA sample preparation kit ( Illumina Inc ) according to manufacturer’s instructions . Genome sequences were determined by Illumina HiSeq 2500 101-nucleotides paired-end sequencing . Reads from each strain were aligned to the reference genome Leishmania infantum JPCM5 ( TriTrypDB version 8 . 0 ) using Burrows-Wheeler Alignment ( bwa-mem ) [51] with default parameters . The maximum number of mismatches was 4 , the seed length was 32 and 2 mismatches were allowed within the seed . Several python and bash scripts were created for the detection of copy number variations . Briefly , chromosomes were divided into genomic windows of 5 kb and the number of reads mapping to each windows determined and normalized to the total number of reads before inter-strains comparisons . Alignments were also performed using the Lumpy-sv and Delly software [41 , 52] with default parameters for split-reads alignments and discordant read pairs and only translocations found with both software were kept for validation . PCR amplification of the new junction created in the translocation was performed using primers within 750 nucleotides from the translocation breakpoint on each involved chromosome . Optimal primer length was 20 nucleotides and optimal melting temperature ( Tm ) was 55°C . Primer sequences are presented in S1 Table . Digestion of the Psp72-α-ZEO-α plasmid [98] using PciI and XbaI enzymes was performed and the isolated α-ZEO-α fragment was used to target the alpha-tubulin loci in order to monitor the integration efficiency . Briefly , 2x106 cells from WT , MRE11-/- and MRE11-/-RAD50-/- strains were transfected with 5μg of the linear α-ZEO-α construct . After 24h following electroporation , cells were plated on SDM-Agar plates ( 1% ) containing Zeocin ( Invitrogen ) at 1mg/ml . All strains were also transfected with the plasmid Psp72-α-ZEO-α and with sterile water as controls . Colonies were counted after 10–15 days of plating . RNAs were extracted using RNeasy plus mini kit ( Sigma ) according to the manufacturer recommendations . The cDNA was synthesized using Oligo dT12-18 and SuperScript II RNase H-Reverse Transcriptase ( Invitrogen ) and amplified in SYBR Green Supermix ( Bio-Rad ) using a rotator thermocycler Rotor Gene ( RG 3000 , Corbett Research ) . The expression level was derived from three technical and three biological replicates and was normalized to constitutively expressed mRNA encoding glyceraldehyde-3-phosphate dehygrogenase ( GAPDH , LinJ . 36 . 2480 ) . The sequences of the primers used in this assay are listed in S1 Table . L . infantum WT , RAD50-/- Psp-RAD50 , MRE11-/- and MRE11-/-RAD50-/- were resuspended at a concentration of 5x106 cells/ml and exposed to increasing doses of MMS ( Sigma–Aldrich ) . Cells were counted after 72h and reported in survival rate . Reactions ( 10 μl ) contained 40nM of Leishmania infantum RAD50 or RAD50K42A ( purified by double affinity purification accordingly to [99] ) in 50mM Tris-HCl pH 7 . 5 , 1mM Mg ( CH3COO ) 2 , 1mM DTT and 100 μg/ml BSA supplemented with 50 nCi [ɣ-32P]ATP ( 3000 Ci/mmole; Perkin Elmer Life Sciences ) . Aliquots ( 5 μl ) were removed , stopped by addition of EDTA , and the percentage of ATP hydrolyzed was determined by thin layer chromatography followed by quantification using a Fujifilm Phosphoimager . The data set supporting the results of this article is available at the EMBL-EBI European Nucleotide Archive ( http://www . ebi . ac . uk/ena ) under study accession number PRJEB11440 with sample accessions ERS934506 , ERS934507 and ERS934508 for L . infantum MRE11-/-RAD50-/- , L . infantum MRE11-/- and L . infantum JPCM5 , respectively . L . infantum 263 WT sequencing data is available under the study ERP001815 and sample accession number ERS179382 . | The parasite Leishmania relies on gene rearrangements to survive stressful conditions . However , maintaining a minimum level of genomic integrity is crucial for cell survival . Studies in other organisms have provided evidence that the DNA repair proteins MRE11 and RAD50 are involved in chromosomes organization , protection of chromosomes ends and therefore in the maintenance of genomic integrity . In this manuscript , we present the conditional inactivation of the Leishmania infantum RAD50 gene that was only possible in MRE11 deficient cells and suggest the genetic background is crucial for RAD50 inactivation . We demonstrate the occurrence of chromosomal translocations in the MRE11 and RAD50 deficient cells and described a MRE11-independent microhomology-mediated end-joining mechanism at the level of translocation breakpoints . We also suggest a possible involvement of RAD50 in subtelomeric regions stability . Our results highlight that both MRE11 and RAD50 are important for the maintenance of genomic integrity in Leishmania . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"Methods"
] | [
"cell",
"hybridization",
"molecular",
"probe",
"techniques",
"biological",
"cultures",
"parasitic",
"protozoans",
"microhomology-mediated",
"end",
"joining",
"protozoans",
"leishmania",
"molecular",
"biology",
"techniques",
"translocations",
"dna",
"research",
"and",
"analy... | 2016 | Chromosomal Translocations in the Parasite Leishmania by a MRE11/RAD50-Independent Microhomology-Mediated End Joining Mechanism |
Pathogens causing acute fever , with the exception of malaria , remain largely unidentified in sub-Saharan Africa , given the local unavailability of diagnostic tests and the broad differential diagnosis . We conducted a cross-sectional study including outpatient acute undifferentiated fever in both children and adults , between November 2015 and June 2016 in Kinshasa , Democratic Republic of Congo . Serological and molecular diagnostic tests for selected arboviral infections were performed on blood , including PCR , NS1-RDT , ELISA and IFA for acute , and ELISA and IFA for past infections . Investigation among 342 patients , aged 2 to 68 years ( mean age of 21 years ) , with acute undifferentiated fever ( having no clear focus of infection ) revealed 19 ( 8 . 1% ) acute dengue–caused by DENV-1 and/or DENV-2 –and 2 ( 0 . 9% ) acute chikungunya infections . Furthermore , 30 . 2% and 26 . 4% of participants had been infected in the past with dengue and chikungunya , respectively . We found no evidence of acute Zika nor yellow fever virus infections . 45 . 3% of patients tested positive on malaria Rapid Diagnostic Test , 87 . 7% received antimalarial treatment and 64 . 3% received antibacterial treatment . Chikungunya outbreaks have been reported in the study area in the past , so the high seroprevalence is not surprising . However , scarce evidence exists on dengue transmission in Kinshasa and based on our data , circulation is more important than previously reported . Furthermore , our study shows that the prescription of antibiotics , both antibacterial and antimalarial drugs , is rampant . Studies like this one , elucidating the causes of acute fever , may lead to a more considerate and rigorous use of antibiotics . This will not only stem the ever-increasing problem of antimicrobial resistance , but will–ultimately and hopefully–improve the clinical care of outpatients in low-resource settings . ClinicalTrials . gov NCT02656862 .
Acute fever is one of the main reasons for healthcare seeking worldwide . In tropical settings , and especially sub-Saharan Africa , malaria is the first cause to be ruled out , which is done increasingly so following the World Health Organization’s testing before treating policy of 2010 –through microscopic blood slide examination or a rapid diagnostic test ( RDT ) [1] . Following the introduction of this policy , together with the roll-out of the highly efficacious artemisinin-combination therapy as first-line malaria treatment and efficacious vector control , the overall malaria burden declined over the last decade [1] . Accordingly , clinicians face a relatively higher number of malaria-negative patients for whom they do not have a clear diagnosis [2–4] . In sub-Saharan Africa , where healthcare settings are often resource-limited , healthcare providers face the daunting challenge pinpointing the causing agent of acute fever in an adequate and timely fashion , with little to no diagnostic means other than a malaria-RDT . They mostly rely on history taking and physical examination to determine the focus and cause of infection , of which acute respiratory infection ( ARI ) , gastroenteritis ( GE ) and urinary tract infection ( UTI ) are the three most prevalent syndromes reported [4] . However , for some patients presenting with acute fever no focus of infection can be found , thus labeling them as ‘undifferentiated’–knowing that their differential diagnosis is broad , ranging from viral , bacterial , parasitic to fungal infections . Some studies have found that this ‘undifferentiated’ group among fever patients represents 20 to 40% of the grand total [4] . Although viral illnesses are often suspected , both prescription and over-the-counter usage of antimicrobials is rampant in this group globally and their licentious usage in low-resource settings fuels the global burden of antimicrobial resistance [5 , 6] . More insight in the exact causes of this group of ‘undifferentiated fevers’ may help curb the usage of antimicrobials and improve the clinical care of patients in low-resource settings more broadly [7] . Still , evidence on the causes of ‘undifferentiated acute fever syndromes’ is scarce and is coming almost entirely from inpatient settings . Indeed , to our knowledge only one study in sub-Saharan Africa , namely in Sierra Leone in 2012–2013 , looked in a prospective way at the etiologies of acute fever–using RDTs–in patients with self-reported or clinically confirmed fever with a maximum duration of 7 days , finding 5% acute dengue virus ( DENV ) infection and 39% acute chikungunya virus ( CHIKV ) or other alphavirus infection [8] . Nonetheless , limited outbreaks and sporadic clinical cases of DENV have been reported over the last 50 years in 22 African countries [9] . Seroprevalence studies have demonstrated DENV IgG-antibodies , indicating past-infection , in 12 . 5% of study participants in Cameroon , 36% in Burkina Faso and 45% in Nigeria [9] , although in other areas seropositivity remained zero [10] . In Tanzania past infection rates are higher , reaching 50 . 6% in health-facility based studies and 11% in community-based studies [11] . Despite the presence of all four DENV serotypes , severe disease epidemics are rarely reported in Africa [12] . The DENV burden in Africa is , based on modeling , estimated at 16 million symptomatic clinical infections or 16% of the global total [13] . In East-Africa , CHIKV outbreaks and circulation are described , such as in Kenya with a past-infection rate of 67% [14] and the reports of epidemics in 2004 in Kenya [15] , in 2013 in Tanzania [16] and in 2018 in Mozambique [17] . These viral vector-borne diseases are also circulating in the Central African region . This is illustrated by CHIKV outbreaks in Kinshasa in both 2000 [18] and 2012 [19] and Brazzaville in 2011 [20] , and the 2013 DENV [21] and 2016 yellow fever virus ( YFV ) outbreak in Angola [22] , along with the first Zika virus ( ZIKV ) case reported in Angola in 2017 [23] . These outbreaks are only possible because the Aedes mosquito , the vector of the aforementioned arboviruses , thrives in this region . Furthermore , although not conclusive for vector competence and local transmission capability , alphaviruses ( chikungunya ) and flaviviruses ( species not specified ) were demonstrated by RT-PCR in Aedes mosquitoes in Kinshasa in 2014 [24] . In the Democratic Republic of Congo ( DRC ) the circulating pathogens causing uncomplicated acute undifferentiated fever , are unknown [25] . However , outside the above documented epidemics , CHIKV and DENV probably circulate continuously . Of travelers returning from Africa ( 2007–2012 ) and attending the outpatient clinic of the Institute of Tropical Medicine in Antwerp , Belgium , 22% of those diagnosed with a CHIKV infection came from DRC [26] and up to that point there was an increasing number of confirmed DENV infections in travelers coming from a large set of African countries , including DRC [27] . In Eastern DRC , a few DENV cases have been found during an outbreak of West-Nile fever in 1998 [28] and between 2003 and 2012 when testing samples negative for YFV [29] . In this study in DRC , we aim to quantify the importance of four major arboviruses as a cause of acute undifferentiated fever . Furthermore , we aim to describe the case presentation and the presence of arbovirus/malaria co-infections , since Plasmodium falciparum is still responsible for an estimated 25 million cases nationwide–with 97% of the country being a ‘high transmission’ region , ranking DRC among the top 3 countries in sub-Saharan Africa with the highest malaria burden [30] .
This study was approved by the ethical review boards of the School of Public Health of the University of Kinshasa ( DRC ) , the Institute of Tropical Medicine of Antwerp ( Belgium ) and the University Hospital of Antwerp ( Belgium ) . The study was registered in a public repository ( https://www . clinicaltrials . gov/ct2/show/NCT02656862 ) . Written informed consent was obtained from every adult or–in case of minors–from their caretaker . This study was conducted in full compliance with the principles of the latest amended Declaration of Helsinki and of the International Conference Harmonization ( ICH ) guidelines , plus adhering to local laws and regulations . The study took place in Lisungi health center in Pumbu , an area of about 14 , 000 inhabitants , belonging to the peri-urban health district Mont Ngafula 1 , at the southern side of Kinshasa . The climate is tropical with a rainy season between October and May , and a dry season from June to September . The Lisungi health center is the only public health facility in the area , with a medical staff of 40 persons averaging 250 patient encounters per week–which are provided for a small out-of-pocket contribution , as is commonplace throughout DRC . It treats mainly outpatients , but several inpatient beds are available for short time follow-up of more complicated cases . Over the years , around 70% of patients mention fever as the reason for medical care seeking , of whom half tested positive for malaria on RDT ( personal communication with Dr . Blaise Fungula ) . There are no means for other microbiological testing . The Lisungi health center has recently performed a Good Clinical Laboratory Practice compliant trial and has been involved in other febrile illness investigations , specifically on malaria [31] . The study was designed as a cross-sectional study with prospective patient inclusion . As the proportion of pathogens can change over time , especially for epidemic-prone diseases , we included patients proportionally from November 2015 to June 2016 . Only patients of at least 2 years old , presenting at the outpatient department with a history of acute fever ( i . e . ≥ 2 days and ≤ 7 days ) or having an axillary temperature of ≥ 37 . 5°C , were eligible . Patients with any history of an acute injury , trauma or poisoning , suspicion of meningitis/encephalitis , recent hospitalization or women who gave birth in the preceding two weeks , were excluded . Reported recent intake of antimicrobials was not an exclusion criterion , but was recorded accordingly . There were two categories of patients included: the ‘undifferentiated fevers’ , with as case definition history of acute fever and without any clear clinical focus of infection and the ‘differentiated fevers’ , with a history of fever and with acute respiratory infection ( ARI ) , gastroenteritis ( GE ) or urinary tract infection ( UTI ) categorized on clinical grounds . Of the first group , a maximum of 6 patients per day ( 3 children and 3 adults ) , and of the second group a maximum of 2 patients per day ( 1 child and 1 adult ) were to be included . The latter category was included given the non-specificity of the signs and symptoms of a possible arboviral infection and in order to estimate the burden of co-infection–which was also the reason not to exclude the confirmed malaria cases based on laboratory analysis . Our main interest was the distribution of viral pathogens among the undifferentiated fevers . To be able to detect the presence of a disease whose prevalence is 5% with a precision of 2 . 5% at a confidence level of 95% , 290 patients needed to be included . Increased with 10% for incomplete data or loss of biological samples , we came to a minimal sample size of 320 .
Over the period November 2015 to June 2016 , 342 patients were included from whom clinical data and on the spot malaria-RDT results were recorded . Clinical diagnoses were as follows: 70 . 2% undifferentiated fever , 17 . 0% ARI , 6 . 1% GE and 6 . 7% UTI–all three further labeled as ‘differentiated fevers’ . The study population ( Table 1 ) consisted of 183 ( 53 . 5% ) female participants and 180 ( 53 . 1% ) were under the age of 18 . Only 10 ( 2 . 9% ) reported having a chronic disease such as diabetes or sickle cell anemia and 50 ( 14 . 6% ) patients were in need of hospitalization . The reported YFV vaccination rate was as low as 1% . The vast majority ( 77 . 8% ) of patients presenting in the four first days after the onset of fever , came from the commune Mont Ngafula , and a minority from neighbouring Selembao and Ngaliema–acute and past arbovirus infections were detected in patients from these three communes ( Table 2 ) . Before going to the health center , 15 . 5% of the participants already self-medicated with one or more tablets of an anti-malarial drug . At presentation , malaria RDT for Plasmodium falciparum was positive in 155 ( 45 . 3% ) of the total number of participants ( 47 . 9% and 39 . 2% in undifferentiated and differentiated fever groups , respectively , p = 0 . 139 ) . However , 300 ( 87 . 7% ) received an antimalarial treatment , which was significantly more in the undifferentiated ( 92 . 1% ) than the differentiated fever group ( 77 . 5% ) ( p<0 . 001 ) . Antibacterial drugs were frequently prescribed: 64 . 3% of participants received at least one antibacterial drug , significantly more in the differentiated ( 72 . 5% ) than the undifferentiated group ( 60 . 8% ) ( p = 0 . 039 ) and 11 . 7% received even more than one antibacterial drug ( 29 . 4% in the differentiated and 9 . 6% in the undifferentiated fever group , p<0 . 001 ) . In the 235 participants further tested for arboviruses , 19 ( 8 . 1% ) fulfilled the criteria of an acute DENV infection , of which 14 were confirmed by RT-PCR . Both serotypes DENV1 and DENV2 were detected ( Table 3 ) . Five participants were presumptively infected with DENV , based on the presence of IgM antibodies alone . All NS1 positive patients were RT-PCR positive . In contrast , only two acute CHIKV infections were suspected based on the presence of IgM antibodies . The majority of CHIKV IgM ELISA positive samples was not confirmed by IFA . On these non-congruent samples , PCR to detect Plasmodium was performed and revealed an actual malaria infection in 18 of the 22 CHIKV IgM ELISA positive /IgM IFA negative . There was a temporal heterogeneity in the appearance of DENV infections ( Fig 1 ) . In June , a dry season month which had less than 25mm of rain ( www . infoclimat . fr ) , there was an increased risk of 6 . 13 ( adjusted OR 95% CI 2 . 24–17 . 81 ) of presenting with acute DENV in comparison to the rainy season ( S1 Table ) . Of the acute DENV and CHIKV cases , 31 . 6% and 0% had a positive malaria RDT , respectively . When focusing on the malaria negative cases , we observed that 8 . 7% ( 13/149 ) tested positive for acute DENV , 1 . 3% ( 2/149 ) for acute CHIKV , none for acute ZIKV , nor for YFV . With regard to the clinical presentation of both DENV and CHIKV infections , we found no specific signs or symptoms to be statistically significantly–let alone clinically relevant–associated with acute DENV or CHIKV versus the other febrile patients ( S2 Table ) . None of the acute DENV or CHIKV cases were clinically severe enough to require hospitalization and there was no apparent leucopenia or hemoconcentration ( as often seen in severe DENV cases ) . We found 71 ( 30 . 2% ) patients with anti-DENV IgG antibodies , of which 60 ( 75 . 0% ) contained relatively high levels of anti-DENV antibody ( O . D . ≥ 1 . 9 ) , the latter not depending on age ( S1 Fig ) . The PRNT on the subsample was only positive in 5 . 6% and 66 . 7% for DENV , and 25% and 20% for YFV , on samples with IgG ratio below 1 . 9 and above 1 . 9 , respectively ( Table 3 ) . For these past infections , all 4 serotypes were detected: DENV1 , DENV2 , DENV3 and DENV4 in 4 , 2 , 1 and 1 patient , respectively ( of the 6 positively tested DENV PRNT ) – 2 patients were reactive to all 4 serotypes , the other 4 only to 1 serotype . Past exposure to CHIKV was suspected with IFA IgG in 26 . 4% of the study participants . When taking the DENV and CHIKV positive IgG samples together , 56 . 6% of the study participants were suspected having been exposed to at least one arbovirus . The prevalence of past DENV and CHIKV infections increased with age , raising from 18 . 9% and 5 . 4% under 5 years of age , to 80% and 40% over 65 years of age , respectively ( Fig 2 ) . The association of age is statistically significant for the past infections with CHIKV ( p = 0 . 01 ) and DENV ( p<0 . 01 ) ( S2 Table ) . Having been exposed to DENV was also statistically significant associated with recent travel ( p = 0 . 01 ) . The congruence between the RDT and PCR/ELISA results for DENV was variable: in comparison to PCR the sensitivity of NS1 was 90% , in comparison to IgM ELISA the IgM RDT had a sensitivity of 30% and in comparison to IgG ELISA the IgG RDT had a sensitivity of 7 . 6% . The specificities were all above 99 . 3% .
Although no large epidemics were reported recently , our study showed ongoing transmission of arboviruses in Kinshasa , DRC . Acute DENV , caused by DENV1 and DENV2 , and CHIKV infection was demonstrated in 8 . 1% and 0 . 9% of the patients attending a first line health center with acute undifferentiated fever , respectively . Importantly , neither DENV nor CHIKV was clinically suspected , nor considered in the clinical differential diagnosis and 64 . 3% of patients were treated with at least one antibacterial drug of whom almost one in eight ( 11 . 7% ) received dual or triple antimicrobial therapy . A possible explanation of the apparent absence of clinical and/or severe acute DENV cases in our study , and in other African settings too , is that African heritage is described to genetically protect against severe DENV . More specifically , the lower OSBPL10 expression profile in Africans is protective against viral hemorrhagic fever and dengue shock syndrome [36 , 37] . However , diagnostic testing for arboviruses has several shortcomings . Hereafter , we will highlight the limitations of the tests used for acute and past infections . Five out of 19 acute DENV and 2 out of 2 acute CHIKV infections were diagnosed based on the presence of IgM antibodies only and were therefore only presumptive infections . This could have led to an overestimation of the number of acute infections . In addition , IgM antibodies can be present for several months and it is therefore possible that DENV/CHIKV was not the cause of the fever at the moment of presentation . Nonetheless , the presence of IgM antibodies suggests that DENV/CHIKV was recently circulating in the area . On the other hand , as we did not have repeated measurements or convalescent samples to demonstrate seroconversion to IgG or a four-fold increase in IgG titer , we may have missed some acute secondary dengue cases , which may have undetectable IgM antibody levels [38] . The number of positive CHIKV IgM ELISA test results was unexpectedly high . As the results could not be confirmed with IFA , a technique which is considered to be more specific , interference with malaria was suspected based on our experience . Indeed , false-positive reactions as a result of polyclonal B-cell activation is a phenomenon that we experienced before with ZIKV ELISA [39] . The detection of Plasmodium by PCR in the CHIKV ELISA positive/IFA negative samples strongly supported this hypothesis . The use of IFA for IgM detection may result in false-positive reactions although the risk is limited in case of experienced readers [40] . Cross-reaction of anti-CHIKV antibodies with antibodies against other members of the Semliki-forest serogroup , notably O’nyong nyongvirus , could not be excluded as no neutralization assays were performed . ZIKV was not suspected to be circulating in the area and indeed , no RT-PCR positive cases were found . YFV was actively circulating in the region at the time of study , transgressing the border with Angola [22] , but we did not detect any RT-PCR positive case . Detecting YFV and ZIKV only based on RT-PCR diagnostics could have resulted in an underestimation of acute ZIKV and YFV infections . However , ZIKV and YFV IgM testing was not done , because almost 80% of the patients included in this study presented in the first four days after the onset of fever , which is the period with the highest probability to detect the virus with molecular methods and there are several shortcomings with the IgM testing for these viruses [41 , 42] . Nowadays , there is evidence that in urine samples ZIKV is longer detectable by ZIKV PCR , but at the time of study this was not known . Thus , no such samples were collected [43] . We demonstrated past exposure to arboviruses too , with 30 . 2% of the participants having detectable IgG against DENV , which is on the higher end of the spectrum compared to other studies in Africa reporting an overall flavivirus seroprevalence ranging between 0 and 35% with a mean of 18 . 1% [9 , 25] . The IgG-seroprevalence increased with age , thus suggesting a continuous exposure to flaviviruses over time . The 26 . 4% past CHIKV infection rate was in line with the estimated seroprevalence of 34 . 4% in Congo Brazzaville before the outbreak of 2011 [20] and was on par with other African sites reporting an overall alphavirus seroprevalence oscillating between 0 and 72% [25] . Remarkably , CHIKV IgG was also detected in small children , born after the 2011 epidemic , pointing towards an endemic circulation of the virus . DENV IgG testing was done with an ELISA test . Although it is widely known that there is cross-reactivity among flaviviruses , ELISA is still the most affordable–hence most commonly used–test [25] . Since YFV vaccination is an expected cause of cross-reaction with DENV IgG , we performed PRNT for DENV and YFV in a subset of samples , and found that the majority of samples negative with PRNT for DENV , were also negative with PRNT for YFV , suggesting that the high flavivirus IgG positivity is not likely to be the result of YFV vaccination . Due to operational reasons and the scope of the study , we did not further test for other flaviviruses and hence , we cannot rule out that the DENV IgG positivity in our study is due to other flaviviruses exposure [44] . We noted that the congruence between DENV ELISA and PRNT was lower than in American and Asian settings [45–49] , but similar to the observation in other studies in DRC [50] and in Ethiopia [51] . The report of YFV vaccination was very low ( below 1% ) , but as YFV vaccination is included in the childhood vaccination program over the last decade , this may indicate that the population is not aware of which vaccines their children get . However , the increasing prevalence of flavivirus IgG antibodies with age is congruent with a history of increasing exposure to the pathogens over lifetime and thus could not be explained by YFV vaccination . The CHIKV IgG testing was done with two tests: screening with ELISA and confirmation with IFA . We cannot exclude that positive results are due to cross-reactivity with other known ( e . g . O’nyong nyong virus , Semliki Forest virus or Sindbis virus ) or unknown togaviruses . In a recent study conducted at the Lisungi health center , it was reported that 62% of patients–both children and adults–with acute fever had neither malaria nor bacteremia [52] . For the first time we were able to demonstrate the fact that arboviruses , more specifically DENV and CHIKV , circulate in the capital of DRC . The highest number of acute cases was reported in June ( a dry season month ) , but cases were also confirmed in the other months , indicating that despite an epidemic profile , transmission persists over the rainy season in Kinshasa . This finding is consistent with observations over the past decades in Asia [53] and Latin-America [54] and adds to mounting–although still scarce–evidence that arboviruses are endemic in large parts of sub-Saharan Africa [9 , 17] . Furthermore , we were able to document the common practice of over-prescription of antimicrobials , including antimalarial drugs , in malaria RDT-negative patients , as is apparently the case nationwide in DRC as recently shown by Ntamabyaliro et al [55] . Indeed , while not even half of the patients ( 45 . 3% ) tested positive for malaria–a figure just below average national RDT positivity rates [30] , close to 90% received antimalarial treatment , in addition to 15 . 5% of patients treated with over-the-counter antimalarials prior to presentation at the clinic . It could be questioned whether the rigorous implementation and usage of RDTs has any added benefit . A recent meta-analysis , including studies from Afghanistan , Cameroon , Ghana , Nigeria , Tanzania , and Uganda , evaluated data from over half a million children and adults and showed that the introduction of a malaria RDT simply shifted the antimicrobial overuse from one antimicrobial class to the other , mainly from antimalarial to antibacterial and anthelmintic drugs [56] . Consequently , the increasing prescription rate of antimicrobials–including antibacterial , anthelmintic and antimalarial drugs , is extremely worrisome in terms of the growing global problem of antimicrobial resistance , including against Plasmodium falciparum [57] . Although the sampling design of this study was adequate to evaluate the proportion of arboviruses causing acute undifferentiated fever , sample size was small and patients were only recruited from a single health center . However , the Lisungi health center is well visited by the surrounding population and all ages were represented in our study population . Moreover , the median age of our study population is 17 years , which approximates the median age of 18 . 6 years in the DRC ( UNDESA 2017 and CIA World Factbook 2017 ) . Another limitation in our study was the impossibility due to operational reasons to include participants over an entire year . The study was halted at the end of June , which was apparently the month with the highest number of infections . We did not investigate bacterial causes of fever through culture of blood or other bodily fluids , tests typically done in hospital settings , hereby possibly underestimating the burden of concomitant bacterial ( super ) infection . We therefore encourage further research elucidating the broad range of pathogens causing acute undifferentiated fever and the distribution of the insect vectors involved in arboviral transmission in urban and rural sub-Saharan African settings . Based on our findings , we recommend to include arboviral infections , namely DENV and CHIKV , in the differential diagnoses of acute fever presentation in Kinshasa . In conclusion , we state that among undifferentiated acute fever cases in a peri-urban health center of Kinshasa , dengue–both DENV-1 and DENV-2 –and chikungunya infections were demonstrated , but no acute cases of Zika or yellow fever were detected . Apart from these acute infections , we showed that about one third of participants showed evidence of past arboviral exposure , as evidenced by positive IgG antibodies titers . | Malaria remains one of the most important causes of fever in sub-Saharan Africa . However , its share is declining , since the diagnosis and treatment of malaria have improved significantly over the years . Hence leading to an increase in the number of patients presenting with non-malarial fever . Often , obvious clinical signs and symptoms like cough or diarrhea are absent , probing the question: “What causes the fever ? ” Previous studies have shown that the burden of arboviral infections–like dengue and chikungunya–in sub-Saharan Africa is underestimated , which is why we screened for four common arboviral infections in patients presenting with ‘undifferentiated fever’ at an outpatient clinic in suburban Kinshasa , Democratic Republic of Congo . Among the patients tested , we found that one in ten presented with an acute arboviral infection and that almost one in three patients had been infected in the past . These findings suggest that clinicians should think about arboviral infections more often , thereby refraining from the prescription of antibiotics , a practice increasingly problematic given the global rise of antimicrobial resistance . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"dengue",
"virus",
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"togaviruses",
"chikungunya",
"infection",
"pathogens",
"immunology",
"tropical",
"diseases",
"microbiology",
... | 2019 | Dengue and chikungunya among outpatients with acute undifferentiated fever in Kinshasa, Democratic Republic of Congo: A cross-sectional study |
Pitch is one of the most important features of natural sounds , underlying the perception of melody in music and prosody in speech . However , the temporal dynamics of pitch processing are still poorly understood . Previous studies suggest that the auditory system uses a wide range of time scales to integrate pitch-related information and that the effective integration time is both task- and stimulus-dependent . None of the existing models of pitch processing can account for such task- and stimulus-dependent variations in processing time scales . This study presents an idealized neurocomputational model , which provides a unified account of the multiple time scales observed in pitch perception . The model is evaluated using a range of perceptual studies , which have not previously been accounted for by a single model , and new results from a neurophysiological experiment . In contrast to other approaches , the current model contains a hierarchy of integration stages and uses feedback to adapt the effective time scales of processing at each stage in response to changes in the input stimulus . The model has features in common with a hierarchical generative process and suggests a key role for efferent connections from central to sub-cortical areas in controlling the temporal dynamics of pitch processing .
Modelling the neural processing of pitch is essential for understanding the perceptual phenomenology of music and speech . Pitch , one of the most important features of auditory perception , is usually associated with periodicities in sounds [1] . Hence , a number of models of pitch perception are based upon a temporal analysis of the neural activity evoked by the stimulus [2]–[5] . Most of these models compute a form of short-term autocorrelation of the simulated auditory nerve activity using an exponentially weighted integration time window [6]–[13] . Autocorrelation models have been able to predict the reported pitches of a wide range of complex stimuli . However , choosing an appropriate integration time window has been problematic , and none of the previous models has been able to explain the wide range of time scales encountered in perceptual data in a unified fashion . These data show that , in certain conditions , the auditory system is capable of integrating pitch-related information over time scales of several hundred milliseconds [14]–[22] , while at the same time being able to follow changes in pitch or pitch strength with a resolution of only a few milliseconds [14] , [15] , [21]–[24] . Limits on the temporal resolution of pitch perception have also been explored by determining pitch detection and discrimination performance as a function of frequency modulation rate [25]–[27] , the main conclusion being that the auditory system has a limited ability to process rapid variations in pitch . The trade-off between temporal integration and resolution is not exclusive to pitch perception , but is a general characteristic of auditory temporal processing . For instance , a long integration time of several hundred milliseconds is required to explain the way in which the detectability and perceived loudness of sounds increases with increasing sound duration [28] , [29] . In contrast , much shorter integration times are necessary to explain the fact that the auditory system can resolve sound events separated by only a few milliseconds [28]–[30] . Therefore , it appears that the integration time of auditory processing varies with the stimulus and task . Previously it was proposed that integration and resolution reflect processing in separate , parallel streams with different stimulus-independent integration times [28] . More recently , in order to reconcile perceptual data pertaining to temporal integration and resolution tasks , it was suggested that the auditory system makes its decisions based on “multiple looks” at the stimulus [31] , using relatively short time windows . However , to our knowledge no model has yet quantitatively explained the stimulus- and task-dependency of integration time constants . Another major challenge for pitch modelling is to relate perceptual phenomena to neurophysiological data . Functional brain-imaging studies strongly suggest that pitch is processed in a hierarchical manner [32] , starting in sub-cortical structures [33] and continuing up through Heschl's Gyrus on to the planum polare and planum temporale [34]–[36] . Within this processing hierarchy , there is an increasing dispersion in response latency , with lower pitches eliciting longer response latencies than higher pitches [37] . This suggests that the time window over which the auditory system integrates pitch-related information depends on the pitch itself . However , no attempt has yet been made to explain this latency dispersion . In this study , we present a unified account of the multiple time scales involved in pitch processing . We suggest that top-down modulation within a hierarchical processing structure is important for explaining the stimulus-dependency of the effective integration time for extracting pitch information . A highly idealized model , formulated in terms of interacting neural ensembles , is presented . The model represents a natural extension of previous autocorrelation models of pitch in a form resembling a hierarchical generative process [38] , [39] , in which higher ( e . g . , cortical ) levels modulate the responses in lower ( e . g . , sub-cortical ) levels via feedback connections . Without modification , the model can account not only for a wide range of perceptual data , but also for novel neurophysiological data on pitch processing .
The role of the feed-forward process ( solid lines in Figure 1 ) is to predict the pitch of the incoming stimulus . The perceived pitch of periodic sounds corresponds approximately to the reciprocal of the repetition period of the sound waveform . This is why temporal models of pitch perception , such as autocorrelation models , usually analyze the periodicities of the signal within the auditory-nerve channels , and then use these periodicities to derive a pitch estimate by computing the reciprocal of the periodicity that is most prevalent across frequency channels [2] . The cochlea in the inner ear acts as a frequency analyzer , in that different sound frequencies activate different places along the cochlea , which are in turn innervated by different auditory nerve fibres [1] . Thus , the cochlea can be modelled as a bank of band-pass filters . In the current model , each cochlear filter was implemented as a dual resonant nonlinear gammatone filter , which accounts for the sound level-dependent non-linear properties of cochlear processing [40] . The filter output was then passed through a hair cell transduction model [41] to simulate the conversion of the mechanical cochlear response into auditory-nerve spiking activity . The model was implemented using DSAM ( Development System for Auditory Modelling http://www . pdn . cam . ac . uk/groups/dsam/ ) . It contained a total of 30 frequency channels with centre frequencies ranging from 100 to 10000 Hz on a logarithmic scale . The hair cell transduction model generates auditory-nerve spike probabilities , p ( t , k ) , as a function of time , t , in each frequency channel , k . The first processing stage ( open boxes in Figure 1 ) computes the joint probability that a given auditory nerve fibre produces two spikes , one at time t and another at t-l , where l is a time delay or lag [10] . These joint probabilities are generated by computing the cross-product of the auditory-nerve firing probability , p ( t , k ) , with time-delayed versions of itself for a range of time delays . The cross-products are then summed across all frequency channels , k , to generate the output of the first stage of the model A1 ( t , l ) : ( 1 ) The activity at the second processing stage , A2 ( t , l ) ( green circles in Figure 1 ) , is computed as a leaky integration , ( i . e . , a low-pass filter using an exponentially decaying function [42] ) of the input activity , A1 ( t , l ) , using relatively short time constants , τ2 . It may therefore be assumed to represent sub-thalamic neural populations [43]–[46] . The time constants at the second stage are lag-dependent ( τ2 = τ2 ( l ) ) , as suggested by recent psychoacoustic studies [23] , [37] . However , for clarity , the lag dependency will not be explicitly stated in the following equations . In the third stage , A3 ( t , l ) ( red circles in Figure 1 ) , the output of the second stage is integrated over a longer time scale , τ3 , as suggested by neuroimaging studies of pitch in the cortex [37] , [47] . This stage is assumed to be located more centrally . Both integration stages can be simply described as time-varying exponential averages , ( 2 ) In equation ( 2 ) , Δt is the time step of the integration and En ( t ) is the instantaneous exponential decay rate of the response at each integration stage ( En ( t ) ≤τn ) , which will henceforth be referred to as the effective integration window . Establishing an appropriate time constant is as has been mentioned one of the major difficulties in formulating a general model of pitch perception . Hence , the value of En ( t ) in the model proposed here is not constant but is controlled by changes in the properties of the stimulus . The control of En ( t ) will be explained below . The factors gn ( t ) normalize the input to each stage by the corresponding integration window ( g2≡1; g3 ( t ) = E2 ( t ) /τ2 ) . At each time step An ( t , l ) will have a maximum at some value of l which we will write as Ln . The inverse of this lag for the output of stage 2 , 1/L2 ( t ) , represents the intermediate pitch prediction of the model ( see Figure 1 ) . Similarly , the inverse of the lag corresponding to the maximum response in stage 3 , 1/L3 ( t ) is the final pitch prediction . For convenience , we refer to the final pitch prediction from the preceding time step 1/L3 ( t-Δt ) as the pitch expectation , 1/LE . In all simulations presented in the current study , we used 200 lags , with reciprocals logarithmically distributed , representing pitches between 50 to 2000 Hz [48] . As an example , Figure 2 shows the model response to a sequence of pure tones ( Figure 2A ) with random frequencies and durations . Figure 2B shows the first stage of the model A1 ( t , l ) and Figure 2C the effective integration windows . Figure 2D shows the final model output; the red colour highlights the lag-channels with strong responses . The lag of the channel with the maximum response at a given time corresponds to the reciprocal of the pitch predicted by the model . Note that the response A3 ( t , l ) in Figure 2D was normalized to a maximum of unity after each time step and mapped exponentially onto the colour scale to make the plot clearer . However , this transformation is monotonic and thus does not affect the model predictions . The necessity for stimulus-driven modulation of the effective integration time , En ( t ) , becomes clear from a consideration of existing autocorrelation models . If E2 ( t ) were constant over time , i . e . , E2 ( t ) ≡ τ2 , then A2 ( t , l ) would correspond to the summary autocorrelation function ( SACF ) proposed by Meddis and colleagues [6] , [7] . If , in addition , E3 ( t ) ≡ τ3 then A3 ( t , l ) would represent an additional leaky integrator with a longer time constant . This is equivalent to the cascade autocorrelation model proposed by Balaguer-Ballester et al . [13] . The right panel in Figure 3A illustrates the success of the purely feed-forward model in response to a click train stimulus with alternating inter-click intervals [49] , [50] . The arrow indicates the average pitch reported by listeners . The pitch of such alternating click train stimuli has been difficult to predict with autocorrelation models consisting of only one integration stage with a short time constant ( see right panel in Figure 3B ) . However , the longer time scale used in the second stage of the cascade autocorrelation model prevents the detection of rapid pitch changes such as in the sequence of pure tones shown in Figure 2 . The left panel in Figure 3A clearly shows that the cascade autocorrelation model fails to distinguish the pitches of individual tones in the tone sequence used in Figure 2 , while the left panel in Figure 3B shows that the SACF model does so fairly well . Therefore , stimulus-dependent changes in the effective integration windows are required . Autocorrelation is usually considered to be a simplified phenomenological model of pitch perception , which is not straightforward to implement in a biologically plausible way [8] , [43] . This is also the case for the proposed model . Nevertheless , an alternative , more formal way to express the second and third model stages ( equation 2 ) is shown in equation ( 3 ) , below . This is equivalent to an expression for the response of a neural population which integrates activity from the previous stage [42]: ( 3 ) The dot indicates a partial temporal derivative and τn is defined as the processing time constant of an idealized homogeneous population of neurons at stage n . The “activation” functions , Ψn , in equation ( 3 ) , which typically use a fixed sigmoid function in standard models of neural assembles [51] , are in the model proposed here time-dependent multiplicative gains: ( 4 ) where ω1/λ1≡0; and ωn , λn are defined in the next section . Substituting equation ( 4 ) into equation ( 3 ) and integrating , allows us to obtain the effective integration windows , En ( t ) , used in equation ( 2 ) : ( 5 ) In contrast with the feed-forward model , the goal of the feedback processing ( dotted lines in Figure 1 ) is to detect unexpected changes in the input stimulus , such as the offset of a tone in a sequence , and to modulate the integration times involved in the feed-forward processing when such changes occur . In the case where the stimulus is constant the pitch predictions at successive time steps will not differ . However , if the stimulus changes then the height of the peak corresponding to the current pitch prediction 1/Ln ( t ) will change from one time step to the next . A mismatch between the pitch predictions at each level and the pitch expectation therefore indicates a change in the input stimulus . A stimulus change typically requires a fast system response , so that information occurring around the time of the change can be updated quickly; this corresponds to using small En ( t ) values . Thus , during periods when there is a significant discrepancy between the current and expected pitch estimates , the effective integration time windows at both integration stages should become very short , so that the “memory” component of the model response is reduced to near zero and essentially reset . Similar rapid changes of activity in response to variations in the input have been previously reported in neural ensemble models [51] , [52] . Figure 2C illustrates the dynamics of E2 ( t ) ( solid green line ) and E3 ( t ) ( dotted red line ) in response to a random tone sequence , the spectrogram of which is shown in Figure 2A . After the end of each tone , both time constants , E2 and E3 , decrease for a brief period of time and then recover back to their maximum values ( En ( t ) ≈τn ) when the next tone begins . As E2 is lag-dependent , the values plotted in Figure 2C represent the integration time constant at the lag , L2 ( t ) , corresponding to the current maximum of A2 ( t , l ) . The small overshoots after the initial dips in E2 reflect transient variations in L2 before a new stable prediction is achieved . The effective integration windows , En ( t ) , can vary over a large range of values , far exceeding the range of plausible neural time constants . However , it should be noted that the neural processing time constants used in the model , τn ( see equation 3 ) , only take on biologically plausible values ( shown in Table 1 ) . The effective integration windows , derived from the activation functions ( equation 5 ) , do not represent neural processing time constants . This aspect will be further addressed in the Discussion section . During the steady-state portions of each tone , the model essentially behaves like the cascade autocorrelation model [13] . The feedback mechanism simply allows the model to adapt quickly to changes in the stimulus . A natural measure of the mismatch between pitch expectations and pitch predictions is the relative error gradient of the maximum response in An ( t , Ln ) , ( 6 ) where the expected lag , LE , is fixed in the temporal derivative; and Ln ( t ) is the lag corresponding to the maximum response at each time step as defined earlier . The gradient at stage three in the model , is an “error” measure: if there is mismatch between the expected pitch estimate and the current prediction , i . e . , LE ≠ L3 ( t ) , then ρ3<0 . Similarly , at the second stage , ρ2<0 represents a mismatch , or error , between the expected pitch and the current intermediate prediction at stage two , 1/L2 ( t ) . The goal of the feedback modulation triggered by changes in the stimulus is to adjust the effective time constants En ( t ) . The error gradients ρn give us a measure of stimulus change therefore , when ρn is negative enough ( compared to a threshold value θn ) there is a discrepancy between the pitch prediction and the pitch expectation which requires that the time constants be adjusted . This is achieved by temporarily activating the recurrent term in equation 4 , i . e . , by defining ( 7 ) where Θ ( x ) is the Heaviside function ( equal to unity if x>0 and zero otherwise ) and θn are small positive thresholds for the error terms , ρn . For example , during the gaps between tones in a sequence of tones , ρn<−θn and the gains ωn ( t ) /λn ( t ) temporarily become nonzero , thereby modulating the effective temporal integration windows , En ( t ) . This approach leads to a problem with the model as described so far in that the response to stimuli where there is a continuous discrepancy between expectations and predictions , very short effective time windows ( En ( t ) ≪τn ) produce oscillatory responses which do not correspond to the stable pitch perceived by listeners ( see , for example , Figure 3B , right panel ) . The dynamics of the ‘adaptation’ variable , λn ( t ) , defined in equation 8 below , serve to modulate uncontrolled corrections to the effective integration windows . Initially the value of λn ( t ) is small ( λn ( 0 ) ≪τn ) so that when change is first detected En ( t ) also becomes small ( equation 5 ) . However , in situations where there is a continuous mismatch between the predicted and the expected pitch , λn ( t ) grows and En ( t ) recovers to a value closer to τn . Then , when there is no longer any discrepancy between expectation and prediction , λn ( t ) recovers to a small value again but without affecting En ( t ) because , in the absence of a mismatch , ωn = 0 . Therefore , the dynamics of λ are described in general by: ( 8 ) Where η and μ are the constants that control the rate of increase in λ during periods of mismatch and the rate of decay in λ during periods where no mismatch occurs . Figures 2C and 8B illustrate two opposite instances of the effect of this top-down processing . In response to a sequence of tones , the effective integration windows shorten precisely at the tone offsets before returning to their maximum values , τn , during the tones ( Figure 2C ) . In response to a click train with alternating inter-click intervals ( Figure 8B ) , the window length settles to a maximum value after a longer period of transient fluctuations . Figure 4 illustrates the discrete processing steps of the model in the form of a flowchart . Table 1 gives the set of parameter values used in the simulations . Further neurobiological justifications for the model are presented in the Discussion . A Matlab-based software implementation of the model is freely available from the first author .
Hall and Peters' experiment highlighted an unsolved problem concerning the balance between synthetic and analytic listening in response to a sequence of pure tones [14] , [15] . The stimuli of the pioneering Hall and Peters' study [14] consisted of three tones played sequentially either in quiet ( Figure 5A , left panel ) or against a background of white noise ( Figure 5A , right panel ) . Each tone lasted 40 ms and was separated from the following tone by a gap of 10 ms . Tone frequencies were 650 , 850 and 1050 Hz ( similar results were obtained with a harmonic sequence ) . The overall level of the noise was about 15 dB above the level of the tones . The individual tones in the sequence were perceived in both conditions . In the experiment , listeners were instructed to match the lowest pitch that they perceived , and in the quiet condition , this was the first of the tones ( 650 Hz ) . However , in the noise condition , the non-simultaneous tones combine to create a lower global pitch of about 213 Hz , which is not perceived in the quiet condition . Recently , it was shown that the cascade autocorrelation model , which used two fixed integration stages , could account for the perception of the global pitch in the noise condition when the time constant of the second stage was long enough [13] . However , the same , long , integration stage could not be used to simultaneously predict the perception of the individual tones in quiet . Figure 5B shows the responses A3 ( t , l ) over time . As in Figure 2 , the responses after each time step have been normalized for visualization purposes ( however , it should be noted that their real magnitudes , which are close to zero during the silent gaps , are not evident in the figure ) . The maximum of A3 ( t , l ) correctly predicts the pitches perceived in quiet , which correspond approximately to the frequencies of the individual tones at each moment in time ( left plot ) . Thus , the peak in the profile of the final response at the end of the stimulus correctly reflects the period of the last tone of the sequence at 0 . 95 ms , and the lowest reported pitch corresponds to the first tone in the sequence ( horizontal arrow in Figure 5B ) . However , when background noise is present ( Figure 5B , right plot ) , a global pitch gradually emerges ( horizontal arrow in the right plot ) , and the peak in the final response occurs at the reciprocal of the perceived pitch of 213 Hz ( 4 . 7 ms , right panel of Figure 5C ) . The above results match precisely the listeners' responses in this study [14] . Many other studies have explored more explicitly the characteristics of temporal integration in pitch perception . Earlier findings showed that the accuracy of pitch discrimination increases with stimulus duration [1] , [19] , depends on the resolvability of the harmonics [20] , and on the sudden onsets and offsets of overlapping tones [21] , [22] . In Figure 6 , another example of the model's ability to simulate the integration of pitch information across noise-filled gaps is presented [17] , [18] . Figure 6A shows a sequence of two unresolved complex tones of 20-ms duration , containing 100 harmonics of a 250-Hz base frequency , high-pass filtered from 5500 to 7500 Hz . After the first of the tones , there was either a short silent gap ( silent-gap condition ) or a noise-filled gap , having a similar mean level to the harmonic complex ( noise-burst condition ) . Background noise was added to mask distortion products . In their study , Plack and White reported that subjects perceived pitch continuity through the gap in the noise-burst condition , but not in the silent-gap condition [17] . The normalized model output A3 ( t , l ) ( Figure 6C ) is qualitatively consistent with a continuous pitch sensation in the noise-burst condition ( right panel ) , which does not occur in the silent-gap condition ( left panel ) . Conditions under which pitch encoding is affected by the presence of other sounds have been also studied using non-simultaneous stimuli such as temporal “fringes” ( consisting of complex tones played immediately before and after a “target” tone ) [16] , [53] , [54]; and by mistuning delayed harmonics of the complex [12] , [55]–[57] . The model described here also accounts for the “reset” of pitch information occurring for large frequency differences between fringe and target tones [53] ( data not shown ) . The previous section shows the model's ability to generate stimulus-dependent changes in the effective time scale of temporal integration for extracting pitch information . This raises the question of whether the ability of the model to adjust the effective integration windows could also account for the temporal resolution of the auditory system . While there is substantial evidence for temporal integration in pitch perception , temporal resolution in pitch perception is perhaps still poorly understood . Therefore , we conducted a psychoacoustic experiment specifically to investigate the temporal resolution of pitch information . It should be stressed that this experiment was conducted independently of the model development and was subsequently used to test the model's predictions . Figure 2B showed that the model uses very short integration times for pitch information when a change in pitch occurs . However , it is possible to construct a class of stimuli , in which the periodicities change continually over very short time scales but which nevertheless elicit a single pitch [49] , [50] , suggesting that pitch information is integrated across these rapid changes in periodicity . The stimuli in question are high-pass-filtered click trains where the interval between successive clicks varies . Previously we showed that the cascade autocorrelation model with fixed integration times [13] predicted the pitch percept elicited by a range of click train stimuli , which had proved problematic for conventional autocorrelation models [49] , [50] , [60]–[63] . Here , we test whether the current model ( which generalizes the model reported in [13] by including variable integration times ) retains this ability . This is an important question , because a rapid reset of pitch information is apparently in contradiction with the long-term integration used in [13] , as was illustrated in the Methods section ( Figure 3 ) . As an example , Figure 8 shows the response of the model to one of these stimuli . In this case , the inter-click intervals alternate between 4 and 6 ms , but listeners usually report a single pitch somewhere in between these extremes and closer to the longer interval . Carlyon et al . [49] , [50] presented the click trains with a duration of 400 ms . Stimuli were band-pass-filtered with cut-off frequencies of 3900 and 5300 Hz in order to avoid the harmonic spectral components being resolved by the cochlear filters . They also added a pink noise to avoid audible distortion products . Carlyon et al . [50] demonstrated that the combined auditory nerve responses , measured as compound action potentials ( CAPs ) , were stronger for the largest inter-click interval ( 6 ms ) than for the shorter interval ( 4 ms ) . Therefore , they suggested that a population of more central neurons , which respond only when their inputs exceed a fixed threshold value , would respond preferentially to the longer intervals , thereby explaining listeners' preference for matching a pitch close to 6 ms . Figure 8C shows that the predicted pitch of the model ( red highlight ) varies almost randomly for approximately 80 ms and then progressively stabilizes at a lag in the region of 5 . 5–6 ms ( see horizontal arrow in Figure 8C ) . Thus , the model prediction is in good agreement with the geometric average of the reported pitch values ( shown by vertical dashed line in Figure 8D ) . While the final snapshot of A3 ( tfinal , l ) ( Figure 8D ) peaks close to the geometric mean of the reported pitches ( vertical dashed line ) , there are other prominent peaks in A3 ( tfinal , l ) close to this maximum; this is consistent with the large variability in reported pitches for these alternating click trains . A prediction of the model yet to be tested is that no reliable pitch estimate would be possible for stimuli shorter than 100 ms . To conclude , it is worth remarking that this model can similarly account for the pitches of the other click train stimuli considered in [13] . The model proposed here is not a formal model of neural populations; nevertheless , it is neurophysiologically based ( see Methods and Discussion sections ) . This raises the question as to whether the model can explain aspects of the responses of neural ensembles in a pitch perception task . Krumbholz et al . [37] identified a transient neuromagnetic response in Heschl's Gyrus , which they termed the “pitch onset response” ( POR ) . In their experiment , they used iterated rippled noise ( IRN ) stimuli with delays of 4 , 8 , 12 and 16 ms . IRN differs from the RN stimulus described previously in that the delay-and-add process is iterated N times . Increasing the number of iterations , N , increases the degree of serial correlation and therefore the pitch strength . Figure 9A shows the spectrogram of an IRN stimulus with a 12 ms delay and 16 iterations . Neuromagnetic responses were recorded to the onset of an IRN , which was directly preceded by an uncorrelated noise with the same energy and spectral composition . Recordings showed that the transition from noise to IRN produced a reliable POR with a mean latency of approximately four times the delay , d , plus a constant offset of about 120 ms ( left panel in Figure 9D , solid blue line ) . The authors concluded that the POR reflects pitch-related processing within Heschl's Gyrus in the human auditory cortex . This has been supported by other more recent studies [36] . Figure 9B shows the output of the model , ( A3 ( t , l ) without any normalization , in contrast to previous plots ) , for the example shown in Figure 9A . After some time the maximum of A3 ( t , l ) ( red colour ) stabilises and becomes prominent . The predicted pitch is the reciprocal of L3 = 12 ms , which corresponds to the delay of the IRN stimulus . However , the maximum value of A3 ( t , L3 ) in Figure 9B emerges gradually . Therefore , there seems to be no obvious correlate of the latency at around 150 ms of the measured cortical response in the model . A number of previous studies have suggested that the temporal derivative of the neural population responses at lower levels of processing might correlate with the measured activity in higher ( i . e . , cortical ) levels [44] , [45] , [64] . Therefore we investigated whether the latency of the pitch onset response might correspond to the latency of the peak in the derivative of A3 ( t , L3 ) . Here , we calculated a smoothed version of the temporal derivative of A3 ( t , L3 ) by convolving A3 ( t , L3 ) with the first differential of a Gaussian function ( representing connection efficacies to higher areas [44] , [45] ) . We then used the first maximum of this smoothed derivative to predict the latencies of the POR for different pitch values . Figure 9C illustrates the smoothed derivative of A3 ( t , L3 ) for the example shown in Figure 9A ( red dotted line ) . The derivative has a maximum at approximately 168 ms , which is consistent with the POR latency for this condition . The green solid line shows the variance of A3 ( t , l ) ( calculated at each fixed t ) for the same stimulus . It appears that the variance of A3 ( t , l ) , which might be taken to represent the uncertainty of the pitch estimate , reaches a minimum at a similar time as the derivative of A3 ( t , L3 ) reaches a maximum ( in general , however , the smoothed derivative is a more accurate predictor of the experimental latencies ) . The red dotted line in Figure 9D ( left plot ) shows the time at which the smoothed derivative of A3 ( t , L3 ) reaches its first maximum as a function of pitch value , which appears to correlate remarkably well with the POR latencies ( solid line ) . Krumbholz et al . [37] also found that the POR latency mainly depended on the delay of the IRN stimulus and was influenced little by the number of iterations . The right panel in Figure 9D shows the latencies when the delay is fixed at 16 ms and the number of iterations varies ( solid line ) . Consistent with experimental results , the number of iterations of the stimulus do not significantly affect the smoothed derivative of A3 ( t , L3 ( t ) ) ( dotted line ) . To conclude , it is worth mentioning that the model also accounts for the minimum duration of IRN stimuli for reliable perceptual discrimination of the pitch , also reported in [37] . The solid line in Figure 10 indicates the average perceptual results . The dashed line shows the duration of the transient period in A3 ( t , L3 ) , i . e . , the time window during which the pitch prediction is not stable ( e . g . around 100 ms in the stimulus shown in Figure 8C ) . Clearly , the model simulations match the data extremely well ( dashed line in Figure 10 ) . Therefore , the initial period in which the model output varies rapidly seems to correlate with unstable pitch perception . This model prediction might be valid not only for IRN stimuli but also for other pitched sounds .
We propose a neurocomputational model to explain the observed paradox between temporal integration and temporal resolution in the auditory processing of pitch information . Our goal was to capture essential elements in the temporal dynamics of pitch perception within a unified framework . This model is an extension of the autocorrelation theory of pitch perception formulated in terms of equations describing the activity of neural ensembles [51] , [65] , and extended to include feedback processing . The principal novelty of the model is the suggestion that top-down connections to sub-cortical areas determine the temporal dynamics of auditory perception , and that this influence is mediated through feedback modulation of recurrent inhibitory circuits . As a result , the responses at each stage adapt to recent and relevant changes in the input stimulus; i . e . , feedback in the model essentially determines the dynamics of the “effective” integration window used at each stage . This approach is consistent with the available neuroimaging data: a sustained pitch response ( SPR ) in lateral Heschl's Gyrus has been shown to adapt to the recent temporal context of a pitch sequence , enhancing the response to rare and brief events [36] . The successful explanation of the latency of the Pitch Onset Response ( Figure 9D , left plot ) further supports the neurobiological validity of this model . Therefore , we hypothesize that the model captures some fundamental processing aspects of pitch processing , occurring up to Heschl's Gyrus [37] . Consistent with this , a recent study also suggests that the auditory sensory thalamus processes fast changes in speech , which appears to be modulated by slower contextual states [66] . It should be noted that efferent connections to the auditory peripheral model have not yet been implemented , although there is evidence for those connections too [67] . The addition of this connection could provide a method for controlling the cochlear model , a current focus of our investigations . Although highly idealized , the model uses a minimal set of biologically plausible parameters . The values shown in Table 1 were optimized for generating the correct temporal dynamics of the effective integration windows in the global pitch of non-simultaneous tones , and the pitches of click trains , described in the Results section . Neither the gap detection threshold nor the POR latency experiments were used for parameter optimization; they therefore stand as tests of the generalization of the model . The current model might thus serve as a basis for more realistic neurophysiological models in the future . In fact , the model responses during the offsets of tones are similar to responses of neurons to amplitude modulated pure tones measured in the superior paraolivary nucleus ( SPON ) of rats [68] . Remarkably , a short gap between tones was found to produce a significant burst of spikes; i . e . , a change in neural activity of several orders of magnitude in less than a millisecond during discontinuities between the tones . Consistent with this data , the model responses An ( t , l ) vary very quickly at tone offsets , because the effective integration windows become very short at these discontinuities ( Figure 2C ) . Interestingly , this very fast offset response in SPON neurons is not a feed-forward process , but is modulated by feedback from neurons in the medial nucleus of the trapezoid body , which inhibit the SPON [68] . The model architecture shown in Figure 1 is similar to this type of feedback inhibitory circuit . In some ways ( see Text S1 ) the model can be understood as a special case of a more general class of models: the hierarchical generative models ( HGMs ) of sensory processing [38] , [39] , [69] . In the HGM approach , it is assumed that higher areas have access to more abstract and contextualized information , and therefore produce a more refined expectation of the next sensory input . Lower areas deal with more detailed information and generate intermediate predictions [38] . A mismatch between these two predictions generates an error , which propagates from the upper level to the level immediately below and minimizes the free energy of the model [38] , [69] . This is shown in Text S1 , where a comparison between the proposed model and HGMs is presented . Very recently , Kiebel and colleagues also showed that the minimisation of the free energy can be used to invert temporal hierarchies in the processing of bird songs [70] . In summary , we propose a unified model to explain the stimulus-dependency of the time constants of temporal processing in auditory perception . We suggest that one possible role for efferent connections in the auditory system is to detect perceptually relevant changes in the temporal patterns of afferent activity and to adapt the effective processing time constants to the stimulus characteristics . Currently , we are not aware of any studies that have explicitly tested the role of efferent signals in pitch perception , thus , this hypothesis has yet to be tested . Nevertheless , a prediction of the model is that blocking the feedback circuits would impair the ability to separate sounds over time . Recent experimental studies in cortical cooling [71] may provide a methodology for further testing this proposal . | Pitch is one of the most important features of natural sounds . The pitch sensation depends strongly on its temporal context , as happens , for example , in the perception of melody in music and prosody in speech . However , the temporal dynamics of pitch processing are poorly understood . Perceptual studies have shown that there is apparently a wide range of time scales over which pitch-related information is integrated . This multiplicity in perceptual time scales requires a trade-off between temporal resolution and temporal integration , which is not exclusive to pitch perception but applies to auditory perception in general . As far as we are aware , no existing model can account simultaneously for the wide range and stimulus-dependent nature of the perceptual phenomenology . This article presents a neurocomputational model , which explains the temporal resolution–integration trade-off observed in pitch perception in a unified fashion . The main contribution of this work is to propose that top-down , efferent mechanisms are crucial for pitch processing . The model replicates perceptual responses in a wide range of perceptual experiments not simultaneously accounted for by previous approaches . Moreover , it accounts quantitatively for the stimulus-dependent latency of the pitch onset response measured in the auditory cortex . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"otolaryngology/audiology",
"computational",
"biology/computational",
"neuroscience"
] | 2009 | Understanding Pitch Perception as a Hierarchical Process with Top-Down Modulation |
Nelfinavir is a potent HIV-protease inhibitor with pleiotropic effects in cancer cells . Experimental studies connect its anti-cancer effects to the suppression of the Akt signaling pathway , but the actual molecular targets remain unknown . Using a structural proteome-wide off-target pipeline , which integrates molecular dynamics simulation and MM/GBSA free energy calculations with ligand binding site comparison and biological network analysis , we identified putative human off-targets of Nelfinavir and analyzed the impact on the associated biological processes . Our results suggest that Nelfinavir is able to inhibit multiple members of the protein kinase-like superfamily , which are involved in the regulation of cellular processes vital for carcinogenesis and metastasis . The computational predictions are supported by kinase activity assays and are consistent with existing experimental and clinical evidence . This finding provides a molecular basis to explain the broad-spectrum anti-cancer effect of Nelfinavir and presents opportunities to optimize the drug as a targeted polypharmacology agent .
Tremendous effort has been directed at rational drug design where one strives to understand , and subsequently optimize , how a small molecule interacts with a single protein target and impacts a disease state . However , such approaches are less fruitful in discovering effective and safe therapeutics to treat complex diseases such as cancer . It is suggested that the inhibition or activation of a single specific target may fail owing to the inherent robustness of the underlying biological networks causing the disease state [1] , [2] , [3] , [4] , [5] , [6] . The goal then is to perturb multiple relevant targets . Perturbation may be achievable through the use of drug cocktails , or possibly through a single drug that has the appropriate polypharmacological effect [1] , [2] , [4] , [6] , [7] , [8] , [9] , [10] , [11] . To rationally design such a drug is a very complex problem that begins by identifying the targets to which that drug binds . Here we address a much simpler problem , that is , to take a drug that is already believed to show this effect and attempt to explain why it might be so . Nevertheless , we must still begin by identifying the multiple targets to which it binds . To this end , we have developed an off-target pipeline to identify protein-drug interaction profiles on a structural proteome-wide scale . The off-target pipeline integrates our previous chemical systems biology approach [12] , [13] , [14] with algorithms that accurately estimate binding affinity . We then use the target list predicted from the off-target pipeline to suggest physiological outcomes from the associated biological networks and determine how well these outcomes map to what is observed clinically . The extension to our previous approach presented here is to better estimate the binding affinity in forming a protein-ligand complex , as both experimental and theoretical studies suggest that even weak binding to multiple targets may have profound impact on the overall biological system [15] , [16] , [17] . Available computational tools that quantitatively study protein-ligand interactions are based predominantly on protein-ligand docking and free energy calculations for the protein-ligand complex [18] , [19] . A formidable task then is to include protein flexibility into the binding affinity calculation since errors in scoring mainly result from the use of rigid protein conformations [20] . The modeling of protein flexibility requires computationally intensive molecular dynamic ( MD ) simulations . However , it is impractical to apply MD simulation to the whole structural proteome . Our approach pre-filters the structural proteome to find the most likely cases to apply MD . Specifically , we undertake a human structural proteome-wide ligand binding site comparison using previously developed algorithms [21] , [22] , [23] and add intensive binding free energy calculations , based on protein-ligand docking , MD simulation and MM/GBSA free energy calculations . We apply this strategy to explore the molecular mechanism for the observed anti-cancer effect of Nelfinavir , a human immunodeficiency virus ( HIV ) protease inhibitor . Recently , Nelfinavir has been repurposed for cancer treatment [24] , [25] , [26] . However , its molecular targets remain unknown . The majority of published data indicates that the drug suppresses the Akt signaling pathway [27] . In human , the Akt family includes the serine/threonine protein kinases Akt1 , Akt2 and Akt3 . These proteins are involved in cell survival , protein synthesis and glucose metabolism and are considered markers for many types of cancer [28] , [29] , [30] . Akt3 is also known to be stimulated by platelet-derived growth factor ( PDGF ) , insulin and insulin-like growth factor 1 ( IGF1 ) [31] . Thus inhibition of the Akt pathway may also cause insulin resistance and diabetes , a phenomenon observed as a side effect of treatment by HIV protease inhibitors . Currently , there is no experimental evidence to suggest that Nelfinavir binds directly to members of the Akt family , rather it has been suggested that the drug acts upstream of the Akt signaling pathway [32] . Using our structural proteome-wide off-target pipeline , we find that multiple members of the protein kinase-like superfamily as off-targets of Nelfinavir . Most of these protein kinases are found upstream of the Akt , MAPK , JNK , NF-κB , mTOR and focal adhesion pathways . We hypothesize that this weak but broad spectrum protein kinase inhibition by Nelfinavir contributes to the therapeutic effect against different types of cancer . Our hypothesis is supported by kinase activity assays and consistent with other existing experimental and clinical observations . This suggests that the next challenges are specifically to optimize Nelfinavir as a targeted polypharmacology agent , and more generally , to determine whether our computational protocol can be applied to other systems .
The steps in our off-target pipeline are shown in Figure 1 . In the first step , the Nelfinavir binding pocket in the HIV protease dimer structure ( PDB Id: 1OHR ) was used to search against 5 , 985 PDB structures of human proteins or homologs of human proteins using the SMAP software ( see Supporting Information ( SI ) and methods for details ) , which is based on a sensitive and robust ligand binding site comparison algorithm [21] , [22] , [23] . Hits are considered significant if the SMAP p-value <1 . 0e-3 . In step 2 , the binding poses and affinities of Nelfinavir to these putative off-targets are estimated using two docking methods , Surflex [33] and eHiTs [34] , starting from the superimposed binding sites . If the docking score indicates severe structural clashes between Nelfinavir and the predicted binding pocket , the protein is removed from the off-target list . After filtering by SMAP and the two docking programs , 92 putative off-targets remained for further analysis ( SI , Table S1 ) . Among them , the top 7 ranked off-targets belong to the aspartyl protease family that is the fusion form of the primary target HIV protease dimer . The remaining 85 proteins belong to different global folds from the primary target . These off-targets are dominated by protein kinases ( PKs ) ( 51 off-targets ) and other ATP or nucleotide binding proteins ( 17 off-targets ) . The distribution of the 51 protein kinases on the human kinome tree [35] is shown in Figure 2 . Even though these protein kinases have a broad distribution among the different protein kinase families , the majority of predicted off-targets belong to the tyrosine kinase , cAMP-dependent , cGMP-dependent and protein kinase C families . This distribution is more pronounced with a stringent SMAP p-value smaller than 1 . 0e-4 ( green in Figure 2 ) . The 12 top-ranked PKs with p-value smaller than 1 . 0e-4 were subject to detailed protein-Nelfinavir docking and 10 of them were further investigated through computational intensive molecular dynamic simulations and MM/GBSA binding free energy calculations . The SMAP alignments between the PKs and the Nelfinavir binding sites reveal that ATP and its competitive inhibitors bind in the vicinity of the predicted binding sites . An example is shown in Figure 3 for the case of epidermal growth factor receptor ( EGFR ) protein kinase domain ( PDB id: 2J6M ) . The superimposed Nelfinavir is accommodated in the protein kinase inhibitor binding pocket and overlaps with the co-crystallized EGFR inhibitor ( PDB ligand id: AEE ) . If amino acid residues with atomic distances less than 5 . 0 Å to the inhibitors are considered as the binding site , approximately 73% ( 16/22 ) of the known AEE binding site residues are included in the predicated binding site of Nelfinavir to EGFR . Binding poses and affinities of Nelfinavir to the identified PK binding sites are firstly estimated using the docking software eHiTs [34] and compared to the binding affinities of co-crystallized inhibitors in those PKs . The binding pose of Nelfinavir is optimized from its superimposed conformation obtained from the SMAP output rather than by an ab initio global conformational search . Systematic errors in the scoring function are cancelled out by using a normalized docking score ( NDS ) [13] . A large negative value for the NDS indicates a higher likelihood of binding . The predicated binding affinity of Nelfinavir is comparable to that of co-crystallized inhibitors for several classes of PKs , notably EGFR ( SI Table S2 ) . The NDS for EPHA2 is 1 . 328 , which implies that the docking score of Nelfinavir to EPHA2 is higher than for randomly selected molecules . This protein was removed from further calculations . In order to get more accurate estimates for the binding affinities , MM/GBSA calculations were performed on 10 PK hits filtered by the SMAP binding site similarity search and ligand docking scores . Since in reality binding is dynamic , the structure will adopt different conformations during binding , and this should be anticipated . Hence one should generate a statistically sufficient ensemble from molecular-dynamics trajectories and compare the resulting ensemble averages to obtain a more reliable binding free energy value . Recent studies on MM/GBSA binding free energy calculations show that a nanosecond scale MD simulation is sufficient to perform a meaningful MM/GBSA calculation [36] , [37] , [38] . Here , the binding free energies averaged over 200 snapshots from the last 2 ns trajectory of an 8 ns MD simulation are listed in Table 1 for the complex structures of protein kinases bound with Nelfinavir and co-crystallized ligands . To estimate the stability of the MD simulations , structural root-mean-square-deviations ( RMSDs ) for receptor backbone atoms and ligand non-hydrogen atoms are examined as a function of time ( Supporting Information Figure S1 ) . The RMSD is calculated based on superimposed structures fitting to the first frame of the 8 ns MD simulation using the coordinates of the receptor backbone atoms . Thus , RMSD values for the ligands reflect both internal and rigid body movements relative to the protein . In all cases , RMSDs for the receptor backbone atoms are well below 3 Å for the last 2 ns simulation , indicating robust simulations and reasonable samplings for the MM/GBSA binding free energy calculation . The conformational fluctuation of Nelfinavir bound to EPHB4 and FGFR are higher than in other targets . Structural analysis of their trajectories shows that Nelfinavir moves out of the binding pockets of EPHB4 and FGFR during simulation , which indicates EPHB4 and FGFR may not be good candidates for Nelfinavir interaction . Here , the MM/GBSA binding free energy calculation includes gas-phase energies , solvation free energies and entropy contributions . As shown in Table 1 , if only gas-phase energies and solvation free energies , i . e . , total binding enthalpy , are taken into account , Nelfinavir shows comparable binding affinity to the co-crystallized ligands . However , when considering the loss of entropy during binding , Nelfinavir becomes less favorable than the co-crystallized inhibitors due to its larger size and flexibility . For example , when AEE enters the binding pocket of EGFR , the entropy change for the whole system is 14 . 16 kcal/mol . However , the binding of Nelfinavir to EGFR causes an 18 . 12 kcal/mol entropy losses for the whole system . Thus , even though the entropy contribution is smaller than the enthalpy contribution , the binding free energy difference between Nelfinavir and AEE comes predominantly from the entropy change and this part of the free energy cannot be omitted in providing a reliable estimate of binding affinity . Ligand binding pose and atomic interactions between ligand and protein kinases are also important factors when measuring ligand binding . The predicted binding pose of Nelfinavir significantly overlaps with the known inhibitors of EGFR , IGF-1R , FAK , Akt2 , CDK2 , ARK and PDK1 ( SI , Figure S2 and S3 ) . The structure of Nelfinavir can be fragmented into five moieties: the 2-methyl-3-hydroxy-benzamide portion A , the S-phenyl group B , the tert-butyl carboxamide moiety C , the lipophilic dodecahydroisoquinoline ring D and the central hydroxyl group E ( SI , Figure S4 ) . The benzamide ring A in the predicted conformations superimposes well onto the aromatic groups of the co-crystallized inhibitors for these protein kinases , and plays a critical role in molecular recognition [39] . For other predicted protein kinases , the binding pose of Nelfinavir still partially overlaps with their respective co-crystallized inhibitors and occupies the ATP-binding pockets . Most of the hydrogen-bond interactions and hydrophobic interactions between protein kinases and co-crystallized inhibitors could be found between Nelfinavir and the respective protein kinases . As shown in Figure 4 , the hydrogen bond between the pyrrolopyrimidine core of AEE and the main chain amide of Met793 on EGFR is maintained between benzamide hydroxy O38 of Nelfinavir and the same atom on EGFR . This hydrogen bond interaction is critical for protein-ligand binding in EGFR . Missing this hydrogen bond will cause ∼3 , 700-fold loss of inhibitor potency in EGFR [40] . Residues that form hydrophobic interactions with AEE are also close to Nelfinavir and provide appropriate hydrophobic interactions as shown in Figure 4 . These conserved hydrogen bond interactions and hydrophobic interactions support the binding of Nelfinavir to EGFR . Similar conserved hydrogen bond interactions and hydrophobic interactions are observed for other protein kinases , excluding FGFR , EPHB4 and Abl , that is , where Nelfinavir partially overlaps with the co-crystallized ligands . The binding free energies for Nelfinavir to FGFR and EPHB4 also indicate that the binding affinities of Nelfinavir to these two proteins are weaker than the other eight protein kinases . In summary , MM/GBSA binding free energy , ligand binding pose , conserved hydrogen bond interactions and hydrophobic interactions supports the direct interaction of Nelfinavir with EGFR , which has been shown as a possible Nelfinavir target based on ligand binding site similarity and from experimental studies by others [25] . For FGFR , EPHB4 and Abl , the results from MD simulation and MM/GBSA free energy calculations indicate that Nelfinavir is unlikely to bind to these three targets . For other targets , IGF-1R , FAK , Akt2 , CDK2 , ARK and PDK1 , the calculated binding free energies and predicted ligand binding poses suggested the possible inhibition by Nelfinavir , even though there is no experimental support at this time . Given that computationally EGFR and Akt2 show favorable binding affinities for Nelfinavir , MD simulation and MM/GBSA binding free energy calculations were extended to other members of the EGFR ( ErbB2 , ErbB4 ) and Akt families ( Akt1 and Akt3 ) . As shown in Figure 5 , the binding free energies for EGFR , ErbB2 , ErbB4 , Akt1 , Akt2 and Akt3 are -15 . 60 , -25 . 76 , -31 . 83 , -15 . 39 , -19 . 25 and -12 . 13 kcal/mol , respectively . A HTRF® TranscreenerTM ADP Assay of 20 µM Nelfinavir was undertaken for EGFR , ErbB2 , ErbB4 and Akt ( Akt1 , Akt2 and Akt3 ) in an effort to verify the predictions from the MM/GBSA calculations . Weak inhibition by Nelfinavir is detected for ErbB2 ( Figure 5 ) . The lower binding free energy of ErbB2 is consistent with its higher inhibition rate and the experimental and computational results both show inhibition of the EGFR family by Nelfinavir . Considering that a prescribed dose of Nelfinavir is 1 , 250 mg ( 2 . 2 mmol ) ( http://www . rxlist . com/viracept-drug . htm ) , the plasma concentration of Nelfinavir in HIV patients can reach 7-9 µM [41] . However , these concentrations only achieve a partial reduction of cancer cell proliferation and are not efficient in inducing apoptosis in cancer cells . Most cellular activity studies require concentrations of Nelfinavir greater than 20 µM [42] . At such high concentration , Nelfinavir demonstrated specific anti-cancer activity with no reports of non-specific binding . As such , it is not likely that the specific in vivo and in vitro anti-cancer activity when using a high concentration of Nelfinavir is due to its aggregation . Likewise , when the same concentration of Nelfinavir is used in our kinase assay , it is unlikely that Nelfinavir is aggregated [43] . Since the assay may not be sensitive enough to detect weak bindings , most of assay results are inconclusive . It is necessary to develop more robust assay methods for determining weak bindings . The inhibition of EGFRs by Nelfinavir is consistent with Gills et al . ’s work on exploring the effect of HIV protease inhibitors on endogenous and growth factor induced Akt activation [25] . In their study , 20 µM Nelfinavir reduced the activation of EGFR , IGF-1R and Akt signaling pathways . The decreased phosphorylation of EGFR , IGF-1R and Akt directly in response to EGF or IGF-1 indicates that Nelfinavir can compete with EGF or IGF-1 and act at the plasma membrane to inhibit growth factor receptors . However , the inhibition of Akt activation by Nelfinavir is weaker than that observed using a known PI3K inhibitor and the effect is transient , which may suggest a weaker inhibition of EGFR or IGF-1R by Nelfinavir . No obvious inhibition of Akt1 and Akt3 by 20 µM Nelfinavir is observed . Even though the ADP assay was not applied to every predicted protein kinase , the comparable computational results indicate the possibility that Nelfinavir may also inhibit other protein kinases through weak interactions . Nelfinavir is the most potent inhibitor in cell proliferation and Akt activation studies [25] . To compare Nelfinavir with other protease inhibitors , MD simulation and MM/GBSA binding free energy calculations were applied to two other protease inhibitors , Saquinavir and Indinavir . Saquinavir has the most similar inhibition effect to Nelfinavir in the cell proliferation analysis involving 60 cell lines derived from nine different tumor types and Indinavir has the weakest effect on cell proliferation [25] . Autodock Vina [77] was applied to get the starting structures for Saquinavir and Indinavir when bound to EGFR , ErbB2 and ErbB4 . The docking energies for Nelfinavir , Saquinavir and Indinavir are listed in SI Table S3 and show that there is no significant difference between these three inhibitors . However , the conserved hydrogen bond between Nelfinavir and EGFR cannot be found for either Saquinavir or Indinavir . The calculated MM/GBSA binding free energies for Saquinavir are -8 . 51 , -10 . 12 and -9 . 37 kcal/mol when bound to EGFR , ErbB2 and ErbB4 , respectively and -1 . 11 , -1 . 68 and -2 . 51 kcal/mol , respectively , for Indinavir . Compared with the calculated MM/GBSA binding free energies for Nelfinavir , the less negative values for the binding free energies of Saquinavir indicate weaker binding affinities . This is consistent with the observed effect of these HIV protease inhibitors on Akt activity . The unfavorable binding of Indinavir to the EGFR families is also supported experimentally [25] . Putting together the results from the off-target predictions , docking experiments , MD simulation , MM/GBSA free energy calculations , and kinase activity assays , it appears that Nelfinavir binds to different protein kinase ( PK ) off-targets through relatively weak interactions . The majority of our top ranked Nelfinavir off-targets belong to the receptor tyrosine protein kinase family , including EGFR , IGF-1R , Abl , FGFR and ephrin receptor . The PKs in this family are high affinity cell surface receptors that not only regulate normal cellular processes but also play a critical role in the development of many types of cancers . There are also other PKs identified as off targets for Nelfinavir , such as CDK2 , ARK2 , FAK1 , Akt2 and PDK1 . By examining pathways associated with each individual predicted off-target , we constructed an integrated off-target interaction network ( Figure 6 ) . To simplify the whole network , we only present the interactions between predicted off-targets and the major pathways involved in cancer development and insulin resistance . Effects of these off-targets are not limited to these pathways . Predicted off-targets , represented by yellow circles in the network , regulate PI3K , MAPK , JNK , mTOR , NF-κB and focal adhesion pathways through direct or indirect interactions with intermediate proteins connecting the pathways . Inhibition of predicted off-targets is predicted to down-regulate these pathways , and hence reduce cancer risk and increase insulin resistance . Consider EGFR as an example to show how inhibition by Nelfinavir can result in an anti-cancer effect . Some major effects of EGFR on cellular functions come from its regulation of the PI3K/Akt pathway . As a receptor tyrosine protein kinase , EGFR can be activated by epidermal growth factor and then induce activation of Phosphoinositide 3-kinases ( PI3K ) , resulting in the formation of a PtdIns ( 3 , 4 , 5 ) P3 molecule ( PIP3 in Figure 6 ) . Akt will then bind to PtdIns ( 3 , 4 , 5 ) P3 and be phosphorylated and activated by PDK1 and mTOR . As a consequence , the activation of Akt triggers the downstream response of the Akt pathway , such as phosphorylation of the Bcl-2-associated death promoter ( BAD ) , activation of the NF-κB pathway and inhibition of the retinoblastoma protein ( Rb ) . The inhibition of EGFR by Nelfinavir will reduce Akt signaling , consistent with current experimental evidence . Along with the regulation on the PI3K-Akt pathway , EGFR can also induce the activation of the MAPK and JNK pathway through interaction with Ras [44] , [45] . All these activities have the potential to increase cell survival and cell proliferation and prevent cell apoptosis , as shown in Figure 6 . Conversely , over-activation of EGFR and the associated down-stream pathways could result in uncontrolled cell growth and division . Other predicted off-targets of Nelfinavir , for example , IGF-1R , Abl , FGFR , EPHB4 and FAK , have similar effects to EGFR , again by controlling activation of PI3K and Ras . According to our calculations , Nelfinavir can also bind to PDK1 and ARK . While a different mechanism than EGFR inhibition , it is hypothesized this can lead to regulation of the MAPK and mTOR pathways . PDK1 is crucial for the activation of Akt through direct phosphorylation . CDK2 is also implicated by our off-target analysis . CDK2 is part of the downstream regulation of the PI3K/Akt pathway , and depending on cellular location , can either promote cell cycle progression or cell death [46] . The presence of active nuclear CDK2 during the transition to the G2 phase inhibits the cell cycle progression while Akt-regulated nucleo-cytoplasmic CDK2-relocation is required for cell cycle progression . The dual control of CDK2 on cell proliferation and apoptosis makes it an interesting anti-cancer target . Jiang et al . showed that Nelfinavir can inhibit CDK2 activity in melanoma cells [30] in keeping with our computational findings . In summary , the dominant effect of Nelfinavir through off-target binding to a variety of protein kinases comes from up-stream regulation of the PI3K/Akt pathway . These protein kinases are also hypothesized to regulate other cancer pathways such as MAPK , JNK , NF-κB , mTOR and the focal adhesion pathway . Similarly , Nelfinavir is predicted to inhibit IGF-1R , which regulates the insulin/insulin-like growth factor signaling pathway , and offers one possible explanation for the observed side effects of Nelfinavir on insulin resistance and diabetes .
This study indicates that Nelfinavir is capable of a broad based polypharmacological effect against a number of protein kinases as targets . Determining the total number of possible targets is limited by the availability of the 3-D structures ( or models ) of human proteins . A second limitation might arise based on the versatility of Nelfinavir itself . The binding sites determined here map to the image of the ligand in the conformation it is found when bound to an HIV-1 protease . It might bind to a different target using a different conformation with higher affinity than observed here and these would not be found since the binding pocket itself would be different . Given that existing experimental data indicate that the off-targets to Nelfinavir are involved in the Akt pathway , other potential strong binding off-targets upstream of the identified receptor tyrosine kinases also need to be considered . One of the most likely alternatives is the β-arrestin regulated G-protein coupled receptor signaling transduction pathway which regulates MAPKs , SRC , PI3K , and Akt , and mediates EGFR transactivation [47] . Two major non-kinase proteins involved in the kinase regulation and transactivation of the GPCR signaling pathway are the GPCR and β-arrestin . If the GPCR or β-arrestin is strongly inhibited by Nelfinavir , it is expected that the cellular functions such as GPCR internalization , translocation of smoothened to the primary cilium , and chemotaxis control , which are mediated by the β-arrestin , should be affected [47] . However , the related phenotype changes have not been reported . In addition , no significant hits ( p-value <1 . 0e-5 ) are found for Nelfinavir using the Similarity Ensemble Approach ( SEA ) , which is one of the most sensitive methods to identify GPCR related off-targets [48] , [49] . The SMAP similarity between β-arrestin and HIV protease is not significant ( p-value >1 . 0e-2 ) . Although more analyses are required to determine if Nelfinavir binds to other proteins that indirectly regulate EGFR pathways , the data reported here at least suggest that the pleotropic effect of Nelfinavir comes from the direct inhibition of a variety of protein kinases . A fundamental question raised by this work is whether weak binding of a drug to multiple targets can cumulatively cause strong phenotypic changes ? Existing studies of biological networks have shown that the malfunction of multiple nodes more likely causes the system to fail than the removal of a single node as a result of diversity , redundancy and system control of the biological network . Multiple node failures have been called “fail-on” [50] , and used to explain neurological disorders [51] and cancer [52] , [53] in recent genome-wide studies . Addressing the fail-on phenomenon would require a polypharmacological effect . The therapeutic efficacy of multiple protein kinase inhibitors suggested here has already been demonstrated by less specific protein kinase inhibitors which attack tumors through multiple mechanisms and are used in more than one type of cancer therapy [54] . For example , Sunitinib is the first cancer drug simultaneously approved for two different cancer treatments , namely , renal cell carcinoma and imatinib-resistant gastrointestinal stromal tumor . A protein kinase assay against 113 different kinases shows that Sunitinib can bind to 73 additional kinases apart from its primary target [55] . In another example , moderate micomolar RAF inhibitor PLX4720 is potent in inhibiting downstream signaling and proliferation of the cell harboring BRAF , and in treating melanoma cell lines [56] . In contrast , Sorafenib that was developed as a potent nanomolar RAF inhibitor failed in the clinical trial due to its low anti-melonoma efficacy . Araujo et al . demonstrated the synergistic effect of multiple low-dose inhibition of upstream processes on the attenuation of downstream signals in the EGFR signaling pathway [57] , and suggested that low-dose combination therapy may reduce drug side effects and resistances in the treatment of cancer [58] . Nelfinavir is a potential lead compound in the design of the next generation of anti-cancer drugs . As indicated by the MM/GBSA binding free energies for different protein kinases , the binding affinity of Nelfinavir is weaker than for the original inhibitors . Entropy changes during binding contribute significantly to the differences in binding affinity since Nelfinavir consists of more rotatable bonds and is more flexible than many small molecule protein kinase inhibitors . Covalent bonds could be added to the Nelfinavir structure to reduce the degree of freedom and increase the specificity and binding affinity . On the other hand , it can be hypothesized that the weak binding of Nelfinavir to multiple protein kinases helps avoid severe side effects , but still impacts the system enough to have a positive effect . That is , weak inhibition of multiple protein kinases may be just enough to return the system to a normal state , as suggest by dynamic analysis of model systems [57] , [58] . There are a number of unmet computational challenges in exploiting the concept of multiple weak interactions and designing selective polypharmacology therapeutics , from target identification to lead optimization . Computational techniques that are able to identify optimal combination targets and their inhibition windows in cellular networks have been developed , but their scope is still limited [59] , [60] , [61] . It is well accepted that an optimal lead should balance binding potency and molecular size [62] . A highly potent lead compound usually leads to a drug candidate with high molecular weight , which is often linked to a higher risk of failure in drug development [63] . Analysis of the binding affinity of marketed drugs and natural products indicates that therapeutic efficacy is not necessarily associated with high binding affinity [63] . Moreover , drug-target interactions in vivo are different from those in vitro . An increasing body of evidence suggests that the drug-target residence time , a measurement of the lifetime of the drug-target complex , better correlates to drug efficacy than does the binding affinity [64] , [65] . This suggests that lead optimization should focus on the drug-target residence time instead of binding affinity . Although methodologies have been proposed for multi-target screening based on binding affinity [66] , there are simply no computational tools available for the efficient and accurate a priori estimation of the drug-target residence time from molecular structures . A detailed understanding of the effect of multiple interactions on the biological network requires innovative systems biology approaches . The qualitative description of the biological network presented here is limited in its predictive power , considering the highly dynamic nature of signal transduction pathways . A mathematical modeling approach will be more powerful than the static approach as we have demonstrated recently in a study of CETP inhibitors [67] . Existing mathematical modeling methods such as ordinary differential equations , Petri nets , and pi-calculus require a large number of kinetics parameters to simulate the dynamic behavior of the biological system [68] . In practice many of these parameters may not be available . Thus the network model has to be reduced . The qualitative properties derived from off-target binding network may help to develop restrained but functional dynamic models that are suitable for parameter optimization and mathematical modeling . In conclusion , by integrating methods from structural bioinformatics , molecular modeling and network analysis , we propose that the observed anti-cancer effects of the HIV protease inhibitor Nelfinavir derive from weak binding to multiple protein kinases that are mostly upstream of the PI3K/Akt pathway . Our computational approach , enhanced from previous work with the use of MD simulation and MM/GBSA free energy calculations , is supported by kinase activity assays and existing experimental and clinical evidence . This type of approach has the potential to be generalized as a form of rational polypharmacological drug design .
The structural proteome-wide off-target pipeline is outlined in Figure 1 . Firstly , the Nelfinavir binding pocket in the HIV protease ( PDB Id: 1OHR ) was used to search against 5 , 985 PDB structures of human proteins or homologous of human proteins using the SMAP software [21] , [22] , [23] . Secondly , the binding poses and affinities of Nelfinavir to these putative off-targets are estimated using two docking methods , Surflex [33] and eHiTs [34] . If the docking score indicates severe structural clashes between Nelfinavir and the predicted binding pocket , the protein is removed from the off-target list . Finally , the remaining putative off-targets are subject to MD simulation , MM/GBSA calculation , network reconstruction , and kinase activity assay . 5 , 985 PDB structures that are homologous to human proteins ( sequence identify >30% , alignment coverage larger than 90% ) are searched against the HIV-1 protease dimer ( PDB id: 1OHR ) using SMAP , which can be downloaded from http://funsite . sdsc . edu . The detailed algorithms implemented in SMAP are presented elsewhere [21] , [22] , [23] . In brief , proteins are represented using Cα atoms only and characterized by a geometric potential [21] . Then two proteins are aligned to identify similar local binding sites using the Sequence Order Independent Profile-Profile Alignment ( SOIPPA ) algorithm [22] . The statistical significance of the binding site similarity is estimated using an extreme value distribution model [23] . The binding affinity of Nelfinavir to the putative off-targets with SMAP p-value less than 1 . 0e-3 are estimated by two docking methods , Surflex [33] and eHiTs [34] . First , the complex structure of HIV-1 protease with Nelfinavir is superimposed onto these proteins according to the SMAP alignment . The superimposed structure of Nelfinavir is used as the starting conformation for docking . The binding pose of Nelfinavir in these statistically significant off-targets is locally optimized and scored starting from the starting conformation using Surflex 2 . 1 ( default setting ) and eHits 6 . 2 ( the fastest setting ) . The docking score is normalized using the protocol described in reference [13] . MM/GBSA [69] , [70] was developed for free energy calculations and has been used to estimate the binding affinity for several protein or DNA systems [71] , [72] , [73] , [74] . Here we perform ensemble average MM/GBSA binding free energy calculation on the snapshots from the MD simulation to compare binding affinity of Nelfinavir with that of the co-crystallized ligands . Explicit solvent molecular dynamics simulations were performed with NAMD [75] on the structures of the Nelfinavir-protein kinase complexes and co-crystallized ligand-protein kinase complexes . The starting structure for Nelfinavir in each protein kinase is the lowest energy conformation obtained through Autodock Vina [76] . These complex structures are embedded in rectangular boxes of TIP3P water [77] molecules to mimic the solvent environment . The smallest distances between the edge of the boxes and the atoms of the complex structures are adjusted to be at least 10 Å . Ions are added to neutralize these systems and satisfy the salt concentrations . The salt concentration is obtained from individual experimental condition for each protein kinase . The long distance cut-off for both van der Waals interactions and electrostatic interactions is set as 14 Å . A switching function is used to truncate the van der Waals energy smoothly at the cut-off distance . The Particle Mesh Ewald ( PME ) [78] method is applied to treat the long range electrostatic interactions . All covalent bonds involving hydrogen atoms are constrained by the SHAKE algorithm [79] . In order to simulate the NPT ensemble ( system with a fixed pressure P , temperature T , and number of atoms N ) , the Langevin piston Nose-Hoover method [80] , [81] in NAMD together with the periodic boundary conditions is used to maintain a constant pressure and temperature for these systems . Systems are first minimized by a five-stage minimization protocol . Hydrogen atoms are optimized in the first stage , keeping all other atoms fixed . Then water molecules and side chain atoms are relaxed in the second and third stage , respectively . All atoms are optimized in the fourth stage , with position restraints on backbone atoms of proteins and ligands . Minimization is completed by an additional 25 , 000 steps , without any restraints , to remove bad contacts . All minimizations are preformed with the conjugate gradient energy minimization method [82] in NAMD . The optimized systems are then gradually heated from 0 K to 50K , 100 K , 150 K , 200 K , 250 K , and experimental temperature ( about 298 K ) with position restraints on the backbone atoms . The structures are equilibrated at each temperature for 250 ps with a 1 . 0 fs time step . The force constant of restraints is 4 . 0 kcal/mol·Å-2 . After the systems are heated to the experimental temperature , position restraints are removed in the following 120 ps simulation by gradually reducing the force constant . Subsequently 8 ns NPT MD simulations are carried out on these systems with 1 . 0 fs time step at the experimental temperature . 200 snapshots are extracted from the last 2 ns simulations with 10 . 0 ps time intervals to generate representative configurations for the MM/GBSA binding free energy calculation . The binding free energy can be calculated through the following equation: ( 1 ) where Gcomplex , Greceptor , Gligand are the free energies of the complex , receptor and ligand respectively . The free energy of each molecular on the right hand side can be considered as the sum of molecular mechanical energy in gas phase , solvation energy and entropy term , as shown in the following formula: ( 2 ) EMM is calculated by the molecular mechanics method with standard force field parm9 in AMBER9 package [83] , [84] . The electrostatic contribution to the solvation free energy is determined by the Generalized Born ( GB ) model [85] , [86] , [87] , [88] , a widely used continuum solvent model . The “OBC” model with modified Bondi radii ( mbondi2 ) [89] , [90] , [91] in AMBER9 is applied to calculate this part of energy . This model is newer than the original version of the GB model and provides a significant improvement and is recommended for both proteins and nucleic acids . The interior dielectric constant of the molecule of interest is set as 1 . 0 and the exterior or solvent dielectric constant is set as 78 . 5 . The non-polar contribution to the solvation free energy is proportional to the solvent-accessible surface area [89] , [92] . The surface area is calculated by the LCPO model [93] and the surface tension used to calculate the non-polar part is taken as 0 . 0072 kcal/mol·Å-2 . The entropic term is the most time-intensive part of the MM/GBSA calculation but is found to be indistinguishable among different conformational states and contributes less than the other two terms in many application for estimating relative binding free energies [69] , [94] , [95] . The entropy change associated with ligand binding is estimated by normal mode analysis [96] in AMBER9 . For each system , the MM/GBSA calculation is carried out on the 200 snapshots extracted from the last 2ns of the MD simulation . HTRF® TranscreenerTM ADP Assays were performed on EGFR , ErbB2 and ErbB4 Akt1 , Akt2 and Akt3 by GenScript ( New Jersey , U . S . A ) . Nelfinavir Mesylate was purchased from Toronto Research Chemicals ( North York , Canada ) . The compound is diluted to a 10 mM concentration with acetone and stored at -20°C . Inhibition of Nelfinavir at 20 µM was tested on EGFR , ErbB-2 , ErbB-4 and Akt ( Akt1 , Akt2 , Akt3 ) . | The traditional approach to drug discovery of “one drug – one target – one disease” is insufficient , especially for complex diseases , like cancer . This inadequacy is partially addressed by accepting the notion of polypharmacology – one drug is likely to bind to multiple targets with varying affinity . However , to identify multiple targets for a drug is a complex and challenging task . We have developed a structural proteome-wide off-target determination pipeline by integrating computational methods for high-throughput ligand binding site comparison and binding free energy calculations to predict potential off-targets for known drugs . Here this method is applied to identify human off-targets for Nelfinavir , an antiretroviral drug with anti-cancer behavior . We propose inhibition by Nelfinavir of multiple protein kinase targets . We suggest that broad-spectrum low affinity binding by a drug or drugs to multiple targets may lead to a collective effect important in treating complex diseases such as cancer . The challenge is to understand enough about such processes so as to control them . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biotechnology",
"medicine",
"molecular",
"dynamics",
"drugs",
"and",
"devices",
"pharmacology",
"chemistry",
"biology",
"drug",
"discovery",
"systems",
"biology",
"biochemistry",
"drug",
"research",
"and",
"development",
"computational",
"chemistry",
"computational",
"bi... | 2011 | Drug Discovery Using Chemical Systems Biology: Weak Inhibition of Multiple Kinases May Contribute to the Anti-Cancer Effect of Nelfinavir |
Most natural odors have sparse molecular composition . This makes the principles of compressed sensing potentially relevant to the structure of the olfactory code . Yet , the largely feedforward organization of the olfactory system precludes reconstruction using standard compressed sensing algorithms . To resolve this problem , recent theoretical work has shown that signal reconstruction could take place as a result of a low dimensional dynamical system converging to one of its attractor states . However , the dynamical aspects of optimization slowed down odor recognition and were also found to be susceptible to noise . Here we describe a feedforward model of the olfactory system that achieves both strong compression and fast reconstruction that is also robust to noise . A key feature of the proposed model is a specific relationship between how odors are represented at the glomeruli stage , which corresponds to a compression , and the connections from glomeruli to third-order neurons ( neurons in the olfactory cortex of vertebrates or Kenyon cells in the mushroom body of insects ) , which in the model corresponds to reconstruction . We show that should this specific relationship hold true , the reconstruction will be both fast and robust to noise , and in particular to the false activation of glomeruli . The predicted connectivity rate from glomeruli to third-order neurons can be tested experimentally .
Although it is still debated how many different odorants humans can perceive , the most commonly cited number is on the order of 104 [1–3] , much greater than the 500 olfactory receptor neuron ( ORNs ) types . Many other species , including both vertebrates and insects , have the same order of magnitude of ORN types or even fewer ( around 1000 in mice , 50 in Drosophila ) . The order of magnitude difference between the number of odorants and ORN types implies that humans as well as other species rely on compressed representations , potentially following the principles of compressed sensing [4–7] . In the compressed sensing framework [4] , sparse high dimensional signals can be accurately reconstructed using a small number of measurements provided that the input signals are sparse . Natural odors are sparse in the sense that they are dominated by a few molecular components [8–10] . The relevance of compressed sensing algorithms to olfactory coding is reinforced by the anatomical organization of the olfactory system . High dimensional odor signals are compressed into a low-dimensional representation in terms of the activity of a relatively small number of glomeruli in the olfactory bulb , in the case of vertebrates , or the antennal lobe in the case of invertebrates . The standard compressed sensing algorithm performs signal reconstruction as a constrained ℓ1 minimization [4] . Such optimization can be solved through neural dynamics [5 , 6] , but the resulting reconstructions were considerably less fault tolerant than observed experimentally . For example , mice olfactory discrimination remains essentially intact when half of glomeruli are disabled [11] whereas theoretical reconstructions fail at this level of signal interference [5] . Furthermore , signal reconstruction based on dynamical optimization by construction requires more time for signal recognition compared to feedforward reconstruction schemes . Here we describe a feedforward reconstruction scheme based on compressed sensing ideas that is both fault tolerant and matches the main features of the organization of the olfactory system . The results demonstrate that a purely feedforward network is capable of robustly compressing/decompressing binary signal without dynamical optimization .
We begin by reviewing the main results from compressed sensing literature as they pertain to olfactory coding . The odor signal s0 can be described as a binary vector of length N where each element is either 1 or 0 depending upon whether a given molecular component is present or not in the odor . We refer to the number K of nonzero components in the odor as the odor sparsity . The main premise of compressed sensing is that a sparse signal s0 can be compressed into a vector x = As0 of length M < N and then recovered with high reconstruction quality provided K ≪ N . The encoding matrix A has dimensions M × N; its matrix elements can be chosen randomly . With this setup , the original signal s0 can be recovered exactly from the convex ℓ1 optimization problem [4] s ^ = arg min | | s | | 1 subject to x = A s 0 . ( 1 ) Although the ℓ1 minimization problem can be solved in polynomial time , it is not straightforward to implement such optimization algorithms in a neural circuit . One solution involves a two-layer neural network that perform similar ℓ1 minimization through neural dynamics [6] . However , this imposes certain requirements on the structure of recurrent connections in the second layer together with a static nonlinear activation function . Another alternative implementation relies on ℓ2 minimization instead of ℓ1 . In this case , the reconstruction is obtained simply as s ^ = ( A T A ) - 1 A T x where the −1 represents a pseudo-inverse relation . However , such an approach does not produce exact signal reconstruction [7] and would predict much larger errors than observed in olfactory experiments . We now propose a model for the olfactory system , which can compress and robustly recover sparse binary signal with high probability , without using any dynamical optimization . The solution is based on a nonlinear binary encoding model instead of the linear encoding model used in the conventional compressed sensing approach . Specifically , the compressed vector x has the form of a threshold function x i = H ( x i l - θ c ) where xl = As0 and H is the Heaviside step function with H ( 0 ) = 1 . We assume that the measurement matrix ( affinity matrix ) A is a M×N random binary matrix where each element is chosen independently to be either 1 or 0 with equal probability p and 1 − p , respectively . It is worth mentioning that while we use a random connectivity matrix in our model , we do not assume that this matrix differs across individuals . Rather , the randomness is meant to characterize how well the system works in the absence of specificity between odorants and glomeruli identity . By extending the definition of H to vectors , the measurement vector x can be compactly written as x = H ( A s 0 - θ c ) , ( 2 ) where θc = 1 , reflects that all measurements larger than 1 are set to 1 so that x is binary . This corresponds to a binary model of glomeruli activity described by the binary vector x . The threshold value of θc = 1 corresponds to a logical OR operation , so that glomerulus k will be activated if any of the odor components that are associated with inputs to this glomerulus are activated . To reconstruct the original signal , the glomeruli activity x are projected to another layer of neurons ( neurons in the olfactory cortex of vertebrates or Kenyon cells in the mushroom body of insects ) which has the same dimension as the original signal s0 . The activity of neurons in this layer is denoted by vector s ^ which has the same dimensionality N as the original signal s0 . The reconstructed signal can be computed as s ^ = H ( W T x - θ r ) , ( 3 ) where θr is the activation threshold for neurons in the reconstruction layer . The reconstruction matrix W equals the measurement matrix A normalized to 1 by column , i . e . Wki = Aki/∑k Aki . With this normalization , the reconstruction threshold θr = 1 corresponds to logical AND operation . That is , odor component i will be detected as present if all glomeruli that feed signals to node i in the reconstruction layer are activated . Below we will present most of the results for θr = 1 and then analyze how the reconstruction quality and recovery robustness depend on this threshold . We will also determine the optimal connectivity ratio from the compression to the reconstruction layer that maximizes the fidelity of reconstructions . Our feedforward model can be thought of as an information transmission channel that compresses , transmits , and decompresses a sparse binary signal . To find the optimal network configuration , we seek to maximize mutual information between the input and output of the channel as has been done to characterize performance in the visual and other sensory systems . The mutual information between s0 and s ^ is given by I ( s 0 , s ^ ) = ∑ s 0 ∑ s ^ P ( s ^ | s 0 ) P ( s 0 ) log 2 P ( s ^ | s 0 ) P ( s ^ ) . ( 4 ) For a given signal sparsity K , the conditional probability P ( s ^ | s 0 ) of the reconstructed signal s ^ given the original signal s0 can be computed as: P ( s ^ | s 0 ) = p false N e r r ( 1 - p false ) ( N - K - N e r r ) , ( 5 ) where p false ≡ P ( s ^ i = 1 | s i 0 = 0 ) is the probability of false detection for an odor component and N e r r = | | s ^ | | 0 - K is the number of false detection events for the odor s0 . We note that for θr = 1 , the probability to miss an odor component is zero provided this odor component activates at least one of the glomeruli . In this regime , the information is fully determined by the false detection rate pfalse , and as we show below decreases proportionally with pfalse . Assuming a uniform prior over individual odor components P ( s 0 ) = 1 / ( N K ) , one can also compute the probability distribution of reconstructed signals: P ( s ^ ) = ∑ s 0 P ( s ^ | s 0 ) P ( s 0 ) = ( K + N e r r K ) ( N K ) p false N e r r ( 1 − p false ) ( N − K − N e r r ) ( 6 ) Putting together Eqs ( 4 ) – ( 6 ) , the mutual information can be written as I ( s 0 , s ^ ) = log 2 ( N K ) − ∑ N e r r = 0 N − K ( N − K N e r r ) p false N e r r ( 1 − p false ) ( N − K − N e r r ) log 2 ( K + N e r r K ) . When ( N − K ) pfalse ≪ 1 , the summation above can be well approximated by its leading nonzero term ∑ N e r r = 0 N − K ( N − K N e r r ) p false N e r r ( 1 − p false ) ( N − K − N e r r ) log 2 ( K + N e r r K ) ≈ ( N − K ) p false log 2 ( K + 1 ) , ( 7 ) so that the expression for the mutual information becomes: I ( s 0 , s ^ ) ≈ log 2 ( N K ) − ( N − K ) p false log 2 ( K + 1 ) . ( 8 ) Thus , for given N and K , maximizing I ( s 0 , s ^ ) can be approximated by minimizing the probability of false detection pfalse .
The false detection rate that appears in Eq 8 can be computed as p false ≡ P ( s ^ i = 1 | s i 0 = 0 ) = ∑ k = 1 M P ( s ^ i = 1 | | | T i | | 0 = k ) P ( | | T i | | 0 = k | s i 0 = 0 ) = ∑ k = 1 M 1 - ( 1 - p ) K k ( M k ) p k ( 1 - p ) M - k 1 - ( 1 - p ) M = 1 1 - ( 1 - p ) M ∑ k = 0 M 1 - ( 1 - p ) K k ( M k ) p k ( 1 - p ) M - k - ( 1 - p ) M 1 - ( 1 - p ) M = 1 - p ( 1 - p ) K M - ( 1 - p ) M 1 - ( 1 - p ) M , ( 9 ) where Ti ≡ {xk ∈ x|Aki = 1} , and p is the average connectivity rate from the compression to the reconstruction layer . In the last line above we use the binomial expansion . Because we are interested in the regime where M is large , we have ( 1 − p ) M ≪ [1 − p ( 1 − p ) K]M ≪ 1 as long as p is not too small . Thus , Eq 9 can be approximated with great accuracy by the following simple equation: p false = 1 - p ( 1 - p ) K M . ( 10 ) As shown in the inset of Fig 1B , Eq 10 provides an accurate approximation when the connectivity p is not too sparse . Since our main interest is near the optimal connectivity rate ( see below ) where Eq 10 is very accurate , we will use Eq 10 unless specified . As expected , the false detection rate pfalse decreases as the number of glomeruli M increases and as the signal sparseness K decreases . Importantly , for a given M and K , there is an optimal p , which we refer to as pm , that minimizes pfalse , as shown in Fig 1B . Taking ∂pfalse/∂p = 0 leads to p m = 1 K + 1 . ( 11 ) It is worth noticing that the optimal connectivity pm is independent of the number of glomeruli M , and depends only on the signal sparseness K . Thus , optimal connectivity depends exclusively on the level of sparseness of signals in the environment and can be determined prior to any measurements on neural circuits . For an optimal connectivity p = pm , the probability of fault activation decreases exponentially as M increases and thus can be very small . This indicates that the proposed feedforward compression-reconstruction scheme from Fig 1A can achieve exact recovery with high probability . To test the reconstruction quality , we compute the signal-to-noise-ratio ( SNR ) of the recovered signal . Since all nonzero components in the original will be recovered , the only source of errors in the reconstructed signal are due to false detection rates . Therefore , we can define the SNR of recovered signal as SNR = | | s 0 | | 0 < | | s ^ | | 0 > - | | s 0 | | 0 = K ( N - K ) p false , ( 12 ) as shown in Fig 2A–2C , where < ⋅ > denotes the expectation value . We can see from Fig 2B that the SNR increases exponentially with M . For our case where K ≪ N , we can achieve a high SNR for a number of glomeruli M much smaller than the number of odor components N or , equivalently , the number of third-order neurons . A key characteristic of a compression algorithm is the compression ratio α ≡ M/N . In previous compressed sensing frameworks , the critical compression ratio αc above which the signal can be perfectly recovered was shown to only depend on the relative signal sparsity f ≡ K/N . As f → 0 , αc ( f ) ∼ −f log f [12] . To compute the critical compression ratio for our reconstruction algorithm , we note that from Eq 12 , log pfalse = log f − log ( 1 − f ) − log SNR . In the strong compression limit where f ≡ K/N is small , this yields log p false ≈ log f - log SNR . ( 13 ) On the other hand , for the optimal connectivity rate pm and large K , log pfalse can also be simplified using Eq 10 as follows: log p false = M log 1 - 1 K + 1 1 + 1 K - K ≈ M log 1 - 1 K + 1 e - 1 ≈ - M e K = - α SNR e f . ( 14 ) where αSNR is defined as the compression rate to achieve a certain SNR . Combining Eqs 13 and 14 , in the limit of strong compression where f → 0 , the critical compression ratio behaves as αSNR ∼ −f log f . We note that care should be taken when the SNR becomes comparable to or larger than N because 1/f = N/K ≤ N , so that log SNR cannot be neglected when f → 0 . The obtained critical compression rate can be compared to its theoretical limit . The latter corresponds to the minimal number of bits required to encode a sparse signal: M m i n = ⌈ log 2 ( N K ) ⌉ , ( 15 ) where ⌈x⌉ is the smallest integer not less than x . When N and K are large but f ≡ K/N is small , using Stirling’s approximation , we obtain that M m i n × log 2 ≈ N log N - K log K - ( N - K ) log ( N - K ) ≈ K log N - K log K + K = K - K log f , ( 16 ) This yields that the theoretically possible compression ratio αmin in the strong compression limit of f → 0 as α m i n → f log 2 e / f , ( 17 ) which also yields αmin ∼ −f log f as f → 0 . Notice that although both αSNR and αmin behave as −f log f for f → 0 , they have different proportionality coefficients . To be more specific , αSNR ∼ ef log 1/f while αmin ∼ ( log 2 ) −1 f log 1/f . As a result , αSNR/αmin → e log 2 ≈ 1 . 88 as f → 0 . Thus , the number of glomeruli needed in our model is about twice the theoretical limit but is achieved here with an extremely simple feedforward encoding model . As shown in Fig 2D , the number of required glomeruli increases sub-linearly with K , and logarithmically with SNR . In practice , with only a few times more glomeruli than the theoretical limit , a very high SNR can be achieved . Advances in experimental techniques provide opportunities to test our theory under the circumstances of extreme genetic manipulations . For example , following a genetic manipulation that caused most olfactory receptor neurons to express a single odorant receptor M71 , the M71 ligand acetophenone activates half of the glomeruli . Despite this drastic manipulation , mice can still readily detect other odors in the presence of acetophenone , while their discrimination performance is only moderately compromised [11] . This result is consistent with our model . Assume there are M glomeruli in our model and half of them are always turned on ( corrupted ) . Such a system is equivalent to a model with only M/2 glomeruli , since the anomalously activated glomeruli will not affect signal recovery . Thus , the odor signal can still be recovered , but the SNR is decreased , which is in agreement with the experimental result . As a comparison , in previous compressed sensing framework , one can only allow a small percentage of corrupted glomeruli even when M > N [4] . In another set of experimental studies , part of the glomeruli in mice are removed or disabled [13–15] . It is shown that the ability to discriminate odors and simple odor mixtures is not impaired even when most of the glomeruli are removed or disabled . This seemingly surprising finding is also consistent with our model . From previous results , one can see that decreasing M will only lead to larger noise in the recovered odor signal but not to a failure of the system if the activation threshold for neurons in the reconstruction layer can be properly adapted to the new M . Assume the mice need SNR > ν to discriminate odors . When K is small , the minimal M needed for discrimination is M l o w = log K N ν log [ 1 - p ( 1 - p ) K ] . ( 18 ) From experiment data , p ≈ 0 . 05 ( although this is a very rough estimation , see [11 , 16–18] ) . One can check that the equation above is insensitive to variations in K and Nν over a broad range . If we assume K < 10 ( as in the experiments ) and Nν is within the range of 104 ∼ 105 , then Mlow is roughly between 200 and 300 , or around 20% of the glomeruli , which is in good agreement with the data in those experiments . On the other hand , our model can tolerate negative gloleruli noise ( false negative ) by changing its recovery threshold θr . Although we use θr = 1 in our results for analytical solution , it is very likely that real biological systems would use a lower threshold θr . With θr < 1 , the SNR is somewhat lower , as shown in Fig 3 , yet the system is more robust to noise in the reconstruction stage since the activation of a third-order neuron doesn’t require all of its connected gloleruli to be active and it also leaves room for odor generalization and pattern completion [19] . Indeed , when the threshold at the reconstruction stage is less than 1 , the reconstruction can tolerate some incompleteness in the glomeruli activation patterns . Real biological systems likely have the ability to adaptively change the activation threshold in order to balance the needs of high quality reconstruction and pattern completion . Our model is shown to be very robust and fault tolerant , and this robustness is achieved with accuracy . As one can see , each glomerulus in the model only contains part of the information about the original signal . Because the measurement matrix A is random , no single glomerulus or cluster contains more or unique information , so any subset of the glomeruli could recover the original signal . The more glomeruli there are , the better recovery quality ( SNR ) can be achieved . Thus , removing or disabling part of the glomeruli will not change the system qualitatively , but will make the recovered signal more noisy , up to a point where noise becomes comparable to the true signal at which point the reconstruction fails . For a real biological system , it is reasonable to assume that the recovered signal has very high SNR , which also means high redundancy , as is observed experimentally .
From our analysis we observed that for a given level of signal sparseness K , there is an optimal connectivity rate pm that maximizes SNR as well as the mutual information . Assuming that the biological system is adapted to a given value of odor sparseness in its environmental niche , one can essentially make predictions on the connectivity rate of matrix A . This is followed by another prediction that the percentage of glomeruli activated by a single odorant should be close to the percentage of glomeruli that could activate a neuron in olfactory cortex or a Kenyon cell , and this number should be similar among species which operate in similar olfactory environments . The latter prediction should be easier to test , since the number of coexisting odorants in the environment is hard to measure . Fortunately , previous experiments have gathered sufficient data to test our prediction indirectly . It has been shown that in Drosophila , 9% of the glomeruli have a strong response to an odorant [20] , while the connectivity rate between glomeruli and Kenyon Cells is 6 . 5% [21] to 12 . 5% [22] . ( The latter number is obtained based on the average number of claws per Kenyon cell measured in [22] ) These estimates are consistent with model predictions . Furthermore , in the locust , a typical projection neuron responds to about half of the odorants [23] , while the connectivity rate between projection neurons and Kenyon Cell is also around 50% [24] , which is also consistent with our prediction . We can see that the connectivity rate is very different between species . Such differences can be unified in our model as the adaptation to different environmental niches . The locust has an anomalously high connectivity rate ( 50% ) , which in our model implies that its olfactory system is adapted to extreme odor sparseness tuned to odors with primarily a single component ( pm = 0 . 5 when K = 1 ) . Similarly , Drosophila is adapted to sense odors composed of a mixture of about 10 odor components , while mice are tuned to detect a mixture of about 20 mono-molecular odors . In general , our model predicts that species with sparse connectivity will behave better in environments with complex odor mixtures , while species with dense connectivity have better performance in detecting simple odor mixtures . In addition to the predictions above , further experimental evidence supports the structure of our model , in particular the approximate logical OR/AND operations associated with the compression/reconstruction stages , respectively . For example , it has been observed experimentally that Kenyon Cells in Drosophila receive convergent input from different glomeruli and require several inputs to be co-active to spike [25] . This is consistent with our threshold activation function which at the reconstruction stage uses a logical AND operation . Functionally , experiments have shown that locust Kenyon cells are individually much better than projection neurons from glomeruli at detecting a single odorant; Kenyon cells that respond to an odorant also often respond to odor mixtures containing it [26] . This observation agrees with our assumption that each Kenyon cell only responds to one odorant and it will respond when an odor mixture contains that odorant . Since the affinity matrix A is determined genetically , all the connections in our model are predetermined before birth . There is some debate about such stereotypy versus random connectivity , and a compressed sensing model of olfaction based on random connections from glomeruli to mushroom body has been proposed [27] . Yet , our model supports both stereotyped and non-stereotyped projection from glomeruli to the mushroom body/olfactory cortex because the model is invariant under the exchange of neurons within the same layer . In order to verify such predetermination , one needs to obtain a detailed connectivity map from glomeruli to the mushroom body/olfactory cortex for different individuals , which is experimentally very challenging . An indirect approach to verify the predetermined connectivity hypothesis could be through an examination of innate behaviors that should depend primarily on predetermined connections . If one could relate innate behaviors to projections between glomeruli and the mushroom body/olfactory cortex , it would then provide additional supporting evidence for the genetically predetermined structural connectivity of the feedforward model . The feedforward structure of our model is an effective approximation to the more complicated structure of biological olfactory system where recurrent and feedforward-feedback connections exist . For example , it has been observed that inhibitory interneurons modulate neuronal responses in the olfactory bulb [28 , 29] . In linear dynamic systems , such feedforward-feedback structure could be mathematically modeled as a pure feedforward system with different effective feedforward connectivity . Suppose that we add a layer of interneurons z in Fig 1 that is connected to the glomeruli layer x by feedforward-feedback connectivity B . Then the linear dynamics of the system are x ˙ = - x + A s 0 - B T z and z ˙ = - z + B x , where we assume B is feedforward excitatory and feedback inhibitory . The steady state solution is x = ( I + BT B ) −1 As0 , which is the same for a pure feedforward system , except that connectivity A is replaced by ( I + BT B ) −1 A . This analysis is not exact if the activation function is nonlinear . In general , the feedforward-feedback system in steady state with a nonlinear activation function does not have an equivalent feedforward system , but one can still write the linear perturbation when neurons receive only weak inputs , which allows a feedforward approximation . Such a feedforward approximation is supported by experimental observations that the representations of odor mixtures in mouse glomeruli can be explained well by the summation of the glomeruli responses to their components [30] . One advantange of the effective feedforward model is that it enables an adaptive affinity matrix even with pre-determined connectivity . In the feedforward-feedback architecture mentioned above , the effective affinity matrix is ( I + BT B ) −1 A , where A is the pre-determined affinity matrix encoded in the genes , while B could be a learned matrix adapted to the environment . From this perspective , the existence of interneurons in both insects and vertebrates [31 , 32] , as well as adult neurogenesis in the olfactory bulb of mammals [33] , could play the role of adjusting the effective affinity matrix for the purpose of adaptation . We compare the performance of our feedforward architecture with the often-used LASSO ℓ1 minimization algorithm [34] provided by the Python scikit-learn library min s ^ 1 2 M | | A s ^ - x | | 2 2 + β | | s ^ | | 1 , ( 19 ) where N = 1000 , M = 500 , β = 0 . 001 are used . Linear measurement x = As0 is used for LASSO . For each K , we conduct 100 experiments with different random measurement matrices and signals , and compute the average of the reconstruction errors | | s ^ - s 0 | | 1 as well as the number of iterations used in LASSO . We also compute the mean reconstruction error when only 5 iterations are used in LASSO as a comparison . The results are shown in Fig 4 . As shown in the figure , the feedforward architecture has a lower reconstruction error when the signal is very sparse , while LASSO has a lower reconstruction error than the feedforward architecture when K becomes larger . However , the number of iterations also increases as the signal becomes denser . If we restrict the number of iterations to 5 in the LASSO ( equivalent to setting a maximum response time ) , LASSO performs much worse when the signal is very sparse . But as K increases , it still has a lower reconstruction error than the feedforward architecture . One drawback of this feedforward architecture is that it may not be able to achieve both compression and high-quality reconstruction simultaneously when the signal is not sparse . Unlike the ℓ1 minimization method where the number of measurements required to reconstruct the signal will never exceed signal length N ( N/2 for binary signal ) [35 , 36] , the feedforward architecture may need more measurements than the signal length to accurately reconstruct the signal . This can be seen by restoring the term in Eq 13 that we have previously neglected assuming that f is small log p false = log f - log ( 1 - f ) - log SNR . ( 20 ) Combining this with Eq 14 that remains the same when f is not small , we obtain: α SNR = e f log SNR + e f log ( f - 1 - 1 ) , ( 21 ) which could be larger than 1 when f is not small . Thus , the feedforward computation may require number of measurements that are larger than the input dimensionality to achieve reliable reconstruction . From another perspective , we can compute the upper bound on the reconstruction SNR that can be achieved for a given compression level . From Eq 21 and αSNR < 1 we get log SNR < 1 e f - log ( f - 1 - 1 ) , ( 22 ) which only depends on signal sparsity . For example , if f = 0 . 1 , then SNR < 4 . 4 , and the reconstructed signal will not be accurate . Although our analysis above is based on a binary signal / measurement matrix / glomeruli activity and threshold activation function , our results can be extended to positive real-valued signal / measurement matrix / glomeruli activity and any monotonically increasing activation function . Consider the case where the signal s0 and the element of measurement matrix Aij could take any positive value rather than just 0 and 1 . Denoting xl = As0 , and letting the activation function g be any monotonically increasing function , the output at the glomerulus stage can be written as x i = g ( x i l ) . Now , signal reconstruction can proceed based on the evaluation of a minimum function ( rather than the logical AND function that was used in the case of binary inputs and binary measurement matrices ) . Indeed , when the ith component of the reconstructed signal s ^ i is computed as the smallest value {g−1 ( xj ) /Aji} across the set of its inputs ( i . e . where Aji ≠ 0 ) , then our analysis remains valid . The only modification is that now the distribution of the signal and the measurement matrix elements are both required to compute the noise magnitude . This procedure ensures that the recovered components are still recovered exactly , while corrupted components are still corrupted . As a practical aside , we note that the minimum function can be implemented by short-term synaptic plasticity , see S1 Text and S1 and S2 Figs . | Many olfactory systems are capable of accurately sensing a minimum of thousands of different odorants using as few as hundreds of different receptors . This compression raises the possibility that the mathematical properties of compressed sensing might be relevant to olfaction , similar to how these properties were found relevant to other sensory systems . In olfaction , previous applications of compressed sensing algorithms relied on the dynamics of neural circuits to reconstruct high dimensional signals . Such approaches are relatively temporally inefficient and sensitive to noise . To overcome these problems , we propose a purely feedforward compressed sensing model of the olfactory system where high dimensional signals can be recovered with a single feedforward layer of neural processing . The reconstructions are shown to be robust to noise , account for a number of experimental observations , and because of the feedforward structure are temporally efficient . Using the model , we make predictions that can be tested in future experiments with respect to optimal connectivity within the olfactory system . Our results indicate that feedforward neural architectures can provide an efficient way to implement compressed sensing in neural systems . | [
"Abstract",
"Introduction",
"Models",
"and",
"Methods",
"Results",
"Discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"brain",
"vertebrates",
"neuroscience",
"animals",
"animal",
"models",
"optimization",
"drosophila",
"melanogaster",
"model",
"organisms",
"mathematics",
"odorants",
"materials",
"science",
"drosophila",
"research",
... | 2016 | A Robust Feedforward Model of the Olfactory System |
Trypanosoma brucei belongs to a group of unicellular , flagellated parasites that are responsible for human African trypanosomiasis . An essential aspect of parasite pathogenicity is cytoskeleton remodelling , which occurs during the life cycle of the parasite and is accompanied by major changes in morphology and organelle positioning . The flagellum originates from the basal bodies and exits the cell body through the flagellar pocket ( FP ) but remains attached to the cell body via the flagellum attachment zone ( FAZ ) . The FP is an invagination of the pellicular membrane and is the sole site for endo- and exocytosis . The FAZ is a large complex of cytoskeletal proteins , plus an intracellular set of four specialised microtubules ( MtQ ) that elongate from the basal bodies to the anterior end of the cell . At the distal end of the FP , an essential , intracellular , cytoskeletal structure called the flagellar pocket collar ( FPC ) circumvents the flagellum . Overlapping the FPC is the hook complex ( HC ) ( a sub-structure of the previously named bilobe ) that is also essential and is thought to be involved in protein FP entry . BILBO1 is the only functionally characterised FPC protein and is necessary for FPC and FP biogenesis . Here , we used a combination of in vitro and in vivo approaches to identify and characterize a new BILBO1 partner protein—FPC4 . We demonstrate that FPC4 localises to the FPC , the HC , and possibly to a proximal portion of the MtQ . We found that the C-terminal domain of FPC4 interacts with the BILBO1 N-terminal domain , and we identified the key amino acids required for this interaction . Interestingly , the FPC4 N-terminal domain was found to bind microtubules . Over-expression studies highlight the role of FPC4 in its association with the FPC , HC and FPC segregation . Our data suggest a tripartite association between the FPC , the HC and the MtQ .
Cell polarisation requires precise positioning and connection of organelles during the cell cycle [1 , 2] . This applies to the unicellular , pathogenic parasite Trypanosoma brucei , the aetiological agent of human African trypanosomiasis [3] . In T . brucei , organelle positioning and segregation during the cell cycle show a high degree of coordination and control [4–7] . Essential for pathogenicity , cytoskeleton remodelling occurs also during the parasite life cycle with major morphological alterations and changes in organelle positioning [8–11] . The T . brucei characteristic fusiform shape is mainly maintained by a sub-pellicular microtubule-based cytoskeleton [12] . The flagellum extends from the basal body tethered to the kinetoplast ( the mitochondrial genome ) [13 , 14] , and exits the cell through the flagellar pocket ( FP ) , an invagination of the plasma membrane . It then runs along the length of the cell whilst remaining attached to the cell body via the flagellum attachment zone ( FAZ ) , a complex structure that mediates lateral attachment of the flagellum [12 , 15 , 16] . A set of four specialised microtubules ( the microtubule quartet MtQ ) nucleates at the basal bodies and extends around the FP , inserts into the sub-pellicular microtubules array and runs as part of the cytoplasmic portion of the FAZ up to the anterior end of the cell body [17–19] . MtQ polarity is considered to be opposite to the polarity of the sub-pellicular microtubules ( the latter having their + end towards the posterior end of the cell ) , but the same polarity as the axoneme microtubules that have their + end towards the distal tip of the flagellum [4] . The functional role of the MtQ is still unknown . The membrane of the FP is devoid of sub-pellicular microtubules ( MTs ) and is the sole site for endo- and exocytosis processes . As such , the FP is a key player in protein trafficking , cell signalling and immune evasion through the removal of surface-bound host immune factors [20–23] . The FP encircles the flagellum at the FP neck , the latter being maintained by the flagellar pocket collar ( FPC ) , a cytoskeletal structure situated at the exit point of the flagellum [24 , 25 , 19] . The Golgi apparatus is precisely positioned between the kinetoplast and the nucleus in procyclic form ( PCF—found in the tsetse fly ) and is considered to be physically connected to the FP or the neck region of the FP [19 , 26 , 27] . Overlapping the FPC is the bilobe , an essential structure thought to be involved in protein entry in the FP [28 , 29] . The bilobe is a cytoskeleton-associated structure that overlaps the FPC and contains the cell cycle progression regulators centrin2 , centrin4 , the Polo-like kinase TbPLK that is associated transiently [28 , 30–32] , and the tubulin co-factor TBCCD1 [33] , together with other uncharacterised proteins [34–38] . RNAi knockdown ( RNAi ) of any of the above-named proteins leads to defects in organelle duplication , segregation and cytokinesis , and disorganisation of the bilobe structure [28 , 30 , 33] . The bilobe is composed of two domains; the hook complex ( HC ) , and the centrin arm , which is adjacent to the stem of the HC . The MtQ threads between these two structures [29] . To date , only three HC-specific proteins have been characterised as essential for cell survival in PCF or bloodstream form ( BSF , found in the mammalian hosts ) . The tubulin co-factor TBCCD1 is associated with filament-based structures in the cytoskeleton and is essential in PCF [33] . The Leucine-Rich Repeat Protein LRRP1 is a Ran regulator with essential role in FAZ assembly , flagellum inheritance , and cell division and is essential in PCF [39 , 40] . The Membrane Occupation and Recognition Nexus protein 1 ( MORN1 ) may be involved in the endomembrane balance and in facilitating protein entry into the FP and has been shown to be essential in BSF [41 , 42] . LRRP1 and MORN1 co-localise at the HC and are the landmarks of the HC . However , at the “hook” region of the HC , both proteins are super-imposed on top of the FPC [43] . The FPC plays essential roles in the biogenesis and function of the FP and thus in the viability of PCF parasites [25] . However , the mechanisms behind FPC biogenesis and function remain elusive , mostly due to poor knowledge of its molecular composition . The FPC is also a complex structure , and in addition to its ring/horse-shoe shape and attachment to the flagellum , it is attached to the sub-pellicular microtubule cytoskeleton [19 , 24 , 25] . To date , BILBO1 is the only FPC component that has been functionally characterised , and importantly , it is required for FPC biogenesis in PCF [44] . RNAi of BILBO1 abolishes the biogenesis of a new FPC , FP and FAZ , and the newly formed flagellum is positioned at the extended posterior end of the cell and is detached from the cell body . This means that the FPC is clearly required for the biogenesis of numerous structures and functions in the cell . BILBO1 is a structural protein with four main functional domains . The N-terminal domain ( NTD ) folds into a ubiquitin-like fold and deletion or mutation of key residues within this domain affects cell viability , suggesting that it could be involved in the interaction with partner proteins . Two EF-hand domains follow the NTD , then a central coiled-coil ( CC ) and a C-terminal leucine zipper ( LZ ) . The LZ is necessary , but not sufficient , for FPC targeting . The CC allows the formation of antiparallel dimers that can extend into filaments by interdimer interaction between adjacent LZ . Deletion of the NTD or mutation of key residues in either the NTD or the EF-hand domains influence the shape of the polymers formed and affect cell viability , suggesting that BILBO1 forms a molecular frame on which other FPC proteins can interact with [25 , 45 , 46 , 44] . The FPC and the HC remain super-imposed during the trypanosome cell cycle and are in close proximity with the microtubule cytoskeleton and the MtQ [43] . Indeed tomographical data demonstrate that the MtQ actually traverses the FPC [19] , and the shank of the HC is adjacent to , if not part of , the MtQ [43] . Therefore , it is thus reasonable to imagine that one or more proteins can link the FPC and the HC to the MtQ . To understand more about FPC composition and function , we carried out a genomic yeast two-hybrid screen using BILBO1 as bait . Among several putative BILBO1 binding partners , we identified Tb927 . 8 . 6370 , which was given the annotation FPC4 for Flagellar Pocket Collar protein 4 . Using a series of biochemical and cellular biology approaches , we demonstrate that FPC4 is a bona fide BILBO1 partner and is a microtubule-binding protein that potentially plays a role in FPC segregation .
Yeast two-hybrid ( Y2H ) is a well-established technique for analysing and determining protein-protein interactions . Using BILBO1 as bait in a T . brucei 927 genomic Y2H screen ( Hybrigenics ) , we identified several putative partners . Among them was Tb927 . 8 . 6370 , which we named FPC4—for Flagellar Pocket Collar protein 4 . FPC4 is a 48 . 9 kDa protein consisting of 444 amino acids with a calculated pI of 10 . 61 . It is encoded by a kinetoplastid specific gene whose synteny is conserved in trypanosomes , but not in Leishmania species . The BILBO1-binding domain of FPC4 that was identified in the Y2H screen lies within the region of amino acids ( aa ) 357 to 444 ( from now on named BILBO1 binding domain—B1BD ) . FPC4 primary sequence analysis did not predict any functional domains beside a putative poly-proline motif between aa 28–40 potentially for SH3 binding , and a coiled-coil domain between aa 218–252 ( embnet . vital-it . ch/software/COILS_form . html [47] ) . The interaction between FPC4 and BILBO1 via the FPC4 B1BD was confirmed by yeast two-hybrid interaction tests using full-length BILBO1 and full-length FPC4 as well as truncations of both proteins ( Fig 1A ) . Notably , BILBO1-FPC4 interaction was abolished when either the N-terminal domain of BILBO1 ( NTD , aa 1–170 ) or the FPC4 B1BD domain ( FPC4-ΔB1BD , aa 1–356 ) were deleted . Further , the BILBO1 NTD alone interacts with FPC4 B1BD , thus demonstrating that the N-terminal domain of BILBO1 and the FPC4 B1BD are required and sufficient for BILBO1-FPC4 interaction . It was previously described that BILBO1 forms polymers in vitro [46] or in vivo in mammalian cells [44] . We simultaneously expressed BILBO1 and FPC4-GFP ( or several GFP tagged truncations of FPC4 ) in U-2 OS cells to further characterise the BILBO1-FPC4 interaction ( Fig 1B ) . Expression of BILBO1 alone forms polymers with globular structures at the extremities , as previously described [44] ( Fig 1Ba ) . FPC4-GFP or FPC4 minus the B1BD domain ( FPC4-ΔB1BD-GFP ) was observed in linear filaments ( Fig 1Bb , c ) , whilst FPC4-B1BD-GFP was cytoplasmic ( Fig 1Bd ) . Some nuclear labelling was also observed , most probably due to a weak bipartite nuclear localisation sequence between the amino acids ( aa ) 120–154 that was predicted by NLS mapper [48] . Since FPC4 never localises to the nucleus in trypanosomes , we did not further address the nuclear localisation in the U-2 OS cells ( see below ) . When co-expressed with BILBO1 , FPC4-GFP or GFP-tagged B1BD ( FPC4-B1BD-GFP ) co-localised with BILBO1 polymers , whilst FPC4 deleted of its B1BD ( FPC4-ΔB1BD-GFP ) was not recruited to the BILBO1 assembly ( Fig 1B e—g ) . Consequently , using two different in vivo heterologous systems ( yeast and U-2 OS cells ) , we were able to demonstrate that FPC4 is a BILBO1 partner protein and that the aa 357–444 domain of FPC4 is involved in its interaction with the BILBO1 NTD . To assess the stoichiometry of the FPC4-B1BD–BILBO1-NTD complex , we co-expressed maltose binding protein ( MBP ) -tagged FPC4-B1BD ( aa 357–440 ) and 6HisBILBO1-NTD’ ( aa 1–120 ) in E . coli , and the complex was purified by nickel affinity chromatography . After removing the MBP and the 6xHis tags , the purified complex was checked by size exclusion chromatography ( Superdex-200 16/60 ) , which resulted in two elution peaks ( Fig 2A ) . SDS-PAGE analysis ( inset ) indicated that the peak at the retention volume of 86 . 88 ml corresponded to the BILBO1-NTD’–FPC4-B1BD complex , and that the peak at the retention volume of 94 . 51 ml corresponded to the BILBO1-NTD’ only . Analysis by static light scattering ( SLS ) showed that the complex had a molecular mass of 24 . 9 ± 1 . 0 kDa , which corresponded to one molecule of BILB01-NTD' ( aa 1–120 , MW = 14 . 1 kDa ) and one molecule of FPC4-B1BD ( aa 357–440 ) ( MW = 10 . 0 kDa ) ( Fig 2B ) . It was reported previously that a series of aromatic and hydrophobic residues forms a crater-like structure on the solvent-exposed face of the conserved surface patch of BILBO1-NTD . Previously described mutations of the residues that form the “rim” ( F12A/K15A/K60A/K62A , previously reported as mut1 ) or the “bottom” of the crater-like structure ( W71A/Y87A/F89A , previously reported as mut2 ) impaired BILBO1 function , demonstrating that they are essential for FPC function [45] . To further characterise the interaction between BILBO1-NTD and FPC4-B1BD , we generated two novel sets of mutations where the “rim” residues Lys-60 and Lys-62 ( being half of the rim mutation mut1 and named Edge-Mut ) and the “bottom” residues Trp-87 and Phe-89 ( named Centre-Mut ) were substituted with alanine . After expression in bacteria and purification , we assessed by isothermal titration calorimetry ( ITC ) the interaction of the mutant proteins with the FPC4-B1BD ( Fig 2C ) . Wild-type ( WT ) BILBO1-NTD bound strongly to the FPC4-B1BD ( dissociation constant Kd ≈ 5μM ) . Edge-Mut partially reduced the binding affinity ( Kd ≈ 18μM ) , whilst Centre-Mut completely abolished the interaction . Additionally , FPC4-B1BD and BILBO1 Centre-Mut or Edge-Mut were co-expressed and immuno-labelled in U-2 OS cells ( Fig 2D ) . Interestingly , the polymers formed by BILBO1-Edge-Mut and BILBO1-Centre-Mut were slightly different to non-mutated BILBO1 alone , because no annular termini were observed as described in Fig 1B and in [44] . This suggests that mutation of the residues on the BILBO1-NTD involved in FPC4 binding modifies the polymer shape formed by BILBO1 . Similar polymer shape modification was previously shown in trypanosome when mut1 and mut2 mutants were expressed [45] . When co-expressed with BILBO1-Centre-Mut , FPC4-B1BD was not associated with the BILBO1 polymers , but rather gave a cytosolic pattern ( Fig 2D ) . Results of co-expression of FPC4-B1BD and BILBO1-Edge-Mut are consistent with the ITC data ( Fig 2D ) . Using western-blotting we next assessed the expression of the Ty1-tagged Centre-Mut and Edge-Mut BILBO1 in trypanosomes upon induction with tetracycline ( Fig 2E , upper panel ) , and followed cell growth ( Fig 2E , lower panel ) . Induction of the expression of BILBO1 slightly reduced cell growth as previously described [44 , 45] . However , expression of Centre-Mut was lethal after 2 days of induction , whereas Edge-Mut did not affect cell growth . These in vitro ( Chromatography , SLS , ITC ) and in vivo data ( expression of mutated BILBO1 in U-2 OS cells and in trypanosomes ) demonstrate that residues Trp-87 and Phe-89 are critical for the interaction between BILBO1-NTD and FPC4-B1BD . Taken together , these data demonstrate that the N-terminal domain of BILBO1 and the FPC4-B1BD can form a complex with a 1:1 stoichiometry , which is required and sufficient for the BILBO1-FPC4 interaction . To localise FPC4 in the trypanosome , we produced a rat antibody raised against recombinant full-length FPC4 ( anti-FPC4 ) . We tested the specificity of anti-FPC4 by probing T . brucei whole cell extracts and extracts of bacteria expressing recombinant 6HisFPC4 or purified histidine-tagged FPC4 aa 1–260 and FPC4 aa 357–440 by western-blot ( WB ) ( S1A , S1B and S1C Fig ) . In these blots , the anti-FPC4 polyclonal antibody recognizes the full-length and the truncated FPC4 ( aa 1–260 ) purified proteins but not the FPC4 aa 357–440 . The specificity of the anti-FPC4 antibody was also tested by immunofluorescence on FPC4 RNAi induced cells ( S2 Fig ) . Unfortunately , neither the anti-FPC4 nor the anti-myc antibodies were able to detect the endogenous FPC4 ( S1B Fig ) or endogenously myc-tagged FPC4 ( S1D Fig ) respectively from trypanosome cell extracts . However , the proteins could be detected when over-expressed in trypanosomes . Considering that FPC4 and BILBO1 interact in vitro and in the U-2 OS cells , our co-labelling immunofluorescence experiments using anti-BILBO1 and anti-FPC4 on T . brucei cytoskeletons ( CK ) revealed , surprisingly , merely a close localisation of the two proteins ( Fig 3Aa ) . This was also observed using anti-myc in the cell line expressing endogenous C-terminal myc-tagged FPC4 ( Fig 3Ab ) . Clear co-localisation of FPC4 and BILBO1 was however observed in a tetracycline-inducible cell line that was over-expressing N-terminal myc-tagged FPC4 ( Fig 3Ac ) . The structure highlighted by the anti-FPC4 and the anti-myc labelling resembled the shank and the hook of the HC; indeed , FPC4 and MORN1 co-localised as seen in Fig 3Ad and 3D a-d , suggesting that FPC4 is a component of the HC . Unfortunately , endogenous FPC4 or myc-FPC4 were not detectable by immuno-gold electron microscopy ( iEM ) . However , iEM on flagella , derived from PCF cells over-expressing myc-FPC4 , showed that FPC4 localised on the HC ( Fig 3B ) . Immunofluorescence on CK from BSF using the anti-FPC4 showed a clear labelling proximal and distal to the FPC ( Fig 3C ) suggesting a better accessibility for the antibody or higher expression of FPC4 in BSF . Unfortunately , as WB does not detect the endogenous protein , we could not compare protein expression levels between BSF and PCF that perhaps could explain the more intense IF labelling in BSF . In PCF , endogenous FPC4 was detectable by immunofluorescence at every stage of the cell cycle and was present on the old and the new FPC and HC , and always associated with MORN1 and BILBO1 ( Fig 3D a-d ) and ( Fig 3D e-h ) . During the later stages of this FPC4 project , a novel 10xTY1 endogenous tagging plasmid was developed by the Gull laboratory ( University of Oxford , England , U . K ) . This plasmid was generously donated to our laboratory . We used the 10xTY1 plasmid to endogenously tag FPC4 in PCF and from the subsequent clonal cell lines we obtained excellent anti-TY1 immuno-fluorescence signal . Further , we could detect 10xTY1-FPC4 by western blotting ( S1E Fig ) demonstrating a better detection using the 10xTY1 tag over the myc-tag or over the anti-FPC4 polyclonal antibody . This increased detection permitted us to also obtain immuno-gold labelling ( see below ) . The clearer and better 10xTY1 signal , compared to other tags , confirms the FPC4 localisation overexpression data ( see below ) . Our endogenously N-terminal 10xTY1 tagged FPC4 ( 10xTY1-FPC4 ) permitted labelling of FPC4 , whilst also probing cells with anti-BILBO1 and anti-MORN1 . This was done to further analyse FPC4 localisation using wide-field and STimulated Emission Depletion ( STED ) confocal microscopy . Wide-field immunofluorescence analysis confirmed that 10xTY1-FPC4 co-localises on the shank and hook of the HC with MORN1 , but it also co-localises with BILBO1 on the FPC ( Fig 4A ) . Immuno-electron microscopy data indicated that 10xTY1-FPC4 localised above BILBO1 , and co-localises with MORN1 ( Fig 4B ) . Results of STED analysis confirmed this data and further described FPC4 as a series of substructures regularly positioned close to the MORN1-labelled shank and hook of the HC . Moreover , MORN1 and FPC4 labelling showed close localisation on the hook region of the HC where it overlaps with , and is on top of , BILBO1 ( Fig 4C ) . Three-dimensional reconstructions of all three immunolabelling signals show the close proximity of the three proteins ( Figs 4C and 3D snapshots ) . Normalised intensity profile plots also highlight the proximity of these proteins , and show that FPC4 is sandwiched between MORN1 and BILBO1 ( Fig 4D ) . The partial superposition of the peaks confirmed the interplay between the HC and the FPC . Interestingly , under the conditions used for STED , BILBO1 also appeared to be on the MtQ between the FPC and the basal bodies ( Fig 4C ) . Taken together , these data indicate that FPC4 is a cytoskeletal protein expressed in PCF and BSF parasites . FPC4 localises mainly at the HC together with MORN1 , but is also present close to BILBO1 at the FPC , which suggests that the HC and the FPC share and overlap with FPC4 . High-resolution STED analysis also showed that the FPC4-labelled region of the HC overlaps the FPC , and FPC4 is sandwiched at discrete points between BILBO1 and MORN1 suggesting that FPC4 is in simultaneous contact with MORN1 and BILBO1 and may be able to connect or link the FPC to the HC . In PCF , BILBO1 RNAi knockdown induces new flagellum detachment from the cell body but it remains anchored to the cell through its basal body . BILBO1 knockdown also induces new flagellum relocation to the posterior end of the cell as well as absence of new FP and FPC . Moreover , the FAZ structure is affected because no new FAZ is associated with the new flagellum [25] . Immunofluorescence assays show that FPC4 localisation is also affected in BILBO1RNAi cells ( Fig 3E ) . In cells displaying the typical BILBO1RNAi phenotype , anti-FPC4 antibody labelled the old FPC ( where the old FPC remains present or intact ) . In these cells , FPC4 was detected in 70% of the new-detached flagella ( n = 205 ) . The flagella , however , did not have an FPC because induction of BILBO1 RNAi prevents new FPC formation [25] . In 82% ( n = 72 ) of these new flagella , FPC4 labelling localised to the base of the axoneme . Importantly , in 18% of the new flagella that were labelled , the FPC4 labelling could be observed at foci along the length of the flagellum , suggesting that FPC4 was able to traffic into the flagellum in absence of the FPC . A similar labelling was observed for MORN1 in 51% of the new-detached flagella ( n = 45 ) . In these flagella , 78% had a signal at the base of the axoneme , and 22% were labelled within foci on the flagellum ( n = 23 ) . These data demonstrate that the correct localisation of FPC4 and MORN1 depends directly or indirectly on the presence of BILBO1 . One possible hypothesis to explain this could be that in the absence of the FPC or MtQ in BILBO1RNAi cells , FPC4 enters the flagellum and binds to axoneme MT . Furthermore , FPC4 and MORN1 both co-localise in the new detached flagellum suggesting that they may interact and are able to , perhaps , sequester each other . To further characterise FPC4 , we generated a PCF tetracycline-inducible , FPC4 RNAi cell line . RNAi induction decreased FPC4 mRNA levels , but total depletion was not achieved even after 10 days of induction ( S2A Fig ) . RNAi knockdown of FPC4 in PCF had no impact on cell growth ( S2B Fig ) or cellular morphology , even though the protein was not detected by immunofluorescence after 48h of induction ( S2C Fig ) . This suggests that even small amounts of protein might be sufficient for cell survival . Alternatively , other HC proteins with no predicted function , such as Tb927 . 4 . 3120 , might compensate for FPC4 RNAi knockdown [34] . However , knockout attempts were unsuccessful in PCF and BSF , suggesting that FPC4 might be essential . We generated PCF tetracycline-inducible cell lines that overexpress the N-terminal GFP tagged FPC4 or myc-tagged form of FPC4 . We also made tetracycline-inducible cell lines that express an FPC4 mutant that is deleted of its B1BD ( myc-FPC4-ΔB1BD ) and a mutant expressing only the FPC4 B1BD ( myc-FPC4-B1BD ) ( Fig 5A ) . The GFP tag did not affect FPC4 localisation because GFP-FPC4 was directed to the FPC/HC upon short time of induction ( Fig 5A ) . Immunofluorescence labelling showed that myc-FPC4 , GFP-FPC4 and myc-FPC4-ΔB1BD proteins also localised to the FPC/HC in cytoskeletons . However , myc-FPC4-B1BD was not detectable on cytoskeletons and instead was cytosolic . These data indicated that the FPC4-B1BD is neither sufficient nor required to target FPC4 to the FPC and that the domain consisting of aa 1–356 ( FPC4-ΔB1BD ) contains the FPC targeting sequence . Interestingly , myc-FPC4-ΔB1BD seemed to localise mostly on the shank of the HC compared to myc-FPC4 . Indeed , a clear hook and shank localisation could be observed in 75% of the cells expressing myc-FPC4 ( n = 50 ) , whilst the labelling of the hook section could be observed in only 17% of the cells expressing myc-FPC4-ΔB1BD ( n = 67 ) . Also , the myc labelling on the shank was longer in cells expressing myc-FPC4-ΔB1BD ( 3 . 32 ± 0 . 15 μm ) compared to cells expressing myc-FPC4 ( 1 , 87 ± 0 . 07 μm ) ( n at least 200 cells ) . In vivo , over-expression of a protein , or a mutant polypeptide that disrupts the activity of the wild-type protein , can induce dominant-negative phenotypes . Indeed , over-expression of several BILBO1 mutant proteins induced BILBO1RNAi-like phenotypes [44] . Because FPC4 RNAi knockdown did not provide information about FPC4 function , we decided to analyse the localisation of two N-ter-tagged-FPC4 ( GFP and myc ) , and to analyse any possible phenotypes induced by long-term expression . GFP or myc tags were both used to confirm that the cell expressing either of these tagged proteins did not show modified targeting to the FPC . Indeed , neither tag affected FPC4 localisation ( Fig 5A ) . However , expression of GFP tagged FPC4 induced cell death 3 days post induction ( Fig 5B ) . The induction of the expression of myc-FPC4 or myc-FPC4-B1BD did not affect cell growth , whilst growth rate was reduced during expression of myc-FPC4-ΔB1BD ( Fig 5B ) . Because of these differences in growth rates , and the fact that FPC4 could only be detected by western blot when tagged with 10TY or overexpressed , we measured the expression levels and integrity of the recombinant proteins by western blot using rat anti-FPC4 antibody . Quantification of expression levels showed that myc-FPC4-ΔB1BD and GFP-FPC4 were expressed 4x fold more than myc-FPC4 ( Fig 5C ) . We acknowledge that the phenotypes observed ( slower growth for myc-FPC4-ΔB1BD and lethality for GFP-FPC4 ) might be due either to the dominant negative effect of the over-expression of the proteins per se , or by the absence of the B1BD , or in the latter case , the presence of the GFP tag . It was , however , informative to study in more detail the phenotypes induced by the over-expression of GFP-FPC4 and myc-FPC4-ΔB1BD . Long induction ( 72h ) of GFP-FPC4 induced the formation of a long GFP-positive filament , within cells , that connected two FPCs together ( Fig 6A ) . Electron microscopy of thin sections of such cells confirmed the presence of an unusual electron dense fibre associated with the FPC ( Fig 6B ) . This filament was positively decorated when probed with anti-BILBO1 by immunofluorescence ( Fig 6A ) . Over-expression of myc-FPC4-ΔB1BD produced cells where 7 . 5% ( ±0 . 43 ) of the new kinetoplasts were positioned between the two nuclei ( NKKN configuration ) instead of a proximal position relative to the new nucleus ( KNKN ) , without affecting cell body length . FAZ labelling appeared normal in these cells ( Fig 5C ) . A filament was also observed between both old and new FPC in these cells ( Fig 6C ) . Electron micrographs of thin sections from the GFP-FPC4 induced cell lines illustrated the presence of unidentified material in the FP suggesting an alteration in the structure , function or integrity of the structure ( Fig 6B ) . Contrary to the GFP-FPC4 cell line , over-expression of myc-FPC4-ΔB1BD did not induce these phenotypes or cell death ( Fig 6C ) . However , phenotypic similarities were observed; for example , ( a ) a fibre connecting the two FPCs , and ( b ) formation of epimastigote-like cells . We occasionally observed the localisation of myc-FPC4-ΔB1BD together with MORN1 at the distal anterior end of the cell body and adjacent to the flagellum ( Fig 6C , asterisk in d ) . Further , a connection between two FPCs was clearly observed by electron microscopy on isolated flagella ( Fig 6D ) . Double immuno-gold labelling of isolated flagella from myc-FPC4-ΔB1BD cells illustrates that both FPCs are decorated with anti-myc and anti-BILBO1 gold , but the connecting fibre was only decorated with anti-myc . Examples of a myc-FPC4-ΔB1BD expressing cell with a mispositioned kinetoplast are shown by IF and EM ( Fig 6Cb and 6E ) . This mis-positioning was observed in 1K1N , 2K1N and 2K2N cells with different configurations . The different categories of misplaced kinetoplast containing cells represented 30% of the total cell population after 48h of induction ( n = 3 with at least 400 cells ) . These categories were epimastigote-like ( 37% ) , 1N1K1N ( 21% ) , 1N2K1N ( 24% ) , and 1K1N1K , 1K2N , 1K2N1K , KKNN grouped as others ( 17% ) ( Fig 6F ) . In WT 2K2N cells used in this work , the average distance between 2 kinetoplasts is 4 . 5 μm ( ± 0 . 2 ) . This distance was reduced in myc-FPC4-ΔB1BD to 3 . 2 μm ( ± 0 . 3 ) ( n = 3 , at least fifty 2K2N cells ) , with only a very slight cell body length reduction , if any ( cell body length was 20 . 2 μm ( ± 0 . 3 ) in WT , and 19 . 3 μm ( ± 0 . 5 ) in myc-FPC4-ΔB1BD ) . When expressed in U-2 OS cells , FPC4-GFP localised on filamentous structures resembling microtubules ( MTs ) ( Fig 1B ) . This MT localisation was confirmed by co-immunolabelling FPC4-GFP with acetylated tubulin ( Fig 7Aa ) , suggesting that FPC4 is a MT-binding protein . Also , GFP recombinant FPC4-ΔB1BD and FPC-1-217 proteins co-localised with MTs , whilst FPC4-B1BD was cytoplasmic , showing that the B1BD is not involved in MT binding ( Fig 7Ab , c , d ) . Because the basic pI of FPC4 might induce artifactual MT binding [49] , and that FPC4-1-217 domain is basic ( pI 11 ) , its primary sequence was shuffled ( the amino acid composition remains the same , but their sequence within the protein is different , S3 Fig ) . Shuffled-FPC4-1-217 was not able to co-localise with MT and was cytoplasmic , demonstrating the MT-binding specificity of FPC4 ( Fig 7Ae ) . To test FPC4 microtubule binding in vitro , we expressed and purified the B1BD domain of FPC4 ( 6HisFPC4-357-440 ) and FPC4-1-2606His from E . coli . Both recombinant proteins were soluble and could be used in a MT binding assay ( Fig 7B ) . Increasing concentration of FPC4-1-2606His ( 0 . 33 μM to 25 . 52 μM ) and of 6HisFPC4-357-440 ( 0 . 96 μM to 56 . 27 μM ) were mixed with fixed concentration of polymerised tubulin ( 7 . 2 μM ) . After centrifugation ( 16 , 100 g at 22°C ) , supernatants and pellets were loaded on SDS-PAGE and the proteins stained with Instant Blue . The presence of the protein of interest in the pellets , bound to tubulin , demonstrates the MT-binding property of the protein . This was the case for FPC4-1-2606His , which was found in the pellets up to a saturating concentration ( 5 . 6 μM ) ; above this concentration , the protein was also observed in the supernatant . On the contrary , 6HisFPC4-357-440 was found in the supernatants and thus did not bind to the MTs ( Fig 7B ) . FPC4 is a MT-binding protein in vitro and in U-2 OS cells but does it localise to MT containing structures in trypanosomes ? In PCF T . brucei isolated flagella , the MtQ is partially retained and can be decorated with anti-α-tubulin where it is observed as a short structure ( named here the MtQ complex ) close to the BBs ( Fig 8A ) . As expected from the localisation in CK , myc-FPC4 and myc-FPC4-ΔB1BD localised to the proximal end of the isolated flagellum close to the BBs . Moreover , the remnant MtQ complex was partially decorated with myc-FPC4 and fully decorated with myc-FPC4-ΔB1BD ( Fig 8A ) . The co-localisation on the MtQ complex of myc-FPC4-ΔB1BD and tubulin was confirmed by immuno-gold electron microscopy on isolated flagella ( Fig 8B ) . The axoneme and the MtQ were decorated with anti-α-tubulin . The FPC and the MtQ , from the proximal to the distal end , were decorated with anti-myc , but the axoneme was negative for myc label . Considering that FPC4 is a MT-binding protein ( in vitro and in the U-2 OS cells ) , our data imply an interaction between FPC4 and the MtQ , especially since it co-localises with MORN1 and MORN1 localises to the MtQ within the shank of the HC [43] . Taken together , our data demonstrate that FPC4 is a BILBO1 binding protein that localises at the FPC and the HC . Moreover , FPC4 is a MT-binding protein that may associate with the MtQ . This implies that FPC4 has multiple binding partners and is a protein involved in linking the FPC , HC and possibly the MtQ .
The flagellar pocket collar ( FPC ) and the hook complex ( HC ) are two essential T . brucei cytoskeletal structures whose molecular components are inadequately defined [29 , 50] . Furthermore , the molecular processes involved in their segregation during the cell cycle are poorly understood . Although recent data identified the Polo-like kinase TbPLK as being involved in HC duplication , BB segregation , and flagellum attachment [51 , 5 , 32 , 52] , BILBO1 remains the only FPC protein to be identified and characterised . As stated above , BILBO1 is the only FPC protein to be described , whereas MORN1 , TBCCD1 and LRRP1 are the HC proteins that have been characterised in detail [25 , 33 , 39 , 41 , 42] . Following up on results showing that BILBO1 is a structural protein in T . brucei [46 , 44] , we hypothesize that it is unlikely that BILBO1 is the only protein required for the biogenesis and function of the FPC . For this reason , we screened for BILBO1 interacting partners using a yeast two-hybrid genomic library and FPC4 was identified . In this study , we demonstrate that FPC4 is a MT-binding protein that localises together with MORN1 at the HC and is sandwiched between the HC and the FPC . The interface between the HC and the FPC has FPC4 labelling and logically this is where FPC4 interacts with BILBO1 . Also , FPC4 co-localises with MORN1 , which has been shown by immuno-gold labelling to be adjacent to and on the MtQ [43] . Our data imply that FPC4 also localises close or onto the MtQ and provide the first evidence of a possible molecular link between these structures . The precise role of the FPC4 MT binding is still unclear , but one possible hypothesis is that FPC4 connects components of the FPC and of the HC to the MtQ . FPC4 could have MT stabilization function ( as suggested by the presence of FPC4 on acetylated MTs in U-2 OS cells ) . In vitro and in vivo analysis demonstrated that FPC4 interacts with the BILBO1 NTD via its C-terminal domain ( B1BD ) in a 1:1 ratio . BILBO1 NTD is neither required for targeting to the FPC nor for polymerization [45 , 44] . However , it folds into an ubiquitin-like structure most likely playing a role in protein interaction [45] . Here , we have tested two new sets of mutations Centre-Mut and Edge-Mut that both localise at the FPC similar to mut1 and mut2 ( 45 ) , and demonstrated the critical role of Y87 and F89 residues ( Centre-Mut ) in the interaction with the FPC4-B1BD . Centre-Mut over-expression in T . brucei leads to rapid cell death , and in vitro , Centre-Mut completely abolishes the interaction with FPC4 . This demonstrates a direct impact on the interaction of these two residues , which can therefore be used in a drug design or drug-screening project . The different effects in the two mutants , Centre-Mut and Edge-Mut , might attribute to their specific locations at the FPC4-binding site and different contributions to FPC4 binding . The Centre-Mut is located at the centre of the binding pocket and consists of two aromatic residues ( Y87 & F89 ) , which likely provide a strong hydrophobic interaction between the two proteins . The Edge-Mut is at the edge of the pocket and contains two charged residues ( K60 & K62 ) , which may provide only a peripheral contact mediated by a much weaker electrostatic interaction . BILBO1 NTD is not required for FPC4 targeting , and overexpression of the BILBO1 Centre-Mut mutant , which cannot interact with FPC4 , is lethal . This suggests that the mutation may prevent proper binding of other partner proteins and , in consequence , is affecting the function of the FPC . Reduction of FPC4 expression by RNAi knockdown did not affect cell growth or cell morphology similar to the result reported in the genome-wide RNA interference target sequencing analysis by Alsford and colleagues [53] . However , knockout attempts were unsuccessful in PCF and BSF , suggesting that FPC4 might be essential . Further , over-expressing FPC4 ( GFP-FPC4 or FPC4-ΔB1BD ) led to the formation of a single fibre that formed between the old and the new FPC and generated epimastigote-like cells , presumably because this fibre impinges on correct FPC segregation . Several studies have shown that limiting the kinetoplast segregation ( or the FPC segregation ) leads to misplaced kinetoplasts and epimastigote-like cells [54–56] . It was previously reported that transition to epimastigote-like cells can be the result of the knockdown of FAZ proteins [15 , 56] . Indeed , FPC4 has been recently identified by proximity-dependent biotin identification ( BioID ) as a TbSAS-4 near neighbour [37] . TbSAS-4 is concentrated at the distal tip of the FAZ filament and is involved in regulating FAZ length . FPC4 and TbSAS-4 might , therefore , come in close proximity when the new FAZ elongation is initiated . However , FAZ was not obviously affected in the FPC4 over-expressing cell lines , as its immunolabelling appeared normal . The epimastigote-like cells most probably come from a downstream effect of the delayed segregation of FPC—HC due to the connecting fibre between the two FPCs . Interestingly , this fibre contains not only FPC4 but also MORN1 ( S4 Fig ) , and BILBO1 ( in the case of expression of GFP-FPC4 , but not in the case of the expression FPC4 deleted of its B1BD ) , suggesting that FPC4 can sequester BILBO1 under these conditions and also supports the data from the FPC4-BILBO1 binding assays . Because FPC4 binds MT , and the MtQ is not apparent when the fibre is formed , we also questioned whether this fibre is microtubule-based . The fibre has an average width of 56 nm ( n = 33 ) and could , theoretically , accommodate one or two microtubules ( 25 nm in diameter ) . As FPC4 is present on or close to the MtQ , this fibre could be an extension ( or a modification ) of a subset of the MtQ microtubules . However , in our hands , myc and α-tubulin co-immuno-labelling and visualisation of the fibre using electron microscopy was not conclusive , perhaps due to limited access of anti-tubulin antibodies to the fibre . Alternatively , FPC4 or MORN1 could form polymers in the absence of microtubules as previously suggested for MORN1 that was immuno-gold-labelled on the tendril , an uncharacterised filament , devoid of tubulin , that seems to run parallel to the MtQ between the BBs and the FPC [43] . Interestingly , RNAi knockdown of Spef1 , a MtQ-binding protein that localises between the BB and the FPC , leads to the inhibition of the assembly of a new MtQ . Spef1 knockdown also prevented FPC and HC segregation , and resulted in unsegregated and misplaced kinetoplasts [55] . Thus , over-expressing the microtubule-binding domain of FPC4 ( GFP-FPC4 or FPC4-ΔB1BD ) could stabilise the MtQ and prevent FPC separation . The potential association between FPC4 and the MtQ is restricted to a specific region of the MtQ ( the FPC and along the shank of the HC ) . Considering that FPC4 binds to acetylated tubulin in U-2 OS cells ( Fig 7 ) , it could imply that the MtQ tubulin is also post-translationally modified . Unfortunately , MtQ-specific modifications have not been identified so far , although IFA images of flagella probed with anti-acetylated tubulin antibody suggest that the MtQ is acetylated [57] . The restricted MtQ localisation of FPC4 could arise also via the interaction with BILBO1 and MORN1 and other proteins that would retain FPC4 to slide along the MtQ . This is supported by the fact that deletion of the BILBO1-binding domain allows FPC4 to localise to the anterior end of the cell body as observed in Fig 5 , suggesting that the protein was able to be associated with the MtQ when released from the interaction with BILBO1 . FPC4 might bind or sequester MORN1 as suggested by the presence of MORN1 on the fibre connecting the two FPCs and at the anterior end of the cell body ( most likely at the distal tip of the MtQ/FAZ ) , as well as on the structure observed in the flagellum of BILBO1RNAi cells . It is possible that under these over-expression conditions , either FPC4 deleted of its B1BD is free to move along the MtQ or other MTs and can accumulate at the tip of the MtQ , or that the binding sites at the FPC/HC are saturated and the excess protein is also to along the MtQ where it accumulates . The reasons for FPC4 localisation on the detached flagellum of BILBO1RNAi cells is unclear , but can be explained by its affinity for MTs . An alternative explanation is related to Centrin2 . Centrin2 localises to the BBs , the bilobe , and the flagellum [41 , 58] . Since Centrin2 is also part of the bilobe , in induced BILBO1RNAi cells some components of the bilobe could be assembled together and be targeted to the flagellum in absence of a real bilobe structure . Despite not having any homology at the primary sequence level , the folding of BILBO1 NTD is similar to the N-terminal PB1 domain of Par6 [59] . Par6 is part of the Par complex , a membrane associated complex of polarity proteins , consisting primarily of Par6 , Par3 and aPKC ( atypical Protein Kinase C ) . This complex regulates cellular processes involved in epithelial cell polarity or tight junction formation in epithelial cells [60 , 61] . Par3 is phosphorylated by aPKC and binds to MTs . Par3 and Par6 are multi-modal scaffold proteins that bind to each other as well as to other proteins . [62] . The analogy between this Par complex and BILBO1 –FPC4 –MtQ complex is tempting , especially when considering TbPLK , the unique Polo-like kinase homologue in T . brucei , a key player in the initiation of the cell cycle . TbPLK RNAi leads to a BB segregation defect most probably because it causes defects in FPC duplication or segregation [36 , 63 , 51 , 5 , 32 , 54] . Further , phosphoproteomics and BioID approaches to identify TbPLK binding partners and substrates identified several bilobe proteins , but also the uncharacterised FPC protein Tb927 . 11 . 5640 [36] . Thus , FPC4 interaction with one or several partners might be regulated by phosphorylation . We also localised FPC4 on BSF cytoskeletons and also observed similar FPC and HC localisation as described for PCF . However , we also observed in BSF some labelling towards the BB , similar to the region where TbSpef1 label is located [55] . This would suggest higher levels of FPC4 expression in BSF , easier access for the antibody , or higher rate of traffic to the FPC . Interestingly , in Xenopus embryos , Spef1 ( also named CLAMP ) interacts with aPKC and stabilizes MT [64] . Further work on the role of FPC4 might help understanding how the FPC is segregated and if the MtQ is involved in this process . In summary , our results demonstrate that FPC4 is a multi-partner protein that is involved in FPC segregation and is the first identified link between the FPC , the HC and possibly the MtQ . The dual role of FPC4 , as BILBO1 partner and as a microtubule binding protein , highlights the importance of BILBO1 , making it not only a real ring of power for T . brucei but also a potential target for therapeutic intervention . The resolution of the 3D structure of the BILBO1-NTD/FPC4-B1BD complex is on-going , and it is planned to identify in detail the key amino acid residues in FPC4 that are involved in this interaction . The FPC ( as characterised by BILBO1 ) is an essential cytoskeletal structure , which is also present in T . cruzi and Leishmania ssp [65 , 66] and BILBO1 is highly conserved [25 , 45] . Thus , the FPC is a generic target in kinetoplastids as well as an important component in understanding cytoskeleton biogenesis .
Genomic DNA of T . brucei TREU927/4 GUTat10 . 1 [67] was used to amplify by PCR the FPC4 ORF ( Tb927 . 8 . 6370 ) . The work described in this study used the PCF T . brucei 427 29–13 and BSF 427 90–13 strains co-expressing the T7 RNA polymerase and the tetracycline repressor [68 , 69] ( named WT throughout the manuscript otherwise stated ) . PCF cells were cultured at 27°C in SDM79 medium ( Sigma ) containing 10% ( v/v ) heat-inactivated foetal calf serum , 10 μg . ml-1 Hemin , hygromycin 25 μg . mL-1 , and neomycin 10 μg . ml-1 . Cells were transfected as described in [69] with the transfection buffer described in [70] . We generated the SmOxP427 cell line expressing endogenously tagged 10xTY1-FPC4 using the pPOTv4-10xTY1 vector for PCR template as described in [71] . After transfection , the cells were selected using blasticidin at 20 μg . mL-1 . BSF Tb427 90–13 MITat 1 . 2 ( named in this study as BSF wild-type , WT ) were cultured at 37°C as described in [72] in IMDM medium containing 10% ( v/v ) heat-inactivated foetal calf serum , 36 mM sodium bicarbonate , 136 μg . mL-1hypoxanthine , 39 μg . mL-1 thymidine , 110 μg . mL-1 sodium pyruvate , 28 μg . mL-1 bathocuproine , 0 . 25 mM β-mercaptoethanol , 2 mM L-cysteine , 62 . 5 μg . mL-1 kanamycin , 2 . 5 μg . mL-1 neomycin , and 5 μg . mL-1 hygromycin . They were transfected using the AMAXA electroporator ( Lonza ) as described in [70] . Clones were selected after serial dilutions . Over-expression and RNAi were induced with tetracycline at 10 μg . mL-1 for 48 h otherwise stated . U-2 OS cells ( human bone osteosarcoma epithelial cells , ATCC Number: HTB-96 [73] were grown and transfected as described in [44] and processed for immuno-fluorescence 24h post-transfection . Semi-quantitative RT-PCR was performed on total RNA with a primer pair that amplified a ( 673 bp ) fragment of FPC4 . RNA 18S was used as control for RNA integrity and for loading and was amplified with a primer pair amplifying a 181 bp fragment [76] . Briefly , total RNA of 108 trypanosome cells was collected and resuspended in 500 μL of TRIzol RNA ( 5PRIME ) and treated according to the manufacturer’s instructions . Collected RNA was then treated with DNase-TURBO ( Ambion ) for 30 minutes at 37°C . 100 ng of RNA were used for the RT-PCR reaction via SuperScript III One-Step RT-PCR System with Platinum Taq DNA polymerase ( Invitrogen ) according to the manufacturer’s instructions . BL21 ( DE3 ) ( Novagen ) bacterial strain was transformed with pHis-FPC4-His , pFPC4-1-2606His , pFPC4-6His357-440 and grown in 200 mL of LB + kanamycin ( 50 mg . mL-1 ) to OD600 nm of 0 . 5 . Expression of the recombinant proteins was induced by adding 1 mM isopropyl-beta-D-thiogalactopyranoside ( IPTG ) for 2 h at 37°C . Cells were harvested and resuspended in 20 mL of NaPi 50mM pH 8 . 0 or 9 . 0 , NaCl ( 500 mM for pHis-FPC4-His , 250 mM for pFPC4-1-260-His and pFPC4-His-357-440 ) ( buffer A ) with protease inhibitors ( Calbiochem Cocktail III ) and lysozyme ( 1mg . mL-1 ) . Cells were lysed by sonication . Inclusion bodies ( pHis-FPC4-His ) were pelleted by centrifugation ( 10 , 000 g ) for 30 min 4°C and resuspended in 20 mL of buffer A , sonicated and washed twice in buffer A . The pellet was solubilised in buffer B ( buffer A plus 8 M Urea , pH 8 . 0 ) . The 6HisFPC46His protein was purified by affinity chromatography on a 1 . 25 mL HIS-Select Cartridge ( Sigma ) Nickel affinity gel and eluted with a gradient of Imidazole ( 0 mM—250 mM ) in buffer B . The protein was dialysed against buffer A plus 6 M urea at 4°C , and used to immunise rats for antibody production ( Eurogentec ) . FPC4-1-2606His and FPC4-6His357-440 were present in the supernatant after sonication ( in NaPi 50mM pH8 or pH9 , 250 mM NaCl ) and were purified on affinity columns HisTrap FF ( 1ml ) and eluted with a 20-500mM gradient of imidazole . Proteins were dialysed in PEMD ( 100mM PIPES-NaOH pH6 . 8 , 1mM EGTA , 1mM MgCl2 , 1mM DTT ) . Protein concentration was assessed with the Pierce 660nm protein assay kit . For SLS and ITC experiments , transformed bacterial cells were grown at 37°C until OD600nm reached 0 . 6–0 . 8 , and then subjected to cold shock on ice for 10 min . Afterwards , the cell culture was further incubated at 16°C for 30 min before 0 . 5 mM of IPTG was added to the cell culture to induce expression . The cells were harvested the next day by centrifugation in a Sorvall GS-3 rotor ( 6 , 000 g , 12 min , 4°C ) and resuspended in cold lysis buffer ( 20 mM Tris-HCl ( pH 8 . 0 ) , 300 mM NaCl , 20 mM imidazole , and 5% ( v/v ) glycerol ) . The cells were broken by the EmulsiFlex-C3 homogenizer ( Avestin ) and the lysate was cleared by centrifugation at 30 , 000 × g for 30 min . The supernatant was filtered through a 0 . 4-μm filter and loaded onto a Ni-HiTrap column ( GE Healthcare ) pre-equilibrated in the same lysis buffer . The column was washed with 5 × column volume ( cv ) of lysis buffer , and bound protein was eluted by a linear gradient concentration of imidazole ( 20–600 mM , 10 × cv ) in the same lysis buffer . The fusion tags on BILBO1-NTD’ and FPC4-B1BD were removed by incubating with ~1% ( w/w ) of thrombin and ~2% ( w/w ) of TEV , respectively , overnight at 4°C . Target proteins were further purified with a Superdex-200 16/60 column ( GE Healthcare ) pre-equilibrated with 20 mM Tris-HCl ( pH 8 . 0 ) , 100 mM NaCl and 5% ( v/v ) glycerol . Fractions containing each target protein were pooled , concentrated , and used for subsequent analysis . FPC4 and BILBO1 ORFs were cloned into pGADT7-AD and pGBKT7 respectively . The pGADT7-AD ( prey ) and pGBKT7 ( bait ) based plasmids were transformed in the Y187 and Y2HGold yeast cell lines respectively . After production of diploids cells , interaction tests were done on SC-L-W-Histidine media , and control media containing histidine , using the drop test technique as described in [44] . Five mg . mL-1 taxol-stabilised MTs were prepared as in [79] . For co-sedimentation assays , 4 μL of MTs were mixed with different amounts of protein dialysed in PEMD buffer . Volumes were completed to 50 μL with PEMD ( 7 . 2 μM tubulin final concentration ) . Samples were then incubated for 1 h at 22°C and centrifuged 15 min at 16 , 100xg at 22°C . Supernatants and pellets were separated and brought to equal volumes in SDS sample buffer . Equal volumes of pellets and supernatants were analysed by SDS-PAGE . Gels were stained using Instant Blue ( Expedeon ) . Flagella were prepared as previously described in [43] , loaded on Formvar/Butvar covered , charged , carbon-coated nickel grids and incubated for 10–15 min to let them adhere . The grids were then moved to a 250 μL droplet of 1% Nonidet P-40 ( IGEPAL ) in PEME ( 100 mM PIPES-NaOH pH 6 . 9 , 1 mM MgCl2 , 0 . 1 mM EDTA , 2 mM EGTA ) and 1:10 , 000 complete protease inhibitor cocktail ( Calbiochem ) and incubated for 5 min and repeated once on a fresh droplet . The grids were then moved to a droplet containing in addition 1 M KCl , for 30 min on ice . After four washes in PEME buffer with protease inhibitors , the flagella were fixed on a 100 μL 3% PFA in PEME droplet for 5 min . PFA was then neutralised with four incubations on 100 μL droplets of 100 mM glycine in PEME . The grids were then transferred through five blocking droplets for 5 min each ( 0 . 1% BSA , 0 . 1% Tween-20 in PBS ) then to droplets containing the primary antibodies–anti-myc ( mouse monoclonal 9E10 , 1:20 ) , anti-BILBO1 1–110 ( rabbit polyclonal , 1:400 ) , anti-MORN1 1340 ( rabbit polyclonal , 1:400 ) , anti-myc ( rabbit , Santa Cruz , 1:100 ) , and anti-tubulin TAT1 ( mouse monoclonal—a kind gift of K . Gull , 1:50 ) –and were incubated for 2h at RT in PBS 0 . 01% BSA and 0 . 1% Tween-20 . The grids were washed four times in PBS 0 . 01% BSA and 0 . 1% Tween-20 and incubated with gold-conjugated secondary antibodies–EM GAR10 1:20 ( anti-rabbit , 10 nm ) , and EM GAM15 1:50 ( anti-mouse , 15 nm ) –in PBS 0 . 01% BSA and 0 . 1% Tween-20 for 2h at RT . The grids were then washed twice in blocking buffer , twice in PBS and then fixed for 5 min in 2 . 5% glutaraldehyde in PBS . After two washes in milliQ H2O , the grids were negatively stained in 0 . 5% Nanovan for 5–10 sec . For the double immuno-labelling on SmOXP427 cells expressing 10xTY1-FPC4 , flagella were prepared as described above , but with the following modifications . All buffers including blocking buffers contained protease inhibitors . Flagella were not fixed prior to antibody incubation . After washing , the flagella were blocked in 1% or 2% fish skin gelatin , 0 . 01% Tween-20 in PBS pH . 7 . 3 for 30 min . Grids were incubated on antibody droplets sequentially for 60 min starting with mouse monoclonal anti-TY1 ( BB2 ) , diluted 1:5 in blocking buffer . Grids were then washed 3 x 5 min in blocking buffer , then incubated in 30μL droplets of anti-mouse 5nm gold ( British Biocell International BBI solutions ) . Grids were then washed 3 x 5 min in blocking buffer and incubated in 30μL droplets of affinity purified anti-BILBO1 ( mAb 5F2B3 , IgM diluted 1:5 in blocking buffer ) or rabbit anti-MORN1 diluted 1:250 in blocking buffer . Grids were then washed 3 x 5 min in blocking buffer and incubated on 30μL droplets of anti-mouse IgM 15 nm gold ( for grids incubated with anti-BILBO1 ) diluted 1:10 in blocking buffer or anti-rabbit IgG 15 nm gold ( for grids incubated in anti-MORN1 ) diluted 1:10 in blocking buffer . After incubation , grids were washed 3 x 5 min in blocking buffer , 2 x 5min in 0 . 1% fish skin gelatin , 0 . 001% Tween-20 in PBS pH 7 . 3 , 2 x5 min PBS pH 7 . 3 and then fixed in 2 . 5% glutaraldehyde in Milli-Q water for 5 min . Samples were negatively stained with 5–10μL Aurothioglucose for 20 sec . For ultra-thin sections , a mid-log phase culture was fixed for 2h in medium with 2 . 5% glutaraldehyde , and consequently fixed in 2 . 5% glutaraldehyde in 0 . 1 M Sorensen’s Sodium Phosphate buffer pH 7 . 2 . The cells were rinsed twice in milliQ H2O and incubated for 1 h in 1% OsO4 in milliQ H2O pH 7 . 2 . After three washes in milliQ H2O , the cells were stained and fixed in 2% Uranyl-acetate ( milliQ H2O ) at 4°C overnight . After three 10 min washes in milliQ H2O , the cells were dehydrated in EtOH , starting from 30% EtOH , then 50% , 70% and 90% for 2h each and left in 90% EtOH overnight . The cells were incubated three times up to 1h in 100% EtOH at RT and then for 3h each with 30% , 50% , 70% and 90% EtOH: Spurs resin . Finally , the cells were incubated three times for 1h with 100% Spurs resin at RT and then overnight in 100% Spurs ( Low viscosity embedding kit , EMS 14300 ) at RT . The resin was embedded in size “OO” BEEM capsules ( EMS ) and let polymerize overnight at 60°C . Thin sections were cut with Ultramicrotome LEICA EM-UCT at a thickness of approximately 70–90 nm . The sections were deposed on grids and stained with aqueous saturated uranyl acetate for 15–30 min . The grids were washes three times in boiled cooled water for 5 min each and then air-dried . An additional wash with 0 . 1 N NaOH for 30 sec and three washes in boiled cooled water for 5 min . Samples were visualised on a FEI Tecnai 12 electron microscope , camera ORIUS 1000 11M Pixel ( resolution 3–5 nm ) . Images were acquired with Digitalmicrograph and processed with ImageJ . Bacterial extracts , trypanosome whole cell protein lysates and purified proteins were separated on SDS-PAGE gels ( 10%–15% ) and transferred by semi-dry ( BioRad ) blotting 45 min at 25V on PVDF membrane . After a 1 h blocking step in 5% Milk in PBS-0 . 2% Tween , the membranes were incubated with the primary antibodies diluted in blocking buffer–anti-TbSAXO ( mAb25 mouse monoclonal , 1:1 , 000 dilution [80] ) ; anti-FPC4 ( rat , 1:500 dilution ) and anti-myc ( rabbit polyclonal ( Santa Cruz ) , 1:500 dilution ) . After three washes in blocking buffer , the membranes were incubated with the secondary antibodies–anti-mouse antibody HRP-conjugated ( Jackson , 1:10 , 000 dilution ) , anti-rat antibody HRP-conjugated ( Jackson , 1:10 , 000 dilution ) and anti-rabbit antibody HRP-conjugated ( Sigma , 1:10 , 000 dilution ) –and washed twice 10 min in blocking buffer and twice 5 min in PBS . Blots were revealed using the Clarity Western ECL Substrate kit ( Bio-Rad ) with the ImageQuant LAS4000 . The SLS experiment was carried out on a liquid chromatography apparatus ( Wyatt Technology Corp . ) consisting of an HPLC system ( Agilent Technologies ) connected in series with a triple-angle laser light scattering detector ( miniDAWN TREOS ) , a UV detector at 280 nm ( Agilent technologies ) , and a refractive index detector ( RI-101 , Shodex ) . 100 μl of purified BILBO1-NTD’ and FPC4-B1BD were mixed ( molar ratio ~1:2 , total 2 mg/ml ) and , after 30-min incubation at room temperature , eluted from a Superdex 200 10/300 GL column ( GE healthcare ) at a flow rate of 0 . 5 ml per min . Data analysis was carried out using the Astra software ( Wyatt technology ) . Purified BILBO1-NTD’ and FPC4-B1BD were dialysed overnight against a buffer containing 20 mM Tris-HCl ( pH 8 . 0 ) and 50 mM NaCl . Protein concentration was determined by ND-1000 spectrophotometer ( PEQlab ) . ITC experiments were carried out at 25°C using an iTC200 Microcalorimeter ( MicroCal , GE healthcare ) . The cell contained 200 μl of 50 μM BILBO1-NTD’ , which was titrated with an initial 0 . 4 μl injection followed by 19 constitutive injections ( 2 μl each ) of 600 μM FPC4-B1BD with duration of 0 . 8 s . The interval between each two injections was 150 s . The ITC data were analysed using the program Origin version 7 . 0 provided by MicroCal . One-site binding model was used to fit the integrated data to calculate the stoichiometry and binding constants . | Trypanosoma brucei is the parasite responsible for human African trypanosomiasis , a disease also known as sleeping sickness . African trypanosomiasis is present in Sub-Saharan Africa and transmitted by infected tsetse flies . This devastating disease is lethal if untreated , making it important to understand and characterise the basic biology of the pathogen . During the T . brucei cell cycle , organelle positioning and segregation show a high degree of coordination and control . As such , T . brucei is a highly polarised cell with the transition zone of the flagellum present inside an invagination of the pellicular membrane called the flagellar pocket ( FP ) . The FP is the main site of traffic into and out of the cell . At the exit point of the flagellum is a cytoskeletal structure called the flagellar pocket collar ( FPC ) . One component of the FPC , BILBO1 , has been characterised as essential for the FP biogenesis and cell survival . Here , we identify a new partner of BILBO1 called FPC4 , and determine the domains involved in the BILBO1-FPC4 interaction . We further highlight the role of FPC4 in the segregation of the FPC and in the interplay between the FPC and other essential cytoskeletal structures . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"medicine",
"and",
"health",
"sciences",
"rna",
"interference",
"pathology",
"and",
"laboratory",
"medicine",
"microtubules",
"pathogens",
"parasitic",
"protozoans",
"trypanosoma",
"brucei",
"protozoans",
"epigenetics",
"immunologic",
"techniques",
"cellular",
"structures",... | 2017 | Interaction between the flagellar pocket collar and the hook complex via a novel microtubule-binding protein in Trypanosoma brucei |
Protein-protein interactions ( PPIs ) formed between short linear motifs and globular domains play important roles in many regulatory and signaling processes but are highly underrepresented in current protein-protein interaction databases . These types of interactions are usually characterized by a specific binding motif that captures the key amino acids shared among the interaction partners . However , the computational proteome-level identification of interaction partners based on the known motif is hindered by the huge number of randomly occurring matches from which biologically relevant motif hits need to be extracted . In this work , we established a novel bioinformatic filtering protocol to efficiently explore interaction network of a hub protein . We introduced a novel measure that enabled the optimization of the elements and parameter settings of the pipeline which was built from multiple sequence-based prediction methods . In addition , data collected from PPI databases and evolutionary analyses were also incorporated to further increase the biological relevance of the identified motif hits . The approach was applied to the dynein light chain LC8 , a ubiquitous eukaryotic hub protein that has been suggested to be involved in motor-related functions as well as promoting the dimerization of various proteins by recognizing linear motifs in its partners . From the list of putative binding motifs collected by our protocol , several novel peptides were experimentally verified to bind LC8 . Altogether 71 potential new motif instances were identified . The expanded list of LC8 binding partners revealed the evolutionary plasticity of binding partners despite the highly conserved binding interface . In addition , it also highlighted a novel , conserved function of LC8 in the upstream regulation of the Hippo signaling pathway . Beyond the LC8 system , our work also provides general guidelines that can be applied to explore the interaction network of other linear motif binding proteins or protein domains .
A large number of protein-protein interactions ( PPIs ) are mediated by short linear motifs ( SLiMs ) that are recognized by specific globular domains [1] . SLiM-mediated interactions are involved in a wide range of biological functions and can regulate the formation of transient protein complexes , orchestrate subcellular localization , modulate post-translational modification state , and determine the fate of proteins [1] . Such interactions emerged as key mediators of complex regulatory processes in higher eukaryotic cells and their aberrant functioning can contribute to various diseases as well [2] . The key to the essential nature of SLiMs in biological systems lies in their specific properties . SLiMs correspond to a stretch of approximately 3–10 residues that generally reside within intrinsically disordered regions ( IDRs ) . As a result , they usually form transient , weak interactions with micromolar binding affinity [3] . Due to these specific properties , the identification of linear motif sites is challenging both experimentally and computationally [4] . Currently , the most comprehensive collection , the Eukaryotic Linear Motif database ( ELM ) holds only 200–300 motif patterns with a few thousands of experimentally verified instances [5] . This number pales in comparison to the expected number of linear motif mediated interactions in the human proteome , estimated to number at least several hundred thousand [6] . Linear binding peptides have been systematically analyzed only for a few specific interaction domains , such as SH2 , PTB , 14-3-3 , PDZ and SH3 domains [7 , 8] . However , for most motif binding domains , the interaction network is largely incomplete . The identification of linear motif mediated interactions is usually divided into two phases . The first phase is the characterization of the common consensus sequence motif that is shared among the diverse set of binding partners of a common domain , i . e . defining the motif class [9 , 10] . The core motif that mediates interactions with a given domain is usually represented by a sequence pattern or a position specific scoring matrix ( PSSM ) . In the second phase , the core motif is used to identify additional candidate binding sites in the proteome , i . e . finding novel motif instances . However , as the information content of consensus motifs is usually low , predicted motif matches are overwhelmingly dominated by false positive matches that occur purely by chance [3 , 11] . Therefore , this phase involves additional filtering steps to remove matches that are unlikely to be functional and to prioritize motif hits for further experimental characterization . Various computational tools such as ELM , QuasiMotifFinder , MiniMotifMiner , SLiMSearch , ScanProsite or DOReMi [5 , 12–16] have been developed to overcome this problem . These tools scan a defined set of proteins with a single consensus motif and utilize various discriminatory attributes to prioritize motif hits , including structural context , protein disorder , functional ontology , evidence for PPIs and shared cellular localization . Evolutionary conservation can highlight functionally relevant positions in proteins and have been used for globular domains to identify conserved motif-like patterns that are indicative of the function of a protein [12] . However , linear motif sites generally reside within IDRs that are generally less conserved [17] . Within these regions , SLiMs often show a specific pattern of evolutionary conservation that are characterized by a higher relative conservation of the key motif residues compared to their flanking regions [18 , 19] , and this information can be used to highlight true binding motifs . Using various sequence attributes , specific filtering pipelines were constructed to identify novel motif instances for several domains [20–24] . However , the strength and optimality of the filtering steps have never been systematically tested . In order to build optimal filtering protocols , a more systematic approach is needed that can take into account the specific trade-off between reducing false positive hits while capturing biologically relevant motif matches , which is likely to be specific to individual binding domains . In this study , we focused on LC8 dynein light chain and its binding partners as a case study , and explored how the interaction network of a specific linear motif binding protein can be expanded in an optimal way . LC8 is a remarkably conserved eukaryotic hub protein [22] . Although LC8 was originally suggested to function as a cargo adaptor for the dynein motor complex , its extensive interaction network suggests a more general role , independent of dynein [22 , 25] . Recently , the prevailing view has become that LC8 functions as a dimerization or oligomerization engine for various proteins [25 , 26] . Known interaction partners link LC8 to processes such as nuclear transport , tumor suppression , viral replication , DNA damage repair , apoptosis , mitosis and signaling [22 , 25] . In contrast to their functional heterogeneity , LC8 binding partners generally share a common binding mechanism [22 , 27] . The known structures of complexes between LC8 and various bound partners show that the binding groove is formed at the dimerization interface of the homodimeric LC8 , favoring binding partners that are also dimerized [28 , 29] , often promoted by coiled coil ( CC ) regions [20 , 30] . The recognition SLiMs are generally located within intrinsically disordered regions [25] and undergo a disorder-to-order transition upon complex formation . In their bound form , these segments adopt highly similar conformations that augment the central beta sheet of LC8 on each side of the dimer [27] . Many of the binding segments contain a Thr-Gln-Thr ( TQT ) motif with additional positions showing larger variations . Apart from the canonical TQT motif , there are also non-canonical binding motif instances , in which the central Gln is replaced by Met or Asn [22] . An even more unusual Thr-Ser-Pro ( TSP ) binding motif mediates the interaction between Pak1 and LC8 . Altogether , more than 50 LC8 binding motif instances were collected from various eukaryotic species [22] . To reveal the optimal binding motif , a phage display study was also carried out , identifying further LC8 binding motif instances [20] . The suggested general role for LC8 raises the possibility of many additional binding partners in the human proteome . In this work we expanded the interaction network of dynein light chain LC8 using a combination of computational and experimental methods . We introduced a novel measure based on information gain that enabled us to build an optimal bioinformatic pipeline by combining various attributes predicted from the amino acid sequence . We also incorporated known binding partners from PPI databases and exploited the specific evolutionary conservation of binding motifs to increase the likelihood that the motif hit is biologically relevant . The resulting procedure enabled us to drastically reduce false positive predictions among putative novel linear motif instances and to expand the interaction network of LC8 with high-confidence predictions . We experimentally verified the binding of several novel motif instances to LC8 using surface plasmon resonance ( SPR ) assay . One of the most interesting outcomes of the extended interaction network of LC8 revealed a possible new function of the LC8 protein in the Hippo pathway through interaction with WWC and AMOT protein family members . The presented study significantly contributes to the better understanding of the functional and evolutionary properties of the LC8 interactome . Beyond this specific hub protein , it also offers general guidelines for the exploration of additional linear motif-mediated interaction networks .
LC8 recognizes a short linear motif in its partner proteins . In this case , the binding involves both polypeptide chains of the homodimer LC8 . However , to emphasize the similar binding mode to many single modular domains that also bind short linear motifs , LC8 is also referred to as a “binding domain” in this article . We assembled a manually curated database of LC8 interaction partners based on literature search , in which LC8 binding was verified at the motif level ( see Materials and Methods and S1 Table ) . The final dataset contained 53 partners with 67 motif instances and covered multiple eukaryotic species and viruses . 40 motifs in 33 proteins belonged to human or could be directly mapped to a human protein based on close homology . The length of the core binding motif was taken as 8 amino acids ( S1 Fig; for details , see S1 Text ) . The ELM database describes the LC8 binding motif using the regular expression “^P . K . TQT” . From the 67 known motif instances , only 13 matched this regular expression , and 24 further motifs contained only the canonical “TQT” motif core . Considering human partners only , the corresponding numbers were 7 and 17 , respectively . These numbers indicate that the definition in the ELM database is too restrictive to capture the majority of known instances . To better describe the common sequential properties of known LC8 binding sites , we used a position specific scoring matrix . The calculated PSSM is shown in Fig 1 . Positive scores indicate amino acid residues that are favored at a given position . The PSSM clearly captures the frequent occurrence of the canonical “TQT” motif . However , with the exception of Gln in position 0 , there are additional favored amino acids in every position . The strong preference for Lys at position -3 from the central Gln is not supported by the current collection of known partners , as neighboring positions have stronger preferences according to the bitscore . Using the obtained PSSM , we scanned the whole human proteome to score every overlapping eight amino acid long peptide segment ( S2 Fig ) . As expected , all known LC8 binding peptides had a positive score . While only 2% of human peptides had a positive score , these still represented more than one hundred thousand cases . This indicates that on the one hand , positive PSSM scores are strongly associated with true binding motifs and can serve as a valid starting point for novel LC8-binding motif discovery . On the other hand , the very large number of initial motif hits underscore the need for that additional filtering steps . In order to establish additional filtering criteria , we gathered various predicted features of the PSSM-identified peptides . The methods we used included PFAM annotations [31] , average disorder prediction scores ( using IUPred , PONDR VSL2 , Espritz and DISOPRED3 ) [32–35] , average score to be part of disordered binding regions ( using ANCHOR , MORF-CHIBI and DISOPRED3-BR ) [35–37] , and secondary structure prediction scores using PSIPRED [38] . At the protein level , information about predicted cellular localization [20 , 39] and the presence of coiled coil regions was also collected ( See Materials and Methods ) . The various data for all known true positive motifs from human and other species are available at http://gerdos . web . elte . hu/data/LC8/known_results . html . These various features can be associated with known motifs to different extents . The main challenge is to find the best tools and parameter settings that enable the prioritization of peptide segments that are the most likely to be biologically relevant motif hits in an optimal way . We introduced a metric based on weighted information gain derived from the Shannon entropy to globally attest the discriminatory power of each criterion . On a dataset containing 40 known human binding partners and 10 , 000 random human segments from the proteome with a higher than zero PSSM score , the weighted information gain was calculated for each criterion . The suggested measure enabled not only to rank the attributes in terms of their discriminatory power , but also to choose the best parameter settings . According to this protocol , the strongest criterion based on the information gain was the predicted intracellular localization , which was fulfilled by all known motifs , but only 85 . 53% of random peptides . The next strongest filtering criterion was based on PFAM annotations . Using annotation strictly based on the domain type PFAM families , it was possible to filter out 15% of random motifs , while retaining all known motifs . The third strongest criterion was based on PSSM scores . The highest information gain was achieved with the cutoff value of 3 . 3 . This cutoff value was not met by only three known human motifs ( PAK1 , NRF1 , MYZAP ) . At the next level , disorder prediction methods produced the highest information gain . Four methods were tested with different cutoff values corresponding to different false positive rates . The optimal choice was IUPred with the cutoff value of 0 . 42 , which was fulfilled by all but one known motifs . We also tested three methods ( ANCHOR , MoRFchibi , DISOPRED3 ) [35–37] for predicting disordered binding regions also known as molecular recognition features ( MoRFs ) . The information gain was much lower , compared to previous criteria . The optimal information gain was reached using the ANCHOR method with a cutoff value of 0 . 57 . However , even this criterion would filter out 57 . 5% of the positive examples . An additional criterion considered was secondary structure prediction . Based on known structures of the complexes , the binding motif is expected to adopt a beta-strand conformation . Surprisingly , predicted beta-strands occurred in only four out of 40 cases , and even in these cases only very short segments were predicted to be in beta conformation . Helical segments were predicted in five cases , including MYO5A , which has been shown to have helical tendencies in the unbound form [40] . However , the lack of residues predicted to be in helix did not perform well as a filtering criterion as the majority of random hits were also predicted to lack regular secondary structural elements , and the overall information gain was below that of predicting disordered binding regions . Similarly , the presence of coiled coil regions in the partner proteins predicted by NCoils [41] did not have a strong discriminatory power either . As these criteria filtered out many known motifs , they were not included in our filtering protocol . The resulting filtering protocol is shown on Fig 2 , indicating the proportion of random hits and known motifs that were eliminated at each step . By the stepwise application of the four filtering steps , a drastic reduction of random hits could be achieved , while still retaining the majority of known motifs: 90% of known motifs were kept while 99 . 78% of random hits were filtered out ( i . e . only 0 . 22% was kept ) . By applying the filtering protocol for the complete human proteome , 335 candidate motifs remained . We carried out a 3-fold cross validation to measure the generality of the obtained filtering protocol ( see Materials and Methods ) . On average , we were able to correctly categorize 34 examples from the 40 experimentally validated human binding motifs . This was only marginally worse compared to using the complete database where 36 motifs were categorized correctly . The cut-off values for the PSSM and IUPred scores also did not change largely . This indicates that the protocol is robust and can correctly describe the general attributes of the binding event between LC8 and its known binding partners . Given the extreme conservation of LC8 , it is of special interest how the interaction motifs are conserved in partner proteins . To study this , we generated multiple sequence alignments of orthologous proteins harboring known LC8 binding motifs and categorized them into 5 taxonomic groups: Mammalia , Vertebrata , Metazoa , Fungi and Eukarya . The conservation at the level of protein and motif was tested based on the obtained alignments in each taxonomic group . The definition of motif conservation applied here depends on the PSSM score of the motif ( see Materials and Methods ) . Consequently , this approach cannot be applied to motifs that significantly differ from canonical motifs , i . e . their binding motif had a PSSM value below the threshold . These three examples were excluded from the analysis . Nevertheless , this strict criterion ensured that not only the general sequence similarity is maintained , but also the similarity to known human LC8 motifs is preserved . The results of the conservation analysis for the known human motifs are presented in Fig 3 . Among the known partners , the LC8 binding sites located within dynein intermediate chains ( DYNC1|1 , DYNC1|2 ) exhibited the most pronounced conservation , spreading across a wide range of eukaryotic species ( e . g . starlet sea anemone , C . elegans , slime mold ) . In contrast , other motifs identified in human or other mammalian species were generally not conserved beyond Vertebrata ( e . g . , BMF , DLGAP1 , MTCL1 ) . In the majority of cases , not only the LC8 binding motif but the complete protein was lost beyond this evolutionary distance ( e . g . BMF , BSN , SNPH ) . Similarly , limited conservation of LC8 motifs was observed for binding regions experimentally verified in other species ( S3 Fig ) . For example , the EGL protein from Drosophila melanogaster had orthologues at each level , but its binding motif showed conservation only in metazoan species . While the centriole duplication functionality is conserved across metazoan species , the LC8 binding protein ANA2 involved in this process is specific to Drosophila species [42] . Lowering the PSSM cutoff value used to define the conservation of the motifs perturbed the results only in three cases: NRF1 , FAM83D and MYO5A , but had no major impact on the overall trends ( S4A and S4B Fig ) . We can conclude from this analysis of motif conservation that while the LC8 binding interface is highly conserved , known partners and especially the binding motifs located within them are significantly less conserved . We tested the specific conservation pattern of motif binding sites using the SLiMPrints algorithm [19] . This method identifies short stretches of residues within disordered regions that show high relative conservation compared to their flanking region as calculated from multiple sequence alignments of orthologous sequences . In our dataset , 13 out of the 40 known human motifs exhibited high relative conservation , which is only 32 . 5% of all cases . This suggests that the approach based on the island-like conservation identified by SLiMPrints has a limited ability to highlight putative functional binding motifs and overall it is not a good filtering criterion , as it misses the majority of functional motifs . Here , we suggest an alternative filtering criterion based on the observation that known motifs are generally conserved within their own taxonomic group even if they lack motif level conservation over wider evolutionary distances . For the majority of the cases , known motifs in human sequences were conserved in at least 80% of mammalian species . The exceptions included only the three cases that had a PSSM value below our cutoff , and two additional examples ( the MAP4 motif and the second motif of FAM83D at position 405 ) . In contrast , randomly chosen motif hits located within disordered regions generally did not possess this property . Applying this discriminatory technique ( see Materials and Methods ) , we could filter out 163 of the possible candidate peptides , reducing the number of candidate motifs to 172 . Therefore , for human sequences the conservation within mammalian species can be used as an efficient filter to further reduce likely false positives . The predicted motif hits were also analyzed in relation to experimental data in PPI databases . Known interaction partners of LC8 were collected using the integrative PSICQUIC approach [43] , which enabled us to search multiple databases simultaneously and to group the experiments based on detection methods and association types ( see Materials and Methods ) . Altogether , 381 LC8 interaction partners were collected ( S2 Table ) . Several of these interaction partners were supported by multiple lines of evidence , with a total of 782 independent experiments . Known motif partners were well represented in current PPI databases , with only three missing out of these 33 partners . The comparison of interaction partners collected from PPI databases with those that contained known motifs indicated that there is no single criterion that can identify biologically relevant interactions from candidates in PPI databases . However , PPIs from low-throughput and direct experiments are more likely to be biologically relevant , especially when they are supported by multiple independent measurements ( see S5 Fig ) . Overall , there were a large number of potential partners recorded in PPI databases ( 323 ) that contained neither known nor predicted LC8 binding motif instances . By looking at the distribution of partners over the number of supporting experiments , we can see that partners with known and predicted motifs were largely evenly distributed , similarly to the partners detected by direct methods ( Fig 4A ) . Known and predicted motifs dominated in partners with at least 7 supporting measurements , with a single exception corresponding to KANK2 , in which no motif-like segments could be identified . In contrast , most of the partners without a likely motif candidate were detected by only a single , indirect method . Although it cannot be ruled out that some of these interactions represent an alternative binding mechanism to LC8 , the large number of PPIs that are not compatible with the existing binding mode underlines that data from PPI databases should be treated with caution as they may contain several false positives . Among the 172 predicted motifs in 152 proteins that satisfied our filtering criteria , there were 43 novel motif instances predicted by our pipeline that had corresponding PPI data . While most of these interactions were based on indirect , high-throughput experiments , many of them were supported by multiple measurements . 8 of these motifs represented additional instances in proteins that already contained an experimentally verified binding motif . In 35 cases , the likely binding region for LC8 could be identified for proteins whose interaction were studied only at the level of the protein ( see S2 Table ) . Furthermore , the presence of PPIs lends support to the biological relevance of the predicted motifs in these cases . By considering motif hits located in proteins with corresponding PPI data , we created a dataset that contained a list of high confidence motif instances for LC8 . In addition to PPIs , the presence of island-like conservation was also taken as an indication of likely motif hits . Altogether , we collected 71 high confidence motif instances , supported by PPI data ( 29 cases ) , the presence of island-like conservation ( 27 cases ) , or both ( 15 cases ) ( Fig 4B ) . Using these complementary information , the number of interaction partners of LC8 could be practically doubled ( S3 Table ) . The high confidence motifs are available at http://gerdos . web . elte . hu/data/LC8/HUMAN/high_confidence_hits . html . The high confidence set together with the already known motif hits was used to carry out a GO enrichment analysis against the human proteome with the DAVID server [44] . The analysis showed an enrichment of multiple functionalities previously associated with LC8 , such as cytoskeletal organization , microtubule binding or cell morphogenesis ( Fig 5 ) . Besides the already known activities , the enrichment analysis revealed a highly over-represented , yet previously undiscovered function of LC8 that links the protein to the regulation of cell and tissue growth , and in particular , to the Hippo pathway . This result suggests that LC8 might play a critical role in the regulation of the Hippo pathway . In order to understand how LC8 is connected to the Hippo pathway , we took a closer look at the relevant partners . The motif hits highlighted two families involved in the upstream regulation of the Hippo pathway , the WWC ( WWC1 , WWC2 and WWC3 ) and angiomotin ( AMOT , AMOTL1 , AMOTL2 ) families [45 , 46 , 47] . The WWC family member WWC1 , also known as KIBRA , was shown previously to interact with LC8 [48] , and two binding motifs were also identified , located at positions 278 and 887 ( highlighted in Table 1 ) [49] . However , no binding sites have been previously identified in the other two members of the family , WWC2 and WWC3 . Interactions between LC8 and the AMOT family members were previously reported in high-throughput studies to be LC8 interactors [50] , but the binding motif has not been located in any of these proteins as of yet . Here , we identified the likely LC8 binding regions in all WWC and AMOT family members and analyzed their evolutionary conservation . The binding of these peptides was verified and quantitatively characterized by SPR ( Table 1 ) . The list of putative LC8 binding peptides identified by our protocol are shown in Table 1 , together with the measured binding constants , which confirmed the binding of the selected peptides to LC8 . As an example , results are shown for the motif in AMOTL2 in Fig 6 . Although the apparent Kd values indicate weak interactions , these binding affinities are similar to other LC8 interactions [25 , 20] , and in biological settings they can increase in strength due to avidity caused by dimerization [30] . Besides the canonical TQT motifs these peptides also contain SQT and TNT motifs . Additional peptides that further expand the repertoire of compatible amino acids at various positions were also tested for their binding to LC8 ( S4 Table ) . In order to properly assess the conservation of LC8 binding motifs in WWC and AMOT family members , we traced the evolutionary history of these proteins and their identified binding regions and additional functional modules . Almost all vertebrate species have three paralogs of both the WWC and AMOT family members ( Figs 7 and 8 ) . For both families , the multiple paralogs observed at the level of vertebrates could be traced back to a single common ancestor gene , as shown in the case of Florida lancelet . KIBRA orthologs were detected beyond the level of chordates , not only in arthropods , but also in cnidaria ( Fig 7 ) . AMOT was also present in arthropods , but it was missing in the fruit fly ( Fig 8 ) . The most likely explanation for this is that the angiomotin was originally present in all bilaterian animals , but was lost relatively recently in the dipteran lineage that includes Drosophila [51] . Altogether , the WWC family members seem to have a more ancient evolutionary origin compared to angiomotins . Both WWC and AMOT family members showed a highly conserved domain and linear motif organization across a wide range of species ( Figs 7 and 8 ) . While most LC8 binding motifs verified in human sequences are not conserved beyond vertebrates , KIBRA and the other WWC family members represent an important exception in this regard . The two LC8 binding motifs of this protein family are well conserved in all three paralogs in vertebrates , lancelet and fruit fly . Furthermore , one of the motifs is also present in a sea anemone , matching the strong evolutionary conservation of the WW and C2 domains , the defining functional modules of this family ( Fig 7 ) . The AMOT family members contained a single LC8 binding motif , which seems to have emerged more recently . While orthologous sequences could be detected in non-vertebrate species based on the conserved coiled coil region together with the angiomotin domain , these sequences lacked the LC8 binding motif ( Fig 8 ) . The LC8 motif , similarly to the PDZ and WW binding motifs , has become conserved within vertebrate species . In both of these families , the redundancy and conservation underline the functional importance of LC8 binding motifs in the upstream regulation of the Hippo pathway .
In this work , we established a systematic filtering protocol that can be used to expand the interaction network of a given linear motif binding domain with high confidence motif hits and reduce the number of random motif matches . While similar bioinformatic pipelines have been used before [14] , the optimality of the applied filtering steps could not be guaranteed or even assessed due to the lack of appropriate measures . Here , we offered a solution to this problem by introducing a decision tree-like filtering procedure together with the weighted information gain that enabled the construction of an optimal bioinformatic pipeline . The presented approach was applied to expand the interaction network of LC8 , a highly conserved eukaryotic hub protein that binds its partners via a specific linear motif [22] . By combining our motif filtering protocol with data collected from protein-protein interaction databases and the information on the presence of island-like conservation , we created a dataset of 71 novel high confidence motif instances ( S3 Table ) . These novel binding sites significantly enriched the interaction network of LC8 with novel partners and highlighted a previously unknown , important function of LC8 in the Hippo pathway . A list of predicted LC8 binding motifs was also created in an earlier work [20] . In this case , the PSSM was calculated based on a library of binding motifs evolved through phage selection instead of using the collection of naturally occurring motif instances . Although some of the filtering criteria , like IUPred based disorder prediction or intracellular localization were used in both studies , the details of implementation differed significantly , including the optimal cutoff for the PSSM score . The comparison of the two sets showed that among the top 83 hits identified based on the in vitro evolution and absent from the list of known naturally occurring motifs , 58 ( 70% ) were also uncovered by our presented motif discovery pipeline . However , our evolutionary filtering criteria reduced this overlap to 38 ( 46% ) . While to some extent , the limited overlap could be due to the technical differences in the implementation , a more appealing explanation is that binding motifs selected by phage display are optimized for binding strength , while the sequence of naturally occurring motifs were shaped by various evolutionary requirements in the cell , from which strength is just one , and not necessarily the dominant constraint . Despite these differences , in vitro evolution by phage selection [20 , 52] represent a complementary approach to predict potential motif instances for specific binding domains and to explore the binding preferences of such domains . While the filtering protocol implemented here is specific to LC8 , the presented work also has important implications beyond this system . We suggested here a general framework to find the optimal selection of methods for the filtering steps . The importance of this optimization can be best demonstrated through the example of disorder prediction methods . Several methods , such as DISOPRED3 , PONDR VSL2 or ESpritz Disprot [33–35] , that were tested in this work , perform better on specific datasets of ordered and disordered proteins [53–55] . Nevertheless , IUPred achieved the best results for this specific problem , supporting the choice of this approach in motif-centric application of protein disorder [4 , 14 , 19] . The pipeline applied here also enabled us to identify features that had limited discriminatory power and therefore were not incorporated into the current pipeline . For example , all known LC8 binding motifs adopt a β-strand conformation upon binding , but the current tools are not capable of capturing this property efficiently . Additional features , such as predicted disordered binding regions/MoRFs , the presence of coiled coils or island-like conservation also had limited discriminatory power . As these properties are clearly associated with at least a subset of LC8 binding partners , they highlight potential areas where more efficient methods are needed . The analysis of evolutionary information was also highly valuable to increase the biological relevance of potential functional sites , but also provided some unexpected results . LC8 is ubiquitous in all eukaryotes , presenting a highly conserved binding interface for its interaction partners , which is practically identical from Drosophila to human . One of the most well-characterized interaction partner of LC8 , the dynein intermediate chain is also conserved , at least in the animal kingdom . Therefore , it was surprising to find that most of the other known interaction partners of LC8 showed limited evolutionary conservation at the motif level . The overwhelming majority of both known and predicted motifs in the human proteome could not be traced beyond vertebrates . These results are in agreement with the evolutionary plasticity of linear motifs , as SLiMs can be introduced or eliminated by a few point mutations , or created ex-nihilo [56] . However , the lack of motif conservation was unexpected in the light of the strong evolutionary constraints acting upon LC8 . Nevertheless , signs of purifying selection could be detected for the binding motifs over smaller evolutionary distances . In a subset of cases , the SLiMPrints method [19] was able to detect that motif residues showed higher relative conservation compared to their flanking regions . However , this approach failed to highlight many validated binding motifs , either because of problems with alignments , a strong overall conservation or the lack of the sensitivity of the method . In our experience , a more efficient filtering criterion was the conservation of motif residues within mammalian species . This criterion was true for nearly all verified human binding motifs , but was not met by most of the random motif hits that were located within the evolutionary variable disordered regions . Therefore , it is a highly efficient filtering approach . Nevertheless , we understand that by filtering based on mammalian-level conservation comes with a risk of losing some instances that are conserved only in a smaller taxonomic unit within mammalia , e . g . in primates . The expanded interaction network of LC8 highlighted novel functions and drew attention to the potential role of LC8 in the Hippo pathway which controls organ size and cell growth by regulating cellular proliferation and apoptosis [57] . In our high confident hits , two putative LC8 binding partner families , the WWC and the angiomotin family were identified . Both families are tightly connected to the Hippo pathway and are known to be involved in its upstream regulation pathway in multiple ways [58] . WWC and AMOT family members can act either by activating the core kinase modules or by forming complexes with YAP/TAZ sequestering them at cell-cell junctions and preventing their nuclear access [59–64] . Previously , the interaction between WWC family members and LC8 has been established at the protein level [65] , but the binding motifs were identified only in WWC1/KIBRA [49] . In addition , a study of the protein interaction network of the mammalian Hippo pathway also revealed interaction between AMOT proteins and LC8 [66] , confirmed by additional large-scale studies [67 , 68] . A recent work established a connection between LC8 and the Hippo pathway in Drosophila [69] . In the present work , the LC8 binding sites were experimentally verified both in WWC and the AMOT protein families and were shown to be evolutionarily conserved . The verification of binding sites established a direct connection between LC8 and the Hippo pathway through the WWCs and AMOT family proteins , nevertheless , the exact role of these interactions in the regulation of Hippo pathway is still unclear . Recently , LC8 was suggested to promote dimerization of partially disordered proteins by stabilizing their coiled coil regions [25] . This model was based on detailed characterization of the complex formation of several scaffolding proteins , such as Swallow , dynein intermediate chain DYNC1|1 and GKAP [70] . However , there are still open questions regarding the regulation of binding of LC8 to its partners . In agreement with the general model for LC8 function , both families contain coiled coil ( CC ) regions and the dimerization of both AMOT [71 , 64] and WWC family members [72] might be necessary for their scaffolding activity . While the functional role of the CC region in the WWC family has not been explored , the CC regions of AMOT family members were shown to form interactions with various components of the Hippo pathway and influence their activity and localization [59 , 63 , 73 , 64] . Therefore , the potential stabilization of coiled coil regions driven by LC8 could also play important roles in regulation of the Hippo pathway . However , the LC8 function related to intracellular trafficking cannot be ruled out entirely either , as both KIBRA and AMOT proteins alternate between cytoplasmic and nuclear locations depending on various signals [39 , 74] . While LC8 might provide an important additional layer to the regulation of the Hippo pathway , further investigations are needed to understand how the interaction of LC8 affects WWC and AMOT proteins in the Hippo pathway and in other processes . In conclusion , the novel binding partners presented in this work provide exciting new opportunities to study how the binding of LC8 influences the stability and binding properties of various proteins . In addition , a general framework was also proposed here for the optimization of the linear motif filtering . While the protocol was applied specifically to LC8 and takes advantage of its relatively well-characterized interaction network with over 60 known linear motif instances , it can be applied in a similar way for other linear motif binding domains for which similar number of partners have been identified . In a more general sense , the novel motif filtering protocol further underlines that computational approaches can complement large-scale experimental studies of linear motif binding systems , and advance our understanding of how these relatively weak , transient interactions contribute to highly dynamic , complex regulatory processes in the cell .
We collected 76 LC8 binding motifs from the literature , however , upon further investigation we rejected 9 of them due to the lack of motif level evidence . The resulting list is given in S1 Table ( for details , see S1 Text ) . The two vertebrate isoforms of LC8 , DYNLL1 and DYNLL2 were not discriminated in this study . From the collected 67 binding motifs we generated a non-redundant set using the CD-HIT suite tool [75] . The remaining 62 motifs were aligned and used to construct a PSSM . The values in the PSSM matrix capture the preferences of each amino acid for over-representation or under-representation at each position in the binding motif . The elements of the PSSM ( Pi , j ) were expressed as the log-odds score of amino acid frequency in each position in the known binding partners divided by the background frequency: Pi , j=log ( Ai , jDi ) where Ai , j is the frequency of amino acid i at position j in the alignment of known binding partners , and Di is the background frequency of amino acid i . The background frequency was derived from the frequency of amino acids in the UniProt Eukaryotic proteome . As not every amino acid was present in each position in the known partners , we incorporated a pseudo-count correction to account for these zero occurrences . Pi , j=log ( Ai , j+B20m+BDi ) Where B is the pseudocount with a value of 5 , as suggested in previous works [76] , m is the number of sequences , and 20 is the number of amino acids . Additional sequential features were calculated for each known binding partner to identify the attributes of true binding events . The binding partners were analyzed using more than 20 different tools , to predict disorder content ( IUPred , PONDR VSL2 and DISOPRED3 ) [32 , 33 , 35] , the tendency to be involved in disordered binding regions ( ANCHOR , MoRFchibi and DISOPRED3-BR ) [35–37] , secondary structure and subcellular localization predictions . For each potential binding motif , position-based numerical values were averaged . The subcellular localization of a motif was predicted by a combination of methods , and only putative motifs with predicted intracellular localization were considered to be accessible for interaction with LC8 . A motif was considered to be intracellular if it contained no predicted signal sequence according to the SignalP predictor [77] . Proteins with a predicted signal sequence were further analysed using PHOBIUS [78] for transmembrane topology and the motif was categorized according to its result to either be in intracellular ( INT ) , transmembrane ( TM ) or extracellular space ( EXT ) . PFAM identifies conserved sequence families , which can be of different types , including domains , families , motifs , repeats and also intrinsically disordered domains [31] . Each motif was categorized as Domain or Family if it overlapped in at least one position with a respective PFAM annotation [31] . Secondary structure predictions were carried out using PSIPRED [38] . Proteins were checked with NCOILS to predict coiled-coil regions [41] . The calculated features can be found at http://gerdos . web . elte . hu/data/LC8/known_results . html . A protocol was generated to find the optimal filtering criteria based on these sequence features . Due to the limited number of positive examples , the full tree space was not explored , instead , optimal filtering methods and the best parameter settings were selected globally . For this , we defined two sets , one from the known human motifs , and one from 10 , 000 randomly selected peptides from the human proteome with PSSM scores above 0 . Various attributes were calculated for each motif in both sets . We sought to find attributes that can keep all or most of the known motifs while discarding the largest number of the randomly selected peptides . The optimization was guided by using the information gain theory derived from the Shannon entropy . I=Hp−pc1 ( Hc1 ) −pc2 ( Hc2 ) Where I is the information gain , Hp is the Shannon entropy of the parent group , c1 and c2 are the child groups on a specific split , and p is the probability of an attribute in the given group . During the entropy calculations , the number of occurrences with a specific attribute were normalized with the sum of the examples to weight the imbalanced sets . The information gain reaches its theoretical maximum in the ideal case , which results in a clean set , e . g . when a criterion is true for all entries in the known motif set . In these cases , the best attribute was defined as the one that filtered out the largest number of random peptides . We validated the filtering protocol using a 3-fold cross-validation . The set of known human interaction partners and their respective attributes were split into three equal , non-overlapping subsets . Two sets were used to calculate the new maximum points of the information gain for each attribute , and the elements of the remaining third set were used as a test set to evaluate the resulting filters . The number of correctly categorized examples were calculated by summing over all three possible test set . In order to compile evolutionary data on each protein having any putative motif according to our pipeline , we generated a dataset of orthologous sequences . These hits were obtained by running the GOPHER prediction algorithm with default settings against the QFO database [79 , 80] . Then , we constructed the multiple sequence alignments of orthologs for each protein using the MAFFT algorithm ( default parameters ) [81] . The ortholog sequences were classified into the most specific term using the five main evolutionary levels according to the UniProt taxonomic lineage: Mammalia , Vertebrata , Metazoa , Fungi and Eukarya . According to the multiple sequence alignments of orthologs , each aligned instance of the candidate LC8 binding motifs was analyzed in PSSM based conservation terms . For the evolutionary analysis at least two predicted orthologs were required at each level . An aligned motif was considered to be conserved if its PSSM score exceeded the cutoff value ( 3 . 3 ) . In the evolutionary filtering step a candidate motif was kept if it was conserved in at least 80% of the mammalian species . Motif conservation was also analyzed using the SLiMPrints method [19] . This method searches for regions with island-like conservation , i . e . regions that have high conservations relative to their flanking regions . A motif was considered to have an island-like conservation if it had at least 3 overlapping positions with a significant region ( p-value<0 . 05 ) detected by SLiMPrints . The method was applied to known human binding motifs of LC8 as well as predicted motifs . In order to study the evolutionary history of the structural organization of AMOT and WWC families , the domain annotation and coiled coil region predictions were retrieved from the InterPro resource ( version 63 ) [82] . The inter-species motif mapping of the known LC8 binding motifs of WWC and AMOT family members was carried out by using the canonical ELM definition and literary data . The definitions of the C-terminal class 3 PDZ-binding motif and the WW-binding motif of group I were obtained from the ELM database [5] . The first WW-binding motif of the AMOT family members was defined as LPxY [74] . In addition , within the orthologues of AMOTL2 the WW-binding motif of group I was defined as PPxF . The identified ortholog sequences were used to generate the multiple sequence alignment and phylogenetic tree of the families by applying the PhyML algorithm ( default settings ) [83] . To collect PPIs for LC8 from public databases we applied the PSICQUIC approach [43] . Known non-linear motif binding partners ( DYNLL1 , DYNLL2 and UBC ) were omitted . To remove redundant hits , the following four filtering steps were applied: 1 . For each partner and study only the direct interactions were kept , except when the given study had only non-direct interactions for that protein . 2 . The less informative ‘other affinity chromatography technology’ annotations were filtered out when the study had ‘TAP’ , ‘CO-IP’ or ‘pulldown’ annotation as well . 3 . Annotations with unknown association type were filtered out if another database had the same interaction from the same study with full annotation . 4 . Cases where the PubMed identifier could not be retrieved were filtered out if the given partner was annotated with the same detection method in an identifiable study . The process resulted in a collection of 559 interaction partners for LC8 . The partners came from 7 eukaryotic species and 23 virus strains . Finally , for subsequent analysis , the subset of human , mouse and rat data were merged , and this way 381 partners remained . The resulting interaction data is given in S2 Table . PSICQUIC gives the PSI-MI identifier for detection methods and association types . In the case of the detection methods , based on their meaning and position in the PSI-MI hierarchy , we ( i ) assigned the labels “biochemical or biophysical” , “PCA” and “other” to them; ( ii ) merged the detection method descriptions into 8 groups ( see the legend on S5A Fig ) . Similarly , we labelled the association types as “direct” or “non-direct” interaction . PSICQUIC doesn’t provide direct information about the scale of the study , therefore we queried each PubMed identifier in our set . Since the output was , again , redundant , we sorted the number of annotations by source , and took only the highest one . If this number was higher than 50 , we considered it as “high throughput” , otherwise as “low throughput” . The full-length His6-tagged LC8 ( DYNLL2; UniProt accession number Q96FJ2 ) was cloned into a pET21-derived pBH4 vector and expressed as described previously [84] . The 11-residue-long fragments of the predicted binding partners were synthesized using solid-phase technique by GenoSphere Ltd . SPR measurements were performed on a ProteOn XPR36 ( Bio-Rad ) instrument equipped with HTG sensor chip ( Bio-Rad ) . The sensor chip was activated for 300 s with 150 μl 10 mM NiCl2 solution followed by a washing step for 300 s with the running buffer containing 20 mM Hepes , 150 mM NaCl , 0 . 05% Tween-20 , 0 . 1 mM TCEP , 50 μM EDTA , pH 7 . 5 buffer . The immobilization of His6-tagged LC8 was performed on the activated Ni2+-NTA surface at three different densities ( 1200 RU , 900 RU , 600 RU ) . The putative binding peptides were injected onto the chip at five different concentrations simultaneously at a flow rate of 60 μl ∙ min-1 for 400 s , while the dissociation of the peptides was recorded for 600 s . In the sixth analyte channel , running buffer was injected for double referencing . The double referenced data were global fitted to the 1: 1 Langmuir model using the ProteOn software . The presented Kd values were obtained from the mean of the kinetic and the equilibrium analysis delivered Kd values , and the standard deviations were calculated from three individual Kd values . | Fine-tuning of many cellular processes relies on weak , transient protein-protein interactions . Such interactions often involve compact functional modules , called short linear motifs ( SLiMs ) that can bind to specific globular domains . SLiM-mediated interactions can carry out diverse molecular functions by targeting proteins to specific cellular locations , regulating the activity and binding preferences of proteins , or aiding the assembly of macromolecular complexes . The key to the function of SLiMs is their small size and highly flexible nature . At the same time , these properties make their experimental identification challenging . Consequently , only a small portion of SLiM-mediated interactions is currently known . This underlies the importance of novel computational methods that can reliably identify candidate sites involved in binding to linear motif binding domains . Here we present a novel bioinformatic approach that efficiently predicts new binding partners for SLiM-binding domains . We applied this method to the dynein light chain LC8 , a protein that was already known to bind many partners in a wide range of organisms . With this method , we not only significantly expanded the interaction network of LC8 , but also identified a novel function of LC8 in a highly important pathway controlling organ size in animals . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"protein",
"interactions",
"protein",
"interaction",
"networks",
"dyneins",
"molecular",
"motors",
"network",
"analysis",
"sequence",
"motif",
"analysis",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"computer",
"and",
"information",
"sciences",
"ne... | 2017 | Novel linear motif filtering protocol reveals the role of the LC8 dynein light chain in the Hippo pathway |
Mathematical models have been used successfully at diverse scales of biological organization , ranging from ecology and population dynamics to stochastic reaction events occurring between individual molecules in single cells . Generally , many biological processes unfold across multiple scales , with mutations being the best studied example of how stochasticity at the molecular scale can influence outcomes at the population scale . In many other contexts , however , an analogous link between micro- and macro-scale remains elusive , primarily due to the challenges involved in setting up and analyzing multi-scale models . Here , we employ such a model to investigate how stochasticity propagates from individual biochemical reaction events in the bacterial innate immune system to the ecology of bacteria and bacterial viruses . We show analytically how the dynamics of bacterial populations are shaped by the activities of immunity-conferring enzymes in single cells and how the ecological consequences imply optimal bacterial defense strategies against viruses . Our results suggest that bacterial populations in the presence of viruses can either optimize their initial growth rate or their population size , with the first strategy favoring simple immunity featuring a single restriction modification system and the second strategy favoring complex bacterial innate immunity featuring several simultaneously active restriction modification systems .
One of the major challenges in biology is to understand how interactions between individual molecules shape living organisms and ultimately give rise to emergent behaviors at the level of populations or even ecosystems . At the very bottom of this hierarchy , inside single cells , interacting biomolecules such as DNA or proteins are often present in small numbers , giving rise to intrinsic stochasticity of individual reaction events [1 , 2] . As a result , genetically identical organisms occupying identical environments can express different phenotypes [3 , 4] and make different decisions when presented with identical environmental cues [5 , 6] . This molecular noise is known to be the cause of biologically and medically important traits of bacteria such as persistence in response to antibiotics [7 , 8] and competence during acquisition of heterologous DNA [9] . However , while its causes and consequences are relatively well-studied at the organismal level [10–13] , how molecular noise propagates to higher scales of biological organization to affect the ecology and evolution of organisms remains mostly unknown [4] . Recently it has been shown that ecosystems can follow surprisingly deterministic trajectories despite the prevalence of stochastic events [14 , 15] , yet these trajectories could themselves be strongly influenced by molecular noise . Thus , the extent to which ecological interactions are affected by molecular noise , and the extent to which these ecological consequences feed back to reshape individual traits , remain to be explored . Perhaps the most prevalent biological systems in which molecular noise is thought to play an important role are restriction-modification ( RM ) systems [16] . Present in nearly all prokaryotic genomes [17] , RM systems are a highly diverse class of genetic elements . They have been shown to play multiple roles in bacteria as well as archaea , including regulation of genetic flux [18] and stabilization of mobile genetic elements [19] , but have most frequently been described as primitive innate immune systems due to their ability to protect their hosts from bacterial viruses [20] . When a virus ( bacteriophage or phage ) infects a bacterium carrying a RM system , the DNA of the phage gets cleaved with a very high probability , thus aborting the infection . With a very small probability , however , the phage can escape and become immune to restriction by that specific RM system through epigenetic modification , leading to its spread and potentially death of the whole bacterial population in absence of alternative mechanisms of phage resistance [21 , 22] . Thus , in the context of RM systems , molecular noise occurring at the level of individual bacteria can have profound ecological and evolutionary consequences . Because RM systems are ultimately based on only two very well characterized enzymatic activities ( restriction and modification ) [23] , they represent a simple and tractable biological system in which we can investigate propagation of effects of molecular noise across different scales of biological organization . Here , we mathematically model the action of RM systems from individual molecular events occurring inside a single cell , through individual bacteria competing in a population , to interactions between populations of bacteria and phages in a simple ecological setting , as shown in Fig 1 . We demonstrate that , by imposing a tradeoff between the efficiency and cost of immunity , molecular noise in RM systems occurring at the level of individual bacteria has consequences that propagate all the way up to the ecological scale , and that the ecological consequences in turn imply the existence of optimal bacterial defense strategies against phages .
RM systems consist of two enzymes , a restriction-endonuclease R , that recognizes and cuts specific DNA sequences ( restriction sites ) , and a methyl-transferase M , that recognizes the same DNA sequences and ensures that only invading phage DNA can be cut by the endonuclease while the bacterial DNA remains methylated and protected . However , since chemical reactions occur stochastically , RM systems can produce errors and fully methylate invading phage DNA before it is cut and degraded ( phage escape , typically occurring with a probability ranging between 10−2 and 10−8 ) [24 , 25] . Similarly , it is possible that newly replicated restriction sites on the bacterial DNA , which are originally unmethylated , are accidentally cleaved instead of methylated ( self-restriction ) [26] . Inside a single cell , the probability of such self-restriction events depends on the total activity , r , of all restriction endonuclease molecules R , the total activity , m , of all methylase molecules M , as well as the bacterial replication rate λ , since λ determines the rate at which new unmethylated restriction sites are generated . To investigate how self-restriction depends on these parameters , we model the corresponding biochemical reactions at each individual restriction site on the bacterial DNA with the stochastic reaction network displayed in Fig 2a ( see S1 Appendix Section S . 1 ) . The time τS until the first self-restriction event in a given cell—i . e . , until that cell’s death or substantial reduction in growth rate—can be obtained as the time when the first restriction site is cut , that is as τS = mini∈{1 , … , NS} τi , for bacterial DNA with NS restriction sites , where τi , i = 1 , … , NS are the waiting times for cutting events at individual sites . It can be shown that all τi follow a phase-type distribution ( see [27] and Fig 2b and 2c ) : f ( τ i ) = p Q exp ( B τ i ) c 1 , where B = [ - ( r + m ) m 0 λ / 2 - ( m + λ / 2 ) m 0 λ - λ ] and c 1 = [ r 0 0 ] , ( 1 ) with pQ = [p0 p1 p2] being the initial methylation configuration , i . e . , the proportion of restriction sites that are unmethylated ( p0 ) , hemi-methylated ( p1 ) and doubly-methylated ( p2 ) ; see S1 Appendix Section S . 1 . Eq [1] allows us to derive the expected time until self-restriction of a single site as E [ τ i ] = - p Q B - 1 1 , where 1 = [ 1 1 1 ] ⊤ ; ( 2 ) more generally , Fig 2b shows how the distribution of waiting times depends on the restriction rate r ( increasing the probability of the site getting cut when it is unmethylated ) and the magnitude of m relative to λ ( which decreases the probability that the site is unmethylated in the first place ) . Fig 2c shows that time to self-restriction at a single site depends essentially on an unknown quantity , the methylation configuration pQ . We will now proceed to show that when we consider an exponentially growing population of bacterial cells , the configuration pQ can no longer be freely chosen , and has to be determined self-consistently instead . Intuitively , this is because when the bacterial population is in steady-state growth , new unmethylated sites are constantly replenished by replication , while cells with more unmethylated sites are simultaneously and preferentially being removed , as illustrated in Fig 3a and required by Eq [2] . These two forces , generation of new unmethylated sites and their preferential removal , will push any initial pQ towards a unique steady state equilibrium . Mathematically , assuming that the methylation dynamics in all cells are equilibrated and that cells cannot be distinguished , the internal methylation configuration of any randomly chosen cell at any time during growth of the population can be derived from the quasi-stationary distribution pQSD ( r , m ) of the individual-site methylation process in Fig 2a ( see S1 Appendix Section S . 1 ) . pQSD ( r , m ) is the equilibrium distribution of the stochastic process conditional on it not having reached the absorbing state where the DNA is cut and the cell has died ( Fig 3a ) ; in short , methylation and growth equilibrate “in all directions except the one leading towards self-restriction” . Then , setting pQ = pQSD ( r , m ) in Eq [1] reduces the phase-type distribution f ( τi ) for the time τi until self-restriction at an individual restriction site to a single exponential , implying further that the waiting time τS = mini∈{1 , … , NS} τi until self-restriction of any site in the cell is also exponentially distributed . Consequently , we are led to the main result of this section: growth with self-restriction can be rigorously modeled at the population level with a Markov birth-death process for which the expected population size n ( t ) follows a simple ordinary differential equation d d t n ( t ) = ( λ - μ ( r , m , λ ) ) n ( t ) = λ e n ( t ) , ( 3 ) where λe ( r , m , λ ) = λ − μ ( r , m , λ ) is the effective growth rate and μ ( r , m , λ ) is the rate of self-restriction , defined as the inverse of the per-cell expected waiting time until self-restriction μ ( r , m , λ ) = E [ τ S ] - 1 = N S - p QSD B - 1 1 = - γ 1 N S , ( 4 ) with γ1 being the largest eigenvalue of B ( an explicit stochastic simulation validating this analytical result is provided in the S1 Appendix Section S . 2 ) . Eq [4] allows us to straightforwardly evaluate the reduction in the population growth rate due to random self-restriction events in single cells for any given pair of enzyme activities , r and m . To study possible qualitative effects of self-restriction , we explore in Fig 3b a wide range of enzyme activities for a system with NS = 5 restriction sites ( chosen , for illustration purposes , significantly smaller than the typical number of sites recognized by real RM systems ) . We find that the main determinant of self-restriction is the activity m of the methyl-transferase and that the effects of molecular noise can be suppressed by sufficiently increasing m . Furthermore , so long as m is large enough such that unmethylated restriction sites are only rarely available , μ ( r , m , λ ) lies on a large plateau of low self-restriction and changes only little with r and m , suggesting that stochastic fluctuations in enzyme activities would only have minor consequences for the population , especially when they are positively correlated , as would be the case if R and M enzymes were expressed from the same operon ( S1 Appendix Section S . 3 ) . The ( r , m ) plane in Fig 3b contains a transition region that separates the large plateau with low self-restriction from the plateau where self-restriction is severe enough to stop the population growth altogether . We have chosen our reference ( red ) parameter values ( rref , mref ) to lie in this transition region , and explored the regime with an e-fold higher rates ( “large r & m” , indicated by green ) , and with 2e-fold lower rates ( “small r & m” , indicated by blue ) in Fig 3b and 3c . The comparison of these three regimes in Fig 3c is most clear when the effective growth rate is shown as a function of λ , the rate at which the cells , and thus the restriction sites , are replicated . In the “small r & m” regime , self-restriction is so infrequent that it can easily be outgrown by replication ( except at very low λ ) . In the “large r & m” regime , m is sufficiently high to keep the restriction sites protected and thus self-restriction is rare , except at extremely large λ , where the green curve falls below the blue curve . In the reference regime , r is too large and m not high enough to protect , so self-restriction can not be “outgrown”; effective growth thus falls significantly below λ . Our numerical analyses further show that the self-restriction rate μ ( r , m , λ ) grows faster-than-linearly with λ ( S1 Appendix Section S . 1 ) , causing the effective population growth to slow down and ultimately drop to zero at high enough λ . We end this section by highlighting a non-trivial interaction between the single-cell and population-scale processes . While increasing the activity r of the endonuclease always decreases the effective growth rate of the population due to self-restriction , the effect can be smaller than expected from the single-cell analysis ( dashed lines in Fig 3c ) . This is because high values of r feed back through the population scale to bias the steady-state distribution of methylation configurations away from cells with lots of unmethylated sites , as shown in Fig 3a , making self-restriction less likely . Implicit feedback effects of this type frequently give rise to complex dynamics in multi-scale models . RM systems lower the growth rate of the population due to self restriction , especially when the activity m of the methyl-transferase is small . Upon infection by a phage , however , small values of m are advantageous , making it less likely that the unmethylated phage DNA will get methylated and escape the immune system before it can be cut by the restriction endonuclease . Assuming that all restriction sites are identical and independent , the probability of phage escape can be calculated [28] as p V ( r , m ) = ( m r + m ) N V , ( 5 ) where NV is the number of restriction sites on the phage DNA . From Eq [5] it is straightforward to see that pV ( r , m ) is monotonically increasing in m and decreasing in r . One might therefore expect that the balance between avoiding self-restriction that favors high m , Eq [4] , and minimizing phage escape that favors low m , Eq [5] , would impose a tradeoff and thus lead to an optimal value of m . However , this is not the case , because the phage escape probability pV ( r , m ) and the population self-restriction rate μ ( r , m , λ ) can both approach zero so long as r and m both increase to infinity but r does so faster . While mathematically possible , this limit is , however , not biologically relevant: large enzyme expression levels should incur a cost ( metabolic or due to toxicity presumably caused by non-specific protein-DNA interactions in the case of RM systems ) for the cells [29 , 30] , which we sought to incorporate into our model by including a growth rate penalty proportional to the activity of restriction and methylation enzymes , i . e . , λe ( r , m , λ ) = λ − μ ( r , m , λ ) − crr − cmm . Interestingly , it can be verified that our reasoning is valid only because two subsequent demethylation events need to occur to create a restriction-susceptible site on the bacterial DNA ( S1 Appendix Section S . 1 ) . If hemi-methylated sites could be recognized by the restriction endonuclease , or if both methyl groups could be lost in a single event , our initial expectation about the existence of the tradeoff would be correct , and a particular choice of r and m values would simultaneously minimize the phage escape and self-restriction , even in the absence of the expression cost for R and M . Our model can be generalized to multiple coexisting RM systems that recognize different restriction sites and operate in parallel , as is often observed for bacteria in the wild [17] . This provides increased protection from phages since the phage has to escape all RM systems to infect successfully . However , multiple RM systems also imply that the bacteria either have to pay higher expression and self-restriction costs or that they have to re-balance the expression levels of the enzymes such that lower self-restriction rates per RM system are obtained with the same overall enzyme activity . Allowing bacteria to have multiple RM systems , but assuming for the sake of simplicity that these systems are all equivalent in terms of enzyme activities and number of recognition sites , we obtain the phage escape probability for k RM systems as p V ( r , m , k ) = ( m r + m ) k · N V , with the corresponding growth rate being λ e ( r , m , k , λ ) = λ - k · ( μ ( r , m , λ ) + c r r + c m m ) . ( 6 ) What is the combined effect of phage escape and self-restriction in simple bacteria-phage ecologies ? To investigate this question , we first extended an established deterministic model of bacteria-phage ecology [31] to track the population dynamics of bacteria with and without RM systems and both susceptible and methylated phages ( see S1 Appendix Section S . 4 . 2 ) . By numerically integrating this population model for more than a million parameter combinations for the activity of restriction ( r ) and methylation ( m ) enzymes , we find that whether or not phages will ultimately take over the population depends on the ecological parameters ( e . g . phage adsorption rate , rate of spontaneous phage inactivation , etc . ) but is completely independent of RM system efficiency . This result might seem surprising at a first glance , but closer analysis reveals that for efficient RM systems the phage population reaches levels that are so small that they should be considered as extinction from a biological perspective . Nevertheless , even in these cases methylated phage eventually takes over the population . This is because phages cannot go extinct in the mathematical sense and the phage population always remains at levels that are strictly larger than zero if ecology models based on ordinary differential equations are used for the analysis . While this clearly limits the practical relevance of such models , the finding that RM systems apparently cannot provide long term protection if phage escape probability and phage population size remain strictly larger than zero is still interesting since it suggests that the task of RM systems cannot be to prevent phage escape but only to delay it as much as possible to give bacteria enough time to develop alternative mechanisms of phage resistance through genetic mutations [21 , 22] . In line with the above reasoning , we decided to represent phage escape events stochastically and to focus in more detail on how RM systems impact exponentially growing bacterial populations until the first phage escape event . To explore this question , we formulated several efficiency measures that quantify how RM systems can help bacterial populations before the first phage escape event: Here we will show that questions ( i ) - ( iii ) can be answered rigorously if we assume that the size of the phage population remains approximately constant until the first phage escape event . An example of an ecological scenario where this assumption is realistic is that of bacteria colonizing a phage-dominated environment in which the number of phages is much larger than the number of bacteria such that the reduction in the phage population size due to unsuccessful infections is negligible . More generally , any ecological scenario in which the phage population size is for some reason in equilibrium at least until the first phage escapes on a bacterium carrying a RM system , will fulfill this assumption . Mathematically , we consider a bacterial population of initial size n0 trying to colonize an environment containing a phage population of size v . As we have shown before , the bacterial population will initially grow exponentially at rate λe until the time τp at which the first phage escape event occurs . Interpreting these events as random , the crucial unknown is therefore τp , the random time to first phage escape , characterized by its probability distribution , f ( τp ) , which we find to be given by ( see S1 Appendix Section S . 4 . 3 ) : f ( τ p ) = ρ v n 0 p V exp ( ρ v n 0 ( 1 - e λ e τ p ) p V λ e + λ e τ p ) . ( 7 ) Specific examples of this probability distribution are visualized in Fig I in S1 Appendix . In general , larger values of any of the parameters ρ , v , n0 , pV or λe will imply that phage escape is likely to occur at earlier times . Importantly , the waiting time distribution until first phage escape , f ( τp ) , allows us to analytically answer questions ( i ) - ( iii ) , as summarized in Table 1 ( see S1 Appendix Section S . 4 . 3 ) . We note that despite the somewhat intricate form of f ( τp ) the “bacterial performance” metrics derived for all three efficiency criteria turn out to be remarkably simple , depending only on some of the parameters that define f ( τp ) . More concretely , by examining these metrics , we can make two important observations: First , assuming a fixed mutation rate cmut , expressions for bacterial performance in Table 1 are functions of λe and pV , which depend solely on the restriction rate r , the methylation rate m , and the number of concurrently active RM systems , k . This means that optimal bacterial strategies at the ecological level can be found mathematically —and possibly tuned evolutionarily— by adjusting the three parameters , r , m , and k , defined at the single-cell level . Second , despite the dependence of the time to phage escape on the initial population size n0 , the performance of the bacterial population is independent of n0 according to all criteria . This has the important consequence that optimal defense strategies against phages do not depend on the size of the bacterial population and that there exists a single unique best defense strategy that is constant in time: if phage escape has not happened until a certain time during which the bacterial population has grown to a new size , the same defense strategy continues to be optimal with the initial size taken to be the new size , with no need to re-balance the activity levels r and m of the RM enzymes , or the number of RM systems , k . For cases ( ii ) and ( iii ) , we further observe that the results are independent of the effective growth rate λe . Faster growth leads to quicker increases in the probability that immunity conferring mutations happen but this is exactly compensated by the increase in probability of a phage escape event . An in-depth study of the consequences and implications of case ( i ) is presented in the following section while questions ( ii ) and ( iii ) are treated in the S1 Appendix ( Section S . 4 . 4 ) . Taken together , these results show how the mathematical framework developed in this paper can be readily adjusted to analyze trade-offs between the efficiency and cost of immunity in different ecological contexts . For the concrete scenarios that we considered here , we find that ( i ) and ( iii ) imply overall similar results in which bacteria can relatively directly trade cost for efficiency and vice versa . The results for ( ii ) , however , are qualitatively different since ( ii ) implies that increasing the cost beyond a certain point provides only diminishing returns in terms of efficiency ( Fig J in S1 Appendix ) . We conclude that analyzing such trade-offs in practice will require careful consideration of the efficiency criteria according to which bacteria might have been shaped in a particular ecological context . Reversely , different trade-offs and optimal strategies for different efficiency criteria imply that the criterion on which evolution might have been operating in a given ecological context can , in principle , be reverse engineered from observations of phage defense strategies that are employed by the bacteria . Can bacteria tune the single-cell parameters over evolutionary timescales in order to minimize the cost of RM systems , that is maximize the growth rate λe ( r , m , k ) , while also maximizing their efficiency , quantified here as the increase in population size before the first phage escape , n s ( r , m , k ) : = E [ n ( τ p ) ] - n 0 in ( i ) , that is determined by λe ( r , m , k ) /pV ( r , m , k ) ? Eq [6] and Table 1 assert that cost and efficiency are necessarily in a tradeoff and cannot be optimized simultaneously . This tradeoff is the first key result of the section . With no single optimum possible , we look instead for Pareto-optimal parameter combinations , ( r , m , k ) , i . e . , solutions for which λe cannot be further increased without reducing ns and vice versa [32 , 33] . Different Pareto-optimal solutions trace out a “front” in the plot of λe vs ns in Fig 4a that jointly maximizes growth rates and population sizes to the extent possible . Points in the interior of the front are sub-optimal and could be improved by adjusting parameter values , while points beyond the front are inaccessible to any bacterial population . Which Pareto-optimal solution ultimately emerges as an evolutionary stable strategy depends on the actual bacterial and phage species considered as well as their biological context . Rather than focusing on specific examples , we next establish several general results of our analysis , contrasting in particular “fast growth” bacterial strategies that maximize λe with “large size” strategies that maximize ns . We start by examining in Fig 4b the optimal enzyme activities , mopt and ropt , along the Pareto fronts . For the “large size” regime at low λe , the bacterial population primarily needs to defend against phage escape , favoring low m and high r , even at the cost of self-restriction . As we move towards the “fast growth” regime , r can drop to decrease the cost , but m must increase to protect against self-restriction , until maximal mopt is reached . For even higher λe , it is optimal to “shut down” the RM systems altogether to save on the cost , by tuning r and m simultaneously to zero . Numerical analysis ( S1 Appendix S . 4 . 5 ) reveals that along the Pareto front of Fig 4a , the total cost of running the RM systems varies in precise inverse linear relationship with λe . Pareto-optimal solutions are further characterized by the fact that the reduction in growth rate , λ − λe , is split equally between the cost of running RM systems , c ( r + m ) , and self-restriction . If this were not the case and the cost were larger ( or smaller ) than self-restriction cost to growth , cells could always down- ( or up- ) regulate the RM system activity to trade cost for self-restriction and obtain an overall smaller total growth reduction . This universal equality of cost of running RM systems and self-restriction at optimality is the second key result of the section . A detailed examination of the Pareto front in Fig 4a reveals a striking shift in the structure of optimal solutions as we move from “fast growth” to “large size” regime . In situations where fast growth is favored , we observe that a single RM system ( k = 1 ) is optimal . In contrast , large bacterial population sizes favor kopt > 1 RM systems , with the optimal number , kopt , set by the costs , cm and cr , of operating the RM systems . These results are quantitatively robust to changes in replication rate , λ , as shown in Fig 4c , where Pareto fronts for different λ are nearly rescaled versions of each other . These results are also qualitatively robust to changes in the cost c = cr = cm so long as the cost is nonzero , as shown in Fig 4d . Establishing that “fast growth” regime favors simple innate immunity with a single RM system while “large size” regime favors complex innate immunity with multiple RM systems is the third key result of this section . This result can be understood intuitively by considering under what conditions , if any , multiple RM systems could be optimal at “fast growth” . If costs for R and M enzymes are vanishingly small , a single RM system can provide arbitrarily good protection , as we showed previously . If the costs are not vanishingly small , multiple RM systems must be more costly than a single system at comparable phage escape and self-restriction rates: to keep self-restriction constant with k RM systems , not only does the cell require k times more M molecules than at k = 1 , but their individual activities need to be higher as well , leading to a higher cost for M and thus a lower effective growth rate; thus , k > 1 cannot be optimal for “fast growth” and can only be tolerated in the “large size” regime where protection from phages is more important than fast growth . Lastly , we sought to put our results into perspective by relating them to a typical E . coli strain . Recent measurements [26] quantified the self-restriction rate in a bacterial population with the EcoRI system replicating at λ = 0 . 017 min−1 to be around μ ≈ 10−3 min−1 . The cost of RM systems was not detectable in WT strain but could be detected in strains overexpressing M enzymes . Treating the cost c as unknown and assuming that E . coli is Pareto-optimal in light of criterion ( i ) in Table 1 ( black dots in Fig 4c and 4d ) , would lead us to predict the following parameter values for the RM systems: cost c ≈ 3 . 7 ⋅ 10−7 , enzyme activities r ≈ 1 . 2 ⋅ 103 min−1 , m ≈ 1 . 5 ⋅ 102 min−1 , with the optimal number of RM systems being at the boundary between k = 1 and k = 2 . Clearly , this prediction depends on the chosen measure of the efficiency of RM systems , which is determined by the considered ecological scenario and the particular objective that bacteria have in this scenario . Consequently , the concrete numbers presented here should not be understood as general results , but rather as a demonstration of how our framework can be used to calculate optimal bacterial strategies given different modeling assumptions about the phage-bacteria ecology .
Despite the ubiquity of RM systems in prokaryotic genomes [17] , basic ecological and evolutionary aspects of these otherwise simple genetic elements are poorly understood [20] . Although RM systems have been discovered more than six decades ago due to their ability to protect bacteria from phage [34] and this is often assumed to be their main function [35] , only a few experimental studies focused on the ecological and evolutionary dynamics of interactions between RM systems and phage [36 , 37] . Similarly , effects of RM systems on their host bacteria , such as their cost in individual bacteria due to self-restriction , began to be addressed quantitatively only recently [26 , 38] . In this work , we bridged these two scales using mathematical modeling . Our model captures the stochastic nature of RM systems originating at the level of interacting molecules in individual bacteria and extends it all the way to the dynamics of interactions between bacterial and phage populations . Using this approach , we analytically described tradeoffs between the cost and the efficiency in different ecological contexts of immunity conferred by RM systems . The existence of such tradeoffs was previously indicated by quantitative single-cell experiments with two RM systems isolated from E . coli [26] and has since then been reported in the context of other RM systems [39] . We used our mathematical framework to quantify these tradeoffs and to study their ecological consequences , as well as the implications that these consequences have for optimally tuning the R and M enzymatic activities at the level of single cells . Our results for different ecological scenarios suggest that we should expect observed expression levels and enzymatic activities of naturally occurring RM systems to represent adaptations to specific environmental pressures . Such “tuning” of expression levels towards optimality has previously been directly experimentally shown in different molecular systems [29] . The expression levels of both R and M should be readily tunable by mutations in the often complex gene-regulatory regions [40] . With optimal bacterial defense strategies depending on the ecological scenario and the particular objective of the bacteria ( see S1 Appendix Section S . 4 and Table 1 ) , making general predictions on R and M expression levels or numbers of concurrently active RM systems that we should expect to find in bacteria in the wild is difficult . However , we want to highlight that , in a given context , assuming optimality of the bacterial defense strategy allows one to make clear and quantitative predictions about enzymatic activities and the number of RM systems , and improving these predictions to take into account more relevant biological detail ( if needed and known ) remains only a technical , rather than conceptual challenge . Second , for the ecological scenario that we investigated in detail in this paper , parameter values measured for an E . coli RM system put optimal solutions into a regime that permits a large variation in the optimal number of RM systems , between one to six , with relatively small changes in the effective growth rate . This observation allows us to advance the following hypothesis: the number of RM systems in different bacterial strains and species is not a historical contingency , but an evolutionary adaptation to different ecological niches . In other words , the tradeoff between the cost and the efficiency of immunity can be partially alleviated in bacteria employing multiple RM systems . It is therefore interesting to note that many bacterial species carry multiple RM systems and the number of RM systems varies significantly among bacteria with different genome sizes and lifestyles [16 , 17] . Our results indicate that different numbers of RM systems would be optimal in populations under different selection pressures ( phage predation/resource limitation ) . The analytical model presented here makes several simplifying assumptions . First , we consider only interactions between a single species of bacteria and a single species of phage . In natural environments , many bacterial and phage species interact and this diversity will certainly impact the resulting ecological end evolutionary dynamics [36 , 41–43] . Second , we assumed the key parameters such as the numbers of restriction sites in bacterial and phage genomes to be constant in time and thus disregarded the long-term evolutionary dynamics . Bioinformatic studies have shown that many bacteria and phage avoid using restriction sites in their genomes [44–46] . Restriction site avoidance can represent an adaptive mechanism for increasing the probability of escape in phages [45 , 47] and decreasing the probability of self-restriction in bacteria [26 , 48] . The stochastic nature of RM systems observed at the level of individual cells is thus likely to critically shape the ecological and evolutionary dynamics of interactions between bacteria , RM systems and phage . | Mathematical understanding of how randomness at the molecular scale , also known as molecular noise , ultimately affects the fate of organisms and whole populations is widely recognized as a challenging problem in multi-scale modeling . Here , we develop an analytical framework for analyzing how the randomness of individual reaction events in single cells propagates to higher levels of biological organization and affects population and ecology scale dynamics . We deploy our mathematical results to study an example from the ecology of bacteria and bacteriophage viruses . Bacteria defend themselves against viruses by a simple innate immune system composed of a pair of enzymes . Due to molecular noise , however , viruses sometimes escape this immunity , causing bacterial populations to plummet . Noise can also cause the immune system to turn against its own host . By analyzing how such costs and benefits of bacterial immunity balance at the ecological level , we predict the optimal parameters for bacterial innate immune systems . While the focus of this work is on bacteria-phage ecologies , we expect that our results will generally help to better understand population and ecology scale consequences of stochastic cell fate decisions in diverse biological domains ranging from cell differentiation in developmental biology to studies of the microbiome and consequences of stochasticity in apoptotic responses for anti-cancer therapies . | [
"Abstract",
"Introduction",
"Methods",
"and",
"results",
"Discussion"
] | [
"bacteriology",
"organismal",
"evolution",
"innate",
"immune",
"system",
"ecology",
"and",
"environmental",
"sciences",
"medicine",
"and",
"health",
"sciences",
"bacteriophages",
"immunology",
"microbiology",
"viruses",
"methylation",
"microbial",
"evolution",
"theoretical"... | 2019 | Molecular noise of innate immunity shapes bacteria-phage ecologies |
Trachoma is a disease that can lead to visual impairment and ultimately blindness . Previous estimates of health losses from trachoma using the Global Burden of Disease methodology have not , however , included the stage prior to visual impairment . We estimated the burden of all stages of trachoma in South Sudan and assessed the uncertainty associated with the severity and duration of stages of trachoma prior to full blindness . The prevalence of trachoma with normal vision , low vision and blindness in the Republic of South Sudan has been estimated previously . These estimates were used to model the incidence and duration of the different stages employing DISMOD II . Different assumptions about disability weights and duration were used to estimate the Years Lived with Disability ( YLD ) . We have estimated the total burden of trachoma in South Sudan to be between 136 , 562 and 163 , 695 YLD and trichiasis with normal vision contributes between 5% and 21% of the total depending on the disability weight applied . Women experience more of this burden than men . The sensitivity of the results to different assumptions about the disability weights is partly dependent upon the assumed duration of the different disease states . A better understanding of the natural history of trachoma is critical for a more accurate burden estimate .
Trachoma is a neglected tropical disease that is endemic in the Republic of South Sudan and more than 50 other countries in the world [1] . Globally , trachoma is the leading infectious cause of blindness and the eighth most common blinding disease [2] . When susceptible individuals come into contact with the bacterium Chlamydia trachomatis , they may become infected . Recurrent re-infection by the bacterium can eventually lead to the repositioning of the eyelashes back towards the cornea , a condition known as trichiasis . Without treatment this state eventually begins to impair the individual's vision and ultimately leads to blindness . Many neglected tropical diseases cause years of suffering . When the Global Burden of Disease methodology is used to assess the public health importance of chronic conditions such as trachoma , the resulting Disability Adjusted Life Years ( DALYs ) estimates are dominated by the estimated years lived with disability ( YLD ) . The burden estimates will thus be sensitive to the disability weights applied to different health states . While this is an issue for all nonfatal conditions , it can be a particularly important issue for conditions that are highly prevalent with lower levels of severity [3] . This is because small absolute changes to the already small disability weights can make proportionately large differences to the overall estimate . Uncertainty about disability weights may therefore be most consequential where chronic , low severity conditions exist in large parts of the populations . This describes the case of trachoma in South Sudan . Previous assessments of the global burden of trachoma have not included health losses from stages prior to visual impairment . Prior to the loss of visual acuity , however , trichiasis is associated with pain , photophobia and sensitivity to smoke and dust and therefore reduced capabilities in everyday life [4] . Frick et al [5] , for example posited that trichiasis with normal vision may result in economic burden comparable to trachomatous low vision and one study has suggested that the inclusion of trichiasis with normal vision in the burden estimate could add up to 50% to the total burden of trachoma [6] . Between 2001 and 2005 Ngondi et al [7] collected data on visual acuity in the Mankien district providing the most reliable estimates of the prevalence of trichiasis with normal vision , low vision and blinding available for South Sudan . According to these observations and extrapolating to the whole of South Sudan , it is clear that many of those with trichiasis are not visually impaired and have thus not been included in the attempts to estimate the burden of trachoma . The first aim of this study is to determine whether leaving out the least severe end of the disease spectrum is likely to make a material difference to the estimate of disease burden . A secondary aim is to employ assumptions derived from previous studies and empirical sources to assess the sensitivity of the YLD estimates of trichiasis with low vision and trichiasis with normal vision to uncertainty in the natural history model in terms of both the duration of different states of trichiasis and their disability weights .
The detailed methods for the population based trachoma prevalence surveys have been described elsewhere [7] . In brief , surveys were conducted in ten districts in South Sudan between September 2001 and May 2005 . Using a two-stage cluster random sample survey design a total of 23 , 139 people were examined for trachoma signs using the WHO simplified grading system [8] . Visual acuity testing was conducted in Makiem payam district to assess distribution of vision status in people presenting with trichiasis . Three stages of vision loss were recorded according to the level of presenting visual acuity ( Table 1 ) . A two-stage cluster sampling method was used to select the households for visual acuity ( VA ) testing . Villages were selected in the first stage and households in the second . A total of 3 , 567 present members of the selected household were tested for VA . VA testing was conducted using the Snellen E chart at 6 meters . Those with VA<6/60 , were evaluated with the Snellen chart at 3 meters and those with VA<3/60 were evaluated by counting fingers , hand movement and light perception as appropriate . An ordinal logistic regression model to the observed VA data was used to explore the age and sex distribution of the three categories of vision status in participants with trichiasis: normal vision; low vision; and blindness . All participants in whom trichiasis and cataract was identified were excluded from analysis; however , it was not possible to adjust for visual impairment due to other causes such as refractive errors [9] , [10] . Age-grouped and sex specific prevalence of trichiasis was estimated from the cross-sectional surveys and a logistic curve fitted to smooth the prevalence estimates across age groups . The prevalence of vision status in the sample population was calculated by multiplying the predicted probability of each vision status with the overall prevalence of trichiasis ( 9 ) . Abridged life tables for Sudan for the year 2001 were obtained from the WHO Statistical Information System ( WHOSIS ) for males and females separately [11] . Demographic estimates for South Sudan are based on model life tables because vital registration data are either poor or not available . Life tables were collapsed to represent 5-year age groups from age zero to 75 years and above . Years Lived with Disability ( YLD ) estimates were estimated using prevalence proportions in South Sudan applied to one third of the total Sudanese population ( South Sudan made up about one third of the total population of Sudan in 2001 ) . Like the authors of the Global Burden of Disease study in 2000 , we maintained the same assumptions here and applied relative risks of 2 . 5 and 1 . 5 of death for blindness and low vision respectively [12] . The disabling sequelae of an established trachoma infection do not remit unless treated early enough . It was assumed , therefore , that the population has no access to treatment and that there is no remission once trichiasis has developed . The public-domain disease modelling software , DisMod II [13] , was used to model trachoma in South Sudan and estimate unavailable parameters . DisMod II is a generic mathematical disease model which describes the relation between incidence , prevalence and mortality ( IPM model ) and can be used to supplement observational data producing internally consistent epidemiological estimates . Here the prevalence of each of the three stages of trichiasis in the district of Mankien , as well as assumptions about the remission rates and relative risks of mortality , were used to generate the estimated incidence and duration of each stage of the condition for the population of South Sudan . Incidence was first calculated for blindness using the prevalence , the assumed relative risk of mortality of 2 . 5 and a remission rate of zero . While the only exit from blindness from trachoma is death , low vision has two exits , death and blindness . DisMod models only a single disease specific exit ( case fatality ) from the prevalent state , therefore to account for these double exits the incidence of blindness was added to the case fatality for trichiasis with low vision . Similarly , the same logic was applied to trichiasis with normal vision which we assumed had no excess mortality , by setting the case fatality of trichiasis with normal vision in the DisMod model equal to the incidence of trichiasis with low vision . Years Lived with Disability ( YLD ) estimates were calculated with the standard formula [14] , [15] . Estimates were calculated with and without age weighting and discounting at 3% . Estimates presented here follow conventional age-weighting and discounting ( 3 , 1 ) . Duration estimates for each age group and each disease stage were obtained from the DisMod II output . In our analysis we have used durations gained from two sources . For the first aim of our study , to assess the additional burden of trichiasis prior to loss of vision , we employed the durations obtained from modelling the observed prevalence , relative mortalities ( or case fatalities ) and remission rates using DisModII . For the second aim , the assessment of different assumptions about duration and disability weights , we have used two scenarios for both trichiasis with normal vision and with low vision using evidence gathered from literature . Ghambir et al [16] reviewed all studies which reported an estimate of trachoma disease incidence and trachoma disease duration . Results varied greatly . In Tanzanian women , 27% of those with trichiasis had developed corneal opacity ( CO ) within 10 years [17] . In The Gambia 15% of a sample with trichiasis developed visual impairment or blindness in 12 years [18] . The range of durations cited in the literature suggests that environmental and contextual factors play an important role in the natural progression of the disease [16] . As the largest study of its kind , the estimates of transition observed by Munõz et al [17] were used as a basis here to make some assumptions about duration of trichiasis with normal vision . In their study of 4 , 898 women in Tanzania , 10-year cumulative incidence of CO from trichiasis was up to 35 . 1% in age groups under 35 and between 42 . 7% and 53 . 5% in age groups 35 and above . CO has not been further graded in the literature and there is no evidence to directly suggest duration of the low vision state . One study in The Gambia [18] found that over a 12 month period between 5% and 17% of those with incident CO transitioned to progressive CO . From the above evidence we conservatively conjectured , that it could take about ten years for 50% of those 35 and above to transition to CO and about 20 years for those under 34 and younger to do so . We constructed an arbitrary range of durations positing that it may take between 15 to 25 years for 50% of those under 35 , and between 5 and 15 years for those 35 and above , to progress from trichiasis with normal vision to trichiasis with low vision . It was also assumed that transition from low vision to blindness could take on average between 3 to 10 years ( Table 2 ) . The burden of trichiasis with normal vision has never been assessed before so it was necessary to derive a disability weight for the condition . Using DWs developed for other conditions for the GBD we have assumed the DW of trichiasis with normal vision to be within the range of 0 . 024 and 0 . 12 . The lower and upper limit of this range are intended to represent consistent low level pain or discomfort and chronic severe pain respectively , while the pain and discomfort associated with Onchoceriasis is place as about equivalent to our mid-way DW of 0 . 068 . . The DW used for trichiasis with low vision has varied between different burden studies . In order to reflect the uncertainty of the contribution of trichiasis with normal vision to total trachoma burden a range of DWs for trichiasis with low vision was also used corresponding to those used in previous studies ( Table 3 ) .
Incidence rates and duration estimates for each of the trachoma disease states by age were modelled using DISMOD II and are presented here in Table 4 . Employing these modelled durations and the disability weights of 0 . 245 and 0 . 068 for trichiasis with low vision and trichiasis with normal vision , respectively , all states of trachoma combined , incident in 1 year , resulted in 174 , 550 Years Lived with Disability in South Sudan ( or 16 . 46 YLD/1000 person years ) . Trichiasis with normal vision contributes 19 , 219 YLDs , or 11% , to the total burden ( Figure 1 ) . Women carry more of the burden than males for all stages of the disease . The inclusion of trichiasis with normal vision contributes 11 . 73% to total burden experienced by females ( with age-weighting ) and is equivalent to 30% of the burden amongst women due to trichiasis with blindness . Most of the burden associated with trichiasis with normal vision is experienced during childhood and while the burden of trichiasis with low vision is placed upon young adults and blindness is mostly experienced in later life ( Figure 2 ) . Age weighting emphasizes the contribution of trichiasis with normal and low vision on the total burden of trachoma ( Figure 1 ) . Employing the different disability weights described above in Table 3 , Table 5 reports the resulting range of Years Lived with Disability ( YLD ) estimates . Estimates of total YLD were sensitive to the application of different weights to trichiasis with both normal and low vision . The difference between the lowest and highest disability weight for trichiasis with normal vision was 5 times higher and 1 . 6 times higher for trichiasis with low visions . Depending on the disability weight employed , the burden due to trichiasis with normal vision can contribute between 5% and 21% of the total burden of trichiasis adding up to 33 , 917 YLD to the total . Figure 3 presents the sensitivity of the results to the assumed durations of the two disease states . As would be expected , the longer the duration the more sensitive the results are to the different disability weights . Similarly , because trichiasis with low vision results in greater burden in South Sudan than trichiasis with normal vision , it exhibits more sensitivity to the assumptions employed . Using the longer duration model 2a can add up to 0 . 84/1000 person years ( or 8 , 883 YLD ) for trichiasis with normal vision as compared with the short duration model 1a and using model 2b for trichiasis with low vision can add up to 3 . 69/1000 person years ( or 39 , 171 YLD ) as compared with the short duration model 1b ( in both cases the highest DW have been applied ) .
Valuing health states is a contentious business and disability weights have attracted a fair amount of critical attention since the development of the DALY . One of the major challenges in developing disability weights arises when a disease is characterized by a spectrum of severity levels and when there are multiple stages of differing severity . A series of efforts have attempted to construct DWs capable of capturing this disability by foregoing the assumption underlying the standard DALY which requires independence between duration and disability [21]–[23] . The secondary aim of this study was to assess the impact of using both a range of disability weights and durations . Many of these assumptions were arbitrary and were not scientifically derived but they were intended to show the extent to which a range of inputs , providing a reasonable reflection of the uncertainty around these parameters , may impact upon the outcome of such an analysis . In the case of trachoma in South Sudan , the burden estimates are particularly sensitive to the range of durations employed here . Regardless of the assumptions used , the health burden of trachoma is consistently under-estimated because the disease state of trichiasis without vision loss had not previously been included in estimations . Including this state increases burden estimates considerably . Underestimating the disease burden caused by trachoma understates its importance not only to those directly affected but also to those at risk . Understating the burden leads to neglect of disease; not recognizing the impact of successful control measures and not prioritising trachoma control for neglected populations . Burden estimates of a number of diseases classified as ‘neglected tropical diseases’ have been revisited recently . The burdens estimated for schistosomiasis , leishmaniasis , diarrhoea and rabies for example have all been re-assessed with the aim of ensuring that the complex patterns of disability characteristic of these conditions is taken into account [24]–[29] . In this study we have looked at a disease that is characterised by graded severity and noted that the priority is for more robust estimates of disease duration . Until a better understanding of trachoma's natural history is gained , efforts to perfect the disability weights to be assigned to the different severity levels will be mostly wasted [23] . Estimating disease burdens is an iterative process which requires on-going , self-assessment and self-correction as more data become available . In the case of trachoma , as with other non-fatal conditions , there remains considerable uncertainty around the occurrence parameters , and the duration of different stages of the disease in particular . Improving these estimates remains a priority . | Trachoma is an infectious disease that is endemic to the Republic of South Sudan . In the absence of appropriate treatment recurrent re-infection in an individual will lead to progressively severe states of trachoma , eventually leading to the loss of visual acuity and finally blindness . Here we distinguish between three separate states of disease: trachoma with normal vision , trachoma with low vision and trachoma with blindness . The first of these states , trachoma with normal vision , is the least severe and the impact of this state on a population has not been well investigated . Trachoma , even before any loss of vision , comes with a great deal of pain and social consequences , and thus disability . In this study we employ data from South Sudan and estimate the burden caused by trachoma with normal vision for the first time . In doing so , we also reveal the extent of the gaps in our knowledge surrounding the natural history of trachoma and highlight areas of research that require urgent attention . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"infectious",
"diseases",
"public",
"health",
"and",
"epidemiology",
"epidemiology",
"global",
"health",
"neglected",
"tropical",
"diseases"
] | 2012 | The Burden of Trachoma in South Sudan: Assessing the Health Losses from a Condition of Graded Severity |
Toxocariasis is a worldwide helminthic zoonosis caused by infection with the larvae of the ascarid worms that comprise the Toxocara spp . Children are particularly prone to infection because they are exposed to the eggs in sandboxes and playgrounds contaminated with dog and cat feces . Certain behaviors , such as a geophagy habit , poor personal hygiene , a lack of parental supervision , close contact with young dogs , and ingestion of raw meat , as well as gender , age , and socioeconomic status , affect the prevalence of the disease . However , previous studies of the risk factors for toxocariasis have generally produced inconsistent results . An epidemiological cross-sectional study was conducted to evaluate the seroprevalence of IgG anti-Toxocara spp . antibodies and associated factors in schoolchildren from a region in the southeast of Brazil . A total of 252 schoolchildren aged 1 to 12 years ( 120 males and 132 females ) were assessed . An enzyme-linked immunosorbent assay based on Toxocara canis larval excretory-secretory antigens was used to determine outcomes . A questionnaire was used to collect information on children , family , and home characteristics . Clinical and laboratory data completed the dataset investigated in this study . Seroprevalence was 15 . 5% ( 95%CI 11 . 5–19 . 8 ) . Geophagy ( aPR 2 . 38 [95%CI 1 . 36–4 . 18] , p-value 0 . 029 ) and the habit of hand washing before meals ( aPR 0 . 04 [95%CI 0 . 01–0 . 11] , p-value ≤0 . 001 ) were factors associated with increased and decreased seroprevalence , respectively . The income factor and its related variables lost statistical significance after adjustment with a multiple Poisson regression model . The current study confirms that toxocariasis is a public health problem in the evaluated area; modifiable factors such as soil contact and personal hygiene appear to have a greater influence on the acquisition of infection than sociodemographic attributes , thus representing direct targets for disease prevention and control .
Human toxocariasis is a helminthic zoonosis caused by the larvae of the ascarid worms of Toxocara spp . In the early 1950s , Toxocara canis was recognized as a human pathogen [1] , and the term “visceral larva migrans” was so widely used that human toxocariasis is also known as visceral larva migrans syndrome . Two species of roundworms , Toxocara canis and Toxocara cati , are recognized as causative agents of human toxocariasis [2] , [3] , [4] , [5] . The adults of both of these species parasitize the small intestines of their definitive hosts , which are canids and felids , respectively [6] . Toxocara spp . have a worldwide distribution and tend to be more prevalent in tropical regions , including industrialized countries , where they are considered the cause of the most frequent form of helminthiasis . Toxocariasis has been described as the most common human parasitic worm infection in developed countries [2] , [7] , [8] . Humans are infected via the accidental ingestion of embryonated eggs containing Toxocara spp . larvae . This ingestion most commonly results from contact with contaminated soil and rarely from the ingestion of undercooked meat containing Toxocara larvae . Children are notably susceptible to infection because they are exposed to the eggs during recreational activities in sandboxes and playgrounds contaminated with dog and cat feces [2] , [8] , [9] . The clinical manifestations of toxocariasis are related to the location and degree of damage to host tissues caused by the inflammatory response to larval migration . These manifestations can vary from asymptomatic infection to the most severe: visceral toxocariasis ( VT ) , characterized by larval migration through major organs including the liver , lungs and , more rarely , the central nervous system ( CNS ) [8] , [10] , [11]; ocular toxocariasis ( OT ) , which occurs when the larvae of the parasite affect the human eye , causing severe inflammation and potentially producing partial or total loss of vision [12]; and less severe syndromes , characterized by nonspecific signs and symptoms , which have been primarily described in children . These syndromes include covert toxocariasis [13] and common toxocariasis [14] in adults . The literature reports that behaviors such as geophagy , poor personal hygiene and a lack of parental supervision , close contact with young dogs and ingestion of raw meat as well as gender , age , and socioeconomic status affect the frequency of the disease [7] . Nevertheless , the results of various studies of the risk factors for toxocariasis have generally been inconsistent [15] , [16] . Between 2007 and 2008 , soil contamination by eggs of soil-transmitted helminths was evaluated in the same area investigated in the current study [17] . The authors observed a high rate of soil contamination by parasites with zoonotic potential ( 30 . 2% ) , including Toxocara spp . ( 79 . 3% ) , Trichuris spp . ( 13 . 8% ) and Ancylostoma spp . ( 6 . 9% ) in the evaluated samples from sandboxes and playgrounds located in public squares , an observation that prompted a serological evaluation of children in the region . The aim of this paper is to present the methodology and preliminary results of a cross-sectional study conducted to estimate the seroprevalence of IgG anti-Toxocara spp . and associated factors in schoolchildren aged 1 to 12 years from a region in southeast Brazil .
Between 2007 and 2010 , a cross-sectional study was conducted in urban areas of Fernandópolis ( S . 20°17'02'' W . 50°14'47'' ) , a city with 59 , 580 inhabitants located in the northwestern São Paulo State in the southeast of Brazil ( Figure 1 ) . Fernandópolis has a high Human Development Index ( a comprehensive statistic that incorporates population life expectancy , education and income ) of 0 . 83 , whereas the HDI value for Brazil is 0 . 71 [18] . The population of individuals aged 1 to 12 years in the city was estimated to total approximately 22 , 000 children studying in 31 schools , 18 of which were public ( funded by the state or municipal government , with no cost to students ) , 8 philanthropic ( funded in whole or in part by churches , with no cost to students ) , and 5 private schools ( self-funded by charging their student body ) [19] . The sample size ( n = 252 ) was based on the following parameters: 95% confidence level and a 5% margin of error , a statistical power of 80% , design effect of 1 . 5 , and an estimated prevalence of 12 . 4% ( average proportion of two sampling strata ) , as observed by Campos Junior et al . [20] . A complex sample ( multi-stage cluster and strata sampling ) [21] , [22] was prepared . In the first stage , the number of schools with students within the age range being studied was identified ( n = 31 schools ) , and then , schools were selected with a probability proportional to the type of school ( public , philanthropic , or private ) . In total , 2 public schools , 2 philanthropic schools and 1 private school were selected . In the second stage , simple random sampling was used to select groups , the number of classrooms with students within the age range under study was identified ( n = 40 classrooms ) , and 25 classrooms were selected ( 5 classrooms for each school ) with a probability proportional to the eligible age ranges . Three operational strata were used: 1 to 4 , 5 to 8 , and 9 to 12 years old . The representativeness of the age groups was thus conserved intra-classroom . The children outside the specific age group within a classroom were excluded from the final stage , in which children were selected within each classroom by simple random sampling with replacement ( n = 10 children per classroom for original sample members and n = 3 children per classroom for replacement sample members ) . Additionally , 2 children were randomly selected from the 25 classrooms to fulfill the prespecified sample size of 252 . In this stage , the sampling also preserved 2 income strata ( with similar population proportions ) : less than or equal to 2 minimum wages were included in stratum stA , and incomes higher than 2 minimum wages were included in stratum stB . The replacement sample members were identified as alternates and placed on a waiting list; they were included in the study if original sample members refused to participate . The process of identifying the classrooms and the students was based on an updated list of the students in each participating school , with birth dates previously provided by the 5 schools that participated in the study . This list permitted all procedures to be performed randomly . After receiving a formal request and information on the importance , objectives and methodology of this study , the school board of each selected school gave permission for the study to be conducted . Selected children and their families were approached by interviewers to validate the prospective subjects' consent to participate in this research , and over the next week , the interviewers visited the families to collect data and schedule a medical visit for the child at the Support Center for Infectious and Parasitic Diseases in Fernandópolis . Some of the variables investigated in blocks were as follows: 1 ) characteristics of children ( collected from family interview questionnaire ) : gender , age , type of school ( public , philanthropic , or private ) , geophagy , onychophagy and hand-washing habits; 2 ) family and home characteristics ( collected from family interview questionnaire ) : SUS users only ( SUS - Brazilian Unified Health System ) and income ( wage of head of household ) ; 3 ) clinical characteristics ( collected from medical records ) : history or presence of hepatomegaly , splenomegaly , adenomegaly , cutaneous manifestations ( urticarial , pruritus , or eczema ) , lung manifestations ( wheezing , bronchitis , or cough ) , pneumonia , and seizures; 4 ) laboratory characteristics ( collected from laboratory reports ) : complete blood count ( CBC; absolute and relative eosinophils , and other cells ) , intestinal parasites ( larvae and eggs in fecal samples ) , and anti-Toxocara spp . IgG antibodies evaluated with an enzyme-linked immunosorbent assay ( ELISA ) . Data from different sources were collected by a team of 8 interviewers who were previously trained in pilot studies over a 1-week interval at 2 schools that were not included in the final sample . The kappa coefficient was used to verify agreement between the 2 questionnaire applications . The agreement between the questionnaire applications was considered very good ( kappa coefficient = 0 . 94 ) . Refusal at the blood collection or interview stage was observed in 3 . 17% ( 8 ) of individuals without prejudice to the study because the alternate participants were then included in the study . For this specific study , 2 binary ( yes/no ) outcomes were considered: ( 1 ) positivity by ELISA test for anti-Toxocara spp . antibodies , described in the following methodology sections , and ( 2 ) physician diagnosis that was performed by an infectious and parasitic diseases expert based on the clinical history of characteristic signs and symptoms ( described in item 3 of this section ) and the assessment of laboratory findings , the most relevant of which are the serology ( ELISA anti-Toxocara spp . ) and eosinophilia ( from CBC ) . As the main objective of the study was to evaluate the risk factors for contact with the parasite , greater emphasis was placed on the factors associated with the first outcome . The study was reviewed and approved by the Research Ethics Committee of the Clinics Hospital , São Paulo University Medical School ( Protocol Number #0518-07 ) , in accordance with Brazilian and international laws . Legal guardians were informed about the study ethical criteria and procedures by letter and by interview with the research assistant; they all signed the informed consent for the children that participated in this study . Blood samples were collected by the Clinical Laboratory School from the Fernandópolis Educational Foundation into Vacutainer™ tubes ( BD , Franklin Lakes , NJ , USA ) with and without EDTA . The blood in the EDTA tubes was used for CBC counts and differentiation . Red and white cell counts were obtained with a Coulter T890 automated hematology analyzer ( Beckman Coulter , Brea , CA , USA ) , and differential leucocyte counts were evaluated via the microscopic examination of stained blood smears . Eosinophil levels ( to characterize relative eosinophilia ) were assessed as the percentage of total leucocytes represented by eosinophils [23] . Serum samples were separated from the blood in EDTA-free tubes by centrifugation and stored at −20°C prior to use . The samples were sent to the Seroepidemiology and Immunobiology Laboratory of the São Paulo Tropical Medicine Institute for the IgG anti-Toxocara spp . ELISA test . A series of stool samples for parasitological examinations were collected from the schoolchildren on 3 different days in the same container with the liquid preservative MIF ( merthiolate-iodine-formaldehyde ) . The following parasitological methods to detect the larvae and eggs of parasites were used in the laboratory analyses: the flotation technique in a saturated sodium chloride solution with a density of 1 . 20 g/mL [24] , the spontaneous fecal sedimentation technique [25] , and the formol-ethyl acetate concentration technique [26] . A preparation of infective Toxocara canis larval excretory-secretory antigens ( TES ) was obtained as described by Rubinsk-Elefant et al . [27] , with several modifications briefly described as follows . Eggs obtained from the uteri of female worms were incubated in 2% formalin for approximately 1 month at 28°C . Formalin was removed after exhaustive washing with physiological sodium chloride solution ( 0 . 85% NaCl ) , and the embryonated eggs were artificially hatched in serum-free Eagle's medium . Larvae were recovered by transferring the suspension on a loose cotton wool plug to a Baermann apparatus . After 18 h , larvae were collected from the bottom of the apparatus , and cultures were incubated at 37°C . The supernatants that contained the antigens were removed and replaced with fresh Eagle's medium at weekly intervals . Different batches were pooled , concentrated in an Amicon apparatus , dialyzed against distilled water , centrifuged ( 18 , 500 g ) at 4°C for 60 min , and filtered with a 0 . 22 mM Millipore membrane ( Millipore , Danvers , MA , USA ) . The protein content was determined with the Lowry protein assay [28] . Antibodies cross-reacting to Ascaris spp . were removed from the sera by preincubating with an adult worm extract ( AWE ) of Ascaris suum [27][29] . In the enzyme-linked immunosorbent assay ( ELISA ) , all sera were preincubated with a solution ( 25 µg/mL ) of AWE in 0 . 01 M phosphate buffered saline ( PBS , pH 7 . 2 ) containing 0 . 05% Tween-20 ( PBS-T ) for 30 min at 37°C . Ninety-six-well microtitration polystyrene plates ( Corning , Costar , New York , NY ) were coated ( 100 µL/well ) with TES antigen solution ( 1 . 9 mg/mL in a 0 . 1 M carbonate–bicarbonate buffer , pH 9 . 6 ) , incubated for 2 h at 37°C , and then incubated for 18 h at 4°C in a humidified chamber . After the wells were washed 3 times with 0 . 01 M ( PBS , pH 7 . 2 ) containing 0 . 05% Tween-20 ( PBS-T ) , the plates were blocked ( 200 µL/well ) with 1% bovine-serum albumin in PBS-T ( BSA , Sigma , St . Louis , MO , USA ) for 1 h at 37°C . After 3 washing cycles with PBS-T , the plates were incubated ( 100 µL/well ) with serum samples ( dilution: 1/320 ) for 40 min at 37°C . After incubation with serum samples , the plates were washed 3 times and incubated ( 100 µL/well ) with horseradish peroxidase-conjugated goat anti-human IgG ( Sigma ) at a 1∶10 , 000 dilution in PBS-T for 40 min at 37°C . The plates were washed 3 times and incubated ( 100 µL/well ) with ortho-phenylenediamine ( 0 . 4 mg/mL , OPD-Fast , Sigma , Dorset , United Kingdom ) and H2O2-urea ( 0 . 4 mg/mL ) in 0 . 05 M citrate buffer for 15 min at 37°C . The reaction was stopped ( 50 µL/well ) with 2 M H2SO4 . The assay was monitored by including negative and positive serum samples in addition to a blank ( no serum sample ) . Absorbance at 492 nm was determined in an automatic microplate reader ( Titertek Multiskan MCC/340 , Lab-system , Helsinki , Finland ) . A cut-off absorbance value was defined as the mean absorbance reading for 96 negative control sera plus 3 standard deviations . Antibody levels were expressed as reactivity indices ( RIs ) , which were calculated as the ratio between the absorbance values of each test sample and the cut-off value; positive samples had RIs greater than 1 . Data were entered by 2 research assistants on a specific EpiData 3 . 1 form ( The EpiData Association , Odense , Denmark ) using the double entry method with subsequent validation . The agreement observed was high ( kappa = 0 . 92 ) , and inconsistencies were resolved to complete agreement ( kappa = 1 . 00 ) . The database was exported to the Statistical Package for the Social Sciences ( SPSS ) 20 for Windows ( International Business Machines Corp , New York , USA ) , to STATA 12 ( StataCorp LP , Texas , USA ) , and R version 3 . 0 . 2 ( http://www . r-project . org/ ) for statistical treatments . The selected variables were initially studied in conjunction with confidence intervals ( CI ) estimated from 1 , 000 bootstrap samples [30] . The design effect ( deff ) was estimated to study the variance between and within the clusters ( classrooms ) in relation to the income sampling stratum for all the presented variables [31] , [32] , [33] , [34] . A post-hoc proportion Z-test was conducted between sample and population categories [35] . The variables analyzed were gender , age , income ( wage of head of household ) , and school type . A univariate analysis was performed with the Pearson chi-squared test ( χ2 ) , while a chi-square test for linear trends or Fisher's exact test were initially used to examine the association between positivity in the IgG anti-Toxocara spp . test and the analyzed factors [36] . A crude prevalence ratio ( cPR ) was applied to assess the impact of individual factors on outcomes [37] . A multiple Poisson regression with robust variance was used to estimate the adjusted prevalence ratio ( aPR ) , and a 95% confidence interval ( 95% CI ) was also applied [38] , [39] , [40] , as recommended for high-prevalence outcomes [41] , [42] in conjunction with consideration of deff [43] , [44] . The variables significantly associated in the univariate model ( p-value <0 . 05 ) remained in the final models ( all entered simultaneously ) . To improve the final model , the predictor variables were tested for collinearity with variance inflation factor ( VIF ) and for the presence of influential values . The accuracy of the model was evaluated using a cross-validation system [45] , [46] . The significance level was set at p-value <0 . 05 .
The general prevalence of IgG anti-Toxocara spp . antibodies was 15 . 5% , with equal variability between and within clusters considering income strata ( deff 1 . 0 ) . The descriptive analyses of the evaluated attributes , socio-demographic characteristics ( including sample and population comparisons ) , behaviors that increase soil ingestion , and clinical laboratory characteristics including the prevalence of IgG anti-Toxocara spp . antibodies are shown in Table 1 . The variable gender showed an equal distribution in the sample , already age showed high percentages for the categories 4 |- 6 years and 6 |- 9 years . Income was concentrated in the first category , <1 minimum wage , and in the last category , ≥5 minimum wages . The public and philanthropic categories for the type of school were combined due to the similarity of individuals and the representativeness of the sample . The category of exclusive users of SUS completed the socio-demographic characteristics block . For the socio-demographic variables ( excluding SUS users only ) , all the tested attributes were found to be comparable to the statistical distribution in the population . The data on behaviors that increase soil ingestion showed that participants without geophagy , without onychophagy , and with the habit of hand washing dominated the sample ( Table 1 ) . Prevalence according to the sampling strata was studied for two possible outcomes , IgG anti-Toxocara spp . antibodies and physician diagnosis , with statistical significance ( p-value ≤ 0 . 001 ) observed for both variables and a greater impact on stA in both cases ( Table 2 ) . The unadjusted analysis showed that factors associated with IgG anti-Toxocara spp . antibodies that contributed to higher prevalence were SUS users only , geophagy , and onychophagy . Relative eosinophilia was also suggested as an associated factor . Age was associated only considering a linear trend effect but was not significant as an estimator of risk . Income in the categories of ≥5 minimum wages , 4 |- 5 minimum wages , private school , and the habit of hand washing were factors that contributed to low prevalence for this outcome . After adjustment in the Poisson model , however , most analyzed variables lost their statistical significance and remained associated only with the habit of hand washing and geophagy ( Table 3 ) . Intestinal parasite infections or commensal species were detected in 33 ( 14 . 9% ) children with prevalences of 9 . 7% for protozoa and 5 . 8% for helminths . The predominant species were Entamoeba coli ( 4 . 5% ) , Giardia spp . ( 4 . 1% ) and Strongyloides stercoralis ( 3 . 1% ) . No significant association was observed between intestinal parasites and anti-Toxocara spp . antibody positivity ( Table 4 ) .
The seroprevalence of Toxocara spp . in schoolchildren aged 1 to 12 years from urban areas of Fernandópolis , in northwestern São Paulo State , was examined with an anti-Toxocara spp . ELISA test and estimated as 15 . 5% ( 95% CI 11 . 5–19 . 8 ) . In previous studies of human toxocariasis in Brazilian children based on use of the same method , higher rates were reported: 51 . 6% [47] and 36 . 8% [48] . Other studies , such as Manini et al . [49] , demonstrated rates ( 17 . 8% ) similar to that found in the current study . Worldwide , Toxocara seroprevalence ( based on an ELISA test ) in children has been reported to range from 7 . 3% to 62 . 3% [50] , [51] , [52] , [53] , [54] . The sample investigated in this study showed good inferential quality for several reasons . First , the application of the resampling method helped to obtain narrower CIs to compare the sample with the population , and all CIs included the corresponding population parameter . Furthermore , no statistical significance was observed in the classic Z-test for proportions , also applied for this purpose . Thus , the sample showed similarities to several characteristics of the original population ( compare the CI sample with the population proportion and the p-values in Table 1 ) . The impact of the complex sample design upon variance estimates is measured by the deff . It is defined as the ratio of the variance of a statistic that accounts for the complex sample design to the variance of the same statistic based on a hypothetical simple random sample of the same size . Accordingly , a deff value of 1 indicates that the variance for the estimate under cluster sampling is the same as the variance under simple random sampling . In contrast , a deff value greater than 1 indicates that the effective sample size is less than the number of sampled persons but greater than the number of clusters; moreover , there is a loss of precision and a reduction in the effective sample size because individuals are chosen within clusters rather than sampled randomly throughout the population [22] , [34] . In this study , a deff of 1 . 5 was used in the sample size calculation step to prevent a loss of precision , but an initial analysis demonstrated that the variables of age , income , type of school , SUS users only , onychophagy and physician diagnosis were associated with higher deff values . For this reason , the univariate and multivariate analyses incorporated the design effect to ensure more accurate inferences [43] , [44] . In addition to the enabling environment for toxocariasis transmission observed in a previous study [17] , the municipality under study has a large dog population . The most recent official data , based on a rabies vaccination campaign , reported approximately 18 , 000 dogs ( an average based on the years 2008–2010 ) [55] , corresponding to a ratio greater than 3 people for every dog . This phenomenon is difficult to measure and may contribute to the epidemiological context for toxocariasis in Fernandópolis . In general , variation in the rate of seropositivity can occur due to the presence of dog and cat populations with high prevalences of T . canis and T . cati [56] , close relationships between pets and humans indoors [57] , [58] , [59] , and the defecation habits of pets or infected stray animals in streets and public squares . These factors produce contamination of the environment , particularly the soil , and can create an environment suitable for human infection [55] , [56] . However , certain factors do not represent the life cycle of the parasite or the natural history of the disease but rather the research design and the outcome of the diagnostic method . First , the sampling design should consider the principles of randomness , uniformity and stratification of individuals relative to the group composition to ensure that the exposure factors and outcome found for the sample will be similar to the population . Considerable variation among laboratory methods , technical modifications in the production of TES , and disagreement over the cutoff definition must also be considered [27] . These factors can generate results that do not necessarily reflect the prevalence in the general population . Differences in seropositivity for IgG anti-Toxocara spp . antibodies according to income strata have been found in previous studies . These differences always indicate a greater impact on the lower income classes [16] , [20] , [60] . The same effect was observed , although to a lesser extent , for the appearance of clinical disease ( only VT cases were observed ) , but the reason for this outcome is that the basis of the medical diagnosis was the serologic test [3] , [7] , [8] . Other authors have stated that the duration of human IgG responses elicited by Toxocara larvae remains undetermined [61] . Viable larvae may persist in tissues and excrete/secrete antigens for several years , and no simple method is available to confirm parasite death after chemotherapy . Consequently , a single-sample IgG-ELISA titer cannot distinguish between past and current infection [62] . Based on this , the physician diagnosis was included in this evaluation to obtain more specific diagnoses of clinical disease . Thus , patients with a positive physician diagnosis received antiparasitic and/or anti-inflammatory therapy when needed , while those with positive serology without any apparent signs and symptoms were placed for clinical follow-up . Many factors that predict Toxocara contact have been identified in human pediatric populations , but the results have been inconsistent [15] , [29] . The results for gender have been contradictory . Specifically , the male gender has been observed to face increased risk [51] , [57] , to protect the subject [63] , [64] and to show no association with Toxocara contact [52] , [65] , [66] . A young age [63] , low socioeconomic status [16] , [20] , [60] , low parental education [52] , poor sanitation [53] , [67] , onychophagy [65] , [66] , geophagy [57] , [59] , an absence of hand washing before meals [54] , eosinophilia [51] , [58] , and dog ownership [57] , [58] , [59] , [68] are factors associated with Toxocara contact or infection . None of these known factors , except for geophagy and hand washing before meals , was significantly associated with Toxocara exposure in the population investigated in the current study , with p-values ranging between 0 . 094 and 0 . 954 in the final Poisson model ( Table 3 ) . The significant associations between positivity of IgG anti-Toxocara spp . antibodies found in the univariate analysis , as well as the significant associations between this outcome and the magnitude of income , type of school , SUS users only , onychophagy , and relative eosinophilia , were lost after a multivariate analysis . The 2 variables that remained associated with this positivity variable represent the modifiable behavioral habits that underscore the proximity of humans to the parasite life cycle . The permanence of these in the multiple model , combined with the partial modification of geophagy and no modification of the habit of hand washing before meals ( compare p-values and cPR vs . aPR in Table 3 ) , suggests that income and the other sociodemographic attributes may be confounding factors in this relationship . The gender variable was retained in the final model , although it was not significantly associated in the univariate analysis because it implied a contradiction . In practice , geophagy and a lack of hand washing before meals are behaviors that increase soil ingestion ( directly or indirectly ) , facilitating human contact with the eggs of the parasites . We also suggest the hypothesis that hand washing before meals is more common in children whose socioeconomic status is high . For this reason , an association ( due to a confounding effect ) can be expected between poverty and the presence of IgG anti-Toxocara spp . antibodies . In summary , a confounding effect occurs if the association between exposure and outcome is distorted by the presence of another variable , the confounding factor [69] , [70] , [71] . Geophagy should be a phenomenon observed in both economic strata , in which hand washing demonstrates only a partial effect . In general , income and related sociodemographic variables have been strongly emphasized as causal factors in the human-Toxocara relationship based on interpretations of unadjusted [20] , [60] , [72] or adjusted models , considering few [65] , [73] or no [16] , [74] , [75] attributes that are important and directly related to the parasite life cycle , such as soil contact and personal hygiene . This paper , however , suggests that these modifiable factors are more important than sociodemographic attributes and are thus a direct target for disease prevention and control . As a simple example , we suggest that modifiable factors discussed here , such as the habit of washing hands , can be taught within schools . A study in Brazil with students from elementary school addressed health education in toxocariasis prevention; among the most interesting findings , the researchers suggested that the use of only 1 approach is insufficient to change risk behaviors in children , which reinforces the idea that the educational process should be continuous throughout the stages of child development [76] . The strengths of this study are as follows: a complex probability sample with a multi-stage cluster design and stratified sampling of income , samples collected from different schools with the same methodology , and appropriate statistical analyses controlling for potential confounding factors and also consideration of the sampling design . The main limitations of this study is its cross-sectional design , which makes it impossible to establish causality and makes it difficult to relate the seropositivity found in the study to environmental contamination present in Fernandópolis . In terms of the impact of sensitivity and specificity , we do not expect variation beyond that expressed by the confidence interval for the seroprevalence , especially because false positive results may occur due to trichinosis and fascioliasis , which are unusual infections in this population . Strongyloidiasis and other parasite infections was observed but were not significantly associated with serological status . The literature also reports that false negative results are rare and only occur in certain localized early or very old infections ( OLM ) [77] . A previous study found that the ELISA technique , accompanied by absorption of the serum with Ascaris suum antigens , demonstrated a sensitivity of 80% and a specificity of 90% [78] . For physician diagnosis , one laboratory parameter used by the physician was eosinophilia as determined from the CBC , which represented a non-specific test that could be altered for allergic conditions and parasitic infections . In this study , eosinophilia was used in conjunction with the presence of signs and symptoms and serology results to obtain a more specific clinical diagnosis . Additional studies that consider a longitudinal perspective , such as a prospective-cohort design or statistical models employing hierarchical or multilevel analyses , are needed to make new contributions to the discussion of the factors related to contact with Toxocara spp . , as the vast majority of previous research has involved cross-sectional studies using classical models of statistical analysis .
The current study confirms that toxocariasis is a public health problem in urban areas of Fernandópolis , in northwestern São Paulo State , southeast Brazil . The presence of modifiable behaviors that increase soil ingestion , such as the habits of geophagy and a lack of hand washing , contributes to the seroprevalence rates observed in the evaluated schoolchildren . | Human toxocariasis is a neglected parasitic infection occurring in Brazil and worldwide . The combination of the close proximity of canines and felines to humans , environmental contamination by infectious forms of the parasite and neglect by public health officials provides a favorable situation for the spread of this zoonosis , which primarily affects pediatric populations . An epidemiological study in an area of known parasitic soil contamination was conducted to evaluate the impact of this disease and the factors that influence its occurrence . A sample of 1- to 12-year-old children was obtained with random selection of schools , classrooms and schoolchildren . A questionnaire was used to collect socioeconomic and behavioral information , laboratory tests determined the presence of antibodies against Toxocara spp . parasites and a medical evaluation noted clinical features of the disease . The study found that antibodies is present in more than 15% of subjects and can be in as many as 30% depending on individual characteristics . The analysis also indicated that behaviors that increase soil ingestion , such as nail biting , eating dirt or poor personal hygiene , contribute to increased frequency of the disease and may be even more important than the socioeconomic status of individuals . These behaviors therefore represent real and simple targets for the control of this neglected tropical disease . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion",
"Conclusions"
] | [
"public",
"and",
"occupational",
"health",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"epidemiology",
"pediatrics"
] | 2014 | Seroprevalence and Modifiable Risk Factors for Toxocara spp. in Brazilian Schoolchildren |
Natural products have moved into the spotlight as possible sources for new drugs in the treatment of helminth infections including schistosomiasis . Surprisingly , insect-derived compounds have largely been neglected so far in the search for novel anthelminthics , despite the generally recognized high potential of insect biotechnology for drug discovery . This motivated us to assess the antischistosomal capacity of harmonine , an antimicrobial alkaloid from the harlequin ladybird Harmonia axyridis that raised high interest in insect biotechnology in recent years . We observed remarkably pleiotropic effects of harmonine on physiological , cellular , and molecular processes in adult male and female Schistosoma mansoni at concentrations as low as 5 μM in vitro . This included tegumental damage , gut dilatation , dysplasia of gonads , a complete stop of egg production at 10 μM , and increased production of abnormally shaped eggs at 5 μM . Motility was reduced with an EC50 of 8 . 8 μM and lethal effects occurred at 10–20 μM within 3 days of culture . Enzyme inhibition assays revealed acetylcholinesterase ( AChE ) as one potential target of harmonine . To assess possible effects on stem cells , which represent attractive anthelminthic targets , we developed a novel in silico 3D reconstruction of gonads based on confocal laser scanning microscopy of worms after EdU incorporation to allow for quantification of proliferating stem cells per organ . Harmonine significantly reduced the number of proliferating stem cells in testes , ovaries , and also the number of proliferating parenchymal neoblasts . This was further supported by a downregulated expression of the stem cell markers nanos-1 and nanos-2 in harmonine-treated worms revealed by quantitative real-time PCR . Our data demonstrate a multifaceted antischistosomal activity of the lady beetle-derived compound harmonine , and suggest AChE and stem cell genes as possible targets . Harmonine is the first animal-derived alkaloid detected to have antischistosomal capacity . This study highlights the potential of exploiting insects as a source for the discovery of anthelminthics .
Natural compounds represent one of the richest sources for the discovery of new active compounds against cancer , infections , or other threats to human health . From 1981 to 2010 , 33% of approved drugs represented natural compounds and derivatives , mostly from plants , algae , and fungi [1] . In recent years , the search for novel anthelmintic compounds from natural sources has been intensified with the aim to identify new hit and lead compounds for drug development [2] . So far , medicinal plants and their metabolites ( like alkaloids , terpenes , and peptides ) have been widely exploited as sources of novel natural compounds with anthelmintic activity . In contrast , only few studies have focused on animal-derived molecules [3] . Surprisingly , although insects are among the most successful and widespread organisms on earth , especially regarding their diversity and adaptability , they are still rather underrated as sources of compounds with medical importance . Along these lines , insects have been almost completely neglected with respect to anthelminthic discovery [4] . The chemical defense of insects against pathogens and parasites relies on effector molecules such as antimicrobial peptides and secondary metabolites . The invasive harlequin ladybird Harmonia axyridis , which is also known as Asian ladybird or Multicolored ladybird , represents an outstanding example in this regard [5] . Its immune system encompasses more than fifty antimicrobial peptides , the highest number ever reported for an animal [6] . In addition , its hemolymph contains extraordinarily high concentrations of the constitutively expressed antimicrobial alkaloid harmonine ( ( 17R , 9Z ) ‐1 , 17‐diaminooctadec‐9‐ene ) , an aliphatic , long-chain diamine which displays antimicrobial activities [7] . Its superior immune system promotes its invasive success in a multifaceted manner . The beetle’s antimicrobial peptides have been demonstrated to mediate resistance against pathogenic bacteria [8] , whereas harmonine has been postulated to keep its microsporidia under control . Microsporidia are highly specialized relatives of fungi that propagate as intracellular parasites in insects and other taxa [9] . H . axyridis carries a high load of microsporidia which can infect and kill native competitors such as the two-spotted ladybird Adalia bipunctata and the seven-spotted ladybird Coccinella septempunctata when transmitted e . g . during intraguild predation . Therefore , these parasites have been postulated to function in invaded areas like bioweapons to successfully outcompete native competitors [6 , 10 , 11] . In addition , first tests against human pathogens showed that harmonine displays a broad-spectrum antimicrobial activity including in vitro effects against mycobacteria as well as protozoan parasites [7 , 12] . Activity against helminths has so far not been investigated , even though novel active compounds are highly needed for a whole list of helminth infections , which includes neglected tropical diseases ( NTDs ) in humans and veterinary infectious diseases [13 , 14] . Schistosomiasis is among the helminthic diseases causing highest disability and mortality in humans worldwide [15] . In endemic areas , schistosomiasis occurs as a chronic disease , which derives from the longevity of blood-resident adult schistosomes persisting many years in the host . Pathology is mainly caused by tissue deposition of eggs which are produced in hundreds per day by the female worm [16] . Approximately 700 million people are at risk of schistosomiasis [15] , and there is no vaccination that might prevent infection . Treatment relies on a single drug , praziquantel ( PZQ ) . PZQ is active against all three major schistosome species , Schistosoma mansoni , S . haematobium and S . japonicum , it can be produced at low cost and is well tolerated . It is therefore used in mass treatment programs and as preventive chemotherapy for people at high risk [17] . This and its continued use since the 1980s have increased the risk of resistance development . Indeed , evidence was obtained for schistosomes with lowered PZQ susceptibility in human drug administration programs and in experimental animal models [18–20] . Therefore , finding alternative treatment options has become an urgent issue [21] . Due to their outstanding species number and diversity , insects constitute a huge “drug cabinet” to be explored which might also include novel compounds with anthelmintic activity . To make a start , we focused in our study on the antimicrobial alkaloid compound harmonine derived from the harlequin ladybird . The aim of this study was to reveal whether this insect-derived compound shows anthelmintic capacity against adult S . mansoni worms . By physiological , cellular , and molecular analyses we observed a complex multifaceted phenotype comprising tegumental damage , gut dilatation , gonadal dysplasia , egg-production deficits as well as cellular and molecular effects on stem cells . Furthermore , we found first evidence for AChE as one potential target . With the results of our study , we want to promote insect-derived compounds and move them into spot-light as possible sources of new drugs in the treatment of schistosomiasis and further parasitic diseases .
Animal experiments were performed in accordance with the European Convention for the Protection of Vertebrate Animals used for experimental and other scientific purposes ( ETS No 123; revised Appendix A ) and have been approved by the Regional Council ( Regierungspraesidium ) Giessen ( V54-19 c 20/15 c h 02 GI 18/10Nr . A 1/2014 ) . A Liberian strain ( Bayer AG , Monheim ) of S . mansoni was used to infect freshwater snails of the genus Biomphalaria glabrata as intermediate host and Syrian hamsters ( Mesocricetus auratus ) as final host [22 , 23] . Eight week-old hamsters were obtained from Janvier ( France ) , infected by the “paddling method” [24] , and sacrificed at 46 days p . i . to collect adult worm couples by perfusion . Unisexual worm populations were generated by monomiracidial intermediate-host infection [23] . Worms were cultured in M199 medium ( Sigma-Aldrich , Germany; supplemented with 10% Newborn Calf Serum ( NCS ) , 1% HEPES [1 M] and 1% ABAM-solution [10 , 000 units penicillin , 10 mg streptomycin and 25 mg amphotericin B per ml] ) at 37°C and 5% CO2 . For in vitro culture of adult couples with harmonine , worms were cultured in 6-well plates in supplemented M199 medium with 10 worm couples per well . Harmonine was synthesized as described by Nagel et al . 2015 [12] and kindly provided by W . Boland ( Max-Planck-Institute for Chemical Ecology , Jena , Germany ) . Harmonine was dissolved in DMSO and added in final concentrations of 2 . 5–50 μM as indicated . As negative control , M199 medium was adjusted to the same concentration of DMSO as used for the highest inhibitor concentration . The worms were incubated at 37°C and 5% CO2 for 72 h , medium and harmonine were exchanged every 24 h . Harmonine-induced morphological effects were assessed every 24 h using an inverted microscope ( Leica , Germany ) . Worm motility was scored with a system following recommendations by WHO-TDR [25] , with the scores 3 ( normal motility ) , 2 ( reduced motility ) , 1 ( minimal and sporadic movements ) , 0 ( no movements within 30 sec was considered dead ) . For depletion of proliferating cells , 20 mM hydroxyurea was added to the culture for 72 h , and medium plus hydroxyurea was refreshed every 24 h . For visualization of proliferating cells , EdU ( 5-ethynyl-2-deoxyuridine ) was added to a final concentration of 10 mM for the last 24 h of in vitro culture . Thereafter , worms were either processed for confocal laser scanning microscopy ( CLSM ) , or subjected to RNA extraction for quantitative real-time PCR ( qPCR ) analysis . Freshly perfused or in vitro-cultured worm couples were separated by gender , and RNA was extracted from male and female worms using the PeqGOLD TriFast reagent ( Peqlab , Germany ) according to the manufacturer’s protocol . RNA quality was checked using the Agilent RNA 6000 Nano kit and an Agilent Bioanalyzer 2100 instrument ( Agilent Technologies , USA ) , followed by reverse transcription using the Quantitect RT-Kit ( Qiagen , Germany ) . Expression levels of the S . mansoni orthologs of the stem cell markers nanos-1 ( Smp_055740 ) and nanos-2 ( Smp_051920 ) , and the AChE ortholog ( Smp_154600 ) were determined by qPCR using the SYBR Green method [26] with the PerfeCTa SYBR Green SuperMix ( VWR , Germany ) , the Rotor-Gene Q instrument and Rotor-Gene Q Series Software ( Qiagen ) . All samples were pipetted in technical triplicates . Ct values were normalized against the geometric mean of three references genes selected based on stable expression in both sexes ( Haeberlein et al . , submitted ) : orthologs of LETM1 ( Smp_065110 ) , phosphatase 2A ( Smp_166290 ) and proteasome-beta ( Smp_073410 ) . Relative expression levels were calculated either by the delta delta Ct method [27] or by expressing the data as n-fold difference by the formula: relative expression = 2−delta Ct × f , with f = 100 as an arbitrary factor ( as indicated in the figure legends ) . The following primers were used , which were confirmed by test qPCRs to have efficacies between 0 . 9–1: LETM1_fw 5’-GAAGGTGATCAAGCTCCATTGT-3’ , LETM1_rev 5’-TTGTACTGCATGGATAGGTGGT-3’; phosphatase-2A_fw 5’-GTAAAACTGGTCCATTTGAAGAAC-3’ , phosphatase-2A_rev 5’-TACCGAATAGGAAATGTTGAACGA-3’; prot-beta_fw 5’-GGTCTGGTGGTTTCTCGTTC-3’ , prot-beta_rev 5’-GTACCTTCTGTTGCCCGTG-3’; nanos-1_fw 5’-ACTTGTCCATTATGCGGTGCT-3’ , nanos-1_rev 5’- GGTTCCAACAAACCAGCTTCA-3’; nanos-2_fw 5’-GCCGTGTTATGACCTCTGG-3’ , nanos-2_rev 5’-GACGATCTGGAGACTCTGG-3’; AChE1_fw 5’-GATGATGATGATGAACGACCG-3’ , AChE1_rev 5’- CAGTAACTAATGATTATCGTATACCA-3’; AChE2_fw 5’-TAAGACACGAAATGATGATTCACG-3’ , AChE2_rev 5’-TACTTCATATTGTGTAGTTGATTGAC-3’ . Native protein lysates were prepared from 50 male or 150 female worms as described [28] . Briefly , worms were sonicated in PBS supplemented with protease inhibitors . The protein concentration was determined by the advanced protein assay reagent ( APAR , Cytoskeleton Inc . , USA ) and measured at 590 nm in a microplate reader ( VARIOSKAN FLASH; Thermo Fisher Scientific , USA ) . AChE activity was determined in native protein lysates of adult male or female S . mansoni using the Amplite Colorimetric Acetylcholinesterase Assay Kit ( AAT Bioquest , Biomol , Germany ) following the instruction of the manufacturer , which is based on Ellman’s method [29] . For inhibition studies , the protocol was adapted as follows: for each harmonine concentration , 200 μg/ml of native male protein or 100 mU/ml of the AChE standard ( from the electric eel Electrophorus electricus ) were added . A negative control containing an equivalent amount of solvent of harmonine ( DMSO ) was included . To start the reaction , an AChE reaction mixture was added to each well . The final concentration of the substrate acetylthiocholine was 0 . 5 mM . Absorbance by the reaction product TNB-thiocholine , which is proportional to the AChE activity , was measured by the VARIOSKAN FLASH microplate reader at 405 nm with reads every 10 min at 37°C . For morphological analysis by CLSM , worms were fixed and stained with carmine red ( CertistainH; Merck , Germany ) as described before [30 , 31] . For EdU labelling and detection of proliferating cells , the Click-iT Plus EdU Alexa Fluor 488 Imaging Kit ( Thermo Fisher Scientific ) was used . After 24 h of incubation with EdU , couples were separated , fixed and stained as described [32] . Worms were counterstained with Hoechst 33342 in a final concentration of 8 μM . Stained worms were examined on an inverse CLSM ( Leica TSC SP5; Leica , Germany ) . Hoechst was excited with a 405 nm laser , and Alexafluor488 as well as carmine red with an argon-ion laser at 488 nm . Laser power as well as gain and offset of all photomultiplier tubes ( PMTs ) were optimized for minimizing possible bleaching effects and for full range intensity coding using the CLUT-function ( color look-up table ) of the Leica LAS AF software . Background signals and optical section thickness were defined by setting the pinhole size to airy unit 1 . The software package “IMARIS for cell biologists” ( Bitplane , Switzerland ) was used to quantify Hoechst- and EdU-positive cells in ovaries and testes of worms . Z-stacks acquired by CLSM were used as input data . First , a so-called surface was created manually for each organ to extract it in silico from the surrounding tissue . Next , surfaces were created for all EdU- and Hoechst-positive cells . This allowed quantifying stained cells per organ . To minimize counting of artifacts or background noise , a threshold was set prior to surface creations that excluded objects <3 μm . Statistical analysis was performed using an unpaired t-test . A p-value < 0 . 05 was considered significant .
To investigate whether the insect-derived compound harmonine has anthelminthic activity , adult S . mansoni worms were cultured in vitro with different concentrations of the compound over a period of 72 h . Harmonine reduced the pairing stability of worm couples in a dose-dependent manner ( Fig 1A–1C ) . With ≥ 10 μM harmonine , all couples separated within 48 h and were detached from the culture-well surface . With 5 μM , around 50% of worm couples separated ( calculated EC50: 5 . 6 μM ) , and 90% were detached after 72 h . A similar time- and dose-dependent pattern was found for the reduction of worm motility by harmonine ( Fig 1D ) , with a calculated EC50 of 8 . 8 μM ( Fig 1E ) . Specifically , with concentrations of 50 and 20 μM , all worms died within 2 h and 48 h , respectively . With 10 μM , worms had a significantly reduced motility with an average motility score of 1 after 48–72 h , indicating only minimal and sporadic movements; 25% of worms were dead after 72 h . Interestingly , most of the females were unaffected with 10 μM after the first day , whereas treated males showed reduced motility , indicating a gender bias in the effect of harmonine . Concentrations below 10 μM caused a weak , time-dependent reduction of motility . Next to its effects on motility , viability and pairing stability , harmonine affected worm tissue structure in remarkable ways . Concentrations as low as 5 μM induced bubble formation on the tegumental surface in both males and females ( Fig 2A and 2B ) . In addition , severe gut dilatations were observed in both sexes of S . mansoni by bright-field microscopy , which was confirmed by CLSM ( Fig 2C–2E ) . Taken together , harmonine showed antischistosomal activity and induced a dose-dependent spectrum of phenotypic effects in schistosomes , ranging from tissue damage and pairing-instability with 5 μM , severe impairment of worm motility up to death at 10 μM , and finally 100% lethality at 20 μM . Reduced motility , pairing-instability and gut dilatations might point to a target of harmonine involved in ( neuro ) muscular activity . In addition , the observed tegumental effects suggested target localization within the tegument . Previous in silico docking analyses with Leishmania proteins suggested AChE as a possible target of harmonine [34] . AChE is a well-characterized enzyme , also in schistosomes , and it is the presumed target of metrifonate , a formerly used antischistosomal drug [35] . AChE was found to be abundantly expressed in the tegument of schistosomes [36 , 37] . Therefore , we hypothesized that AChE might be one target of harmonine in S . mansoni . By SMART analysis of protein sequences from genes electronically annotated as AChEs in GeneDB and WormBase ParaSite , we identified two potential orthologs of AChE in S . mansoni: Smp_125350 and Smp_154600 , which both show the typical carboxyl-esterase domain ( S1 Fig ) . We suggest the name SmAChE1 for Smp_154600 because of its first characterization in a previous study [38] , and SmAChE2 for Smp_125350 . Based on preliminary data of a transcriptomics study [39] , we first characterized the expression levels of both SmAChE genes in the different sexes as well as the enzymatic activity of their protein lysates . AChE transcript levels were determined by qPCR and revealed a sex- and pairing-dependent expression ( Fig 3A and 3B ) . Notably , expression in females after pairing contact ( from bisexual infection , bs F ) was significantly decreased compared to females in unpaired state ( from single-sex infection , ss F ) or males . For females , this difference in transcript level correlated well with enzyme activity since we determined a significantly lower AChE enzymatic activity in protein lysates of paired females compared to unpaired females . A similar trend was observed for paired females vs . males ( Fig 3C and 3D ) . Next we investigated a possible effect of harmonine on AChE transcript levels . The expression of both genes was reduced in a dose-dependent manner in harmonine-treated compared to control male and female worms ( Fig 4A and 4B ) . Because of the low AChE activity in protein lysates of paired females , male lysates were used to test the capacity of harmonine for inhibiting schistosomal AChE activity . Harmonine ( 10 – 100 μM ) decreased the turnover of the substrate acetylthiocholine in a dose-dependent manner ( Fig 4C ) . Interestingly , the inhibition of AChE from a common test organism ( electric eel ) [40] was even more efficient , and it occurred at lower concentrations ( Fig 4D ) . With an IC50 of 5 . 5 μM against electric eel AChE ( Fig 4E ) , harmonine can be considered a moderate inhibitor of AChE activity , whose potency might be species-dependent . Also compared to IC50 values of the known AChE inhibitor physostigmine ( electric eel , 148 . 4 nM; schistosome lysate , 724 . 2 nM; S2A–S2D Fig ) , harmonine was clearly less potent . As increasing excess concentrations of the substrate ACh at constant harmonine concentration did not restore full AChE reaction velocity , an inhibition mechanism other than competitive inhibition is likely ( S2E Fig ) . Besides motility and morphology , compound-induced effects on reproduction are also of high interest because schistosome eggs are essential for maintaining the life-cycle and causative for the pathology of schistosomiasis . We therefore determined the quantity and quality of egg production during 72 h treatment of adult worm couples with harmonine . Egg production ceased completely with 20 μM harmonine after 48 h and with 10 μM after 72 h , respectively ( Fig 5A ) . 5 μM harmonine reduced the number of eggs to 31% compared to the control , but of note , up to 57% of these eggs were of abnormal size and shape . In addition , free vitellocytes were found in the culture medium indicating egg-production deficits ( Fig 5C and 5D ) . Overall , the EC50 of harmonine for the reduction of egg production was 3 . 6 μM ( Fig 5B ) . To investigate whether impaired egg production was related to gonadal tissue defects , harmonine-treated worms were stained with carmine red for subsequent CLSM analysis , which allowed the detection of morphologic abnormalities at the organ level . As expected [30 , 31] , in control females the vitellarium was found to be tightly packed with cells , and it was arranged similar to a zipper with cell rows interlocking with the opposite side ( Fig 6B ) . Treatment with harmonine at a concentration of 10 μM led to the formation of numerous unstained , hole-like areas , giving the whole vitellarium a swiss cheese-like tissue pattern ( Fig 6E ) . In the ovary of control females , the small immature oogonia are located within the anterior part and the bigger , mature oocytes within the larger posterior part ( Fig 6A and 6C ) , as shown before [30 , 31] . Already at 10 μM , harmonine clearly disrupted the ordered structure of the ovary ( Fig 6F ) . The ovary appeared shrunken , oogonia of smaller size and mature oocytes poorly separated from each other . The cytoplasm stained weaker compared to control ovaries , and unstained , hole-like areas were found similar to the phenotype seen in the vitellarium . In control males , testes consist of distinct lobes which are tightly packed with spermatogonia [30 , 31] . At the anterior end , mature spermatocytes gather in the seminal vesicle ( Fig 6D ) . After harmonine treatment , lobes also appeared shrunken and showed similar porous areas as seen in female tissues . The seminal vesicle contained less spermatocytes and an undefined cell mass instead , which probably represents undifferentiated spermatogonia ( Fig 6G ) . All observed phenotypes were occasionally observed also after treatment with 5 μM harmonine . To sum up , treatment with harmonine led to a dramatic reduction and malformation of schistosomal eggs , which might be related to the structural disruption of ovary , vitellarium , and testes . Two findings led to the hypothesis that harmonine might affect stem cell proliferation . First , one of the non-neuronal functions of AChE described for humans is related to stem-cell activity [41] . Second , the observed harmonine-induced impairment of gonadal cell organization might point to a defect in a preceding step , i . e . gonadal stem-cell proliferation . Therefore , we used the thymidine analogue EdU to visualize proliferating cells in schistosomal tissues of harmonine-treated worms vs . controls by CLSM . As background staining we used Hoechst . Indeed , sublethal concentrations of 5 and 10 μM harmonine reduced the number of proliferating cells in both ovaries and testes compared to organs of control worms ( S3 Fig ) . In order to quantify the number of proliferating stem cells per organ as an objective measure , we established a procedure to separate gonads in silico from the surrounding worm tissue with the help of the analysis software IMARIS . 3D visualization of ovaries revealed that proliferating stem cells are exclusively located in the anterior part ( Fig 7 ) . In addition , stem cells are not homogeneously nested between oogonia , but preferentially located at the outer edge of the organ ( Fig 7C and 7D; S1 Movie ) . Next , we determined the percentage of EdU-positive stem cells per total Hoechst-positive cells for each ovary or testis . The percentage of proliferating stem cells per organ was significantly reduced in harmonine-treated worms compared to control worms . While control ovaries and testes showed on average 30% and 47% EdU-positive cells , respectively , the percentage dropped to 5% and 7% after exposure to harmonine ( Fig 8A–8D ) . This resulted in a merely scattered distribution of stem cells within the gonads . To support these findings , we investigated whether harmonine might also affect the transcript level of nanos-1 ( Smp_055740 ) , a gene described as germline-specific stem cell marker in adult S . mansoni [42 , 43] . Indeed , the transcript level of nanos-1 was significantly reduced in harmonine-treated male and female worms ( Fig 8E and 8F ) . In addition to the gonads , proliferating stem cell-like cells are also present in the parenchyma of S . mansoni [44] . These so-called neoblasts are thought to provide replenishment of tegumental and gastrodermal cells . Impairment of these cells by a compound like harmonine would therefore be a useful way to interfere with worm survival . Similar to proliferating cells in the reproductive organs , the amount of EdU-positive cells decreased after treatment with harmonine in males and females . In the representative control female depicted in Fig 9A ( top , left image ) , a huge number of EdU-positive cells was also detected in the vitellarium , while after treatment , only some residual cells were found ( Fig 9A , bottom , left image ) . Nanos-2 ( Smp_051920 ) is a stem cell marker expressed both in germline and somatic stem cells . Nanos-2 transcripts were found to be reduced after irradiation-mediated depletion of somatic stem cells [44] . We therefore used it as a molecular measure for compound-induced effects on neoblasts . Nanos-2 transcript levels were determined by qPCR after treatment of worms with harmonine or hydroxyurea for comparison . Hydroxyurea is a mitotic inhibitor that was successfully used before to deplete the majority of neoblasts in helminths [45 , 46] . Compared to control worms , the expression of nanos-2 was significantly decreased by almost 50% after treatment with 20 μM harmonine ( Fig 9B ) . Hydroxyurea reduced expression fourfold at a concentration of 20 mM ( Fig 9C ) . Taken together , harmonine reduced the number of proliferating stem cells , both gonadal and parenchymal ones , in S . mansoni , which was proven by gonad-specific quantification of EdU-positive cells and a reduced expression of the stem cell markers nanos-1 and nanos-2 .
Antischistosomal compounds , whether synthetic or natural , often induce alterations in vitality and motility of the adult worms , or in their reproductive fitness ( disruption of mating , diminished egg production ) , in the integrity of the protective tegument , or in the functionality of the parasite nervous system . Remarkably , we found that harmonine affected not just one or few , but all of these parameters simultaneously . Our in vitro studies showed that harmonine reduced the motility of adult worms in a dose-dependent manner , with an EC50 of 8 . 8 μM and lethality at 10–20 μM within 3 days of culture . These effective concentrations are desirably low , also compared to other natural compounds with described schistosomicial activity , many of which were found to be active in the range of 50–100 μM and even higher [2] . Motor activity alterations belong to the most important indicators of schistosomicidal activity . If the neuromuscular system is affected , mating may be disrupted because the female is released from the gynaecophoric canal of the male partner which eventually leads to degeneration of the female and a complete cessation of egg production [47] . In addition , intact muscle function is required for the suckers to allow the worm to attach to the host endothelial wall , and for functionality of the digestive , reproductive and excretory organs which are lined by musculature [48] . Indeed , harmonine-treated male and female worms separated from each other , were unable to attach with their suckers to the culture well , and stopped egg production at 10 μM . In vivo , reduced motor activities would very likely result in removal and degradation of worms by the host and , because egg production is affected , to reduced pathology and the interruption of transmission . Harmonine also caused adverse effects on the tegument at concentrations as low as 5 μM , which plays crucial roles for nutrient uptake , secretion , osmoregulation , and immune evasion [49] . Damage of the tegument could facilitate the penetration of schistosomicide compounds but also of antibodies to deeper-lying tissues , which may culminate in greater damage to the parasite including disruption of the above-mentioned physiological processes and the ultimate elimination of worms [50 , 51] . The tegument is an important target structure for drug discovery , and consequently , compounds affecting the tegument were found to make the parasites more sensitive to the host immune response in vivo [52] . Tegumental damage is also induced by PZQ , which may contribute to the compound’s efficacy [17] . Previous work showed the induction of early necrosis in Leishmania parasites , which involved the loss of cell-membrane integrity [12] . Also in S . mansoni , induction of necrosis might contribute to the detrimental effects of harmonine . Besides the tegument , the nervous system of helminths has been considered a promising target for drug discovery . Several components of the neuronal system are targets of currently approved anthelminthics , including monepantel , levamisole and pyrantel [35 , 53 , 54] . We found a reduction of schistosomal AChE activity and of AChE transcript levels by harmonine , which suggests two molecular mechanisms that contribute to the observed motility reduction and subsequent paralysis of the parasite’s musculature . We conclude that the reduced enzymatic activity in schistosome protein extracts is not exclusively a consequence of reduced gene transcription , because worms used to prepare extracts have not been treated with harmonine , instead harmonine was added directly to the enzyme assay . According to diverse studies in S . mansoni , S . haematobium , S . japonicum , and S . bovis , AChE fulfills two different functions: a classical role in the neuromuscular system for motor activity , and a non-classical role in the tegument related to glucose import from host blood [37 , 55] . The tegumental damage observed in harmonine-treated worms might be correlated with the high tegumental expression of AChE [38 , 56 , 57] , while the observed paralysis might be linked to the neuromuscular role . At cholinergic synapses , the neurotransmitter ACh binds to nicotinic ACh receptors ( AChRs ) and thereby mediates muscular contraction via membrane depolarization . AChE plays a central role in the termination of transmission by hydrolysis of ACh . Treatment of adult schistosomes with either ACh agonists or inhibition of AChE ( resulting in higher levels of ACh ) led to excessive stimulation , desensitization , and flaccid paralysis of worms [56 , 58] . If harmonine inhibits AChE , this should likewise induce paralysis . Indeed , this was observed in our study . To clarify whether the antischistosomal effects of harmonine are mainly due to targeting of AChE , a gradual knock-down of AChE expression to a similar degree as found for the reduction of AChE activity by harmonine might be of interest for future work . While we found a generally high expression and enzymatic activity of AChE in males and unpaired females , both parameters were strongly decreased in paired females . This might reflect a reduced need in motor activity of the female after pairing and/or a different need for AChE-mediated glucose uptake . Up-and down-regulation of genes in females after pairing and separation , respectively , is a known phenomenon [59 , 60] . We also found other genes involved in the function of the neuromuscular synapse being downregulated , such as genes annotated as nicotinic AChR ( Smp_176310 , Smp_139330 , Smp_180570 ) and as choline acetyltransferase ( Smp_146910 ) , the enzyme involved in ACh synthesis [39] . Altogether , it seems likely this contributes to a reduced motor activity of females after pairing . Our findings of a sex- and pairing-dependent AChE expression and activity pattern adds novel aspects to the wealth of data on AChE function in schistosome biology , and might explain why females appeared less sensitive than males towards harmonine during the first 24 h of treatment . The druggability of AChE was demonstrated by the use of the AChE inhibitor metrifonate against S . haematobium in the past [35] . However , metrifonate was withdrawn from the market because of the need for multiple doses , its toxicity to the host , and its unsatisfying efficacy against other schistosome species [61] . Nonetheless , AChE remains an interesting antischistosomal target for rational drug design . That species-specific optimization is achievable was for instance demonstrated by re-engineering inhibitors toward higher specificity for Anopheles AChE than human AChE [62] . Also for harmonine , rational drug design might be used to obtain even better efficacies . Our conclusion that there are further harmonine targets beyond AChE is based on three observations: ( 1 ) the merely weak AChE inhibitory capacity of harmonine , ( 2 ) the multifaceted phenotypes observed after treatment with harmonine , and ( 3 ) effects by the AChE inhibitor metrifonate that were different in strength involving a rapid paralysis within only 1–3 hours , which was reversible [63] . Additional targets might be related to stem-cell biology . Due to the fundamental role of stem cells in the parasite life cycle and parasite survival , it was recently proposed that many helminth infections may be considered as stem cell diseases [45 , 64] . Compounds targeting stem cells are thus particularly attractive candidates for drug development . In order to not only assess but objectively quantify stem cell effects , we established a 3D reconstruction approach of schistosome ovaries and testes to quantify proliferating stem cells . Harmonine significantly reduced the number of proliferating stem cells in gonadal tissues and the parenchyma where they are known as neoblasts [44] . Furthermore , we used for the first time quantification of nanos gene expression as a measure for compound-induced effects . Nanos-1 was characterized as a germline-specific stem cell marker in adult S . mansoni [42 , 65] and is known as a conserved regulator of germ cell development [43] . Somatic stem cells , so-called neoblasts , were only recently identified in adult schistosomes and are characterized by expression of nanos-1 and nanos-2 [44] . Nanos-2 is a post-transcriptional regulator responsible for the formation , development , and maintenance of pluripotent cells in many metazoans [66] . Knock-down of nanos-1 and nanos-2 in S . mansoni resulted in loss of mature germ cells and in degenerated testes , and it was concluded that nanos-1 and nanos-2 are required for germ-cell differentiation [42] . Since the levels of nanos expression were reduced upon harmonine treatment , one could speculate about a reduction of total stem cell numbers by induction of cell death . However , it is not clear whether nanos expression might also be affected in alive but cell cycle-arrested stem cells . Therefore , it remains open whether harmonine induces cell-cycle arrest or cell death in gonadal and somatic stem cells . In the case of cell-cycle arrest , removal of harmonine before EdU addition to the culture would lead to a resumption of cell proliferation and may clarify this question . AChE as the target of harmonine in schistosomal stem cells is another attractive hypothesis . Indeed , according to previous RNAseq analyses , transcripts of a variety of ACh receptors were found in ovaries and testes of S . mansoni [39] , suggesting a non-neuronal role of the AChE-ACh-AChR axis in gonads . In addition , AChE is known to be expressed in certain stem cells where it regulates cell proliferation . For instance , embryonic stem cells of human and mouse express AChE and showed reduced proliferation upon ACh stimulation [67 , 68] , just like an ACh over-abundance by inhibition of AChE activity might do . Furthermore , inhibition of AChE by organophosphates in mesenchymal stem cells or by donezepil in neural stem cells led to suppression of proliferation , differentiation and self-renewal ability [41 , 69] . Therefore , future studies should address the possible role of ACh signaling in schistosomal stem cells in detail , and we propose RT-qPCR analysis of sorted stem cells as a start . This might reveal new insights into AChE function in schistosomes and aid in developing novel therapies . As a side note , it appears biologically astonishing how H . axyridis is able to resist the toxic potential of harmonine which accumulates in the hemolymph up to 7 μg per mg body mass [5] . This resistance is even more remarkable seeing the potential of harmonine to inhibit such crucial processes like stem cell proliferation , and will fuel the curiosity in future research . Alkaloids are natural compounds that are widespread in plants , bacteria , fungi , and animals . In vitro schistosomicidal activity was demonstrated for several plant-derived alkaloids , such as piplartine from Piper species [2 , 3] . Harmonine is the first animal-derived alkaloid found to have antischistosomal capacity . Notably , the chemical structure of the alkaloid harmonine , an aliphatic long-chain diamine , is quite different to these other active alkaloids . To date , only very few other studies addressed animal-derived compounds and their antischistosomal activities . These are compounds from sea cucumbers , skin secretions of a South-American tree-frog species , and undefined compounds from snake venom [3 , 70–72] . From insects , so far only two complex product mixtures , but no defined compounds , were described to have direct or indirect anthelminthic effects . Propolis ( “bee glue” ) is a complex resinous bee hive product and was shown to have various anti-protozoal [73] , fungicidal , and antimicrobial properties [74] . Furthermore , in vivo treatment of S . mansoni-infected mice with propolis produced by Apis mellifera reduced the worm burden but increased granuloma diameter [4] , which pointed to an immune-modulatory effect rather than a direct effect on worm vitality . Unfortunately , there are no data in the literature on in vitro-culture experiments that demonstrate any direct effects of propolis or propolis-derived compounds on schistosomes . The second anthelmintic insect-derived product described in literature is bee venom . Bee venom is a complex mixture of enzymes , biogenic amines and peptides with described anti-inflammatory capacity and has been studied as an alternative medicine for inflammatory diseases and cancer [75 , 76] . As with propolis , a reduction of worm burden was found upon treatment of S . mansoni-infected mice [4] , but unfortunately without analysis of direct effects , mode of action or possible schistosomal targets . Thus harmonine and its effects on AChE activity and stem cell gene expression provides the very first insight into possible mechanisms of action of an insect-derived compound . In view of the broad in vitro activity of harmonine found against microbes [7 , 10] , protozoans [7 , 12] and helminths ( this study ) , it is tempting to promote this insect-derived compound as a novel universal weapon , a kind of swiss-army knife against multiple pathogens . However , we found first evidence of cytotoxicity against HepG2 cells at concentrations of 50 μM and higher , indicating that harmonine in its current form is not yet an ideal lead candidate , but certainly a valuable basis for structure/activity-based compound development . The EC50 of 8 . 8 μM obtained for the reduction of schistosome motility by harmonine is already a promising start , as this concentration did not lead to any cytotoxic effects against HepG2 cells . In summary , this study provided clear evidence for the antischistosomal activity of the lady beetle-derived compound harmonine together with biologically highly interesting effects on AChE activity , inhibition of stem-cell proliferation and gene expression . This is the first time to proof a direct effect of a defined insect-derived compound on schistosomes , and harmonine may serve as basis for the development of new antischistosomal , or even broader antiparasitic compounds . This study highlights the potential of exploiting insects for the discovery of anthelminthics and motivates the screening of insect compound libraries for novel anthelminthic compounds in the future . | Natural compounds represent one of the richest sources for the discovery of new active compounds against diseases such as cancer or infections , including helminth infections that cause the highest disease burden in tropical countries . Surprisingly , insects have been almost completely neglected with respect to anthelminthics discovery although they represent the most species-rich class of animals known on earth , producing a wide spectrum of compounds with biological activities . In insect biotechnology , the harlequin ladybird Harmonia axyridis raised high interest being a rich source of antimicrobial compounds such as the alkaloid harmonine . Harmonine is thought to act as a chemical weapon keeping otherwise detrimental microsporidia in the beetle under control . Testing the antiparasitic potential of harmonine against adult Schistosoma mansoni , one of the most harmful helminths worldwide , resulted in multifaceted negative effects . The compound damaged tissues essential for survival and reproduction of schistosomes ( tegument , intestine , gonads ) and also affected stem-cell proliferation . Furthermore , we obtained first evidence for acetylcholinesterase as one potential molecular target , which was partially inhibited by harmonine . This is the first time to proof a direct effect of a defined insect-derived compound on a helminth parasite , a finding that will encourage further studies to explore insects as sources of novel anthelminthics . | [
"Abstract",
"Introduction",
"Materials",
"&",
"methods",
"Results",
"Discussion"
] | [
"schistosoma",
"invertebrates",
"schistosoma",
"mansoni",
"medicine",
"and",
"health",
"sciences",
"alkaloids",
"reproductive",
"system",
"chemical",
"compounds",
"helminths",
"light",
"microscopy",
"animals",
"confocal",
"laser",
"microscopy",
"microscopy",
"confocal",
"... | 2019 | Insects in anthelminthics research: Lady beetle-derived harmonine affects survival, reproduction and stem cell proliferation of Schistosoma mansoni |
Widely used chemical genetic screens have greatly facilitated the identification of many antiviral agents . However , the regions of interaction and inhibitory mechanisms of many therapeutic candidates have yet to be elucidated . Previous chemical screens identified Daclatasvir ( BMS-790052 ) as a potent nonstructural protein 5A ( NS5A ) inhibitor for Hepatitis C virus ( HCV ) infection with an unclear inhibitory mechanism . Here we have developed a quantitative high-resolution genetic ( qHRG ) approach to systematically map the drug-protein interactions between Daclatasvir and NS5A and profile genetic barriers to Daclatasvir resistance . We implemented saturation mutagenesis in combination with next-generation sequencing technology to systematically quantify the effect of every possible amino acid substitution in the drug-targeted region ( domain IA of NS5A ) on replication fitness and sensitivity to Daclatasvir . This enabled determination of the residues governing drug-protein interactions . The relative fitness and drug sensitivity profiles also provide a comprehensive reference of the genetic barriers for all possible single amino acid changes during viral evolution , which we utilized to predict clinical outcomes using mathematical models . We envision that this high-resolution profiling methodology will be useful for next-generation drug development to select drugs with higher fitness costs to resistance , and also for informing the rational use of drugs based on viral variant spectra from patients .
The rise of drug resistance to antimicrobial agents , frequently a consequence of acquiring mutations de novo that confer resistance , causes failure of current infectious disease treatments and results in continued economic burden [1] . Resistance development is an evolutionary process , often depending on the combined effects of fitness cost and resistance gain of the associated mutations [2] , [3] . It is of paramount importance to systematically explore the evolutionary dynamics of infectious pathogens to assess the likelihood of resistance breakthrough during drug development . In this study , we utilize Hepatitis C virus ( HCV ) and a potent antiviral compound ( Daclatasvir ) as a working model to illustrate the application of a quantitative high-resolution genetic ( qHRG ) platform to interrogate the Daclatasvir-resistance profile of the virus . Persistent infection with HCV is a major cause of human liver damage in over 3% of the world's population , and consequently , these patients are at risk of developing advanced liver diseases and liver cancer [4] . The long-standard treatment for HCV infection is a combination of ribavirin and PEGylated interferon ( PEG-IFN ) , which activates the immune system and thus causes severe side effects [5] . Ever since the discovery of HCV almost 25 years ago , enormous effort has been devoted to understanding the replication life cycle of the virus and developing effective direct-acting antiviral ( DAA ) drugs with the goal of reducing its global health impact . In 2011 , the first two protease inhibitors were approved and used in combination with standard treatment [6] , [7] . Although the viral response rate in patients has markedly improved with the addition of the protease inhibitors , the efficacy of this new therapeutic regimen is observed to be highly dependent on HCV genotype . Moreover , the emergence of resistant mutations further hinders its application [8] , [9] and creates a demand for more effective treatment options . The establishment of the HCV replicon cell [10] , [11] and infectious [12]–[14] systems have paved the way for high-throughput screening of small-molecule inhibitors , and thereby aided the identification of many new classes of antiviral compounds [15] , [16] . However , the ability to systematically define mechanisms of action and determine the genetic barriers of promising compounds poses unmet challenges [17] , [18] . Previously , a chemical screen has identified potent antiviral compounds that target the HCV protein NS5A [19] , [20] , which is a non-enzymatic protein but is indispensable for viral genome replication , viral assembly , and innate immune evasion [21] . The mechanism of action and binding site of many NS5A inhibitors , however , remain unknown . The NS5A inhibitor Daclatasvir was identified as a potent antiviral agent that blocks viral replication at both the genome replication and viral assembly stages [22] . It possesses a potent antiviral activity in cell culture , with a half maximal effective concentration ( EC50 ) in the pico molar ( pM ) range and a cytotoxic concentration in the micro molar ( uM ) range , yielding a large potential therapeutic window [20] . Daclatasvir has also been reported to alter the localization of NS5A [23] , but the mechanism of drug-protein interactions is under investigation and not yet fully understood [22] . Moreover , considering the fast replication rate and error-prone RNA polymerase of HCV , drug escape mutants are expected . Therefore , a systematic investigation of the evolutionary fitness and Daclatasvir sensitivity of all possible variants is imperative to better understand the mechanism of drug action and to design second-generation compounds with higher genetic barriers for resistance . Previous studies passaging wild type ( WT ) HCV clones in the presence of Daclatasvir identified resistant mutations within the NS5A domain IA ( DIA ) , suggesting an interaction between the drug and this region [19] , [20] , [24]–[26]; however such studies are limited by the breadth of genetic variability and can only identify positively selected mutations . Here we have implemented a qHRG platform to simultaneously quantify the effects of all possible single mutations in NS5A DIA on relative replication fitness and sensitivity to Daclatasvir . Our dataset includes all NS5A DIA mutations that both increase and decrease drug sensitivity at any magnitude and therefore provides an informative framework for identifying residues and key chemical contacts that may be involved in protein recognition of the drug . Quantitative analysis of the altered drug sensitivity for each mutant enabled determination of residues mediating drug-protein interactions . Complete analysis of drug sensitivity for all possible single amino acid variants also greatly clarifies the differences in Daclatasvir sensitivity among different HCV genotypes . Moreover , the relative fitness and drug sensitivity profiles generated by qHRG were further utilized to predict genotype specific clinical outcomes using mathematical models . We anticipate that the qHRG approach will be generally applicable to studying other virus-host interactions or drug-protein interactions to understand the underlying mechanisms of drug action .
To systematically map the fitness landscape and assess the drug-resistance profile of NS5A DIA , saturation mutagenesis techniques were used to introduce randomization at each codon position ( amino acids 18–103 ) thereby covering all possible amino acid substitutions . To do this , we substituted each codon individually using synthetic template oligonucleotides that contained 3 continuous random nucleotides ‘N1N2K’ at the codon of interest , where N1 and N2 represent random incorporation of A/T/G/C , and K represents random incorporation of T/G . The randomized codons therefore include 32 nucleotide combinations , allowing representation of all possible amino acids at each position ( Fig . 1A ) . The input DNA library ( pool 0 , Fig . 1B ) was isolated from more than 21 , 000 bacteria colonies to ensure coverage of all possible mutations in the pool 0 ( Table S1A ) . The viral library ( pool 1 ) was reconstituted and subsequently subject to selection in a human hepatocyte cell line ( Huh-7 . 5 . 1 ) for 4 rounds ( pool 2_control through pool 5_control ) of infection at a low multiplicity of infection ( MOI ) ( Fig . 1B , Table S1B ) . Pools labeled “control” indicate replication without the presence of drug . The fitness of each mutant , reflected by the change in frequency of each mutation , was determined by Illumina paired-end sequencing , allowing for the identification and quantification of all mutants with high confidence [27] . From the 23 . 9 million sequence reads that passed quality filtration , each mutant virus was sequenced approximately 1200 times on average , which enabled precise quantification of mutant frequencies . The WT fraction served as an internal benchmark to determine the relative fitness of each mutant . The relative fitness score ( W ) of each mutant was determined by regression analysis of the mutant frequency relative to WT through 5 rounds of selection [28] . The selection coefficient ( s ) of each variant was also calculated ( Methods ) . Interestingly , as shown in figure 2B , the selection coefficients calculated from 5 rounds of selection ( pool 1 and pool 2–5_control ) strongly correlate with those calculated from 3 rounds of selection ( pool 1 , pool 2_control and pool 3_control ) , suggesting that 3 rounds of passaging is sufficient to determine the phenotype of each variant . The selection coefficients ( s ) representing the difference in fitness between all variants and WT are displayed in a heat map representation ( Fig . 2A ) . In agreement with the critical functions of NS5A required for viral replication , stop codons are not tolerated at any position of the region ( Fig . 2A ) , which demonstrates the effectiveness of our selection assay and its reliability in measuring changes in frequency . To verify the accuracy of our fitness profiling method , 16 mutant viruses that span the range of all phenotypes and span a range of functional and structural motifs were constructed on a monocistronic Renilla luciferase HCV reporter virus background ( FNX24_RLuc ) . A reporter virus defective in RNA polymerase activity ( NS5B_GNN contains a double mutation within the RNA-dependent RNA polymerase motif of NS5B that converts GDD into GNN ) served as a negative control [29] and WT as a positive control . The individually determined selection coefficients show strong correlation at high confidence with the profiling data ( Fig . 2C , D ) , demonstrating the accuracy of fitness measurements from the qHRG profile throughout a large dynamic range . The fitness effects enable fine mapping of sequence-function requirements at each position . For example , the N-terminus forms an amphipathic membrane-binding α-helix and we observe sequence requirements in agreement with the three distinct faces ( hydrophobic , acidic , and polar/non-acidic ) as determined by NMR structural analyses ( Fig . 2C , D , 3A ) [30] . Strict sequence requirements at positions within this helix may indicate that this region contributes to the localization of NS5A [31]–[34] . Continuing this trend , the unresolved proline-rich linker region displays a requirement for the sequence KXΦPXΨPGΨP . We illustrated the NMR model of the helix [30] in combination with the linker region modeled as the ubiquitous poly-proline type II helix recognition motif ( Fig . 3A ) [35] , [36] . Our profiling method also identifies strict structural requirements for the unique NS5A DIA fold . In line with an essential structural property [37] , four zinc-associated cysteine residues [37] do not tolerate any substitutions ( Fig . 2A , 3C ) . Furthermore , residues 87–91 encompass a conserved segment that passes through the core of the protein ( Fig . 3D ) which includes a buried polar residue , N91 , that does not tolerate any substitutions . F88 and P89 are also absolutely required , while T87 and I90 only tolerate highly similar residues ( Fig . 2A ) . This analysis may suggest potential targets for peptide-based vaccine designs . We also validated that N82 is absolutely essential as even glutamine at this position is lethal ( Fig . 2A , C ) . N82 is buried and participates in hydrogen bonds with both T87 and the Q40 main chain ( Fig . 3D , Fig . S1C ) highlighting the sensitivity at this position to side chain geometry . The complete mutagenic analysis of each residue in NS5A DIA also provides in-depth insight into potential molecular recognition surfaces . Although NS5A was previously reported as an RNA-binding protein , the exact binding sites remain to be determined [38]–[40] . Our fitness landscape profile shows that position K41 ( R41 in genotype 1b ) and K44 exhibit a consistent requirement for WT-like fitness: although these positions tolerate a diverse set of substitutions , acidic residues are lethal at both positions . Substitution with an acidic residue ( K41D ) results in a defect for genome replication ( Fig . 3E ) . These positions were located in a basic groove large enough to accommodate RNA in one of the dimerized structures [37] ( Fig . S2A ) , providing evidence that these two positions may be important for the RNA binding function of NS5A . To quantify changes in drug sensitivity for all non-lethal mutants and to interrogate the drug-protein interaction surface , we passaged the mutant virus libraries under Daclatasvir treatment ( 20 pM ) for 2 rounds ( Fig . 1C ) . The relative fitness score of each mutant in relation to WT under drug treatment ( Wdrug ) was determined and the selection coefficients are displayed in a heat map ( Fig . S3 ) . Next , we calculated the fold change in relative fitness score ( Wdrug/Wcontrol ) and the resulting drug sensitivity profile is presented as a heat map in figure 4A . The data show that mutations at positions 28 , 31 , 38 , 92 , and 93 are noticeably enriched upon drug selection , suggesting that mutations at these positions confer resistance to the drug . In contrast , the fitness values of variants at positions 21 , 56 , and 58 are significantly diminished due to increased drug sensitivity . The changes in variant fitness at positions 24 , 30 , 62 , and 75 are highly dependent on the property of the substituted amino acids . We validated the drug sensitivity profile by constructing 10 variants for determination of the Daclatasvir EC50 ( Fig . S4A , B ) and demonstrated an excellent correlation to their drug sensitivity ( Fig . 4D ) in the profile . The data are also consistent with critical drug-interacting residues identified in previous adaptation experiments [19] , [20] , [24]–[26] . In addition to corroborating previous results , our profiling method also uncovered new resistance determinants at position 24 and 56 where no resistant mutations were previously identified by virus adaptation studies . Thus , this approach enables identification of all residues participating in the drug interaction network . As a result of our saturation mutagenesis approach , we also identified highly resistant substitutions that were not previously identified at known resistance-conferring positions , including F28C , K30Y , L31I and Y93W . These include substitutions that may require a two-step mutational path in the genetic backgrounds utilized for previous adaptation ( e . g . L31I requires two nucleotide changes and has not been observed in genotype 1a , 1b or 2a ( JFH1 ) , while M31I only requires one nucleotide change and its breakthrough has been detected in genotype 4a ( ED3 ) [26] ) . In this study , although we analyzed the Wdrug of each variant from two rounds of selection ( pool 2_drug and pool 3_drug in Fig . 1C ) to determine the drug sensitivity , the strong correlation of Wdrug between round one and two ( Fig . 3C ) suggests that a single round of selection is sufficiently informative . The completeness of our drug sensitivity profile also clarifies the observed differences in Daclatasvir sensitivity among different genotypes recently reported by Scheel et al . [41] ( Fig . 5A ) . For instance , the difference between the two 1a strains is likely due to an increased sensitivity of viruses with Q at position 30 compared to H ( Fig . 5A ) . Additionally , our profiling data provides insight into why certain strains that naturally carry residues previously demonstrated to confer resistance in other genetic strains remain highly sensitive to the drug . M31 is a relatively resistant residue according to the drug sensitivity profile and also has been demonstrated to confer significant resistance when acquired in genotype 1a or 1b [24] , [42] . However , M31 is the WT residue in the sensitive strain 4a ( ED3 ) , suggesting that there are other critical positions involved . Our data indicate that the identity of residue 30 and 56 may contribute to this property as L30 and T56 are much more sensitive than R/K30 and R56 . H93 is another high resistance-conferring mutation that is naturally carried in a sensitive genotype 7a strain . This can be explained by the extreme sensitivity of the serine residue at position 30 ( Fig . 5A ) . Likewise , our drug profile shows that T93 , found in genotype 5 and 6 , is much more resistant than Y93 in genotype 1–4 , but this may be largely negated by Q30 in 5a or T56 in 6a . Interestingly , the additive total effect of drug sensitivity from these residues approximates the experimentally determined EC50 for 9 out of 10 strains [41] ( Fig . 5A ) . The outlier , genotype 3a ( S52 ) , carries residues A30 and E92 which are two positions that have previously been shown to be genetically linked [24] , indicating that some of the interactions are epistatic . All of these analyses demonstrate that the qHRG approach developed in our study can quantitatively determine the drug sensitivity of all point mutations , including the equally informative negatively selected mutations that would not be identified in adaptation selection studies . The results presented here systematically map the entire panel of drug-sensitivity determinants and predict how substitutions will impact HCV replication upon Daclatasvir treatment . Measurement of drug sensitivity and fitness of all possible point mutations maps the evolutionary space of the virus upon drug treatment and also provides comprehensive information to predict the clinical outcome of single mutations resistant to Daclatasvir treatment for genotype 2a . We combined previous mathematical models of viral evolution [22] , [43]–[45] to assess the probability of emergence for all resistance mutations identified in our screen ( Fig . 6 ) . The model shows that some substitutions , including L31I , F28T/C and Y93W , can cause failure of Daclatasvir monotherapy even with perfect treatment adherence . By incorporating parameter uncertainties into the analysis , we show that several other resistant mutants have substantial probabilities of arising ( Fig . 6 and S5 ) . If treatment is imperfect , many other escape mutants pose a risk of emergence , particularly if consecutive drug doses are missed . Some resistant mutations , such as Y93W , require two nucleotide changes from the WT sequence , and therefore are less likely to arise via natural adaptation from a single clone . Combination therapy is essential to prevent the growth and dissemination of resistant strains within or among infected individuals [46] , and patients whose viral populations contain these mutations at high frequency would not benefit from Daclatasvir monotherapy . Thus , exploring the fitness and drug sensitivity of all possible single-mutant variants with saturation mutagenesis , combined with mathematical modeling , allows for risk assessment of possible evolutionary routes for the emergence of drug resistance . The approach shown here , combined with deep sequencing of clinical samples , would enable rational design of patient-specific combination therapies to minimize the threat of de novo resistance .
Modern molecular medicine has enormously accelerated the rate of developing novel therapeutics . For HCV , the establishment of in vitro replication systems has facilitated discovery of novel compounds that inhibit virus replication in addition to designed NS3 and NS5B inhibitors [18] , [47] . This progress , however , creates a demand for more efficient methods to identify the inhibitory mechanisms of novel therapeutics as well as the determination of genetic barriers to resistance . The HCV-encoded non-enzymatic NS5A protein is a new target identified through chemical genetic screens [17] . Using the NS5A inhibitor Daclatasvir as an example , we have developed the novel qHRG method for systematically profiling drug-protein interactions . Unlike conventional virus adaptation studies where WT virus is passaged under drug treatment to positively select resistant mutations , we are capable of quantifying the drug sensitivity and fitness cost of all possible single amino acid mutations , thereby identifying the entire set of positions that govern particular drug-protein interactions . The profile of mutant fitness and drug sensitivity can be informative for the patient-specific use of Daclatasvir and may facilitate the development of second-generation drugs with higher genetic barriers . NS5A is a multifunctional protein essential for several stages of HCV viral replication . It is a membrane-associated protein through an N-terminal amphipathic α-helix ( amino acids 1–25 ) , followed by an unstructured linker region ( amino acids 26–35 ) and three functional domains . The N-terminal domain ( domain I , amino acids 36–198 ) is the only structured domain [37] , [48] and is targeted by a new class of direct-acting antivirals , such as Daclatasvir [20] . Although NS5A domain I appears to be a key regulator of viral replication [49] and is an effective drug target , its functional mechanisms are largely undefined . The fitness landscape reveals novel strict requirements at positions critical for protein function . Projection of the fitness effects of mutations onto the structure illustrates the sequence-structure-function relationships , including the amino acid property requirements in the N-terminal amphipathic helix , the proline-rich unstructured linker region ( residues 26–35 ) , the conserved stretch of buried residues ( 87–91 ) , and the putative RNA-binding residues essential for viral replication . Together , these results have demonstrated the specific residues critical for the multifunctional roles of NS5A in viral replication . Since the discovery of the potent NS5A inhibitor , Daclatasvir , numerous long-term adaptation studies have been employed to identify drug resistant mutations and gain insight into the mechanisms of drug action [19] , [20] , [24] , [26] . However , the positively selected drug-adaptive resistant clones identified previously only represent a fraction of the mutations that confer resistance . Furthermore , mutations that lead to hypersensitivity to Daclatasvir , which are equally informative , could not be identified in previous adaptation studies . Through qHRG analysis , we have quantitatively determined the drug sensitivity of all possible single amino acid variants in the domain IA of NS5A and thus provided a comprehensive understanding of the positions governing drug-protein interactions . Structural analyses of the NS5A domain I previously revealed simultaneous existence of two different homodimer arrangements with non-overlapping interfaces on the opposite side of each monomer [37] , [48] . Interestingly , in both forms of the dimer structure , the drug sensitivity-determining residues are located away from the dimerization interface , which supports previous results showing that the drug does not interfere with the dimerization of NS5A [23] . Instead , these residues cluster on the surface of domain 1A ( Fig . 5B , C ) and the unstructured linker region ( amino acids 26–35 ) that connects the N-terminal amphipathic helix with the core of domain I . The existence of the two possible dimerization interfaces on the opposite side of the monomer has led to the prediction of superhelical array organization where the monomer is polymerized through alternative interfaces [48] . It is therefore possible that interaction with the drug may induce a conformation shift that disrupts the protein oligomerization , which prevents newly synthesized NS5A from facilitating replication complex formation . Although the oligomerization hypothesis is supported by the extremely low working concentration of the drug even in replicon cell lines that homogenously harbor highly active replication complex , it does not exclude the possibility that the drug directly competes with cellular or viral factors for NS5A binding , and as a consequence , abolishes membranous web formation . In fact , a recent study suggests that the drug also affects genome replication in addition to virus assembly [22] possibly through inhibiting the function of NS5A that is required for genome replication . Several studies have shown that NS5A interacts with many host factors to hijack their cellular functions for facilitating viral replication . PI4KIIIα is a phosphatidylinositol 4-kinase identified as a host kinase targeted by NS5A and relocated to HCV replication complex [50]–[54] . Co-immunoprecipitation of NS5A deletion mutants and PI4KIIIα mapped the interaction to domain I [52] , [53] , and this interaction is critical for regulating the phosphorylation status of NS5A [55] . It is possible that the drug affects the interaction between NS5A and PI4KIIIα , and therefore obstructs genome replication . However , more direct evidence will be needed to elucidate the detailed mechanism . This systematic profiling approach is a combination of forward and reverse genetics , which we refer to as quantitative high-resolution genetics ( qHRG ) ( Fig . 7 ) . It fully utilizes recent advances in DNA sequencing capacity to accurately quantify the fitness of individual variants in a large and diverse population , thereby determine the phenotypic effects of each genetic modification within a single experimental platform . While this platform is limited to single mutant profiling , we found that the sum of individual mutational effects predicted multiple mutant sensitivity/resistance in 9 of 10 alternative genotypes reported by Scheel et al . [41] . Importantly , our ability to characterize mutations that led to increased drug sensitivity was essential in the prediction of multiple mutational effects . Furthermore , the simplicity and rapid time scale associated with this study enables investigation of mutational profiles in alternative mutant backgrounds . For example , if studying a protein-drug interaction system where all resistance mutations confer fitness costs , then qHRG could be used to conduct a secondary screen of a mutant library constructed on the background of the resistant mutations to investigate the existence of compensatory mutations and their likelihood of evolving . Although our approach cannot identify context-dependent resistant mutations , i . e . those that are resistant only in the presence of an additional mutation , this study still identified most common mutations that contribute to escape in clinical specimens . Importantly , the major breakthrough amino acid substitutions conferring resistance in clinical trials were revealed by our qHRG approach ( e . g . positions M28 , Q30 , L31 and Y93 for genotype 1a and L31 and Y93 for genotype 1b ) [42] , [56] . Interestingly , the vast majority of multiple-mutation variants in clinical specimens revealed by clonal sequencing analysis are combinations of these single amino acid substitutions [57] . Therefore , the systematic analysis of drug sensitivity/resistance of single mutations provides a reference for the viral response to Daclatasvir in HCV patients as analyzed by population sequencing . Moreover , likely due to the higher heterogeneity of natural HCV sequences in vivo compared to experimental replicons , the resistant mutation patterns in clinical specimens are more complex than those in vitro [57] . For example , some of the resistant mutations found in clinical samples were not previously detected in vitro ( e . g . M28G , Q30G ) . However , the ability for these mutations to confer resistance was revealed by our complete drug profile ( Fig . 4A ) . This approach offers a quantitative view of the genetic barriers for all possible single amino acid changes during viral evolution , which can be linked to clinical outcome predictions using mathematical models . A profile of drug resistance may inform the rational use and combination of Daclatasvir based on viral mutant spectra from patients . With the continual reduction in cost and improvement in detection sensitivity of next-generation sequencing technology , direct sequencing of clinical specimens before treatment will reveal virus quasispecies in patients , ensuring timely diagnosis of pre-existing drug-resistant strains , enabling design of optimal therapeutic strategies for individuals . It will also enable monitoring of emergence of resistant strains during treatment to prevent the enrichment and spread of resistance . In addition , comprehensive mapping of genetic barriers to drug resistance will facilitate the development of second-generation drugs with higher fitness cost to resistance .
The Huh-7 . 5 . 1 cell line was kindly provided by Dr . Francis Chisari from the Scripps Research Institute , La Jolla . The cells were cultured in Dulbecco's Modified Eagle Medium ( DMEM , Invitrogen ) supplemented with 10% of fetal bovine serum ( FBS ) , 10 mM non-essential amino acids ( Invitrogen , Carlsbad , USA ) , 10 mM HEPES , penicillin ( 100 units/ml ) , streptomycin ( 100 mg/ml ) , and 2 mM L-glutamine at 37°C with 5% CO2 . A plasmid of pFNX-HCV that carries the viral genome was synthesized in our lab based on the chimeric sequence of J6/JFH1 virus [13] . We introduced 7 nucleotide substitutions , resulting in synonymous mutations to the genome to distinguish this construct . The construct and sequence are available upon request . The area to be mutated ( 86 amino acids ) was divided into 5 small regions , with each composed of 17–18 amino acids . For each region , 17 ( or 18 ) oligos , each of which contains one random codon ( N1N2K , N1 and N2 code for A/T/G/C and K codes for T/G to ensure the coverage on every amino acid and minimize the possibility of getting stop codons ) at the desired position were synthesized from IDT . This mutagenesis results in all possible amino acid substitutions at a given position to facilitate the functional exploration of each possible variant . The oligos each contain a BsaI recognition site on each end , which allows the generation of “sticky ends” to match the ends of each cassette . The cassettes were established by amplifying the fragments ( from pFNX-HCV ) flanking the desired mutation region with primers containing a BsaI recognition site , and digested with BsaI enzyme ( NEB ) to produce the “sticky ends” matching the oligos . The oligos and the cassettes were ligated with T4 DNA ligase ( Invitrogen ) overnight at 16°C and purified with PCR purification columns ( Invitrogen ) . The ligated product was subcloned into the pFNX-HCV vector via BamHI and RsrII restriction sites and transformed . Approximately 50 , 000 colonies were collected for the library in total . Each library covers all possible mutations at approximately 50 fold . The HCV NS5A inhibitor BMS-790052 was purchased from company Selleck Chemicals . The mutant virus library ( 12 ml ) was used to infect naïve Huh-7 . 5 . 1 cells ( 4million ) at M . O . I ∼0 . 2 with or without BMS-790052 treatment at 20 pM . The supernatant was collected at 144 hpi and used to infect naïve cells for the second round of selection . After two rounds of selection , the viral genome was recovered from the supernatant , and the mutated regions were PCR amplified and processed following the standard sample preparation protocol for HiSeq 2000 sequencing . Each library was tagged with a unique 6-bp molecular barcode sequence , which allows for the identification and study of relative fitness levels in each selection pool . The sequence of primers and barcodes can be found in the supplementary materials . Burrow-Wheeler Aligner was used to map the pair-end read by allowing 5 mismatches with a Q30 cutoff . Since both forward and reverse reads covered the whole randomized region , sequencing error was corrected by reads pairing . SAMtools and BamTools were employed for sequence analyses . Custom Python script was created for the other downstream data analyses . After determining the number of sequence reads ( Reads ) for each mutant , we then calculated the frequency of each mutant from each pool and the fitness score in relation to the WT . Any frequency that is lower than 0 . 0005 will be considered as noise and discarded , since the mutation frequency of HCV is about 10−5 to 10−4 nucleotide substitutions per nucleotide per round of genome replication [44] . The frequency of a given variant , v , in the pool #N ( ) and the frequency of WT , wt , in the pool #N ( ) were calculated as follows:where indicates the number of sequence reads for the variant ( v ) in pool #N , shows the number of sequence reads for the WT in pool #N , and represents the total reads in the pool #N . The relative fitness score of a given variant ( ) was determined as the antilogarithm of the slope of the regression:where is the logarithm of the relative frequency of a given variant ( v ) in the input library ( pool 0 ) . The relative fitness score of each variant in drug treatment was calculated in the same way , but only with 2 rounds of selection in 20 pM drug treatment . We then calculated the selection coefficient ( ) :To examine the essentialness of each position , we also calculated the fold change of mutations at each position , which was used to color code the protein structure . The fraction at the ith position ( bearing jth amino acid substitution , 19 total ) in the pool N ( ) wasThe fold change of mutations in pool #N compared to the input ( pool 0 ) ( ) was: In total 16 individual mutant viruses containing point mutation as indicated in figure 2C were reconstructed based on a monocistronic Renilla luciferase HCV reporter virus , FNX24-RLuc , and recovered by electroporating the viral RNA genome into Huh-7 . 5 . 1 cells . The supernatant of transfected cells was collected and subjected to infect naïve Huh-7 . 5 . 1 cells for 2 passages . The replication of the viruses in each round was assayed by measuring the Renilla luciferase activity in the infected cells with Renilla Luciferase Assay System from Promega . 10 individual mutant viruses , with a wide range of drug sensitivities in the screen , were reconstructed in order to determine their EC50 . Each mutant , containing a point mutation as indicated in figure S4A , was constructed on the Renilla reporter virus background ( FNX24-Rluc ) . The RNA was transfected into Huh7 . 5 . 1 cells to reconstitute the viruses , which were treated with different concentrations of Daclatasvir ( 0 , 3 pM , 10 pM , 30 pM , 100 pM , 300 pM , 1000 pM , 3000 pM and 10 nM ) at 96 hours post transfection . After 48-hour treatment of drug , the cells were collected and the replication of viruses in the cells was measured by Renilla Luciferase Assay System from Promega . The EC50 of each mutant was obtained by curve fitting using Prism 6 software . To assess the potential to develop drug resistance for each mutant we considered in this study , we used mathematical models incorporating viral infection dynamics , pharmacodynamics of the drug , and level of drug adherence . We first assessed which mutants are resistant to the drug , i . e . that have potential to grow during long-term treatment . We then calculated the probabilities that those resistant mutants avoid extinction during the initial period of treatment ( when target cells are depleted ) , such that they eventually lead to treatment failure . Please refer the supplementary information for detail . | The emergence of drug resistance during antiviral treatment limits treatment options and poses challenges to pharmaceutical development . Meanwhile , the search for novel antiviral compounds with chemical genetic screens has led to the identification of antiviral agents with undefined drug mechanisms . Daclatasvir , an effective NS5A inhibitor , is one such example . In traditional methods to identify critical residues governing drug-protein interactions , wild type virus is passaged under drug treatment pressure , enabling the identification of resistant mutations evolved after multiple viral passages . However , this method only characterizes a fraction of the positively selected variants . Here we have simultaneously quantified the relative change in replication fitness as well as the relative sensitivity to Daclatasvir for all possible single amino acid mutations in the NS5A domain IA , thereby identifying the entire panel of positions that interact with the drug . Using mathematical models , we predicted which mutations pose the greatest risk of causing emergence of resistance under different scenarios of treatment compliance . The mutant fitness and drug-sensitivity profiles obtained can also inform the patient-specific use of Daclatasvir and may facilitate the development of second-generation drugs with a higher genetic barrier to resistance . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"antimicrobials",
"organismal",
"evolution",
"microbial",
"evolution",
"virology",
"microbial",
"control",
"biology",
"and",
"life",
"sciences",
"microbiology",
"antivirals",
"evolutionary",
"biology",
"viral",
"evolution"
] | 2014 | A Quantitative High-Resolution Genetic Profile Rapidly Identifies Sequence Determinants of Hepatitis C Viral Fitness and Drug Sensitivity |
Zoonotic pathogens respond to changes in host range and/or pathogen , vector and host ecology . Environmental changes ( biodiversity , habitat changes , variability in climate ) , even at a local level , lead to variability in environmental pathogen dynamics and can facilitate their transmission from natural reservoirs to new susceptible hosts . Whilst the environmental dynamics of aquatic bacteria are directly linked to seasonal changes of their habitat they also rely on the ecological processes underpining their transmission . However data allowing the comparison of these ecological processes are lacking . Here we compared the environmental dynamics of generalist and vector-borne aquatic bacterial pathogens in the same unit of time and space , and across rural and urban habitats in French Guiana ( South America ) . Using Leptospira sp . and Mycobacterium ulcerans we performed an environmental survey that allowed the detection of both pathogens in urban vs . rural areas , and during rainy vs . dry weather conditions . All samples were subjected to qPCR amplifications of LipL32 ( Leptospira sp . ) and IS2404 and KR ( M . ulcerans ) genetic markers . We found ( i ) a greater presence of M . ulcerans in rural areas compared with Leptospira sp . , ( ii ) that modified urban environments were more favourable to the establishment of both pathogens , ( iii ) that Leptospira sp . presence was enhanced during the rainy season and M . ulcerans during the dry period , and ( iv ) differences in the spatial distribution of both bacteria across urban sites , probably due to the mode of dissemination of each pathogen in the environment . We propose that in French Guiana simplified and modified urban ecosystems might favour leptospirosis and Buruli ulcer emergence and transmission . Moreover , disease risk was also constrained by seasonality . We suggest that the prevention of aquatic bacterial disease emergence in impoverished urban areas of developing countries would benefit from seasonal diseases targeted surveys , which would maximise limited budgets from cash-strapped health agencies .
During the last decades infectious diseases have considerably increased in incidence and new pathogens have emerged and/or re-emerged [1 , 2] . The majority of known pathogenic species are represented by human pathogens ( 61% ) , and most of these are zoonotic [3] . Zoonotic pathogens are widespread in the environment and often transmit from their abiotic reservoir to wild animals ( biotic reservoir ) but also to domesticated animals and humans ( susceptible hosts ) . Moreover , emerging/re-emerging pathogens are opportunists and respond to changes in host range and/or pathogen , vector and host ecology [3 , 4] . Thus environmental changes , even at a local level , leads to variability in pathogen dynamics in the environment and contributes to changes in the infectious risk [5] . Biodiversity changes through fragmentation and degradation of natural habitats , and particularly in tropical areas , increase contacts between wildlife , domestic animals and humans , facilitating the transmission of environmental pathogens from natural reservoirs to new susceptible hosts [2] . Biodiversity loss is now well recognized to be associated with the increase in emergence of infectious diseases [2 , 6 , 7] . Urbanization and agricultural intensification change land-use , population size and population density , but also impact the interactions between pathogens-vectors-hosts and thus may affect the spread of environmental pathogens [7 , 8] . Also , whilst the exact transmission routes of many tropical diseases remain unclear , variability in climate , even at a local scale , has been reported to affect the prevalence of infectious pathogens in the environment , as well as their transmission dynamics [9 , 10] . Many aquatic bacteria are responsable for major public health concerns , and more importantly in developing countries where access to drinking-water and sanitation is often limited ( for example Vibrio cholerae , Salmonella enterica , Shigella sp . , Leptospira sp . , Mycobacterium sp . , etc . ) [11] . Whilst the environmental dynamics of such pathogens are directly linked to their habitat seasonal changes ( i . e . water temperature , pH , oxygen level , salinity , sedimentation/turbidity , presence of biofilm , rainfall patterns , etc . ) , they also rely on the ecological processes underpining their transmission . For instance , a bacterium directly transmitted from the environment might be more constrained by the local habitat parameters when compared to a bacterium disseminated through a vector , thus depending on the availability , abundance and ecology of that particular vector . However , few studies have described such environmental dynamics in a singular unit of time and space , despite the fact that such description would help to better characterize the infectious risk in the environment ( e . g . urban vs . contryside ) , as well as to better monitor the emergence of infectious diseases . Mycobacterium ulcerans and Leptospira sp . are two pathogenic bacteria found in tropical areas that are accidentally transmitted to humans from the aquatic environment [5 , 8 , 12] . These pathogens are responsible for Buruli ulcer and leptospirosis , respectively , that account for significant morbidity and mortalities among impoverished urban settlements [5 , 8 , 12 , 13] . Whilst M . ulcerans is considered as a generalist pathogen ( i . e . associated with different taxa of the aquatic trophic network ) with no clear transmission routes to humans [8] , Leptospira sp . are transmitted to humans through contact of skin lesions or mucous membranes with contaminated surface water or soil [12] , but is mainly disseminated in the environment via urinary secretions of rodent populations which act as a major reservoir for pathogenic leptospires [5] . Land-use changes ( i . e . deforestation ) in tropical areas were correlated with increased prevalences of M . ulcerans in the environment [14] . Urbanization , associated with increased population density and inadequate sanitation ( precarious sewer systems and trash accumulation ) , favours rodent populations expansion and thus increases leptospirosis risk [5] . Also , local weather patterns are important drivers for both diseases transmission since Buruli ulcer cases are associated to rainfall patterns , with cases occurring during the dry season following a flooding event [9] , and leptospirosis outbreaks frequently occur during periods of seasonal rainfall and flooding [15 , 16 , 17] . Therefore , Amazonian environmental conditions are highly favorable for the persistance of both bacteria in aquatic systems [18 , 19] and disease cases are notably reported in French Guiana ( South America ) [13 , 20] . Focusing on these two aquatic pathogens as model systems we performed an environmental survey along the French Guiana coastline with the objective to test the following hypothesis: Finally , we discussed about the likely impact of the transmission mode , e . g . vector-borne versus generalist pathogen , on the spatio-temporal dynamics of both bacteria in the environment .
Between november 2015 and march 2017 a total of 18 rural aquatic sites were sampled monthly for water and sediments . Rural sites were located along the French Guiana coastline and also along the Sinnamary river . These sites were selected based on previous sampling campaigns in French Guiana [19] and represented ponds and oxbows characterized by low water level and stagnant water , either shaded or sunny , composed of a community of aquatic taxa and surrounded either by vegetation or a dense tropical rainforest ( Fig 1 , Table 1 ) . In parallel , 51 urban water bodies were also sampled for water and/or sediments from september 2016 to october 2017 among three urban centers: Cayenne ( 2441 . 27 inhabitants/km2 ) , Rémire-Montjoly ( 519 . 97 inhabitants/km2 ) and Matoury ( 236 . 37 inhabitants/km2 ) ( Fig 1 , Table 1 ) . Urban sites were selected based on the location of leptospirosis and Buruli ulcer cases and corresponded to small ditches . Disease cases were mapped and we selected water bodies that fitted the following conditions: ( i ) being close to Buruli ulcer and/or leptospirosis cases and ( ii ) showing the ecological conditions prone to sustain the bacteria in the environment ( i . e . small water body with low water level and biofilm development for M . ulcerans , and prone to harbour rodents for Leptospira sp . ) . Water was collected in the middle of the water body , from the water column between 0–1 m below the surface and kept in 1 . 5 L plastic bottles stored on ice and transported to the laboratory . Also the first layer of sediments ( 0–1 cm depth ) was collected in 30 mL tubes stored on ice and transported to the laboratory . Samples were kept at 4°C until DNA extractions ( performed within 24 h for water and 48 h for sediments ) . Moreover all sites were surveyed during both the dry ( september 2016/october 2016/july 2017/october 2017 ) and the rainy ( february and may 2017 ) seasons in order to compare the distributions and prevalences of both environmental pathogens in space and time ( Table 2 ) . Disease cases were reported at the Cayenne Hospital and were provided by Dr . Loïc Epelboin ( Infectious Disease Unit ) and Prof . Pierre Couppié ( Dermatology Unit ) . Leptospirosis database provided by Dr . Loïc Epelboin was based on the diagnosis for each patient that has been reported at the Cayenne Hospital and the Centre National de Référence de la Leptospirose at the Pasteur Institute in Paris . Leptospirosis cases reported in French Guiana were analyzed and made available for the period 2014–2017 by the Cayenne Hospital , the Pasteur Institute in Cayenne and the Biomnis laboratory . Buruli ulcer database was built by Prof . Pierre Couppié and colleagues and cases were available from 1969 to 2017 . However to be consistent in our comparison of disease cases and positive environmental sites we mapped only cases diagnosed between 2015–2017 , occurring thus over the same timescale . Water samples ( 1 . 5 L ) were first filtered onto 1 . 6 μm GF/C glass microfiber filters ( Whatman ) and then through 0 . 45 μm cellulose nitrate membrane filters ( Merck Millipore ) . These later filters were air dried and kept at -20°C until further analysis . Total DNA was extracted from filtered water using the DNeasy PowerWater extraction kit ( Qiagen ) following the manufacturer’s recommendations . For sediment samples , 250 mg of sediments were used to extract DNA using the DNeasy PowerSoil extraction kit ( Qiagen ) . Extracted DNA was kept at -20°C . To detect and quantify M . ulcerans DNA in environmental samples , we performed two TaqMan qPCR runs; one targetting the insertion sequence IS2404 and one targetting the ketoreductase B ( KR ) domain of the mycolactone polyketide synthase gene that is specifically found in the virulence plasmid of M . ulcerans strains . To amplify IS2404 genetic marker , we used the following primer and probes: IS2404 forward primer 5’-ATTGGTGCCGATCGAGTTG-3’ , IS2404 reverse primer 5’-TCGCTTTGGCGCGTAAA-3’ and IS2404 probe FAM-CACCACGCAGCATTCTTGCCGT-BHQ1 [21] . For KR amplification we used KR forward primer 5’-TCACGGCCTGCGATATCA-3’ , KR reverse primer 5’-TTGTGTGGGCACTGAATTGAC-3’ , and KR probe FAM-ACCCCGAAGCACTG-MGBNFQ [19] . The qPCR reaction consisted of 1X TaqMan Gene Expression Master Mix ( LifeTechnologies ) , 0 . 3 μM ( final concentration ) of each primer , 0 . 1 μM ( final concentration ) of the probe , 5 μl of DNA and water adjusted to a final volume per reaction of 25 μl . For KR we followed the same protocol except that we used the probe at a final concentration of 0 . 25 μM . An internal positive control ( IPC ) was added in each IS2404 reaction in order to test for the presence of PCR inhibitors in the environmental samples . In each qPCR plate , a positive ( M . ulcerans DNA at a concentration of 105 bacteria/mL ) and negative ( DNA replaced by water ) controls were included . The positive control for M . ulcerans consisted of genomic DNA purified from a cultured strain from French Guiana ( strain 1G897 ) and provided by Laurent Marsollier ( ATOMYCA , Université d’Angers ) . This positive control was also used in our study to run standard curves based on serial dilutions of purified DNA from 105 to 100 bacteria/mL ( in triplicates ) . Standard curves allowed us to determine a threshold value above which we considered our samples as negative ( CT-values > 38 ) . The assays were run in duplicates on an Applied Biosystems 7300 Real Time PCR system , with the following program: one cycle at 50°C for 2 min , one cycle at 95°C for 10 min , followed by 45 cycles at 95°C for 15 sec and at 60°C for 1 min . Only samples with cycle threshold values < 38 for both IS2404 and KR markers in 1 out of 2 replicates were considered as positives . In all assays the negative controls remained negative . To date , 22 species of Leptospira have been described and arranged into 3 groups based on their pathogenicity; pathogenic species ( L . interrogans , L . kirschneri , L . borgpetersenii , L . mayottensis , L . santarosai , L . noguchii , L . weilii , L . alexanderi , L . kmetyi , L . alstonii ) , intermediate species of unclear or low pathogenicity ( L . broomii , L . fainei , L . inadai , L . licerasiae , L . wolffii ) , and saprophytic species which are free-living cells in water and soil and are not infectious ( L . biflexa , L . idonii , L . meyeri , L . terpstrae , L . vanthielli , L . wolbachii , L . yanagawae ) [22] . Whilst pathogenic and intermediate Leptospira species are infectious for humans or animals [22] , most diagnotic PCR tools only detect Leptospira from the pathogenic cluster and fail to detect intermediate species [23] . Among these tools the TaqMan qPCR assay targetting the lipL32 gene is commonly used to detect pathogenic Leptospira [5 , 23 , 24] since it encodes outer membrane proteins and virulence factors found in pathogenic species [22] . This qPCR assay has been optimized for both sensitivity and specificity , allowing thus to detect and characterize Leptospira sp . in low number or in samples that contain high concentrations of non-Leptospira DNA [22 , 23] . However since this target gene is highly conserved among Leptospira species it does not allow the discrimination between species . Therefore , to detect the presence of Leptospira sp . DNA we performed a qPCR targetting the lipL32 gene . To do so , we used forward primer LipL32-45F 5’-AAGCATTACCGCTTGTGGTG-3’ , reverse primer LipL32-Rb 5’-GAACTCCCATTTCAGCGAT-3’ and the probe LipL32-189P FAM-AAAGCCAGGACAAGCGCCG-BHQ1 [23] . The qPCR reaction consisted of 1X TaqMan Gene Expression Master Mix ( LifeTechnologies ) , 0 . 7 μM ( final concentration ) of each primer , 0 . 15 μM ( final concentration ) of the probe , 5 μl of DNA and water adjusted to a final volume per reaction of 25 μl . In each qPCR plate , a positive ( L . santarosai DNA at a concentration of 102 bacteria/mL ) and negative ( DNA replaced by water ) controls were included . The positive control for Leptospira sp . consisted of genomic DNA of the strain L . santarosai that was provided by the Pasteur Institute in Paris ( Pascale Bourhy , Centre National de Référence de la Leptospirose ) . Standard curves were run with serial dilutions of genomic DNA from L . santarosai from 105 to 100 bacteria/mL ( in triplicates ) . The assays were run on an Applied Biosystems 7300 Real Time PCR system , with the following program: one cycle at 50°C for 2 min , one cycle at 95°C for 10 min , followed by 45 cycles at 95°C for 15 sec and at 60°C for 1 min . Only samples with cycle threshold values < 40 were considered as positives . In all assays the negative controls remained negative . The prevalences of Leptospira sp . and M . ulcerans in the environment were calculated for each sampling period based on the ratio between the number of sites found positive and the total number of sites tested ( % ) . Whilst we measured the quantity of DNA rather than the quantity of live bacteria in the environment [25] , the variation in DNA concentration between sites and between each sampling periods was used as a proxy of bacterial abundances in the environment . Statistics were performed with R version 3 . 5 . 1 ( R Development Core Team ) . We used raw data to perform a logistic binomial regression ( package stats , function glm ) in order to test for the effect of seasonality ( dry and rainy seasons ) on the pathogen’s presence in the environment ( significance threshold: p-value < 0 . 05 ) . Maps were created with QGIS ( Las Palmas , version 2 . 18 . 20 ) .
Among the 18 rural sites tested for the presence of Leptospira sp . DNA , only one site ( R9 ) was found positive for this pathogen ( Fig 2A , Table 1 ) . Moreover , this site was found positive at only one sampling period , in february 2017 , leading to a total of 1/201 ( 0 . 5% ) sample recorded positive for Leptospira sp . DNA in rural areas in French Guiana . These 18 rural sites were also tested for the presence of M . ulcerans DNA and we found 4/18 sites ( R5 , R9 , R13 , R17 ) that harboured DNA of this mycobacteria , leading to a total of 12/201 ( 6% ) samples that were positive for M . ulcerans DNA ( Fig 2A , Table 1 ) ; site R5 was positive in november 2015 only , site R9 was positive in january , february , april , july , october and november 2016 , site R13 was positive in december 2015 only , and site R17 in august , october and november 2016 ( Supporting Information S1 Table ) . Therefore these results show a greater number of sites found positive for M . ulcerans in natural aquatic systems compared with Leptospira sp . . Also , M . ulcerans DNA was found in remote prestine sites with no human contact on the upper part of the Sinnamary River ( R5 ) , and one site ( R9 ) was suitable for both pathogens . In parallel , 51 urban water bodies were also sampled for water and sediments from september 2016 to october 2017 around Cayenne , Rémire-Montjoly and Matoury ( Table 1 ) . We found a total of 34/169 ( 20% ) samples positive for Leptospira sp . DNA and 27/169 ( 16% ) samples positive for M . ulcerans DNA ( Fig 2B , Table 2 ) . These results suggest that urban habitats are more favourable to the establishment of Leptospira sp . and M . ulcerans , which exhibited higher number of positive sites through time when compared with rural environments ( 20% vs . 0 . 5% and 16% vs . 6% , respectively ) . The only rural site found positive for Leptospira sp . DNA corresponds to the rainy season ( average rainfall of 270 . 4 mm , Table 2 ) . In contrast , the majority of rural sites found positive for M . ulcerans DNA were positive during the dry periods ( see Table 2 ) . A total of 7/11 rural sites ( 64% ) were found positive for M . ulcerans DNA during the dry season . Among the urban sites , the prevalence of Leptospira sp . DNA increased during the rainy season to 47 . 2% and 22 . 2% in February and May respectively , and reached its minimum during the dry period ( Supporting Information S1 Map , Table 2 ) . Although M . ulcerans DNA prevalence was more stable in the environment across seasons , most positive sites were observed during the dry period from 11 . 8% to 66 . 7% depending on the months ( Supporting Information S2 Map , Table 2 ) . Our results showed a correlation between the prevalences of each pathogen in the environment and seasonality , such that the prevalence of pathogenic leptospires was enhanced by rainfall while M . ulcerans’s prevalence in aquatic sites was more related to drought ( Fig 3 ) . These observations were confirmed by the binomial regression models that showed that the environmental dynamics of Leptospira sp . and M . ulcerans were significantly different between the dry and the rainy seasons . Indeed , environmental sites had 2 . 86 times ( 95% CI 1 . 14–8 . 19; p-value = 0 . 0339 ) higher odds of M . ulcerans positivity during the dry season , while Leptospira sp . had 5 . 20 times ( 95% CI 2 . 32–12 . 63; p-value = 0 . 00012 ) higher odds of being detected in the rainy season ( Fig 4 ) .
Here our aim was to follow and describe the environmental dynamics of two aquatic bacteria , potentially pathogenic to animals and/or humans , across the French Guiana territory including the Amazon tropical rainforest . Our environmental survey describe the presence of both pathogens in rural and urban sites , with M . ulcerans being more often encountered in natural ( undisturbed ) rural habitats compared with Leptospira sp . . As proposed before , our results suggest the ubiquitous nature of the mycobacterium M . ulcerans in the environment [19] . According to Combe et al . [8] , this pathogen is likely to be widely distributed in suitable natural aquatic systems and , under specific environmental conditions could become more abundant in the system , resulting an increased risk of Buruli ulcer emergence . Whilst M . ulcerans DNA was more often found in rural aquatic sites than Leptospira sp . , we observed that one site ( R9 ) was suitable for both pathogens , which was the only rural site located close to poor human settlements and frequently visited by livestock , such as cows and pigs , and domestic animals such as dogs , cats and poultry . This suggest that the presence of Leptospira sp . in rural environments would need the presence and activity of humans , livestock and/or domestic animals , that would represent sustainable host populations for this pathogen . The presence of Leptospira sp . in peri-domestic water samples from rural households has already been reported in southern Chile , with the presence of dogs and a high density of rodent populations being associated with positive puddles in the lower income households [26] . Pathogenic leptospires could be permanently present in the aquatic environment , and more specifically in watered soils as recently showed in New Caledonia [12] , but only detected by qPCR when their abundance is increased by the proximity with human settlements and the availability of animal hosts that could also represent more attractive areas for rodent populations harbouring and sharing the bacteria in surrounding aquatic sites [5] . Also , the results suggest that modified urban environments are more favourable to the establishment of Leptospira sp . and M . ulcerans , with higher positivity for both pathogens when compared with rural environments ( 20% vs . 0 . 5% and 16% vs . 6% , respectively ) . Here the difference in the number of sampling sites and the different sampling scale in rural ( 18 water bodies , 17 months , 1 samples , n = 306 ) vs urban areas ( 51 water bodies , 6 months , 1 sample , n = 306 ) , could constrain our interpretation when comparing the presence of both pathogens in these environments . However , in French Guiana most of the population live along the coastline and only 5 centers can be considered as urban ( based on infrastructure development , population density , etc . ) : Cayenne ( 2441 , 27 inhabitants/km2 ) , Rémire-Montjoly ( 519 , 97 inhabitants/km2 ) , Matoury ( 236 , 37 inhabitants/km2 ) , Kourou ( 12 , 14 inhabitants/km2 ) and Saint-Laurent du Maroni ( 9 , 03 inhabitants/km2 ) [27] . Our sampling effort covered 3 of these urban centers ( Cayenne , Rémire-Montjoly and Matoury ) , and each show higher number of positive sites for Leptospira sp . and M . ulcerans when compared to rural sites . Moreover our sampling stategy was designed to reflect the differences in number of human cases in urban vs rural settings , with 99 , 3% and 8% of human cases for leptospirosis and Buruli ulcer occuring in urban populations versus 0 , 73% and 27 , 3% of cases occuring in rural ones , respectively . As the number of human cases was extremely low in rural settings it was important to have more regular sampling and over a wider geographical area in order to not miss any environmental signal . In urban settings this was not an issue and as such we adopted seasonal based sampling over the whole urban habitat . Therefore our results are consistent with the distribution of human cases across French Guiana , with much less cases in rural than in urban areas [20] . Therefore , we do not observe in villages located in the countryside clusters of human cases as observed in urban environments . In addition , a lack of replicates ( 1 sample/site ) could potentially be a limit as sometimes many samples at a site can result in a single positive sample , or even none ( from our field experience ) . However , our results show that even with such sampling effort ( 1 sample/site ) , we had no difficulties in detecting the presence of both pathogens in the environment , and confirmed their presence across seasons similar to what was previously reported on multiple samples [6 , 9 , 14] . Therefore , the sampling effort did not overestimated the presence of these pathogens . In urban environments these pathogens seems to establish , colonize and share similar ecological niches ( e . g . benthic algae and watered soils from 1–5 cm depth for pathogenic leptospire [12]; algae biofilm and watered soils from 1–10 cm depth for M . ulcerans , personnal data ) , habitats that also exist in rural settings but which are for some biotic reasons less favourable to their development . Casanovas-Massana and collaborators ( 2018 ) found that both sewage water and standing water were reservoirs for pathogenic Leptospira sp . in a urban area in Salvador , Brazil [5] . It shows that simplified urban ecosystems with less predators to rodents , low level trophic networks due to pollution and increased contact with humans would favour leptospirosis and Buruli ulcer emergence and transmission . Other diseases are also known to be more prevalent in urban areas , such as dengue fever due to the availability of suitable human-created micro-environments for Aedes sp . mosquitoes breeding and eggs laying [28] , or even for water-borne and enteric diseases with a oral-fecal transmission in areas with poor sanitation infrastructure [7] . Since most human emerging and/or re-emerging infectious diseases are zoonotic , increased urbanization in developing tropical countries would tend to increase the frequency of contact between wildlife and humans , representing an increased risk of disease emergence [7] . Indeed , drivers such as land-use changes ( i . e . from natural toward deforested areas , agriculture intensification , road building ) modify the ecology of the pathogen-vector-host , as well as increase contact between pathogen-vector and humans ( i . e . increase human population densities , pollution , unsanitary conditions ) . The prevention of aquatic bacterial diseases emergence and transmission in tropical areas , where millions of people are currently living and where half of the world’s population will live by 2050 [7] , would necessarily result from improved infrastructure and sanitation in impoverished urban areas of developing countries . Here we showed a synchronisation between the presence of each pathogen in the environment and seasonality , with higher number of positive sites for pathogenic leptospires during the rainy season while M . ulcerans’s presence in aquatic sites was more related to drought ( Fig 3 , Fig 4 ) . Previous findings indicated that leptospirosis outbreaks frequently occurred during periods of seasonal rainfall and flooding events in endemic areas [5 , 15 , 16 , 17] . Moreover , seasonal ( i . e . rain vs drought ) conditions leading to increased human exposure to contaminated water are known to be important drivers for leptospirosis transmission , and the proximity of households to open drainage systems and direct contact with sewage , flooding water and runoff were associated with increased risk of infection [29–33] . Similarly , a recent study conducted in Brazil found higher bacterial concentrations in urban environmental sites ( sewage and standing water ) during the rainy season when compared with the dry period , indicating thus a seasonal effect [5] . Also , the link between Buruli ulcer cases and seasonal patterns has been identified in several studies , showing that Buruli cases occurred during the dry season that followed rainfall events [reviewed in reference 8] . These observations were confirmed by long-term time series of Buruli ulcer cases [9] and climatic models [34 , 35] that revealed robust correlations between disease incidence and seasonality , with the disease being reported in French Guiana after dry periods following periods of heavy rainfall . Moreover in French Guiana Morris et al . ( 2014 ) have linked Buruli ulcer cases with extreme weather events such as La Niña , that are responsible to cause short dry periods during the rainy season [9] . Here the most interesting findings rely on the comparison of the environmental dynamics of the two aquatic bacterial pathogens in the same unit of time and space across both rural and urban environments . Indeed , we found that both pathogens are ubiquitously distributed in aquatic sites ( persistent with low burden ) although this seems more obvious for M . ulcerans in rural settings , and are both able to survive under similar ecological conditions . However , our seasonal survey clearly showed that the presence of each pathogen in the environment were heterogeneous and depended on different climatic patterns; whilst the presence of pathogenic leptospires in the environment was enhanced by rainfall , M . ulcerans emergence was boosted by drought that followed rainfall and flooding events . These results suggested that in French Guiana the infectious risk for each disease does not occur at the same period of the year , and is constrained by seasonality . Such local prevalence variability due to seasonal patterns might result from the pathogen’s life cycle or the dynamics of reservoir and/or host populations . Looking at the spatial distribution of Leptospira sp . and M . ulcerans in the urban environment show that positive sites for Leptospira sp . are much more widely distributed when compared with M . ulcerans positive sites that are locally constrained within small neighborhoods ( Fig 2B ) . We propose that such difference in the spatial distribution of both bacteria across urban sites could also be explained by the mode of dissemination of each pathogen in the environment ( i . e . environmental dissemination of M . ulcerans vs . vector-borne/animal-borne for Leptospira sp . ) . During rainfall with increased oxygen levels and alkaline pH ( up to pH 8 . 0 ) , low salt concentrations , and/or the dilution of sewage toxic compounds , Leptospira sp . flourish [5 , 12 , 36] and are ingested by rodents and other carrier mammals . Flooded urban habitats , favour the re-distribution of rodents across cities , thus dispersing leptospires from one site to another mainly via urinary excretions . It does result in an increase of cases during the rainy season over a wide urban range . Such patterns have been also observed in other settings around the world where leptospirosis epidemics occurred in the rainy season that followed heavy rainfall [37 , 38 , 39] . Alternatively , Ferreira de Albuquerque et al . ( 2017 ) reported that capybaras ( Hydrochoerus hydrochaeris ) were massively infected by leptospires in the western Amazon region [40] . These small mammals are present in urban areas in French Guiana and could thus further play a role in the environmental dissemination of pathogenic leptospires . Unfortunately studies on rodent infections with Leptospira sp . are very scarce and old in French Guiana [41] . In contrast , the ecological conditions favouring M . ulcerans emergence in the environment are known to rely on higher water temperature , low pH , low oxygen levels and the presence of algal biofilm , conditions typically encountered during the dry period in the tropics [reviewed in reference 8] . In addition , M . ulcerans is not transmitted by a specific vector , but rather was found to be associated with a large range of aquatic invertebrates . Several studies showed that aquatic organisms of low/mid trophic level usually exhibit greater bacterial loads compared with organisms of a higher trophic level [reviewed in reference 8] . For instance M . ulcerans seems to have a specific association with gathering collectors and filter feeders . After anthropogenic ( i . e . deforestation ) or natural ( i . e . changes in weather , flooding ) changes , stagnant water bodies are prone to rapid local abiotic changes ( temperature , pH , oxygen , etc . ) associated with a rapid turnover of the biotic community ( i . e . changes in functional diversity ) and leading to an increase in favourable hosts harbouring M . ulcerans [8] . Based on the current knowledge on M . ulcerans ecology it appears clearly that its distribution is locally constrained by habitat and aquatic hosts communities highlighting the relative clustering of human cases within urban units in French Guiana . Whilst the wide presence of M . ulcerans along French Guiana’s coast was already known , this is the first survey that screened for the presence of pathogenic leptospires across the territory ( including the Amazon tropical rainforest ) . Until recently it was assumed that there were few leptospirosis cases in French Guiana , compared to West Indies for instance , probably because of the acidity of the soil across the Guiana shield preventing bacterial development . However our results clearly showed that ( i ) both pathogens are present in the environment in French Guiana , and ( ii ) urbanization and seasonality are two important factors underlying Buruli ulcer and leptospirosis emergence . Also we propose that the mode of transmission ( i . e . generalist vs . vector-borne ) of environmental pathogens might have a detrimental role in disseminating the infectious agent in the environment . To better monitor diseases emergence , we suggest that future studies should focus on determining which specific socio-economic and environmental factors are underlying the spatio-temporal distribution of emerging infectious pathogens . | Many emerging pathogens are zoonotic and transmit from their abiotic reservoir to wild animals , domesticated animals and humans . It is now well known that environmental changes lead to variability in their dynamics in the environment and contribute to changes in the infectious risk . Many aquatic bacteria are responsable for major public health concerns , and more importantly in developing countries where access to drinking-water and sanitation is often limited . Whilst their environmental dynamics are directly linked to seasonal changes of their habitat , they also rely on the ecological processes underpining their transmission , i . e . directly transmitted vs . vector-borne . However , few studies have compared such environmental dynamics despite the fact that it would help to better characterise the infectious risk in the environment , as well as to better monitor the emergence of infectious diseases . Our aim was to provide data on the prevalence of generalist vs . vector-borne aquatic bacterial pathogens in the environment that would further allow the comparison of their environmental dynamics in the same unit of time and space , and across rural and urban habitats . We showed that urbanization and seasonality are two important factors underlying Buruli ulcer and leptospirosis disease emergence in French Guiana ( South America ) , and propose that the mode of transmission of such environmental pathogens might have a detrimental role in disseminating the infectious agent in the environment . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"ecology",
"and",
"environmental",
"sciences",
"leptospira",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"tropical",
"diseases",
"microbiology",
"seasons",
"bacterial",
"diseases",
"urban",
"environments",
"geology",... | 2019 | Comparison of Mycobacterium ulcerans (Buruli ulcer) and Leptospira sp. (Leptospirosis) dynamics in urban and rural settings |
The gambiense form of sleeping sickness is a neglected tropical disease , which is presumed to be anthroponotic . However , the parasite persists in human populations at levels of considerable rarity and as such the existence of animal reservoirs has been posited . Clarifying the impact of animal host reservoirs on the feasibility of interrupting sleeping sickness transmission through interventions is a matter of urgency . We developed a mathematical model allowing for heterogeneous exposure of humans to tsetse , with animal populations that differed in their ability to transmit infections , to investigate the effectiveness of two established techniques , screening and treatment of at-risk populations , and vector control . Importantly , under both assumptions , an integrated approach of human screening and vector control was supported in high transmission areas . However , increasing the intensity of vector control was more likely to eliminate transmission , while increasing the intensity of human screening reduced the time to elimination . Non-human animal hosts played important , but different roles in HAT transmission , depending on whether or not they contributed as reservoirs . If they did not serve as reservoirs , sensitivity analyses suggested their attractiveness may instead function as a sink for tsetse bites . These outcomes highlight the importance of understanding the ecological and environmental context of sleeping sickness in optimizing integrated interventions , particularly for moderate and low transmission intensity settings .
We developed a deterministic model of the West and Central African form ( T . b . gambiense ) of human African trypanosomiasis transmission . The model captures heterogeneity in exposure to tsetse bites , and can allow for the possibility of non-human animals to contribute to transmission . A schematic overview of the model structure is provided ( Fig 1 ) , while the details are given in the methods section , below . A description of all state and rate parameters is provided in Tables 1 and 2 . Since this system of equations can lead to a wide range of equilibrium prevalences when evaluated using a range of reasonable estimates for the rate parameters [39 , 40] , we had to obtain parameter sets that allowed the behavior of the model around realistic prevalence levels to be examined . In active foci of HAT transmission , prevalences are often as low as 0 . 1–1% in humans [41] . Annual incidence thresholds associated with high ( >1/103 and <1/102 ) , moderate ( > 1/104 and <1/103 ) , and low ( >1/105 and <1/104 ) risk categories of HAT have been suggested [42] . Based on the mid points of those ranges , thereby ignoring very high and very low outliers , an assumption that incidence will be comparable to prevalence if mobile units visit afflicted areas infrequently , and an underreporting rate of approximately 3 [12] , we specified prevalence levels of 1 . 65% , 0 . 165% , and 0 . 0165% as being representative of high , moderate , and low transmission settings , respectively . A critical assumption we made is that these prevalence levels were stable . We note that due to the paucity of available data , posterior parameter estimates were unlikely to change significantly from the prior ranges . The intent , however , was not to obtain more precise parameter estimates , but rather to act as a filter for sets of parameters that would lead to unrealistic prevalences . We obtained parameter sets using a Bayesian framework of importance resampling [43 , 44] . Median values of the resampled estimates for the animal reservoir and no animal reservoir versions at high , medium and low transmission intensities are provided ( Tables 2 and S1 , respectively ) . For both the version with and without the possibility of an animal reservoir , as transmission intensity decreased ( i . e . , from the high to moderate to low intensities ) , the number of parameter sets in the resampled sets did as well , presumably because the parameter space leading to such very low prevalence levels was sparser than that leading to high levels of prevalence . Stable levels of sleeping sickness prevalence associated with high , moderate , and low transmission in field settings were found using both the model version without the possibility of transmission from non-human animals to tsetse , and the model version allowing for animal reservoirs . In both cases a role for heterogeneity in exposure to tsetse bites was suggested , whereby the smaller the proportion of humans at high risk , the lower prevalence could be , particularly in the absence of an animal reservoir . For instance , the estimated median ratio of humans commuting to an area with potentially higher tsetse biting rates ( N2/N1 ) was 0 . 03 ( without animal reservoirs ) and 0 . 06 ( with animal reservoirs ) in the high transmission settings . To gain insight into the parameters that drive HAT prevalence at the levels encountered in the field , we performed a global sensitivity analysis . We created 500 parameter sets by drawing Latin hypercube samples for each parameter from uniform ranges between the largest and smallest of the median values estimated for the high , moderate and low transmission settings for the model versions with ( S1 Table ) and without animal-tsetse transmission ( Table 2 ) . We simulated the equations until the asymptotically stable equilibrium point was reached and the prevalence of infection in humans was recorded for each iteration . The sensitivity of prevalence to model parameters was investigated for 500 samples by calculating the partial rank correlation coefficients , which gives an indication of the degree to which an output parameter is related to an input parameter when controlling for the effects of other parameters . The sensitivity analyses were performed using SaSAT [45] . The sensitivity of sleeping sickness prevalence to model parameters sheds light on the drivers of heterogeneous exposure ( Fig 2 ) . In the case where non-human animals cannot contribute to transmission directly , the parameters with the greatest influence on prevalence were the proportion of humans commuting to this area ( N2/N1 ) , and the tsetse biting preference for peridomestic animal hosts ( σa1 ) . The density of vectors to humans in the village ( V/H1 ) , the biting preference for humans ( σh ) , as well as parameters related to transmission efficiency ( b , c ) were also of importance . When allowing for animal-tsetse transmission a number of the same components stand out as drivers of prevalence , namely the efficiency of transmission to humans and animals ( b ) , the vector density in the village ( V/H1 ) , and the biting preference for humans ( σh ) . To gain further insight into how the efficacy of different parameters and potential control approaches ( screen & treat and vector control ) depend on variation in other parameters , we investigate the impact on the basic reproduction number , R0 , by looking at zero-growth isoclines , i . e . , the parameter space where R0 = 1 ( Fig 3 ) . As the human populations are more separated ( with increasing ξ ) a greater removal rate is required to interrupt transmission , and removal rates leading to R0 < 1 become smaller as vector mortality increases ( left plot ) . When we compare the impact of screening humans in the low risk setting ( ra ) and screening commuting humans ( rb ) , it is clear that screening the humans commuting to high exposure areas is critical , as transmission can be sustained by this group even at very high rates of screening the non-commuting population ( ra ) . Depending on the density of animals in the low risk area ( which here do not contribute to transmission , but can lead to “wasted bites” among tsetse ) , it can be sufficient to target only the commuting population . When animal densities are low , resulting in higher biting rates on humans in the village ( Nh1 ) , then transmission could also be sustained among this population and all populations will need to be screened ( middle plot ) . When animals can contribute to transmission ( ca = 0 . 003 ) , treatment of infected humans will not lead to interruption of transmission above a certain vector-human density threshold , and tsetse control will have to be relied on ( right plot ) . In addition to the feasibility of interrupting transmission at various levels of efficacy , the timelines under which prevalence decreases to near zero can also be a relevant consideration . Simulations were performed to assess these timelines and probabilities , allowing for the range of model parameter uncertainty , as well as a range of uncertainty in control intervention efficacy , and structural model uncertainty regarding the ability of animals to contribute to sleeping sickness transmission . Note that when elimination is used in this context , it is strictly more accurate to speak of elimination as a public health problem . This is because we used a deterministic model and had to specify an arbitrary threshold below which we assume elimination is likely in reality to occur . We used a prevalence in humans of < 1 x 10−6 , below the lowest threshold associated with very low risk HAT foci [42] . A stochastic model allowing for immigration and importation of infections , with discrete numbers of infected individuals , would be required to investigate actual elimination . For these simulations we investigate high transmission settings only , as these are the ones where a priori a combination of interventions may be considered [12] . A wider range of interventions and their cost-effectiveness using this modelling approach over all transmission settings will be reported on elsewhere ( Sutherland et al , in prep ) . The interventions we considered were case detection in humans ( screen and treat ) , potentially supplemented with vector control using targets or traps . The effectiveness of traps may differ by the species of tsetse and environmental conditions , as well as issues related to spacing and maintenance . Estimates from the literature suggest attaining a 5% daily mortality is often attainable [46 , 47 , 48] , and we used a broad range of 1–10% to capture this variability . We assume screen and treat operates by removing 1st and 2nd stage infected people at a daily rate . We use the formula specified by Artzrouni & Gouteux [14] who relate a percentage , d , of humans effectively screened in a given period ( a month in their model , here , a year ) : d = 100 ( 1−e−365 r ) and the daily removal rate , r , ( which appears in eqs ( 3 ) through ( 5 ) and ( 8 ) through ( 10 ) ) is therefore: r = −ln ( 1−d100 ) /365 . The percentage of humans screened at a yearly basis will depend on the frequency of visits of mobile teams , as well as other factors such as the achieved level of coverage and compliance , and sensitivity of the diagnostic test . Here , we investigate a broad range of 20 to 80% of the population . In one analysis where we investigate the impact of having different screening rates on humans in low and high exposure populations , instead of a single parameter r , we then specify ra and rb for the removal rate of Nh1 and Nh2 , respectively . On average , these simulations suggest that if animals do not contribute to transmission , deployment of either a screen & treat strategy or an integrated approach of screening together with vector control is likely to result in interruption of transmission ( Fig 4 ) . However , the rate at which prevalence declines toward zero is considerably faster when using the integrated approach . If biting an animal can result in a tsetse becoming infected , screening and treatment of humans by itself is less likely to lead to interruption of transmission , although prevalence in humans will still be reduced . To explore the potential of using an integrated approach of screening and treatment with vector control in more detail in different transmission settings , and depending on whether animals were assumed to to be capable of transmitting infections , we ran sets of 500 simulations whereby both approaches were varied from absent , to low , moderate , or high coverage . For each set we determined the mean time required to reach elimination ( i . e . , a prevalence < 1 x 10−6 ) and the proportion of 500 simulations that led to elimination . In the high transmission setting in the absence of animal reservoirs , to achieve the goal of elimination within 20 years and with greater than 80% probability , either vector control at the highest level of efficacy ( resulting in a mean time to reach elimination of 19 years ) , or an integrated approach of screen & treat and vector control was required . A combination of both at our lowest levels of efficacy ( 20–40% coverage for screening , and a mortality rate between 1–4% for vectors ) was sufficient to reach this threshold , and either of the two interventions could be intensified in order to achieve greater probabilities of success or to reduce time to elimination . In general , increasing the intensity of vector control led to greater increases in the probability of eliminating , while increasing the intensity of human screening led to a sharper reduction in the time required to eliminate ( Fig 5 ) . In both the moderate and low transmission settings the time required to reach elimination was on average shorter , and the patterns overall were similar . In these settings , while an integrated approach resulted in high probabilities of achieving elimination within the shortest period , both approaches ( screen and treat and vector control ) employed by themselves at the high levels of efficacy also resulted in elimination , albeit at a somewhat lower probability and over a longer period ( Figs 6A and 7A ) . If there are foci where animal reservoirs do contribute to transmission , the patterns are in general similar to those from the model that assumes no animal reservoir , except that the probability of eliminating without vector control is reduced ( Figs 5B , 6B and 7B ) . The conclusions are similar in regard to an integrated approach of screen & treat and vector control likely being necessary to eliminate with a greater than 80% probability and within 20 years , although a strategy of vector control by itself may also be acceptable , especially in moderate or low transmission zones .
Increased understanding of the ecological , environmental and behavioral drivers of HAT transmission is critical to designing effective control programs that maximize the probability of achieving elimination . In this study we focused on the role of non-human animals and the importance of heterogeneous exposure of humans to infected tsetse bites . A key insight is that regardless of whether or not non-human animals contribute to the transmission cycle of T . b . gambiense , animal populations play a crucial , but varying , role in the epidemiology of HAT . Although the consensus currently suggests that non-human animals do not play an important role in HAT transmission , we found that when—in perhaps a minority of foci—animal populations are capable of harboring T . b . gambiense and infecting tsetse , then a strategy based only on case detection and treatment of humans is less likely to lead to interruption of transmission , and additional vector control interventions are likely necessary . We did not disentangle the factors that allowed screen & treat to interrupt transmission when animal-tsetse transmission could occur , but speculate that whether animal populations constitute a reservoir will depend on a combination of their efficiency of transmission , densities of animals and tsetse flies , and contact patterns of humans . It is perhaps worth noting that the current thought in HAT epidemiology that animals do not contribute to transmission is based largely on experiences where screening and treatment of human populations alone , without vector control , has led to the local elimination of HAT in certain foci ( e . g . , [49] ) . However , it may be dangerous to base such conclusions on limited natural experiments . Our simulations suggest that even if animals can contribute to transmission , whether vector control is required may depend on more specific epidemiological or ecological factors . For instance with animal-tsetse transmission , at the highest rate of screening without vector control , the probability of achieving elimination varied between 0 . 35–0 . 65 for the different transmission settings ( Figs 5A , 6A and 7A ) . Further validation of these results is required , but what likely occurred in our model was that although animals could contribute to transmission in these simulations , their type reproduction number in a subset of the parameter sets may have been below 1 , and maintenance may have been due to humans . In the other subset of parameter values , it appears that infection in humans may have been due to spillover from animals . When we did not allow for animal-tsetse transmission in our model—a common assumption in models of T . b . gambiense transmission [18 , 40]—the role of animal populations was one of dilution or zooprophylaxis: essentially leading to tsetse “wasting” bites on inefficient hosts [50] . Similar insights were derived in a previous modelling study of T . b . rhodesiense , where in certain areas between 50–90% of meals may be taken by G . fuscipes fuscipes on monitor lizards , which are incompetent hosts [15] . This was evident from a comparison of the sensitivity analyses for both model versions . Without animal-tsetse transmission , one of the most important parameters with a negative relation to prevalence was a preference for tsetse to bite animals instead of humans in the area surrounding the main human population . In a similar study , although with different equations for the host choice model , Davis et al [40] found comparable outcomes where the most important parameters in a sensitivity analysis of R0 were the proportion of bites on humans , followed by the susceptibility of Glossina fuscipes to infection , the tsetse to human ratio , and vector mortality . Under this scenario where only humans can transmit infection to tsetse , our results suggest that an integrated approach of screening humans along with vector control will be the most efficacious and perhaps required: none of the screening coverage levels without vector control resulted in > 80% of simulations leading to elimination in the high transmission setting , whereas adding even the lowest level of vector control led to large increases in the probability of eliminating and reduced the time required to do so . Additionally , increasing the intensity of vector control was more likely to eliminate HAT transmission , while increasing the intensity of human screening reduced the time to elimination . Whether adding vector control is cost-effective however will require model extensions that include health impacts and implementation costs , and will be explored in a follow-up study on control and elimination strategies ( Sutherland et al , in prep ) . We found that if a strategy for interrupting transmission only employs human case detection , it is necessary and sometimes sufficient that those humans who are at the highest risk , i . e . , exposed to higher levels of tsetse bites , can be reached effectively . The opposite , an inability to target high exposure groups , which would hamper the ability to interrupt transmission [24] which may occur , for example , if people working in areas of high tsetse densities , like plantations , are less likely to attend mobile screening events due to opportunity costs [51 , 52] . Our outcomes thus suggest that in order to better identify risk factors leading to heterogeneity in HAT transmission and determine their implications for control strategies , understanding the regulation of tsetse population dynamics and biting behavior in relation to humans and non-human animals , as well as the suite of human activities that lead people to visit high risk areas , should be a priority . Similarly , measuring the efficiency with which animals can infect tsetse under natural field conditions would greatly reduce the structural uncertainty in our understanding of HAT transmission . The latter may also have important implications for optimizing elimination strategies by transmission setting: at the high levels , regardless of the presence of an animal reservoir , our results support an integrated approach . At the moderate and low levels , the conclusions are less straightforward: if an animal reservoir does not exist , either human treatment or vector control by themselves may be sufficient and still allow for elimination to be reached with reasonable probabilities and within a reasonable timeframe , whereas if an animal reservoir does exist , vector control by itself may be sufficient , but relying on human treatment alone likely will not be . There are several important assumptions implicit in this modelling study that could affect our outcomes . The median level of heterogeneity in our parameterizations is high compared to the common assumption that 20% of the people account for 80% of transmission [24] . However , prevalence data is often recorded at larger spatial scales ( such as provinces or districts ) than the scales at which transmission occurs ( such as villages and hamlets ) . One such example of heterogeneity comes from a description of epidemiological data in the Bipindi focus ( of high transmission ) , where 95% of cases were located in only 2 out of 15 villages [20] within the focus . In addition to this spatial heterogeneity , the behavioral and environmental factors described in the introduction would further increase the heterogeneity in exposure . It is possible that other factors , for instance the occurrence of long-term asymptomatic or chronic carriers [1 , 53] , may also have important effects on HAT transmission . Recently evidence for very long term asymptomatic carriage in stage I has been reported [54] . However , it remains unclear how common such occurrences are . In our model we assume a constant rate of progression from stage I to stage II . This leads to an exponential distribution of time spent within the stage , with a tail of a few people who would be expected to remain infectious for a very long time [53] . In the absence of better data on the frequency with which these chronic carriers occur , we thus assume that the extent to which they occur in our model based on an exponential distribution is reasonable . Additionally , the impact of heterogeneity , as well as the animal host species and other factors may vary between foci . Model fitting to data ( as it becomes available ) that encompasses human movement and activities as well as heterogeneity in tsetse density and biting behavior would be required to reduce parameter uncertainty beyond that which we were able to do here , and corroborate the importance and extent of heterogeneity in HAT transmission . We also assumed that prevalence of HAT was at a stable equilibrium in the absence of control interventions , which is at odds with historical data on a number of epidemic resurgences of the disease . The reasons for which are unclear , although they have been linked to ecological and societal disruptions caused by colonization [55] . Furthermore , our current model for screen & treat and vector control may ignore essential spikes and lapses in coverage by translating into continuous rates , what are in reality pulses associated with the roll-out of mobile teams or the set-up of tsetse traps . We have also made the assumption that passive detection of cases through the regular health only occurs after they have progressed beyond stage I , i . e . , that passive detection has no effect on transmission . If this view proves to be too pessimistic , a baseline value for the removal rate of humans due to case detection , r , greater than zero should be used . As longitudinal data of both prevalence in humans , animals , tsetse , and the performance of both the active and passive systems in specific foci becomes available , model fitting should take these considerations into account . Additionally , we simulated vector control by increasing the daily mortality rate of tsetse , but not decreasing their abundance , likely underestimating the impact of tsetse control . A more detailed model could include tsetse population dynamics [56] and the impact of seasonality ( in tsetse population dynamics , human behavior , and control interventions ) and could be extended to include temperature-dependent tsetse life history traits [57] . Although we included a metapopulation structure at a small scale ( e . g . , village and plantation ) , we did not allow for connectivity between different foci and resulting migration of either infected humans , animals , or tsetse [58] . Considering a larger spatial scale and movement patterns would allow for reinvasion following local interruption of transmission and likely make elimination more difficult to achieve than the results presented here . We assumed that areas at all ranges of transmission intensities allow for transmission in the absence of immigration of infected hosts , but if in reality there are instead sources ( active foci with high transmission ) and sinks ( areas with ( very ) low intensity that cannot sustain transmission ) , then this would have obvious implications for elimination strategies . A large scale metapopulation modeling study would be a useful follow-up to our current work . It should be noted however that including such additional sources of complexity will come at a cost of model tractability , i . e . , understanding the model drivers becomes more difficult and idiosyncratic results harder to explain . Likewise , data fitting and validation against different data sets may become more difficult; and before adding more complexity , it may be prudent to perform such validation exercises with the current model . One area where additional fitting and validation may shed light is the very low to low transmission intensity settings . As indicated , in the version without animal reservoirs , our fitting procedure led to repeated sampling of very few parameter sets for this transmission setting . This may be an indication of a model struggling to capture these patterns , and could indicate the existence of a reservoir of some type ( whether this is an animal reservoir , or migration of infected cases from nearby higher prevalence zones ) . Until further data fitting and validation can be undertaken , the results for the low prevalence setting should thus be interpreted with care . In conclusion , these results show the potential importance of heterogeneous exposure of humans to tsetse bites , and interactions between abundance and attractiveness of non-human animals to tsetse , as important drivers of HAT epidemiology , even when animals do not constitute hosts for T . b . gambiense . An increased understanding of these ecological dimensions of sleeping sickness is not only expected to lead to a better understanding of risk factors for transmission , but also to tailoring foci-specific integrated and cost-effective control approaches .
Our deterministic model of human African trypanosomiasis ( T . b . gambiense ) transmission is defined by a system of ordinary differential equations for compartments of two tsetse , animal and human populations , indicated by subscript i , where i below is 1 or 2 . The state variables and rate parameters of the model are described in Tables 1 and 2 , respectively . We denote each compartment with upright upper case Latin letters and the total number of individuals in each compartment by italicized upper case Latin letters . For humans , we have susceptible ( Shi ) , incubating ( Ihi ) , and two compartments related to the distinct stages of sleeping sickness , the 1st or asymptomatic stage ( Ahi ) , and the 2nd stage where trypanosomes have reached the cerebro-spinal fluid ( Rhi ) . Finally , we consider a treated compartment ( Thi ) , predominantly to enable downstream application of the model to health impact and cost-effectiveness studies . The total human population is given by: NThi ( t ) =Shi ( t ) +Ihi ( t ) +Ahi ( t ) +Rhi ( t ) +Thi ( t ) . We incorporate the assumption made by Artzrouni & Gouteux [18] that humans in advanced stages of the disease ( Rhi ) or those receiving treatment ( Thi ) are not available to tsetse bites , and therefore the population sizes used to calculate biting preferences exclude these two compartments , Nhi ( t ) =Shi ( t ) +Ihi ( t ) +Ahi ( t ) . To better understand the role of animals as reservoir hosts , we develop two separate models . In the first model , we assume that trypanosomes cannot infect animals and model two animal populations as constant parameters , Nai . In the second model , we divide the animal populations into susceptible ( Sai ) , incubating ( Iai ) , infectious ( Aai ) , and recovered ( Rai ) classes . We adopt the assumption that animals can recover from infection and are then temporarily immune , Rai , before returning to the susceptible state [17] . The total animal populations are , Nai ( t ) =Sai ( t ) +Iai ( t ) +Aai ( t ) +Rai ( t ) . In both models , we assume that all animals are subject to bites by tsetse files . For tsetse , we have susceptible ( Svi ) , incubating or latent ( Evi ) and infective ( Ivi ) compartments , so that the total vector population is given by: Nvi ( t ) =Svi ( t ) +Evi ( t ) +Ivi ( t ) . The equations describing the changes of numbers in human compartments ( Fig 1 ) are given by: dSh1dt=βh1+ r3Th1− μhSh1−b f θh1 , 1 Iv1 ( t ) Nh1 Sh1 ( t ) , ( 1 ) dIh1dt =b f θh1 , 1 Iv1 ( t ) Nh1 Sh1 ( t ) − ( μh+η ) Ih1 , ( 2 ) dAh1dt= η Ih1− ( μh+s1+r ) Ah1 , ( 3 ) dRh1dt = s1Ah1− ( μh+μs1+r ) Rh1 , ( 4 ) dTh1dt = rAh1+ rRh1− ( μh+ μt+r3 ) Th1 , ( 5 ) dSh2dt = βh2+ r3Th2− μhSh1−b f ( θh2 , 1Iv1 ( t ) Nh2+ θh , 2 Iv2 ( t ) Nh2 ) Sh2 ( t ) , ( 6 ) dIh2dt = b f ( θh2 , 1Iv1 ( t ) Nh2+ θh , 2 Iv2 ( t ) Nh2 ) Sh2 ( t ) − ( μh+η ) Ih2 , ( 7 ) dAh2dt= η Ih2− ( μh+s1+r ) Ah2 , ( 8 ) dRh2dt = s1Ah2− ( μh+μs1+r ) Rh2 , ( 9 ) dTh2dt = rAh2+ rRh2− ( μh+ μt+r3 ) Th2 . ( 10 ) The population size is assumed to be stable , by allowing the birth terms , βhi , to consist of the deaths in all compartments . The rate at which hosts are bitten depends on the frequency at which tsetse take blood meals , and the relative preference for human and non-human animals . The probability of biting a human for Nv1 is: θh , 1=σh ( Nh1+ ( 1−ξ ) Nh2 ) σh ( Nh1+ ( 1−ξ ) Nh2 ) +σa1Na1 which can also be specified for the two human populations exposed to the vector population: θh1 , 1=σhNh1σh ( Nh1+ ( 1−ξ ) Nh2 ) +σa1Na1 θh2 , 1=σh ( 1−ξ ) Nh2σh ( Nh1+ ( 1−ξ ) Nh2 ) +σa1Na1 and for non-human animals: θa , 1=σa1Na1σh ( Nh1+ ( 1−ξ ) Nh2 ) +σa1Na1 where σi represents the relative preference for human and non-human host types . The probability of biting a human for Nv2 is: θh , 2=σhξNh2σhξNh2+σa2Na2 and of biting a non-human animal: θa , 2=σa2Na2σhξNh2+σa2Na2 The dynamics of infections in animals are given by , dSaidt =βai+ r4Rai − b f θa , i Ivi ( t ) Nai Sai ( t ) −μaiSai , dIaidt =b f θa , i Ivi ( t ) Nai Sai ( t ) − ( μai+η ) Iai , dAaidt= η Iai− ( μai+sai ) Aai , dRaidt = saiAai− ( μai+ r4 ) Rai , where i is either 1 or 2 . As for humans , the population sizes of animals are assumed to be stable , by allowing the birth terms , βai , to consist of the deaths in all compartments . The forces of infection on vectors are: Λv1= cfθh1 , 1Ah1Nh1Sv1+ cfθh2 , 1Ah2Nh2Sv1+ca1fθa , 1Aa1Na1Sv1= f ca1σa1Aa1+c σh ( Ah1+ ( 1−ξ ) Ah2 ) Na1σa1+σh ( Nh1+ ( 1−ξ ) Nh2 ) Sv1 Λv2= cfθh , 2Ah2Nh2Sv2+ca2fθa , 2Aa2Na2Sv2= f ca2σa2Aa2+c σhξ Ah2Na2σa2+σhξ Nh2Sv2 . In the models where animals cannot get infected , Aa1 = Aa2 = 0 , or equivalently , ca1 = ca2 = 0 , in the equations above and tsetse flies can only get infected from humans although they can also bite animals . The ordinary differential equations describing changes in the vector compartments ( with i indicating population 1 or 2 ) are: dSvidt = βvi ( t ) − μvSvi ( t ) −Λvi ( t ) dEvidt =Λvi ( t ) − ( μv+ve ) Evi dIvidt = veEvi−μvIvi For ease of fitting and analysis we allow for the simplification that the birth terms βvi consist of the death terms of all compartments , thereby ensuring stable tsetse populations . The impact of including tsetse population dynamics , density-dependence , migration and seasonality will be considered in a follow-up paper . We obtained parameter sets using a Bayesian framework of importance resampling [43 , 44] . This entailed defining uniform ranges for parameter values; generating 50000 random samples of sets of parameter estimates drawn from the uniform priors ( the same set of random samples was used for the six different scenarios—except for ca1 and ca2 which were set to 0 for the scenarios without animal reservoirs ) ; running the model for 400 years ( in order to avoid resampling simulation runs that move toward their equilibrium state at very slow rates ) and obtaining a measure of the goodness of fit using a binomial likelihood function: Li=∏j=114 ( Nx ) px ( 1−p ) N−x where N is the human population size , p the target prevalence levels associated with high , medium or low transmission , and x is the simulated number of infected humans at j times 10 thousand days for the simulation run with parameter set , i; and randomly sampling 500 parameter sets from these 50000 proportional to their likelihood , ΔLi=LiΣL which are samples from a distribution that forms an approximation of the posterior distribution . Based on these resampled parameter sets , questions of interest can be investigated such as the impact of vector control or screen and treat on prevalence over time . The basic reproduction number , R0 , is defined as the number of secondary infections that arise from a single infected case in a fully susceptible population . It provides insight into whether a pathogen can invade a population , and into which parameters of the disease system to target with control interventions [59] . We derive R0 using a next-generation matrix approach [60] . To do so , we separate the Jacobian matrix of the system into T , a matrix with the terms denoting infection events , and Σ , a matrix with the stage transitions ( due to progression through the incubation period , death , etc . ) . A 16x6 matrix E of zeroes with a 1 in each column at the row corresponding to the infection events in T is also specified . The next-generation matrix K is then given by –E’T Σ-1E: K = [0000k1500000k25k260000k35000000k46k51k52k530000k620k6400] where the elements kij represents the average number of infections ( of vectors or hosts of population i ) arising over the infectious lifetime of one individual of host type j . The columns j correspond to Ah1 , Ah2 , Aa1 , Aa2 , Iv1 , and Iv2 , respectively . The basic reproduction number , R0 , is equal to ρ ( K ) 2 , the spectral radius or dominant eigenvalue of the next generation matrix , squared to reflect the interest in transmission from host to vector to host . | Sleeping sickness , a disease that strikes predominantly poor populations in sub-Saharan Africa , has been targeted for elimination as a public health problem . Despite decades of control operations the disease remains enigmatic and is capable of persisting in populations at low levels of prevalence . Two mechanisms are investigated here that could allow persistence at such levels . Heterogeneous exposure of humans to tsetse is modelled as a subset of humans commuting to areas of high vectorial capacity . Additionally , non-human animals may act as reservoir species . We developed , parameterized , and investigated a model of sleeping sickness transmission to gain insight into the impact of these assumptions on the prospects of elimination using screening and treatment of humans and vector control . Supplemental use of vector control increased the probability of elimination and decreased the duration until elimination was achieved . This was more pronounced when animals did contribute to transmission , or when coverage and compliance of humans with screening operations was lower , for instance due to an inability to reach the humans at greatest risk of exposure . These results can provide insights to public health officials as to when to consider supplementing human treatment with additional measures , and thereby improve the prospects of elimination of this disease . | [
"Abstract",
"Introduction",
"Discussion",
"Methods"
] | [] | 2015 | Implications of Heterogeneous Biting Exposure and Animal Hosts on Trypanosomiasis brucei gambiense Transmission and Control |
TAR-DNA-binding protein-43 ( TDP-43 ) C-terminus encodes a prion-like domain widely presented in RNA-binding proteins , which functions to form dynamic oligomers and also , amazingly , hosts most amyotrophic lateral sclerosis ( ALS ) -causing mutations . Here , as facilitated by our previous discovery , by circular dichroism ( CD ) , fluorescence and nuclear magnetic resonance ( NMR ) spectroscopy , we have successfully determined conformations , dynamics , and self-associations of the full-length prion-like domains of the wild type and three ALS-causing mutants ( A315E , Q331K , and M337V ) in both aqueous solutions and membrane environments . The study decodes the following: ( 1 ) The TDP-43 prion-like domain is intrinsically disordered only with some nascent secondary structures in aqueous solutions , but owns the capacity to assemble into dynamic oligomers rich in β-sheet structures . By contrast , despite having highly similar conformations , three mutants gained the ability to form amyloid oligomers . The wild type and three mutants all formed amyloid fibrils after incubation as imaged by electron microscopy . ( 2 ) The interaction with nucleic acid enhances the self-assembly for the wild type but triggers quick aggregation for three mutants . ( 3 ) A membrane-interacting subdomain has been identified over residues Met311-Gln343 indispensable for TDP-43 neurotoxicity , which transforms into a well-folded Ω-loop-helix structure in membrane environments . Furthermore , despite having very similar membrane-embedded conformations , three mutants will undergo further self-association in the membrane environment . Our study implies that the TDP-43 prion-like domain appears to have an energy landscape , which allows the assembly of the wild-type sequence into dynamic oligomers only under very limited condition sets , and ALS-causing point mutations are sufficient to remodel it to more favor the amyloid formation or irreversible aggregation , thus supporting the emerging view that the pathologic aggregation may occur via the exaggeration of functionally important assemblies . Furthermore , the coupled capacity of TDP-43 in aggregation and membrane interaction may critically account for its high neurotoxicity , and therefore its decoupling may represent a promising therapeutic strategy to treat TDP-43 causing neurodegenerative diseases .
The TAR-DNA-binding protein-43 ( TDP-43 ) was initially identified as a factor capable of binding to the TAR DNA of HIV and repressing transcription [1] , which are well conserved among Caenorhabditis elegans , Drosophila , mouse , and human [2] . In 2006 , the human TDP-43 was identified as the major constituent of the proteinaceous inclusions that are characteristic of most forms of amyotrophic lateral sclerosis ( ALS ) and the most common pathological subtype of frontotemporal dementia-frontotemporal lobar degeneration with TDP-43-positive inclusions ( FTLD-TDP ) [3 , 4] . TDP43 is an intrinsically aggregation-prone protein [3–14] , and its irreversible aggregation has been found in ~97% ALS and ~45% FTD patients . Additionally , TDP-43 immunoreactive inclusions have also been observed in an increasing spectrum of other neurodegenerative disorders , which include ALS/parkinsonism–dementia complex of Guam , Alzheimer disease ( AD ) , dementia with Lewy bodies ( DLB ) , Pick’s disease , argyrophilic grain disease and corticobasal degeneration ( reviewed in 5 , 12 ) . Very recently , TDP‑43 has been identified as a key player in the clinical features associated with Alzheimer disease , in particular , cognitive impairment [15] . TDP-43 is a member of the heterogeneous nuclear ribonucleoprotein ( hnRNP ) family , which includes some of the well-known splicing modulators , such as hnRNP I , hnRNP A/B , and hnRNP H [5–7 , 12–14 , 16] . Previously , the 414-residue TDP43 was established to be composed of a nuclear localization signal ( NLS ) , two RNA recognition motifs ( RRM1 and RRM2 ) hosting a nuclear export signal ( NES ) , and C-terminal Q/N/S/G-rich domain ( Fig 1A ) . The NLS and NES regulate the shuttling of TDP-43 between the nucleus and the cytoplasm [17] , while the RRM1 and RRM2 have been characterized to bind to a large variety of nucleic acids including single- or double-stranded DNA/RNA [17–21] . Intriguingly , the C-terminal domain over residues 274–414 has a low-complexity sequence abundant in Gln , Asn , Ser , and Gly residues , which shares 24 . 2% sequence identity with the N-terminal yeast prion domain of Sup35 [13 , 14 , 22] . The critical role of the C-terminus in ALS pathogenesis has also been strongly highlighted by the fact that this domain hosts almost all known ALS-associated mutations and has also been proposed to be responsible for the prion-like spreading of ALS [2 , 5–7 , 12–14] , thus called prion-like domain [22–25] . Noticeably , such low-complexity domains have been shared by a large number of RNA-binding proteins , many of which were identified to be involved in neurodegenerative disease [12–14 , 22 , 25 , 26] . The functional studies revealed that the TDP-43 prion-like domain functions to form reversible oligomers or high-order granule by self-association or complexing with protein partners such as other hnRNPs [5–7 , 12–14 , 16 , 22 , 25–27] . For example , TDP-43 was shown to form 50–250 nm granule in the nucleus , providing a scaffold for other functionally related sub-compartments , that participates in transcriptional repression as well as alternative splicing [14 , 28] . On the other hand , in the cytoplasm , TDP-43 also participates in forming RNP granule , which include processing bodies ( P-bodies ) and stress granule ( SGs ) [5 , 13 , 29 , 30] . As such , it has been recently proposed that the physiological and reversible structures of TDP-43 serve as precursors , which transform into irreversible inclusions under certain pathological conditions [5 , 13 , 29 , 30] . Since its discovery , the mechanism for the TDP-43 aggregation has become a central focus , although the exact role of the aggregation in TDP-43 neurotoxicity remains controversial [31] . The TDP-43 oligomerization/aggregation appears to be cooperatively mediated by several regions , particularly by N- and C-termini which have very high tendency to aggregate in vitro and in vivo [31–33] . Surprisingly , recent studies revealed that like superoxide dismutase 1 ( SOD1 ) , TDP-43 is also capable of becoming associated with mitochondria to impair mitochondrial dynamics and function in motor neurons [34 , 35] . Previously , it has been extensively demonstrated that ALS-causing mutations transform the cytosolic SOD1 into a membrane-interacting protein that thus became associated with organelles such as mitochondria [36–38] . Indeed , recently the peptides derived from the TDP-43 prion-like domain were identified to have the membrane-damaging capacity [39] . Due to its extremely important role in a large spectrum of neurodegenerative diseases , the neurotoxicity of the wild-type and mutant TDP-43 have been extensively investigated by various cell and animal models [5–8 , 11–14 , 16 20 , 26 , 27 , 31 , 32 , 34 , 35 , 40–44] . On the other hand , the elucidation of the high-resolution structures , dynamics , and self-association of the TDP-43 domains , particularly for the prion-like domain , represents a crucial step towards delineating the mechanistic aspects of its physiological and pathological functions . Unfortunately , however , due to the intrinsic propensity of aggregation , this task has been significantly retarded except for the RRMs [21] . Consequently , only dissected prion-like fragments have been previously investigated by nuclear magnetic resonance ( NMR ) [43–45] . Very recently , as facilitated by our previous discovery that protein aggregation can be significantly minimized by reducing salt concentrations [46 , 47] , we have successfully decrypted that the TDP-43 N-terminus unexpectedly encodes a novel ubiquitin-like fold coexisting with its unfolded form in equilibrium , thus rationalizing its high tendency in aggregation [33] . Here by circular dichroism ( CD ) , fluorescence , electron microscopy , and NMR studies on the full-length prion-like domains of the wild type and three ALS-causing mutants , namely A315E , Q331K , and M337V , we aimed to gain insights into three important aspects associated with the TDP-43 proteinopathies: ( 1 ) What are the high-resolution pictures of the conformation , dynamics and self-assembly of the wild-type prion-like domain ? ( 2 ) How do the ALS-causing mutations affect these properties ? ( 3 ) Which unique feature might account for the high neurotoxicity of the TDP-43 inclusion ? Our study reveals that although the TDP-43 prion-like domain is intrinsically disordered only with some nascent secondary structures , it has the capacity to assemble into dynamic oligomers by self-association or interacting with nucleic acid . Unexpectedly , all three ALS-causing point mutations are able to significantly perturb this capacity . Furthermore , we identified a region previously demonstrated to be indispensable for the TDP-43 neurotoxicity is in fact a membrane-interacting subdomain . Taken together , our results support the emerging view that the pathologic aggregation of proteins including TDP-43 in neurodegenerative diseases may occur via the exaggeration of functionally important and reversible assemblies [5 , 6 , 14 , 22 , 25 , 26 , 29 , 30 , 48–50] . Our study also implies that the coupling of the TDP-43 aggregation and membrane interaction might at least partly account for its high toxicity .
The TDP-43 C-terminus over residues 263–414 containing the full-length prion-like domain ( 274–414 ) and a short linker ( 263–273 ) ( Fig 1A ) was cloned , expressed , and purified as described in Methods . It is highly soluble in Milli-Q water ( pH 4 . 0 ) with a protein concentration up to 600 μM , similar to what we extensively found on other “insoluble” proteins , including the TDP-43 N-terminus [33 , 38 , 46 , 47] . As judged by its far-ultraviolet ( UV ) CD spectrum with the maximal negative signal at 199 nm and no positive signal at 190 nm ( Fig 1B ) , it appears to be highly disordered without any stable secondary structure . Moreover , it has a 1H-15N heteronuclear single quantum coherence spectroscopy ( HSQC ) spectrum with very narrow 1H ( 0 . 86 ppm ) and 15N ( 17 . 84 ppm ) spectral dispersions in which three Trp residues have their sidechain HSQC peaks largely overlapped ( Fig 1C ) . These observations indicate that it also has no tight tertiary packing . Nevertheless , the HSQC peaks are well separated , and the sample at 600 μM showed no detectable changes in CD and NMR spectra for several months in Milli-Q water ( pH 4 . 0 ) , thus allowing the collection of a large set of high-quality NMR spectra . By diluting the protein in Milli-Q water into phosphate buffer , we were able to prepare the samples in 1 mM phosphate buffer at different pH values for CD and NMR characterization . In pH 5 . 0 buffer , the prion-like domain has a far-UV CD spectrum almost identical to that in pH 4 . 0 Milli-Q water , and an HSQC spectrum with the majority of peaks superimposable to those in pH 4 . 0 Milli-Q water , except for those of His-tag and several N-/C-terminal residues that become shifted or disappeared ( Fig 1C ) . Evidently , HSQC peaks for Gly residues distributed over the whole sequence are almost completely superimposable under two conditions . Furthermore , the samples at 100 μM also showed no significant changes in CD and NMR spectra in pH 5 . 0 buffer for several weeks , suggesting that it has no significant conformational difference under two conditions , as well as no significant self-association . We also characterized it in phosphate buffers at higher pH including 6 . 0 and 6 . 8 . As shown in Fig 1B , the sample at pH 6 . 8 immediately prepared has a CD spectrum almost superimposable to that in Milli-Q water , implying that it also has no significant difference of secondary structures under two conditions . Consistent with CD results , except for the disappearance of peaks of the His-tag and some N-/C-terminal residues , many HSQC peaks including most of Gly residues are still superimposable under two conditions . This provides residue-specific evidence that the prion-like domain has no significant difference of the solution conformation at pH 4 . 0 and 6 . 8 . As temperature has large effects on the disordered proteins , we also systematically assessed the temperature-induced conformational changes at pH 4 . 0 , 5 . 0 and 6 . 0 by CD spectroscopy ( S1 Fig ) . Conformational changes are relatively small , in particular below 40°C . Therefore , in the present study we conducted all CD and NMR characterization at 25°C in aqueous solutions . Interestingly , the sample , even at a concentration of 20 μM , started to self-associate in pH 6 . 8 buffer , as monitored by CD ( Fig 1B ) . After 1 d , the CD signal intensity showed a slight reduction , while after 4 d , the CD spectrum changed dramatically which is similar to what have been observed on the soluble β-stranded oligomers formed by the peptides derived from the TDP-43 prion-like domain [43–45] . After 8 d , no further changes were detected and also no visible aggregate was formed . This implies that the wild-type prion-like domain is able to progressively assemble into larger but soluble oligomers at neutral pH . The deconvolution analysis of the CD spectra revealed that the soluble oligomer formed at 8 d contains ~3% helical but ~72% β-sheet and β-turn , as well as ~25% random coil conformations . We also monitored the changes in pH 6 . 8 buffer by HSQC spectra at a protein concentration of 100 μM . At this higher protein concentration , the formation of larger oligomers appeared to be much faster . Even after 15 min , many HSQC peaks became broadened ( Fig 1D ) , implying the occurrence of dynamic oligomerization . After 4 hr ( Fig 1E ) and 9 hr ( Fig 1F ) , most peaks became very broad and consequently the intensity became weak . After 1 d , most peaks became too broad to be detectable ( Fig 1G ) . Furthermore , the NMR sample in pH 6 . 8 buffer formed hydrogels after 1 d , which could change back to solution upon shaking , similar to what was also observed on the fused in sarcoma ( FUS ) prion-like domain [30 , 49] . However , higher protein concentrations such as at 200 μM would result in the rapid precipitation with white aggregates upon dilution into the buffer at pH 6 . 8 . To slow down the self-association , particularly for the A315E and M337V mutants to allow detailed NMR characterization , we also acquired an array of one-dimensional 1H and HSQC spectra of the wild type and three mutants at a protein concentration of 40 μM in 1 mM phosphate buffer at pH 6 . 8 . Fig 2 presents NMR spectra over 0 . 6–0 . 96 ppm , which are from the non-labile methyl protons . As seen in Fig 2A , in Milli-Q water at pH 4 . 0 , there are only three large clusters of peaks and lacking of very up-field peaks , indicating that the wild-type prion domain is highly disordered , consistent with the above CD and NMR results . However , 15 min after the dilution of the wild type into 1 mM phosphate buffer at pH 6 . 8 , two very up-field NMR peaks manifested respectively at 0 . 680 and 0 . 689 ppm ( Fig 2A ) . At 14 hr , the intensity of the two peaks became the largest although the intensity of other peaks reduced significantly due to the self-association to form large oligomer . After 14 hr , the intensity of the two peaks started to reduce and mostly disappeared at 24 hr ( Fig 2A ) . For proteins , such very up-field peaks are resulting from the methyl protons which have tight stack interaction with aromatic ring . As no new peaks manifested in the corresponding HSQC spectra , the manifestation of the up-field two peaks is most likely to result from the oligomeric form whose HSQC peaks were too broad to be detected . Interestingly , at a protein concentration of 40 μM in 1 mM phosphate buffer at pH 6 . 8 , hydrogel was formed in NMR tube only after 1 week , much slower than that observed for the NMR sample at 100 μM . To understand why some point mutations on the intrinsically disordered TDP-43 prion-domain are sufficient to trigger ALS , we successfully generated recombinant proteins of three ALS-causing mutants , A315E in the Ω-loop , Q331K in the middle of the helix , and M337V in the C-half of the helix ( S2A Fig ) . As shown in S2B Fig , only Q331K has a CD spectrum slightly different from that of the wild type , whereas A315E and M337M have CD spectra almost superimposable to that of the wild type . The result clearly indicates that in aqueous solution , all three mutants are similarly disordered as the wild type . Indeed , except for the mutated residues , HSQC peaks of three mutants are also highly superimposable to those of the wild type . For example , the A315E mutant only has slight peak shifts of residues Gly314 , Phe316 , Ser317 , and Ile318 which are close to the mutation site A315E in sequence ( S2C Fig ) . On the other hand , the M337V mutant has slight peak shifts of residues Gly335 and Gly338 which are close to the mutation site , but of additional residues Leu340 , Ser342 and Gln343 ( S2E Fig ) . Interestingly , the Q331K mutant has slight peak shifts of the most extensive residues , which include Ala326 , Gln327 , Ala328 , Ala329 , Leu330 , Ser333 , Trp334 , and Gly335 , as well as Ala324 and Met337 ( S2D Fig ) . Nevertheless , minor shifts of HSQC peaks of three mutants indicate that they have monomeric conformations very similar to that of the wild type , consistent with CD results . We further conducted CD characterization of the self-association of three mutants at a protein concentration of 20 μM in 1 mM phosphate buffer at pH 6 . 8 . Very unexpectedly , as shown in Fig 3A–3D , unlike the wild type , all three mutants finally transformed into the conformations which have CD spectra typical of the amyloid oligomers , as previously well-documented [43–45] . Interestingly , the A315E and M337V mutants transformed into the amyloid oligomers respectively after 18 hr ( Fig 3B ) and 1 d ( Fig 3D ) , whereas it took 9 d for the Q331K mutant to complete the transformation ( Fig 3C ) . We also monitored the changes of NMR spectra over 0 . 6–0 . 96 ppm for three mutants together with the wild type under the exactly same conditions ( Fig 2 ) . Interestingly , three mutants showed no manifestation of two very up-field peaks during the self-association . Consistent with CD results ( Fig 3 ) , the A315E and M337V mutants largely completed the self-association at 8 hr ( Fig 2B and 2D ) , while the Q331K mutant took 8 d ( Fig 2C ) . Unlike the wild type , A315E and M337V mutants whose self-associations were significantly speeded up by higher protein concentrations used for NMR ( 40 μM ) , the amyloid-formation of Q331K is less concentration-dependent as it showed only a small difference at protein concentrations for CD ( 20 μM ) and NMR ( 40 μM ) . We further utilized fluorescence spectroscopy to characterize the self-association of the wild type and three mutants , which is one of the most common techniques to identify the formation of amyloid-like structures [51] . Briefly , we monitored the time-lapsed changes of three fluorescence probes , which include the intrinsic UV , visible fluorescence , and induced fluorescence by binding to Thioflavin T ( ThT ) [51–56] . The TDP-43 prion domain contains three Trp residues: Trp334 , Trp385 , and Trp412 and consequently has detectable intrinsic UV fluorescence [51] . Indeed , as shown in S3 Fig , the wild type and three mutants have very similar emission spectra in Milli-Q water ( pH 4 . 0 ) , with the emission maxima at ~351 nm , implying that three Trp residues have similar exposure in the wild type and three mutants . The slight intensity differences are most likely due to minor changes of the chemical environments triggered by mutations [51] . Interestingly , upon dilution into 1 mM phosphate buffer ( pH 6 . 8 ) , the wild type has the largest blue-shift of the emission maximum from 351 to 347 nm ( S3A Fig ) , implying that the wild type started to self-assemble immediately , and consequently its Trp residues became more buried . After 1 d , the emission maximum of the wild type further blue-shifted to 342 nm , then to 339 nm after 6 d , and no more significant change occurred after 8 d ( S3A Fig ) . Similar patterns of changes were observed for the three mutants . The results , particularly with the blue-shift of the emission maxima , suggest that the wild type and three mutants are all able to undergo the self-association into the oligomers in which Trp residues become more shielded from bulk solvent [51] . It is well established that the binding of ThT is a diagnostic probe for the formation of the β-rich amyloid structures although the exact molecular details still remain elusive [51] . So , we also monitored the binding of the wild type and three mutants to ThT at different incubation time points . Interestingly , as seen in S3E Fig , immediately upon dilution into the 1 mM phosphate buffer ( pH 6 . 8 ) , the wild type showed a large intensity of the ThT-binding induced fluorescence with the emission maximum at ~488 nm , implying its fast formation of the β-rich amyloid-like structure . After 1 d , the intensity was further increased and then reached the highest point after 2 d , followed by the reduction of the intensity afterward . The similar patterns were observed on A315E and M337V , but for Q331K , the intensity reached the highest point only after 4 d . The decrease of the ThT binding after long incubation times has been extensively observed because the fibrils may become packed together in such a way that the surface for ThT binding becomes less accessible [51 , 56] . Recently , it has been found that an intrinsic visible fluorescence develops during the β-rich fibrillar aggregation of amyloid-β ( 1–40 ) and ( 1–42 ) , lysozyme as well as tau [54] . More specifically , this intrinsic visible fluorescence has been characterized to be independent of the presence of aromatic side-chain residues but to have its origin in the formation of special hydrogen bonds involved in the backbone C = O and N-H atom groups of peptide bonds , which already have electron delocalization to some degree . The formation of such hydrogen bonds will further enhance electron delocalization and thus allow low energy electronic transitions [53–55] . As shown in Fig 3E , the incubation buffer had no detectable visible fluorescence . However , even in water at pH 4 . 0 , this intrinsic visible fluorescence could be detected for the wild type and three mutants even at 40 μM concentration , which have similar emission maxima: 446 nm for the wild type ( Fig 3E ) , 447 nm for A315E ( Fig 3F ) , 446 nm for Q331K ( Fig 3G ) , and 449 nm for M337V ( Fig 3H ) . This implies that for the TDP-43 prion-like domain , a large number of hydrogen bonds involved in backbone atoms already exist even in the highly-disordered monomeric states , because previously this visible fluorescence only became detectable for lysozyme ( emission maximum at 425 nm ) and γ-crystallin ( emission maximum at 465 nm ) in solution at a very high concentration: 100 mg/ml [53] . Remarkably , only 5 min after the dilution of the wild type into 1 mM phosphate buffer ( pH 6 . 8 ) , the intensity of this fluorescence was almost doubled ( from reading of 24 to 45 ) , with a slight red-shift of the emission maximum from 446 to 450 nm ( Fig 3E ) . After 2 d , this fluorescence reached the highest with no significant shift of the emission maximum , followed by the reduction of the intensity afterwards . By a sharp contrast , the changes of this fluorescence showed a different pattern for three mutants . Upon immediate dilution into 1 mM phosphate buffer ( pH 6 . 8 ) , three mutants only showed slight intensity increases with slight red-shifts of the emission maxima ( Fig 3F–3H ) . Nevertheless , after 1 d , the emission spectra of three mutants became very different from that of the wild type . For example , after 1 d , the emission maxima have significantly red-shifted to 475 nm for A315E ( Fig 3F ) , 468 nm for Q331K ( Fig 3G ) , and 472 nm for M337V ( Fig 3H ) . Only after 6 d , the intensities of all three mutants reached the highest , with the emission maxima significantly red-shifted as compared to that for the wild type ( 450 nm ) : 479 nm for A315E , 476 nm for Q331K , and 479 nm for M337V . The red-shift of the emission maximum of this intrinsic visible fluorescence has been established to be correlated to the amount and arrangement of the hydrogen bonds involved in the backbone peptide bonds: the higher the amount of hydrogen bonds arranged in the β-rich amyloid-like structures , the larger the red-shift will be [52–56] . For example , lysozyme with a mixture of α-helix and β-sheet secondary structures has the emission maximum at 425 nm , while γ-crystallin with β-sheet dominant secondary structures has its emission maximum significantly red-shifted to 465 nm [53] . Remarkably , the emission maxima for three mutants are even larger than that of an amyloid nanofibrils formed by Poly ( ValGlyGlyLeuGly ) peptide ( ~468 nm ) . Strikingly , this nanofibril has been found to become electricity-conductive due to its well-formed amyloid β-structures , which thus owns a highly ordered hydrogen bond network so as to allow radical electron delocalization [55] . Therefore , it appears that even in the highly disordered states of the wild type and three mutants in water at pH 4 . 0 , there already exist a large amount of hydrogen bonds involved in the peptide bonds , as evidenced by their detectable intrinsic visible fluorescence . Furthermore , upon dilution into the buffer at pH 6 . 8 , the prion-like domains of the wild type and three mutants will start the self-association , which thus leads to the formation of inter-molecular β-sheet structures , and consequently results in significant intensity increases of this fluorescence . Together with CD and NMR results , the results with this visible fluorescence suggest that the wild type self-assembles into the oligomer very different from those by three mutants . It appears that the oligomer formed by the wild type may still contain a small portion of disordered/dynamic regions lacking of hydrogen bonds which are involved in peptide bond atoms and arranged in β-amyloid structures . Consequently , the oligomer formed by the wild type has the emission maximum at 450 nm , larger than that of lysozyme ( 425 nm ) but smaller than that of γ-crystallin ( 465 nm ) . By contrast , the three mutants , although upon immediate dilution they behave similarly to the wild type , acquired the capacity to further transform into the well-formed amyloid structures with very different CD spectra and emission maxima of this fluorescence ( Fig 3 ) . In the future , it is of fundamental interest to investigate whether the fibrils formed by the TDP-43 prion like domains , particularly by three mutants , are also conductive , and if yes , whether it has any relevance to the physiological functions or/and pathological roles . We further used electron microscope ( EM ) to visualize the morphology of the self-association for the wild type and three mutants at the incubation times of one and two weeks . As seen in Fig 4A–4D , after one week , the wild type and three mutants were all able to form amyloid fibrillar structures , with the widths of fibrils ranging from 15 to 30 nm , similar to the fine structures that were detected in the neuronal TDP-43 inclusions in patients’ brain tissues [57] . After 2 wk , the amyloid fibrillar structures remained similar but they appeared to become slightly more clustered/condensed together ( S4 Fig ) . On the other hand , amorphous structures of diverse sizes were observed for the aggregates rapidly formed by diluting the wild type into 1 mM phosphate buffer at pH 6 . 8 to reach a concentration of 200 μM ( Fig 4E ) . It is interesting to point out that the fibrillar structures in patients’ brain tissues have been demonstrated to fail to significantly bind ThT , thus representing non-classical amyloid fibrils . In the future , it is of fundamental interest to investigate whether the inability of the pathological TDP-43 aggregates to bind ThT might be due to the loss of accessibility of the ThT-binding sites as we observed here after long incubation times ( S3E and S3H Fig ) . Here , by analyzing three-dimensional NMR spectra including CCC ( CO ) NH , HN ( CO ) CACB , HSQC-TOCSY ( total correlation spectroscopy ) , and HSQC-NOESY ( nuclear Overhauser effect spectroscopy ) , we have successfully achieved the sequential assignments of all non-Proline residues except for residues Gln286-Gly287 , Gly290-Asn291 , Gly295 , Asn301-Asn302 , Asn371-Asn372 , Ser393-Ser395 , and Phe401-Gly402 , whose NMR resonances were either undetectable or severely overlapped . Fig 5A presents the ( ΔCα-ΔCβ ) chemical shifts , which represent a sensitive indicator of the residual secondary structures in disordered proteins [38 , 58 , 59] . The small absolute values of ( ΔCα-ΔCβ ) chemical shifts over the whole sequence clearly indicate that it is indeed lacking of any stable secondary structure , completely consistent with its CD results ( Fig 1B ) . Nevertheless , several regions have relatively large deviations . For example , residues Pro320-Gln331 all have the ( ΔCα-ΔCβ ) larger than 1 ppm , suggesting that this region is populated with helical conformation to some degree . Furthermore , despite very small , residues Ser317-Ile318-Asn319 have negative ( ΔCα-ΔCβ ) values , implying that they may adopt a relatively extended conformation , consistent with a previous finding that these residues were involved in forming an intermolecular β-sheet in the isolated peptide Met307-Asn319 [45] . To gain quantitative insights into the populations of different secondary structures , we further analyzed NH , N , Hα , Cα and Cβ chemical shifts of the prion-like domain by SSP program [60] . As seen in Fig 5B , all residues have the absolute values of SSP scores less than 0 . 5 , confirming that the whole domain has no stable secondary structure . Nevertheless , residues Pro320-Leu340 have SSP scores larger than 0 . 2 while residues Ser403-Met414 have SSP scores larger than 0 . 15 , thus implying that they are populated with helical conformations to some degree , resembling nascent helix . Indeed , residues Pro320-Leu340 were previously found to adopt stable helical conformation in an isolated peptide of Met311-Gln360 [43] . However , a detailed comparison with our present results is impossible as the previous NMR data and structure was not deposited . Interestingly , some short segments also have negative SSP scores , which include but are not limited to Gly274-Gly277 , Gly314-Ser317 , Pro363-Gly368 , Ala382-Gly384 , implying that these regions might have intrinsic capacity to adopt extended conformations to some degree . They may serve as seeds/nucleation sites to form amyloids or/and inclusions which are rich in extended β-conformations . We also assessed the backbone rigidity by collecting heteronuclear NOEs , which provides a measure to the backbone flexibility on the pico- to nanosecond ( ps-ns ) timescale [33 , 38 , 58] . As shown in Fig 5C , the backbone is overall flexible on ps-ns time scale as judged from the small or even negative heteronuclear NOEs ( hNOEs ) , with an average value of only 0 . 07 ( Fig 5C ) . However , residues Gly314-Met339 have hNOE all larger than 0 . 25 , implying that this region has relatively-restricted backbone motions , which is likely due to the presence of partly-populated secondary structures . Furthermore , analysis of the HSQC-NOESY spectrum indicated that even for Met311-Ala341 , only some dαN ( i , i+2 ) and dNN ( i , i+2 ) , but no dαN ( i , i+3 ) NOEs could be found ( Fig 5E ) , suggesting that the helical conformations over this region are only dynamically populated in the full-length domain . The lack of stable secondary structure appears due to the abundance in polar residues Gln , Asn , Ser , and Gly in the TDP-43 prion-like domain . Indeed , all residues have negative hydrophobicity score [61] , except for two regions Met311-Met339 and Ile383-Ser387 ( S5A Fig ) . An interesting question is why the TDP-43 prion-like domain showed no significant oligomerization in water at pH 4 . 0 , or even in 1 mM phosphate buffer at pH 5 . 0 , but started to oligomerize at pH 6 . 8 although the initial solution conformations are very similar at these pH values ? On the other hand , it is also interesting to observe very minor shifts of HSQC peaks at different pH values ( Fig 1C and 1D ) , because for disordered proteins , their amide protons are expected to be highly exposed to bulk solvent . As a consequence , even minor changes of solution conditions such as pH will trigger significant shifts of HSQC peaks due to the changes of the chemical environments , although the conformation may remain largely unchanged [58] , as we previously observed on the disordered ALS-causing SOD1 mutant [38] , P56S-MSP domain [62] and isolated dengue NS3 protease domain [63] . This observation , together with the above results with the intrinsic visible fluorescence , strongly implies that the majority of the backbone amide protons might be involved in hydrogen bonding in the TDP-43 prion-like domain . Therefore , we measured NMR temperature coefficients of the backbone amide protons in Milli-Q water at pH 4 . 0 , in 1 mM phosphate buffers at pH 5 . 0 and pH 6 . 0 ( Fig 5D ) , which represents a sensitive NMR probe for the involvement of the amide protons in hydrogen bonding [58 , 59 , 64] . Surprisingly , at pH 4 . 0 , the backbone amide protons of the most residues have very small temperature coefficients , with an average of 4 . 3 , which are not only much smaller than those of the disordered peptides [58 , 59] but also around or even smaller than 4 . 6 ppb/K , which was defined as an indicator of the involvement in well-formed hydrogen-bonds based on the studies on 793 backbone amides derived from 14 well-folded proteins [64] . This implies that the majority of the backbone amide protons are engaged in hydrogen bonding at pH 4 . 0 . Previously , it has been found that the side chain atoms of Asn , Gln and Ser have particularly strong capacity in forming hydrogen bonds with the backbone atoms in both well-folded proteins as well as short peptides , which include the hydrogen bonds between the side chain oxygen and backbone amide protons [65 , 66] . One the other hand , it has been proposed that poly-Q sequences behave as polar zippers , which are able to oligomerize by forming inter-molecular hydrogen bonds between side-chain and backbone atoms [22 , 67] . As such , here we propose that in the TDP-43 prion-like domain at pH 4 . 0 , most backbone amide protons are involved in forming hydrogen bond networks with the side-chain oxygen atoms , thus resulting in the very small NMR temperature coefficients and minor shifts of HSQC peaks at different pH , as well as manifestation of the intrinsic visible fluorescence ( Fig 3E ) . These hydrogen bond networks might be most likely to be intra-molecular , rather than inter-molecular , as we have also collected CPMG-based 15N relaxation data on the TDP-43 prion-like domain as we previously did on SOD1 [38] but found no response , thus implying no significant inter-molecular association as protein association occurs on the μs-ms time scale , which should be detected by CPMG-based dispersion experiments [68] . In particular , this successfully rationalizes the above observation that the wild type and three mutants could have detectable intrinsic visible fluorescence even at a very low concentration ( 40 μM ) in the water ( pH 4 . 0 ) , in which the TDP-43 prion-like domains are mostly monomeric and highly disordered ( Fig 3E–3H ) . On the other hand , at pH 5 . 0 , most backbone amide protons have slightly larger temperature coefficients , with an average of 4 . 7 ( Fig 5D ) . Furthermore , at pH 6 . 0 , for the detectable HSQC peaks , they have further increased temperature coefficients ( with an average of 5 . 2 ) , in particular over Ser369-Asn378 . These observations might be explained by the possibility that at higher pH , the exchange rates of most backbone amide protons increase due to the base-catalyzed exchange , as the prion-like domain is highly disordered with backbone amide protons largely accessible to bulk solvent [69] . As a result , at higher pH , the dissociation between amide proton and nitrogen atoms would become increased , thus leading to the disruption of the intra-molecular hydrogen-bonds involved in backbone amide protons of the TDP-43 prion-like domain at pH 4 . 0 . As a consequence , at neutral pH , the side chains of Asn , Gln , and Ser will be mostly liberated and become available to form intermolecular polar zipper [22 , 67] , or/and steric zipper [70] . We also determined the temperature coefficients of three mutants at different pH values ( S6 Fig ) . However , for the A315E and M337V mutants , a large portion of HSQC peaks became too broad or even disappeared at pH 6 . 0 when temperature was above 30°C . As such , we were unable to obtain the temperature coefficients of the A315E and M337V mutants at pH 6 . 0 . As shown in S6A and S6B Fig , most A315E residues have the temperature coefficients ( with an average of 4 . 8 at pH 4 . 0 and 4 . 9 at pH 5 . 0 , respectively ) larger than those of the wild type . Similarly , most M337V residues ( S6F and S6G Fig ) also have the temperature coefficients ( with an average of 4 . 9 at pH 4 . 0 and 5 . 0 at pH 5 . 0 , respectively ) larger than those of the wild type . By contrast , many Q331K residues have the temperature coefficients ( with an average of 4 . 2 at pH 4 . 0 , 4 . 6 at pH 5 . 0 and 4 . 9 at pH 6 . 0 , respectively ) slightly smaller than those of the wild type . This set of results thus implies that the incubation time needed for completing the self-association appears to be correlated to , but the conformations of the final self-associated states have no clear correlation to the overall stability of the inter-molecular hydrogen-bonding network , because the Q331K mutant has slightly lower average temperature coefficients than the wild type , but still transforms into the amyloid oligomer after long incubation time as reported by its CD ( Fig 3C ) and fluorescence ( Fig 3G ) spectra . The low-complexity sequences homologous to the TDP-43 prion-like domain have been extensively identified in DNA/RNA-binding proteins , which function to form oligomers for binding a large spectrum of nucleic acids including single- and double-stranded DNA/RNA . Remarkably , a large set of such domains alone without RNA-binding motifs , which include those from FUS and TDP-43 , have been recently characterized to be sufficient to bind nucleic acids to facilitate the assembly into dynamic β-dominant oligomers [30 , 49] . Therefore , here we used CD and NMR spectroscopy to characterize the interactions of nucleic acid with the wild-type and three mutant TDP-43 prion-like domains . As RNA triggered rapid self-associations or even aggregation , and thus did not allow the differentiation between the wild type and the mutants , here we used single-stranded DNA ( ssDNA ) which has been identified to bind TDP-43 [33] . To minimize the nucleic-acid-independent self-association at neutral pH as we showed above , we conducted the binding characterization in 1 mM phosphate buffer at pH 5 . 0 , in which both wild-type and mutant TDP-43 prion-like domains showed no significant self-associations in one week . Indeed , as characterized by CD spectroscopy , ssDNA was able to bind and trigger the significant conformational changes for the wild-type prion-like domain ( S7A Fig ) . Interestingly , no aggregate was formed during the titration and the CD sample formed hydrogels even at pH 5 . 0 and a protein concentration of 20 μM shortly after reaching the ratio of 1:1 , thus preventing from further adding ssDNA . By contrast , all three mutants showed the behaviors different from that of the wild type upon titrations with ssDNA . Once the ratio exceeded 0 . 3 for A315E ( S7B Fig ) , 0 . 6 for Q331K ( S7C Fig ) and 0 . 4 for M337V ( S7D Fig ) , the mutant proteins precipitated immediately with white aggregates and thus no good-quality CD spectra could be acquired further . We also monitored the interaction of ssDNA with the wild-type and mutant prion-like domains by one-dimensional 1H NMR and HSQC spectra ( Fig 6 ) . For the wild type , the dilution of the protein into 1 mM phosphate buffer at pH 5 . 0 would not initiate the self-association as observed at pH 6 . 8 ( Fig 2A ) , as evidently from the absence of any very up-field two NMR peaks ( Fig 6A ) . However , upon addition of ssDNA at a ratio of 1:0 . 5 ( protein:ssDNA ) , two very up-field peaks started to manifest and their intensity become much higher at a ratio of 1:1 ( Fig 6A ) . Interestingly , the two peaks induced by ssDNA at pH 5 . 0 have chemical shifts ( at 0 . 683 and 0 . 690 ppm respectively ) similar to those of the oligomer ( at 0 . 681 and 0 . 689 ppm respectively ) formed at pH 6 . 8 without ssDNA . Furthermore , we also monitored the interactions by HSQC spectra ( Fig 6B ) . Very interestingly , for the wild type , although at a ratio of 1:0 . 5 many peaks became too broad to be detected mostly due to the involvement in forming the large oligomer , a set of HSQC peaks is still detectable and mostly superimposable to those in the free state ( Fig 6B ) . Based on the sequential assignment , the remaining peaks were all from the C-terminal residues Gln343-Met414 except for those from Gly304 and Gly309 ( S8 Fig ) . This implies that the C-terminal residues Gln343-Met414 still remain largely flexible even in the oligomer of the wild-type prion-like domain complexed with ssDNA . By contrast , all three mutants behaved very different from the wild type in interacting with ssDNA , as evidenced firstly by the absence of the very up-field peaks for A315E ( Fig 6C ) , Q331K ( Fig 6E ) , and M337V ( Fig 6G ) . Secondly , upon exceeding certain ratios of protein:ssDNA , the mutant proteins precipitated rapidly with white aggregates and their 1D and HSQC peaks became too weak to be detected ( Fig 6C–6H ) , consistent with CD results . The capacity in disrupting the plasma and organelle membranes has been identified in all aggregation prone proteins causing human diseases including amyloid-β peptide [71–73] , the diabetes related peptide IAPP [73–75] and SOD1 mutants [36–38] . Very recently , imaging studies revealed that mutant huntingtin inclusions appeared to be “engulfed” in the nuclear membrane to lead to eventual neuronal death [76] . Indeed , membrane-interacting fragments/domains have been identified in these proteins and their NMR structures have been determined in membrane environments [36 , 71–75 , 77 , 78] , such as in lipid-mimetic dodecylphosphocholine ( DPC ) micelle , which is amenable to liquid NMR characterization at atomic resolution . Here , we addressed the question of whether the TDP-43 prion-like domain also contains any membrane-interacting subdomain . To achieve this , we first titrated the prion-like domain with the large bicelle composed of DMPC/DHPC at a q value of 4 , which resemble native bilayer membranes [79] . As seen in Fig 7A , in the presence of bicelle at a ratio of 1:200 ( prion:bicelle ) , the prion-like domain has a far-UV CD spectrum very different from that in aqueous solution , with the maximal negative signal shifted from 198 to 203 nm; and a new negative signal at 222 nm , implying that the prion-like domain indeed has the subdomain which can interact with membranes to form helical conformation . Furthermore , in the presence of bicelle , many HSQC peaks underwent shifts and some even completely disappeared , as well as the peaks of three Trp side chains became well-separated ( Fig 7B ) . This observation suggests that upon interacting with bicelle , each of Trp side chains has different chemical environment . Strikingly , the disappeared peaks were identified to be from residues Met307-Gln344 , indicating that these residues became tightly associated with the large bicelle , consequently their HSQC peaks became too broad to be detected due to shortening of their T2 values [79] . We thus conducted further NMR characterization in DPC micelle which has much smaller size than bicelle , thus allowing the collection a large set of high-quality three-dimensional NMR spectra for determining the structures and dynamics [38 , 71–75 , 77–79] . We titrated the prion-like domain with DPC as monitored by CD and NMR HSQC . As judged from CD spectra in the presence of DPC at different ratios ( Fig 7C ) , gradual addition of DPC induces progressive increase of the helical conformation , with the significant transition occurring over the ratios 50–100 ( prion:DPC ) . Interestingly the CD spectra are very similar in DMPC/DHPC bicelle and DPC micelle ( Fig 7A ) , suggesting the secondary structures of the prion-like domain are highly similar in two membrane environments . Furthermore , consistent with the CD results , HSQC titrations also showed that the prion-like domain underwent conformational changes with gradual addition of DPC , as indicated by the significant shifts of some HSQC peaks ( Fig 7D and 7E ) . However , above the ratio of 1:100 , the HSQC spectra showed only minor changes ( Fig 7F ) . Similar to what is observed in bicelle , the HSQC peaks of three Trp side chains also becomes well-separated in DPC micelle . It is also interesting to note that many HSQC peaks in the presence of DPC even at a ratio of 1:400 ( prion:DPC ) are still superimposable to those in aqueous solution ( Fig 7E and 7F ) . This clearly indicates that very different from what was observed on the SOD1 and P56S-MSP mutants that a large portion of residues became embedded in the membrane environment [38 , 78] . For the TDP-43 prion-like domain , only a small portion of the residues became tightly associated with DPC micelle . We also successfully achieved sequential assignments of all non-Proline residues in the presence of DPC at a ratio of 1:200 ( prion:DPC ) , except for residues Lys263-Ser266 , Asn352 , Asn371 , Ser393-Gly394 , and Gly402 . Fig 8A presents its ( ΔCα-ΔCβ ) chemical shifts . Interestingly , in DPC micelle , residues Met311-Leu340 have ( ΔCα-ΔCβ ) chemical shifts significantly different from those in aqueous solution . More specifically , the ( ΔCα-ΔCβ ) chemical shifts of residues Met311-Asn312 and Phe316-Ile318 become more negative , implying that they become more extended . By contrast , the ( ΔCα-ΔCβ ) chemical shifts of residues Asn319-Leu340 become much more positive , strongly indicating that this region becomes highly helical . Indeed , SSP score analysis ( Fig 8B ) revealed that in DPC micelle , residues Gly309-Ile318 adopt more extended conformation while Asn319-Leu340 become well-formed helical conformation . In particular , residues Ala321-Gln331 all have SSP score larger than 0 . 9 . The results together reveal that the residues Met311-Leu340 constitute the main region for interacting with both bicelle and DPC micelle . The C-terminal residues Trp442-Gly413-Met414 have slight increase in the helical conformation ( Fig 8B ) , implying that these residues may transiently interact with DPC . We collected CD spectra of the prion-like domain in DPC micelle at a ratio of 1:200 ( prion:DPC ) in Milli-Q water ( pH 4 . 0 ) , 1 mM phosphate buffers ( pH 5 . 0 and 6 . 8 ) , which are almost superimposable ( S9A Fig ) . This result implies that the DPC-embedded conformations have no significant difference at three solution conditions . We also collected their HSQC spectra , and the results showed that at pH 5 . 0 , most HSQC peaks remain superimposable to those at pH 4 . 0 , only with some HSQC peaks disappeared which are from His-tag residues and several N-/C-terminal residues ( S9B Fig ) . At pH 6 . 8 , more peaks shifted and disappeared , but those from Gly310-Gln343 remain almost unperturbed ( S9C Fig ) , implying that the backbone NH protons of these residues are not accessible to the bulk solvent . We collected heteronuclear NOEs in DPC micelle ( Fig 8C ) . Consistent with chemical shift changes ( Fig 8A ) , residues Met311-Ala341 in the presence of DPC have significantly more positive hNOEs than in aqueous solution , with an average of 0 . 53 ( 0 . 3 in aqueous solution ) . Additionally , the C-terminal residues Ser403 , Asp406 , Gly411-Met414 also have increased hNOEs . It is also intriguing to note that in DPC micelle , some residues such as Ser292-Gly309 have more negative hNOEs , implying that they become more flexible upon interaction with DPC . Further analysis of the HSQC-NOESY spectrum indicated that many dαN ( i , i+2 ) and dNN ( i , i+2 ) , as well as dαN ( i , i+3 ) and dαN ( i , i+4 ) NOEs manifested in DPC micelle ( Fig 8D ) . In particular , a large amount of NOEs were found over Gly309-Gln344 . Surprisingly , 9 long-range NOEs were found over segment Met311-Ala325 ( Fig 9A ) . We explored the solvent accessibility of the residues of the prion-like domain in the membrane environment by titrations with a paramagnetic agent gadolinium ( III ) 5 , 8-bis ( carboxylatomethyl ) -2-[2- ( methylamino ) -2-oxoethyl]-10-oxo-2 , 5 , 8 , 11-tetraazadodecane-1-carboxylate hydrate ) , or gadodiamide , as we previously performed on the DPC-embedded SOD1 [38] and major sperm protein ( MSP ) mutants [78] . Gadodiamide with the paramagnetic gadolinium ( III ) coordinated has a large molecular volume and thus is only accessible to the protein atoms exposed to the bulk solvent . Upon titration to 10 mM , except for those of residue Met311-Gln343 and of several C-terminal residues , HSQC peaks of other residues have the intensity ratios less than the average value plus a standard deviation ( 0 . 59 ) ( Fig 8E ) , implying that the amide protons of these residues are mostly exposed to the bulk solvent while those of Met311-Gln343 and several C-terminal residues are embedded in the membrane environment , thus inaccessible to gadodiamide [38 , 78] . SSP analysis and manifestation of a large amount of NOEs indicate that the membrane-interacting subdomain transforms into a well-folded structure in the membrane environment . Therefore , 50 NMR structures of the TDP-43 Met307-Ser347 were calculated by CYANA software package [38 , 80] , with 328 NOE-derived distance and 46 TALOS-based dihedral angle [81] restraints ( S1 Table ) . Six structures with the lowest target functions were selected for further refinement with AMBER force field [38 , 82] and the calculation statistics and structure quality are summarized in S1 Table . Fig 9B presents the superimposition of six lowest-energy NMR structures , which are very similar with RMS deviations of 1 . 11 Å and 0 . 35 Å respectively for all and backbone atoms of the residues Met311-Gln344 ( S1 Table ) . Fig 9C and 9D show the electrostatic surfaces of the lowest-energy structure . Due to the location of the hydrophobic side chains on the surface of the structure , the majority of the surface is highly hydrophobic ( Fig 9C and 9D ) . While residues Met322-Gln344 adopt well-defined α-helix structure , residues Met311-Asn319 assume a well-defined but irregular loop structure ( Fig 9E and 9F ) , which can be classified into Ω-loop [83–86] . Interestingly , completely different from a cytosolic Ω-loop of ephrin-B2 determined in aqueous solution , in which the hydrophobic sidechains formed a relatively buried core [86] , in the TDP-43 Ω-loop formed upon being embedded in the membrane environment , the hydrophobic sidechains of Met311 , Phe313 , Ala315 , Phe316 , and Ile318 are pointed out to constitute a hydrophobic surface . This membrane-induce exposure of the hydrophobic side chains may play a key role in interacting with hydrophobic phase of membranes ( Fig 9C–9F ) . In fact , the TDP-43 Ω-loop is similar to that of the human prothrombin gamma-carboxyglutamic acid-rich ( GLA ) domain which is required to anchor clotting proteins onto membrane surfaces in order to increase their local concentration for effective clotting [84 , 85] . In the Ω-loop of the GLA domain , three hydrophobic side chains are also highly exposed , which have been demonstrated to be absolutely essential for penetrating into the phospholipid bilayer [84 , 85] . Remarkably , the TDP-43 sequence forming the Ω-loop appears to be extremely critical for manifesting its neurotoxicity . Previously , a cellular model of the TDP-43 aggregation using the sequence 331–369 ( lacking of the Ω-loop but containing the C-half of the helix ) repeated 12 times has been established but surprisingly its aggregation is not toxic per se but instead has a protective role [31] . On the other hand , residues Met311-Asn319 exactly forming the Ω-loop ( Fig 9F ) has been identified as the minimal region to manifest neurotoxicity [44] . Furthermore , the peptide Met307-Met322 covering the Ω-loop was identified to be capable of disrupting liposomes but amazingly the deletion of Met307-Met311 completely eliminated the membrane-damaging capacity of the peptide Asn312-Met322 [39] . To characterize the conformations of three ALS-causing mutants in membrane environments , we first titrated them in 1 mM phosphate buffer at pH 5 . 0 with DPC , and S10A Fig presents their CD spectra in the presence of DPC at a molar ratio of 1:200 ( protein:DPC ) . Interestingly , although three mutation sites are located on the membrane-embedded subdomain , they have CD spectra very similar to that of the wild type , thus implying that the conformations of three mutant prion-like domains are very similar to that of the wild type in DPC micelle . Indeed , a further examination of their HSQC spectra shows that except for the mutation residue , the residues of three mutants , namely A315E ( S10B Fig ) , Q331K ( S10C Fig ) and M337V ( S10D Fig ) , have their HSQC peaks largely superimposable to those of the wild type . Moreover , in DPC micelle , the CD and HSQC spectra of the wild type and three mutants have no significant changes even after 1 month . Subsequently , we titrated the wild type and mutants with the DMPC/DHPC bicelle which has relatively large and flat surface , thus better mimicking the bilayer membrane . Interestingly , the wild type and mutants also have very similar CD spectra 5 min after the addition of the bicelle at a ratio of 1:200 ( Fig 10A ) . Furthermore , immediately after adding DMPC/DHPC bicelle , the residues of three mutants also have their HSQC peaks largely superimposable to those of the wild type ( Fig 10B–10D ) . Strikingly , however , 1 d after embedded in DMPC/DHPC bicelle , the CD spectra of the wild type showed no significant change , whereas those of three mutants underwent very large changes ( Fig 10A ) . In particular , after 1 d , A315E had a CD spectrum typical of the amyloid oligomer . On the other hand , if compared their HSQC spectra , after 1 d , no new peak or significant peak shifts were observed . Instead , the peak intensity for three mutants reduced significantly . After 1 wk , visible aggregates were observed for three mutants but not the wild type . This implies that the changes of the CD spectra observed on three mutants ( Fig 10A ) are largely resulting from the self-association of three mutant proteins in the bicelle . As well-established [71–75 , 79] , the DPC micelle has very high curvature as well as small volume which may stably accommodate only one molecule with the size of the TDP-43 prion-like domain . As such , the self-association/aggregation is anticipated to be largely inhibited in DPC micelle . By contrast , the DMPC/DHPC bicelle has the relatively large and flat membrane surface , and thus can host more than one protein molecules in one bicelle . As a consequence , the local concentration of the proteins will be significantly increased , and their self-association might be significantly enhanced . This explains why the self-association is very minor even for the three ALS-causing mutants in the DPC micelle , but become much faster in DMPC/DMHC bicelle . Indeed , nature has used this strategy to increase the local protein concentration for effective clotting by anchoring the gamma-carboxyglutamic acid-rich ( GLA ) domain of human prothrombin onto membrane surfaces [84 , 85] .
Our views about ALS pathogenesis are undergoing a paradigm shift triggered by the discovery that the pathogenic inclusions of TDP-43 presented in ~97% ALS and ~45% FTD patients regardless of being familial or sporadic [5] . Furthermore , TDP-43 inclusions have also been observed in many other neurodegenerative diseases and recently it was clinically revealed that only the Alzheimer patients with TDP-43 inclusions have significant cognitive impairment [15] . On the other hand , dynamic assembly into functional oligomers mainly mediated by the prion-like domain has been extensively demonstrated to be essential for the physiological functions of TDP-43 and consequently it was recently proposed that ALS pathogenesis may be initiated by a transition from the reversible assembly to irreversible aggregation under pathological conditions . Indeed , the wild-type TDP-43 itself is intrinsically aggregation-prone as well as toxic but the ALS-causing mutations appear to significantly exaggerate it . In the present study , as facilitated by our previous discovery [33 , 38 , 46 , 47 , 62 , 63 , 68] , for the first time , to the best of our knowledge , we have successfully determined the conformations and dynamics of the full-length TDP-43 prion-like domain at atomic-resolution by NMR spectroscopy . In aqueous solution , the TDP-43 prion-like domain is intrinsically disordered , which only contains some nascent secondary structures with highly-unrestricted backbone motions on ps-ns time scale . Unexpectedly , despite being mostly monomeric and disordered , the TDP-43 prion-like domain in water at pH 4 . 0 contains a large number of inter-molecular hydrogen bonds between side chain and backbone atoms , as particularly evidenced by the manifestation of the intrinsic visible fluorescence and low temperature coefficients of backbone amides . Furthermore , at neutral pH , the monomeric prion-like domain starts to assemble into the oligomer with a transition from the disordered to β-sheet rich conformations as reported by CD and three fluorescence probes , which has amyloid-like fibrillar structures as imaged by EM ( Fig 4A ) . Mechanistically , the assembly appears to be initiated by the liberation of the QNS side chains from intra-molecular hydrogen bonding to form inter-molecular “hydrogen bonds/steric zippers” as previously proposed [22 , 67 , 70] . Nevertheless , at high protein concentrations , even the wild type would become precipitated rapidly with the formation of white aggregates of amorphous structures ( Fig 4E ) . By a sharp contrast , despite having average conformations highly similar to that of the wild type , all three ALS-causing point mutants namely A315E , Q331K and M337V gain the ability to transform into well-formed amyloid oligomers as particularly indicated by CD and intrinsic visible fluorescence spectra , which are very different from those of the wild type . Three mutants also form amyloid fibrils ( Fig 4B–4D ) , with the morphology similar to that by the wild type ( Fig 4A ) , as shown by EM , although their secondary structures might have some difference from those of the wild type . Previously , it has been revealed that the prion-like domains alone of FUS and TDP-43 are sufficient to bind nucleic acids to initiate the functional assembly [30 , 49] . Here we confirm this discovery and further show that the interaction with ssDNA facilitates the assembly of the wild type into the hydrogel . By a dramatic contrast , at the same ratios ( protein:ssDNA ) , the interactions with ssDNA trigger immediate and irreversible precipitation with white aggregates for three mutants . The changes observed here in the self-assembly and interaction with nucleic acids may significantly reduce the functional capacity of the ALS-causing mutants , thus contributing to “loss of the functions” in ALS pathogenesis . So how could we rationalize such radical effects of the ALS-causing point mutations on the TDP-43 prion-like domain , which is in fact highly disordered ? Previously it has been well established that the intrinsically disordered proteins ( IDPs ) , such as the TDP-43 prion-like domain , are in a dynamic equilibrium between different sets of conformations . So the monomeric state is characteristic of a relatively flat but rugged energy landscape with numerous local energy minima separated by low energetic barriers [22–24 , 30 , 49 , 50 , 87–89] . As a consequence , IDPs are , in fact , predicted to have very high specificity in binding as well as self-assembly . Indeed , it has been previously revealed that yeast uses the conformational diversity of its prion protein Sup35 to dictate its seeding specificity [22–25] . Therefore , the prion-like domains , such as in yeast Sup35 and human TDP-43 , appear to represent a subgroup of the intrinsically disordered proteins that utilizes extremely high specificity in the assembly to achieve their functions . The wild-type TDP-43 prion-like domain appears to have an energy landscape to allow the self-assembly of reversible and functional oligomers only under limited conditions ( Fig 11A and 11B ) . For instance , if upon pathological overexpression , the TDP-43 concentration is too high , the prion-like domain may directly jump to form irreversible aggregates with amorphous structures ( Fig 4E ) , which were most frequently identified in the neuronal inclusions . Alternatively , if the dissociation of the functional oligomers is inhibited due to a long stress time or by other pathological conditions , they may lose the reversibility , thus transforming into irreversible aggregates or amyloid fibrils ( Fig 4A ) , which was also detected in patients’ brain tissues [57] . As the assembly is extremely specific , an ALS-causing point mutation even like M337V which only has very minor change of the side chain is sufficient to remodel the energy landscape , at least partly by perturbing the hydrogen network , to more favor the formation of irreversible aggregates or/and amyloid oligomers ( Fig 11C and 11D ) . In the future , more ALS-causing mutants should be studied to better understand the molecular mechanism by which the mutations globally remodel the energy landscape . On the other hand , only aggregation of TDP-43 alone appears insufficient to cause “gain of neurotoxicity” because it was recently shown that the in vivo aggregation by the sequence 331–369 repeated 12 times in fact prevented TDP-43 neurotoxicity [31] . Here , our identification of a membrane-interacting subdomain over residues Met311-Gln343 may reconcile the paradox that on one hand , the TDP-43 prion-like domain is highly neurotoxic . On the other hand , its aggregation could be protective . Previously , all aggregation-prone proteins causing human diseases have been demonstrated to contain membrane-interacting domains/fragments . Therefore , here we propose that the coupled capacity of the membrane-interaction and aggregation might critically account for the high neurotoxicity of TDP-43 . Indeed , previous studies have extensively demonstrated that the membrane-interacting regions , particularly that forming the Ω-loop , are absolutely indispensable for the neurotoxicity of the TDP-43 fragments [39 , 43 , 44] , and its absence in fact converted the aggregation to become protective [31 , 45] . Here we also found that upon being associated onto the large surface of the DMPC/DHPC bicelle , the self-association of three ALS-causing mutants has been significantly sped up as compared to that in solution . Therefore , the membrane disruption by TDP-43 inclusion may also follow similar mechanisms previously established for other disease-causing proteins such as amyloid-β and IAPP ( 20–29 ) peptides . Briefly , under pathological conditions , the wild-type TDP-43 or its mutant molecules associated onto the membranes may further form large aggregates/inclusions to trigger nonspecific fragmentation of the lipid membrane [71–75] . In the future , it is also of interest to explore whether the membrane-interaction is essential for the physiological function of TDP-43 . In fact , it has been realized that for the prion-like domains of FUS and TDP-43 , the protein concentrations required for the specific assembly in vitro are much higher than the physiological concentrations in cells . So , it has been proposed that the additional binding with nucleic acids by RNA binding motifs may act to increase the local concentration to allow the specific assembly [30 , 48 , 49] . Here we speculate that the dynamic association with membranes might offer an alternative mechanism to effectively increase the local concentration as well as properly align the individual prion-like domain for the specific assembly , as found with the human GLA domain [84 , 85] . Nevertheless , if this physiological membrane-association is exaggerated by pathological factors , TDP-43 might become significantly aggregated in membranes , leading to the loss of its physiological functions as well as gain of neurotoxicity . As such , to decouple the aggregation and membrane interaction may represent a promising therapeutic strategy to treat neurodegenerative diseases .
The DNA encoding the full-length prion-like domain over residues Lys363-Met414 was amplified by PCR reactions from the full-length TDP-43 gene and subsequently cloned into a modified vector pET28a with six His residues at C-terminus . Three ALS-causing mutations , A315E , Q331K , and M337V , were introduced into the TDP-43 prion-like domain by use of the QuikChange Site-Directed Mutagenesis Kit ( Stratagene , La Jolla , CA , United States ) as previously described [62 , 68] . The expression vectors were subsequently transformed into and overexpressed in Escherichia coli BL21 ( DE3 ) cells ( Novagen ) . The recombinant wild-type and mutant proteins were all found in inclusion body . As a result , the pellets were first dissolved in a phosphate buffer ( pH 8 . 5 ) containing 8 M urea and subsequently purified by a Ni2+-affinity column ( Novagen ) under denaturing conditions in the presence of 8 M urea . The fractions containing the recombinant proteins were acidified by adding 10% acetic acid and subsequently purified by reverse-phase ( RP ) HPLC on a C4 column eluted by water-acetonitrile solvent system . The HPLC elutions containing pure recombinant proteins were lyophilized . The generation of the isotope-labeled proteins for NMR studies followed a similar procedure except that the bacteria were grown in M9 medium with the addition of ( 15NH4 ) 2SO4 for 15N labeling and ( 15NH4 ) 2SO4/[13C]-glucose for double labelling [28 , 33 , 48 , 62 , 68] . The purity of the recombinant proteins was checked by SDS–PAGE gels and their molecular weights were verified by a Voyager STR matrix-assisted laser desorption ionization time-of-flight-mass spectrometer ( Applied Biosystems ) . The concentration of protein samples was determined by the UV spectroscopic method in the presence of 8 M urea . Briefly , under the denaturing condition , the extinct coefficient at 280 nm of a protein can be calculated by adding up the contribution of Trp , Tyr , and Cys residues [90] . All CD experiments were performed on a Jasco J-810 spectropolarimeter equipped with a thermal controller using 1-mm path length cuvettes . Data from five independent scans were added and averaged . The wild-type and mutant TDP-43 prion-like domain samples were prepared at a protein concentration of 20 μM either in Milli-Q water ( pH 4 . 0 ) , or 1 mM phosphate at pH 5 . 0 , 6 . 0 and 6 . 8 respectively . For characterizing the interactions of ( TG ) 6 ssDNA we previously used for the TDP-43 N-domain [33] with the wild-type and three mutant prion-like domains , the protein samples at 20 μM were prepared in 1 mM phosphate buffer at pH 5 . 0 while the ssDNA was dissolved in the same buffer . The CD spectra of the proteins at different ratios of protein:ssDNA were obtained by calibrating the dilution factor as well as subtracting the CD contribution of ssDNA at the corresponding ratios . CD spectra were further analyzed to estimate secondary structure contents as we previously conducted [33 , 63] by CDPro software package ( lamar . colostate . edu/∼sreeram/CDPro/main . html ) . In the present study , two membrane-mimetic systems , namely the DPC micelle and DMPC/DHPC bicelle , were used to identify the membrane-interacting regions of the TDP-43 prion-like domains . In aqueous solution , dodecylphosphocholine ( DPC ) self-assembles into the micelle structure typically containing ~50–100 molecules . The large bicelle to better mimic bilayer membrane was prepared by mixing up dimyristoylphosphatidylcholine ( DMPC ) and dihexanoylphophatidylcholine ( DHPC ) at a q value of 4 as previously described [79] . At this ratio , the disk-shaped bicelle with a diameter of ~460 Å is formed in which DMPC constitutes a bilayered section surrounded by a rim of DHPC [79 , 91] . All NMR experiments were acquired on an 800 MHz Bruker Avance spectrometer equipped with pulse field gradient units as described previously [33 , 78] . For characterizing the conformations in aqueous solutions , a pair of triple-resonance experiments HNCACB , CBCA ( CO ) NH were collected for the sequential assignment on a 15N-/13C-double labelled sample of 500 μM , while 15N-edited HSQC-TOCSY and HSQC-NOESY were collected on a 15N-labelled sample at a protein concentration of 500 μM . For achieving assignments in DPC micelle , triple-resonance experiments HNCACB , CBCA ( CO ) NH and HCCH-TOCSY were acquired on 15N-/13C-double labelled samples at a protein concentration of 500 μM in the DPC micelle ( H-DPC ) at 100 mM . For obtaining NOE connectivities , 15N-edited HSQC-TOCSY and HSQC-NOESY were collected on a 15N-labelled sample at a protein concentration of 500 μM in DPC micelles at 100 mM . NMR data were processed with NMRPipe [92] and analyzed with NMRView [93] . For assessing the backbone dynamics on the ps-ns time scale , {1H}-15N steady-state NOEs were obtained by recording spectra on the 15N-labeled sample at 500 μM in either aqueous solution or DPC micelle ( 100 mM ) , with and without 1H presaturation with duration of 3 s plus a relaxation delay of 6 s at 800 MHz . To assess conformational exchanges over μs-ms , 15N transverse relaxation dispersion experiments were acquired on the 15N-labeled sample at 500 μM in either aqueous solution or DPC micelle ( 100 mM ) , on a Bruker Avance 800 spectrometer with a constant time delay ( TCP = 50 ms ) and a series of CPMG frequencies , ranging from 40 Hz , 80 Hz , 120 Hz ( x3 ) , 160 Hz , 200 Hz , 240 Hz , 320 Hz , 400 Hz , 480 Hz , 560 Hz , 640 Hz , 720 Hz , 800 Hz , and 960 Hz ( ×3 indicates repetition ) as we previously performed [68 , 78] . To probe the accessibility of the prion-like domain residues in DPC micelle [78] , HSQC spectra at 100 μM in the presence of 20 mM DPC were acquired by gradual addition to 10 mM of gadodiamide ( gadolinium ( III ) 5 , 8-bis ( carboxylatomethyl ) -2-[2- ( methylamino ) -2-oxoethyl]-10-oxo-2 , 5 , 8 , 11-tetraazadodecane-1-carboxylate hydrate ) . All fluorescence spectra were measured at 25°C with a RF-5301 PC spectrophotometer ( Shimadzu , Japan ) as previously established [51–56] , at different time points of the incubations of the wild-type and three mutants at a protein concentration of 40 μM in 1 mM phosphate buffer ( pH 6 . 8 ) . The rectangular fluorescence quartz cuvette has the pathlength dimension of 10 x 10 mm and the general settings are: PMT at low sensitivity and scan speed of medium speed ( 200 nm/min ) . For the intrinsic UV fluorescence , the emission spectra were measured with the excitation wavelength at 280 nm and slit widths: excitation at 5 nm and emission at 10 nm . For the intrinsic visible fluorescence , the emission spectra were measured with the excitation wavelength at 375 nm and slit widths: excitation at 20 nm and emission at 10 nm . For Thioflavin-T ( ThT ) binding assay , a 2 mM ThT stock solution was prepared by dissolving ThT in milli-Q water and filtered through a 0 . 22 μm Millipore filter . The fresh working solution was prepared by diluting the stock solution into 1 mM phospate buffer ( pH 6 . 8 ) to reach a final ThT concentration of 50 μM . A 10 μL aliquot of each incubation solution , or 10 μL aliquot of the incubation buffer ( 1 mM phosphate at pH 6 . 8 ) as the control , was mixed with 130 μL of the ThT working solution in the dark for 10 min . The fluorescence emission spectra were acquired for three repeats with the excitation wavelength at 442 nm and slit widths: excitation at 5 nm and emission at 10 nm . Incubation samples of the wild type and three mutants at 40 μM were imaged at one and two weeks of the incubation in 1 mM phosphate buffer ( pH 6 . 8 ) , by a TEM microscope ( Jeol Jem 2010f Hrtem , Japan ) operating at an accelerating voltage of 200 kV . The aggregates of the wild type were prepared immediately before EM imaging , by diluting the stock protein sample in Milli-Q water ( pH 4 . 0 ) into 1 mM phosphate buffer to reach a concentration of 200 μM ( pH 6 . 8 ) . For EM imaging , a 5 μl aliquot of the incubation or aggregate solutions was placed onto the Cu grids ( coated with carbon film; 150 mesh; 3 mm in diameter ) and negatively stained with 5 μl of 2% neutral , phosphotungstic acid ( PTA ) . This aliquot was allowed to settle on Cu grid for 30 s before the excess fluid was drained away . The Cu grid was later air-dried for another 15 mins before being imaged . Structure calculation was conducted on residues Met307-Ser347 with a large amount of NOEs in DPC micelle . Backbone dihedral angles were generated with TALOS+ by inputting backbone 1H , 15N and 13C chemical shifts . NOE-based distance constraints were extracted only from 15N-edited NOESY spectrum as the side-chain 13C NMR resonances are too broad to assign NOE connectivities as we previously observed [38 , 78] . The NMR structures in the DPC were calculated with the input of distance and dihedral angle constraints by CYANA . Six lowest target-function CYANA structures with no NOE violation >0 . 4 Å and no dihedral angle violation >4 degrees were selected for further refinement by use of GROMACS version 4 . 5 . 3 in Amber99sb-ildn force field , with dihedral angle and distance restraints converted and incorporated into the topology file . The structure coordinate of the TDP-43 membrane-interacting subdomain over residues Met307-Ser347 in DPC micelle has been deposited in PDB with ID of 2N2C and the associated NMR data were also deposited in BMRB with ID of 25595 . | Amyotrophic lateral sclerosis ( ALS ) is the most prevalent fatal motor neuron disease . It was identified ~140 years ago , but the exact mechanism underlying the disease has still not been well defined . TAR-DNA-binding protein-43 ( TDP-43 ) was identified as the major component of the proteinaceous inclusions present in ~97% ALS and ~45% frontotemporal dementia ( FTD ) patients , and has also been observed in an increasing spectrum of other neurodegenerative disorders , including Alzheimer disease . The TDP-43 C-terminus is a key domain—it encodes a prion-like domain and , crucially , hosts almost all ALS-causing mutations . Here we have successfully determined the conformations , dynamics , and self-associations of the prion-like domains of both wild type and three ALS-causing mutants in both aqueous solutions and membrane environments . The study suggests that the TDP-43 prion-like domain appears to have a unique energy landscape , which allows the assembly of the wild-type sequence into specific oligomers only under very limited conditions . Intriguingly , ALS-causing point mutations remodel the energy landscape to favor amyloid formation or irreversible aggregation , thus supporting the emerging view that pathologic aggregation may occur via the exaggeration of functionally important assemblies . Furthermore , the coupled capacity of TDP-43 in aggregation and membrane interaction may partly account for its high neurotoxicity; decoupling these may therefore represent a promising therapeutic strategy to treat TDP-43-mediated neurodegenerative diseases . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2016 | ALS-Causing Mutations Significantly Perturb the Self-Assembly and Interaction with Nucleic Acid of the Intrinsically Disordered Prion-Like Domain of TDP-43 |
There is increasing evidence that genetic risk variants for non-syndromic cleft lip/palate ( nsCL/P ) are also associated with normal-range variation in facial morphology . However , previous analyses are mostly limited to candidate SNPs and findings have not been consistently replicated . Here , we used polygenic risk scores ( PRS ) to test for genetic overlap between nsCL/P and seven biologically relevant facial phenotypes . Where evidence was found of genetic overlap , we used bidirectional Mendelian randomization ( MR ) to test the hypothesis that genetic liability to nsCL/P is causally related to implicated facial phenotypes . Across 5 , 804 individuals of European ancestry from two studies , we found strong evidence , using PRS , of genetic overlap between nsCL/P and philtrum width; a 1 S . D . increase in nsCL/P PRS was associated with a 0 . 10 mm decrease in philtrum width ( 95% C . I . 0 . 054 , 0 . 146; P = 2x10-5 ) . Follow-up MR analyses supported a causal relationship; genetic variants for nsCL/P homogeneously cause decreased philtrum width . In addition to the primary analysis , we also identified two novel risk loci for philtrum width at 5q22 . 2 and 7p15 . 2 in our Genome-wide Association Study ( GWAS ) of 6 , 136 individuals . Our results support a liability threshold model of inheritance for nsCL/P , related to abnormalities in development of the philtrum .
Orofacial clefts are malformations characterised by a failure of fusion between adjacent facial structures in the embryo [1] . Cleft lip with/without cleft palate ( CL/P ) is a sub-type of orofacial cleft , consisting of individuals presenting with a cleft of the upper lip , with or without a cleft of the palate . Approximately 70% of CL/P cases are non-syndromic , where the facial cleft is not accompanied by other apparent developmental or physical abnormalities [2] . The non-syndromic form of CL/P ( nsCL/P ) is a multifactorial trait with both genetic and environmental risk factors [1] . A possible polygenic threshold model of inheritance is supported by the identification of more than 20 common genetic risk variants for nsCL/P from genome-wide association studies ( GWAS ) [3–9] and single nucleotide polymorphism ( SNP ) heritability estimates of around 30% [6] . Facial morphology in the general population is also likely to be highly polygenic; genome-wide significant loci have been found for multiple facial phenotypes across diverse ethnic populations [10–14] . In some cases , the genes associated with normal-range variation in facial shape have also been implicated in nsCL/P ( e . g . MAFB ) [12] . Likewise , previous studies using candidate SNPs have found overlap between nsCL/P risk loci and facial phenotypes in the general population [11 , 15 , 16] . For example , the strongest nsCL/P GWAS signal , intergenic variant rs987525 on chromosome 8q24 , was found to be associated with more than half of the 48 facial phenotypes studied in a European population [11] while in a Han Chinese population , rs642961 in IRF6 ( a major nsCL/P-associated gene ) strongly predicted lip-shape variation in females [16] . However , associations between nsCL/P genetic variants and facial morphology were not consistently replicated , possibly because of methodological differences in measuring facial phenotypes or population differences between cohorts [10] . The use of individual markers to demonstrate genetic overlap between two phenotypes has notable limitations; a large number of statistical tests are introduced , and interpretation is difficult when some SNPs show an association and others do not . Polygenic risk scores ( PRS ) involve incorporating multiple markers , including those not reaching genome-wide significance , into a genetic score that serves as a proxy for a trait [17] . PRS have been previously shown to be suitable predictors for nsCL/P [6] suggesting they can be used to estimate genetic overlap between nsCL/P and normal-range facial morphology . Interpreting genetic overlap between nsCL/P and a facial phenotype is difficult because the development of the face and development of an orofacial cleft are largely synchronous . One possibility is that differences in the facial phenotype are a sub-phenotypic manifestation of genetic liability to nsCL/P ( see Fig 1 ) . The inheritance of dichotomous traits can be modelled on the liability scale , where every individual has an underlying normally distributed liability to the trait determined by genes , environment and chance . Individuals above a liability threshold develop the trait , while increased liability may cause related phenotypic differences in individuals without the trait [18–20] . For example , increased liability to developing a cleft palate ( CP ) has been hypothesised to be associated with delayed movement of the palatal shelf , which may in turn result in a CP , dependant on other factors such as shelf and head width [20] . In order to evaluate the coherence of the liability-related sub-phenotype model , we apply the principles of Mendelian randomization ( MR ) . MR is an instrumental variable approach , testing causality of an “exposure” and an outcome by using genetic instruments to mimic a randomised controlled trial [21] . MR relies on several strict assumptions; firstly , genetic variants must be robustly associated with the exposure ( in this instance , genetic liability to nsCL/P ) ; secondly , the variants cannot influence the outcome through a pathway independent of the exposure; and thirdly , the variants should not be associated with confounders of the relationship between the exposure and the outcome [22] . If these assumptions are met , bidirectional MR can be used to test the hypothesis that genetic liability to nsCL/P is causally related to facial morphology [22] . In the absence of large-scale publicly available GWAS summary data for nsCL/P , we used individual level genotype data from the International Cleft Consortium to Identify Genes and Interactions Controlling Oral Clefts ( ICC ) and GWAS summary statistics from the Bonn-II study [8] to replicate the meta-analysis GWAS summary statistics from the previously published Ludwig et al 2012 GWAS [3] . Next , we investigated genetic overlap between nsCL/P and normal-range facial morphology in the general population , using PRS derived from the GWAS summary statistics . Finally , in the instance of genetic overlap , we used bidirectional MR to explore the relationship between nsCL/P and implicated facial phenotypes . A flowchart detailing the primary analyses is contained in Fig 2 .
We performed a GWAS of nsCL/P using the TDT on 638 parent-offspring trios and 178 offspring duos of European descent , and then meta-analysed our results with GWAS summary results previously published on 399 cases and 1 , 318 controls in the Bonn-II study [8] . This yielded comparable results to a previously published GWAS [3] , which used a very similar data-set with slightly different quality control and analysis methods ( S1 Table ) . We also evaluated the predictive accuracy of nsCL/P that could be achieved using different PRS constructed from these summary data by comparing the strength of association at different inclusion thresholds of the PTDT . We determined that including independent SNPs that surpass a P-value threshold of 10−5 was the most predictive of nsCL/P liability in both European and Asian trios ( S2 Table ) . Therefore , this threshold was used for generating polygenic risk scores from the meta-analysis summary statistics . SNPs included in the selected score are listed in S3 Table . Prior to testing the performance of our nsCL/P PRS on predicting facial morphology , we calculated the minimum genetic correlation required to detect an association between the PRS and the facial phenotypes . We found that the minimum genetic correlation required ranged from 0 . 17 to 0 . 28 with differences attributable to different heritability estimates across the facial phenotypes ( S4 Table ) . We evaluated the performance of our nsCL/P PRS for prediction of seven facial morphological traits . Facial distances used in the analysis are shown in Fig 3 . We found evidence of an association between the nsCL/P PRS and philtrum width in the ALSPAC children , where a 1 S . D . increase in nsCL/P PRS was associated with a 0 . 07 mm decrease in philtrum width ( 95% C . I . 0 . 02 , 0 . 13; P = 0 . 014 ) ( Table 1 ) . We attempted to replicate this finding in the 3DFN study and found a consistent effect of 1 S . D . increase in nsCL/P PRS being associated with a 0 . 14 mm decrease in philtrum width ( 95% C . I . 0 . 07 , 0 . 21; P = 1 . 7x10-4 ) . Meta-analysing these results; indicated that a 1 S . D . increase in nsCL/P PRS is associated with a 0 . 10 mm decrease in philtrum width ( 95% C . I . 0 . 054 , 0 . 146; P = 2x10-5 ) . To generate SNP-philtrum width association information for MR analyses , we performed GWAS of philtrum width in both ALSPAC and 3DFN separately , before meta-analysing . The combined sample included 6 , 136 individuals of recent European descent . We identified two novel chromosomal regions associated with philtrum width with genome-wide significance at 5q22 . 2 ( lowest P value for rs255877 , P = 3 . 8x10-10 ) , within the non-coding RNA intronic region of an uncategorised gene ENSG00000232633 , and 7p15 . 2 ( rs2522825 , P = 1 . 4x10-8 ) , an intergenic SNP near HOXA1 ( S5 Table ) . We found some evidence that the two lead SNPs may be eQTLs for nearby genes ( S6 Table ) . The two lead SNPs of the genome-wide significant loci , rs255877 and rs2522825 , were used as genetic variants associated with philtrum width in subsequent MR analyses . The GWAS summary statistics are available at the University of Bristol data repository , data . bris , at https://doi . org/10 . 5523/bris . 1kz9y0moa8sgj2lxlk53mdmlbj [23] . We used MR to investigate the possible causal mechanism that would give rise to the genetic overlap between nsCL/P and philtrum width . Firstly , we determined whether genetic variants contributing to liability of nsCL/P cause changes in philtrum width , by testing SNPs strongly associated with nsCL/P for association with philtrum width . A 1-unit log odd increase in liability to nsCL/P was associated with a 0 . 11mm ( 95% C . I . 0 . 04 , 0 . 19; P = 0 . 0036 ) decrease in philtrum width . Sensitivity analyses suggested there was no strong evidence for pleiotropy or heterogeneity and validated the consistency of the instrument . Leave-one-SNP-out analysis showed consistent effect estimates after exclusion of each SNP ( Table 2 ) . Secondly , we determined whether genetic variants associated with philtrum width also affect liability to nsCL/P , by testing two independent SNPs associated with philtrum width at genome-wide significance ( derived in the ALSPAC and 3DFN cohorts ) for association with nsCL/P . Utilising strong LD proxies ( S7 Table ) , weak evidence was found of an association between philtrum width-associated variants and liability to nsCL/P ( LogOR = 0 . 30; 95% C . I . -0 . 26 , 0 . 86; P = 0 . 30 ) . Sensitivity analyses for pleiotropy were limited , with only 2 SNPs . Thirdly , we used the MR-Steiger test of directionality to test the direction of effect between philtrum width and liability to nsCL/P . The results suggested that the true direction of effect is that genetic variants contributing to liability to nsCL/P cause changes in philtrum width ( P <10−10 ) . The rationale for interpretation of the bidirectional MR analysis is contained in Fig 4 . Strong evidence was found for genetic liability to nsCL/P causing decreased philtrum width , weak evidence was found for heterogeneity or assumption violations in the forward-MR , and weak evidence was found for the reverse-MR of philtrum width-associated variants on liability to nsCL/P . Therefore , we conclude that the most likely explanation for the genetic overlap between nsCL/P and philtrum width is that genetic liability to nsCL/P is causally related to decreased philtrum width .
In this manuscript , we have shown that there is genetic overlap between nsCL/P and normal-range variation in philtrum width , and furthermore , that genetic risk SNPs for nsCL/P consistently cause decreased philtrum width in the general population . Notably there was weak evidence for genetic overlap between nsCL/P and upper lip width despite the observational correlation between the widths of the upper lip and philtrum . There are two main implications of these results . First , our findings demonstrate the aetiological relevance of the formation of the philtrum to nsCL/P . The medial nasal and maxillary processes are responsible for development of the upper lip and philtrum [24] . Developmental anomalies within these processes may result in a cleft lip [25] and our findings show that even when there is successful fusion , as in our study populations , the genetic variants which give rise to a CL/P cause decreased philtrum width . Secondly , the non-heterogeneous additive effect of common nsCL/P risk variants , on a related phenotype in the general population , supports a polygenic threshold model of inheritance for nsCL/P . Although previous studies have looked at nsCL/P related sub-phenotypes , this study uses causal inference methods to more formally investigate the relationship . Our identification of phenotypic differences related to nsCL/P liability are consistent with previous studies [26–31] observing sub-clinical facial phenotypes in individuals with nsCL/P and their unaffected family members , particularly a previous study which observed reduced philtrum width in unaffected parents of individuals with nsCL/P [31] . A polygenic threshold model of inheritance related to development of the philtrum is consistent with a previously proposed mechanism for the inheritance of cleft palate [20] , the identification of numerous common nsCL/P genetic risk variants [3–7] and estimation of a substantial SNP heritability for nsCL/P [6] . We do not replicate associations between nsCL/P and other facial morphological dimensions found in previous studies [11 , 15 , 31] using candidate SNPs but note that polygenic risk score methods are methodologically distinct and are used to investigate a different research question to single SNP analyses . We extend the investigation of the association between nsCL/P and facial morphology in two important ways . We demonstrate that the association is present not only in unaffected family members but also in the general population , and use MR to demonstrate that this relationship is present on the liability scale . Conventionally MR is used to test possible causal effects of a modifiable continuous exposure such as cholesterol or alcohol on disease outcomes [32 , 33] . Here we exploit the principles of MR to test the threshold hypothesis , by inferring a causal relationship between genetic variants contributing to liability of nsCL/P and philtrum width in a non-clinical population . We interpret this causal relationship as evidence that smaller philtrum width is a sub-phenotypic manifestation attributable to the same genetic variants that cause nsCL/P . In addition to investigating the relationship between facial morphology and nsCL/P , we also performed the first GWAS of philtrum width , and identified two novel genome-wide significant loci . Notably one of the loci , rs2522825 at 7p15 . 2 , was associated with gene expression at several nearby genes in the homeobox gene family , which are known to play important roles in embryonic development [34 , 35] . The causal inference made in this study was achieved through the use of two independent cohorts as discovery and replication samples which greatly reduces the risk of false positives and demonstrates that results can be generalised to different populations . Detailed facial phenotyping data on a large number of individuals in our cohorts along with other detailed phenotype and genotype data enabled us to identify philtrum width as being the most relevant facial morphological feature from amongst seven biologically likely candidates . Statistical power does limit the detection of other features that may have mechanistic relationships with smaller effect sizes ( S4 Table ) . In this study , we combined CL/P and cleft lip only ( CLO ) , however there is evidence suggesting that there are distinct aetiological differences between these traits , [5 , 36 , 37] which could reduce our statistical power , and complicates interpretation . For example , the philtrum may be more related to CLO , but we did not have sufficient data to compare nsCL/P subtype differences . An additional limitation is that there are few well-characterised genetic risk loci for philtrum width , so our MR analysis testing if genetic variants associated with a narrow philtrum width also affect liability of nsCL/P , may be underpowered . We conclude that genetic liability to nsCL/P is causally related to variation in philtrum width and that this finding supports a polygenic threshold model of inheritance for nsCL/P , related to abnormalities in development of the philtrum . Further research looking at the relationship between genetic liability for nsCL/P and severity of cleft would provide further evidence for the polygenic threshold model .
A bidirectional two-sample Mendelian randomization analysis was performed using the TwoSampleMR R package [54] , testing both the forward direction ( the effect of genetic risk variants for nsCL/P on implicated facial measurements ) and the reverse direction ( the effect of genetic risk variants for implicated facial measurements on liability to nsCL/P ) . The Inverse-Variance Weighted method was used as the primary analysis . Several sensitivity analyses were performed to test the assumptions of MR; the heterogeneity test was used to measure balanced pleiotropy , MR-Egger [55] was used to test for directional pleiotropy , the weighted median method [56] was used to test if the result is consistent assuming that at least half of the variants are valid and the weighted mode method [57] was used to test if the result is consistent assuming that the most common effect is valid . The Steiger test [58] was used to determine the likely direction of effect . | Non-syndromic cleft lip/palate ( nsCL/P ) is a birth defect , primarily affecting the upper lip and hard palate . Individuals with nsCL/P and their unaffected family members sometimes present with other minor craniofacial anomalies and may have differences in facial morphology compared to the general population . Here , we investigate the shared genetic relationship between nsCL/P and facial morphology in a sample of around 6 , 000 Europeans from two different studies . We demonstrate that genetic risk factors for nsCL/P are associated with decreased philtrum width in unaffected individuals from the general population and that the relationship is likely causal . This finding is important because it demonstrates that nsCL/P , which is often treated as a dichotomous trait , may have a continuous dimension and offers insight into potential biological mechanisms that result in nsCL/P . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"genome-wide",
"association",
"studies",
"medicine",
"and",
"health",
"sciences",
"face",
"computational",
"biology",
"developmental",
"biology",
"mathematics",
"statistics",
"(mathematics)",
"genome",
"analysis",
"cleft",
"palate",
"molecular",
"genetics",
"morphogenesis",... | 2018 | Investigating the shared genetics of non-syndromic cleft lip/palate and facial morphology |
The evolutionary divergence of mitochondrial ribosomes from their bacterial and cytoplasmic ancestors has resulted in reduced RNA content and the acquisition of mitochondria-specific proteins . The mitochondrial ribosomal protein of the small subunit 34 ( MRPS34 ) is a mitochondria-specific ribosomal protein found only in chordates , whose function we investigated in mice carrying a homozygous mutation in the nuclear gene encoding this protein . The Mrps34 mutation causes a significant decrease of this protein , which we show is required for the stability of the 12S rRNA , the small ribosomal subunit and actively translating ribosomes . The synthesis of all 13 mitochondrially-encoded polypeptides is compromised in the mutant mice , resulting in reduced levels of mitochondrial proteins and complexes , which leads to decreased oxygen consumption and respiratory complex activity . The Mrps34 mutation causes tissue-specific molecular changes that result in heterogeneous pathology involving alterations in fractional shortening of the heart and pronounced liver dysfunction that is exacerbated with age . The defects in mitochondrial protein synthesis in the mutant mice are caused by destabilization of the small ribosomal subunit that affects the stability of the mitochondrial ribosome with age .
Mitochondria are composed of proteins encoded by the nuclear and mitochondrial genomes . Most of the mitochondrial proteins including the ribosomal proteins and translation factors that are responsible for the expression of the mitochondrial genome are synthesized on cytoplasmic ribosomes and imported into mitochondria post-translationally . In chordates , the mitochondrial genome encodes 22 tRNAs , 2 rRNAs and 11 mRNAs that are translated on mitochondrial ribosomes ( mitoribosomes ) into 13 polypeptides , all members of the oxidative phosphorylation complexes [1] . Mutations in mitochondrial genes or nuclear genes coding for mitochondrial proteins result in mitochondrial dysfunction and impaired energy production that cause mitochondrial diseases ( reviewed in [2] ) . The most common cause of mitochondrial diseases are defects in the translational machinery ( reviewed in [3] ) , however the mechanisms of mitochondrial protein synthesis are not well understood . Mammalian mitoribosomes are 55S particles consisting of a 28S small subunit that includes the 12S rRNA and ~ 29 proteins and a 39S large subunit , which contains the 16S rRNA and ~ 48 proteins [4–7] . Mitoribosomes are distinct from their bacterial and cytoplasmic counterparts; they have reduced RNA content and an increased number of proteins [8] . The mitoribosome consists of proteins that share homology with bacterial ribosomes , some of which can have mitochondria-specific extensions , and additional , mitochondria-specific proteins that decorate the ribosomal surface , the mRNA entrance site and form some of the bridges linking the small and large subunits [6 , 7 , 9 , 10] . The increased ribosomal protein content does not entirely compensate for the loss of rRNA as many of the mitochondria-specific ribosomal proteins do not replace the missing RNA helices but instead have unique positions on the exterior of the mitochondrial ribosome [11 , 12] . Recent cryo-electron microscopy ( cryo-EM ) reconstructions indicate that the additional protein elements may fulfil roles necessitated by the unique features of the mitochondrial leaderless mRNAs , likely in their recognition , as well as facilitating the translation of the particularly hydrophobic proteins they encode and making the contacts between the small and large ribosomal subunits [6 , 7 , 9 , 10 , 12] . The mitoribosomes are located inside the matrix , the site of the transcriptome , and associate closely with the mitochondrial inner membrane through MRPL45 [7] . This positioning allows for co-translational insertion of the hydrophobic proteins , which are translated by the mitoribosome , into the inner membrane and their assembly into oxidative phosphorylation ( OXPHOS ) complexes [13] . There is very little known about the functions and roles of the mitochondria-specific ribosomal proteins , therefore characterizing their function within the ribosome and mitochondria will provide valuable insights into mitochondrial translation . The mitochondrial ribosomal protein of the small subunit 34 ( MRPS34 ) has been identified as one of 15 mitochondria-specific proteins that are part of the small ribosomal subunit [14 , 15] . Although MRPS34 has been found localized to mitochondria and associated with the human homolog of the Drosophila discs large tumor suppressor protein ( hDLG ) [16] , its role in mitochondria and protein synthesis has not been identified or characterized . Here we investigated the role of MRPS34 in mice carrying a homozygous mutation in the nuclear gene encoding this protein that causes a significant decrease of this protein . MRPS34 is required for protein synthesis of all 13 mitochondrially-encoded polypeptides and stability of the 12S rRNA and specific mRNAs . Dysfunction in the efficiency of mitochondrial protein synthesis leads to reduced mitochondrial oxygen consumption and respiratory complex activity in the mutant mice indicating that MRPS34 is essential for the stability of actively translating ribosomes . The Mrps34 mutation causes tissue-specific molecular and pathological changes that result in alterations in fractional shortening of the heart and pronounced liver steatosis that leads to fibrosis with age . Mitochondrial dysfunction caused by the Mrps34 mutation is likely caused by decreased levels of mitochondrial ribosomal subunits and translationally competent mitoribosomes .
A mouse line carrying an ENU-induced T203C point mutation in the Mrps34 gene that converts a leucine residue at position 68 to proline ( Fig . 1A ) was identified by whole exome sequencing [17] . Sanger sequencing shows that the Mrps34mut/mut mice are homozygous for the mutation that is absent in age and littermate matched control Mrps34wt/wt mice ( Fig . 1B ) . The leucine residue at position 68 is conserved in vertebrates ( Fig . 1A ) , and mutation to proline is predicted to disrupt the formation of an alpha helix . To determine whether the mutation caused changes in the abundance of the MRPS34 protein we carried out immunoblotting of mitochondrial lysates isolated from liver and heart of Mrps34 mut/mut and Mrps34wt/wt mice . The MRPS34 protein was reduced in the heart and liver of the mutant mice ( Fig . 1C ) , indicating that the mutation causes instability of the protein . The MRPS34 protein is expressed in all examined tissues including brain , colon , heart , kidney , liver , thymus , pancreas , skin and testis and the Mrps34 mutation results in decreased levels of the protein in these tissues ( S1 Fig ) . Mutations in nuclear genes encoding proteins that are part of the translational machinery have been shown to cause mitochondrial diseases with varying age of onset and diverse clinical pathologies that affect a range of different tissues [2 , 3] . Next we sought to determine the effects of the Mrps34 mutation , in young ( 6–8 week ) and aged ( 28–30 week ) mice , on mitochondrial function and the downstream effects on energy metabolism and disease pathology . The mitochondrially-encoded 12S and 16S rRNAs form the scaffolds for the mitochondrial ribosomes that use 22 mitochondrial tRNAs to translate the 11 mRNAs [12] . Therefore we analyzed if the Mrps34 mutation affected the steady-state levels and stability of mitochondrial RNAs in heart and liver by northern blotting . The 12S rRNA levels were reduced in hearts and livers of the young Mrps34mut/mut mice , and the 16S rRNA level was unaffected ( Fig . 2A ) , suggesting that the MRPS34 is required for the stability of the 12S rRNA . In addition , we observed that the levels of the mt-Nd5 mRNA were decreased in the livers , but not in the hearts of the young mutant mice ( Fig . 2A ) , indicating that the Mrps34 mutation causes tissue specific effects on mitochondrial RNA metabolism . The 12S rRNA was also reduced significantly in the heart and livers of the mutant aged mice ( Fig . 2B ) . Furthermore , the levels of specific mitochondrial mRNAs , mt-Co1 , mt-Nd1 and mt-Nd5 were decreased in the livers of aged mice . However , in the hearts of aged Mrps34wt/wt and Mrps34mut/mut mice only mt-Nd5 was decreased but not mt-Co1 or mt-Nd1 suggesting that in the liver the stability of specific mRNAs is more severely affected by the Mrps34 mutation with age ( Fig . 2A and 2B ) . The levels of mitochondrial tRNAs were unaffected in heart and liver mitochondria , suggesting that MRPS34 is necessary for the stability of 12S rRNA and for the stability of specific mRNAs . Since MRPS34 is a ribosomal protein we analyzed how decreased levels of this protein , as a result of the Mrps34 mutation , affect translation by measuring de novo protein synthesis of the 13 mitochondrially-encoded polypeptides in heart and liver mitochondria from Mrps34mut/mut and Mrps34wt/wt mice . The overall decrease of mitochondrial protein synthesis in the young and aged mutant mice compared to controls ( S2 Fig ) indicates that MRPS34 is required for mitochondrial protein synthesis . To investigate the effects of the mutation on the rate of protein synthesis between heart and liver mitochondria we measured mitochondrial translation over time . Interestingly we observed that the initial rate of translation in control liver mitochondria is faster compared to that of heart mitochondria ( Figs . 3A , S2 ) . In addition , we observed that the Mrps34 mutation causes a decrease in mitochondrial protein synthesis ( Fig . 3A ) , however the initial rate in translation may account for the severity of the molecular changes found in the liver compared to heart mitochondria . Next we analyzed the effects of the Mrps34 mutation on the steady-state abundance of mitochondrial proteins by immunoblotting . The levels of the mitochondrially-encoded COXI and COXII were reduced in the livers and hearts of young Mrps34mut/mut mice compared to controls ( Fig . 3C ) . In the aged mutant mice we observed a more pronounced decrease in COXI and COXII in both the heart and liver , suggesting that the effects of decreased MRPS34 levels on the steady state levels of these proteins are cumulative and the molecular defects have a late onset in the heart ( Fig . 3D ) . In addition , we observed reduction in nuclear-encoded mitochondrial proteins , such as NDUFA9 and COXIV , possibly in a retrograde response to the decreased levels of mitochondrially-encoded proteins in the livers of aged Mrps34mut/mut mice ( Fig . 3D ) . We investigated the effects of Mrps34 mutation on the abundance of the mitochondrial respiratory complexes by blue native polyacrylamide gel electrophoresis ( BN-PAGE ) . The reduction in the abundance of the respiratory complexes in hearts of Mrps34mut/mut mice is more apparent in the aged mice compared to the young ( Fig . 4A and 4B ) . The reduction of the respiratory complexes was more significant in the livers of Mrps34mut/mut compared to Mrps34wt/wt mice ( Fig . 4A and 4B ) and this was confirmed by immunoblotting of each complex following BN-PAGE ( Fig . 4C and 4D ) . Complexes I and IV were reduced in the hearts of both young and aged mice , whereas in the livers , Complexes I , III , IV and V were reduced ( Fig . 4C and 4D ) . Taken together these findings provide evidence that MRPS34 is required for protein synthesis and that decreased mitochondrial translation as a result of the Mrps34 mutation can have varied effects on the abundance of mitochondrial proteins and respiratory complexes in different tissues . Because we observed a decrease in the abundance of mitochondrial respiratory complexes we measured their enzyme activities in heart and liver mitochondria from young and aged Mrps34mut/mut and Mrps34wt/wt mice . The activities of the respiratory complexes were not significantly decreased in hearts of young Mrps34mut/mut mice ( Fig . 5A ) , although there was a trend towards a decrease in in the activity of Complex IV , likely as a result of its decreased abundance ( Fig . 4C ) . The activities of Complexes III and IV were significantly decreased in livers of the young mutant mice ( Fig . 5B ) , consistent with the observed decrease in the levels of these complexes relative to those in control mice . In the hearts of aged Mrps34mut/mut mice only the activity of Complex IV was decreased ( Fig . 5C ) , whereas the activities of both Complex III and IV were most significantly affected in the livers of these mice ( Fig . 5D ) as a result of reduction in the abundance of these complexes identified by immunoblotting ( Fig . 4C and 4D ) . Measurements of oxygen consumption confirmed that mitochondrial respiratory function was affected more in the livers than hearts of young Mrps34mut/mut mice and this reduction was more dramatic with age in both tissues ( Figs . 5E , 5F , and S3 ) . Consistent with the observed molecular changes in response to the Mrps34 mutation , mitochondrial dysfunction was more pronounced in the liver compared to the heart of Mrps34mut/mut mice ( Fig . 5E ) , suggesting that these two organs have different capacities to cope with changes in translational efficiency at different ages as observed when we measured mitochondrial translation ( Fig . 3A ) . Mutations in genes encoding mitochondrial ribosomal proteins have been shown to cause cardiomyopathy in patients that result in impaired mitochondrial protein synthesis and consequently reduction in the levels and activities of respiratory complexes [18 , 19] . The Mrps34mut/mut mice appeared similar to their Mrps34wt/wt mice littermates at birth and no developmental or fertility differences were observed compared to the control mice . However , with age the Mrps34mut/mut mice develop physiological changes that affect multiple tissues to varying extents . We observed slight reduction in vision in the Mrps34mut/mut compared to the Mrps34wt/wt mice , measured using optokinetic drum experiments , although the optic nerve was not affected ( S4A–S4C , S4H Fig ) . The motor coordination , strength , and balance between the Mrps34wt/wt and Mrps34mut/mut mice was not significantly different ( S4D–S4G Fig ) , although a slight difference at day 4 may suggest a potential motor learning deficit . We observed centralization of nuclei in the muscle of aged Mrps34mut/mut mice and decreased COX activity ( S4I Fig ) . To investigate if reduced Complex IV activity , as a result of the Mrps34 mutation , affects heart function we carried out echocardiography on the Mrps34mut/mut and Mrps34wt/wt mice . We found that the hearts of mutant mice have increased fractional shortening , a thickening ( hypertrophy ) of the posterior wall and associated decreased oxygen consumption ( Fig . 6A ) that can be a common consequence of mitochondrial dysfunction . However , because we found greater mitochondrial dysfunction in the livers of Mrps34mut/mut mice we also investigated the effects of the mutation on the morphology and function of the livers in these mice compared to Mrps34wt/wt mice . Morphological examination revealed increased lipid accumulation in the livers of the young mutant mice and this was more pronounced in the aged mutant mice , suggesting that they have developed liver steatosis ( Fig . 6B ) . Oil red O staining revealed extensive accumulation of lipid droplets in the livers of Mrps34mut/mut mice that was significantly exacerbated with age ( Fig . 6C and 6D ) and correlated with increased levels of alanine aminotransferase ( ALT ) in the serum of these mice ( Fig . 6E and 6F ) , which is a marker of liver dysfunction , commonly associated with liver steatosis [20] . Extensive liver dysfunction can cause fibrosis , therefore we used Gomori’s trichome to stain the livers of Mrps34mut/mut and Mrps34wt/wt mice ( Fig . 6B and 6C ) . In the young Mrps34mut/mut mice infiltration of collagen was found around portal tracts ( Fig . 6B , arrows ) compared to control mice , that was more pronounced in the aged Mrps34mut/mut mice , disrupting the morphology of the liver , encapsulating the tissue ( arrow ) and forming nodules that are markers of liver fibrosis ( Fig . 6C ) . Our findings indicate that the Mrps34 mutation causes mitochondrial dysfunction that can affect multiple tissues , making these mice a model system to investigate how nuclear mutations can cause pathology with varying severity in different tissues . To understand the role of MRPS34 within the mitochondrial ribosome we investigated the effects of the Mrps34 mutation on the levels of mitochondrial ribosomal proteins from the small and large subunit in Mrps34mut/mut compared to Mrps34wt/wt mice by immunoblotting . It has previously been shown that the loss of certain mitochondrial small ribosomal subunit proteins can disrupt the assembly of the small subunit and cause the loss or reduction of other small ribosomal subunit proteins [21] , while loss of other small subunit proteins does not [22] , potentially revealing their roles in ribosome assembly . We investigated the levels of MRPS16 , MRPS25 , and MRPS35 mitochondrial small ribosomal subunit proteins , and found that their levels closely paralleled those of MRPS34 ( Fig . 7A ) , suggesting that the Mrps34 mutation causes destabilization of the small ribosomal subunit . This reduction in the small ribosomal subunit proteins is consistent with the reduced levels of the 12S rRNA ( Fig . 2A and 2B ) . Although in the young mutant mice the levels of the large ribosomal subunit protein MRPL44 were slightly increased by the Mrps34 mutation ( Fig . 1C ) , likely as a compensatory response to decreased small ribosomal subunit proteins ( as previously observed in [23] ) , we observed that in the aged mutant mice the MRPL44 and MRPL23 proteins were also decreased compared to controls ( Fig . 7A ) both in heart and liver mitochondria . As ribosomal proteins are required in tightly regulated ratios , decrease in MRPS34 over time leads to reduction in other mitoribosomal proteins . Therefore we conclude that MRPS34 is required for the steady-state levels of small subunit ribosomal proteins , and thereby the stability of the small ribosomal subunit , which is necessary for decoding of mitochondrial mRNAs when coupled with the large subunit and consequently translation of the mitochondrially-encoded proteins . Next we investigated if the residual levels of the mutant MRPS34 protein can interact with the mitochondrial ribosome using sucrose gradients followed by immunoblotting for this protein in heart and liver mitochondrial from the mutant and control mice ( Fig . 7B ) . We observed that there are significantly reduced levels of the MRPS34 protein in both heart and liver mitochondria , however the mutant protein can associate with the small ribosomal subunit and constitute part of the actively translating mitoribosome . To determine if MRPS34 is required for the stability of the mitoribosome we analyzed the mitochondrial ribosome profile of the large and small subunit , the monosome and polysome on sucrose gradients followed by immunoblotting of ribosomal marker proteins ( Fig . 7C and 7D ) . In heart and liver mitochondria from young Mrps34mut/mut mice we observed a decrease in the presence of the small ribosomal proteins MRPS16 and MRPS35 and consequently decrease in the presence of actively translating mitochondrial ribosomes . Similarly , we observed a decrease in the actively translating mitochondrial ribosome in young heart and liver mitochondria when we immunoblotted for the large ribosomal subunit markers MRPL44 and MRPL23 and a small increase in the presence of the large subunit consistent with the slight increase in the steady state levels of these proteins ( Fig . 1C ) , that are more apparent in heart mitochondria ( Fig . 7C ) . The decreased levels of the small ribosomal subunit result in decreased formation of mitochondrial ribosomes . In liver and heart mitochondria from aged mutant mice we observe decreased levels in both the small and the large ribosomal subunits and consequently reduced levels of the actively translating mitoribosomes ( Fig . 7D ) . In the aged Mrps34mut/mut mice we found that both the ribosomal proteins from small and large mitochondrial subunits are redistributed compared to control mice and we observe reduced polysome formation in the mutant mice , indicating destabilization of mitoribosomes and polysomes with age ( Fig . 7D ) , likely as a result of decreased MRPS34 levels ( Fig . 7A and 7B ) . These results indicate that reduced abundance of the MRPS34 protein affects the rate of translation through destabilization of the small ribosomal subunit and the 12S rRNA that are required for mitoribosome formation .
The mitochondrial ribosome is a unique molecular machine that has been honed through evolution to cope with the compaction of the mitochondrial genome [24] . The lack of substantial untranslated regions has given rise to leaderless mitochondrial mRNAs that are somehow recognized by the mitoribosome for translation . These ribosomes have acquired additional mitochondria-specific proteins that predominantly decorate the surface of the ribosome , giving clues that these proteins may play a role in specific recognition and recruitment of mRNAs to the mitoribosome , modulation of translation according to mitochondrial environmental changes , and the exclusive translation of hydrophobic membrane proteins [7 , 10] . The mitochondria-specific proteins are largely uncharacterized and understanding their roles in the mitoribosome would provide insight into the unique features of mitochondrial protein synthesis that are different from cytoplasmic and prokaryotic systems . The MRPS34 protein was identified as a constituent of the small ribosomal subunit [6 , 25] , it shares little homology to other proteins and is conserved from humans to zebrafish . Here we show that a point mutation in the mouse Mrps34 gene , causing a leucine to proline change in a conserved alpha helix , results in a significant reduction of the MRPS34 protein in heart and liver mitochondria . Consequently , mitochondrial protein synthesis is impaired in the mice homozygous for the Mrps34 mutation leading to pathologies with varying severities in different tissues . Decreased levels of MRPS34 affect the stability of mitochondrial RNAs and this is particularly pronounced in the liver . The 12S rRNA is specifically decreased in the hearts and livers of the mutant mice , although this reduction is more profound in the liver , particularly with age , indicating that MRPS34 is required for the stability of the 12S rRNA . Although the mutation causes significant reduction of MRPS34 , the remaining protein is assembled within the small subunit of the mitoribosome . However the significant decrease in 12S rRNA levels as a result of the MRPS34 reduction suggests that this protein may be required early in the assembly of the small ribosomal subunit and is likely a 12S rRNA-binding protein . We found that reduction in MRPS34 levels affects the abundance of small ribosomal subunit proteins . In the young mice the proteins of the large ribosomal subunit were present in substantial amounts , suggesting that the large subunit was still fully assembled despite the significant loss of the small subunit . This is further corroborated by the reduction in abundance of the 12S rRNA , which is readily degraded unless incorporated into a ribosomal subunit [26] , and the uninterrupted presence of the 16S rRNA . This finding suggests that there is no regulatory cross-talk monitoring the levels of the mitochondrial large and small subunit proteins , instead there seems to be a compensatory mechanism that increases the level of large ribosomal proteins in response to loss of small ribosomal proteins . However this may not be long lasting as with age persistent loss of the small ribosomal subunit also leads to reduction of the large ribosomal subunit proteins and destabilization of the actively translating ribosomes in the Mrps34mut/mut mice . In the mutant mice we observed re-distribution of mitochondrial ribosomal proteins suggesting that the assembly of mitochondrial ribosomes is compromised when there are insufficient levels of MRPS34 . The differential distribution of the ribosomal subunits in the mutant compared to control mice suggests that there is accumulation of ribosome assembly intermediates due to the reduced MRPS34 levels . In the young mice this re-distribution is observed solely in the small ribosomal subunit proteins indicating that the small ribosomal subunit is dissociating while the large remains stable . This may result in the formation of inaccurately assembled ribosomes , which are easily dissociated into subunits or are unable to bind to docking sites on the inner mitochondrial membrane ( IMM ) affecting protein synthesis and OXPHOS biogenesis . As a result of this we find that over time , in the aged mice , the stability of the mitochondrial ribosome is affected causing a reduction in polysome formation . These changes may account for the more obvious mitochondrial dysfunction in the aged mutant mice . Furthermore , as the ribosomal proteins are required in a particular stoichiometry , loss of a ribosomal protein would destabilize this balance and would result in decreased assembly of de novo mitochondrial ribosomes . This is confirmed by our data where persistent decrease in MRPS34 in the aged mutant mice leads to reduction of actively translating ribosomes as a result of a decrease in small and large ribosomal subunit proteins . We found decreased levels of specific mRNAs , including mt-Nd1 , mt-Nd5 and mt-Co1 in the livers of mutant mice , while only the mt-Nd5 mRNA was reduced in the hearts of aged mutant mice . The levels of other mRNAs such as mt-Co2 and the tRNAs were unaffected in both heart and liver , and with age . Taken together these findings reveal that MRPS34 is necessary for the stability of specific mitochondrial mRNAs and may indicate that the active translation of certain mRNAs is linked to their stabilities . The different effects on the mitochondrial mRNAs between heart and liver reflect tissue-specific changes that have been found in mitochondrial disease patients previously [18 , 27 , 28] , as well as recently in a mouse model of mitochondrial disease [29] . In the heart , mitochondrial RNAs account for at least 30% of the total RNA [30] reflecting the dependence of the heart on OXPHOS . It is likely that in the heart there is excess production of mitochondrial mRNAs relative to the required threshold for normal function , such that decreases in mRNA levels do not compromise mitochondrial function in the short term . This is not the case in the liver where the decrease in mitochondrial RNAs has more profound effects on energy metabolism . Furthermore we found that MRPS34 is more abundant in the heart compared to the liver of control mice , which may protect the heart from a more pronounced decline in function in the mutant mice . We find that the levels of the other mitoribosomal proteins in control mice is higher in the heart compared to the liver , likely due to the higher mitochondrial content in the heart . Finally we observe that the liver has a faster initial rate of mitochondrial translation that is compromised in the mutant mice and may contribute to the more pronounced defect in the liver compared to the heart . It could be that after the initial burst of translation in the liver the levels of mitoribosomes or their recycling become limiting so that subsequently translation plateaus , whereas in the heart the higher abundance of mitoribosomes contributes to the steady rate of translation . Unlike the heart , the liver is highly proliferative and it may require rapid bursts of mitochondrial translation for its normal function and during regeneration . Mitochondrial diseases caused by mutations in nuclear genes encoding mitochondrial proteins can affect the function of many different tissues with varying severity ( reviewed in [2 , 3] ) . Furthermore , nuclear mutations in mitochondrial proteins can result in remarkably heterogeneous defects with varying severity in different tissues that are poorly understood currently [2] . We observed similar varying defects in different tissues in the Mrps34mut/mut mice , which have a fractional shortening of their hearts , centralized nuclei in their muscles and increased lipid accumulation in their livers that causes liver steatosis . Hepatopathies have been identified along with a range of different symptoms in many mitochondrial disorders , although hepatopathy was the main consequence of a mutation in the tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase ( TRMU ) gene that produces the enzyme responsible for 2-thiouridylation of the tRNAGlu , tRNAGln and tRNALys , causing a severe but reversible infantile hepatopathy [31–33]; the mutation in the Mrps34 gene is the first to show pronounced liver defects in mice providing the means to investigate the contribution of mitochondrial function to hepathopathy in the future . Many mutations in nuclear genes that encode protein components of the translational machinery result in compromised biogenesis of specific or all mitochondrial respiratory chain complexes and lead to decreased OXPHOS and some of these have been shown to cause accumulation of lipids in hepatocytes [3 , 34] . Although we observe overall decrease in mitochondrial protein synthesis as a result of the Mrps34 mutation , the greatest reduction is in the activity of Complex IV in both heart and liver , and this reduction is greater in the liver . Decrease in the mt-Co1 mRNA and consequently the COXI protein , that is necessary for the biogenesis of Complex IV [35] likely contributes to this pronounced decrease in its abundance and activity . Our work has shown that MRPS34 plays a role in maintaining the stability of the small ribosomal subunit and the 12S rRNA , which are necessary formation of actively translating mitoribosomes . In addition , MRPS34 is required for the stability of specific mRNAs , indicating that the mitochondria-specific ribosomal proteins might have unique roles in mitochondrial RNA metabolism . Because the Mrps34 mutation in mice is not embryonic lethal it has provided a means to investigate how mitochondrial dysfunction can lead to disease in the whole body and the progression of the disease with age . Establishment of mouse models where mitochondrial dysfunction causes disease are particularly important for understanding the causes of human disease pathology , identification of drug targets and for the development of future treatments for these diseases .
Male age- and litter-mate matched ( 6–8 weeks ‘young’ and 30 weeks ‘aged’ ) wild-type ( Mrps34wt/wt ) and homozygous ( Mrps34mut/mut ) ENU mutant mice on a C57BL/6J background were obtained from the Australian Phenomics Facility . The Mrps34 mice were bred onto a C57BL/6J background for 8–10 generations . Animals were singly housed in standard cages ( 45 cm × 29 cm × 12 cm ) under a 12-h light/dark schedule ( lights on 7 a . m . to 7 p . m . ) in controlled environmental conditions of 22 + 2°C and 50 + 10% relative humidity . Normal chow diet ( Rat & Mouse Chow , Speciality Foods , Glen Forrest , Western Australia ) and water were provided ad libitum . The study was approved by the Animal Ethics Committee of the UWA ( AEC 03/100/526 ) and performed in accordance with Principles of Laboratory Care ( NHMRC Australian code for the care and use of animals for scientific purposes , 8th Edition 2013 ) . Behavioral and motor tests are described in Supplemental Methods ( S1 Text ) . Tissues were homogenised in 100 μl of 100 mM Tris , 2 mM Na3VO4 , 100 mM NaCl , 1% Triton X-100 , 1 mM EDTA , 10% Glycerol , 1 mM EGTA , 0 . 1% SDS , 1 mM NaF , 0 . 5% deoxycholate , 20 mM Na4P2O7 , pH 7 . 4 containing PhosSTOP Phosphatase Inhibitor Cocktail and EDTA-free Complete protease inhibitor cocktail and the supernatant was collected after centrifugation at 10 , 000 g . Mitochondria were isolated from homogenized hearts and livers and isolated by differential centrifugation as described previously [36] with some modifications . Livers were homogenized in buffer containing 250 mM sucrose , 5 mM Tris , 1mM EGTA , pH 7 . 4 with EDTA-free Complete protease inhibitor cocktail ( Roche ) and hearts were homogenized in 210 mM mannitol , 70 mM sucrose , 10 mM Tris , 0 . 1mM EDTA pH 7 . 4 containing EDTA-free Complete protease inhibitor cocktail . Liver ( 1 . 2 mg of protein ) and heart ( 0 . 8 mg of protein ) mitochondria were lysed with 1% n-Dodecyl β-D-maltoside in 10mM Tris-HCl , pH 7 . 4 , 260 mM sucrose , 100 mM KCl , 20 mM MgCl2 in the presence of RNasin and protease inhibitors for 30 min , the lysate centrifuged at 10 , 000 g for 45 min at 4°C , the clarified lysate was loaded on a continuous 10–30% sucrose gradient ( in 10 mM Tris-HCl , pH 7 . 4 , 100 mM KCl , 20 mM MgCl2 in the presence of RNasin and protease inhibitors ) and centrifuged at 71 , 000 g in an Optima Beckman Coulter preparative ultracentrifuge as described before [6] . Fractions were collected and precipitated with 30% trichloroacetic acid , washed in acetone , and the entire fraction was resolved by SDS-PAGE . Protein markers of the mitochondrial ribosomal subunits were detected by immunoblotting as described below . RNA was isolated from heart and liver mitochondria using the miRNeasy Mini kit ( Qiagen ) incorporating an on-column RNase-free DNase digestion to remove all DNA . RNA ( 5 μg ) was resolved on 1 . 2% agarose formaldehyde gels , then transferred to 0 . 45 μm Hybond-N+ nitrocellulose membrane ( GE Lifesciences ) and hybridized with biotinylated oligonucleotide probes specific to mouse mitochondrial tRNAs , mRNAs and rRNAs . The hybridizations were carried out overnight at 50°C in 5x SSC , 20 mM Na2HPO4 , 7% SDS and 100 μg . ml-1 heparin , followed by washing . The signal was detected using streptavidin-linked infrared antibody ( diluted 1: 2 , 000 in 3x SSC , 5% SDS , 25 mM Na2HPO4 , pH 7 . 5 ) using an Odyssey Infrared Imaging System ( Li-Cor ) . Specific proteins were detected using rabbit polyclonal antibodies against: MRPL44 , MRPL23 , MRPS16 , MRPS35 , MRPS25 ( Proteintech , diluted 1:1000 ) , MRPS34 ( Sigma , diluted 1:1000 ) and mouse monoclonal antibodies against: porin , NDUFA9 , Complex II , Complex III , COXI , COXII , COXIV and Complex V subunit ( Abcam , diluted 1:1000 ) , in Odyssey Blocking Buffer ( Li-Cor ) . IR Dye 800CW Goat Anti-Rabbit IgG or IRDye 680LT Goat Anti-Mouse IgG ( Li-Cor ) secondary antibodies were used and the immunoblots were visualized using an Odyssey Infrared Imaging System ( Li-Cor ) . Tissue specific immunoblotting analysis was performed on a Proteintech mouse tissue blot ( Cat . No . M10005 ) . Mitochondrial de novo protein synthesis was analyzed using Expres35S Protein Labelling Mix [35S] ( 14 mCi , Perkin–Elmer ) as described before [23] . Liver and heart mitochondrial lysates were resolved by BN-PAGE to detect the respiratory complexes by Coomassie staining as described previously [8 , 37] or by immunoblotting as described above . The enzyme activities of all five respiratory complexes and citrate synthase were measured in a 1 ml cuvette at 30°C using a Perkin Elmer lambda 35 dual beam spectrophotometer as described in [38] . Mitochondrial respiration was evaluated as O2 consumption in isolated heart and liver mitochondria according to Kuznetsov et al . Mitochondria were supplemented with substrates for either complex I ( 10 mM glutamate/malate , Sigma ) , II ( 10 mM succinate , Sigma ) or III ( 1 mM TMPD/1 mM ascorbate , Sigma ) . After addition of 1 mM adenosine diphosphate ( ADP , Sigma ) to the recording chamber , State 3 respiration activity was measured . ADP independent respiration activity ( State 4 ) was monitored after addition of oligomycin ( 2 μg/ml , Sigma ) . Echocardiographic studies to measure left ventricular function were performed on mice under light methoxyflurane anesthesia with the use of an i13L probe on a Vivid 7 Dimension ( GE Healthcare ) . Echocardiographic measurements were taken on M-mode in triplicate from each mouse and the quantitative measurements represent the average . M-mode recordings were made at a sweep speed of 200 mm/s . Measurements of left ventricular end diastolic diameter ( LVEDD ) , left ventricular end systolic diameter ( LVESD ) , fractional shortening ( FS ) , left ventricular posterior wall in diastole ( LVDPW ) , left ventricular posterior wall in systole ( LVSPW ) , intraventricular septum in diastole ( IVDS ) , and intraventricular septum in systole ( IVSS ) were made . Fractional shortening ( FS ) was calculated by the formula [ ( LVEDD-LVESD ) /EDD] x 100 . Fresh sections of the liver and muscle were frozen in Optimal Cutting Temperature ( OCT ) medium , sectioned and stained with Haematoxylin and Eosin , Gomori’s Trichrome and Haematoxylin , Oil Red O and Haematoxylin , COX or NADH stains . Images were acquired using a Nikon Ti Eclipse inverted microscope using a Nikon 40x objective and Oil Red O staining was quantified as described previously [39] . | Mitochondria make most of the energy required by eukaryotic cells and therefore they are essential for their normal function and survival . Mitochondrial function is regulated by both the mitochondrial and nuclear genome . Mutations in nuclear genes encoding mitochondrial proteins lead to mitochondrial dysfunction and consequently diminished energy production , a major symptom of metabolic and mitochondrial diseases . The molecular mechanisms that regulate mitochondrial gene expression and how dysfunction of these processes causes the pathologies observed in these diseases are not well understood . Messenger RNAs encoded by mitochondrial genomes are translated on mitochondrial ribosomes that have unique structure and protein composition . Mitochondrial ribosomes are a patchwork of core proteins that share homology with prokaryotic ribosomal proteins and mitochondria-specific proteins , which can be unique to different organisms . Mitochondria-specific ribosomal proteins have key roles in disease however their functions within mitochondria are not known . Here we show that a point mutation in a mammalian-specific ribosomal protein causes mitochondrial dysfunction , heart abnormalities and progressive liver disease . This mouse provides a valuable model to elucidate the pathogenic mechanisms and progression of metabolic diseases with age , while enabling a more thorough understanding of mitochondrial ribosomes and protein synthesis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Mutation in MRPS34 Compromises Protein Synthesis and Causes Mitochondrial Dysfunction |
Bunyamwera orthobunyavirus is both the prototype and study model of the Bunyaviridae family . The viral NSs protein seems to contribute to the different outcomes of infection in mammalian and mosquito cell lines . However , only limited information is available on the growth of Bunyamwera virus in cultured mosquito cells other than the Aedes albopictus C6/36 line . To determine potential functions of the NSs protein in mosquito cells , replication of wild-type virus and a recombinant NSs deletion mutant was compared in Ae . albopictus C6/36 , C7-10 and U4 . 4 cells , and in Ae . aegypti Ae cells by monitoring N protein production and virus yields at various times post infection . Both viruses established persistent infections , with the exception of NSs deletion mutant in U4 . 4 cells . The NSs protein was nonessential for growth in C6/36 and C7-10 cells , but was important for productive replication in U4 . 4 and Ae cells . Fluorescence microscopy studies using recombinant viruses expressing green fluorescent protein allowed observation of three stages of infection , early , acute and late , during which infected cells underwent morphological changes . In the absence of NSs , these changes were less pronounced . An RNAi response efficiently reduced virus replication in U4 . 4 cells transfected with virus specific dsRNA , but not in C6/36 or C7/10 cells . Lastly , Ae . aegypti mosquitoes were exposed to blood-meal containing either wild-type or NSs deletion virus , and at various times post-feeding , infection and disseminated infection rates were measured . Compared to wild-type virus , infection rates by the mutant virus were lower and more variable . If the NSs deletion virus was able to establish infection , it was detected in salivary glands at 6 days post-infection , 3 days later than wild-type virus . Bunyamwera virus NSs is required for efficient replication in certain mosquito cell lines and in Ae . aegypti mosquitoes .
Bunyamwera virus ( BUNV ) is the prototype of both the Orthobunyavirus genus and the Bunyaviridae family . It was originally isolated from a pool of several Aedes spp . mosquitoes collected in the Semliki Forest in Uganda [1] . Based on detection of antibodies to BUNV in human sera and isolations of BUNV from patients suffering febrile illness , the virus is widely distributed in several regions of sub-Saharan Africa [2]–[4] . BUNV is maintained in nature by a propagative cycle involving blood-feeding mosquitoes and susceptible vertebrate hosts , probably small rodents [5] . BUNV can replicate efficiently in both vertebrate and invertebrate cells in culture but with different outcomes: in mosquito cells no cytopathology is observed and persistent infection is established , whereas in mammalian cells infection is lytic and leads to cell death [6]–[8] . From a practical standpoint , this is shown by the ability of the virus to form clear lytic plaques in cells of vertebrate origin but not in those derived from insects . Like all bunyaviruses , BUNV is an enveloped virus containing a tri-segmented , single stranded negative-sense RNA genome that encodes four common structural proteins: an RNA-dependent RNA polymerase ( L protein ) on the large ( L ) segment , two glycoproteins ( Gc and Gn ) on the medium ( M ) segment and the nucleoprotein ( N ) on the smallest ( S ) segment . BUNV also codes for two non-structural proteins , NSm on the M segment and NSs on the S segment [9] . The segmented nature of the genome allows for reassortment between closely related orthobunyaviruses to generate viruses that may have altered biological properties , such as Ngari virus , which is associated with human haemorrhagic fever in East Africa , whose genome comprises L and S segments from BUNV and M segment from Batai virus [10]–[12] . BUNV continues to serve as a convenient laboratory model to study the molecular biology of bunyaviruses in general , and an understanding of many aspects of gene function and viral replication have been developed using BUNV , including the establishment of a robust reverse genetics system [13]–[15] . The BUNV NSs protein is a nonessential gene that contributes to viral pathogenesis . It has been shown that in mammalian cells , NSs induces shut-off of host protein synthesis [16] , [17] , which leads to cell death . It also counteracts the host cell antiviral response and seems to be the main virulence factor [18]–[21] , acting at the level of transcription by inhibiting RNA polymerase II–mediated transcription [22] . In mosquito cells neither host cell transcription nor translation are inhibited [17] , and although so far no function for the orthobunyavirus NSs protein has been found in mosquito cells [23] , it seems the differential behaviour of NSs could be one of the factors responsible for different outcomes of infection in mammalian and mosquito cell lines . To date , studies on BUNV replication in mosquito cells have been limited to the C6/36 line [24] from Aedes albopictus [6]–[8] , [16] , [17] , [25]–[27] . As differences in the appearance of cytopathic effects , cell death and viral morphogenesis were observed in various Ae . albopictus subclones infected with Sindbis virus [28]–[32] , we have compared the replication of BUNV in two additional Ae . albopictus cell clones , C7-10 [33] and U4 . 4 [34] . All three cell lines were independently obtained from the original Singh cell line derived from Ae . albopictus neonate larvae [35] , which has been shown to comprise a heterogeneous population of cell types . More recently it has been reported that these cell lines differ in their RNAi responses: C7-10 [36] and C6/36 [37]–[39] cells have impaired Dicer 2-based RNAi responses , whereas the U4 . 4 cell line encodes a fully functional Dicer 2 gene [36] , [40] . Secondly , we have investigated the ability of BUNV to replicate in Aedes aegypti ( Ae ) cells , and have compared this to the infection in living Aedes aegypti mosquitoes . Here we report that Ae . albopictus U4 . 4 and Ae . aegypti Ae cells are refractory to infection with a recombinant BUNV lacking the NSs gene ( rBUNdelNSs2; [16] , [17] ) , and that expression of NSs influences the efficiency of viral replication in Ae . aegypti mosquitoes .
Aedes albopictus C6/36 , C7-10 and U4 . 4 and Aedes aegypti Ae cells were grown at 28°C in Leibovitz 15 medium ( Gibco ) supplemented with 10% foetal calf serum ( FBS ) and 10% tryptose phosphate broth . Vero E6 cells were maintained at 37°C in Dulbecco's modified Eagle's medium ( Gibco ) supplemented with 10% FBS . Working stocks of wild-type Bunyamwera virus ( wtBUNV ) , the recombinant NSs deletion virus ( rBUNdelNSs2; [17] ) , rBUNGcGFP [41] and rBUNdelNSs-GcGFP [42] were prepared as described previously [18] , [43] . Mosquito cells were infected at an MOI of 1 PFU/cell . After 1 h incubation at 28°C , the inoculum was removed and the cells were washed once to remove unattached virus . Supernatants were harvested at various times post-infection and assayed for virus by plaque assay on Vero E6 cells as previously described [13] , [44] . In our laboratory , we routinely titrate BUNV on Vero E6 cells as these give the most easily discernible plaques . We have also performed immunostaining assays of viral foci produced in C6/36 cells , and observed that the efficiency of plating compared to Vero E6 cells is in the range 0 . 5 to 1 ( data not shown ) . This is similar to the plating efficiency for other arboviruses , like Dengue , West Nile and St Louis encephalitis viruses [45] , when comparing titration on C6/36 and vertebrate cells . At different times after infection , cell lysates were prepared and equal amounts of sample were separated on 12% SDS-PAGE . Separated proteins were transferred to Hybond-C Extra membranes ( Amersham ) . The membranes were incubated with rabbit anti-BUN N protein antibody ( 1∶2000 dilution ) and mouse anti-tubulin antibody ( Sigma; 1∶10000 ) as a loading control , followed by incubation with anti-rabbit horseradish peroxidase-coupled antibody ( Cell Signaling Technology ) . Immunocomplexes on the membranes were detected by SuperSignal West Pico Substrate ( Pierce ) according to manufacturer's instructions . Mosquito cells were cultured on glass coverslips and infected with rBUNGceGFP or rBUNdelNSs-GcGFP at MOI of 1 PFU/cell . At various times post infection , cells were fixed with 4% formaldehyde and washed with PBS . The cells were incubated with rabbit anti-BUN N ( 1∶200 ) and mouse anti-tubulin ( Calbiochem; 1∶100 ) antibodies , followed by incubation with Texas Red-conjugated anti-rabbit ( Cell Signaling Technologies; 1∶200 ) and CY5-conjugated anti-mouse ( Sigma; 1∶400 ) antibodies . Nuclei were stained with DAPI . Prepared slides were examined with a Zeiss LSM confocal microscope . Laboratory-bred Ae . aegypti ( Paea strain ) mosquitoes were reared as previously described [46] . Female mosquitoes were selected and exposed to blood-meal containing approx . 108 PFU/ml of wtBUNV or rBUNdelNSs2 as described previously . At various times post infection , mosquitoes were anesthetized and either homogenized whole or the midguts and salivary glands were dissected . Organs and whole mosquitoes were homogenized in 100 µl of Leibowitz 15 medium ( Gibco ) . Twenty-five microliters of each sample were used for titration by plaque assay on Vero E6 cells . dsRNA approx . 1000 bp long was prepared using the Megascript RNAi kit ( Ambion ) . Virus specific dsRNA were prepared from linearized plasmids containing full-length cDNAs of the BUNV genome segments pT7riboBUNS ( + ) and ( − ) , pT7riboBUNM ( + ) and ( − ) and pT7riboBUNL ( + ) and ( − ) [13] . Renilla-specific dsRNA were prepared from plasmid phRL-CMV by adding a T7 promoter sequence at each end by PCR as described previously [40] . Purified dsRNA was stored in aliquots at −80°C . A total of 3×105 cells per well were cultured in 24 well plates . 100 ng of Renilla dsRNA or 100 ng of a 1∶1∶1 mixture of S- , M- , and L-segment specific dsRNA was used for transfection . One microliter of Lipofectamine-2000 ( Invitrogen ) was used per 100 ng and the transfection mixes were prepared in the final volume of 100 µl as per the manufacturer's protocol; this was then applied onto the cells with 400 µl fresh complete L15 medium . After 5 hours incubation at 28°C , 500 µl of fresh medium was added . The cells were infected with wtBUNV or rBUNdelNSs2 24 h post-transfection , supernatants were collected at various times thereafter for assay of released virus by plaque titration on VeroE6 cells .
In mosquito cells no shut-off of host cell transcription or translation has been observed . Therefore , we first confirmed that BUNV NSs protein is expressed in these cells . Ae . albopictus C6/36 cells were infected at MOI of 10 pfu/cell and the cells were harvested for protein analysis by Western blotting at different time points ( Figure 1 ) . Although BUNV N and NSs proteins are translated from the same mRNA , NSs was produced , though slightly later in the course of infection than the nucleoprotein . This is in line with previous observations [8] based on radiolabelling of infected cells . We next compared virus replication in Ae . albopictus C6/36 , C7-10 and U4 . 4 cell lines . Cells were infected at an MOI of 1 PFU/cell and the samples were collected at various times post infection . wtBUNV was able to replicate in all three cell lines but with different kinetics ( Figure 2A ) . Growth was most efficient in C6/36 cells , with maximum titres of released virus exceeding 108 pfu/ml , 10-fold and 100-fold higher than in C7-10 or U4 . 4 cells , respectively . rBUNdelNSs2 grew more slowly than wtBUNV in C6/36 cells and yielded maximum titres about 100-fold less , confirming previous results [16] , [17] . In C7-10 cells , the mutant virus showed similar kinetics to wt BUNV . In marked contrast , no increase in titre of rBUNdelNSs2 in culture medium of U4 . 4 cells was detected . ( The detection of some rBUNdelNSs2 virus in supernatant medium ( Figure 2A ) represents residual virus that remained after removal of the inoculum and replacement with fresh medium; these cells did not adhere firmly enough to the culture vessel to permit extensive washing ) . These data were supported by Western blot analysis to detect accumulation of the viral N protein ( Figure 2B ) . In C6/36 cells infected with wtBUNV , N could be seen as early as 16 h post-infection ( hpi ) , whereas in cells infected with the NSs deletion , mutant N was not detected until 36 hpi . In the C7-10 cell line , no significant difference between the two viruses was observed in terms of accumulation of N protein , though growth was slower than in C6/36 cells , with N protein first detectable at 48 hpi . In U4 . 4 cells infected with wtBUNV , N was detectable as early as 42 hpi whereas no N protein could be detected in cells infected with rBUNdelNSs2 . These results suggest that BUNV NSs protein is not essential for growth in either C6/36 or C7-10 cell lines , but is required for successful replication in U4 . 4 cells . To analyse how BUNV spreads in mosquito cells , Ae . albopictus C6/36 , C7-10 and U4 . 4 cells were infected with recombinant BUNV expressing green fluorescent protein [41] , either rBUNGc-eGFP or rBUNdelNSs-GcGFP at an MOI of 3 PFU/cell . Three different stages of infection , as proposed by Lopez-Montero et al . ( 2009 ) , were observed in all the Ae . albopictus cell lines infected with rBUNGc-eGFP . These phases were defined by changes in cell morphology due to virus replication , most notably that infected cells produced projections extending towards neighbouring cells ( Figure 3A ) . During the early phase of infection with the wild type version of the eGFP-expressing BUNV , the levels of viral proteins were relatively low and the cells resembled uninfected cells in shape ( Figure 3B ) . However , as infection progressed and the levels of viral proteins increased , cells transitioned into the acute phase . This phase was characterised by formation of filopodia that were most abundant between the 24 hpi and 48 hpi; over this time period , cells maintained physical contact via the projections . During the acute phase , the levels of eGFP-tagged Gc increased and the glycoprotein was also found in the filopodia ( Figure 3B ) . The late phase of the infection was characterised when the cellular filopodia started to disappear ( from 48 hpi ) , which was also manifested by elevated levels of Gc glycoprotein detected in the cells . Later on in the infectious cycle , the cells returned to their normal form ( Figure 3B ) . These results were consistent for infection of C6/36 and C7-10 cells . Infection in U4 . 4 cells was slower , and fewer cells were infected in comparison to C6/36 and C7-10 cells . Also , U4 . 4 cells produced fewer filopodia ( Figure 3B ) . Comparison of the levels of viral proteins in infected cells revealed that BUNV replication in U4 . 4 cells was less intense than in C6/36 and C7-10 cell lines , as relatively lower amounts of N and Gc proteins were produced . These observations corresponded with the differences in viral titres observed in the previous experiments ( Figure 2 ) , where U4 . 4 cells produced less infectious particles than the two other cell lines . The course of infection with the NSs deletion recombinant , rBUNdelNSs-GcGFP , was similar to rBUNGc-eGFP in terms of levels of viral protein expression , but the infection started later in C6/36 and C7-10 cells . Investigation of cell morphology showed that fewer filopodia were produced and they were less pronounced in rBUNdelNSs-GcGFP infected cells , suggesting involvement of NSs in this process . Minimal disruption of the normal cell morphology was observed; those changes that were seen were attributed to cell movement and consecutive attachment to the surface , as similar changes were observed in uninfected cells ( Figure 3B ) . These data suggested that NSs protein contributed to the efficiency of viral replication , but was a non-essential protein . When U4 . 4 cells were infected with rBUNdelNSs-GcGFP , a few cells ( less than 5% ) were observed to harbour replicating virus in that synthesis of tagged Gc protein could be observed by its autofluorescence ( Figure 3B ) . This suggests that infection by rBUNdelNSs-GcGFP is abortive and few , if any , new infectious particles were produced , in line with the titration data shown in Figure 2 . In order to rule out the possibility that changes in morphology were due to alterations of Gc glycoprotein caused by fusion of the GFP sequences , another fluorescent Bunyamwera virus was used , with eGFP fused in-frame into NSm [46] . Infection of the C6/36 , C7-10 and U4 . 4 cells with rBUNM-NSm-EGFP , but not with rBUNdelNSs-NSm-EGFP , resulted in similar marked morphological changes as shown by rBUNGc-eGFP ( data not shown ) . It has been previously documented that the outcome of infection of C6/36 cells with wtBUNV is the establishment of a persistent infection [6]–[8] . To determine whether persistent infections could also be established in C7-10 and U4 . 4 cells , and if the NSs protein participates in this process , cell monolayers were initially infected with wtBUNV or rBUNdelNSs2 at an MOI of 0 . 1 PFU per cell . After four days , supernatant fluids were collected for titration of released virus , and the infected cells were passaged ( split ratio of 1∶5 ) and grown until confluent . Thereafter , they were regularly passaged and maintained for about 7 months ( 25 passages ) . All lines initially infected with wtBUNV or rBUNdelNSs2 continued to shed infectious virus at each passage , though the titres fluctuated widely ( Figure 4A ) . Similarly , N protein was detected by Western blotting in all passaged cells , though the levels varied . Thus wtBUNV could establish persistent infections in all three cell lines , and rBUNdelNSs2 in C6/36 and C7-10 cells . As expected , since U4 . 4 cells did not show evidence of productive infection by rBUNdelNSs2 , no indication of a persistent infection was obtained . Thus the NSs protein was not essential for establishment of persistent infection in C6/36 or C7-10 cells . In previous studies of C6/36 cells persistently infected with BUNV , the generation of viruses displaying different plaque phenotypes when titrated in mammalian cells was observed [6] , [7] . Similarly , when titrating the two viruses released from all persistently infected cell lines described above in Vero cells , we observed the appearance of large , small and cloudy plaques from different passages ( data not shown ) . As mentioned above C7-10 and C6/36 cells are reported to have defective Dicer 2-based RNAi responses [36]–[39] , whereas the U4 . 4 cell line encodes a fully functional Dicer 2 gene [36] , [40] . To determine whether the cells could mount an RNAi response against BUNV infection , we transfected C6/36 , C7-10 and U4 . 4 cells with long virus-specific dsRNA , or Renilla luciferase-specific dsRNA as a control , and then infected the cells with either wtBUNV or rBUNdelNSs2 at a MOI of 5 PFU per cell . Supernatant fluids were collected at various times post-infection and assayed for the presence of infectious virus by plaque formation . Each infection was done in triplicate and the experiment was repeated twice . As shown in Figure 5 , no effect on virus growth was seen in C6/36 or C7/10 cells infected with either virus . However , in U4 . 4 cells transfected with virus-specific dsRNA and then infected with wt BUNV , virus replication was inhibited , and no increase in virus titre was observed . In contrast , transfection of Renilla luciferase-specific dsRNA had no effect on wtBUNV growth in any cell line . These results are consistent with U4 . 4 cells having a functional dsRNA-mediated interference system . Unfortunately , because rBUNdelNSs2 virus does not replicate productively in U4 . 4 cells it was not possible to determine whether BUNV NSs protein could be involved in evasion of an RNAi response in mosquito cells . Although Ae . albopictus cell lines are widely used to investigate arbovirus replication , the genome sequence of Ae . albopictus has yet to be determined , thus limiting detailed investigation of molecular details of virus-host interaction . To date three mosquito genome-sequencing projects have been completed , Anopheles gambiae , Aedes aegypti and Culex quinquefasciatus [47]–[49] , and hence we examined whether cell lines from other mosquito species would be useful to monitor BUNV replication . In preliminary experiments , no evidence for BUNV growth in the Sua4 cell-line derived from An . gambiae Suakoko strain [50] was obtained ( data not shown ) . However , the Ae cell line [51] derived from Ae . aegypti was shown capable of supporting wtBUNV replication ( Figure 6 ) . The virus grew to titres approaching 107 PFU/ml , and accumulation of BUNV N protein from 48 hpi was detected by Western blotting . In contrast , rBUNdelNSs2 appeared unable to replicate in these cells , as N protein was not detected by Western blotting ( Figure 6B ) and no increase in titre of infectious virus in the supernatant medium was measured . ( Again as these cells did not adhere firmly to the culture vessel , extensive washing to remove the virus inoculum was not possible , and only residual infecting virus was detected ) . To investigate if BUNV was capable of establishing persistent infection in the Ae cell line , cells were infected with wtBUNV or rBUNdelNSs2 at an MOI of 0 . 1 PFU per cell . Supernatants were collected and infected cells were then passaged using a fifth of gathered cells as it was done with Ae . albopictus cell lines . Cells were maintained for 7 passages . Analysis for the presence of infectious virus in the supernatants showed the Ae cell line to be persistently infected with wtBUNV ( Figure 6C ) . However the levels of infectious virus production remained low and there was no significant difference between consecutive passages as was seen in Ae . albopictus cell lines . Infection with rBUNdelNSs2 showed a different pattern . Infectious virus particles were detected after the first passage , but no virus was detected in the supernatants for the next two passages . However , from passage 4 , low titres of virus were detected ( Figure 6C ) . This pattern was reproducibly observed in two further independent infections of Ae cells with rBUNdelNSs2 , and also when another Ae . Aegypti cell line , A20 [51] was infected with rBUNdelNSs2: in all cases no virus could be detected by plaque assay of supernatants from the second and third passages , but virus was detected at subsequent passages at low levels ( data not shown ) . Having demonstrated that BUNV could replicate in Ae . aegypti cell cultures , we next studied the infection of Ae . aegypti mosquitoes . It has been reported previously that wtBUNV is capable of replicating in laboratory raised Ae . aegypti mosquitoes and of being transmitted via mosquito saliva to mice [52] . By comparing infection with wtBUNV and rBUNdelNSs2 , we investigated the role of NSs in an insect vector . Female Ae . aegypti ( Paea strain ) mosquitoes were allowed to feed on a blood-meal containing wtBUNV or rBUNdelNSs2 ( approx . 108 PFU/ml in the blood meal ) . After feeding , only engorged mosquitoes were kept for the course of the experiment . Three engorged females were collected from each infection group immediately after feeding to determine the amount of virus ingested . Based on the titration results , it was estimated that each mosquito imbibed two to three microliters of blood ( data not shown ) . These blood-meal volumes were similar to those previously published for various mosquito species , including Ae . aegypti [53]–[56] . The survival rates following the blood-meal were calculated by recording the number of dead mosquitoes daily . Survival rates for each virus were above 98% , which suggests that neither wtBUNV nor rBUNdelNSs2 had any detrimental effects on mosquito viability . At various times post feeding , 8 to 10 female mosquitoes were collected , homogenised and the levels of infectious virus they contained were titrated by plaque assay . Infection rates were calculated by dividing the number of infected female mosquitoes by the total number of engorged mosquitoes tested at a given day post-feeding ( Figure 7 ) . For wtBUNV , >70% of mosquitoes contained infectious virus at all time points . In contrast , at day 1 after feeding , only 20% of mosquitoes fed rBUNdelNSs2 showed evidence of infectious virus . However , by 5 days post-feeding , virus was detected in 60% of mosquitoes , and this rose to 80% by day 9 ( Figure 7 ) . Thus the lack of NSs seemed to delay the progress of infection . To confirm that lack of NSs results in delayed BUNV replication , in another experiment , midguts and salivary glands were dissected from engorged mosquitoes and examined for presence of virus . At days 1 to 13 post-feeding , nine mosquitoes , and at days 15 to 21 post-feeding , six mosquitoes , were collected , and pools of three isolated organs were made . Infectious virus could be detected in the midgut by 2 days post-infection for both viruses . By 4 days post-feeding , midgut infection rates ( calculated as percentage of virus positive midguts among engorged mosquitoes tested ) for wtBUNV reached about 80% , and stayed at high levels for the duration of the experiment ( Figure 8A ) . In comparison infection rates by rBUNdelNSs2 were significantly lower , and more variable . The titres of virus in midguts were also different between the viruses , with wtBUNV titres being about 100-fold higher than those of rBUNdelNSs2 ( Figure 8B ) . In order to be successfully transmitted by a mosquito vector to a vertebrate host , following replication in the midgut the virus must disseminate to secondary tissues like muscles , haemolymph , fat body and eventually the salivary glands . If the virus is found only in the midgut , infection is regarded as non-disseminated . Therefore the transmission potential of BUNV in infected mosquitoes was estimated by calculating disseminated infection rates to salivary glands . Both viruses managed to disseminate to salivary glands successfully , though with different kinetics . wtBUNV was detected in salivary glands by 3 days after feeding whereas the NSs-deletion mutant was not detected in salivary glands until 6 days post-feeding ( Figure 7C ) . The titres of wtBUNV were generally higher in salivary glands than those of rBUNdelNSs2 ( Figure 7D ) . These data suggest that the virus lacking NSs had more difficulty in overcoming the cellular defences in the midgut , but when rBUNdelNSs2 did manage to overcome the midgut escape barrier it was capable of spreading throughout the rest of the body , including to salivary glands .
The BUNV NSs protein has been widely studied in mammalian cells where it is has been shown to be a major virulence determinant . NSs counteracts the host innate immune response mainly by globally inhibiting RNA polymerase II-mediated transcription [16] , [18]–[21] . On the other hand , BUNV NSs does not affect cellular transcription in infected mosquito cells [17] , and for the related La Crosse orthobunyavirus ( LACV ) , no specific function for its NSs protein was found in mosquito cell lines [23] . However , our results presented above show that the BUNV NSs protein could be a crucial factor for efficient infection in certain cultured mosquito cells and in live mosquitoes . Comparison of wtBUNV and rBUNdelNSs2 showed that NSs is nonessential for replication and establishment of persistent infection in Ae . albopictus C6/36 and C7-10 cells . By contrast , rBUNdelNSs2 seemed unable to replicate productively in neither the Ae . albopictus U4 . 4 cell line nor the Ae . aegypti Ae cell line , indicating a requirement for the NSs protein . Unfortunately , attempts to express NSs exogenously in U4 . 4 cells , and thus enable rBUNdelNSs2 replication , have so far been unsuccessful . Comparison of the levels of released wtBUNV and the recombinant NSs deletion mutant suggests that NSs protein enables high-level virus replication in all cells except Ae . albopictus C7-10 , where expression or not of NSs had little effect . It is becoming clearer that , much like with mammalian cell lines , different mosquito lines differ in their response to viral infection and their ability to reproduce accurately events in whole mosquitoes [31] , [57] , [58] . Our data suggest that of the Ae . albopictus lines , the U4 . 4 cell line could similarly be a good tissue culture model to study BUNV replication and the role of NSs in infection of mosquito cells . The three phases of infection , early , acute and late , described by Lopez-Montero and Risco [27] in BUNV-infected C6/36 cells were also identified in the other two Ae . albopictus cell lines . These phases were characterized by changes in the location of viral proteins and changes of the cell morphology throughout each stage . Microscopic observations of the wild type and NSs-deleted viruses showed that lack of NSs reduced the extent to which mosquito cells underwent morphological changes during the acute stage of infection . Possibly these changes are driven by a defense mechanism that allows the cells to cope with severe viral infection , and Lopez-Montero and Risco [27] suggest that the filopodia-like projections could be involved in spreading “protective signals” among the cells . Branch-like projections have also been observed in cells infected with a rodent-transmitted bunyavirus , Sin Nombre hantavirus , and suggested to be sites where progeny particles were released [45] , [59] . A release method that does not involve rupturing the cell membrane could explain why virus replication does not kill mosquito cells and persistence is maintained . To date the only mosquito-borne bunyavirus that was shown to induce RNAi response in mosquito cells is La Crosse orthobunyavirus [23] , [37] , [39] . We have also detected , by conventional Northern blotting analysis , virus-specific small RNAs ( <30 nucleotides ) in all Ae . albopictus cell lines infected with wtBUNV ( data not shown ) . Soldan et al . ( 2005 ) showed that La Crosse virus replication could be inhibited in C6/36 cells pre-treated with virus-specific small interfering RNAs [60] . Here we showed that an RNAi response could be efficient in inhibiting BUNV infection too . When we transfected cells with long virus-specific dsRNA , BUNV replication was only reduced in U4 . 4 cells , which have previously been shown to have fully functional Dicer 2 [37] , [39] , suggesting that the dsRNA was processed efficiently to generate small inhibitory RNAs . Bunyaviruses efficiently avoid dsRNA-based RNAi responses by coating their RNA segments with the nucleoprotein , thereby avoiding the formation of dsRNA species [61] . Our transfection experiments showed that if specific dsRNA species were produced in abundance , mosquito cells could overcome wtBUNV infection . Interestingly , the BUNV NSs deletion mutant was capable of efficient replication only in Dicer 2 incompetent cell lines . There are no studies showing involvement of any bunyavirus NSs protein in overcoming RNAi response in mosquito cells , but the NSs protein of La Crosse virus has been shown to inhibit RNAi antiviral activity in mammalian cells [60] . Further work is required to investigate whether BUNV NSs has an effect on mosquito Dicer 2 activity , or if exogenous expression of NSs in U4 . 4 cells would render them permissive for rBUNdelNSs2 replication . The lack of genomic sequence data for the Ae . albopictus mosquito makes it a less attractive model in which to study host-virus interactions . Therefore , we investigated whether cells derived from Ae . aegypti , whose sequence has been determined [48] , were permissive for BUNV replication . Our results showed that BUNV growth in Ae cells resembled that in U4 . 4 cells , and that the NSs protein also proved to be necessary for efficient replication . Thus Ae . aegypti cells could be a useful tool in studying BUNV infection and identification of cellular components that are important for viral replication . There is only one study of BUNV replication in mosquitoes; Peers [52] reported that BUNV multiplied in the gut , disseminated to salivary glands and was transmitted to suckling mice in Ae . aegypti mosquitoes more efficiently than in Ae . vexans and Ae . canadensis . Our results for wtBUNV showed similar kinetics of viral replication and dissemination in Ae . aegypti to those obtained by Peers . Fewer mosquitoes were infected with the NSs-deletion mutant . In addition , we showed that the NSs protein contributes to high-level virus replication in that the mutant virus lacking NSs grew to lower titres . Similarly , the wild-type virus disseminated to salivary glands more efficiently than rBUNdelNSs2 , and the lower levels of rBUNdelNSs2 in salivary glands could affect the transmission potential of the virus . This requires further investigation . Our experiments showed that BUNV replication in Ae . aegypti mosquitoes resembled replication in Ae . aegypti Ae cells , as well as Ae . albopictus U4 . 4 cells . These data corroborate previous conclusions that these cell lines are the most appropriate mosquito cell culture models to study arbovirus infection . The NSs protein was required for efficient replication in both mosquito cells with a competent Dicer 2-RNAi system and in adult mosquitoes . In a proportion of mosquitoes , however , the mutant virus could eventually overcome host defences , though remained constrained as evidenced by lower virus titres . As BUNV is a relatively fast growing virus perhaps its reproduction rate is able to counteract the host's inhibitory responses . The NSs protein of Rift Valley fever phlebovirus ( also in the family Bunyaviridae ) , although being quite distinct in size , amino acid sequence and expression strategy from BUNV NSs , plays a similar role in mammalian cells in overcoming innate immune responses via global shut-down of cellular transcription [62] . Two recent papers investigated the role of Rift Valley fever virus NSs in infection of mosquitoes , and neither observed any difference in infection or dissemination rates between wt and NSs-deleted viruses [63] , [64] . Interestingly , deletion of another non-structural protein , NSm , from Rift Valley fever virus almost completely abolished its ability to replicate in mosquitoes [64] . These results illustrate the diversity and complexity of virus-host interactions within the Bunyaviridae family . | Bunyamwera and serologically related viruses are widely distributed in tropical and sub-tropical regions , and cause febrile illness in man . The viruses possess a trisegmented genome and can evolve by genetic reassortment generating viruses with different pathogenicity , like Ngari virus , a reassortant between Bunyamwera and Batai viruses , which causes haemorrhagic fever in humans . Like other arthropod-transmitted viruses , Bunyamwera virus can replicate efficiently in both mosquito and mammalian cells . Infected mammalian cells are killed by the virus whereas mosquito cells become persistently infected . Understanding the molecular basis for this difference may be crucial in designing new approaches to control bunyavirus disease . The viral non-structural NSs protein is the major virulence factor , which counteracts the innate immune defences of mammalian cells . In contrast , the role of this protein during infection of vector mosquito cells is unknown . We compared the replication of wild type virus and a genetically engineered virus that does not express NSs in various cultured mosquito cell lines and in Aedes aegypti mosquitoes . We showed that some cells did not support mutant virus replication , implying a role for the NSs protein . NSs protein was also important for efficient replication and dissemination in potential vector species . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"virology",
"biology",
"microbiology"
] | 2012 | Role of Bunyamwera Orthobunyavirus NSs Protein in Infection of Mosquito Cells |
Comparative research on food web structure has revealed generalities in trophic organization , produced simple models , and allowed assessment of robustness to species loss . These studies have mostly focused on free-living species . Recent research has suggested that inclusion of parasites alters structure . We assess whether such changes in network structure result from unique roles and traits of parasites or from changes to diversity and complexity . We analyzed seven highly resolved food webs that include metazoan parasite data . Our analyses show that adding parasites usually increases link density and connectance ( simple measures of complexity ) , particularly when including concomitant links ( links from predators to parasites of their prey ) . However , we clarify prior claims that parasites “dominate” food web links . Although parasites can be involved in a majority of links , in most cases classic predation links outnumber classic parasitism links . Regarding network structure , observed changes in degree distributions , 14 commonly studied metrics , and link probabilities are consistent with scale-dependent changes in structure associated with changes in diversity and complexity . Parasite and free-living species thus have similar effects on these aspects of structure . However , two changes point to unique roles of parasites . First , adding parasites and concomitant links strongly alters the frequency of most motifs of interactions among three taxa , reflecting parasites' roles as resources for predators of their hosts , driven by trophic intimacy with their hosts . Second , compared to free-living consumers , many parasites' feeding niches appear broader and less contiguous , which may reflect complex life cycles and small body sizes . This study provides new insights about generic versus unique impacts of parasites on food web structure , extends the generality of food web theory , gives a more rigorous framework for assessing the impact of any species on trophic organization , identifies limitations of current food web models , and provides direction for future structural and dynamical models .
Ecological network research is a powerful framework for assessing ecosystem organization , dynamics , stability , and function , topics that are central to ecology [1]–[7] . For example , comparative studies of food web structure have revealed regularities in how consumer–resource interactions ( Box 1 ) among species are organized [8]–[12] , produced successful simple models to characterize such structure [13]–[16] , and supported research on the robustness ( Box 1 ) of food webs to species loss [17]–[20] . These and other insights , however , have been largely based on analyses of interactions among free-living species , and have generally neglected parasites . Parasites comprise a significant part of the earth's biodiversity [21] , can achieve substantial biomass in some ecosystems [22] , can have similar abundance and productivity to free-living species of comparable body size and trophic level [23] , and likely extend the generality of the metabolic theory of ecology [24] . Further , in terms of their trophic relations , parasites have consumer–resource body-size ratios inverse to those of most free-living predators [23] , which enhances their ability to regulate host species abundances [25]; they have durable physical intimacy with their hosts [26]; they often have complex life cycles , sometimes requiring multiple phylogenetically distant hosts of widely varying body sizes over a lifetime [27]; they may have different patterns of trophic specialization than free-living predators [28]; they may differentially associate with hosts in different topological positions in food webs [29] , [30]; and their manipulation of hosts can reorganize communities and alter ecosystem function [31] . These and other ecological factors might alter how parasites fit into , and affect the structure of , food webs compared to free-living organisms . For example , although some parasites appear to be trophic generalists ( Box 1 ) , when their hosts are aggregated over their whole life cycle , they are actually temporal serial specialists ( Box 1 ) , with particular hosts at particular life stages [32] . Taking this into account increases the likelihood that primary species loss will lead to secondary extinction of such parasites and also decreases the robustness of the food web in question [32]–[35] . In general , the great diversity and unique habits and roles of parasites suggest that their explicit inclusion in food webs may alter our understanding of species coexistence and ecosystem structure , stability , and function [35]–[40] . Consistent with these types of expectations , prior studies of the network structure of food webs that include parasites have suggested that adding parasites alters food web structure [41]–[49] . This type of thinking is rapidly becoming conventional wisdom , as evidenced by a statement in a 2013 paper in Trends in Ecology and Evolution that “recent advances have shown that native parasites dramatically alter food web structure” [50] . However , there are two problems with this assertion . First , prior studies of parasites in food webs do not distinguish between changes in diversity and complexity and changes to network structure ( Box 1 ) . In food web studies , measures of diversity , such as species richness ( S ) , and of complexity , such as link density ( links per species , L/S ) and connectance ( the proportion of possible links actually observed , C ) , provide simple ways to characterize the numbers of nodes and links in those networks ( Table 1 , Metrics 1–4 ) . However , in the general [51] and ecological [6] network literature , network structure refers to patterns of how links are distributed among nodes . As noted in a recent perspective in Science , “Network approaches to ecological research emphasize the pattern of interactions among species ( the way links are arranged within the network ) ” [6] . While adding parasites , or any species , to food webs necessarily increases the numbers of species and links and can alter link density and connectance [45] , such changes to diversity and complexity should not be characterized as changes in food web structure . Second , while adding parasites and their links generally does alter network structure properties , as noted by prior studies for a few metrics [41]–[49] , there is usually an assumption that such changes result from unique aspects of parasite biology . However , those studies did not account for generic structural effects of adding any type of species and their links to a food web . One of the key insights of the last dozen years of comparative food web research regards the scale dependence ( Box 1 ) of food web structure , which refers to the empirically well-supported hypothesis that most aspects of network structure change systematically with changes in the diversity and complexity of food webs , regardless of the identity of the species in the webs [52]–[56] . Thus , the overall hypothesis we test is whether changes to network structure arising from the addition of parasites to food webs are attributable to the unique trophic roles that parasites play in food webs , or , alternatively , are generic effects of adding any type of species and links to webs . We conducted comparative analyses of the structure of seven highly resolved food webs that include detailed metazoan parasite data [42] , [57]–[60] . The food webs are from coastal areas and include a variety of habitats including estuaries , salt marshes , tidal basins , and mudflats . We assessed many metrics of food web structure ( Table 1 , Metrics 6–22 ) as well as degree distributions ( Box 1 ) and motifs ( Box 1 ) , most of which have not been evaluated previously for food webs with parasites . To our knowledge , this is the broadest set of food web structure properties yet evaluated in a single study . Together they provide a wide range of ways to understand network structure , from system-level properties to types of taxa present in the system to local structure to the occurrence of specific links . We did not analyze robustness ( Box 1 ) [17] , [61] , as it has been explored extensively for food webs with parasites elsewhere [32]–[34] , including an analysis of the seven food webs studied here [35] . That literature includes the only other study known to us that sought to disentangle generic from unique effects of parasites on network structure , by analyzing “whether the reduction in food web robustness after the inclusion of parasitism is due to factors associated with the characteristics of parasites , or simply an inevitable artefact of the addition of new nodes and links to an existing network” [34] . By comparing models with similar species richness ( S ) and connectance ( C ) , that study showed that only those models that incorporated parasite life-cycle constraints resulted in substantial reductions in robustness as well as higher vulnerability of parasites to random species loss . Thus , the general finding of reduced robustness of food webs with parasites to species loss [32]–[35] was attributed to the complex life cycles of many parasites , rather than to generic changes in S and C [17] , [54] . We also used a model-based strategy to assess whether changes in food web properties due to the addition of parasites are attributable either to their unique trophic roles or to generic effects of adding any species . The MaxEnt model for degree distributions [62] , the niche model [12] , [13] , and the probabilistic niche model [63] , [64] ( see Box 1 for brief definitions of the three models ) incorporate scale dependence . In particular , the MaxEnt and niche models use S and C as input parameters , while the probabilistic niche model matches S and C of empirical webs . The scale dependence of structure implicit in those models has been corroborated by analyses that show that these and related models generate networks with structure similar to that observed in empirical food webs [13]–[16] , [62] , [64] . The current study uses these models as a normalization tool—they provide a way to meaningfully compare the structural properties of empirical webs with different numbers of species and links , and they have been critical in identifying generalities in food web structure across space and time [10] , [11] , [54] , [55] . In addition , these models display a fit to empirical data that is scale dependent , with decreasing model fit associated with food webs that have greater diversity and complexity . This second form of scale dependence of food web structure provides another way to assess whether parasites have generic or unique impacts on structure . To summarize , our study improves on prior studies in the following ways: it distinguishes changes in diversity and complexity from changes in network structure; it accounts for the generic effects of the addition of species and links on food web structure; it examines a wide range of local to system-level structural properties; it uses trophic species aggregation ( Box 1 ) [65] , which is a necessary step for model-based comparative analysis [10]–[16]; it considers the role of concomitant links ( Box 1 ) , the numerous trophic links that occur when a predator concurrently eats parasites infecting its prey [38] , [47] , [66]; and it analyzes seven highly resolved webs , compared to the one to five webs of previous studies , some of which lacked high resolution and/or comprehensiveness . Our results underpin a more comprehensive assessment than previously undertaken of whether adding parasites alters food web structure in unique ways and whether parasites play similar or different roles compared to other consumers and resources in ecological networks . Teasing apart the generic effects of increased diversity and complexity on observed food web structure from the specific effects of the unique topological roles of parasites , or other types of organisms not considered here , is an important and necessary step for developing a fundamental understanding of ecological networks that includes a more detailed accounting of the full diversity of ecosystems .
We analyzed three versions of each web , one without parasites , one with parasites but no concomitant links ( Box 1 ) , and one with parasites and concomitant links . Each original species web version was aggregated into a trophic species web ( Box 1 ) , used as the basis for comparative network structure analyses . Species richness ( S; Table 1 , Metric 1 ) of the seven trophic species webs without parasites ranged from 56 to 117 ( Table 2 ) . The number of trophic links ( L; Table 1 , Metric 2 ) in the webs ranged from 358 to 1 , 085 ( Table 2 ) . Adding parasites increased S 1 . 2 to 1 . 9 times ( range of 109 to 185 ) and L 1 . 4 to 3 . 4 times ( range of 576 to 2 , 838 ) , while adding concomitant links increased L 1 . 8 to 5 . 7 times ( range of 1 , 252 to 4 , 671 ) . S was reduced by seven to 33% and L by four to 51% in trophic species webs compared to original species webs ( Table S1 ) . The majority of the metazoan parasites ( 72% to 100% ) in the original species webs have complex life cycles , where the parasites use two or more sequential hosts [27] . Those trophic shifts are often accompanied by an abrupt ontogenetic change in parasite morphology [67] . The use of sequential hosts by many of the metazoan parasites in these webs contrasts with the high degree of trophic specialization ( i . e . , only one host ) reported for parasitoids in other ecological networks [68] , [69] . In addition , the current webs have a large number of trematode parasites that tend to have relatively low specificity for the final host . Parasites comprised 15%–28% of taxa and were involved in 22%–74% of links , while free-living species were involved in 91%–100% of links in trophic species webs ( Table S2 ) , similar to original species webs ( Table S3 ) . Links can be divided into four categories based on the different possible relationships between free-living species ( FL ) and parasite species ( Par ) : classic predation ( FL-FL ) , classic parasitism ( Par-FL ) , parasites consuming parasites ( Par-Par ) , and predation of parasites ( FL-Par ) ( Table S2 ) . In trophic species webs with parasites , classic predation comprised 42%–78% of links , classic parasitism comprised 13%–38% , parasites consuming parasites comprised <10% , and predation of parasites comprised 0%–21% . Adding concomitant links decreased the shares of classic predation ( 26%–60% ) and classic parasitism ( 1%–23% ) , barely altered parasites consuming parasites ( <10% ) , and greatly increased predation of parasites ( 27%–52% ) . The number of classic predation links exceeded classic parasitism links except in the trophic species version of the Bahia Falsa web . The diversity of parasites of prey of free-living consumers resulted in predation-of-parasite links exceeding classic predation links in five of the seven webs with concomitant links . The addition of parasites usually increased link density ( L/S ) and connectance ( C ) ( Table 1 , Metrics 3 and 4 ) , and adding concomitant links resulted in further obligatory increases in L/S and C ( Tables 2 and S1 ) . The inclusion or exclusion of concomitant links changes the appropriate connectance measure to consider [45] . In webs that include concomitant links , the conventionally used “directed connectance” ( C = L/S2 ) is the appropriate measure , as it allows for the possibility of any link occurring between any two taxa [70] . In webs that exclude concomitant links , an “adjusted connectance” ( Cadj = L/ ( F•S ) , where F is the number of free-living species ) is the better measure ( Table 1 , Metric 5 ) , as it accounts for the exclusion of links from free-living to parasite species , as discussed in detail elsewhere [45] . Example images of the Estero de Punta Banda trophic species food webs show how diversity and complexity increased as parasites and concomitant links were added to the food web ( Figure 1 ) . Degree distributions , the distribution of the number of links associated with each node , are a commonly studied feature of networks of all types [51] . For a given food web it is most useful to report separate resource and consumer distributions [10] . Resource distributions give the pattern of numbers of links each species has to its prey or host species , and thus describe the balance of trophic specialization and generality ( Box 1 ) in an ecosystem . Consumer distributions give the pattern of numbers of links each species has to its predator species , and thus describe the balance of trophic vulnerability and invulnerability ( Box 1 ) in an ecosystem . Most extant food webs studied thus far have cumulative degree distributions that map closely onto universal exponential-type scaling functions once data are normalized for link density ( L/S ) [8] , [10] . The exponential shape indicates that the distribution of links in food webs is skewed across taxa [8] , [10]—for example , most taxa are specialists ( Box 1 ) that have one or a very few resources , while a few are generalists ( Box 1 ) that have many resources [10] . The normalized cumulative degree distributions for resource ( Figure S1 ) and consumer ( Figure S2 ) links for the three versions of the seven webs studied here , with and without parasites , followed similar curves , with exponential-type shapes similar to those of previously studied webs [10] . The most variability appeared in the tails of consumer distributions , but the effect of adding parasites or concomitant links did not follow any particular pattern ( Figure S2 ) . A more rigorous way to compare the shapes of these distributions , and to determine whether adding parasites alters the patterns of skewness of generality and vulnerability ( Box 1 ) in food webs , is to assess to what degree they differ from the expectations of a null model , in this case , a MaxEnt model ( Box 1 ) . MaxEnt is a non-mechanistic statistical approach that predicts the most likely distribution of some property given known constraints on information about the system . It has been used successfully to predict various macroecological patterns [71] . When applied to food web degree distributions , MaxEnt produces distributions with an exponential shape similar to what has been observed previously in empirical food webs [62] . It provides a more ecologically realistic null scenario for evaluating and comparing food web degree distributions than models that distribute links randomly [72] and does not assume an exponential distribution like the niche model ( Box 1 ) does [13] . Among the 21 current web versions , nine consumer distributions were significantly narrower , or less skewed , than MaxEnt expectations , in particular in webs with parasites , with or without concomitant links ( Table S4 ) . This means that in those nine food webs , the most vulnerable taxa ( those consumed by the most species ) had fewer consumers than expected compared to the most vulnerable taxa in the other 12 webs , whose consumer distributions did not differ from the MaxEnt expectation . Only one resource distribution , for the Flensburg Fjord web with both parasites and concomitant links , was significantly different ( wider ) than the MaxEnt expectation , meaning that its most generalist consumers fed on more species than expected compared to the other webs . Eight consumer and seven resource distributions were well fit by the MaxEnt model in terms of both the goodness of fit of the model ƒG and the expected width of the distribution W95 ( Table 1 , Metrics 20 and 21 ) . Only two web versions ( of the Ythan Estuary web ) had both consumer and resource distributions well fit by the MaxEnt model . To evaluate whether the significantly narrower than expected consumer distributions for many webs with parasites were likely a result of the unique roles of parasites versus a result of scale dependence ( Box 1 ) of network structure , we investigated a previously reported relationship between the width of the consumer distribution ( W95 Cons ) and L/S [62] . We combined the seven current webs without parasites with 28 prior food webs ( Table S5; Methods S1 ) and found a significant decrease of W95 Cons with C and a marginally significant decrease with L/S ( Figure 2; Table 3 ) . When results for webs with parasites were added , they were consistent with the observed scale dependence of W95 Cons with L/S ( Figure 2A ) , but fell below the scale dependence trend for C ( Figure 2B ) . However , several previously studied webs without parasites also fell in a similar space below the trend line . In terms of 14 commonly studied network structure properties that have well-documented ecological meaning and associated bodies of research ( Table 1 , Metrics 6–19 ) , the niche model ( Box 1 ) [13] fit the webs relatively poorly , especially when parasites were added . Model errors ( MEs ) for properties related to types of taxa ( Table 1 , Metrics 6–12 ) show that for one-third or more of the 21 webs the niche model significantly underestimated the fractions of taxa that are top species , that are herbivores , and that occur in loops , and significantly overestimated the fractions of basal taxa , omnivores , and cannibals ( Table S6 ) . For other web properties the niche model often significantly underestimated the variability in the number of links per species and the number of consumers per species , as well as mean trophic level ( Table S7 ) . It generally overestimated the mean maximum trophic similarity of pairs of species ( Table S7 ) . Across all 14 properties , webs without parasites had the most properties well fit by the niche model ( mean = 8 . 14 ) , compared to webs with parasites ( mean = 4 . 86 ) and webs with parasites and concomitant links ( mean = 6 . 14 ) . However , the reduced fit of the niche model in webs with parasites compared to webs without parasites appears consistent with scale dependence of model fit . When the current seven web versions lacking parasites were combined with ten previously studied webs ( Table S5; Methods S1 ) , there was a significant increase in mean absolute ME with S and a marginally significant increase with L ( Table 3; Figure 3A ) , consistent with prior results [12] . Niche model results for webs with parasites were consistent with the observed scale dependence of mean absolute niche ME with S for webs without parasites ( Figure 3A ) . In other words , as species richness increases , the fit of the niche model decreases , and there is no evidence that webs with parasites deviate from this trend . For three-node motif ( Box 1 ) representation—the frequency with which every possible pattern ( 13 in total ) of interactions among three species occurs in a web relative to its frequency in randomized webs—the seven food webs without parasites showed patterns similar to the typical pattern exhibited across most previously analyzed food webs and in the niche model ( Figures 4A and S3A ) [11] . The most notable differences were underrepresentation of omnivory ( motif S2 ) and overrepresentation of exploitative and apparent competition ( motifs S4 and S5 ) . These deviations , however , were also observed in a few previously studied food webs [11] . Adding parasite links resulted in a similar overall pattern ( Figure 4B ) . This result suggests that interactions involving parasites were distributed across motifs in a manner similar to that of interactions involving free-living species , as confirmed by the results of the compartmented randomization ( Figure S3B ) . However , the addition of concomitant predator–parasite links substantially changed the motif pattern ( Figure 4C ) . These changes were most pronounced in motifs D1 to D8 and indicate that bidirectional interactions made up of one parasite–host interaction and one concomitant link are distributed differently across motifs involving free-living species links and appear far more frequently in some motifs than in others . This observation was confirmed by marked differences between patterns of motif representation when webs with concomitant links were compared across the standard and compartmented randomizations ( Figures 4C and S3C ) . In the compartmented randomization , the addition of concomitant links also changed the over- and under-representation of motifs S1 to S5 to a pattern inconsistent with all empirical webs previously studied [11] , as well as the currently studied webs without parasites and webs with parasites but not concomitant links . These results suggest that patterns of prey selection in food webs were altered by the addition of parasites and concomitant links from predators to the parasites of their prey [11] , as a result of the trophic intimacy of parasites with their hosts . A recently proposed probabilistic niche model ( Box 1 ) uses maximum likelihood methods to parameterize the niche model directly against food web data [63] , [64] . It returns parameter estimates for each species in a web , and relaxes niche model assumptions about parameter distributions and hierarchical ordering of taxa . It also provides a probability of each link occurring , which can be compared to the actual links observed . A one-dimensional probabilistic niche model correctly predicted 0 . 601 to 0 . 756 ( mean ƒL = 0 . 654 ) of links for webs without parasites , 0 . 516 to 0 . 631 ( mean ƒL = 0 . 577 ) of links for webs with parasites but no concomitant links , and 0 . 555 to 0 . 657 ( mean ƒL = 0 . 596 ) of links for webs with parasites and concomitant links ( Table S8 ) . In each of the seven empirical food webs , ƒL was ∼10%–20% greater for webs without parasites than for webs with parasites , indicating a significantly lower ƒL in webs with parasites ( binomial test , seven of seven food webs , p = 0 . 0156 ) . In most cases , ƒL was similar for webs with parasites with or without concomitant links . A two-dimensional probabilistic niche model resulted in greater ƒL for all 21 web versions , ranging from 0 . 624 to 0 . 927 , with means of 0 . 801 , 0 . 737 , and 0 . 758 for webs without parasites , with parasites , and with parasites and concomitant links , respectively . Decreases in Akaike Information Criterion values indicated that the two-dimensional model performed better than the one-dimensional model for all 21 web versions ( Table S8 ) . However , the decrease in the fraction of links correctly predicted by the probabilistic niche model from webs without parasites to webs with parasites appears consistent with scale dependence of model fit . When the current seven webs without parasites were added to 28 previously studied webs ( Table S5; Methods S1 ) , ƒL significantly decreased with both increasing numbers of species ( S ) and links ( L ) ( Figure 3B and 3C; Table 3 ) , consistent with prior results [64] . The results for the current webs with parasites with or without concomitant links were consistent with the observed decrease of ƒL with increasing S ( Figure 3B ) . For webs with >1 , 500 links ( i . e . , most of the webs that include parasites ) , a minimum ƒL of ∼0 . 50 appeared to hold ( Figure 3C ) . A possible lower bound on ƒL in relation to L was suggested in an earlier study [64] . Using maximum likelihood estimates ( MLEs ) of niche model parameters , we ordered consumers by the position of their feeding range ( ci ) along the x-axis in Figure 5 , with their resources ordered by their niche value ( ni ) along the y-axis , and then marked documented links at the intersection of consumers and resources . This provides visualization of whether the resources of generalists tend to be dispersed along the niche axis or are concentrated with a near-contiguous core ( referred to hereafter as “trophic niche structure” ) , and whether parasite feeding ranges tend to clump or disperse along the niche axis ( Figure 5 ) . The trophic niche structure of generalists in the web without parasites showed that their resources' most likely niche values tended to arrange in a nearly contiguous core interval of niche space ( Figure 5A ) , with gaps ( i . e . , discontinuities in a column of links ) occurring more frequently towards the edges of the consumer's trophic niche , consistent with previously studied webs [64] . When parasites were added , the most likely feeding range positions of most parasites tended to group together ( Figure 5B ) . The parasites with multiple hosts also displayed a core trophic niche structure , but compared to those of generalist free-living consumers , parasites' links to resources spread across a larger interval of niche space , there were more gaps in their trophic niches , and in some cases there appeared to be secondary trophic niches separated from the main trophic niche . When concomitant links were added ( Figure 5C ) , the parasites with multiple hosts displayed similar patterns , and the breadth of trophic niches of generalist free-living species expanded greatly but still appeared to have a single nearly contiguous core . All seven webs displayed qualitatively similar patterns ( Figures 5 and S4 , S5 , S6 ) .
Our analyses corroborate previous findings for how parasites alter diversity and complexity of food webs [45] . As occurs with the addition of any species to food webs , adding parasites to the trophic networks studied here increased the number of species ( S ) and links ( L ) , and also usually increased link density ( L/S ) . Increases in links and link density were especially dramatic with the inclusion of concomitant links , the numerous links from predators to the parasites of their prey . Adding parasites also increased connectance ( C ) in most of the food webs analyzed here , especially when concomitant links were included or when connectance was adjusted to account for the non-inclusion of those links [45] . However , our study offers clarification of a prior finding that parasites “dominate” food web links , based on a comparison of classic parasitism links to classic predation links in an earlier version of the Carpinteria Salt Marsh web [45] . For the current seven webs , classic predation links outnumbered classic parasitism links in most cases , including in the Carpinteria Salt Marsh web . Overall , parasites were sometimes involved in >50% of food web links , particularly as prey when concomitant links were included , but free-living taxa were always involved with >90% of links because the vast majority of parasite links included free-living species . Thus , strictly speaking ( and by necessity ) , free-living species are involved in more food web links than are parasites . However , parasites are involved in substantial fractions of food web links , and if excluded , datasets would often account for less than 50% of the links in a given food web . It is important to note that any particular observation of the proportions of types of taxa and links , and thus the relative “dominance” of particular types of taxa or links , can be strongly influenced by the levels of taxonomic and trophic resolution [70] and sampling intensity [68] , [73] , [74] of the ecological networks in question . For example , in the current seven food web datasets , free-living bacteria and protozoa are either absent or highly aggregated . However , parasitic bacteriophages and protozoa are also absent . When we consider that worldwide , ∼60 , 000 vertebrate species may host ∼300 , 000 parasite species [21] , undersampling likely leads to greater underestimates of parasites and their links than of free-living species . Prior studies have shown that variability in the raw values and distributions of network structure properties , as observed for food webs with and without parasites , often masks generalities in ecological network structure . Such generalities emerge only after appropriate normalization for diversity and complexity [8] , [10] , [53] . The MaxEnt , niche , and probabilistic niche models ( Box 1 ) are used in this study as tools that provide normalizations that allow comparison of the structure of webs with different numbers of species and links . These models have previously performed well , revealing generalities in the structure of food webs [10]–[13] , [54] , [62] , [64] . In this study , the models generally did a worse job describing the structure of food webs with parasites than food webs without parasites . This would seem to corroborate prior assertions that adding parasites alters food web structure in unique ways [41]–[48] . However , the webs with parasites in this study have species richness values of 109 to 185 , greater than that of most webs without parasites previously studied . Each of the models used to evaluate network structure in our study has known scale dependence with diversity and complexity , such that the fit of the models decreases in relation to S , L , L/S , or C of the empirical web being analyzed [12] , [62] , [64] . When the current seven webs without parasites are compared to prior webs that lack parasites , significant scale dependencies of model fit are corroborated and extended: the width of the consumer distribution narrows with C and L/S; the absolute mean niche ME increases with S and L; and the fraction of links correctly predicted by the probabilistic niche model decreases with S and L ( Table 3 ) . The network structure of webs with parasites is in most cases consistent with these scale dependencies observed in webs without parasites ( Figures 2 and 3 ) . This suggests that apparent differences in several commonly studied aspects of network structure for webs with and without parasites are not attributable to special topological roles that parasites might play in food webs . Instead , they appear to result from generic changes in network structure due to the increasing diversity and complexity of food webs when parasites are added . Specifically , we found that changes in consumer and resource distributions , 14 commonly studied food web metrics , food web motifs ( when concomitant links are excluded ) , and link probabilities are consistent with generic changes in food web structure associated with changes in diversity and complexity , regardless of species identity . Also , in prior work , relative nestedness , a measure of network structure not considered in the current analysis , was found to change very little with inclusion of parasites and classic parasitism links [45]–[47] , but it increased greatly with the further inclusion of concomitant links in the Carpinteria Salt Marsh web [45] . This change may be attributable to a positive relationship of nestedness with connectance [74] , [75] , which increases with the addition of concomitant links . This should be investigated more explicitly with regard to scale dependence in future research . Our findings suggest that many aspects of previously identified generalities in food web structure across habitats and deep time [10] , [11] , [54] , [55] likely extend from free-living species food webs to those that include parasite species . This is consistent with macroecological patterns showing that parasites and free-living species play by similar rules when it comes to the relationship between body size , abundance , and trophic level [23] , in addition to similarities observed in other aspects of the metabolic theory of ecology [24] . Our analyses do highlight some patterns that need clarification with more data in the future . Specifically , a possible lower bound on the fraction of links correctly predicted by the probabilistic niche model ( ƒL∼0 . 50 ) at ∼1 , 500 links , as suggested by webs with parasites , needs to be examined for other webs without parasites , but with high numbers of links . Also , the rate of decrease in the width of consumer distributions with increasing connectance needs to be clarified with additional data for webs with C>0 . 1 . In general , because the scale dependencies based on webs without parasites reflect ranges of species richness and numbers of links lower than those for webs with parasites , additional data for more diverse webs without parasites , as well as highly resolved webs with parasites from other habitats , will allow more rigorous assessment of the scale dependence of model fit and whether webs with parasites are as consistent with those trends as initially indicated by this study . This brings us to another important point—our analyses reveal limitations of current simple models of food web structure . The majority of webs used to evaluate network structure thus far generally have trophic species richness less than 100 . The simple models used here and elsewhere appear to fit the structure of food webs with S<100 reasonably well , but , as we show , that fit decays systematically with increased diversity and/or complexity of the food web [12] , [62] , [64] . Our results suggest that the availability of more diverse , comprehensive , and highly resolved data requires development and testing of new network structure models , and may require a shift from low- to higher-dimension approaches . Beyond generic scale-dependent effects of greater diversity and complexity on network structure and model fit when parasites are added , two of our analyses suggest that parasites play certain unique topological roles in these food webs . First , the addition of parasites with concomitant links resulted in large and consistent differences in motif representation compared to webs without parasites , webs with parasites but no concomitant links , and niche model webs , all of which had similar motif frequencies . This was especially the case for motifs that included at least one set of two-way ( bidirectional ) links between a pair of taxa . These results imply that , topologically , the roles of free-living species as prey are similar whether they are consumed only by free-living species or by parasites . However , the roles played by parasites as concomitant prey are substantially different from the roles played by free-living species as prey or hosts . This is attributable to the close physical intimacy of parasites with their hosts [26] , which ensures that parasites are also eaten when their host is eaten , something that is generally not the case for classic predator–prey interactions . Thus , inclusion of concomitant links increases the amount of intraguild predation , predation that occurs between taxa that feed on the same prey species [76] , [77] . However , it increases such predation only from predators to parasites , and not the reverse , and these patterns would be useful to quantify in future research . Second , analysis of the most likely trophic niche structure of species reveals some differences between parasites and free-living species . While most generalist consumer species , whether free-living or parasite , tend to have a core , near-contiguous trophic niche with gaps occurring more frequently towards the edges of the range [63] , [64] , the trophic niches of parasites tend to be broader and have more gaps , and in some cases parasites display a smaller , secondary trophic niche . Also , the positions of the trophic niches of parasites tend to group together and are not dispersed throughout the niches of free-living species . A contiguous or near-contiguous trophic niche is a central assumption of the niche and related models [13]–[16] , with near contiguity observed in empirical data [78] . The weakening of the near-contiguous trophic niche pattern for parasite species , including occasional secondary trophic niches , may result from the complex life cycles of many parasites [42] . Parasites can have multiple hosts that diverge from each other in a variety of ways such as body size and phylogeny , factors that are thought to be important for structuring food webs [15] , [79] , [80] . As an example , trematodes are a common parasite group in most of the webs we examined . They use mollusks as first intermediate hosts , fish and invertebrates as second intermediate hosts , and fishes and birds as final hosts [57]–[60] . The inability of the one-dimensional probabilistic niche model to assign a strong contiguous trophic niche to many parasites , and the fact that it tends to group parasites together , may also be related to body size . While free-living consumers are usually larger than their resources by one or more orders of magnitude [81] , parasites are smaller than their resources by similar orders of magnitude [82] , which may result in parasites' feeding being less restricted to contiguous ranges of body sizes . The single niche dimension embodies the concept of a hierarchical species ordering . Body size is a favored hypothesis for how taxa may be ordered [79] , but inclusion of parasites will disrupt any single-dimensional body-size-based ordering in a food web [23] , [42] . Even for webs without parasites , the importance of body size can vary substantially across webs [83] , [84] , and hierarchical ordering itself may often not apply [64] . Increases in intraguild predation and the inclusion of species that lack strongly contiguous , one-dimensional trophic niches should tend to drive food web structure away from niche model expectations . However , our findings suggest that such shifts may be dominated and masked by concurrent scale-dependent shifts in network structure . Future research could address how much additional intraguild predation as well as deviations from niche contiguity , both of which appear to be associated with parasites in food webs , are required to noticeably shift network structure patterns such as link distributions and structural metrics away from empirical and model expectations . Also , future work should focus on more quantitative assessment of patterns and relationships of probabilistic niche model parameter estimates . Such research could quantify differences in the contiguity of the trophic niches of parasites versus free-living predators in one and two dimensions , as well as differences in the contiguity of the trophic niches of free-living consumers with and without inclusion of concomitant links . These analyses would be one way to test the hypothesis presented here , that parasites tend to have more complex trophic niches than free-living taxa . Our work provides a framework for evaluating future claims that adding any particular type of species changes food web structure in unique ways . For example , protozoa , endosymbionts , bacteria , and viruses have yet to be adequately represented in food webs , and , like parasites , are small , can be cryptic , and can be subject to concomitant predation . Terrestrial insects and their interactions are thus far very poorly resolved in food webs , and primary producers are often aggregated . The impact of fixing any of these or other biases on ecological network structure has to be assessed relative to generic impacts of altering the diversity and complexity of food webs [29] , [54] , [55] . In addition , the impact of parasites on the network structure of terrestrial systems may be different from that observed in the coastal aquatic systems analyzed here if terrestrial parasites tend to play significantly different kinds of roles as resources and consumers in those systems compared to estuary or marine-based parasites . The current findings also have important implications for modeling . The inverse niche model was recently proposed for food webs with parasites [85] . This model assigns links between parasites and hosts by inverting two niche model rules [13] . First , the parasite's niche value ( ni ) and feeding range ( ri ) are assigned as usual , but the position of the feeding range ( ci ) is higher , rather than lower , than the parasite's ni , resulting in a reverse hierarchy for parasites . Second , the size of parasites' ri decreases , and thus specialization increases , as parasites' ni increases . The niche model's assumption of trophic niche contiguity still holds—parasites feed on all taxa in their feeding range . Free-living species follow the usual niche model rules . While this model , which treats parasites differently from free-living species , was not compared directly to a niche model that does not distinguish between parasites and non-parasites ( i . e . , the way the niche model was implemented for the current analyses ) , it did fit data for Carpinteria Salt Marsh better than various null models . The current results suggest that if parasites are treated differently in models , the assumption of contiguous parasite feeding niches should be altered to account for greater breadth , more gaps , and the occasional presence of secondary niches . Alternatively , focusing on life stages with distinct diets as nodes in food webs may resolve this issue . Also , the inverse niche model excluded parasite–parasite links and any consumption of parasites by free-living species . Food web data should document , and associated models should allow for , the potential occurrence of links between any two taxa , which then sets directed connectance ( C = L/S2 ) as the appropriate connectedness measure . In the webs studied here , there are instances of all types of interactions , including more uncommon links such as free-living species feeding on free-swimming parasitic stages . Producing an empirically well-supported model of the network structure of food webs with parasites and all types of links will also be important for dynamical modeling of parasites in food webs . Obvious questions are how parasites augment or inhibit the dynamical persistence and coexistence of species , and how parasites alter the likelihood of secondary extinctions given bottom-up , top-down , and indirect effects . For example , one approach to modeling food web dynamics starts by generating network structure with the niche model or a similar model and then implements nonlinear bioenergetic equations constrained by metabolic scaling and allometric relationships to model the biomass dynamics through time of each species in that network [86]–[89] . This approach needs to change when parasites are included to reflect the topological differences noted in this study , without violating the strong scale dependence of many features of food web structure . Other differences between parasite–host , predator–prey , and predator–parasite relationships will need to be integrated in future models , such as differences in consumer–resource body-size ratios , the role of host as both food and habitat for parasites , the role of concomitant links , the complex life cycles of parasites , and potential differences in biomass flow between predators and prey and parasites and hosts . Key emerging aspects of global change research include understanding how interactions among organisms mediate ecological function at multiple scales [5] , [7] , as well as understanding the dynamic relevance of the structural roles of species [90] . Given the diversity of parasites in every ecosystem and at every trophic level , future food web models used in global change studies need to better encompass the topology and dynamics of complex interactions among parasites and free-living species , while also taking account of well-supported scale dependencies of network structure and model fit .
We analyzed seven highly resolved coastal marine or estuarine food webs with detailed metazoan parasite data . Three North American Pacific coast webs were recently compiled by one research group [57]: Carpinteria Salt Marsh in California , US ( an earlier version was published in [45] ) ; Estero de Punta Banda in Baja California , Mexico; and Bahia Falsa in Bahia San Quintín , Baja California , Mexico . Three additional coastal webs in Europe and New Zealand were recently compiled by a second research group: Flensburg Fjord on the Baltic Sea between Germany and Denmark [58]; Sylt Tidal Basin on the North Sea between Germany and Denmark [59]; and Otago Harbor in Dunedin , New Zealand [60] . A seventh food web published in 1996 for the Ythan Estuary on the North Sea near Aberdeen , Scotland [42] , was also used , as it has a resolution of free-living taxa and metazoan parasites comparable to that of the other six webs . This set of seven webs with parasites has been analyzed in one other paper focused on the effects of including parasites in food webs on food web robustness [35] . We excluded from analysis two freshwater webs with parasites [46] , [47] because they have lower diversity and resolution . In general , the compilation of data for the seven webs used in this analysis made use of consistent methodologies for identifying links [91] . Individuals of free-living species sampled in each habitat were dissected to identify metazoan parasites . This approach was combined with a strategy that emphasized searching for more individuals of rare free-living species to reduce the bias towards underrepresentation of parasites of uncommon hosts . These directly sampled data were augmented with literature-based data for the particular sites or nearby sites , as well as with inferences based on current understanding of host and parasite biology . Another bias that leads to underestimation of parasite diversity is the non-identification of certain classes of parasites altogether . For example , in the seven webs analyzed here , bacteriophages and protozoans were either not identified or were under-identified . Both of these biases , underreporting rare taxa and failing to resolve or include whole groups of cryptic or small taxa ( e . g . , microbes ) , are a problem for both parasite and free-living taxa , but likely result in greater underestimation of parasite diversity , given the fact that most host taxa have more than one parasite species . The original seven datasets [42] , [57]–[60] included ontogenetic life stages of parasite species with complex life cycles as separate food web nodes . However , for our analysis we aggregated parasite life stages and their feeding links into a single parasite node and set of links [92] . While species-level analysis masks temporally distinct resource use by many parasite taxa whose juvenile and mature forms have different diets , comparative studies of food web structure generally use the species as the lowest level of resolution , and ontogenetic diet data are not yet available for most free-living species , some of which also undergo ontogenetic and trophic life-stage shifts . We analyzed data for three versions of each food web [92]: a free-living species web , a web with parasites but no concomitant links , and a web with parasites and concomitant links . Concomitant links were inferred by assuming predators eat all parasites of infected prey . All datasets except for Ythan Estuary also included some documentation of parasite–parasite links and targeted ( non-concomitant ) consumption of parasites by free-living species . We focused our analyses on the trophic species ( Box 1 ) versions of the 21 webs . For each web , we generated cumulative degree distributions ( Box 1 ) across species for the number of links from predators ( “consumer distribution” ) and links to prey or hosts ( “resource distribution” ) per node , normalizing the link counts by L/S for each web [8] , [10] . We tested the fit of a maximum information entropy MaxEnt model for food web degree distributions ( Box 1 ) [62] to empirical food web link distributions . MaxEnt models generate the least biased probability distributions by maximizing the information entropy for a system after applying information-containing constraints . For food web degree distributions , S and C serve as such constraints , and we included an additional constraint , the number of basal species for resource distributions and the number of top species for consumer distributions [62] . We tested the fit of MaxEnt predictions by calculating goodness of fit , ƒG , and relative width of the degree distribution , W95 ( Table 1 , Metrics 20 and 21 ) . ƒG≤0 . 95 indicates that the empirical web's link distribution does not differ significantly from the model distribution at the 95% confidence interval [62] . When −1≤W95≤1 , the empirical distribution is neither significantly narrower ( W95<−1 ) nor significantly broader ( W95>1 ) than the distribution predicted by the model at the 95% confidence interval . A distribution is considered well fit by a model when both criteria are met: ƒG≤0 . 95 and −1≤W95≤1 . We calculated link density ( L/S ) and directed connectance ( C = L/S2 ) for each web , as well as adjusted connectance ( Cadj = L/F•S ) ( Table 1 , Metrics 3–5 ) for webs with parasites but no concomitant links , to account for exclusion of such links in those web versions [45] . We calculated 14 network structure properties [12] , [55] for each web ( Table 1 , Metrics 6–19 ) : the fractions of top , intermediate , and basal species ( Top , Int , Bas ) ; the fractions of cannibals , herbivores , omnivores , and species in loops ( Can , Herb , Omn , Loop ) ; the standard deviations of normalized total links , generality , and vulnerability ( LinkSD , GenSD , and VulSD ) ; the mean short-weighted trophic level of all species ( TL ) ; the mean maximum trophic similarity of species ( MaxSim ) ; the mean shortest number of links between species pairs ( Path ) ; and the mean clustering coefficient ( Clus ) . We generated 1 , 000 niche model webs with the same S and C as the 21 webs , and for each property for each web , calculated ME , the normalized difference between the model's median value and the empirical value [12] . ME>|1| indicates that the empirical property falls outside the most likely 95% of model values , with negative and positive MEs indicating model underestimation and overestimation of the empirical value , respectively . We investigated over- and underrepresentation of the 13 unique motifs ( Box 1 ) that can occur among three species [11] . Motifs S1 to S5 include only single-directional links between taxa pairs , while motifs D1 to D8 include bidirectional links ( i . e . , mutual predation ) between at least one species pair . The frequency of a motif in an empirical food web was compared to the same in an ensemble of randomized webs , yielding a z-score for each motif i that measures the degree that the empirical web deviates from the null hypothesis . We used two randomizations: “standard , ” in which all links are shuffled , with the restriction that single-directional and bidirectional links are only shuffled with each other [11] , and “compartmented , ” which proceeds in the same fashion but with the additional restriction that links are shuffled only with those of the same type ( links between free-living taxa , between parasites and free-living hosts , etc . ) . For a given web , we quantified the motif structure with a vector of z-scores Z = {zi} , which has one component for each of the 13 three-species motifs . To compare webs , we plotted the normalized profile , the vector of z-scores normalized to length 1 . This aids in graphical comparison because larger and more densely connected webs tend to exhibit more pronounced patterns of motif representation . The occurrence of motifs in empirical webs was compared to niche model expectations . We used a probabilistic niche model ( Box 1 ) [63] , [64] based on maximum likelihood methods [16] to parameterize the niche model directly against each empirical food web . The probabilistic niche model tests the overall model fit to the data rather than to partial aspects of structure . It produces a MLE of the niche model parameters for each species i in a given web: its niche position ni , position of feeding range ci , and feeding range ( or “trophic niche” ) ri . This allows computation of the probability of each link in a web according to the model , and the overall expected fraction of links ( fL ) in a web predicted correctly by the model ( Table 1 , Metric 22 ) . The one-dimensional probabilistic niche model outperforms [64] other recently proposed structural models [15] , [16] . We calculated fL for one- and two-dimensional versions of the model and compared their performance for each web using the Akaike Information Criterion [93] . The MLE parameter sets were used to explore the trophic niche structure of parasite and free-living species . | Food webs are networks of feeding interactions among species . Although parasites comprise a large proportion of species diversity , they have generally been underrepresented in food web data and analyses . Previous analyses of the few datasets that contain parasites have indicated that their inclusion alters network structure . However , it is unclear whether those alterations were a result of unique roles that parasites play , or resulted from the changes in diversity and complexity that would happen when any type of species is added to a food web . In this study , we analyzed many aspects of the network structure of seven highly resolved coastal estuary or marine food webs with parasites . In most cases , we found that including parasites in the analysis results in generic changes to food web structure that would be expected with increased diversity and complexity . However , in terms of specific patterns of links in the food web ( “motifs” ) and the breadth and contiguity of feeding niches , parasites do appear to alter structure in ways that result from unique traits—in particular , their close physical intimacy with their hosts , their complex life cycles , and their small body sizes . Thus , this study disentangles unique from generic effects of parasites on food web organization , providing better understanding of similarities and differences between parasites and free-living species in their roles as consumers and resources . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"community",
"ecology",
"ecology",
"food",
"web",
"structure",
"coastal",
"ecology",
"biology",
"microbiology",
"marine",
"biology",
"parasitology"
] | 2013 | Parasites Affect Food Web Structure Primarily through Increased Diversity and Complexity |
Brucellosis is a highly contagious zoonosis affecting livestock and human beings . The human disease lacks pathognomonic symptoms and laboratory tests are essential for its diagnosis . However , most tests are difficult to implement in the areas and countries were brucellosis is endemic . Here , we compared the simple and cheap Rose Bengal Test ( RBT ) with serum agglutination , Coombs , competitive ELISA , Brucellacapt , lateral flow immunochromatography for IgM and IgG detection and immunoprecipitation with Brucella proteins . We tested 208 sera from patients with brucellosis proved by bacteriological isolation , 20 contacts with no brucellosis , and 1559 sera of persons with no recent contact or brucellosis symptoms . RBT was highly sensitive in acute and long evolution brucellosis cases and this related to its ability to detect IgM , IgG and IgA , to the absence of prozones , and to the agglutinating activity of blocking IgA at the pH of the test . RBT was also highly specific in the sera of persons with no contact with Brucella . No test in this study outperformed RBT , and none was fully satisfactory in distinguishing contacts from infected patients . When modified to test serum dilutions , a diagnostic titer >4 in RBT resulted in 87 . 4% sensitivity ( infected patients ) and 100% specificity ( contacts ) . We discuss the limitations of serological tests in the diagnosis of human brucellosis , particularly in the more chronic forms , and conclude that simplicity and affordability of RBT make it close to the ideal test for small and understaffed hospitals and laboratories .
Brucellosis is a highly contagious zoonosis caused by the Gram-negative bacteria of the genus Brucella . B . abortus , B . suis and B . melitensis , three of the so-called smooth ( S ) brucellae , preferentially infect cattle , swine and sheep and goats , respectively . These animals are the source of most cases of human brucellosis , a grave and debilitating disease that may leave disabling sequelae . Its incidence is very high in some countries of the Mediterranean basin and bordering areas and , in all likelihood , in developing countries throughout the world . The reported incidence in these countries varies widely ( from <0 . 01 to >200 per 100 , 000 ) , reflecting the difficulties in recognizing a disease that lacks pathognomonic symptoms [1] , [2] . This absence of specific symptoms makes it difficult to distinguish brucellosis from several febrile conditions that often occur in the same areas , including malaria [3]–[11] so that laboratory tests are essential for diagnosis [12] . Among these tests , only the isolation of the microorganism provides absolute proof of infection but bacteriological diagnosis is expensive and dangerous . On the other hand , serological tests are easier to implement and a great aid in diagnosis . The humoral immunoresponse to S brucellae is dominated by antibodies to the polysaccharide ( PS ) section of the Brucella S lipopolysaccharide ( S-LPS ) and it shows a typical IgM/IgG ( and IgA ) shift . In acute cases ( i . e . , short evolution ) IgM is present in the serum; then this immunoglobulin returns progressively to background levels , so that IgG ( and IgA ) are dominant in the sera of long evolution ( i . e . chronic ) patients before treatment . Moreover , non-agglutinating antibodies ( detected in the Coombs test ) increase over agglutinating antibodies ( active in the classical serum agglutination test [SAT] ) during the course of the infection [12] , [13] . Accordingly , the SAT-Coombs combination has been classically used both to increase sensitivity and to evaluate the stage of evolution of the infection . Other S-LPS ( or PS ) tests proposed more recently include the lateral flow immunochromatography assay ( LFiC ) for IgM and IgG assessment , a fluorescence polarization assay , a variety of indirect ELISA , and the immunocapture Brucellacapt test ( for a recent review , see [14] ) . In addition , a competitive ELISA ( cELISA ) has been proposed [15] . Because these tests require well equipped laboratories and/or adequate budgets , they cannot be implemented in many laboratories in endemic areas . The Rose Bengal test ( RBT ) is a rapid slide-type agglutination assay performed with a stained B . abortus suspension at pH 3 . 6–3 . 7 and plain serum . Because of its simplicity , it is often used as a screening test in human brucellosis and would be optimal for small laboratories with limited means . However , there is confusion about the value of this test so that present WHO guidelines recommend that RBT results be confirmed by other tests [14] , [16] . Points of concern expressed by several authors include low sensitivity [17] particularly in long evolution ( chronic ) cases [18] , [19] and relatively low specificity in endemic areas [20] , [21] . Moreover , some authors consider that prozones make strongly positive sera appear as negative in RBT [22] . In the present work , we have addressed these points and reexamined the usefulness of RBT for the diagnosis of human brucellosis using sera of patients with no brucellosis , culture-positive brucellosis patients , and healthy persons that had had contact with the pathogen .
The sera used in this work were obtained during clinical practice in the 1975–2001 period . Their use in this research was approved by the Ethical Boards of Clínica Universidad de Navarra ( Pamplona , Spain ) , Hospital de Navarra ( Pamplona , Spain ) , Hospital Universitari de Bellvitge ( Barcelona , Spain ) , Hospital Clínico Universitario Virgen de la Victoria ( Málaga , Spain ) , Hospital de la Inmaculada de Huércal-Overa ( Almería , Spain ) , Hospital General Universitario de Albacete ( Albacete , Spain ) , and Hospital Clínico Universitario de Valladolid ( Valladolid , Spain ) . For RBT , 30 µL of plain serum were dispensed on a white glossy ceramic tile and mixed with an equal volume of RBT antigen ( Veterinary Laboratory Agency; England , United Kingdom; http://www . defra . gov . uk/vla/ ) ( previously equilibrated at room temperature and shaken to resuspend any bacterial sediment ) using a toothpick . The tile was then rocked at room temperature for 8 minutes ( instead of the 4 minutes recommended for animal brucellosis [23] ) , and any visible agglutination and/or the appearance of a typical rim [23] ( Figure S1 ) was taken as a positive result . Positive sera were tested further as follows . Eight 30 µL drops of saline were dispensed on the tile and the first one mixed with an equal volume of the positive plain serum ( 1/2 serum dilution ) . Then , 30 µL of this first dilution were transferred to the second drop with the help of a micropipette and mixed to obtain the 1/4 dilution . From this , the 1/8 to 1/128 dilutions were obtained by successive transfers and mixings taking care of rinsing the pipette tip between transfers . Finally , each drop was tested with an equal volume ( 30 µL ) of the RBT reagent , so that the final dilutions ranged from 1/4 to 1/256 . The SAT and Coombs test in microtitter plates , Brucellacapt ( Vircell S . L , Santa Fe , Granada , Spain ) and LFiC ( kindly provided by Dr . H . Smits , KIT Biomedical Research , Royal Tropical Institute/Koninklijk Instituut voor de Tropen , Amsterdam , The Netherlands ) were performed as described before [24] , [25] . For the Coombs test , a titer ≥ two times the SAT titer in the same serum was considered as positive . In some cases ( see Results ) , SAT was also performed in the citrate buffer ( pH 5 ) provided as diluent in the Brucellacapt kit . To this end , the bacteria in a volume of the SAT suspension were collected by centrifugation , washed with citrate buffer and resuspended in an equal volume of the same buffer . cELISA was performed according to the instructions of the manufacturer ( Svanova Biotech , Uppsala , Sweden ) . Antibodies to Brucella proteins were detected by counterimmunoelectrophoresis ( CIEP ) using an S-LPS free extract obtained from a B . melitensis rough mutant [26] . The following groups of sera were used: ( i ) , two hundred and eight sera of an equal number of patients with brucellosis confirmed by bacteriological culture ( all B . melitensis ) that were diagnosed at the above-mentioned institutions in the 1975–2001 period; a subset of patients in this group for which the IgM and IgG profile could be determined ( by LFiC ) were classified as short ( IgM dominant ) or longer evolution ( IgG dominant with low or no IgM ) ( see Results ) and correspond broadly to the concepts of acute and chronic brucellosis; ( ii ) , the sera of 20 persons ( Table 1 in Supporting Information S1 ) that had had professional contact ( veterinarians , slaughter house workers , shepherds , etc . ) with B . melitensis-infected animals or their products or had accidentally injected themselves with vaccine B . melitensis Rev 1 and that were followed for a period of at least two years; ( iii ) , eleven sera from brucellosis patients that had been collected in a different study because they showed the prozone effect; and ( iv ) , one thousand five hundred and fifty-nine sera from patients with no symptoms of brucellosis sent to the laboratory for the serological diagnosis of other infections . The sera were aliquoted and kept frozen at −20°C . Care was taken not to thaw and freeze repeatedly these sera . Controls showed no deterioration under these conditions .
The 1559 sera from patients with no brucellosis yielded only one positive result in the standard RBT . The patient was asymptomatic and re-examination of the medical history showed that he had suffered from brucellosis in the past . The sera of 19 of the 20 persons that had had professional contact with B . melitensis-infected animals or had accidentally injected themselves with vaccine Rev 1 showed reactions in the standard RBT despite the fact that these persons were consistently asymptomatic . None of these sera , however , had a titer >1∶4 when tested in the modified RBT ( Table 1 in Supporting Information S1 ) . In one case ( C-20 , Table 1 in Supporting Information S1 ) , seroconversion was observed at the time when symptoms compatible with brucellosis developed , and this patient was successfully treated with antibiotics . Concerning other tests , 3 of these 20 persons had SAT titers equal to 160 , 8 had Brucellacapt titers ≥320 , 16 had a positive Coombs , 4 and 8 were LFiC-IgM and -IgG positive , respectively , and 5 showed antibodies to cytosolic proteins . Table 1 compares the results of SAT and RBT obtained with the sera of the 208 culture positive patients . Whereas 185 had SAT titers ≥160 , the standard RBT identified as positive all the 208 sera . When performed on serum dilutions , a RBT titer discriminating all healthy contacts ( ≥1∶8; previous paragraph ) would identify correctly 180 sera of the culture positive patients ( 176+4; Table 1; 87 . 4% sensitivity ) . A SAT titer similarly discriminating all healthy contacts ( >1∶160; previous paragraph ) would identify only160 of these patients ( 76 . 9% sensitivity ) . In those cases that could be studied with more detail , RBT titers varied from 4 to 256 in the sera with weak or negative Coombs and anti-S-LPS IgM but no IgG , and from 4 to 128 in the sera with a positive Coombs and anti-S-LPS IgG stronger than IgM ( Tables 2 and 3 in Supporting Information S1 ) . The 23 sera ( 208 -185 ) of culture positive patients with SAT titers <160 included 6 showing the blocking phenomenon . This is a rare event appearing in some prolonged brucellosis cases when non-agglutinating IgA are in amounts higher than other anti-S-LPS antibodies and it represents the extreme case of prozones [27]–[30] . Table 2 shows that neither RBT nor Brucellacapt were affected by the blocking phenomenon and that , as expected , SAT titers increased upon IgA removal . Since RBT and Brucellacapt have in common the use of acid buffers ( pH 3 . 65 and 5 . 0 , respectively ) , we hypothesized that an acid pH could promote agglutination and overcome the presence of blocking IgA . To test this , we substituted citrate buffer pH 5 . 0 for the saline in the standard SAT bacterial suspension ( see Material and Methods ) . Under these conditions , the blocking activity in SAT disappeared . To confirm that the use of an acid buffer removes the agglutination-inhibitory effect of IgA and accounts for the absence of prozones in RBT , we examined 11 sera showing 1/40 to 1/80 prozones . These prozones disappeared upon absorption with anti-IgA and were not observed in SAT at pH 5 , RBT or Brucellacapt . Of the 23 sera with SAT titers <160 , 2 ( n°s 1 and 4 , Table 2 ) and 17 ( n°s 7 to 23 , Table 3 ) had RBT titers ≤4 . Using these 23 sera and those of the 20 persons that had had professional contact with B . melitensis-infected animals or had accidentally injected themselves with vaccine Rev 1 ( see above ) , we examined whether other tests could complement RBT and discriminate the sera of infected patients from those of healthy contacts ( Table 1 in Supporting Information S1 ) . The numbers of false positives/false negatives were: CIEP-proteins , 5/4; LFiC-IgM , 4/13; LFiC-IgG , 12/5; Brucellacapt ( ≥1∶640 ) , 2/2; and Coombs ( ≥two times the corresponding SAT titer ) 18/1 , and cELISA ( cut-off at 30% inhibition ) 8/0 . Finally , 11 culture positive patients could be followed periodically . SAT , LFiC-IgM , LFiC-IgG and RBT became negative between months 1 and 16 after the beginning of a successful antibiotic treatment . However , 4 and 2 of these patients remained positive in Coombs , and cELISA , respectively , and 2 had Brucellacapt titers equal to 1∶320 . Although greatly diminished in titer and number of precipitin lines , 6 patients remained positive in CIEP ( a single precipitin line in all cases ) .
One of the early findings in brucellosis was the observation that the sera of infected individuals contained agglutinating antibodies that could be detected in SAT . This test was soon adapted to the more practical slide-agglutination format but this method was prone to false negative results because of prozones and blocking and non agglutinating antibodies [31] . We show here that RBT overcomes these three problems . Moreover , we confirm that it is highly sensitive and demonstrate that a simple adaptation to test serum dilutions improves its specificity and considerably reduces the need for additional serological tests . This simple modification makes RBT close to the ideal test for small laboratories . Although the overall sensitivity reported for RBT varies widely , there could be several reasons for this . Variations in sensitivity have been demonstrated in the past for RBT antigens of various sources [32] , [33] and the use of good quality antigens made by experienced or reference laboratories is of the utmost importance . Although this has been occasionally considered as a weakness of RBT [14] , it is well know that a good quality control is necessary in all brucellosis serological tests because of the tendency of S brucellae to dissociate into rough variants lacking the diagnostically significant S-LPS epitopes [23] . Also , the use of white opaque glossy surfaces is important [34] , and awareness of the various agglutination patterns ( Figure S1a proper incubation time is critical . With regard to the latter , the literature shows from 2 to 5 minutes [20] , [34] , [35] . However , it has been known for a long time [36] that some human sera require a longer incubation to become positive in the RBT-like card test . In our experience , sera without blocking antibodies or prozones are strongly positive in less than 4 minutes , but sera with blocking IgA or with high titers of non-agglutinating antibodies ( high Coombs titers ) may need up to 8 minutes to develop the bacterial clumps or the characteristic rim . These antibodies are typical of long evolution brucellosis and , therefore , the low sensitivity ( 54 to 61% ) reported in chronic brucellosis by some authors [18] , [19] could be accounted for by a non optimized RBT protocol . Our results clearly show that RBT was equally useful in the IgM-negative ( longer evolution ) and IgM-positive ( shorter evolution ) groups of patients and that the use of an acidic pH abrogates prozones and blocking phenomena . Consistent with the demonstration that RBT detects both S-LPS specific IgM , IgG and IgA and that neither prozones nor blocking antibodies are sources of false negative results , most authors have reported a high sensitivity in culture-positive patients , equal or better than that of SAT , ELISA-IgG , ELISA-IgM , or LFiC for IgM plus IgG [20] , [37]–[39] . In this work , we have also used the sera of culture positive patients as the reference , and this point deserves attention for a correct understanding of our results in a clinical context . Careful studies with appropriate bacteriological procedures have shown that the rate of success in isolating Brucella is higher during the initial disease than in relapses ( c . a . 80 versus 65% , respectively , in ref . [40] ) and lowest in the more chronic forms [41] . The reasons for this consistent observation are not understood but , as illustrated for the case of hepatosplenic abscesses [25] , [41] , it is in a fraction of the more chronic cases where serology ( by RBT or other tests ) and culture are sometimes not conclusive . Indeed , the scarce RBT negative results that have been well documented correspond to a few patients with focal forms of brucellosis [25] , [41]–[43] . In these difficult cases , a combination of serological tests and clinical findings and a careful follow-up of the patients are in order . The evidence obtained in a limited number of these cases suggests that the Coombs test provides the best indication of the seroconversion that parallels the relapses and the evolution during treatment [25] , [44] ( see also below ) . The specificity of the RBT and other S-LPS tests is also worth commenting on . Febrile conditions including tuberculosis , malaria , typhoid fever , Still's disease , lupus erythematosus , rheumatoid arthritis , sarcoidosis , and active lymphoma are not a source of RBT false positive results [5] . On the other hand , S-LPS cross-reactivity with Vibrio cholerae , Francisella tularensis and Yersinia enterocolitica 0:9 is a potential source of unspecific results in all S-LPS tests . However , this is of little importance in clinical practice . Although positive cases have been reported in V . cholerae vaccinated individuals [45] , [46] , there are no RBT observations in cholera patients and this illustrates that the clinical picture is widely different . With regard to tularemia , in a series of 5 patients , 3 were RBT positive ( T . Marrodán , Ph . Thesis , University of Navarra , Spain ) but these were easily differentiated by the clinical picture and other tests . Yersiniosis by Y . enterocolitica O:9 elicits antibodies that react in all Brucella S-LPS tests including RBT [26] , [47] but there are tests with protein antigens that discriminate Y . enterocolitica O:9 and S Brucella infections [26] ( see also below ) . Indeed , antibodies to Brucella S-LPS persist for protracted periods in a percentage of recovered patients in all S-LPS tests [13] . Therefore , a past history of brucellosis is a cause of unspecific serological results that has to be evaluated by the physician . Finally , some authors consider that RBT has a limited usefulness in endemic areas [20] , [21] . However , Ruiz-Mesa et al . [48] compared the sera of individuals that had had repeated contact with Brucella with those with no regular exposure or history of brucellosis , and reported specificities of 91 . 7 and 94 . 3% . This same problem was addressed by Gómez et al . [39] who found 100 and 97% sensitivity and specificity , respectively . These studies indicate that the specificity problem of the standard RBT is not so critical and that , as illustrated by our results , other S-LPS tests are also affected . In this regard , it is important to stress that the diagnosis of human brucellosis has to be made on the basis of compatible symptoms , clinical findings and a thorough anamnesis , that it cannot rely exclusively on a weak positive result in any S-LPS serological test and that there are no cut-off diagnostic titers in any single S-LPS test . An alternative to S-LPS tests is the use of protein tests [12] . It has been known for a long time that a large proportion of brucellosis patients develop antibodies to soluble Brucella proteins [26] . However , in the present study , 5 of the 20 healthy persons that had had professional contact with infected livestock developed anti-protein antibodies ( Table 1 in Supporting Information S1 ) . Recently , a large number of Brucella proteins have been evaluated by Liang et al . [49] using brucellosis patients that were culture and RBT-positive and had SAT titers ≥1∶160 as well as healthy persons . For a combination of the 5 top serodiagnostic proteins , these authors reported a specificity of 96% ( 95% sensitivity , both values optimized by ROC analysis ) . Although further studies are necessary to reach a definite conclusion , these data suggest that protein antigens may not completely solve the specificity problems in human brucellosis serodiagnosis . In summary , when complemented with appropriate anamnesis and clinical findings , RBT is a very useful test for the diagnosis of human brucellosis . It needs no complicated infrastructure or sophisticated training , it is exceedingly cheap , highly sensitive and easily adaptable to test serum dilutions . On these bases , a simple scheme for the diagnosis of human brucellosis can be proposed that could avoid confirmation of a large proportion of positive results in the standard RBT protocol ( i . e . , those RBT titers ≥1∶8 ) ( Figure 1 ) . If an assessment of the stage of evolution of a particular case is necessary , a complementary test assessing IgM and IgG levels could be used , and for this purpose the simple LFiC seems the appropriate complement to RBT in small laboratories . Indeed , it is not possible to predict the proportion of RBT results that need confirmation ( RBT titers <1∶8; Figure 1 ) in a given population but it is expected that long evolution cases with low levels of antibodies in RBT or other S-LPS tests would be more common in endemic areas with no adequate awareness of the human disease . As stressed above , these cases would need a careful assessment by physicians , and further serological studies in well equipped laboratories using tests like Coombs or Brucellacapt . Moreover , culture should be attempted because , even though its sensitivity is low in these cases , a positive result is a definite proof of infection . | The Rose Bengal Test ( RBT ) for brucellosis serological diagnosis was adapted to test serum dilutions and its usefulness evaluated using sera of Brucella culture positive patients , persons with contact with Brucella but no symptoms , veterinarians accidentally injected with vaccine Rev 1 who had not developed the disease and normal persons . Using the standard protocol , RBT was not outperformed by more sophisticated and expensive tests ( serum agglutination , Coombs , competitive ELISA , Brucellacapt , and lateral flow immunochromatography for IgM and IgG detection ) in identifying Brucella infected patients . All tests failed to discriminate with total specificity the sera from contacts or Rev 1 injected individuals . However , none of these sera was positive in the modified RBT adapted to test serum dilutions at titers higher than 1>4 . When there is suspicion of brucellosis , RBT is recommended as the first test and , depending upon the titer , a positive result does not need confirmation by other ( usually more expensive , sophisticated and time consuming ) tests . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/bacterial",
"infections",
"microbiology/medical",
"microbiology"
] | 2011 | The Rose Bengal Test in Human Brucellosis: A Neglected Test for the Diagnosis of a Neglected Disease |
Groups can make precise collective estimations in cases like the weight of an object or the number of items in a volume . However , in others tasks , for example those requiring memory or mental calculation , subjects often give estimations with large deviations from factual values . Allowing members of the group to communicate their estimations has the additional perverse effect of shifting individual estimations even closer to the biased collective estimation . Here we show that this negative effect of social interactions can be turned into a method to improve collective estimations . We first obtained a statistical model of how humans change their estimation when receiving the estimates made by other individuals . We confirmed using existing experimental data its prediction that individuals use the weighted geometric mean of private and social estimations . We then used this result and the fact that each individual uses a different value of the social weight to devise a method that extracts the subgroups resisting social influence . We found that these subgroups of individuals resisting social influence can make very large improvements in group estimations . This is in contrast to methods using the confidence that each individual declares , for which we find no improvement in group estimations . Also , our proposed method does not need to use historical data to weight individuals by performance . These results show the benefits of using the individual characteristics of the members in a group to better extract collective wisdom .
Francis Galton was the first to experimentally demonstrate the advantages of collective estimations [1] . At a farmers’ fair , he found that the median of the independent estimations made by 784 farmers of the weight of a slaughtered ox was better than any of their individual estimations . Since then , collective estimations , computed as mean , median or geometric mean values of the group , have been shown to improve upon the estimations of most individuals of a group in several different contexts , an effect popularly known as wisdom of crowds ( WOC ) [2–8] . However , human crowds can also be notoriously bad at making collective estimations for many estimation tasks [7 , 9] . Social interactions can have an additional negative effect in biased crowds [8 , 9] . When individuals learn the estimations of the other members of the group , they typically change their own estimation towards the more common values . After social influence , the collective has thus a distribution of estimations more strongly peaked around the biased solution . This can give the collective perception of an agreement but the value agreed upon can be far from the truth [9] . We propose to turn the negative effect of social interactions to our advantage and improve collective estimations . We do so by taking into account the individuality of the members of the group . Francis Galton argued for each individual counting the same in the collective estimation [1] . But for situations in which most individuals are strongly biased , we would be in a better position with methods selecting the unbiased individuals . Of course , this can be done by finding how well each individual performs in a domain of knowledge and weight them accordingly for similar tasks [10–12] . Here we do not consider the case of access to a classification of individuals by performance . Instead we used the impact of social interactions on estimations to extract individuals in the following way . We first obtained a model of estimation in a collective and used it to measure how much each individual of the collective resists social influence . We tested the model by reanalyzing a dataset in which subjects made estimations before and after social influence [9] . This is a rich dataset that can be used as a reference to test models of social influence [13] . In these experiments subjects were asked to privately estimate the answer to six questions [9]: ‘What is the length of the border between Switzerland and Italy in kilometers ? ’ , ‘How many rapes were officially registered in Switzerland in 2006 ? ’ , ‘How many assaults were officially registered in Switzerland in 2006 ? ’ , ‘What is the population density of Switzerland in inhabitants per square kilometer ? ’ , ‘How many murders were officially registered in Switzerland in 2006 ? ’ and ‘How many more inhabitants did Zurich gain in 2006 ? ’ After their private estimation for each question , each subject could receive social interactions consisting in either receiving on a computer screen a diagram depicting the private estimations of each member of the group ( ‘full information’ condition ) or more simply their arithmetic mean ( ‘aggregated information’ condition ) . To test that the observed effects were due to social interactions , they also used control groups that also estimated twice but without social influence in between ( ‘no information’ condition ) . The experimental data was obtained using 144 people organized in 12 groups of 12 people . Each group was asked 6 questions , 2 in each of the three conditions . We used our model to classify each individual by their resistance to social influence as a measure of confidence on their private information . Our proposal is then to use the geometric mean of the estimations of individuals with high social resistance as a better estimator than the WOC , as we show for the dataset from reference [9] .
To understand the effect of social interactions in estimation , we first tested whether we could model each person in a group as an estimator of some quantity according to their private and the social information . We have already used this modeling approach for fish and ant groups choosing among a low number of options [14 , 15] . Here we adapt it to the case of human data in which individuals estimate quantities that can take any positive real number and the distribution of estimations before social interactions is a log-normal [9 , 16–18] . For the analysis of experimental data it is thus useful to take the logarithm of the raw estimations {xi} to obtain {yi ≡ logxi} , whose distribution is then a Gaussian . We obtained that if before social interactions this Gaussian has mean μp and standard deviation σp , N ( μp , σp ) , after social interactions the distribution of estimations is predicted to be of the form ( see S1 Text ) fY ( y ) =N ( wpμp+wsμs , σp1−ws ) . ( 1 ) The predicted distribution in Eq 1 is also a Gaussian in the logarithm of estimations , but its mean and standard deviation have changed . The mean μf is at a value combination of the private mean μp and a parameter μs that summarizes the impact of social information , μf=wpμp+wsμs , with wp and ws the private and social weights with values between 0 and 1 and with wp+ws=1 . The form of μs in Eq 1 depends on the type of social interactions , and we considered two types . One in which each individual receives the estimations from all members of the group , for which we found that μs is of the form μs ≡ log ( xs ) , where xs is the geometric mean of the estimations ( see S1 Text ) : xs= ( ∏i=1nxi ) 1/n . ( 2 ) We also considered a second form of interaction in which each individual receives only the mean of the estimations of the group , for which the social information is the mean of estimations ( see S1 Text ) xs=1n∑i=1nxi . ( 3 ) These two types of social information impact Eq 1 differently , with only the second of them changing the mean after social interactions . This is because in the first case , as the expected value of the geometric mean of a sample following a log-normal distribution is the median of the population [19 , 20] , then we have on average that xs = exp ( μp ) ( see S1 Text ) , making the mean the same as before social interactions , μf = μp . In the second form of social interactions via the arithmetic mean , the expected value is xs=exp ( μp+σp2/2 ) ( see S1 Text ) , making the mean to shift to higher values after interactions , μf=μp+wsσp2/2 . Social interactions can change not only the mean but also the standard deviation of estimations . The predicted standard deviation after social interactions in Eq 1 is reduced to σf= σp1−ws , more reduced the higher the social weight ws , making the group to agree more around the final mean . We first tested that the predicted distribution in Eq 1 is consistent with the experimental data in [9] . We standardized the estimations made by each individual using a z-score as z ≡ ( y-μp ) /σp , with y the logarithm of the estimation and μp and σp the mean and standard deviation for each time in which a group answered a question . This transformation of variables allowed us to pool together estimations from different groups and questions , each having its own mean and standard deviation . The distribution of the z-score values before social influence has mean 0 and standard deviation 1 , N ( 0 , 1 ) ( Fig 1A and 1B , blue ) . It transforms after social influence according to Eq 1 into N ( 0 , 1−ws ) for the ‘full information’ condition ( see S1 Text ) . This correctly predicts that the distribution of z-score values after social interactions cannot be distinguished from a Gaussian ( p>0 . 27; Kolmogorov-Smirnov test ) , does not change the mean ( p = 0 . 14 , permutation test; see Methods ) and reduces the standard deviation ( p<10−9 , permutation test ) . Unless otherwise stated , in the remainder of the paper we use permutations to obtain p-values . The predicted form N ( 0 , 1−ws ) gives a very good fit to the data and the standard deviation of the data corresponds to a value of the social weight in Eq 1 of ws=0 . 53 ( Fig 1A , red ) . For the ‘aggregated information’ condition , the Gaussian distribution in Eq 1 for the z-score values is after social interactions of the form N ( wsσp/2 , 1−ws ) ( see S1 Text ) . The value of the final mean depends on the standard deviation of estimations before the interaction , σp , that is different for each of the 24 experiments [9] in which each of the 12 groups answered two questions in the ‘aggregated information’ condition . Using for each experiment the value of σp before social interactions and the value of ws for the same group but in the ‘full information’ condition , we can predict the shift in mean and the reduction of standard deviation for the 24 experimental cases ( S1 Fig ) . However , a simpler analysis can be made neglecting the variability of values in σp across the 24 experiments , and instead pool all the data and consider the prediction only using the mean value σp¯ as N ( wsσp¯/2 , 1−ws ) , with σp¯ = 1 . 39 , and ws=0 . 53 from the ‘full information’ condition . The predicted shift in the mean , wsσp¯/2=0 . 39 , and the reduction in standard deviation , 1−ws=0 . 68 , correspond well with the experimental data ( Fig 1B , red ) and with the more complete prediction using the sum of 24 Gaussians predicted for each experiment ( S1 Fig ) . It correctly predicts a shift of the mean to higher values ( p<10−6 ) and a reduced standard deviation ( p<10−6 ) that was not found to be different to the one in the ‘full information’ condition ( p = 0 . 45 ) . An alternative Bayesian test [21] shows similar results for the problems studied here ( see Methods and S1 Table for a summary of permutations and Bayesian significance tests ) . In addition , this test obtains when two quantities are likely taking the same value and not only when they are not found to be different , as in the case of the standard deviation in the ‘aggregated information’ and ‘full information’ conditions ( S1 Table ) . In the ‘no information’ condition , subjects repeat the estimation with no social interactions in between and we found no significant change in the parameters of the distribution of estimations ( S2 Fig , p>0 . 5; see also Bayesian test in S1 Table ) . This shows that the effects seen after social interactions are due to the interaction and not to a repetition of the estimation . Once we tested the close correspondence between the statistical model in Eq 1 and the experimental data , we considered a simple model for an individual that is consistent with the statistical predictions . Specifically , an individual that privately estimates x1 and , upon reception of the social information , gives a new estimation x2 related to x1 through a linear combination in the logarithmic domain , y2=wpy1+wsμs . ( 4 ) with {y1 , 2 ≡ logx1 , 2} , is consistent with the statistics in Eq 1 . This implies that the second estimation can be predicted from the first estimation and the social information as logx2=wplogx1+wslogxs , which is found to be a good approximation for the data with ws=0 . 53 ( Fig 1C ) . A more common rule used in the modelling of social influence in humans is the linear combination rule x2=wpx1+wsxs [13 , 22–24] , but Eq 4 is a linear combination in the logarithmic domain or , equivalently , a weighted geometric mean between private and social information , x2=x1wpxsws . ( 5 ) So far we have assumed that each individual uses the same value of the social weight ws . However , there might be individual differences , with some individuals less influenced by social information . Using wp+ws=1 and Eq 4 we can obtain a different value of the social weight for each individual as ws=y2−y1μs−y1 . ( 6 ) The distribution across the group of values of the social weight ws in Eq 6 shows a striking structure of individual differences ( Fig 1D ) . Some individuals resist social influence ( peak at ws=0 in Fig 1D ) , others shift almost completely to the social information ( peak at ws=1 ) , others combine private and social information ( values between 0 and 1 ) , and even some shift to values farther from the private value than the social value ( ws>1 ) or to values in a direction opposite to the social value ( ws<0 ) . We took advantage of the individuality and extracted the geometric mean of those individuals that resist social influence . To gain intuition on how to perform this extraction we considered the following exploration of the data . We obtained the joint density of social weights ws and private estimations y = log ( x1 ) for the question ‘What is the length of the Swiss/Italian border ? ’ ( Fig 2A ) . To obtain different levels of resolution , we used the following Gaussian smoothing of the data [25] f ( ws , y ) =12πσwsσyn∑i=1nexp ( − ( ws−ws , i ) 22σws2− ( y−yi ) 22σy2 ) ( 7 ) with ws , i and yi = log ( x1 , i ) the social weight and the private estimation of individual i , respectively , σy≡σ^yn−1/γy and σws≡σ^wsn−1/γws with σ^y and σ^ws the sample standard deviation of each variable . We varied the resolution coefficient γws while keeping γy at its optimal value of γy = 6 [25] to see whether there is a consistent tendency for individuals with different social weights to give different estimations ( Fig 2A ) . At the lowest resolution considered , γws=6 , there is a clear tendency of individuals with lower social weight to give higher estimations of the border length between Switzerland and Italy ( Fig 2A , γws=6 ) . At resolution γws= 2 , 3 and 4 the density splits into two peaks , one at high ws and another at low ws ( Fig 2A , γws= 2 , 3 , 4 ) . It is thus clear that for this question the individuals with lower social weight tend to give higher estimations . We then extracted the individuals with lowest social weight . A simple method consists in extracting all individuals with a social weight below the value that gives a result significantly different to WOC ( Fig 2B ) . Specifically , we started from the complete group and its geometric mean as the WOC value . For this case , the WOC value is 302 km ( Fig 2B ) . We then eliminated individuals one by one from highest to lowest values of the social weight keeping those with | ws |≤ω , with ω a decreasing positive real number . With the remaining individuals , we computed the geometric mean . For ω in the interval between 0 . 1 and 0 . 5 of individuals with high resistance , the geometric mean increases to values close to 800 km . At the lowest values of ω there is a drop in the geometric mean , but the number of individuals is also low . To isolate the relevant individuals , we found which values of ω give a geometric mean significantly different from the WOC ( Fig 2B , green dots for p<0 . 05 and red dots for p<0 . 01 ) . The significant values of ω are in the interval from 0 . 06 to 0 . 45 , which correspond to groups whose geometric mean lies between 816 and 464 km , respectively . We then tested that we obtain similar estimations using the complete interval of significant values of ω or only the value of ω giving the highest significance . Specifically , for the complete interval of significant ω we used the following measure that weighted more the values of ω with higher significance as resist 1≡∫0 . 50q ( ω ) x1geom ( |ws|≤ω ) dω∫0 . 50q ( ω ) dω ( 8 ) with x1geom ( |ws|≤ω ) the geometric mean of the estimations of individuals with social weight | ws |≤ω , q ( ω ) =0 . 05−p ( ω ) if the p-value obeys p ( ω ) <0 . 05 and q ( ω ) =0 otherwise , and only counting those groups with sufficiently low social weight , ω≤0 . 5 . The prediction obtained in this way is 714 km , that deviates only -2 . 7% from the true value of 734 km while the WOC value of 302 km deviates -59% ( Fig 2B , ‘resist 1’ , ‘truth’ and ‘WOC’ ) . An alternative to Eq 8 would also use the values of ω giving significance but weighted all of them equally , giving 689 km , -6 . 2% off the true value ( Fig 2B , ‘resist 2’ ) . Another variant would only take into account a single value of ω with the highest significance ( p = 0 . 0002 ) that corresponds to ω=0 . 25 . This gives the prediction of 780 km , 6 . 3% off the true value ( Fig 2B , ‘resist 3’ ) . The three variants give very similar predictions and a large improvement over WOC . We also used a second class of methods based on the finding that resisting individuals can form peaks in the joint distribution of estimations and social weight ( Fig 2A ) . Methods using the peaks will in general use less individuals but should be valuable when the peaks are clear in the distribution , that is , when they are sharp and separated from other peaks . Specifically , we used clustering by Gaussian mixtures [26] . The advantage of this method is that , although it depends on the distribution and therefore on the value of the resolution γws , it is very robust to changes in its value . For the question about the length of the Swiss/Italian border , we obtained that the geometric mean of the cluster of people with low social weight is 422 , 481 , 512 and 491 km for γws = 2 , 3 , 4 and 6 , respectively ( Fig 2C ) . In particular , it is not necessary that the value γws chosen for the clustering corresponds with a distribution showing peaks . For example , the distribution with γws = 6 does not show peaks and it is clustered into approximately the same two clusters than the distribution with γws = 3 that shows two clear peaks . The values obtained are -42% , -34% , -30% and -33% off the true value of 734 km . The cluster at high social weight correspond to individuals with larger errors ( -69% , -67% , -71% and -67% for γws = 2 , 3 , 4 and 6 , respectively ) . WOC is typically a value between the ones at low and at high social weights , here 302 , -59% off the true value . So far we have seen that using the individuals with lowest social weight we can estimate ‘What is the Swiss/Italian border length ? ’ better than using WOC . The results were robust under changes in the method to extract the individuals with low social weights , with a total of 7 variants of the methods used improving over WOC ( Fig 2D ) . We then applied the same methods to the remaining 5 questions from the experiments in [9] . We found a subpopulation with a significant resistance to social influence in 3 of the remaining questions ( Fig 3 and Table 1 for a summary; see S4 Fig for the other two questions ) . For the question of ‘Number of rapes in 2006 in Switzerland’ the geometric mean of individuals of low social weight as measured by Eq 8 and its two variants gives the same value as there is a single significative group at a value of 624 , much larger than the WOC result of 257 ( Fig 3A , ‘resist 1 , 2 , 3’ ) . This corresponds to a much smaller error ( -2 . 3% ) than the WOC ( -60% ) respect to the truth at 639 . The distribution of estimations does not show a structure of two peaks separated at low and high social weight ( Fig 3B , γws = 3 , 4 , 6 ) and at high resolution there are too many peaks with very few individuals each ( Fig 3B , γws = 2 ) so a method based on peaks is not appropriate for this question . For the ‘Number of assaults in 2006 in Switzerland’ , the geometric mean in Eq 8 and the two variants considered have a large deviation from the WOC value of 3685 to 6654 , 6313 and 7557 , respectively ( Fig 3C , ‘resist1’ , ’resist 2’ , ’resist 3’ ) . They correspond to errors of -28% , -32% and -18% , respectively , much lower than the -60% error of WOC . The clustering method obtains the same value of 7699 for γws = 3 , 4 and 6 ( Fig 3D , γws = 3 , 4 , 6 ) and for γws = 2 the resolution is too high and reveals at least four peaks with very few individuals per peak ( Fig 3D , γws = 2 ) . For γws = 3 , 4 , and 6 the error is -17% of the true value 9272 compared to the -60% error of the WOC of 3685 . For the question about the ‘Population density of Switzerland’ the geometric mean in Eq 8 does not find a subpopulation resisting social influence with estimations significantly different to WOC ( Fig 3E ) . The clustering method finds for γws = 2 , 3 , 4 and 6 the values 174 , 177 , 177 and 171 , respectively ( Fig 3F , γws = 2 , 3 , 4 , 6 ) . Compared to the true value of 184 , these values are -5 . 7% , -4 . 0% , -4 . 0% and -7 . 2% off the true value of 184 while the WOC value of 115 is -38% off . Our analysis shows that estimation is improved when there is a subpopulation significantly resisting social influence . The seven variants of the methods improve upon WOC and in many cases the improvement is very large ( Table 1 ) . The success of the method rests in the correlation between resistance to social influence and closeness to the true value seen in the data . It is also interesting to consider some properties of the resisting individuals . The proportion of these individuals is 25±13% using the methods based on Eq 8 and 10±3% for the methods based on the peaks of the distribution . The individuals that resist social influence are not the same in all questions . We only find a significant overlap between questions 1 and 2 ( S5A Fig , p<0 . 05 ) . Resistance to social information may be viewed as a behavioral measure of confidence , and the estimation of those resisting social influence as ‘wisdom of the confident’ . Its success is not a trivial result as other measures of confidence like declared confidence in a scale from 1 to 6 does not improve accuracy [28–31] . We thus decided to compare why the two measures give different results . We found a significant but very low correlation between resistance to social information and declared confidence ( S5B Fig , p<0 . 001 , R2 = 0 . 03 ) . While there are approximately equal numbers of resisting and non-resisting individuals ( Fig 1D ) , most of the population declares low values of confidence , even the majority of those resisting social influence ( S5C Fig , triangles ) . Individuals declaring high values of confidence ( S5D Fig , triangles ) , in general resist social influence more than those with low values , but a relevant proportion does not resist social influence . The two measures are correlated but are very different and it is then unsurprising than a method like the one proposed here for social resistance does not work for declared confidence ( S6 Fig ) .
We have here proposed to extract information from the collective using those individuals resisting social influence . The methods proposed extract the information a collective considers of high private quality . We obtained better collective estimations than the ‘wisdom of crowds’ [1–9] using the data from [9] , especially for cases in which the crowd shows a very large bias . The methods work because resistance to social influence correlates with closeness to the true value . The correlation does not need to be very strong , that is , we do not need experts [10–12] . Instead , we use the geometric mean of those individuals that get influenced less by social information and this group can still show a large standard deviation . We used two types of methods . One based on Eq 8 , taking all individuals below a value of social weight that give a result different from WOC . This method gave predictions very close to true values for those cases in which the joint distribution of estimations and social weight does not show a complex structure at low social weights . When this method does not give significant results , one can resort to a method based on clustering in the space defined by estimations and social weights . This second type of methods takes into account less individuals , but we found they improve upon WOC . The two methods together can be used to understand the relevant subjects in the estimation . For example , Eq 8 does not give significant results for the question on the ‘Population density of Switzerland’ ( Fig 3E ) . Inspection of the density shows that while there is a strong peak at low social weight with an estimation very different from WOC ( Fig 3F ) , there are individuals giving much lower estimations and thus making the geometric mean of individuals with low social weight not different from WOC . Our proposal makes use of individuality to improve upon WOC . It is interesting to speculate what type of individuality is most compatible with our results . One type of individuality would simply be that all individuals use a similar procedure to answer a question but their levels of noise are different . One way to model this would be to extend our models to incorporate that all individuals are most likely to give the correct answer but they have different levels of noise ( S1 Text ) . This model gives very poor predictions ( S1 Text ) . The reason is that the data seems more compatible with different subgroups of people with different biases from the truth , for example the low and high peaks in the joint density in Fig 2A . This can be modelled in that the most probable estimation is shifted away from the true value with different biases in different individuals . As biases are defined respect to truth , this extension of the models would not be predictive . Instead , we propose the methods in the main text , by which we extract the subgroup of individuals of low social weight as the more accurate ones on average . The idea that different individuals or subgroups of individuals have different biases is compatible with the existence in the population of different procedures to solve a problem , each of them with a different bias . According to this view , a possible origin of the data for the question about the Swiss/Italian border as an example could be the following . This question might be answered estimating the approximate length of a straight line separating the two countries , which is 288 km as measured from a map in http://www . freemaptools . com/measure-distance . htm . Interestingly , the cluster of individuals with highest social weight is characterized by an estimation of 216±157 km compatible with these very low values . A procedure more sophisticated than simply the length of a straight line consists in using the shape of the border . Another procedure is to use memorized data to retrieve its value . The cluster at low social weight is characterized by an estimation of 512 ±269 km and the geometric mean at low social weights by values in the interval 650–800 km , compatible with these more sophisticated procedures . This idea of different procedures might also explain the different susceptibilities to social information . Those individuals using the shape of the Swiss/Italian border would in general not consider as very important social information with values so much lower than their estimations . This is because these values would be incompatible with the shape , for example values closer to a straight line . In contrast , individuals using a straight line approach might be willing to consider higher values , as they might have only taken this approach as a very rough approximation they could make because they had difficulties finding how to estimate the full shape . All individuals might declare low confidence levels as they can be very noisy within their approach , but they might still consider differently values more compatible with other approaches . A second and complementary explanation of individuality is that individuals have different levels of expertise on the subject or even in general exercises of estimation . This level of expertise is probably not high enough for the individuals to declare it , but it would be enough to act upon it when confronted with social influence . The methods proposed to improve upon WOC do not correspond to a common situation in which humans interact naturally . Instead , it is a protocol that can be used to extract high quality information in human collectives even if it is present only in a minority of the group . Its value relies on improving upon WOC by eliminating the people that are not confident in their private estimations . And using how much each individual is influenced by others as a measure of confidence seems to extract the correct individuals , unlike methods based on declared confidence [28–31] . Our results point to measures of confidence not based on declaration as a means to gather high quality private information in a group . Response time , perseverance or pay-offs in decision systems might be implementations to test experimentally . An open problem is in which circumstances social influence or these other measures of confidence can be used by humans to improve individual and collective decisions in naturalistic settings .
The distributions were calculated using Gaussian kernel smoothing [25] . The 1D version of Gaussian kernel smoothing was applied for social weights ws in Fig 1D . f ( ws ) =12πσn∑i=1ne− ( ws−ws , i ) 22σ2 , ( 9 ) with { ws , i } the values of the social weights obtained from experiments using Eq 6 , n the length of the sample and σ≡σ^n−1/γ the bandwidth with σ^ the standard deviation of the sample and γ the resolution coefficient . We set the resolution coefficient to half its optimal value [25] , γ=52 , a value that allows the visualization of the main structure of the distribution . We were interested in the interval [0 , 1] and did not then consider points outside ( -1 , 2 ) in our calculations of the bandwidth , avoiding tail effects . The 2D case of Gaussian kernel smoothing is described in the main text , Eq 7 . A complete list of significance tests can be found in S1 Table . In the main text , unless otherwise stated , we computed p-values explicitly without assumptions about the data as the probability that the experimental result is obtained at random . For example , to find whether two distributions have a significantly different value of some parameter θ ( in our case , the mean or the variance ) , we performed a permutations method . We mixed the two samples and randomly divided the resulting set into two subsets . Then , we computed the sample value of the parameter in each of the subsets and extracted the difference d ≡ |θ1-θ2| . We repeated this process 106 times , obtaining a distribution of differences d . The significance p is the proportion of d values bigger than the difference of the parameters between the two original samples . To find whether the group of individuals with ws≤ω in Figs 2 and 3 has geometric mean significantly different from WOC , we used the following procedure . Each ω corresponds to a subgroup of nω individuals . We obtained 105 random sets of nω estimations from the whole crowd and computed the geometric mean of each set , g . The significance of x1geom ( ws≤ω ) is the proportion of values of g at least as far to the wisdom of the crowd ( geometric mean ) as x1geom ( ws≤ω ) . To divide the region of maximum density into two clusters , we performed an Expectation Maximization ( EM ) algorithm to obtain a mixture of two Gaussians [26] . More specifically , for each value of γws we selected those individuals whose social weight and estimation ( wsi , logxi ) lied in the zone of maximum probability , defined as that where the probability in Eq 7 is at least equal than half of the maximum . Then an EM algorithm was applied to the selected data points to find the maximum likelihood estimates of the parameters of a Gaussian mixture with two components . To find whether two questions shared a significant number of individuals with low | ws | , we used the exact expression for the probability that two samples from a finite population have a certain number of elements in common ( see S1 Text ) . | We modelled how humans interact , and used the models to find strategies that can make groups more accurate . Each individual in a group combines private and public information to make estimations . But when the public information is biased , social information has the effect of making groups agree even more on an incorrect collective estimation . We reasoned that not all individuals should be influenced equally by the incorrect public information . We obtained a model to understand how private and social information are combined , and used it to obtain a value of social resistance for each individual . We then used these values of social resistance obtained from the model to extract the subgroup of people resisting social influence , and found that they give an improved collective estimation . Collective intelligence is thus maximal when taking into account individuality in human behavior . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Improving Collective Estimations Using Resistance to Social Influence |
Juvenile dermatomyositis ( JDM ) is a chronic inflammatory myopathy and vasculopathy driven by genetic and environmental influences . Here , we investigated the genetic underpinnings of an analogous , spontaneous disease of dogs also termed dermatomyositis ( DMS ) . As in JDM , we observed a significant association with a haplotype of the major histocompatibility complex ( MHC ) ( DLA-DRB1*002:01/-DQA1*009:01/-DQB1*001:01 ) , particularly in homozygosity ( P-val = 0 . 0001 ) . However , the high incidence of the haplotype among healthy dogs indicated that additional genetic risk factors are likely involved in disease progression . We conducted genome-wide association studies in two modern breeds having common ancestry and detected strong associations with novel loci on canine chromosomes 10 ( P-val = 2 . 3X10-12 ) and 31 ( P-val = 3 . 95X10-8 ) . Through whole genome resequencing , we identified primary candidate polymorphisms in conserved regions of PAN2 ( encoding p . Arg492Cys ) and MAP3K7CL ( c . 383_392ACTCCACAAA>GACT ) on chromosomes 10 and 31 , respectively . Analyses of these polymorphisms and the MHC haplotypes revealed that nine of 27 genotypic combinations confer high or moderate probability of disease and explain 93% of cases studied . The pattern of disease risk across PAN2 and MAP3K7CL genotypes provided clear evidence for a significant epistatic foundation for this disease , a risk further impacted by MHC haplotypes . We also observed a genotype-phenotype correlation wherein an earlier age of onset is correlated with an increased number of risk alleles at PAN2 and MAP3K7CL . High frequencies of multiple genetic risk factors are unique to affected breeds and likely arose coincident with artificial selection for desirable phenotypes . Described herein is the first three-locus association with a complex canine disease and two novel loci that provide targets for exploration in JDM and related immunological dysfunction .
Juvenile dermatomyositis ( JDM ) is an autoimmune vasculopathy that causes a characteristic skin rash ( heliotrope rash across the eyelids and Gottron’s papules on the bony prominences ) and proximal muscle weakness [1] . It is the most frequently diagnosed childhood inflammatory myopathy , comprising 80% of all cases [1] and affecting 3 . 2 in every million children between the ages of 2 and 17 within the USA [2] . Prognosis is positively correlated with early diagnosis and swift treatment with corticosteroids and/or immunosuppressants [1 , 3] . While treatment of JDM is much improved overall , many children still suffer from chronic flare-ups [1] . Though the etiology is unknown , JDM is thought to be triggered by exposure to a virus or other environmental agent . Manlhiot et al . [4] reported that 71% of JDM patients had a clinical history consistent with infection prior to symptoms . Investigations into the class II major histocompatibility complex ( MHC ) , TNF , and IL1 identified several susceptibility and protective alleles , but their collective contribution to pathogenesis is poorly understood [5–8] . Recent genome-wide association studies ( GWASs ) to identify additional susceptibility loci in JDM confirmed a strong association with the MHC but failed to detect novel major risk factors , likely because of a paucity of biological samples and genetically heterogeneous populations [9 , 10] . In domestic dogs , an inflammatory vasculopathy of the skin and muscle , also termed dermatomyositis ( DMS ) , is clinically , histologically , and immunologically similar to JDM and provides the only animal model available to study genetic risk factors [11–16] . The earliest clinical signs of DMS are crusting and scaling on the face , ears , tail tip , and/or across the bony prominences of the limbs and feet [17–19] ( S1 Fig ) . Alopecia and more extensive skin lesions may develop over time , resulting in dermal scarring associated with erythema and mottled pigmentation [17 , 19] . Lesions persist for weeks to months , and may or may not chronically recur [17] . Muscle wasting manifests as atrophy of the head musculature; difficulty eating , drinking , and swallowing; and an atypical , high-stepping gait [17 , 19] . Similar to JDM , DMS is an immune-mediated disease [13 , 18 , 20] that typically develops following an environmental trigger , such as vaccination or viral infection , and is exacerbated by subsequent stressors like exposure to UV light [13 , 17 , 21 , 22] . Anecdotal reports indicate that rabies vaccination , parvovirus infection , owner surrender , or mistreatment commonly precede disease onset . Consistent with an environmental trigger , age at onset is variable with many cases occurring between seven weeks and six months of age , but others not developing until well into adulthood [17–19 , 23] . DMS is diagnosed almost exclusively in the genetically [24] and phenotypically similar collie and Shetland sheepdog breeds , suggesting the presence of a strong heritable component ( s ) arising from ancestors common to both breeds . A 1980s study of disease transmission in the collie eliminated simple Mendelian modes of inheritance [14] . In two test matings , an affected male collie sired litters from an affected collie and a healthy Labrador retriever . All six collie puppies were affected with variable degrees of severity , while three of the four mixed breed puppies developed milder forms of DMS . Retrospective pedigree analyses of the collie sire and dam showed a complete absence of affected ancestors [14] . The availability of a naturally-occurring canine model provides a new opportunity for the identification of genetic risk factors of JDM . The conserved genomic backgrounds of genetically isolated dog breeds have enabled detection of risk loci in complex diseases that are often obscured by heterogeneity within human cohorts [25–30] . Here , we evaluated class II MHC haplotypes , performed multibreed GWASs , and generated whole genome and transcriptome sequencing data to dissect the genetic basis of DMS . We uncovered common polymorphisms of the MHC and two novel loci that are strongly associated with DMS , as well as patterns of allelic inheritance that explain 93% of cases studied . A genetic test is now available to determine the likelihood of a dog developing DMS and to facilitate breeding decisions that avoid progeny having high-risk genotypes .
Given the involvement of MHC genes in JDM , we first determined alleles of the highly polymorphic canine MHC class II dog leukocyte antigen ( DLA ) genes: DLA-DRB1 , -DQA1 , and -DQB1 . Two locus ( DLA-DRB1 and -DQB1 ) and three locus ( DLA-DRB1 , -DQA1 , and -DQB1 ) haplotypes were first generated for 50 collies and 117 Shetland sheepdogs , respectively . Because all observed haplotypes contained a unique DLA-DRB1 allele , the 355 remaining dogs were genotyped for this locus only and the haplotype was inferred ( Table 1 ) . We observed remarkably low DLA diversity among collies , with only three haplotypes present in 225 collies worldwide . This lack of heterogeneity precluded detection of associations with DMS , as 91% of collies were homozygous for the haplotype DLA-DRB1*002:01/-DQA1*009:01/-DQB1*001:01 . In 297 Shetland sheepdogs , we identified two predominant haplotypes , 002:01/009:01/001:01 and 023:01/003:01/005:01 . The former was over-represented among cases ( P-val = 0 . 0010 , OR = 2 . 20 ) , primarily because of increased homozygosity ( P-val = 0 . 0001 , OR = 2 . 98 ) . We therefore conclude that 002:01/009:01/001:01 is a risk factor for DMS and that homozygosity confers increased susceptibility . Under the assumption that the causal alleles derive from an ancestor common to both breeds , we extrapolate the observed DLA risk to collies . The high frequency of the DLA risk haplotype in both populations indicates that additional loci must influence disease progression . We conducted an independent GWAS for each breed , using a total of 97 cases ( 31 collies , 66 Shetland sheepdogs ) , 68 controls ( 23 collies , 45 Shetland sheepdogs ) , and 98 , 520 SNPs after filtering . In collies , a single signal ( P-val = 1 . 47X10-8 ) composed of 17 SNPs at the centromeric end of chromosome 10 exceeded Bonferroni significance ( Fig 1A ) . In Shetland sheepdogs , this association was replicated ( P-val = 2 . 56X10-7 ) , and a second signal ( P-val = 1 . 83X10-9 ) composed of 11 SNPs surpassing Bonferroni significance was detected on chromosome 31 ( Fig 1B ) . Both associations persisted in a combined breed analysis ( chr10: P-val = 2 . 3X10-12 , chr31: P-val = 3 . 95X10-8 ) ( S1 Table , S2 Fig ) ; although the breeds possessed a common haplotype on chromosome 31 , the Shetland sheepdogs appeared to drive this association . No associated SNPs were detected near the MHC loci on chromosome 12 , likely a result of high homogeneity in our cohort and poor SNP coverage on the array [25] . On chromosome 10 , 97% of all affected dogs were homozygous or heterozygous for the risk alleles of the lead SNPs . On chromosome 31 , 88% of affected Shetland sheepdogs , but only 39% of affected collies , shared the risk alleles of the lead SNPs . As neither locus appeared to be independently necessary for disease development , we surveyed the extent of regional linkage disequilibrium ( LD ) to demarcate candidate intervals of ~1 . 33 Mb on chromosome 10 ( Fig 2A ) and ~696 kb on chromosome 31 ( Fig 2B ) , harboring ~65 and 6 genes , respectively . The large size of the chromosome 10 region is attributed to lower recombination rates near the centromere and a dearth of informative SNPs . Whole genome resequencing was performed for four affected dogs ( three collies and one Shetland sheepdog ) and two unaffected collies , resulting in 17X to 41X coverage . Variants were filtered for those lying within our delineated intervals ( chr10:1–1 , 333 , 693; chr31:24 , 026 , 411–24 , 722 , 836 ) and following the inheritance pattern of the most significantly associated SNPs in the affected dogs ( S2 and S3 Tables ) . Five intergenic and two intronic variants were unique to these breeds ( i . e . , not present in the Boxer reference genome , dbSNP , or 27 whole genome sequences from 16 other breeds ) ; however , most were in repetitive regions and none were in conserved positions . Thus , the pathogenic variants were likely to be common polymorphisms , so we next prioritized variants within predicted exons and splice sites of genes expressed in skin for further study . To confirm exon/intron boundaries predicted by Ensembl 79 and establish expression of variants in affected tissue , we generated RNAseq data . We obtained a minimum of 89 million reads per tissue from active skin lesions of two affected dogs ( one collie and one Shetland sheepdog ) and skin from one unaffected Australian shepherd dog . All genes expressed in skin within the candidate regions were also expressed in affected tissues . Seventeen exonic variants were expressed; seven of these were nonsynonymous and evaluated using in silico programs [31–33] ( Table 2 ) . We genotyped a subset of our mapping population for three nonsynonymous SNPs on chromosome 10 ( ANKRD52 g . 565958G>C , PAN2 g . 627760G>A , and STAT6 g . 1239562G>A ) that were predicted to be deleterious or probably damaging by more than one in silico program . The ANKRD52 and PAN2 variants were more strongly associated with DMS than the lead SNP . These variants were in perfect linkage disequilibrium with each other; however , PAN2 g . 627760G>A was assigned damaging scores with higher confidence by in silico programs ( Table 2 ) . We therefore focused further studies on PAN2 g . 627760G>A , encoding p . Arg492Cys ( XP_531635 . 3 ) , although ANKRD52 cannot be excluded . On chromosome 31 , we genotyped Shetland sheepdogs for a SNP ( g . 24132343A>C ) and an indel ( c . 383_392ACTCCACAAA>GACT , XM_846337 . 4 ) , both located in a 5′ non-coding exon of MAP3K7CL . Only the indel was associated with DMS ( Table 2 , S3 Fig ) . In an expanded , combined population ( 132 affected and 390 unaffected collies and Shetland sheepdogs ) , both the PAN2 ( P-val = 2 . 08X10-35 ) and MAP3K7CL ( P-val = 1 . 45X10-33 ) variants were highly associated with DMS ( S4 Table ) . PAN2 ( or USP52 ) encodes the catalytic subunit of the poly ( A ) nuclease deadenylation complex ( PAN2-PAN3 ) and is one of two exonucleases involved in mRNA degradation in eukaryotes [34 , 35] . Deadenylation plays a role in translational regulation of inflammatory response [36] . Independent of this function , PAN2 also stabilizes HIF1A transcripts via their 3′-UTR , which contain AU-rich elements ( AREs ) , and may be involved in regulating other transcripts having AREs [37] . HIF1A , a key regulator of inflammation [38] , and other ARE-containing transcripts , such as IL-6 [39] , are upregulated in JDM [40 , 41] . PAN2 is widely expressed and highly evolutionarily conserved [35]; human ( NP_001120932 . 1 ) and dog ( XP_013972628 . 1 ) amino acid sequences share 98% identity . MAP3K7CL ( also known as TAK1L or C21orf7 ) is a poorly studied kinase gene that is transcriptionally active in immunological tissues and expressed primarily in peripheral blood leukocytes [42 , 43] . Human ( NP_001273546 . 1 ) and dog ( XP_013965340 . 1 ) MAP3K7CL protein sequences share >90% identity . The transcription factors RUNX3 and EP300 bind the 5′ non-coding exon of human MAP3K7CL ( UCSC Genome Browser ENCODE Transcription Factor ChIP-seq track ) . In this exon , the c . 383_392ACTCCACAAA>GACT indel causes the loss of six conserved base pairs , omitting a RUNX3 binding motif ( P-val = 2 . 07X10-3 from TOMTOM [44] ) ( S3 Fig ) . RUNX3 has known roles in inflammatory response ( e . g . , thymopoiesis [45 , 46] and the TGF-β signaling cascade [47] ) , and it has been directly implicated in a number of immune-related diseases [48–50] . Furthermore , SNPs disrupting RUNX binding motifs in target genes confer susceptibility to autoimmune rheumatic diseases , including psoriasis [51] and systemic lupus erythematosus [52] . We next considered three-locus genotypes in our expanded , combined population ( 132 affected and 390 unaffected dogs ) where A = the PAN2 variant encoding p . Arg492Cys , B = MAP3K7CL c . 383_392ACTCCACAAA>GACT , and C = DLA-DRB1*002:01 , lower-case letters denote wild type alleles ( c represents any alternate allele of DLA-DRB1 ) . Only 4% of dogs possessed a three-locus genotype with cc , barring further analysis of those nine genotypes . We considered nine of the remaining genotypes to be low-risk , as less than 6% of dogs with these allelic combinations had DMS ( Table 3 ) . Among healthy dogs ( Fig 3A ) , the most frequently observed genotypes were AabbCC ( 24% ) and aabbCC ( 15% ) . Based on penetrance , we classified five genotypes as conferring moderate risk ( 33–50% ) and four as high risk ( 90–100% ) for DMS . All cases possessed at least two risk alleles and all but one were homozygous for at least one risk allele . The most common genotypes of affected dogs ( Fig 3B ) were AaBBCC ( 20% of cases ) , followed by AAbbCC , AABbCC , and AABBCC ( 17% of cases each ) . Interestingly , only affected dogs possessed AABBCc or AABBCC ( n = 29 ) , indicating that DMS is fully penetrant in dogs having these combinations . Epistasis plots illustrated that genotypes with at least one a or b allele confer a lower probability of disease when a c allele is present , compared to their CC counterparts ( Fig 3 , compare 3C and 3D ) . The plots also depicted a greater probability of disease than expected under a strictly additive model , providing evidence for additive-by-additive epistasis between the chromosome 10 and 31 loci [53 , 54] . We noted at least one ARE in MAP3K7CL , presenting a mechanism for interaction with PAN2 . No difference in gene interactions was observed between the sexes ( S4 Fig ) . Information regarding age at onset or diagnosis was available for 91 dogs . We compared dogs having two , three , or four risk alleles across PAN2 and MAP3K7CL and observed an inverse correlation between age of onset and number of risk alleles ( S5 Fig ) . Dogs having four risk alleles developed DMS at a significantly younger median age ( 5 months ) than did dogs with only two risk alleles ( 18 . 5 months ) . The complete penetrance of AABB genotypes , combined with an early age of onset , suggest that these dogs may be hypersensitive to commonplace environmental stimuli ( e . g . , routine puppy vaccinations ) . All three identified variants associated with DMS are polymorphisms present in several breeds , raising the question: why are other breeds rarely , if ever , affected by DMS ? We genotyped five or more unrelated individuals from each of 30 diverse breeds for all three loci ( Fig 4 ) . The only other breeds to possess all three risk alleles were Cardigan Welsh corgis and Cairn terriers . Three Jack Russell terriers had moderate-risk genotypes ( AAbbCc ) , as did one Cardigan Welsh corgi ( AABbCc ) ; both breeds are occasionally diagnosed with dermatomyositis-like disease [55 , 56] . None of the 229 individuals possessed a high-risk genotype ( S5 Table ) . Interestingly , Labrador retrievers had both B and C , which could have enabled moderate or high risk genotypes ( AaBBCc or AaBBCC ) in puppies from the outcross mating described by Haupt et al . [14] . Combined frequencies of risk alleles in other breeds were dramatically lower than those observed among collie and Shetland sheepdog populations , and homozygosity for a risk allele ( a characteristic of all moderate- to high-risk genotypes ) was rare . Additionally , breeds having a high frequency of one risk allele had few or no risk alleles at the other loci . For example , Cairn terriers had a high frequency of A ( 75% ) but low frequencies of B ( 19% ) and C ( 3% ) , and no high- or moderate-risk genotypes were observed among 18 individuals . These findings suggest that independently the polymorphisms are neither deleterious nor selected against . It is likely that recent artificial selection for phenotypes shared by collies and Shetland sheepdogs led to increased frequencies of A . We propose that persistence of A in these two breeds is attributed to linkage disequilibrium ( D′ = 0 . 998 ) between wildtype PAN2 ( a ) and another chromosome 10 allele , Merle of PMEL . In heterozygosity , Merle causes a popular pigmentation pattern ( see collie in Fig 1A ) , but homozygosity for the allele results in severe hypopigmentation often with auditory and ocular defects [58] . Wildtype PMEL occurred on chromosomes with either A or a , whereas Merle was only found in conjunction with a . Accordingly , the Merle phenotype was underrepresented in affected dogs ( P-value = 0 . 0018 ) , 64% of which were homozygous for A . Consistent selection for heterozygosity ( but not homozygosity ) for Merle would simultaneously encourage maintenance of both PAN2 alleles . To our knowledge , there are no loci on chromosome 31 under positive selection for maintenance of a characteristic phenotype of collies and/or Shetland sheepdogs . Across five collie genomes , we observed a 1 . 2 Mb selective sweep on chromosome 12 ( S6 Fig ) encompassing the MHC class II loci and leading to near fixation of the 002:01/009:01/001:01 haplotype . We suggest that essentially all purebred collies have increased susceptibility for DMS , conferred by homozygosity for allele C . The Shetland sheepdog population has retained a less common second haplotype that permits heterozygosity at the DLA loci , associated with a lower risk for developing DMS . Ironically , this reduced risk may have masked the presence of otherwise high-risk genotypes ( i . e . , AABb and AaBB ) and hindered selection against A and B alleles . We observed striking allele frequency differences between the two affected breeds at PAN2 and MAP3K7CL: collies had a higher frequency of A , 42% ( 25% in Shetland sheepdogs ) , whereas Shetland sheepdogs had a higher frequency of B , 38% ( 5% in collies ) ( Fig 4 ) . Consequently , the frequency of observed allelic combinations varied between the breeds . The most common genotypes in healthy dogs were aaBbCC in Shetland sheepdogs , 16% ( 5% in collies ) , and AabbCC in collies , 38% ( 11% in Shetland sheepdogs ) . Among affected dogs , AaBBCC predominated in Shetland sheepdogs , 25% ( 8% in collies ) , whereas AAbbCC was the most frequent genotype in diseased collies , 38% ( 8% in Shetland sheepdogs ) . The latter finding is interesting given that AAbbCC is only a moderate-risk genotype . Among collies having this genotype , 67% were unaffected by age 8 , whereas only 30% of Shetland sheepdogs with this genotype were disease-free . This discrepancy in disease probabilities between breeds was unique to this genotype . Further studies will be necessary to determine if other loci confer additional risk for or protection from DMS . The contribution of alleles from multiple loci explains the spontaneous appearance of the disease in lines with no prior history [14] and has hindered elimination of DMS in the absence of a genetic test . For example , a mating between two healthy dogs having low risk genotypes ( e . g . , AaBbCC x AaBbCC ) can produce puppies with low , moderate , or high risk for DMS . This study has led to the first three-locus genetic test for a complex disease of dogs , which will allow breeders to carefully reduce the frequency of A and B among collie and Shetland sheepdog populations while preserving genetic diversity and desirable breed characteristics . In a canine model of JDM , we have identified a complex pattern of causation involving three independent loci , two of which offer new targets for exploration in JDM . Furthermore , these data provide support for the involvement of genetic risk factors independent of the MHC in human inflammatory myopathies . While further experiments are necessary to determine the exact contribution of the chromosome 10 and 31 loci , our findings suggest that DMS may result from an inability to properly regulate inflammatory response . This work highlights the utility of the canine model for unraveling genetic susceptibility conferred by common polymorphisms and/or gene-gene interactions in complex diseases .
All samples were obtained with informed consent according to protocols approved by the Clemson University Institutional Review Board ( IBC2008-17 ) and IACUC ( 2012–039 ) . SNP data are available in dbSNP under BioProject number PRJNA338128 . All whole genome and transcriptome data generated for this study were deposited in SRA genomes under accesssion number SRP081080 . Accession numbers for eight of the 27 other breeds used in variant filtering are SRX1360633 , SRX1360635 , SRX1360637 , SRX1360639 , SRX1022256 , SRX1022262 , SRX1022286 , and SRP081080 . | Juvenile dermatomyositis ( JDM ) is an autoimmune disease of the skin and muscle influenced by both genetic and environmental components . Although genes independent of the MHC are thought to contribute to disease pathogenesis , their identification has been complicated by a paucity of biological samples , disease heterogeneity , and genetically diverse subjects . A naturally occurring inflammatory disease of domestic dogs , also termed dermatomyositis ( DMS ) , is analogous to JDM and is the only animal model available for genetic study . We first determined that , as in JDM , a particular MHC haplotype confers susceptibility to DMS . Capitalizing on the genetic isolation of dog breeds and extremely low MHC diversity within affected breeds , we used multibreed genome-wide association studies to identify two novel loci . Through genome resequencing and additional genotyping , we uncovered highly associated polymorphisms in conserved positions of PAN2 and MAP3K7CL . Analysis of three-locus genotypes revealed uniquely high frequencies among affected breeds and nine allelic combinations that confer moderate or high risk for DMS . The pattern of disease probability illustrates the presence of gene-gene interaction , as well as an inverse correlation between age of onset and number of risk alleles . This study highlights the utility of canine models for mapping susceptibility loci in complex diseases and detecting patterns of genetic interactions obscured in diverse human populations . | [
"Abstract",
"Introduction",
"Results",
"and",
"discussion",
"Methods"
] | [
"genome-wide",
"association",
"studies",
"medicine",
"and",
"health",
"sciences",
"variant",
"genotypes",
"immunology",
"vertebrates",
"dogs",
"animals",
"genetic",
"mapping",
"mammals",
"clinical",
"medicine",
"genome",
"analysis",
"mammalian",
"genomics",
"major",
"hi... | 2017 | Beyond the MHC: A canine model of dermatomyositis shows a complex pattern of genetic risk involving novel loci |
The replacement of histone H2A with its variant forms is critical for regulating all aspects of genome organisation and function . The histone variant H2A . B appeared late in evolution and is most highly expressed in the testis followed by the brain in mammals . This raises the question of what new function ( s ) H2A . B might impart to chromatin in these important tissues . We have immunoprecipitated the mouse orthologue of H2A . B , H2A . B . 3 ( H2A . Lap1 ) , from testis chromatin and found this variant to be associated with RNA processing factors and RNA Polymerase ( Pol ) II . Most interestingly , many of these interactions with H2A . B . 3 ( Sf3b155 , Spt6 , DDX39A and RNA Pol II ) were inhibited by the presence of endogenous RNA . This histone variant can bind to RNA directly in vitro and in vivo , and associates with mRNA at intron—exon boundaries . This suggests that the ability of H2A . B to bind to RNA negatively regulates its capacity to bind to these factors ( Sf3b155 , Spt6 , DDX39A and RNA Pol II ) . Unexpectedly , H2A . B . 3 forms highly decompacted nuclear subdomains of active chromatin that co-localizes with splicing speckles in male germ cells . H2A . B . 3 ChIP-Seq experiments revealed a unique chromatin organization at active genes being not only enriched at the transcription start site ( TSS ) , but also at the beginning of the gene body ( but being excluded from the +1 nucleosome ) compared to the end of the gene . We also uncover a general histone variant replacement process whereby H2A . B . 3 replaces H2A . Z at intron-exon boundaries in the testis and the brain , which positively correlates with expression and exon inclusion . Taken together , we propose that a special mechanism of splicing may occur in the testis and brain whereby H2A . B . 3 recruits RNA processing factors from splicing speckles to active genes following its replacement of H2A . Z .
Histones , the key proteins that compact all eukaryotic DNA into chromatin , have attracted much attention recently because of their impact on all aspects of genome function [1] . Histones form the core structure of chromatin , the nucleosome core , in which ~ 145 base pairs of DNA is wrapped around a histone octamer comprising of a ( H3-H4 ) 2 tetramer flanked by two H2A-H2B dimers . Importantly , the structure and function of a nucleosome can be regulated by the substitution of one or more of the major core histones with their variant forms . Despite being discovered by Chadwick and colleagues over a decade ago , the function of the histone H2A variant , H2A . Bbd remains unknown [2] . In vitro biophysical and transcription studies revealed that H2A . Bbd , and the mouse orthologue , which we designated H2A . Lap1 ( Lack of an acidic patch ) [3] , could not compact chromatin in vitro [4] . Functionally , this permitted high levels of RNA polymerase ( Pol ) II transcription [4] . Adopting the new nomenclature for histone variants [5] , H2A . Bbd and H2A . Lap1 will hereafter be referred to as H2A . B and H2A . B . 3 , respectively . H2A . B histones differ from their canonical counterparts in several important ways . First , H2A . B histones have a reduced acidic patch , a key region on the nucleosome surface required for chromatin compaction [3] . Second , H2A . B histones lack the canonical histone carboxyl-terminal region , which is important for stabilizing the interaction interface between the H3–H4 tetramer and the H2A—H2B dimer [6] . Not surprisingly , H2A . B-containing mononucleosomes are unstable [7 , 8] , which is also consistent with numerous studies showing the unwrapping of nucleosomal DNA from the octamer surface at the DNA entry and exit points [8–10] . This unwrapping of DNA also appears to cause a major reorganization of the histone tails within the nucleosome [11] , and allows octamer formation on DNA fragments smaller than the typical 145 base pairs in vitro that are associated with a canonical nucleosome [12] . Third , the N-terminal tails of H2A . B histones distinctively lack lysine residues , but instead are enriched for arginines . The functional significance of this difference is as yet unknown . H2A . B is a rapidly evolving histone variant family that first appeared in mammals [5 , 13] . Notably , it displays a tissue-restricted expression pattern being highly expressed in the adult testis with some expression in the brain , and is encoded by three genes [3] ( NCBI GEO data sets ) . It also appears to be expressed in mouse embryonic stem cells at a low level [14] . Several studies have overexpressed tagged versions of H2A . B in transformed cell lines [2 , 15 , 16] and following genome-wide analyses revealed it to be preferentially associated with actively transcribed genes . Over expression of H2A . B can have abnormal effects on cell cycle regulation and DNA damage [16] . To begin to understand the possible role ( s ) of H2A . B and its orthologues in its proper physiological context , we previously performed H2A . B . 3 immunofluorescence , ChIP-Seq and expression studies in the mouse testis [3] . A new role for H2A histone variants in the transcriptional activation process was uncovered whereby H2A . B . 3 was specifically targeted to the transcription start site ( TSS ) of active genes , which was previously believed to be nucleosome free [3 , 17] . This location was not observed in the above mentioned H2A . B studies in transformed cell lines . We now show here that H2A . B . 3 is also present at the TSS of active genes in the mouse brain . In order to gain new mechanistic insights into how H2A . B . 3 participates in the gene activation process , here we took a proteomic approach to identify proteins that specifically interact with H2A . B-containing nucleosomes in the mouse testis , analysed its pattern of organisation in germ cell nuclei , uncovered new genomic locations for this histone variant , examined its functional relationship with another histone variant , H2A . Z , and the active H3K36me3 mark , both in the testis and the brain and finally , analysed its interaction with RNA both in vivo and in vitro .
Previously , we identified the mouse orthologue of H2A . B , H2A . B . 3 , and showed that it is expressed between the pachytene stage ( meiosis I , day 19 of spermatogenesis ) and the late round spermatid stage ( immediately following the completion of meiosis II , day 28–30 ) [3] . The expression of H2A . B . 3 peaks at the late round spermatid stage ( its expression is ~8-fold higher at this stage compared to the pachytene stage ) . This is the period of spermatogenesis when the overall level of transcription is extremely high [18] . Our initial transcriptomic and ChIP-Seq analysis revealed that H2A . B . 3 is targeted to the TSS concurrent with gene activation indicating a role in transcription initiation [3] . However , it was unclear whether H2A . B . 3 was also located in the body of an active gene , which would indicate other possible functions for this variant in the expression of a gene . To investigate this possibility , we repeated H2A . B . 3 ChIP-Seq ( at a greater sequencing depth ) and RNA-Seq experiments using micrococcal nuclease prepared mononucleosomes and poly ( A ) -transcripts obtained from 28–30 day old mice testes , respectively . Resulting ChIP-Seq and RNA-Seq libraries were sequenced yielding 100 base pair paired-end reads . First , we produced a testis total input for all genes and a H2A . B . 3 ChIP-Seq profile where the normalised reads ( mean reads per base pair per million reads mapped ( RPM ) ) at each base pair were aligned with the TSS 1 kb upstream and 10 kb downstream ( Fig 1a ) . It is important to note that H2A . B . 3 is not found on all active genes but only a subset ( [19]; see below ) . Several new observations are revealed: ( 1 ) H2A . B . 3 is located within the gene body but interestingly , H2A . B . 3 shows an elevated abundance at the beginning of the gene body compared to the end . This is in contrast to the active gene body mark , H3K36me3 , which is more enriched at the end of a gene [20] . ( 2 ) On average , H2A . B . 3 is ~1 . 4 times more enriched at the TSS compared to exons . Using a 50 base pair region for the TSS ( position -75 to -25 upstream from the TSS ) and the intron—exon boundary ( from the boundary to 50 base pairs within the exon ) , the mean coverage in counts per base pair were 19 . 03 and 13 . 27 , respectively ( P-value = 0 . 03 ) . ( 3 ) A marked H2A . B . 3 depleted region is observed ~ 225 base pairs downstream of the TSS at the location of the +1 nucleosome . Next , we investigated whether H2A . B . 3 was more enriched on exons or introns , or present on both types of sequences . First , input nucleosomes from mice testes were mapped to intron—exon boundaries ( ±1 kb ) of protein-coding genes and ranked according to their expression level ( repressed , low , medium and high ) . Several studies have shown that exons have an increased nucleosome occupancy compared to introns , thus marking them [21 , 22] . Consistent with these previous studies , the normalized nucleosome occupancy profile for all exons showed a nucleosome that is strongly positioned at the exon ( in mice , most exons are between 50 and 200 base pairs long [23] ) ( Fig 1b ) . A negative correlation was observed between the nucleosome occupancy of an exon and the level of expression indicating an overall loss of this nucleosome during transcription ( Fig 1b ) . Normalized H2A . B . 3 ChIP-Seq testis reads were aligned with the intron—exon boundary and indeed an H2A . B . 3-containing nucleosome occupying the exon was observed , which was positively correlated with transcription in contrast to the input nucleosome profile ( Fig 1c ) . A plot showing the distribution of input nucleosomes and H2A . B . 3 at the intron—exon boundary as a heat map illustrates these transcriptional changes in more detail ( S1a and S1b Fig ) . The H2A . B . 3 ChIP-Seq intron—exon plots ( Fig 1c ) reveals that H2A . B . 3-containing nucleosomes are not only located on exons but also on surrounding intronic DNA sequences . Further , a comparison of meta-intron with meta-exon plots shows that H2A . B . 3 is distributed throughout the entire intron , and that exons are not enriched with this histone variant compared to intronic sequences ( S2a and S2b Fig , Fig 1c ) . We conclude that in highly expressed genes , H2A . B . 3 is enriched in both exon and intron regions and that this enrichment is inversely correlated with nucleosome occupancy seen in the input of the same regions . While H2A . B . 3 was found on both introns and exons , we wondered whether upon transcriptional activation if both exons and flanking intronic sequences gained H2A . B . 3 equally well or whether compared to the repressed state , there was a preferential targeting of H2A . B . 3 to exons compared to introns . To examine this , we determined the relationship between the H2A . B . 3/input ratio and log expression averaged across all intron-exon boundaries . At each base position relative to the intron-exon boundary , a linear model was fit to this relationship and the slope of the fitted linear model was determined and plotted ( Fig 1d ) . The results clearly show that exons gain H2A . B . 3 at a higher rate compared to introns when genes are activated . The exonic H2A . B . 3 nucleosome is flanked on both sides by an H2A . B . 3 nucleosome depleted region , which is more pronounced at the intron to exon boundary compared to the exon to intron border ( Fig 1c ) . This can also be seen in the H2A . B . 3 meta-intron plot ( S2a Fig ) and when H2A . B . 3 ChIP-Seq reads were aligned with the exon—intron boundary ( S2c Fig ) . The H2A . B . 3 meta-exon plot revealed that this H2A . B . 3 nucleosome is located closer to the exon-intron boundary than the intron-exon boundary ( S2b Fig ) , and accordingly this asymmetry in the H2A . B . 3 nucleosome position can provide a simple explanation as to why the intron-exon boundary is more accessible . It is attractive to suggest that this nucleosome-depleted region at the intron—exon boundary could facilitate the access of the spliceosome to the nascent RNA . The gene body-associated H3K36me3 modification has been shown to be a modifier of splicing outcome [22 , 24] . To investigate the relationship between H3K36me3 , H2A . B . 3 and transcription , H3K36me3 ChIP-Seq experiments were performed . As expected , this modification is located at exons and is positively correlated with transcription ( although the top 25% of expressed genes display less H3K36me3 compared to moderately expressed genes , Fig 1e ) . A H3K36me3 heat map further illustrates this positive correlation with expression ( S1c Fig ) . Next , a Pearson correlation of the log coverage , calculated across 50 base pair windows , was used to determine if there is any correlation between the presence of H2A . B . 3 and H3K36me3 at each base pair position relative to the intron-exon boundary ( Fig 1f ) . Intriguingly , no correlation between H2A . B . 3 and H3K36me3 at the exon is observed . To examine this relationship further , we separated all intron-exon boundaries into 4 groups that contain very low , low , moderate or high levels of H2A . B . 3 . For each of the four groups , a single line represents the normalised H3K36me3 reads at each base pair aligned with the intron-exon boundary ( Fig 1g ) . This analysis shows that there is no correlation between the degree of trimethylation at H3K36 with increasing levels of H2A . B . 3 incorporation at the exon in the testis . Taken together , with the knowledge that H2A . B . 3 and H3K36me3 are enriched at different regions of the gene body , we suggest that H2A . B . 3 functions independently from H3K36me3 in the process of gene expression . In conclusion , based on the observation that H2A . B . 3 is found both at the TSS [3] and gene body , including the intron—exon boundary , of an active gene , we suggest that this variant may have more than one role in the process of expressing a gene . Further , given that splicing occurs co-transcriptionally , we explore below the possibility that H2A . B . 3 has a role in splicing by determining: ( 1 ) whether this histone variant is also found at intron—exon boundaries in the brain , ( 2 ) its relationship with the gene body repressive mark H2A . Z , ( 3 ) its link with exon inclusion , ( 4 ) whether H2A . B . 3 interacts with RNA processing factors , ( 5 ) whether it can directly interact with RNA and ( 6 ) the nuclear localisation of H2A . B . 3 and its position in relation to the RNA splicing machinery located at splicing speckles . H2A . B . 3 is also expressed in the mouse brain ( S3 Fig ) and therefore we wondered whether H2A . B . 3 is present at the TSS and gene body of genes active in this tissue . To investigate this , H2A . B . 3 ChIP-Seq and RNA-Seq experiments were repeated utilizing the hippocampus and then compared with the testis . Genes transcribed by RNA polymerase II were separated into groups according to their expression level ( repressed , low , medium and high ) . For each group of genes , a single line represents the normalised tag counts at each base pair , which has been aligned with the start site of transcription ( TSS ) ( ±1 kb ) . Similar to the mouse testis ( S4a Fig ) , a H2A . B . 3-containing nucleosome appears at ~ -50 base pairs relative to the TSS with increasing levels of transcription ( S4b Fig ) . Conversely , highlighting the fragile nature of this nucleosome as reported previously [3] , and that H2A . B . 3 is only present on a subset of active promoters ( see below ) , no input nucleosome is observed at an active TSS both in the testis ( S4c Fig ) and the hippocampus ( S4d Fig ) . Intriguingly though , the overall H2A . B . 3 organisation at an active promoter is different between the testis and the brain . In contrast to the testis , a second H2A . B . 3 nucleosome forms at ~ -200 base pairs relative to the TSS on an active promoter ( S4b Fig ) . A plot showing the distribution of H2A . B . 3 nucleosomes at the promoter as a heat map for both the testis and hippocampus , respectively illustrates this difference in more detail ( S4e and S4f Fig ) . Previously , we revealed that H2A . B3 was not present on all active promoters and most interestingly , gene ontology ( GO ) analyses revealed that H2A . B . 3 was particularly enriched on active genes involved in RNA processing and splicing [19] . To investigate this further , gene set enrichment analyses were performed ranked by the mean coverage of H2A . B . 3 over a fixed window size of 50 base pairs at successive distances from the TSS ( ±1 kb ) . Strikingly , this analysis revealed that the enrichment of H2A . B . 3 at the TSS for genes active in the hippocampus displays similar GO terms as active promoters associated with H2A . B . 3 in the testis ( RNA processing ( GO:0006396 ) , translation ( GO:0006412 ) and ribonucleoprotein complex ( GO:0030529 ) ; S4g and S4h Fig ) . Importantly , no such functional enrichment at the TSS was observed for input nucleosomes ( S4g and S4h Fig ) . The RNA-Seq data reveals that these H2A . B . 3 enriched gene sets show high expression , with a mean RNA-Seq expression 8 . 2 and 20 . 7 times greater than the overall mean expression across all genes , for the brain and testes , respectively . We conclude that H2A . B . 3 is a target of the TSS being positioned there both in the testis and the brain . Remarkably , H2A . B . 3 is associated with biological functions that are similar in the brain and testis . H2A . B . 3 is also found in the body of active genes in the hippocampus . Recapitulating the observations of the testis , we find that: ( 1 ) H2A . B . 3 is more enriched at the beginning of the gene body then the end ( S5a Fig ) , ( 2 ) the input nucleosome located at the exon is negatively correlated with transcription whereas H2A . B . 3 is positively regulated ( S5b and S5c Fig ) , ( 3 ) exons gain H2A . B . 3 at a faster rate compared to introns when genes are expressed ( S5d Fig ) , ( 4 ) incorporation of H3K36me3 at the intron—exon boundary is positively correlated with transcription ( S5e Fig ) and ( 5 ) there is no correlation between the presence of H2A . B . 3 and incorporation of H3K36me3 at the intron-exon boundary ( S5f and S5g Fig ) Finally , we investigated whether genes that have H2A . B . 3 enriched at the TSS also have this histone variant at the intron—exon boundary . To test for this correlated role , we separated all intron-exon boundaries into 4 groups that contain very low , low , moderate , or high levels of H2A . B . 3 at the TSS . For each of the four groups , a single line represents the normalised H2A . B . 3 reads at each base pair aligned with the intron-exon boundary ( S6 Fig ) . This analysis shows that there is positive correlation between the degree of incorporation at the TSS and the presence of H2A . B . 3 at the intron—exon boundary both in the testis and the brain ( S6a and S6b Fig ) . Therefore , transcriptional activation is associated with simultaneous H2A . B . 3 incorporation at the TSS and in the intron—exon boundary . This suggests that H2A . B . 3 may provide a link between transcriptional initiation and pre-mRNA splicing ( see below ) . H2A . Z is an essential histone variant that is believed to play an important role in establishing an active chromatin structure at promoters [25–28] . However , studies in in plants and C . elegans have shown that it is also located in the body of genes potentially being involved with repressing gene expression rather than activation [25 , 29] . To investigate whether H2A . Z is present within the body of genes in the mouse testis and brain , and its link with expression , ChIP-Seq experiments were performed and normalized H2A . Z ChIP-Seq reads were aligned with the intron—exon boundary , which were then ranked according to the level of gene expression . Significantly , a H2A . Z-nucleosome was observed on the exon and in contrast to H2A . B . 3 , H2A . Z was negatively correlated with transcription ( Fig 2a and 2b ) . A histone H2A . Z heat map illustrates this negative correlation with expression clearly ( S1d Fig ) . Not surprisingly then , there was also a negative correlation between H2A . Z incorporation at the intron—exon boundary with increasing levels of H3K36me3 ( Fig 2c ) . This result raises the intriguing possibility that there is a dynamic histone variant replacement process whereby during the activation of transcription , H2A . Z is lost from the intron-exon boundary being subsequently replaced with H2A . B . 3 . To demonstrate directly that H2A . B . 3 can replace H2A . Z on the same gene when it becomes activated during development , we used published gene expression data from Namekawa and colleagues [30] where they identified a small number of developmentally regulated genes on the X chromosome . These genes are repressed at the pachytene stage ( day 18 ) but become activated in round spermatids ( day 30 ) . We chose three such genes ( Akap4 , Il2rg and Akap14 ) and performed quantitative H2A . B . 3 and H2A . Z ChIP assays to examine the relative amount of these histone variants at 3 different exons for each gene ( Fig 2d ) . For all exons in all genes , the level of H2A . Z decreases with a corresponding increase in H2A . B . 3 when the genes become activated in round spermatids . Using these representative examples , these results clearly demonstrate that H2A . Z is associated with repressed genes in the coding region and upon transcriptional activation , H2A . B . 3 replaces it ( noting that we previously demonstrated that the targeting of H2A . B . 3 to these X-linked genes was concurrent with gene activation [3] ) . Next , quantitative H2A . B . 3 and H2A . Z ChIP assays were performed examining the relative amount of these histone variants at exons and neighbouring intronic sequences of genes either expressed more highly in the brain ( Ctnnd2 , Mpped1 and Ctnn1 ) or in the testis ( Pkib , Tbata and Slain2 ) ( Fig 2e and 2f ) . In all cases , the highest level of H2A . B . 3 at intron—exon sequences occurred when the gene was active irrespective of whether it is expressed in the brain or the testis ( Fig 2e and 2f ) . Conversely , all genes contain more H2A . Z when they were not expressed . As examples , Mpped1 has ~ 50 fold more H2A . B . 3 at its exon in the brain ( where it is expressed ) compared to the testis , ( Fig 2e ) . On the other hand , this exon has ~ 250 times more H2A . Z in the testis where this gene is not expressed compared to the brain ( Fig 2f ) . Similarly , Tbata has ~ 25 fold more H2A . B . 3 at its exon in the testis where it is expressed ( Fig 2f ) while it has 2 fold more H2A . Z in the brain where it is not expressed ( Fig 2e ) . These findings show that in a tissue specific manner , H2A . B . 3 replaces H2A . Z when a gene becomes activated verifying the genome-wide observations ( Fig 1c , S5c Fig , Fig 2a and 2b ) . The brain , followed by the testis , displays the greatest level of alternative splicing compared to any other tissue [31] . The next obvious question to address was whether the gain of H2A . B . 3 at the intron—exon boundary has a potential role in splicing or whether it is only linked to the process of transcriptional elongation as suggested by previous in vitro experiments [4] . To distinguish between these possibilities , we ranked all alternatively spliced exons into 4 groups dependent upon their inclusion levels ( very low , low , moderate or high; noting that alternatively spliced exons represent a minor population ( 13 . 4% ) compared to constitutive included exons ) ( Fig 3 ) . Significantly , a clear positive correlation exists between the degree of exon inclusion and the level of H2A . B . 3 at the intron—exon boundary in the testis with a similar trend in the hippocampus ( Fig 3a and 3b ) whereas input nucleosomes do not ( Fig 3c and 3d ) . Conversely , H2A . Z nucleosomes are negatively correlated with exon inclusion ( Fig 3e and 3f ) . H2A . B . 3 is not only found on alternatively spliced exons but is also present on constitutively included exons ( S7 Fig ) . These data argue that the presence of H2A . B . 3 at the intron—exon boundary has a role in the pre-mRNA splicing process .
No study to date has examined the function of H2A . B . 3 in its proper physiological contexts i . e . both in the testis and brain . Here we addressed this issue and have uncovered new locations for this histone variant on an active gene . The observed enrichment of H2A . B . 3 at the TSS and the beginning of the gene body is distinctively different compared to any other type of chromatin modification . Evidence is provided that H2A . B . 3 not only has a function in the initiation of transcription in the testis [3] ( Table 1 , Fig 4 ) and the brain ( S4 Fig ) , but also has a role in the processing of RNA ( Figs 3 , 4 and 5f , Table 1 ) thus providing a new link between transcriptional initiation and splicing . An unexpected histone variant replacement process was also uncovered whereby H2A . B . 3 replaces H2A . Z at intron—exon boundaries when a gene becomes active ( Fig 2d–2f ) . Previously , we reported that H2A . B . 3 might also replace H2A . Z at the TSS leading to higher levels of transcription suggesting that this histone variant replacement process may not be limited to intron—exon boundaries [19] . This is also consistent with the observation that in vitro , H2A . Z-containing nucleosome arrays are more refractory to transcription then H2A-containing arrays[4] . To date , H2A . Z has largely been viewed as an activator of transcription with its main function to assemble the TSS into an active chromatin structure [17 , 25–28] . Our results suggest that H2A . Z may have a different repressive function when incorporated into the body of a gene in the testis and brain . We also reveal a novel nuclear organisation in round spermatids where distinct and large domains of highly decondensed H2A . B . 3-containing chromatin exist , which co-localise with splicing speckles . Further , these H2A . B . 3-containing domains appear to be transcriptionally active based on its co-localisation with the initiation and elongation forms of RNA Pol II ( Fig 6 ) . This suggests that splicing speckles are not simply passive sites for the storage of splicing factors but participate in the transcription process , at least in highly transcriptionally active round spermatids . The presence of H2A . B . 3 at highly decondensed domains of chromatin in round spermatids is consistent with the in vitro ability of H2A . B . 3 to destabilise the nucleosome and inhibit chromatin compaction [3 , 4] . Perhaps the most unexpected finding of this study is the observation that H2A . B/H2A . B . 3 are RNA binding proteins consistent with a role in RNA processing and its association with mature transcripts in vivo ( Fig 5 ) . Further , this RNA binding ability appears to negatively regulate its capacity to interact with RNA Pol II and certain other RNA processing factors suggesting that a competition may exist between its capacity to bind to proteins or RNA . On the other hand , the interaction of at least one factor , Rent , was dependent upon RNA . While no representative RNA binding module exists , a common feature is a preponderance of arginine residues , which commonly occurs with serine and/or glycine residues [34] . These features are observed in the N-terminal tails of H2A . B/H2A . B . 3 ( Fig 5b ) . We conclude that H2A . B . 3 is a unique histone variant being able to bind to both RNA and DNA . It is attractive to speculate that these roles of H2A . B . 3 enables a special transcription/splicing mechanism to operate in the testis and the brain , two tissues known to display the highest level of splicing compared to other cell types [31] ( see below ) . What makes H2A . B . 3 a truly remarkably histone though , is not only its ability to bind to RNA but also its capacity to directly interact with proteins . Our mass spec analysis of immunoprecipitated H2A . B . 3-containing nucleosomes revealed an interaction with proteins involved in transcription and RNA processing . However , it was unclear whether these H2A . B . 3-protein interactions were direct because the chromatin was first cross-linked with formaldehyde . Therefore , we repeated these experiments without formaldehyde crosslinking followed by the mechanical shearing of germ cell nuclei ( Fig 5a ) . We demonstrated that indeed H2A . B . 3 could directly interact with RNA Pol II , Spt6 and other splicing factors ( which was greatly enhanced by the removal of RNA ) . On the other hand , the interactions with U1A and Sap18 were lost indicating that these interactions were indirect . A recent study over expressed epitope-tagged H2A . B in HeLa cells [15] . Perhaps not surprisingly , significant differences in H2A . B chromatin organisation are observed between Hela cells and what is observed here in the testis and brain ( major differences are seen at the TSS , at the beginning of the gene body and at the intron-exon boundary , and no role for RNA in regulating H2A . B-splicing factor interactions was observed ) . While also not observing interactions with RNA Pol II , Spt6 and many other factors , Tolstorukov and colleagues did observe an interaction between H2A . B and certain splicing factors ( but noting that , as shown here , some of these interactions may not be direct ) . Now in combination with our observations , this suggests that H2A . B has the intrinsic ability to interact with splicing factors even when expressed in a non-physiological setting , which could have important implications for the understanding of certain types of cancers [16] . A more recent study suggested that H2A . B . 3 is specifically deposited to methylated CpGs within the gene body in mouse ES cells to overcome methylation mediated repression of transcriptional elongation . While we cannot rule out that H2A . B . 3 may have a similar role in the testis and the brain , our results suggest that H2A . B . 3 may have a more universal role in facilitating gene expression that includes non-methylated DNA regions given our finding that H2A . B . 3 replaces H2A . Z during the gene activation process . At a genome-wide level , it has been clearly established that H2A . Z and DNA methylation are mutually antagonistic chromatin marks [39] . Further , for reasons that are unclear , this study also did not observe H2A . B . 3 at the TSS but this does suggest that mouse ES cells ( and Hela cells ) lack the H2A . B . 3 targeting mechanisms that operate in the testis and brain . As noted above , previously we demonstrated that H2A . B/H2A . B . 3 could destabilise the nucleosome , inhibit chromatin compaction and thus promote transcription in vitro [3 , 4] . Given that H2A . B . 3 is located at the TSS as well as in the body of an active gene in the testis and the brain , we suggest that these biophysical properties of H2A . B chromatin facilitates both transcription initiation and elongation . During spermatogenesis , the highest overall level of gene expression occurs at the round spermatid stage [18] , which correlates when H2A . B . 3 is maximally expressed . Previously , we showed that the targeting of H2A . B . 3 to the TSS , in a stage specific manner , is concurrent with the activation of previously silent genes on the X chromosome , and genes that were previously active became more highly expressed in round spermatids [3 , 17 , 19] . Here , we show that the exons of previously inactive X chromosome genes also gain H2A . B . 3 when they become activated in round spermatids , which is consistent with the observation that H2A . B . 3 incorporation at the TSS occurs concurrently with its deposition at the intron—exon boundary ( S6 Fig ) . On average , the abundance of H2A . B . 3 is the highest at the TSS compared to the gene body , and intriguingly , within the gene body , its occupancy is greatest at the beginning of the gene ( Fig 1a ) , which is where transcriptional elongation can be the most inefficient [40] . A current model for the role of chromatin in pre-mRNA splicing is that the nucleosome located at an exon acts as a barrier to slow the progress of the elongating RNA Pol II complex , which creates a window of opportunity for splicing factors to execute their splicing function [21 , 22] . However , as shown here , some gene bodies that are transcribed in the brain and the testis contain H2A . B . 3 nucleosomes , which , as discussed above , might facilitate rather than inhibit the progress of an elongating RNA Pol II complex . This argues that a different splicing process might operate in the testis and the brain . The results presented here show that H2A . B . 3 is not a highly expressed variant ( representing 3 . 7% of the total pool of H2A ) . Consistent with this , it is only found on a subset of active genes and most interestingly , the products of these genes are themselves involved in the processing and function of RNA ( S4 Fig ) [19] . Given that H2A . B . 3 is targeted to a gene concurrent with its activation [3] , this raises the possibility that H2A . B . 3 may be incorporated into the body of a gene as a result of transcriptional elongation . Also consistent with this notion is that H2A . B . 3 is associated with genes that are highly transcribed . Intriguingly though , H2A . B . 3 is not incorporated into the +1 nucleosome , a nucleosome that displays a high occupancy but is also proposed to have a high turn over rate [41] . Therefore , simply a higher rate of nucleosome turnover may not be sufficient to incorporate H2A . B . 3 suggesting that other , currently unknown , mechanisms may be involved in delivering H2A . B . 3 to the body of a gene . Alternatively , other mechanisms may prevent H2A . B . 3 from being incorporated into the +1 nucleosome . Based on our findings that H2A . B . 3: 1 ) is bound to chromatin ( Fig 6b ) , 2 ) is located in the body of an active gene ( Fig 1 ) , 3 ) interacts with RNA processing factors ( Fig 4 , Table 1 ) , 4 ) binds to RNA , which inhibits its interaction with RNA processing factors and RNA Pol II ( Fig 5a , 5c and 5d ) , 5 ) associates with mRNA in vivo ( Fig 5f and 5g ) , 6 ) co-localises with splicing speckles ( Fig 6 ) and 7 ) replaces H2A . Z concurrent with gene activation ( Fig 2 ) , we propose the following speculative model ( Fig 7 ) . Following the replacement of H2A . Z with H2A . B . 3 to assemble active chromatin , H2A . B . 3 directly recruits splicing factors from splicing speckles to an active gene . Upon transcriptional elongation and the synthesis of transcript RNA , H2A . B . 3 binds and ‘holds’ onto the RNA thus releasing the splicing factors to facilitate the splicing process . In conclusion , H2A . B . 3 expands the repertoire of histone functions by being involved in the processing and function of RNA .
Wild type 28–30 day old male Balb/c mice were used for all testis studies . 6–8 week old mice were used for Hippocampus ChIP and RNA-Seq experiments . Mice were housed according to animal ethics protocol at ANU animal facilities ( ANU , Canberra , Australia ) . The preparation of mononucleosomes from seminiferous tubules , as prepared for ChIP-Seq experiments , was carried out as described recently by us . 10–15μg of anti-H2A . B . 3 or anti-H2A . Z antibody was covalently bound to magnetic Dynabeads using the Dynabeads antibody coupling Kit ( Life Technologies ) and incubated with 100–200μg of formaldehyde crosslinked nucleosomes ( obtained by MNase I digest , see below ) in 50 mM Tris-HCl , pH 7 . 4 , 100 mM NaCl , 1% NP-40 , 0 . 1% SDS; 0 . 5% sodium deoxycholate , Roche protease inhibitor cocktail for 4 hours at 4°C with rotation . The immunoprecipitated histone variant protein complexes were washed twice in high-salt wash buffer ( 50 mM Tris-HCl , pH 7 . 4; 1 M NaCl; 1 mM EDTA; 1% NP-40; 0 . 1% SDS; 0 . 5% sodium deoxycholate ) , and twice with wash buffer ( 20 mM Tris-HCl , pH 7 . 4; 10 mM MgCl2; 0 . 2% Tween-20 ) . Beads were resuspended in 20 μl 1x NuPAGE loading buffer ( Life Technologies ) , containing ß-mercaptoethanol , and eluted proteins were loaded on a 4–12% PAGE for electrophoretic separation . The gel was fixed with mass spectrometry-compatible protein stain Instant Blue ( Expedeon ) . Each whole gel lane ( H2A . B . 3-IP or H2A . Z-IP ) was divided in 10–12 segments . The gel segments were dehydrated with acetonitrile ( 60μl per gel piece ) and then dried using a SpeedVac SC100 ( Savant ) . For mass spectrometry analysis , gel slices were destained , reduced and alkylated following the procedure described by Shevchenko et al . [42] . Samples were made up to 120 μl using 0 . 05 M NH4HCO3 , and 40 ng of trypsin ( Promega ) was then added to each gel slice . Samples were incubated for 16 h at 37°C . Each digest solution was removed to a new microfuge tube and the gel slices treated with the following solutions sequentially for 30 min each: 80 μl 0 . 1% ( v/v ) formic acid , 67% ( v/v ) acetonitrile and 80 μl 100% acetonitrile . The pooled digest and peptide extraction solutions for each sample were then dried ( Savant SPD1010 , Thermofisher Scientific ) before resuspending in 20 μl of 0 . 1% ( v/v ) formic acid . Proteolytic peptide samples were separated by nano-LC using an UltiMate 3000 HPLC and autosampler system ( Dionex , Amsterdam , Netherlands ) , and ionized using positive ion mode electrospray following experimental procedures described previously[43] . Single stage mass spectrometry and MS/MS were performed using an LTQ Orbitrap Velos Pro ( Thermo Electron , Bremen , Germany ) hybrid linear ion trap and Orbitrap mass spectrometer . Survey scans m/z 350–2000 were acquired in the Orbitrap ( resolution = 30 000 at m/z 400 , with an initial accumulation target value of 1 , 000 , 000 ions in the linear ion trap; lock mass was applied to polycyclodimethylsiloxane background ions of exact m/z 445 . 1200 and 429 . 0887 ) . Up to the 15 most abundant ions ( >5000 counts ) with charge states of >+2 were sequentially isolated and fragmented via collision induced dissociation ( CID ) with an activation q = 0 . 25 , an activation time of 30 ms , normalized collision energy of 30% and at a target value of 10 000 ions; fragment ions were mass analyzed in the linear ion trap . Peak lists derived from LC-MS/MS were generated using Mascot Daemon/ExtractMSn . exe ( Matrix Science , Thermo Electron ) and submitted to the database search program Mascot ( version 2 . 3 , Matrix Science ) [44] . The following search parameters were employed: instrument type was ESI-TRAP; precursor ion and peptide fragment mass tolerances were ±5 ppm and ±0 . 4 Da respectively; variable modifications included were carbamidomethyl ( C ) and oxidation ( M ) ; and enzyme specificity was trypsin with up to 2 missed cleavages . Searches were conducted using the Swiss-Prot database ( November 2013 release , 541762 sequence entries ) ; separate searches were conducted against all taxonomies and Mus musculus sequences only . Peptide identifications were considered to be high confidence if they were statistically significant ( p<0 . 05 ) according to the Mascot expect metric . Hypotonic spreads of male germ cells were prepared essentially as described[45] . Fixed cells were washed in PBS and then blocked for 1 hour with 3% BSA ( w/v ) in PBS at room temperature . The primary antibody , which was diluted with 1% BSA ( w/v ) , 0 . 1% Tween 20 ( v/v ) in PBS were applied to slides and incubated for 16 hours at 4°C in a humidity chamber . Following three washes with PBS , slides were incubated with fluorophore-conjugated secondary antibodies for 1 hour at room temperature . Following a further three washes with PBS , slides were incubated in 1 μM DAPI for 2 min . Vectashild ( Vecta laboratories ) was applied to prevent slides from photo bleaching . Round spermatids were isolated by sedimentation on a BSA gradient as described elsewhere [46] . Subcellular fractionation of round spermatids was carried out in LSBD buffer ( 50mM Hepes , pH 7; 3mM MgCl2; 20% glycerol ( v/v ) ; 1% NP40 ( v/v ) plus 250 mM KCl to obtain the cytoplasmic and nucleoplasmic fraction or 500mM KCl to obtain the fraction containing loosely bound chromatin proteins . The remaining material was designated as the chromatin fraction . Prior to immunoprecipitation with affinity purified H2A . B . 3 antibodies , the chromatin fraction was sonicated and treated with benzonase ( Millipore ) . Mononucleosomes were prepared from testicular tubules using Micrococcal nuclease I ( Mnase I ) digestion as described [3] . To prepare chromatin from hippocampal tissues , 10 Balb/c male mice ( 6–10 weeks old ) were decapitated and hippocampi were surgically removed into ice cold Hank's Balanced Salt Solution ( HBSS , Sigma ) buffered with 50mM Hepes pH-7 . 6 and supplemented with 0 . 2 mM PMSF and EDTA-free protein inhibitor cocktail ( Roche ) ) . Hippocampal slices were homogenised in a Dounce homogeniser with 5–10 strokes using pestle A with 4 ml of ice-cold HBSS . Cells were counted and 2–5 x 107 cells were fixed for 12 min with rotation at room temperature in 10 ml of fresh medium HBSS in the presence of 1 . 2% ( v/v ) formaldehyde . Mononucleosomes were obtained by digesting purified nuclei with 2–4 units of MNase I ( NEB ) at 37C . Total RNA was isolated using TRIzol ( Invitrogen ) and the RNeasy Mini kit ( Qiagen ) . Samples were treated with TURBO™ DNase ( Life Technologies ) . RNA-Seq libraries ( three biological replicates ) were prepared using NEBNext mRNA Library Prep kit using oligo-dT enrichment module ( New England Biolabs ) following manufacturer’s recommendations . Resulted RNA—Seq libraries were sequenced on HiSeq 2000 sequencer ( Illumina ) using 100 base pairs paired-end reads . Total germ cells from 28–30 day mice testis were UV-irradiated once with 75mJ/cm2 at 254 nm to crosslink RNA/protein complexes in a UV-irradiation oven , Stratalinker 1800 , ( Stratagene ) . 3-4x107cells were lysed in 2 ml of lysis buffer ( 50 mM Tris-HCl pH 7 . 6; 100 mM NaCl; 5mM MgCl2; 1% NP-40 , 1mM DTT , Roche protease inhibitor cocktail , pH 7 . 4 ) for 30 min on ice . Chromatin was mechanically sheared by passing nuclei through a 31G syringe . To remove insoluble material , the lysates were centrifuged at 10 , 000 g . Half of the supernatant was subjected to RNase I treatment ( 800 units , Ambion AM2294 ) for 30 min at 37°C , while the other half was treated in the same way but without the addition of RNase I . The degradation of RNA was monitored by the Qubit HS RNA assay . Both untreated and RNAase I-treated lysates were immunoprecipitated with H2A . B . 3 antibodies exactly as described for IP-MS . To distinguish IPs performed with formalin and UV-crosslinked cells from UV-only-crosslinked cells , we use the term CLIP ( UV-cross-linking immuno precipitation ) for the latter samples only . For CLIP assays , germ cell lysates were prepared in an identical manner as just described . 0 . 25ml of lysate was diluted in 1 ml of 50mM Tris , 100mM NaCl , 5mM MgCl2 , 1mM DTT for RNase treatement with 400 units of RNase I ( Ambion AM2294 ) for 30 min at 37°C . Treated lysates were immunoprecipitated with equal amount of anti-H2A . B . 3 and anti-H2A . Z antibodies ( 10μg each ) , bound to protein A/G dynabeads ( Thermo Fisher ) , for 2 hours . Following washing , twice with high salt wash buffer ( 50 mm Tris-HCl; 1M NaCl; 5mM MgCl2; 1% NP-40 , 0 . 1% SDS , 0 . 5% Na deoxycholate , pH 7 . 4 ) and twice with wash buffer ( 20mM Tris-HCL , 10mM MgCL2 , 0 . 2% Tween-20 , pH 7 . 4 ) ) , immunoprecipitated complexes were dephosphorylated on beads using PNK and than the RNA was 5' end labeled with P32 gamma-ATP . Eluted labeled complexes were electrophoresed through SDS-PAGE and transferred onto a nitrocellulose membrane . The same membrane was uses for a western analysis using anti-H2A . B3 and anti-H2AZ antibodies . Histone H2A-H2B ( and variant ) dimers were produced using standard protocols for refolding of histone complexes [48] . Either a 222 nt or 152 nt RNA was in vitro transcribed from the pcDNA 3 . 1 linearised plasmid template containing the RNA probe of interest [35] using a HiScribe T7 Quick High Yield RNA Synthesis Kit ( New England Biolabs ) . The RNA probe was heated at 95°C for 5 min , then rapidly cooled on ice for 2 min just prior to setting up binding experiments . The RNA ( 20 ng ) was then incubated on ice with the relevant histone dimers in binding buffer ( 10 mM MOPS , pH 7 . 5 , 200 mM NaCl , 5 mM MgCl2 , 10% glycerol ( v/v ) , 1 mM DTT , 0 . 03 mg/ml heparin ) . The binding reactions were analysed on 5% acrylamide 1x TB gels . Gels were stained with SYBR Gold Nucleic Acid Gel Stain and visualised using a Typhoon FLA 9000 . Biotinylated peptides corresponding to the N-terminal tails of histones H2A , H2A . Z , H2A . B , and H2A . B . 3 were purchased from GL Biochem at >90% purity . A Cy3-labelled 25 nt RNA ( Cy3-CAGCGACUCGGGUUAUGUGAUGGAC ) was purchased from Sigma-Aldrich . The biotinylated peptides ( 30 pmol ) were immobilised on steptavidin-coated M-280 Dynabeads ( LifeTechnologies ) and incubated with 150 pmol of the Cy3-labelled RNA in binding buffer for 1 hr at room temperature . The beads were then washed three times with binding buffer . Following the final wash , the beads were resuspended in Novex TBE-urea Sample Buffer ( Life Technologies ) , heated at 70°C for 2 min and then run on 15% Novex TBE-urea gels ( Life Technologies ) . Cy3 fluorescence was visualised on a Typhoon FLA 9000 . Antibodies used were as follows . Anti-H2A . B . 3[3] , Anti-H2A . Z [49] , Anti-H3K36me3 ( ab9050 ) , Anti-Smith Antigen [Y12] ( ab3138 ) ; Anti-H2A ( ab18255 ) , anti-H3 ( ab1791 ) , anti-RNA PolII ( phospho S2 ) ( ab5095 ) , anti-RNA PolII ( phospho S5 ) ( ab5131 ) , anti-Rent1 ( ab109363 ) , anti-Snrpa1 ( ab128937 ) , anti-U1A ( ab155054 ) , anti-SPT6 ( ab32820 ) , anti-Symplekin ( ab80274 ) , anti-SAP18 ( ab31748 ) and anti-Sf3b155 ( ab66774 ) , all from Abcam . Anti-DDX39A ( PA5-31220 , Pierce ) , anti-sheep-HRP ( AP324R; Chemicon ) , anti-rabbit-HRP ( AP322P; Chemicon ) . Chip-Seq reads were adaptor trimmed and mapped to the genome using Bowtie2[50] . Paired end reads were converted to single spanning fragments . Coverage bedGraphs were generated . Plots of the mean coverage anchored at genomic landmarks ( the intron—exon boundaryor the TSS ) for a span of +1000 to -1000 base pairs by aligning local coverage at these landmarks genome-wide using UCSC genes canonical transcript annotations . Units of RPM ( mean reads per base pair per million reads mapped ) were used normalised by the total mapped library size . Metagenes plots were generated by scaling exons to the same normalised x-axis from 0 to 1 . Intron—exon boundaries that are alternatively spliced in mouse hippocampus and testes were determined from paired-end RNA-Seq , which was adaptor , trimmed by Trimmomatic[51] in palindromic mode and mapped using Bowtie 2 . Alternatively spliced exons were called using the MATS software[52]; a constitutive set was formed by removing alternative sites that overlapped the UCSC annotations . To test whether H2A . B . 3 coverage over exons was correlated with alternative or constitutive splicing , the mean RPM H2A . B . 3 coverage was compared over the exons ( bases 0 to 150 downstream of the intron-exon junction ) of alternatively spliced versus and constitutively spliced axons . To ensure the sets were comparable and not confounded by possible biases toward a higher expression level for the alternative spliced set , the larger constitutive set was subsampled to the same size as the alternative spliced set such that it had a matched distribution of gene expression levels . Alternatively spliced exons were called using the Multivariate Analysis of Transcript Splicing ( MATS ) software . This was applied to stranded RNA-seq data ( three replicates per tissue ) . Skipped "cassette" exons were identified to define AS sites . A constitutive set was formed by removing these alternative sites from the full set of UCSC intron-exon annotations . Inclusion fraction relative to the flanking constitutive exons was estimated for each cassette exon by the MATS software . RNA-IP data was mapped with Tophat [53] using bowtie2 . Adapters were trimmed using Trimmomatic . To compare correlations between two ChIP-Seq data sets relative to the intron to exon boundary , we performed Pearson correlation between the vectors of mean coverage formed over a fixed window of size 50 base pairs at a given distance from the intron—exon anchor . To determine if exons versus introns gained H2A . B . 3 preferentially , the association between mean H2A . B . 3/input ratio ( in RPM ) and log expression from RNA-Seq data ( RPM ) was calculated for a 1kb flanking window around all intron-exon boundaries . At each base position relative to the intron exon-boundary , a linear model was fitted to the H2A . B . 3/input ratio versus gene log expression and the gradient of the fitted model plotted ( using the mean H2A . B . 3/input ratio value in a sliding 20 bp window around each bp ) . To determine if exons versus introns gained H2A . B . 3 preferentially , the association between mean H2A . B . 3/input ratio ( in RPM ) and log expression from RNA-Seq data ( RPM ) was calculated for a 1kb flanking window around all intron-exon boundaries . At each base position relative to the intron-exon boundary , a linear model was fit to the H2A . B . 3/input ratio versus gene log expression , using the mean H2A . B . 3/input ratio value in a sliding 20 bp window around each bp , and the slope of the fitted model plotted . When overlapping RNA-IP data with H2A . B . 3 ChIP-Seq data over gene bodies and promoters ( -1000 bp upstream of TSS ) , a minimum coverage of 30 reads per base was used , which was above background reads . To determine the percentage of H2A . B . 3 expression compared to the total H2A pool , the mean TPM was determined for all different H2A subtype genes expressed . | The substitution of core histones with their non-allelic variant forms plays a particular important role in regulating chromatin function because they can directly alter the structure of chromatin , and provide new protein interaction interfaces for the recruitment of proteins involved in gene expression . Despite being discovered over a decade ago , the function of H2A . B , a variant of the H2A class , in its proper physiological context ( being expressed in the testis and the brain ) is unknown . We provide strong evidence that H2A . B has a role in the processing of RNA . It is found in the gene body of an active gene , directly interacts with RNA polymerase II and splicing factors and is located in the nucleus at distinct regions enriched with RNA processing factors ( splicing speckles ) . Most significantly , we show that H2A . B can directly bind to RNA both in vitro and in germ cells . Therefore , H2A . B has the novel ability to bind to both RNA and DNA ( as well as proteins ) thus directly linking chromatin structure with the function of RNA . Taken together , this suggests that a special mechanism of splicing may operate in the testis and brain . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"sequencing",
"techniques",
"dna-binding",
"proteins",
"dna",
"transcription",
"germ",
"cells",
"epigenetics",
"molecular",
"biology",
"techniques",
"rna",
"sequencing",
"chromatin",
"sperm",
"research",
"and",
"analysis",
"methods",
"spermatids",
"chromosome",
"biology",... | 2017 | A new link between transcriptional initiation and pre-mRNA splicing: The RNA binding histone variant H2A.B |
Spatial regulation is often encountered as a component of multi-tiered regulatory systems in eukaryotes , where processes are readily segregated by organelle boundaries . Well-characterized examples of spatial regulation are less common in bacteria . Low-fidelity DNA polymerase V ( UmuD′2C ) is produced in Escherichia coli as part of the bacterial SOS response to DNA damage . Due to the mutagenic potential of this enzyme , pol V activity is controlled by means of an elaborate regulatory system at transcriptional and posttranslational levels . Using single-molecule fluorescence microscopy to visualize UmuC inside living cells in space and time , we now show that pol V is also subject to a novel form of spatial regulation . After an initial delay ( ~ 45 min ) post UV irradiation , UmuC is synthesized , but is not immediately activated . Instead , it is sequestered at the inner cell membrane . The release of UmuC into the cytosol requires the RecA* nucleoprotein filament-mediated cleavage of UmuD→UmuD′ . Classic SOS damage response mutants either block [umuD ( K97A ) ] or constitutively stimulate [recA ( E38K ) ] UmuC release from the membrane . Foci of mutagenically active pol V Mut ( UmuD′2C-RecA-ATP ) formed in the cytosol after UV irradiation do not co-localize with pol III replisomes , suggesting a capacity to promote translesion DNA synthesis at lesions skipped over by DNA polymerase III . In effect , at least three molecular mechanisms limit the amount of time that pol V has to access DNA: ( 1 ) transcriptional and posttranslational regulation that initially keep the intracellular levels of pol V to a minimum; ( 2 ) spatial regulation via transient sequestration of UmuC at the membrane , which further delays pol V activation; and ( 3 ) the hydrolytic activity of a recently discovered pol V Mut ATPase function that limits active polymerase time on the chromosomal template .
Replication of the Escherichia coli chromosome is carried out by replisomes: dynamic multi-protein complexes that coordinate genome duplication by DNA polymerase ( pol ) III [1] . Pol III replisomes are both exceptionally fast [1] and accurate [2] , but are inefficient at synthesising DNA on damaged templates [3] . High levels of DNA damage lead to replication-fork collapse , which can be lethal if not resolved . This situation is usually addressed by the approximately 40 genes of the bacterial SOS response , induced in two stages that reflect two different strategies for restoring replication . The earliest stage features induction of proteins involved in several pathways of error-free DNA repair . If this initial repair does not suffice to restart DNA replication , a later mutagenic process ensues with the induction of the umuDC operon [4] . This operon encodes the translesion synthesis polymerase , pol V [5] . Pol V is responsible for UV- and most chemical-induced chromosomal mutagenesis [6] . SOS mutagenesis also results in more rapid adaptation to stress and development of resistance to antibiotics in the absence of exogenous DNA damage [7 , 8] . Translesion DNA synthesis by pol V allows for resumption of replication on heavily damaged chromosomes , but dramatically increases mutation rates . Pol V is activated only after the cell’s capacity for non-mutagenic DNA repair has been exceeded . An elaborate regulatory regime should not be surprising , but the cellular constraints on pol V activity remain imperfectly understood . The active form of the enzyme , pol V Mut , is produced through a series of steps dependent on nucleoprotein filaments of RecA , denoted as RecA* , that are formed on single-stranded DNA after damage ( Fig 1A ) . The umuDC operon , which encodes the pol V precursors UmuD2 and UmuC , contains a particularly high-affinity binding site for the LexA repressor that limits transcription of the umuDC genes [9] . Intracellular levels of UmuD and UmuC are further kept to a minimum through Lon-mediated proteolytic degradation [10] . As a result UmuD and UmuC only accumulate approximately 30 min after DNA damage [4] . However , UmuD2 and UmuC are not active for DNA synthesis . UmuD2 must first undergo a RecA*-mediated autocatalytic cleavage reaction that removes the N-terminal 24 amino acid residues of each subunit to generate UmuD′2 . However , this reaction is inefficient and leads to the formation of UmuD/D' heterodimers . UmuD' in the UmuD/D' heterodimer is then rapidly targeted for degradation by the ClpXP protease [10] . As a consequence , UmuD' homodimers only accumulate in response to a persistent damage-inducing signal . UmuD'2 then associates with UmuC to form pol V ( UmuD′2C ) [5] . Pol V will only accumulate if the damage , and thus RecA* , persists until after the initial 45 minutes of error-free repair [4] . Pol V has weak catalytic activity in vitro [5 , 11] and is unable to promote translesion DNA synthesis in the absence of RecA in vivo [12] . To facilitate translesion synthesis , pol V must physically interact with RecA* and remove a single RecA-ATP molecule from the 3'-proximal end of the filament to form the mutagenic and highly active pol V Mut ( UmuD'2-UmuC-RecA-ATP ) [13 , 14] . Direct observation of the various activation steps of pol V inside living cells with fluorescence microscopy would allow for the testing of various hypotheses and potentially the construction of a global model for mutasome activity . However , analysis of pol V activity in vivo has long proven difficult due its very low expression levels [15] . Here , we report the use of single-molecule fluorescence microscopy to directly visualize the dynamics of pol V inside living cells . Surprisingly , we observe that pol V precursors are temporarily sequestered at the cellular membrane , adding another prominent factor to those limiting pol V activity during early stages of the DNA damage response . Spatial regulation of a DNA processing enzyme is unprecedented in bacteria and suggests the presence of several layers of spatiotemporal partitioning to regulate the molecular processes underlying genomic maintenance .
In order to visualize the regulation of pol V , we altered the umuDC operon so that the bright red fluorescent protein mKate2 is fused to the C-terminus of UmuC ( Fig 1B ) . Western blotting of the chromosomally expressed UmuC-mKate2 protein revealed expression of the chimeric protein , albeit at steady-state levels somewhat lower than untagged UmuC ( Fig 1C , S1A Fig ) . Nevertheless , the chimeric umuC-mKate2 allele promoted both spontaneous and damage-induced reversion of the hisG4 allele , confirming it is functionally active ( Fig 1D , S1B Fig ) . The level of mutagenesis promoted by UmuC-mKate2 was lower than that of wild-type UmuC , which is consistent with the overall lower steady-state levels of the chimera in the cell . We imaged E . coli K12 MG1655 UmuC-mKate2 cells on a home-built wide-field single-molecule fluorescence microscope . To monitor changes in the cellular levels and location of UmuC as a function of time after UV-induced damage , E . coli was immobilized inside flow cells and a series of time-lapse images were recorded . We developed a novel flow-cell design that allowed for imaging of individual cells with single-molecule sensitivity , while also providing space for cells to grow into filaments ( a hallmark of the SOS response ) and allowing for in situ UV irradiation . Cells were irradiated with UV light from a mercury lamp ( fluence = 1–100 J/m2 , λ = 254 nm ) and UmuC-mKate2 fluorescence was recorded over a course of 3h , imaging once every 5 min . We first measured changes in fluorescence intensity using time-lapse measurements , allowing us to monitor the production of UmuC-mKate2 ( Fig 2A and 2B ) . At all UV doses , very little expression of UmuC-mKate2 was observed during the first 30 min following UV irradiation ( Fig 2A and 2B ) , in good agreement with the previously measured value of 30 min required for expression of UmuC and UmuD [4] . Furthermore , UmuC-mKate2 levels increased at all doses after this initial delay , albeit with different kinetics . While at doses ≤ 3 J/m2 , UmuC-mKate2 levels gradually increased throughout the duration of the experiment; at higher doses the protein levels decreased after reaching a maximum at 90–120 min . At the highest dose , 100 J/m2 , UmuC-mKate2 was produced more slowly than at 30 J/m2 and ultimately reached lower levels , presumably due to inhibition of protein synthesis as the result of extensive DNA damage . The amount of UmuC-mKate2 inside cells was quantified using time-sampling measurements ( Fig 2C ) . Similar to the time-lapse experiments , cells were grown , irradiated and imaged within flow cells . Rather than periodically recording single fluorescence images to monitor the response of the same set of individual cells in time , we recorded video-rate movies of UmuC-mKate2 fluorescence using a new population of cells for every time point after UV irradiation . This procedure provided higher-quality intensity information but led to significant photobleaching . We therefore chose a new field of view with unbleached cells for each new measurement . Movies ( 296 × 34 ms frames ) were recorded over 10 fields of view every 30 min for 3 h , capturing between 70 and 236 cells . For each field of view we also recorded a bright-field image , allowing us to define the outline of each cell . The fluorescence movies were used to determine the concentration of UmuC-mKate2 inside each cell ( S2 Fig ) . Plotting the average cellular concentration of UmuC-mKate2 over time , we observed a sharp increase ( Fig 2C ) , with dynamics consistent with those measured by time-lapse measurements ( Fig 2B ) . At 30 J/m2 , the concentration of UmuC-mKate2 increased from 0 . 9 ± 0 . 04 nM ( S . E . M , n = 236 cells ) to 3 . 7 ± 0 . 3 nM ( n = 70 ) , 180 min after irradiation with UV light . These concentrations correspond to an increase from an average of 1 . 6 ± 0 . 08 ( n = 236 ) molecules per cell to 15 . 7 ± 1 . 8 ( n = 70 ) molecules per cell at 180 min . The numbers of UmuC molecules measured by fluorescence microscopy ( 16 per cell ) are lower than measurements obtained by Western blotting of wild-type cells with anti-UmuC antibodies ( ~ 60 per cell ) [4 , 15] , but again , this observation is consistent with overall lower intracellular steady-state levels of the UmuC-mKate2 chimera compared to the wild-type UmuC protein ( Fig 1C ) . Quantification of UmuC-mKate2 levels in cells periodically sampled from shaking culture showed similar expression dynamics and concentration levels as cells grown in flow cells: no UmuC-mKate2 is produced during the first 30 min , then levels rise to 3 nM in the period 40–120 min after irradiation ( S3 Fig ) . The fast time-sampling movies showed foci of UmuC-mKate2 that move within a thin band along the cell periphery ( Fig 3A , S1 movie ) . This observation suggested that many molecules of UmuC associate with the cell membrane , diffusing slowly enough to result in sharply defined foci on a single image ( 34 ms frame duration ) , but sufficiently mobile to show movement on longer timescales . To examine this apparent membrane association further , we used a plasmid encoding a fluorescently labelled inner membrane protein ( LacY-eYFP ) as a reference for the position of the membrane . Super-resolution images were produced for both UmuC-mKate2 and LacY-eYFP , which revealed that many of the UmuC-mKate2 foci were indeed located with LacY on the cell membrane ( Fig 3B ) . Control experiments with cells expressing free mKate2 ( including the same linker as the UmuC-mKate2 construct ) confirmed that membrane association is a property of UmuC , as opposed to mKate2 or the linker region ( S4 Fig ) . In both the time-lapse and time-sampling datasets , visual inspection of the movies indicated a change in the localization behaviour of UmuC during the course of the DNA-damage response . For the first 90–120 min after UV exposure , UmuC-mKate2 was predominantly located at the cell periphery , while in later images UmuC-mKate2 was distributed throughout the cytosol ( Fig 3C , S5 Fig ) . We developed a novel autocorrelation-based approach to quantify the localization behaviour of UmuC across all cells in the time-lapse measurements . Autocorrelation of images can be conceptualized as follows . First , the cell image to be analyzed is duplicated and superimposed with the original ( schematically depicted in Fig 3D ) so that the intensities of each pixel in the two images are perfectly correlated . One of the images is then shifted sideways in increments ( presented in the figures as the distance lag ) relative to the other image and the correlation between pixel values is re-calculated . This procedure is repeated over a range of different shift sizes ( or distance lags ) to produce an autocorrelation function . The highest correlation of pixels is seen when the duplicate images are perfectly superimposed ( zero lag ) . If significant amounts of UmuC are concentrated in the membranes , additional signal peaks will be seen at two non-zero shift increments where the right membrane of one image passes over the left membrane of the other , and vice versa ( schematically shown in Fig 3D ) . Cytosolic UmuC with its distribution centered along the central cellular axis should not generate such secondary peaks . To illustrate this approach , we simulated images of E . coli cells with a width of 0 . 6 μm , expressing cytosolic or membrane-localized fluorescent proteins . Cells containing cytosolic signal gave rise to an autocorrelation function with a single , broad peak at zero lag ( Fig 3D ) . Cells with membrane-associated fluorescence instead produced a narrow peak at zero lag ( perfect superimposition ) , as well as two distinctive cross peaks when the duplicate images were shifted 0 . 6 μm , arising from correlations between signals on the two sides of the cell . To test if this autocorrelation approach could be used to monitor changes over time in cellular distribution of UmuC-mKate2 , we simulated a time-lapse series with an E . coli cell transitioning from a state with no signal , to a membrane-associated signal , to a cytosolic signal ( Fig 3E; left ) . We graphed autocorrelation functions measured at each time point of the simulation as a 2D contour plot , with blue-to-red coloring indicating low-to-high levels of correlation ( Fig 3E; right ) . At each time point , autocorrelation functions were normalized to the fluorescence intensity of the cell . The amplitude of the central peak ( zero lag ) thus reflects relative changes in fluorescent protein concentration: the amplitude is low ( blue ) at lower concentrations and high ( red ) at high cellular concentrations . When present , membrane-bound fluorescent protein gives rise to weaker secondary signals ( light blue ) to either side of the central peak . Analysis of experimentally acquired data produced similar results: time-lapse images of LacY-eYFP ( an inner membrane protein ) produced an autocorrelation plot with a strong central peak and weaker secondary peaks , while DnaX-YPet ( which is cytosolic/nucleoid associated ) produced only a strong central peak ( S6 Fig ) . Our autocorrelation approach thus provides a means to monitor the distribution of fluorescent proteins across the short-axis of the cell during time-lapse measurements: membrane-associated protein produces distinctive cross-peaks , whereas cytosolic/nucleoid associated protein does not . We then applied autocorrelation analysis to our experimentally acquired UmuC-mKate2 time-lapse images ( Fig 3F ) , beginning at the time in which cells were irradiated at the UV dose indicated . Within the time-lapse images , the orientation of E . coli cells was biased by the flow of medium through the flow cell: the majority of cells aligned with the vertical axis of the images . This orientation bias allowed us to look for the repeating pattern of membrane-localized UmuC-mKate2 signals by measuring image-wide autocorrelation functions across the horizontal axis of the images . The autocorrelation functions were normalized by cellular intensity across the entire dataset of UV doses . At 1 J/m2 , there was little change in the autocorrelation function during the time-course and little detectable membrane localization . At 3 J/m2 , a modest signal from membrane-localized UmuC appears late in the time-lapse . At 10 J/m2 , two distinctive membrane-bound cross peaks appear at about 90 min and persist until 180 min . At 30 J/m2 and 100 J/m2 , membrane localization is prominent from 90 min . From 120–180 min , the autocorrelation is significantly broader , with the valley between the central peak and the membrane peak becoming filled in . This broadening is indicative of UmuC entering the cytosol . Evidence of redistribution is also visible in cross-correlation analysis between UmuC-mKate2 and the inner membrane protein LacY-eYFP ( Fig 4A ) . We found that expression of the LacY-eYFP protein partially induced the SOS response: cells were longer than usual and produced an elevated level of UmuC-mKate2 prior to UV irradiation . At early time-points after UV irradiation , clear cross-peaks are visible due to co-localization of UmuC-mKate2 and LacY-eYFP on the membrane . Each UmuC-mKate2 focus correlates with LacY-eYFP signal on both the same side of the cell ( producing a central peak at lag = 0 ) and the opposite side of the cell ( producing cross-peaks at lag = ± 0 . 8 μm ) . From 90–180 min after irradiation , these cross-peaks angle in towards the central peak and become broader . This is consistent with UmuC-mKate2 foci forming in the cytosol , where the distance from each focus to either side of the membrane falls between 0 and 0 . 8 μm . As UmuC-mKate2 is no longer confined to the membrane , the distance to each side of the membrane becomes more variable , causing the cross-peaks to broaden . The redistribution of native chromosomally expressed UmuC into the cytosol was also followed by Western blotting of whole-cell extracts ( WCE ) and soluble fractions at various time points after UV irradiation . ( Fig 4B ) . As previously reported , basal levels of UmuD and UmuC in a recA+ lexA+ strain are undetectable [15] . The affinity purified UmuC antibodies do , however , cross-react with a protein that is slightly smaller than UmuC ( S1A Fig ) that can be detected in both the WCE and soluble fractions ( Fig 4B ) . Importantly , the solubility of this cross-reacting protein does not change after UV . We can therefore use this cross-reacting protein as an internal control for the amount of cellular material loaded in each lane . We begin to observe the appearance of soluble UmuC 30 minutes after UV irradiation . The amount of UmuC in the soluble fraction increases further 1 hour and 2 hours post-UV , consistent with the release of UmuC into the cytosol . The extent of soluble UmuC shows an excellent correlation to the amount of soluble UmuD' in the cell . Fifteen minutes post-UV , we observe the induction of UmuD and very little conversion to UmuD' . By 30 minutes post-UV , roughly 50% of the UmuD is converted to UmuD' in the WCE . Much more UmuD' is found in the soluble fraction , indicating that UmuD' is inherently more soluble than UmuD . The amount of UmuD' in the soluble fraction continues to accumulate 1 hour and 2 hours post-UV . Finally , mapping the cellular location of individual UmuC molecules visualized in cells grown in shaking culture also indicates redistribution of UmuC-mKate2 from the membrane to the cytosol beginning 90 min after UV irradiation ( S7 Fig ) . These results confirm that the membrane localization is observed independent of whether the cells grow in culture or on cover slips , and independent on whether autocorrelation analysis is used or direct mapping . We observed that at large doses of UV light ( > 30 J/m2 ) , many cells produced a single large burst of UmuC-mKate2 and that the change from membrane-associated to cytosolic distribution seemed to initiate when the level of UmuC reached a maximum . This maximum always occurred more than 90 min after UV . To examine this phenomenon further , we cropped movies of single cells from our time-lapse series ( 100 J/m2 ) and post-synchronized them each individually to the time point with the maximum fluorescence intensity ( Fig 5A and 5B ) . Where appropriate , the cells were also rotated to align with the y-axis of the image to further enhance the sensitivity of autocorrelation analysis . These post-synchronized movies reveal that during bursts of production ( lasting typically 30–60 min ) UmuC is membrane-associated and gradually becomes cytosolic after reaching peak intensity ( Fig 5A ) . Autocorrelation analysis on the post-synchronized movies revealed three distinct phases in the behaviour of UmuC ( Fig 5C and 5D ) . In phase I , very little UmuC is produced and those molecules that are present associate with the cell membrane . In phase II , substantially more UmuC is produced and these molecules accumulate at the membrane . In phase III , production ceases and UmuC redistributes into the cytosol . The redistribution of UmuC into the cytosol is essentially complete 30 min after the cell reaches peak intensity . The fact that UmuC only redistributes to the cytosol during phase III and after its production has ceased indicates that UmuC is being released from the membrane , as opposed to newly synthesized UmuC becoming sequestered in the cytosol . The change in abundance of soluble UmuC exhibits a strong correlation to the appearance of UmuD' ( Fig 4B ) . To test the hypothesis that the interaction between UmuD' and UmuC allows UmuC to relocate to the cytosol , we employed a umuD ( K97A ) strain that cannot convert UmuD to UmuD′ [12] . Compared with the wild-type background ( Fig 6A ) , UmuC-mKate2 in the umuD ( K97A ) strain appeared at about the same time after UV irradiation ( Fig 6B ) . However , in this strain the UmuC was more strongly membrane-associated in images and produced stronger and more persistent membrane cross-peaks upon autocorrelation analysis ( Fig 6B ) . Note that UmuC is stabilized in the cell when in a complex with UmuD′ [10 , 16] , such that degradation of UmuC by the Lon protease limits the formation of UmuC signals when UmuD′ cannot be formed . This indicated a strong tendency for UmuC-mKate2 to accumulate at the membrane , both before and after UV irradiation . Analysis of umuD ( K97A ) cells grown in shaking culture similarly show a strong bias for UmuC-mKate2 to be localized on the cell membrane ( Fig 6B ) . Together , these observations support the hypothesis that conversion of UmuD to UmuD′ is critical for re-localisation of UmuC . To further test this hypothesis , we determined the location of UmuC-mKate2 in a recA ( E38K ) ( also known as recA730 ) background , in which active RecA nucleoprotein filaments ( RecA E38K* ) are formed constitutively in the absence of DNA damage [17] . Under these conditions , the SOS regulon is partially derepressed and much of the temporal regulation normally imposed on pol V is circumvented: pol V is constitutively formed through UmuD cleavage and activated to pol V Mut . However , the ability of RecA ( E38K ) to spontaneously form nucleoprotein filaments depends upon the cellular concentration of the RecA mutant [15] and RecA ( E38K ) can be further activated under various conditions [18] . Indeed , UV treatment results in the apparent induction of UmuC in the recA ( E38K ) lexA+ background , where all LexA-regulated proteins are expected to already be expressed in the absence of DNA damage . As expected , and in contrast to the recA+ lexA+ , background ( Figs 3–5 and 6A ) , in the recA ( E38K ) lexA+ strain UmuC was present at all times after irradiation ( Fig 6C ) . The UmuC signal increased over the first 90 min after irradiation , with the brief appearance of a signature for membrane-associated UmuC-mKate2 at the peak of UmuC production ( ~90 min ) . This observation suggests that higher levels of UmuC after UV treatment went to the membrane first and then became cytosolic . While recA ( E38K ) cells efficiently cleave UmuD to UmuD' in the absence of DNA damage , UmuD' can only be generated from full-length UmuD and it is possible that there is sufficient uncleaved UmuD in these cells to transiently allow binding of UmuC to the membrane . The same cells grown in shaking culture show a general cytosolic distribution of UmuC-mKate2 before and 10–120 min after UV irradiation . These observations again indicate that fully activated pol V Mut is largely cytosolic and that the membrane-localized state corresponds to an intermediate species preceding pol V Mut formation . To rule out the possibility that the cytosolic localization of UmuC-mKate2 in the recA ( E38K ) strain is due to constitutive production of RecA E38K* as opposed to UmuD cleavage , we then determined the location of UmuC-mKate2 in a recA ( E38K ) umuD ( K97A ) background , in which UmuC and UmuD are produced constitutively but UmuD cannot be cleaved to form UmuD' . As already indicated , overall UmuC levels decline because of Lon-dependent degradation of UmuC in the absence of UmuD′ . In this strain , UmuC-mKate2 is seen at the membrane at all time points , increasing after UV induction to a higher and persistent level 90 min after UV induction ( Fig 6D ) . Autocorrelation analysis revealed clear membrane cross-peaks throughout the measurement , indicating that in this background UmuC-mKate2 remains associated with the membrane in both the presence and absence of damage . The membrane association in this strain is corroborated by a clean membrane-associated distribution of UmuC-mKate2 both before and 10–120 after UV irradiation seen in the same cells in shaking culture ( Fig 6D ) . The membrane localization of wild-type UmuC when co-expressed with a non-cleavable UmuD protein was also confirmed directly by immuno-electron microscopy ( Fig 7A and 7B ) . Furthermore , in experiments measuring the solubility of wild-type UmuC in E . coli cell extracts , there appeared to be significantly less soluble UmuC when it was co-expressed with UmuD ( K97A ) than when expressed alone , or with UmuD' ( Fig 7C ) . This is consistent with an active role for UmuD in sequestering UmuC in an insoluble membrane-bound fraction , and/or a requirement for UmuC association with UmuD' to effect release to the cytosol . Together , these results provide clear evidence that conversion of UmuD to UmuD' is a requirement for the spatial relocalization of UmuC to the cytosol . Having revealed a novel membrane binding and release mechanism underlying the regulation of pol V , we next studied the positioning of pol V molecules after they were released from the membrane , in the context of translesion synthesis . For many years , the prevailing model for pol V-catalysed translesion synthesis was one that described a highly concerted series of events: ( i ) replisomes stall at lesion sites , exposing ssDNA; ( ii ) RecA* filaments form on the ssDNA , triggering the SOS response and activating pol V; ( iii ) pol V displaces pol III through mass action and synthesises over the lesion; ( iv ) replisomes restart and the cells return to normal . However , a series of recent in vitro observations has brought this model into question . First , RecA*-dependent activation of pol V can occur in trans , i . e . , remote from the site of DNA synthesis [13 , 19] . Second , replisomes do not necessarily stall at lesions , but can instead skip over them leaving a ssDNA gap in their wake [20] . Finally , the presence of RecA* on the template increases the displacement of pol III by translesion synthesis polymerases , indicating additional complexity above a simple mass action-driven exchange mechanism [21] . Our single-molecule observations allow us to address whether translesion synthesis takes place at replication forks by simultaneously imaging UmuC fused to the red mKate2 and the τ subunit of the DNA pol III holoenzyme fused to the yellow YPet; DnaX-YPet ) [22] . We recorded two-color time-lapse series of wild-type recA+ cells and constitutively active recA ( E38K ) cells following irradiation with UV light at 30 J/m2 . We then measured cross-correlation functions between the two image channels as a function of time . This procedure was similar to the autocorrelation-based one described above , except that image color pairs were correlated with each other rather than each image being correlated against itself . Image pairs containing co-localized fluorescent spots should return a strong , sharp cross-correlation signature , whereas spatially unrelated spots should return weak or no cross-correlation . In the wild-type background , a broad and relatively weak cross-correlation peak was observed , suggesting little co-localization between DnaX-YPet and UmuC-mKate2 molecules beyond their presence within the same cell ( Fig 8A ) . In the recA ( E38K ) background , stronger cross-correlation was observed across a relatively narrow distance regime , consistent with co-localizing spots of diffraction-limited size ( Fig 8A ) . However , the cross-correlation declines with time after UV treatment . This observation suggests that in contrast with the wild-type background , a significant proportion of UmuC-mKate2 molecules ( primarily in the form of pol V Mut in this background ) co-localize with replisomes in the recA ( E38K ) background under normal growth , but that the co-localization declines at later times after UV irradiation . To probe this co-localization further , we analysed individual cells within the time-sampled movies . Pol V Mut synthesizes DNA at a very low rate ( 0 . 3–1 nt·s-1 ) [11] . It can therefore be assumed that pol V Mut molecules forming active mutasomes must remain immobilized on the DNA for at least a few seconds . These molecules should present as static foci in our fluorescence movies ( at 34 ms per frame ) . Making average projections of these movies allowed us to highlight static foci while blurring both transient DNA binders and mobile membrane-associated molecules into the diffuse cellular background ( Fig 8B ) . Foci that persisted for > 300 ms were identified as static foci . Static foci were observed in UV-irradiated , wild-type recA+ cells as well as in both untreated and UV-irradiated recA ( E38K ) cells ( Fig 8B ) . In both backgrounds , the number of static foci increases in response to UV-irradiation in a dose-dependent fashion ( Fig 8C ) . In the recA ( E38K ) background , a large proportion of static foci co-localize with replisomes in the absence of UV ( Fig 8D ) . These observations strongly support the notion that the static foci we observe represent DNA-bound pol V Mut molecules . From this point forward we will refer to these static foci as mutasome foci . In the wild-type background , mutasome foci appear in the cytosol 90 min after damage was induced ( Fig 8B and 8C ) , 30 min after the point where UmuC levels begin to increase ( Fig 2 ) . On average , cells contained 1 . 1 ± 0 . 2 mutasome foci at the peak time of 150 min . Very few of these co-localized with replisomes: the percentage detected as being co-localized fell within the range expected by chance ( based on the area of the cell occupied by replisome foci , ~5–10% , Fig 8D ) . These observations are consistent with the weak cross-correlation signature we observed in the time-lapse analysis ( Fig 8A ) . In the recA ( E38K ) strain , pol V Mut is produced constitutively and is evenly distributed throughout the cytosol both before and after inducing damage ( Fig 6B , S8 Fig ) . Prior to UV irradiation , static foci are observed in the cytosol , averaging 2 . 3 ± 0 . 2 per cell ( Fig 8B and 8C ) . In contrast to the behavior observed in wild-type cells , a significant proportion of these foci ( 27 ± 3% , n = 111 cells ) co-localize with replisomes ( Fig 8D ) . This observation is consistent with the strong cross-correlation observed between DnaX-YPet and UmuC-mKate2 in the corresponding time-lapse measurements ( Fig 8A ) and indicates that pol V Mut gains access to replisomes under these conditions . This observation is consistent with the elevated rate of mutagenesis ( ~ 100-fold ) observed in the recA ( E38K ) ( recA730 ) background in the absence of damage [17] . As seen in the cross-correlation ( Fig 8A ) , we observe that after UV irradiation , the co-localization with replisomes gradually decreases ( Fig 8D ) . By 90 min after UV , relatively few pol V Mut foci colocalize with replisomes ( 16 ± 2% , n = 84 ) and by 150 min co-localization ( 11 ± 1% , n = 90 ) approaches levels expected to occur by chance ( ~5–10% , Fig 8D ) . The proportion of replisomes that co-localized with a static pol V focus remained approximately the same throughout the measurement , indicating that pol V Mut gains similar levels of access to replisomes in both the presence and absence of damage in the recA ( E38K ) background ( Fig 8E and 8F ) . Analysis of cells grown in shaking culture returned similar results: there is no significant co-localization of mutasomes and replisomes in the wild-type background; a significant proportion of mutasomes do co-localize with replisomes in undamaged recA ( E38K ) cells; this co-localization gradually diminishes after UV irradiation ( S9 Fig ) . Overall , our results show that mutasomes form at replisomes in undamaged recA ( E38K ) cells in which the SOS response is constitutive , but few sites for DNA synthesis exist outside of replisomes . After damage , additional mutasomes are assembled at sites that are spatially distinct from replication forks . In the wild-type background , static foci rarely form at replication forks , clearly questioning the canonical view of mass-action driven binding of pol V at stalled replication forks .
A complete list of the Escherichia coli strains used in this study appears in Table 1 . It is possible to determine the number of molecules of a fluorescent protein in a cell by dividing its total fluorescence by the average signal arising from individually measured single molecules . When measuring cells that only contain a small number of fluorescent proteins , however , individual fluorescence images tend to contain a lot of noise . We therefore decided on a different strategy for analysis of our time-sampling data . Having removed contributions from background fluorescence , we fit the loss of fluorescence signal due to photobleaching of mKate2 with an exponential decay function and used the fit amplitudes as a measure for the initial fluorescence . In this way , the total fluorescence of the cell is derived from measurements over all 296 frames of the movie , reducing the contribution of noise dramatically . The concentration of UmuC-mKate2 could then be determined for each cell using its volume as calculated from its corresponding bright field image . For each all fluorescence movies we first flattened the images to correct for uneven illumination profile of the excitation light . To produce and of the illumination beam we averaged together each slice from all 70 movies recorded during a 3h UV-damage experiment , smoothed the resulting image by Gaussian blurring ( 10 pixel radius ) and normalised the intensity as a fraction of the brightest pixel . This showed that illumination was already relatively flat ( at the corners of the image was 70% as intense as that at the centre of the image ) . We then divided each movie by this “beam” movie . For each cell we measured the change in mean fluorescence per pixel due to photobleaching . The raw data included contributions not only from UmuC-mKate2 , but also cellular autofluorescence ( S1A Fig ) and background fluorescence from the glass coverslip ( S1B Fig ) , which were removed prior to quantitating mKate2 fluorescence . Under our conditions , mKate2 follows a single-exponential photobleaching decay with time constant t ≈ 5 s ( S1C Fig ) . We quantified the total mKate2 fluorescence of individual cells by fitting the amplitude of each decay curve . We note that by focussing on the middle portion of the cell , signal arising from the top and bottom of the cell may not be completely captured and thus measured signals may be slightly underestimated . Cell outlines were assigned using MicrobeTracker [58] . To calibrate our measurements for number of UmuC-mKate2 molecules per cell , we measured the fluorescence signals of individual molecules . To do this , we examined the final 50 frames of our movies . Here most of the cellular fluorescence had been bleached , however individual molecules could occasionally be observed as foci as they spontaneously returned to a bright state . We fit these foci with 2D Gaussian functions and made a histogram of the integrated volumes under each curve ( S1D Fig ) . We could then calibrate our cellular measurements by dividing the total fluorescence of each cell by the mean value for single molecules . | Escherichia coli , and many other bacteria , respond to high levels of DNA damage with an inducible system called the SOS response . In this response , bacteria first try to restart replication using non-mutagenic DNA repair strategies . If that fails , replication can be restored using DNA polymerases that simply replicate over DNA lesions , a desperation strategy that results in mutations . DNA polymerase V ( pol V ) is responsible for most mutagenesis that accompanies the SOS response . Because of the risk inherent to elevated mutation levels , pol V activation is tightly constrained . This report introduces a new layer of regulation on pol V activation , with a novel spatial component . After synthesis , the UmuC subunit of pol V is sequestered transiently at the membrane . Release into the cytosol and final activation depends on the activity of RecA protein and the autocatalytic cleavage of UmuD to generate the UmuD' subunit of pol V . The resulting delay in activation represents an additional molecular mechanism that limits the amount of time that this sometimes necessary but potentially detrimental enzyme spends on the DNA . | [
"Abstract",
"Introduction",
"Results",
"Discussion"
] | [] | 2015 | Regulation of Mutagenic DNA Polymerase V Activation in Space and Time |
Rabies is endemic in Sri Lanka , but little is known about the temporal and spatial trends of rabies in this country . Knowing these trends may provide insight into past control efforts and serve as the basis for future control measures . In this study , we analyzed distribution of rabies in humans and animals over a period of 12 years in Sri Lanka . Accumulated data from 1999 through 2010 compiled by the Department of Rabies Diagnosis and Research , Medical Research Institute ( MRI ) , Colombo , were used in this study . The yearly mean percentage of rabies-positive sample was 62 . 4% ( 47 . 6–75 . 9% ) . Three-fourths of the rabies-positive samples were from the Colombo , Gampaha , and Kalutara districts in Western province , followed by Galle in Southern province . A high percentage of the rabies samples were from dogs ( 85 . 2% ) , followed by cats ( 7 . 9% ) , humans ( 3 . 8% ) , wild animals ( 2 . 0% ) , and livestock ( 1 . 1% ) . Among wild animals , mongooses were the main victims followed by civets . The number of suspect human rabies cases decreased gradually in Sri Lanka , although the number of human samples submitted for laboratory confirmation increased . The number of rabid dogs has remained relatively unchanged , but the number of suspect human rabies is decreasing gradually in Sri Lanka . These findings indicate successful use of postexposure prophylaxis ( PEP ) by animal bite victims and increased rabies awareness . PEP is free of charge and is supplied through government hospitals by the Ministry of Health , Sri Lanka . Our survey shows that most positive samples were received from Western and Southern provinces , possibly because of the ease of transporting samples to the laboratory . Submissions of wild animal and livestock samples should be increased by creating more awareness among the public . Better rabies surveillance will require introduction of molecular methods for detection and the establishment of more regional rabies diagnostic laboratories .
Each year 55 , 000 people die of rabies throughout the world , more than 31 , 000 of these deaths occur in Asia [1] , [2] . Sri Lanka is one Asian country where human deaths from rabies has decreased markedly during the past decade [3]; however , rabies is endemic and remains a significant public health problem in this country . Sri Lanka is a tropical island state situated in the Indian Ocean near the southern tip of India . The current population is about 20 . 2 million of this 18 . 3% living in urban areas , 77 . 3% in rural areas , and 4 . 4% living in estate areas ( http://www . statistics . gov . lk ) . According to the Census Ordinance ‘Estate” has been defined as areas with plantations where there are 20 or more acres in land and 10 or more resident laborers . Sri Lanka is one of the fastest growing economies in Asia and only 4 . 3% of the population is living under poverty line . The Sri Lankan Ministry of Health spends a substantial amount of its health budget on anti-rabies treatment for humans . Recent estimates that the cost of post-exposure prophylaxis ( PEP ) per patient is US$173 without immunoglobulin and US$177 with equine immunoglobulin [4] . This includes all direct medical costs associated with PEP . As in other canine rabies-endemic countries , dogs are the main transmitter of rabies to humans in Sri Lanka [5] . Recently , the confirmation of sylvatic rabies virus in a civet submitted to Medical Research Institute ( MRI ) for diagnosis from Moneragala district [6] sparked discussion about whether there are other reservoir species for sylvatic rabies in Sri Lanka and whether these animals are widespread throughout the country or localized to a particular area . To explore other reservoir species of rabies viruses in Sri Lanka , the epidemiology of rabies virus variants must be analyzed in different animals and their geographical locations recorded . The rabies endemic regions may expand or contract as a result of virus transmission and animal population interactions [7] . Therefore a long term study focusing on the molecular epidemiology of rabies in Sri Lankan would provide a better understanding of the epidemiology of rabies throughout the country . Among several factors , natural and anthropometric factors have a direct impact on animal population dynamics [8] , and geographic features may act as barriers or corridors for the spread of rabies [9] . Additionally , unusual animal-dispersal patterns and human-mediated translocation of infected animals have resulted in the unexpected introduction of rabies into new areas [8] . Occasionally , transmission of rabies virus variants to new hosts can perpetuate regionally , and these variants can become enzootic in new reservoir species [8] . All mammals are susceptible to rabies , but some animal species do not serve as a reservoir species for rabies virus and therefore do not normally play a role in transmission of the virus . A clearer understanding of the reservoir species for rabies would help to improve rabies prevention and control in Sri Lanka , a country rich in animal diversity . In recent years , Sri Lanka has experienced population growth , rapid urbanization , deforestation , and construction of new highways , dams , and irrigation systems . All of these changes can affect reservoir species habitat and may influence the epidemiology of rabies in different ways . This study was performed to identify the trends of rabies infection in Sri Lanka during the past 12 years by analyzing the number of rabies specimens received at the MRI , the national rabies laboratory in the country . We expect that the findings will be useful for formulating appropriate strategies to strengthen rabies control activities in Sri Lanka .
The study used accumulated data from 1999 through 2010 collected by the Department of Rabies Diagnosis and Research , Medical Research Institute ( MRI ) , Colombo , which is the national reference laboratory for human and animal rabies diagnosis in Sri Lanka ( Figure 1 ) . There are two regional rabies diagnostic laboratories in Peradeniya and Karapitiya for testing animal samples only . The Peradeniya laboratory was established in 2007 and is attached to the Faculty of Veterinary Medicine and Animal Science , University of Peradeniya . Peradeniya has about 50 , 000 inhabitants and is a suburb of Kandy the administrative capital of Central province . The Karapitiya laboratory is attached to the Karapitiya teaching hospital , Karapitiya , Galle . Galle is the administrative capital of Southern province and has a population of about 90 , 000 . During the tsunami in December 2004 , the Karapitiya laboratory was destroyed , therefore all samples from Galle were sent to MRI . The Karapitiya laboratory is currently functioning modestly after reconstruction and is gradually increasing its capacity to commence on the florescence antibody test ( FAT ) . Before the tsunami and after the reconstruction , samples that were reported as having inconclusive results , and samples that were reported as negative but clinically were suggestive of rabies were sent from Karapitiya laboratory to MRI for confirmation . A number of FAT inconclusive samples were also sent from Peradeniya laboratory to the MRI for confirmation of the results . Therefore a large number of samples sent from Karapitiya and Peradeniya to the MRI were included in this study . From the regional laboratories , brain samples were transported in ice box . Therefore , the samples received at the MRI were representative for all of Sri Lanka . Animal rabies is not a notifiable disease in Sri Lanka . For animal samples , the head of the suspected animal was submitted to the laboratory , and the brain was dissected by a trained person . Animals submitted for testing were exhibiting clinical signs suggestive of rabies with or without a history of biting humans and/or animals . The clinical signs of rabies included aggressive behavior , biting tendency , and/or excessive salivation . Samples from animals suspected of being rabid are usually submitted by general public , a small number of samples are also submitted by veterinarians and forest rangers . There are no community veterinarians in Sri Lanka . To evaluate the correlation between the mean yearly numbers of rabies positive samples in different provinces and their population or population density of 2012 , the Pearson correlation test was performed . Human rabies is notifiable in Sri Lanka . Human brains from clinically diagnosed rabies patients were sent by the Judicial Medical Officers in their respective districts . The samples were packaged without preservative and the cold chain was maintained during transportation . FAT was used as the reference test to confirm the diagnosis of rabies in all human specimens and was performed immediately after the samples were received [10] . In the MRI , in the analysis of animal and human brains , FAT is performed on crushed smears of hippocampus and brain stem , respectively . FAT is performed free of charge and test results are reported by hard copy to the person who submitted the sample . In the case of an emergency , such as when PEP is required , the patient is informed by telephone . Annual data are also reported to the Ministry of Health . All brain samples that tested positive by FAT for the presence of rabies virus were reported to the authorities as being from a rabies infected animal . Similarly , when a human brain sample from a suspected rabies case was positive for rabies by FAT that patient was designated as a laboratory confirmed human case . The number of samples and their results were , and continue to be updated regularly in the laboratory's database . Data on human deaths from suspect rabies were collected from the website of the Public Health Veterinary Services , Ministry of Health , Sri Lanka ( www . rabies . gov . lk ) . A suspect rabies case was defined as having agitation , hydrophobia , aerophobia , photophobia , or hypersalivation plus a history of animal bite . To evaluate the correlation between the number of suspect human rabies deaths and Sri Lankan government's annual expenditure on PEP , the Pearson correlation test was performed . Data on the financial expenditure for PEP was obtained from the Medical Supplies Division , Ministry of Health , Sri Lanka . Data were available from 2005 . ArcGIS 10 . 0 software [Environmental Systems Research Institute , Inc . ( ESRI ) , USA] was used to illustrate surveyed areas in different districts of Sri Lanka . The institutional review committee of the MRI approved the study . The data used in this study were anonymized .
During the 12 years of this retrospective study , a total of 12 , 452 samples were received at the MRI , 7 , 519 of which were FAT positive ( Table 1 ) . Each year an average of , 626 . 6 ( 465–787 ) of samples were FAT positive . There was a decreasing trend in the number of positive samples between 1999 and 2002 followed by an increase that peaked in 2006 and then decreased again . By contrast the percentage of positive samples decreased gradually until 2005 ( 47 . 9% ) , increased in 2006 ( 60 . 8% ) , where it remained at approximately the same level until it decreased again in 2010 ( 46 . 4% ) . The yearly mean percentage of positive samples was 62 . 1% ( 47 . 9–75 . 6% ) . Samples were received from 24 of 25 districts in nine provinces of Sri Lanka . When classified by district , most rabies-positive samples were submitted from Colombo , Gampaha , Galle and Kalutara ( Table 3 ) . When classified by province ( Table 4 ) , the numbers of positive samples ( percentage of the total positive samples from all provinces ) /the numbers of submitted samples during the 12 years originated from , 5 , 701 ( 75 . 8% ) /9 , 837 Western province; 896 ( 11 . 9% ) /1076 Southern province; 321 ( 4 . 3% ) /553 North Western province; 291 ( 3 . 9% ) /493 Sabaragamuwa province; 137 ( 1 . 8% ) /227 Central province; 90 ( 1 . 2% ) /146 Uva province; 49 ( 0 . 6% ) /77 North Central province; 28 ( 0 . 4% ) /35 Eastern province; and 6 ( 0 . 1% ) /8 Northern province . The mean yearly numbers ( range ) of positive samples ( Table 4 ) were found in the following order: 475 . 1 ( 385–605 ) from Western province , 74 . 7 ( 7–149 ) from Southern province , 26 . 7 ( 4–63 ) from North Western province , 24 . 2 ( 4–59 ) from Sabaragamuwa province , 11 . 4 ( 3–28 ) from Central province , 7 . 5 ( 1–21 ) from Uva province , 4 . 1 ( 0–7 ) from North Central province , 2 . 3 ( 0–5 ) from Eastern province , and 0 . 5 ( 0–3 ) from Northern province . The Pearson correlation coefficient value of R indicating that there is a strong correlation between the mean yearly numbers of rabies positive samples in different provinces and their population ( R = 0 . 9458 ) or population density ( R = 0 . 9759 ) . Livestock that were confirmed positive for rabies included pigs , cows , horses , buffalo , and goats . Wild animal species included mongooses , polecats , squirrels , foxes , monkeys , rabbits , bandicoots , wild cats , jackals , rock squirrels , and deer . The mean yearly numbers ( range ) of positive animal samples were 555 . 2 ( 392–704 ) for dog , 51 . 2 ( 28–78 ) for cat , 7 . 3 ( 1–17 ) for wild animal , and 12 . 8 ( 9–16 ) for livestock ( Table 5 ) . Among the livestock ( Table 6 ) , cows ( 110 ) had the highest number of cases of rabies followed by goats ( 26 ) , buffalo ( 11 ) , pigs ( 4 ) , and horses ( 3 ) . Among the wild animals , mongooses ( 37 ) had the highest number of cases followed by civet ( 12 ) and other animals ( Table 6 ) . The mean annual number of mongooses confirmed positive was 3 . 1 . The numbers of rabies positive mongoose samples were highest in 2001 ( 7 ) and 2002 ( 10 ) . In other years number decreased to 0–3 samples ( Figure 2 ) . The distribution of total animal and wild animal rabies cases from 1999 to 2010 was plotted by district on maps ( Figure 3 and 4 ) . There was a yearly fluctuation in the number of rabies cases when all animal submission were grouped together . From 2004 , more cases of rabies were confirmed in the areas from the eastern part of the country . From 2007 , rabies was confirmed in the northern district of Jaffna . Time-line mapping ( Figure S1 and S2 ) and cumulative mapping ( Figure 3 and 4 ) with the data available at the MRI showed that rabies was in high prevalence in the districts of Western and Southern provinces . By contrast , although there was a yearly fluctuation of the number of confirmed wild animal rabies cases , wild animal rabies was confirmed more frequently in the districts of Western , Southern , and Central provinces . No wild animal rabies cases were confirmed from other provinces . The number of laboratory-confirmed human rabies specimens increased from 18 in 1999 to 32 in 2010 ( Figure 5 ) . The lowest number of confirmed human rabies cases was in the years 2000 and 2003 when 15 human cases were detected . The highest number of human rabies cases ( 42 cases ) was confirmed in 2009 . Each year , an average of 24 . 7 ( 15–42 ) human samples was confirmed positive for rabies . The number of samples received increased from 20 in 1999 to 33 in 2010 . The lowest number submitted in 1999 and the highest number of received samples ( 46 samples ) in 2009 . From 1999 to 2003 an average of 25 samples per year were received in the MRI , from 2004 to 2010 the average number increased to 36 . 8 samples per year . According to the Ministry of Health there is a trend toward an overall decrease in the number of human deaths from suspect rabies from 113 in 1999 to 49 in 2010 . From these data , the mean number of human deaths from suspect rabies per year was calculated as 72 . 7 ( 49–113 ) . The Sri Lankan government expenses on PEP for the year 2005 , 2006 , 2007 , 2008 , 2009 , and 2010 were 302 . 9 . 204 . 2 , 197 . 8 , 286 . 9 , 378 . 6 , and 351 . 4 million Sri Lankan Rupees , respectively . The Pearson correlation coefficient value of R was −0 . 5432 , indicating that there is a moderate negative correlation between the annual number of human deaths from suspect rabies and annual expenditure of Sri Lankan government on rabies PEP . Samples were received from 18 of 25 districts in nine provinces of Sri Lanka . When classified by district , rabies-positive samples were mainly submitted from Colombo , Gampaha , Galle and Kalutara ( Table 7 ) . When classified by province , the numbers of positive samples ( percentage of the total positive samples from all provinces ) /the numbers of submitted samples during the 12 years were submitted in the following order: 86 ( 29 . 0% ) /118 from Western province , 60 ( 20 . 3% ) /85 from Southern province , 37 ( 12 . 5% ) /47 from North Western province , 27 ( 9 . 1% ) /33 from Uva province , 22 ( 7 . 4% ) /23 from Sabaragamuwa province , 22 ( 7 . 4% ) /29 from Central province , 20 ( 6 . 8% ) /24 from North Central province , 15 ( 5 . 1% ) /16 from Eastern province , and 7 ( 2 . 4% ) /8 from Northern province . The mean yearly numbers ( range ) of positive samples were confirmed from provinces in the following order: 7 . 2 ( 3–15 ) from Western province , 5 . 0 ( 0–13 ) from Southern province , 3 . 1 ( 0–9 ) from North Western province , 2 . 2 ( 1–4 ) from Uva province , 1 . 8 ( 0–6 ) from Sabaragamuwa province , 1 . 8 ( 0–3 ) from Central province , 1 . 7 ( 0–5 ) from North Central province , 1 . 2 ( 0–6 ) from Eastern province , and 0 . 6 ( 0–3 ) from Northern province .
The number of cases of dog rabies has remained stable , indicating that the decrease in the number of human deaths from rabies reflects the government's efforts in implementing improved access to PEP to people exposed to suspected rabies infection . Public awareness of rabies might be another factor contributing to the decreasing trend in the number of human rabies . Fully integrated dog-control activities to cover the whole country would help decrease the incidence of dog rabies significantly in Sri Lanka . Establishment of regional rabies laboratories in other provinces will improve the surveillance of rabies throughout the country . Public awareness should be increased to understand the effects of rabies in livestock and wild animals . Greater cooperation from forest rangers and personnel dealing with wild animals is necessary to understand the overall picture of sylvatic rabies in Sri Lanka . It is of critical importance to perform genetic typing of samples submitted for rabies testing throughout Sri Lanka to find out whether multiple rabies virus variants are circulating within the country . This would be extremely valuable for describing the epidemiology , making PEP recommendations , and developing a comprehensive rabies elimination strategy for Sri Lanka and could serve as a model for other Asian countries . | Rabies is a public health concern in Sri Lanka . The incidence of dog rabies remains unchanged , but the incidence of suspect human rabies is decreasing gradually in Sri Lanka . This finding indicates the effects of improved access to postexposure prophylaxis by animal bite victims and increased rabies awareness . As in other rabies-endemic countries , in Sri Lanka , human rabies is transmitted mainly by dogs , although domestic and wild animals have been diagnosed rabid , and can pose a risk of exposure to humans . In this study , we analyzed 12 years of data accumulated in the national reference laboratory of Sri Lanka to identify the trends of rabies in this country . This study showed that rabies has been recorded mainly in Western and Southern Provinces of Sri Lanka , possibly because of the ease of communication with rabies diagnostic laboratories from these areas . Regional rabies diagnosis laboratories should be established to improve surveillance of rabies in Sri Lanka . There were few submitted animal samples from livestock and wild animals , indicating that greater awareness is needed among the public regarding the need to submit suspect rabid animals for diagnostic evaluation . These data could help policy makers improve rabies prevention and to control rabies in Sri Lanka . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"disease",
"surveillance",
"infectious",
"diseases",
"rabies",
"plant",
"science",
"veterinary",
"diseases",
"zoonoses",
"medicine",
"and",
"health",
"sciences",
"epidemiology",
"disease",
"surveillance",
"plant",
"pathology",
"biology",
"and",
"life",
"sc... | 2014 | Twelve Years of Rabies Surveillance in Sri Lanka, 1999–2010 |
The intracellular accommodation structures formed by plant cells to host arbuscular mycorrhiza fungi and biotrophic hyphal pathogens are cytologically similar . Therefore we investigated whether these interactions build on an overlapping genetic framework . In legumes , the malectin-like domain leucine-rich repeat receptor kinase SYMRK , the cation channel POLLUX and members of the nuclear pore NUP107-160 subcomplex are essential for symbiotic signal transduction and arbuscular mycorrhiza development . We identified members of these three groups in Arabidopsis thaliana and explored their impact on the interaction with the oomycete downy mildew pathogen Hyaloperonospora arabidopsidis ( Hpa ) . We report that mutations in the corresponding genes reduced the reproductive success of Hpa as determined by sporangiophore and spore counts . We discovered that a developmental transition of haustorial shape occurred significantly earlier and at higher frequency in the mutants . Analysis of the multiplication of extracellular bacterial pathogens , Hpa-induced cell death or callose accumulation , as well as Hpa- or flg22-induced defence marker gene expression , did not reveal any traces of constitutive or exacerbated defence responses . These findings point towards an overlap between the plant genetic toolboxes involved in the interaction with biotrophic intracellular hyphal symbionts and pathogens in terms of the gene families involved .
Most land plant species feed carbon sources to arbuscular mycorrhiza ( AM ) fungi , which in turn deliver phosphate and other nutrients via finely branched intracellular structures called arbuscules [1] . The accommodation of fungal symbionts inside living plant cells involves substantial developmental reprogramming of the host plant cell , initiated by the formation of the pre-penetration apparatus , a transvacuolar cytoplasmic bridge that guides the fungal invasion path [2 , 3] . The oomycete downy mildew pathogen Hyaloperonospora arabidopsidis ( Hpa ) develops intracellular feeding organs , so called haustoria , which exhibit certain structural similarities to the arbuscules of AM fungi [4] . Like arbuscules , haustoria are accommodated inside plant cells and are entirely surrounded by a plant-derived membrane , which keeps the pathogen physically outside the host cytoplasm [5–10] . Structural and functional similarities between accommodation structures for intracellular microbes raised the hypothesis that the corresponding symbiotic and pathogenic associations rely on a similar or even overlapping genetic program [4] . This would imply that filamentous hyphal pathogens exploit an Achilles heel , the presence of the symbiotic program in most land plant species , for their own parasitic lifestyle [11] . In the present study , we tested this hypothesis by focussing on an ancient genetic program comprising the common symbiosis genes ( CSGs ) [12] , which is conserved among angiosperms for the intracellular accommodation of AM fungi [13 , 14] . In legumes and actinorhizal plants , the CSGs are also required for root nodule symbiosis ( RNS ) with nitrogen-fixing rhizobia or actinobacteria , respectively [15–17] . In the legume Lotus japonicus , the products of some of the CSGs including the MLD-LRR-RK Symbiosis Receptor-like Kinase SYMRK [16 , 18–20] as well as the nuclear-envelope localised cation channel POLLUX [21 , 22] , are implicated in a signal transduction pathway leading from the perception of microbial signalling molecules at the plasma membrane to the induction of calcium oscillations in and around the nucleus ( calcium-spiking ) , which finally results in the transcriptional activation of symbiosis-related genes and a developmental reprogramming of host cells [1 , 23] . Furthermore , the nucleoporins NUP85 , NUP133 and SEC13 homolog ( SEH1 ) of the NUP107-160 subcomplex are important for the proper establishment of RNS and AM symbiosis [24–27] . The NUP107-160 complex has been implicated in the transport of membrane proteins to the inner membrane of the nuclear envelope [28] . It has been hypothesised that mutations in the NUP107-160 complex may cause a reduced cation channel ( e . g . POLLUX ) and calcium channel concentration on the inner envelope , which is in accordance with the lack of nuclear calcium spiking in the NUP107-160 mutants ( Jean-Michel Ané , personal communication ) . Arabidopsis thaliana belongs to the Brassicaceae , a plant lineage that lost specific CSGs and with this the ability to establish AM after the divergence of the Brassicales [29–31] . A . thaliana can , however , establish a compatible interaction with the hyphal pathogen Hpa . The striking similarities between arbuscules and haustoria led us to investigate whether the interaction of A . thaliana with Hpa , the fungal powdery mildew pathogen Erysiphe cruciferarum , which forms intracellular haustoria in A . thaliana epidermal leaf cells , or the extracellular bacterial pathogen Pseudomonas syringae rely on a similar gene set as the interaction of plants with beneficial hyphal microbes .
We inspected the A . thaliana genome for the presence of genes related to those required for the interaction with beneficial microbes . Several of the orthologs of key genes required for AM , for example SYMRK , CCaMK and CYCLOPS , are specifically deleted from the A . thaliana genome [14 , 29–32] , and thus inaccessible for analysis in the context of the A . thaliana interaction with Hpa . However , we identified the ortholog of nuclear-localised cation channel POLLUX and members of the NUP107-160 subcomplex ( S1 and S2 Figs; S1 Table ) . Furthermore , we identified candidate genes belonging to the MLD-LRR-RK family as close relatives of LjSYMRK [16] , which we consequently named SYMRK-homologous Receptor-like Kinase 1 ( ShRK1 ) and ShRK2 ( S1 Fig , S2 Fig , S3 Fig , S4 Fig; S1 Table ) . Lines carrying insertional mutant alleles of ShRKs , selected members of the NUP107-160 subcomplex , and the Arabidopsis ortholog of POLLUX ( SNUPO genes ) were analysed for their phenotype in the interaction with Hpa . In pollux , shrk1 , shrk2 , the shrk1 x shrk2 double mutant , as well as a reference line with a T-DNA insertion in Phytosulfokine Receptor 1 ( pskr1 ) [33 , 34] or a reference line with an EMS-induced mutation in Constitutive expression of PR genes 5 ( cpr5 ) [35] , the reproductive success of Hpa isolate Noco2—measured as sporangiophore number per cotyledon or spore production—was significantly reduced compared to wild-type plants ( Fig 1A and 1C; S5 Fig ) . We analysed a range of NUP107-160 subcomplex members and found that the sec13 and nup133 single mutants and the sec13 x nup133 double mutant impaired Hpa reproductive success ( Fig 1A and 1C; S5 Fig ) , while lines with T-DNA insertions in nup43 , nup85 , nup160 and seh1 did not ( Fig 1B ) . All SNUPO genes analysed are constitutively expressed in Arabidopsis leaves , in epidermal and mesophyll cells with no consistent clear up- or downregulation in the context of Hpa infection [36 , 37] . The reduced Hpa reproductive success could not be explained by a decreased frequency of haustoria formation measured as percentage of host cells contacted by hyphae that harbour an haustorium . Other than a slight decrease in pollux , the frequency in the other mutants was indistinguishable from the wild-type ( S6 Fig ) . All A . thaliana SNUPO mutants , but not nup43 , nup85 , nup160 and seh1 mutants , exhibited a significant shift in the ratio of haustoria morphologies ( Fig 2; S7 Fig ) . At 5 days post infection ( dpi ) the majority of haustoria in leaves of the wild-type had a globular , single-lobed appearance . Deviations from such morphology , typically characterised by multiple lobes on a single haustorium ( “multilobed” ) , were observed as well . The amount of multilobed haustoria in the SNUPO mutants was significantly increased , a phenomenon that was alleviated in the tested complementation lines ( Fig 2A and 2B ) . In contrast to the SNUPO mutants , the pskr1 mutant previously reported to antagonistically impact defense responses to biotrophic and necrotrophic phathogens [34] , did not show any signs of altered haustoria development ( Fig 2A ) . In both , wild-type and shrk1 x shrk2 , the percentage of multilobed haustoria increased over time , but was significantly higher in shrk1 x shrk2 at each analysed time point ( Fig 2D ) . To test whether multilobed haustoria are the result of the harsh chemical clearance of the leaf and trypan blue staining procedure , we visualised the haustoria shape in vivo . We made use of a stable transgenic A . thaliana line expressing RPW8 . 2-YFP , a fluorescent protein fusion that accumulates around Hpa haustoria [38] ( Fig 2 ) and could confirm that multilobed haustoria appeared already in living plant cells ( Fig 2 ) . L . japonicus SYMRK belongs to the LRR I-RK family of receptor kinase genes , which consists of 50 members in A . thaliana [39 , 40] . Of these , Impaired Oomycete Susceptibility 1 ( IOS1 ) has been implicated in defence-related signalling [40 , 41] . On the one hand , IOS1 supports the infection of an obligate biotrophic oomycete pathogen , but on the other hand , is important for the resistance against bacteria [40 , 41] . Interestingly , we could only observe enhanced frequency of multilobed haustoria in an ios1 mutant of the A . thaliana ecotype Ler infected with Waco9 but not in an ios1 mutant of Col-0 infected with Noco2 , indicating that the impact of IOS1 on the Arabidopsis/Hpa interaction depends on the genotypes involved in the encounter ( S8 Fig ) . To investigate the specificity of the role of SNUPO genes in plant-pathogen interactions , we infected SNUPO mutants with the haustoria-forming powdery mildew fungus E . cruciferarum . We did not observe a consistent reduction in the reproductive success of E . cruciferarum ( Fig 3; S9 Fig ) . A morphological comparison of haustoria shape was not attempted because of the much more complex and highly variable haustoria morphology in this interaction . Plant evolution resulted in several defence strategies against microbial pathogens . For example , PAMP triggered immunity ( PTI ) is initiated by the perception of pathogen-associated molecular patterns ( PAMPs ) or plant-derived damage-associated molecular patterns ( DAMPs ) and results in plant defence responses accompanied by the activation of defence marker genes or callose deposition [42] . To analyse whether the increased Hpa resistance of the SNUPO mutants is a result of deregulated defence responses , we investigated the basal transcript levels of the three PAMP-induced marker genes—Pathogenesis-Related gene 1 ( PR1 ) , a marker gene for salicylic acid ( SA ) -mediated resistance [43] , Ethylene Response Factor 1 ( ERF1 ) , a marker gene for ethylene-mediated resistance [44] and the plant defensin PDF1 . 2a , a marker gene for jasmonic acid ( JA ) - and ethylene-mediated resistance [45] in the absence of pathogenic attack . Transcript levels in the tested mutants under non-challenged conditions did not differ from transcript levels in the wild-type ( Fig 4A , black circles ) . To examine whether SNUPO mutants show enhanced activation of defence marker genes , we analysed the transcript levels of PR1 , ERF1 and PDF1 . 2a in plants infected with Hpa . All tested genes were upregulated to a similar extent in the SNUPO mutants and the wild-type ( Fig 4A , empty circles ) . Only PDF1 . 2a transcript levels were significantly lower in Hpa-infected shrk1 x shrk2 plants than in infected wild-type plants , however , it is unlikely that this is the reason for the increased pathogen resistance of the double mutant . Moreover , the ability of Hpa to suppress callose deposition around the haustorial neck region [46] was not disturbed in the SNUPO mutants compared to the wild-type ( Fig 4B ) . In addition to the expression levels of defence marker genes , we investigated other symptoms typically associated with deregulated immune responses . We could not observe any developmental or growth defects , which are typically a result of the hyper-activation of the SA-dependent defence pathway [35] , in SNUPO mutants grown side-by-side with the wild-type and the dwarf mutant suppressor of npr1-1 , constitutive 1 ( snc1 [47] ) ( S10 Fig ) . Furthermore , we analysed constitutive and Hpa-induced cell death responses in SNUPO mutants and the wild-type ( Fig 5 ) . Most of the non-infected leaves did not display any sign of cell death ( Fig 5 , first column ) , but dark-blue stained dead cells were sporadically observed with no significant differences between non-infected leaves of both wild-type and SNUPO mutants ( Fig 5 , second column and upper graph ) . In infected leaves of the SNUPO mutants , cell death is detected in , or adjacent to , haustoria-containing cells in a frequency indistinguishable from or even lower than the wild-type ( Fig 5 , third column and lower graph ) . In all genotypes , infected leaves contain hyphal strands growing in the absence of any cell death ( Fig 5 fourth column ) . To examine whether mutation of SNUPO genes has an effect on signalling related to PAMP-triggered immunity , we inspected the transcript levels of Flg22-induced Receptor-like Kinase 1 ( FRK1 [48] ) and ERF1 in the wild-type , the SNUPO mutants and a fls2 mutant ( Fig 6 ) in response to the bacterial flagellin-derived peptide flg22 [49] . Transcript levels in mock-treated samples were not different in the wild-type and the analysed mutants , except for the shrk1 mutant that displayed a slight increase in basal FRK1 transcript levels , which was not detectable in the shrk1 x shrk2 double mutant , and the nup133 x sec13 double mutants that contained slightly decreased basal levels of ERF1 transcripts . The deviations observed for individual mutants are unlikely responsible for the increased Hpa resistance . Six hours after flg22 treatment the tested genes were all upregulated to the same extent in the SNUPO mutants and the wild-type ( Fig 6A ) . Furthermore , we investigated the growth behaviour of P . syringae on the wild-type and on SNUPO mutants ( Fig 6B ) . P . syringae pv . tomato ( Pto ) DC3000 induces the activation of SA-dependent defence signalling in the host , and deregulation of this pathways impairs P . syringae resistance [50] . The growth of Pto DC3000 wild-type or the avirulent ΔAvrPto/PtoB strain was unaltered on the A . thaliana SNUPO mutants , providing further evidence that they do not exhibit constitutive or enhanced activation of SA-dependent defences ( Fig 6B ) .
We observed—based on sporangiophore and spore counts—that A . thaliana SNUPO mutants are impaired in supporting the reproduction of the oomycete pathogen Hpa . This was associated with a shift in the haustoria morphology . Intriguingly , we observed congruent phenotypes in mutants of diverse protein classes with the only connector between these protein classes being their reported involvement in root endosymbioses . The frequency of multilobed haustoria increased in the wild-type as well as in the mutants over time revealing a clear connection between haustoria morphology and the age of the interaction , which is in line with a previously observed continuous growth of Hpa haustoria over time ( Marco Thines; personal communication ) . As the frequency of multilobed haustoria was significantly higher in the A . thaliana SNUPO mutants at all time points analysed , it appears that haustoria development is accelerated in the SNUPO mutants . It has been postulated that haustoria are the main avenue for nutrient acquisition from the host [51] . Considering the agricultural impact of biotrophic hyphal pathogens , surprisingly little is known about the precise function of haustoria and changes thereof during haustoria development . Senescence of haustoria has been associated with encasement which was discussed to likely reduce their functionality [52] . Therefore , it is possible that the accelerated formation of multilobed haustoria is a sign of senescence and thus directly responsible for the reduction in Hpa reproductive success . This scenario would suggest a role of Arabidopsis SNUPO genes in maintaining the single lobed , and presumably functional , stage of Hpa haustoria . In an opposite scenario , the surface increase from single lobed to multilobed haustoria may benefit the oomycete and the genes under study are involved in delaying the progress of haustoria development into the multilobed stage . However , this scenario is not compatible with the reduced sporangiophore and spore count on the mutants . Based on the analysis of a wide range of defence symptoms in the classical PTI assays and three different pathosystems , we conclude that A . thaliana SNUPO mutants display unaltered levels and frequencies of defence responses . Consequentially , the impairment of the Hpa interaction is not due to deregulated defence in these mutants . We did not detect consistent differences in the interaction of A . thaliana SNUPO mutants with the haustoria-forming powdery mildew fungus E . cruciferarum in comparison to wild-type plants . This may be due to the different cell types targeted by Hpa and E . cruciferarum for haustoria formation . Hpa initially colonizes epidermal cells and then progresses to the mesophyll [53] , in which the vast majority of haustoria are formed , while E . cruciferarum forms haustoria solely within epidermal cells [54] . In addition , the genetic requirements for the intracellular accommodation of fungal and oomycete pathogens may differ . The A . thaliana Mildew resistance Locus O ( MLO ) gene , for instance , is an epidermal compatibility factor required for powdery mildew fungus penetration [55] , with no role in the Hpa interaction reported to date . However , it should be noted that the haustoria of E . cruciferarum are structurally very complex and irregularly shaped . Because of this inherent polymorphy , it is possible that we missed subtle structural changes caused by the SNUPO mutations . In contrast to the vast knowledge on genes contributing to disease resistance , relatively few genes have been identified that facilitate pathogen colonization on Arabidopsis , such as PMR4 , PMR5 , PMR6 [56–58] , DMRs [59 , 60] , MYB3R4 [36] , and IOS1 [40] . While dmr3 and dmr6 mutants also exhibited aberrantly shaped haustoria , these mutants strongly differ from the consistent phenotype of the SNUPO mutants . In dmr6 , oomycete growth is arrested after the formation of the first haustoria and the expression of several defence-related genes is elevated in unchallenged dmr6 mutants [59 , 60] . Dmr3 mutants are more resistant to the bacterial pathogen P . syringae pv . tomato , they exhibit elevated PR1 expression , and dmr3 and dmr6 mutants show increased resistance to the powdery mildew pathogen Golovinomyces orontii [59] . Taken together , dmr3 and dmr6 mutants confer broad-spectrum resistance and display perturbed defence-related gene expression [59 , 60]; these data suggest that DMR3 and DMR6 –in contrast to Arabidopsis SNUPO genes—are rather involved in general plant disease resistance than in compatibility with and the accommodation of a hyphal organism . It is a long-standing hypothesis that plant pathogens exploit an Achilles heel of the plant , genetic pathways for the intracellular accommodation of mutualistic symbionts such as AM fungi or nitrogen-fixing bacteria [4 , 11] . Recently the oomycete pathogen Phytophthora palmivora , which forms haustoria that only last for a few hours , has emerged as a hemi-biotrophic model system in two independent laboratories to test this hypothesis in legume mutants . This pathogen quickly progresses to a necrotrophic phase and structural alterations in haustoria development are thus less likely to affect pathogen fitness [61–64] . Huisman and colleagues did not detect any alterations in the infection of Medicago truncatula mutant roots of the CSGs DMI1 , DMI3 and IPD3 by P . palmivora [64] . Considering that the lifetime of infected cells in the P . palmivora interaction is reduced , it appears that the transient biotrophic phase of P . palmivora is too short and progresses too fast into the necrotrophic phase to allow the haustorial maintenance functions of the CSGs–or a related gene set—to take effect [65 , 66] . Therefore , the beneficial effect of these genes or their homologs for hyphal pathogens may only be detectable in longer lasting biotrophic relationships . The too short lifetime of the host cell may also explain the absence of a detectable phenotype in A . thaliana SNUPO mutants and L . japonicus CSG mutants infected with the beneficial fungus Piriformospora indica [67] . Colonisation with the endophytic fungus Colletotrichum tofieldiae , is controlled by the plant phosphate starvation response system and C . tofieldiae only increases plant fertility and promotes plant growth under phosphorus-limiting conditions [68] . It will be an interesting task of future research to investigate whether Arabidopsis SNUPO genes are implicated in the interaction with this beneficial microbe . In contrast , for the symbiosis genes for which a mutant phenotype has been described—namely RAM2 and RAD1—the mutation was not affecting haustoria structure [61–63] . Mutants in the glycerol-phosphate acyl-transferase gene RAM2 of the model legume M . truncatula , for example , had a reduced abundance of small cutin monomers and did not elicit the formation of appressoria by P . palmivora at the surface of roots and caused an altered infection behaviour of the AM fungus within the root cortex [63] . This observation also implies that a compatibility factor can be exploited by microbial pathogens and , likewise , by beneficial symbionts . Similarly , mutants of the GRAS protein gene RAD1 , share a role in AM symbiosis and the P . palmivora interaction [61] . Our data revealed genetic commonalities of symbiosis and disease in the formation of intracellular accommodation structures at a later developmental stage of the plant-microbe association . A key difference between our study and the previous ones that failed to observe a mutant phenotype in pathogenic interactions is the longer lasting biotrophic phase in the A . thaliana—Hpa interaction . Although strongly suggested by the function of the CSGs , it remains unclear whether A . thaliana SNUPO genes are similarly involved in a signal transduction pathway directly supporting oomycete development . It will be therefore interesting to identify the mechanistic commonalities between symbiotic and pathogenic interactions with hyphal organisms that are controlled by corresponding gene sets . The loss of AM in A . thaliana and in four other independent plant lineages was correlated with the absence of more than 100 genes with potential roles in AM [29–31 , 69] . While the exploitation of symbiotic programs by pathogens might explain the consistent deletion of CSGs from five independent plant lineages , it raises the question , which evolutionary forces retained SNUPO genes in the Arabidopsis genome . A housekeeping function was not revealed since no pleiotropic developmental phenotypes were observed in the mutants . While , it might be possible that these genes limit Hpa colonisation , thereby forcing the oomycete to sporulate earlier or more profusely , our results leave us with the unexpected finding that the only detected role for the SNUPO genes in A . thaliana is the support of an oomycete . It will be interesting to find out whether ecological conditions exist , under which oomycete colonization might provide a selective advantage to the host plant , or whether SNUPO genes are also involved in the accommodation of beneficial microbes like C . tofieldiae justifying their retention in the Arabidopsis genome [70] .
All A . thaliana mutants described in the manuscript were of Col-0 ecotype , except for ios1 , which was either of Col-0 or Ler ecotype . Seeds were obtained from "The Nottingham Arabidopsis Stock Centre" - NASC [71] or the GABI-DUPLO double mutant collection [72] . For in vitro experiments , A . thaliana seeds were sterilized by incubation in sterilisation solution ( 70% ethanol , 0 . 05% Tween20 ) for 5 min , followed by incubation in 100% ethanol for 2 min . For Hpa infection , seeds were directly germinated in soil and grown for two weeks under long day conditions ( 16 h light , 22°C , 85 μmol m-2 s-1 ) . For E . cruciferarum inoculation , plants were grown in a 2:1 soil/sand mixture . Seeds were stratified for 48 h at 4°C prior to transfer into a growth chamber ( 10 h light , 120 μmol m-2 s-1 light , 22°C day , 20°C night , 65% relative humidity ) . For elicitor treatment , seeds were placed on half-strength MS plates [73] , stratified for 48 h at 4°C in the dark , and grown for 8 days under long day conditions ( 16 h light , 23°C , 85 μmol m-2 s-1 ) . Floral dipping of A . thaliana was performed as described previously [74] . To collect spores for inoculation , A . thaliana wild-type ( Col-0 or Ler ) leaves with sporulating Hpa isolate Noco2 or Hpa isolate Waco9 , respectively , were harvested seven days post inoculation ( dpi ) and placed into 10 ml deionized H2O and vortexed for 2 s in 15 ml reaction tubes . The spore suspension was then filtered through a Miracloth filter and sprayed onto 12-days-old plants using a spraying device . Subsequently , plants were placed into trays and covered with wet translucent plastic lids . Trays were sealed to maintain high humidity , and plants were grown under long day conditions ( 16 h light , 18°C , 85 μmol m-2s-1 ) . For spore counting , A . thaliana wild-type ( Col-0 ) and mutant seedlings were harvested 5 dpi . Ten seedlings per background were harvested into reaction tubes containing 1 ml deionized H2O and vortexed for 3 min . Spores were counted with a Fuchs Rosenthal chamber . For sporangiophore counting and the investigation of haustoria shape and penetration efficiency cotyledons or leaves were harvested and stained in 0 . 01% trypan-blue-lactophenol for 3 min at 95°C and 5 h at room temperature , followed by overnight clearing in chloral hydrate ( 2 . 5 g/ml ) . Samples were mounted in glycerol for subsequent differential interference contrast microscopy with a Leica DMI6000B . For sporangiophore counting , a minimum of 50 cotyledons per genotype and replicate were analysed and the number of sporangiophores per infected cotyledon was plotted . For investigation of haustoria shape and penetration efficiency , a minimum of five leaves per genotype and replicate were analysed . On each leaf , the percentage of multilobed haustoria per total haustoria or the percentage of haustoria-containing cells per cells contacted by hyphae was calculated for 10 individual strands of hyphae . On average , 1100 haustoria have been analysed per genotype and experiment . The mean for each leaf was calculated and plotted . E . cruciferarum was grown on A . thaliana wild-type ( Col-0 ) to maintain aggressiveness and on susceptible phytoalexin deficient 4 ( pad4 ) mutants [50] for elevated conidia production . Plants were placed under a polyamide net ( 0 . 2 mm2 ) and inoculated at a density of 3–4 conidia mm-2 , by brushing conidia off of pad4 mutant leaves through the net . Two leaves per plant were harvested , cleared and kept in acetic acid ( 25% ) until microscopic analysis . Leaves were stained in acetic acid ( 25% ) ink ( 1:9 ) ( Königsblau , Pelikan , 4001 ) , washed in water , placed in water containing a few drops of Tween20 , washed in water again , and analysed under a bright-field microscope . For each replicate , conidiophores per colony were counted on 10 colonies per leaf , on 5–10 leaves per genotype . Bacterial strains Pto DC3000 or Pto DC3000 ΔAvrPto/AvrPto were grown and used for infection assays on leaves of 4–5 weeks old A . thaliana plants as described previously [75 , 76] . Each experiment was performed at least three times independently . Leaves of A . thaliana wild-type ( Col-0 ) or the shrk1 x shrk2 double mutant were harvested at 5 dpi , cleared in 10% KOH for 5 min , stained with 0 . 05% aniline blue in 67 mM K2HPO4 for 20 min and observed with a Leica SP5 confocal light scanning microscope . Images were edited using ImageJ with the “volume viewer” plugin [77] . For Technovit sections , A . thaliana wild-type ( Col-0 ) or mutant leaves were infected with Hpa as described above and harvested at 7 dpi . Leaves were fixed with 3 . 7% formaldehyde and dehydrated by incubating samples in 30% , 50% , 70% and 100% ethanol . Samples were embedded in Technovit 7100 according to the manufacturer’s instruction . A Leica RM2125 rotary microtome was used to cut 7 μm sections . Sections were stained in 0 . 01% trypan-blue-lactophenol for 3 h at 37°C , followed by clearing in chloral hydrate ( 2 . 5g/ml ) and subsequent differential interference contrast microscopy with a Leica DMI6000B . A . thaliana wild-type ( Col-0 ) plants expressing RPW8 . 2-YFP [38] were infected with Hpa as described above , harvested at 10 dpi and observed with a Leica TCS SP 5 confocal laser scanning microscope equipped with a 63x NA 1 . 2 water-immersion objective ( excitation with an Argon laser 514 nm , detection at 520–560 nm ) . Oomycete-associated callose deposition was analysed on cotyledons of A . thaliana wild-type ( Col-0 ) or SNUPO mutants . Leaves were harvested at 4 dpi , cleared in 10% KOH for 5 min , stained with 0 . 05% aniline blue in 67 mM K2HPO4 for 20 min and mounted in glycerol for observation with a Leica DMI6000B with CFP filter settings . Regions of interest ( ROIs ) covering the neck bands were selected in ImageJ [77] and the signal intensity of each neck band was calculated from single ROIs and mean intensities were plotted . n = 27–72 . For pre-incubation , 8 days old seedlings were transferred to a 12-well plate ( 3 seedlings/well represent one biological replicate ) with half-strength liquid MS medium [73] supplemented with 1% sucrose and incubated overnight under long-day conditions ( 16 h light , 22°C , 100 μmol m-2 s-1; 8 h dark , 18°C ) and 100 rpm shaking . On the following day the medium was exchanged , half of the samples were supplemented with 1 μM flg22 , and the other half was kept in half-strength MS to serve as mock controls . Plants were then incubated for 6 h at 22°C and 100 rpm shaking . For every genotype , three biological replicates of treated and non-treated samples were harvested and immediately frozen in liquid N2 for subsequent RNA extraction . RNA extraction was performed using the Spectrum Plant Total RNA kit ( Sigma-Aldrich ) , followed by DNase I treatment ( amplification grade DNase I , Invitrogen ) to remove genomic DNA . First strand cDNA synthesis was performed from 250 ng total RNA using the SuperScript III First-Strand Synthesis SuperMix ( Invitrogen ) with oligo ( dT ) primers . qRT-PCR was performed in 20 μL reactions containing 1x SYBR Green I ( Invitrogen ) in a CFX96 Real-time PCR detection system ( Bio-Rad ) . PCR program: 2’-95°C; 40 x ( 30”-95°C; 30”-60°C; 20”-72°C ) ; melting curve 95°C– 60°C– 95°C . A primer list can be found in S3 Table . Expression levels of target genes were normalized against the housekeeping genes TIP41-like and PP2A [78] . For every genotype , three biological replicates represented by two technical duplicates each were analysed . Complete annotated genomic sequences were obtained from The Arabidopsis Information Resource ( TAIR - www . arabidopsis . org ) for A . thaliana , and from the KDRI website ( Kazusa DNA Research Institute , Japan; http://www . kazusa . or . jp/lotus/ ) and the GenBank for L . japonicus . BLAST searches were performed on TAIR ( http://arabidopsis . org/Blast/index . jsp ) with the L . japonicus genomic CSG sequences as query . The protein domain organization and the exon-intron structure of the A . thaliana homologs of POLLUX , SEC13 , NUP133 ( GenBank accession number: KM269292 ) , ShRK1 and ShRK2 are identical to that of their L . japonicus counterparts . By sequencing a PCR product amplified from A . thaliana wild-type ( Col-0 ) cDNA , we demonstrated that , contrary to the TAIR prediction , this was also the case for NUP133 . The curated sequence has been submitted to TAIR . For phylogenetic studies , protein sequences of A . thaliana and L . japonicus MLD-LRR-RKs were aligned using MAFFT 7 [79] with default settings . Alignments were used to create phylogenetic trees at the CIPRES web-portal with RAxML 8 . 2 . 10 [80] for maximum likelihood analyses . For RAxML , the JTT PAM matrix for amino acid substitutions was chosen . All statistical analyses and data plots have been performed and generated with R version 3 . 0 . 2 ( 2013-09-25 ) "Frisbee Sailing" [81] , using the packages “Hmisc” [82] , “car” [83] , “multcompView” [84] and “multcomp” [85] or Excel . For statistical analysis , data was either subjected to the nonparametric Wilcoxon-Mann-Whitney test with Bonferroni-Holm correction using Col-0 samples as control group , or was power transformed to improve normality and a one-way ANOVA followed by a Dunnett’s Test with Bonferroni correction was performed using Col-0 samples as control group . | Our work reveals genetic commonalities between biotrophic intracellular interactions with symbiotic and pathogenic hyphal microbes . The majority of land plants engages in arbuscular mycorrhiza ( AM ) symbiosis with phosphate-acquiring arbuscular mycorrhizal fungi to avoid phosphate starvation . Nutrient exchange in this interaction occurs via arbuscules , tree-shaped fungal structures , hosted within plant root cells . A series of plant genes including the Symbiosis Receptor-like kinase ( SYMRK ) , members of the NUP107-160 subcomplex and nuclear envelope localised cation channels are required for a signalling process leading to the development of AM . The model plant Arabidopsis thaliana lost the ability to form AM . Although the ortholog of SYMRK was deleted during evolution , members of the malectin-like domain leucine-rich repeat receptor kinase ( MLD-LRR-RK ) gene family , components of the NUP107-160 subcomplex , and an ortholog of the nuclear envelope-localized cation channel POLLUX , are still present in the Arabidopsis genome , and Arabidopsis leaf cells retained the ability to accommodate haustoria , presumed feeding structures of the obligate biotrophic downy mildew pathogen Hyaloperonospora arabidopsidis . We discovered that both of these plant-microbe interactions utilize a corresponding set of genes including the ortholog of POLLUX , members of the NUP107-160 subcomplex and members of the MLD-LRR-RK gene family , thus revealing similarities in the plant program for the intracellular accommodation of biotrophic organisms in symbiosis and disease . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"plant",
"anatomy",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"intracellular",
"pathogens",
"fungal",
"genetics",
"symbiosis",
"pathogens",
"brassica",
"fungal",
"structure",
"plant",
"science",
"model",
"organisms",
"experim... | 2019 | A set of Arabidopsis genes involved in the accommodation of the downy mildew pathogen Hyaloperonospora arabidopsidis |
The infection cycle of viruses creates many opportunities for the exchange of genetic material with the host . Many viruses integrate their sequences into the genome of their host for replication . These processes may lead to the virus acquisition of host sequences . Such sequences are prone to accumulation of mutations and deletions . However , in rare instances , sequences acquired from a host become beneficial for the virus . We searched for unexpected sequence similarity among the 900 , 000 viral proteins and all proteins from cellular organisms . Here , we focus on viruses that infect metazoa . The high-conservation analysis yielded 187 instances of highly similar viral-host sequences . Only a small number of them represent viruses that hijacked host sequences . The low-conservation sequence analysis utilizes the Pfam family collection . About 5% of the 12 , 000 statistical models archived in Pfam are composed of viral-metazoan proteins . In about half of Pfam families , we provide indirect support for the directionality from the host to the virus . The other families are either wrongly annotated or reflect an extensive sequence exchange between the viruses and their hosts . In about 75% of cross-taxa Pfam families , the viral proteins are significantly shorter than their metazoan counterparts . The tendency for shorter viral proteins relative to their related host proteins accounts for the acquisition of only a fragment of the host gene , the elimination of an internal domain and shortening of the linkers between domains . We conclude that , along viral evolution , the host-originated sequences accommodate simplified domain compositions . We postulate that the trimmed proteins act by interfering with the fundamental function of the host including intracellular signaling , post-translational modification , protein-protein interaction networks and cellular trafficking . We compiled a collection of hijacked protein sequences . These sequences are attractive targets for manipulation of viral infection .
Many studies , mainly from bacteria and unicellular eukaryotes , focus on the exchange of genetic material between viruses and cellular hosts . Sequences are best studied through their structural and functional domains [1] , [2] , [3] , [4] , [5] . The evolution of domains is a significant force for shaping the proteins along the tree of life . Sequence exchange between genomes within and between superkingdoms is evident from the appearance of a domain in a particular phylogenetic branch [6] . The contribution of horizontal gene transfer is not limited to bacteria but has occurred across distant species [3] . For example , some signaling domains in bacteria are the consequence of a horizontal gene transfer [7] . The viruses are parasitic agents that maintain an intimacy with their host cells . Consequently , an extensive horizontal evolution [8] is associated with the viral life cycle . The lack of similarity of viral proteins ( e . g . , capsid proteins ) with any cellular organisms is in accord with their early and unique origin [8] , [9] . Most likely , the modern viruses originated at the early RNA world of the primordial genetic pool . With the increasing numbers of sequenced viruses , similarity among seemingly unrelated viruses was reported . A role of the hosts as vehicles for such cases is proposed . For example , the structural similarities observed between bacterial viruses ( PRD1 , Bam35 ) , Chlorella virus ( PBCV-1 ) and adenovirus in the coat proteins , led to the proposal that all viruses are old , probably preceding the cellular life . Furthermore , it is compatible with polyphyletic virus origins , as opposed to the monophyletic origin of cellular life [10] . Still , assignment of viruses to the phylogenetic tree of life remains unresolved [11] . Notably , viruses as vectors ( mainly RNA viruses ) have the potential to rearrange the genomic material , and thus , to change the domain architecture [12] , [13] , [14] . Studies on horizontal gene transfer focused primarily on viruses infecting bacteria and archaea ( e . g . , bacteriophages ) [15] , [16] . The co-evolution of viruses toward their hosts indicates an active crosstalk on an evolutionary time scale [17] , [18] , [19] . Several studies reported on a handful of cases of functional mimicry by viral proteins [20] . In few cases , evidence for gene transfer from the host to the virus is obvious . For example , the photosynthetic efficiency in cyanobacteria ( Synechococcus and Prochlorococcus ) relies on components of the photosystem II . These critical components express in the respective phages [21] . In the case of the phytoplankton–virus system , the DNA virus EhV that infects the microalge ( Emiliania huxleyi ) , contains a complete metabolic pathway as a result of a horizontal gene transfer [22] . A similar case is demonstrated for the dUTPase genes ( Dut ) that are necessary for regulating the cellular levels of dUTP . Phylogenetic analysis revealed the origin of the viral Dut sequence in a monophyletic cluster of DNA viruses with eukaryotic hosts [23] . The Acanthamoeba polyphaga Mimivirus and the family Phycodnaviridae [24] , contain many genes that are found in cellular organisms . For example , the giant virus Cafeteria roenbergensis virus ( CroV ) includes numerous eukaryotic-like genes for translation factors , ubiquitin pathway components , intein elements , histone acetyltransferase and more [25] . These are extremely large viruses of aqueous environments that infect bacteria , animals and protists [26] . A search for similarities between viral and host proteins has largely been focused on herpesviruses [27] , Hepadnaviridae [28] and others . However , the high mutation rate of RNA viruses [29] and the coexistence between viruses and their hosts for millions of years has most likely blurred the sequence similarity . Recently , several studies challenged the origin of ancient viral segments in metazoan genomes . These sequences that are called EVE ( for endogenous viral element ) encompass all virus-derived genomic loci [30] . In this paper , we present a coherent survey on protein sequences that are shared between viruses and their hosts . We assess the scale of the phenomenon by focusing on the viral-related protein sequences that appear in metazoa . We have used the current archive of all proteins [31] as the basis for identifying sequences with a potentially common origin . Presumably , their appearance in the virus reflects virus-acquired sequences . Of about 190 instances of highly similar viral-eukaryotes sequences , we recognize that only a small number originated from a host origin . We extended the collection of viral proteins that have a host origin by investigating the eukaryotes-viruses Pfam families [32] . We focused on the 670 Pfam cross-taxa families that contain viruses and metazoa . A careful examination reveals that these instances reflect either missed annotations or the remnants of sequence exchange by virus infection . To distinguish these possibilities , we constructed sequence alignment trees for all 670 Pfam families . From the properties of the trees , we focused on 335 families that most likely contain viruses that hijacked sequences from their host . We found that most of the viral proteins in the orthologous families are much shorter and composed of simpler domain architectures . In almost all cases , the number of domains and the sequence of the tails and the inter-domain linkers are considerably shorter in the viral proteins relative to their counterpart host proteins . We discuss the potential of such short viral proteins to interfere with critical cellular functions and thus are candidates for manipulation strategies in defeating viral infection .
Figure 1A shows two over-simplified scenarios in support of a genetic exchange from the virus to the host genome and in the reverse direction , from the host to the viral genome . In the first scenario , a viral sequence is detected in the host ( e . g . , human ) but not in the rest of the phylogenetic branch . The following scenario accounts for viral sequences acquired from the host ( Figure 1A , right ) . Under this scenario , the viral gene sequence is identified in a broad group of organisms that belong to a phylogenetic tree that includes the host ( human ) . Therefore , the sequence in the virus is most likely a reflection of a hijacking event , according to an argument of maximum parsimony . Supporting evidence for the directionality of the genetic exchange of viral and cellular organisms relies on a detailed phylogenetic analysis . The topology of the reconstructed tree is used to support the most parsimonious scenario ( see Materials and Methods ) . The simplified illustrations in Figure 1A do not address the more complicated , realistic instances in which different viruses carry sequences that resemble various organisms . An additional criterion used in supporting the occurrence of sequence acquisition by viruses is the presence of a sequence resemblance in the known host . The origin of viruses is probably preceding the cellular life [8] , [10] . Thus , the ancient events in which viral sequences were incorporated into an ancestor eukaryote cannot be traced by their sequence similarity . Still , a conserved functional or structural similarity could expose such early events [33] . In this study , we have not attempted to date the horizontal transfer event . Furthermore , we will not discuss the events of genetic material exchange ( see discussion in [34] ) , but limit our study to the acquisition of coding sequences in viruses and metazoa . There are about one million viral proteins in the UniProt database ( 990 , 049 , August 2010 ) that represent about 66 , 000 viral strains . This is a highly redundant resource and about half of it composed of medically relevant strains including Hepatitis B viruses ( HBV ) and Human immunodeficiency virus ( HIV ) . We took advantage of a reliable source of UniRef [35] that unifies sequences according to their identity level along the sequence length . We used UniRef90 classification ( see Materials and Methods ) . There are >165 , 000 UniRef90 clusters that contain at least one viral protein ( Figure 1B ) . However , from this set , we only considered 262 instances that contain at least two proteins , where one of them must be a eukaryote ( Figure 1B ) . Of the 5 , 482 cross-taxa clusters that contain sequences from viruses and cellular organisms , 95% are sequences of bacteriophages and plasmids confined to the bacteria [36] . We will not further discuss the events that are confined to bacteria and archaea . A taxonomical view shows the diversity of the organisms that share the UniRef90 clusters with viral proteins ( Figure 1C ) . It shows that the eukaryotes are the most diverse group with 106 species that share their homologues with viral proteins . This result suggests that the phenomenon of shared sequences is quite broad , and many eukaryotes have been subjected to a genetic material exchange . Among the UniRef90 clusters that contain viruses and eukaryotes ( Figure 1B ) , ∼70% are from tetrapoda , 13% plants , 13% arthropoda , 4% fungi and only a smaller percentage of other taxa . We focused on the cross-taxa clusters of viruses and mammals ( 118 clusters include ∼2 , 200 proteins , Figure 1D ) . Viruses from Class I ( dsDNA viruses with no RNA stage ) and class VI ( Retro-transcribing ssRNA , Plus strand ) are prevalent among those that infect mammals [17] . The dominating Class VI viruses are characterized by their ability to integrate sequences into the host ( Figure 1A , left ) . Table 1 lists the Class I viral proteins that share sequences with mammals ( 23 clusters , Figure 1D ) . In Class I viruses , the virus enters the nucleus before its replication ( with the exception of Poxvirus family ) and its infectivity is strongly dependent on the host cell division . Discriminating whether a sequence has originated from the virus or the host is not straightforward . We generated for each cluster a phylogenetic dendogram and analyzed the connectivity of the viral protein in view of its neighboring sequences . Often the analyzed cluster is too small . In such cases , we expanded the cluster to the relaxed UniRef50 classification . We applied additional criteria in support of a virus having acquired protein sequences from the host: ( i ) The tested sequence appears in several organisms ( ≥2 ) on the same evolutionary branch ( as in Figure 1A , right ) ; ( ii ) The tested sequence is not associated with viral contamination . Most analyzed cross-taxa clusters derive from the contamination by viral proteins following integration of the virus to a mammalian genome . Several instances were contaminated by the extensive use of viral vectors as vehicles in variety of molecular manipulations ( e . g . , Adenoviruses , Table 1 ) . Another source of contamination is from cancerous cells infected by viruses ( e . g . , human papilloma virus ) . In such instances , some sequences that are assigned as ‘human’ are incorrectly annotated . In these instances that reflect the incorporation of the virus to the host , different protein sequences from the same virus are identified which are best explained as a result of infection or an integration event . For example , the proteins in UniRef90 clusters P06426 , P06463 , P21735 , P06788 , P36741 and P21736 belong to Human papilloma virus ( Table 1 ) . Studies on the viral sequences that were integrated into the vertebrate germ line and hence shaped the vertebrate genetic heritage were reported [37] , [38] , [39] . Herein , we only consider the protein sequences that are shared by viruses and their metazoan hosts . The principal virus families that infect multicellular eukaryotes are listed in Supportive data Table S1 . For few instances , a support exists for viruses that hijacked sequences from the host . Among Class I viruses ( Table 1 ) the shared functions include interlukin-10 ( IL-10 ) ( Figure 2 ) , beta-1 , 6-N-acetylglucosaminyltransferase ( β1 , 6GnT ) ( Figure S1 ) and Ubiquitin . The β1 , 6GnT and IL-10 are found exclusively in metazoa and the multicellular eukaryotic branch . The key features and the functional amino acids are conserved in the viral and the corresponding mammalian proteins ( Figure 2A , Figure S1 ) . Indeed , in human cells lacking β1 , 6GnT gene , the Bovine herpesvirus 4 ( BoHV-4 ) sequence fully recovered the missing enzymatic activity [40] . Resolving the evolution of the Ubiquitin in the genome of Pestivirus suggested that the virus hijacked Ubiquitin-related sequences in two consecutive events [41] . A browsable table is available at www . protonet . cs . huji . ac . il/virost/tables/UniRef90-Class1 . html . Figure 2 shows a prototypic case of viral proteins that resemble the host protein . Interleukin 10 ( IL-10 ) inhibits the induction of pro-inflammatory cytokines . IL-10 was found in many viruses including Epstein–Barr virus ( EBV ) , equine herpesvirus ( EHV ) and cytomegalovirus ( CMV ) [42] . Presumably , the gene product protects the infected cells from the host defense mechanism . An extended cluster of IL-10 ( Table 1 ) covers 20 viruses and 96 cellular organisms ( UniRef50_P22301 ) . Representatives of viral and metazoan proteins are shown by the multiple sequence alignment ( MSA ) ( Figure 2B ) . Most of the variations in the viral and metazoan protein reside in the sequence of the N-terminal that covers the signal peptide ( Figure 2B ) . Traces of a genomic organization of the host in the viral genome were reported . For example , IL-10 like sequence from the gammaherpesvirus ovine herpesvirus 2 includes 5 exons and 4 introns [43] . Inspecting the UniRef90 clusters that contain proteins from viruses and metazoa ( 187 clusters ) shows a wide variation in the distribution of protein lengths ( Supportive data Table S2 ) . Viruses tend to reduce their production load by deleting and reducing the unessential genetic material [44] . While this length reduction is an absolute necessity for most viruses , some giant viruses ( e . g . , Mimivirus , Chlorovirus , and Cafeteria roenbergensis virus ) include ∼1000 proteins [24] , [25] . The evolution origin of proteins from the Giant viruses remains unknown [45] . Still , 12% of the Acanthamoeba polyphaga mimivirus ( APMV ) proteins constitute a large number of host related sequences . The average length of this subset of the Mimivirus proteins ( 523 amino acids ) is similar to the length of their homologous sequences . The small numbers of cases of viral acquired sequences ( Table 1 , Supportive data Table S2 ) may indicate the sequence divergence that had occurred throughout evolution . We therefore expanded the analysis for remote homologous . We questioned whether the viral protein sequences that were already substantially diverged due to a rapid evolution rate , or a long evolutionary history still maintain the host protein's functional domain . The Pfam provides a comprehensive resource of functional and structural families and domains . Each Pfam entry represents a statistical model with an average sequence identity of 30–40% among the members of the family . Currently , Pfam covers 11 , 912 families , where 1 , 165 families include at least a viral protein and a eukaryotic protein representative . Some Pfam families are extremely large . Among families that contain metazoa and viral proteins are ‘Helix-loop-helix DNA-binding domain’ ( ∼6000 proteins ) and ‘Sugar transporter’ ( ∼12 , 000 proteins ) . Contamination of viral proteins in metazoan proteomes ( e . g . , Capsid , Env , Tat ) occurs mainly as a result of viral vector manipulations in cell lines , leading to incorrect assignment as a viral-eukaryotic cross-taxa family . An example is the GFP family ( PF01353 ) that we have manually removed from the analysis . To reduce such sporadic instances , we considered Pfam families having at least two metazoan proteins , resulting in a list of 667 Pfam families . Supportive data Table S3 lists the species , composition of the domains and the proteins' length . The relatively small numbers of cases of viral acquired sequences ( Table 1 , Supportive data Table S2 ) may indicate the sequence divergence that had occurred throughout evolution . Therefore , we expanded the analysis for remote homologous . We questioned whether the viral protein sequences that were already substantially diverged due to a rapid evolution rate , or a long evolutionary history still maintain the host protein's functional domain . Over 300 cross-taxa Pfam families ( virus-metazoa ) are best explained by a viral acquisition of host sequences . Instances of lateral gene transfer between bacteria and their bacteriophages dominate many of the cross-taxa Pfam families . Other families contain genuine viral proteins contaminated by metazoan proteins . In order to justify the directionality of sequences from the hosts to the virus , we constructed for each of the 667 Pfam families a sequence-based tree ( MSA based on the domain and not the full length sequence ) . We considered Pfam families in which only 1–2 viral proteins are included in the family , and families in which the percentage of the virus proteins in the family is small ( <5% , Figure 3A ) . The vast majority ( 547 families , 82% ) of the analyzed Pfam families fulfilled these criteria ( Figure 3A , blue ) . We also requested that the viral proteins are clustered in sub-trees within the family tree . We counted the number of viral proteins spreading within the sequence alignment tree . We suggest that viral proteins that are clustered in a defined sub-tree ( called Viral Cluster , VC ) are likely to represent a single episode of acquired sequence from the host . Consequently , only a limited diversity among the closely related viruses is expected in view of the rest of the tree . 64% of the 547 Pfam families from the previous selection fulfill the requirement for clustered viral proteins . These are the Pfam families that contain ≤2 viral clusters ( 60% ) , and other families ( 4% ) that are specified by a high degree of condensation ( i . e . the ratio of the viral proteins to the number of VCs is ≥3 ) . These filtration steps further reduced the list of relevant Pfam families to 335 ( Figure 3B ) . We show a tree constructed for one of the 335 families . The IL-6 ( PF00489 ) family contains 10 viral proteins ( Figure 3C , blue ) that are split to two sub-trees of viral clusters ( VC ) and other 136 Metazoan proteins ( marked as collapsed sub-trees ) . The maximal depth in this tree is 19 ( included in the collapsed sub-tree , red triangle ) . The deepest viral protein in the tree is of depth = 9 , and its normalized depth is 9/19 = 0 . 474 . The depth of the viral cluster ( VC , the maximal sub-tree which contains all viral proteins ) is therefore , 4/19 = 0 . 211 . The normalized depth for all the proteins in the 335 Pfam families ( Figure 3D , top ) is analyzed in view of the distribution of the normalized depth of the viral proteins within these families ( Figure 3D , bottom ) . It seems that the two distributions are remarkably different ( Figure 3D ) which is in accord with the notion that the viral proteins are relatively isolated subsets among the proteins from the cellular organisms in the relevant Pfam families . Table 2 shows a sample of these families along with the cellular process and the protein function in the viral life cycle . A full list of the 667 Pfam families with the analyzed properties of their alignment trees is provided in Supportive data Table S3 . One of the families that exemplified the trend found in virus-metazoa Pfam families is the PAAD/DAPIN/Pyrin family ( PAAD_DAPIN , PF02758 ) . This domain family is a diverse family ( 26% average sequence identity ) that includes 34 cellular species and 5 dsDNA viruses that belong to the Poxviridae . The PAAD domain is at the N-terminal regions of proteins . This domain occurs in several multicellular organisms , in the context of inflammation , signaling and apoptosis ( Figure 4 ) . Several observations could be extracted for the PAAD domain: ( i ) Based on a multiple sequence alignment ( MSA ) of the PAAD domain sequences it is evident that the 5 viral proteins were diverged significantly ( Figure 4B ) . All 5 viral proteins reside in one cluster , in the phylogenetic tree , together with other mammals as their sibling in the tree ( Figure 4A , blue font ) . The domain architecture within the protein of the family is best explained by an initial extensive duplication of the PAAD domain ( Figure 4A , green symbol ) . At present we identified ∼50 such proteins in human and mouse ( Figure 4A ) ; ( ii ) Most members of the PAAD family contain additional domains ( in 158/177 occurrences ) . For example , the combination of PAAD and NACHT domains ( Figure 4A , red symbol ) are in 93 proteins , and PAAD , NACHT and LRR are in 2 proteins; ( iii ) The majority of the other domains ( e . g . , HIN-200 , CARD ) function in the regulation of apoptosis; ( iv ) All 5 viral proteins are single-domain proteins with PAAD domain . There are other 19 cases of the single domain proteins ( Figure 4A , red font ) . Note that these proteins spread throughout the sequence-based tree . Presumably , it is a reflection of a domain loss event . Some of these proteins are fragments ( e . g . , Q5T3V8_HUMAN ) , and others include less characterized PfamB domains [32] ( e . g . , IFI4L_MOUSE , Q3UPZ5_MOUSE ) . The initial tests on UniRef90 covered 14 , 000 proteins in relatively small clusters ( <90 proteins on average , Supportive data Table S2 ) . In contrast , the collection of the cross-taxa Pfam families ( Table S3 ) covers 161 , 000 viral proteins and 400 , 000 metazoan proteins . Therefore , focusing on the cross-taxa Pfam families provides an opportunity to increase the statistical power of the tests . Several statistical observations regarding the sequences among the cross-taxa families of viruses and multicellular organisms can be made: ( i ) The average length of the metazoan proteins is 507 amino acids , while the average length for the viral proteins in these families is only 396 amino acids ( P-value of <1 . 0e-17 by the KS-test , Figure 5A ) . ( ii ) For 73% of all families , the viral proteins are shorter than the length of the average metazoan proteins in the family ( P-value<1 . 0e-13 by the Hypergeometric test ) . ( iii ) In 67% of the families , the number of Pfam domain appearances ( including several repeats of the same domain or different ones , Figure 5B ) is smaller in the viral proteins relative to the metazoan proteins in the family ( P-value<1 . 0e-40 by the KS test ) . ( iv ) In 62% of the families , the number of different Pfam domains is higher in the metazoan proteins relative to the viral proteins . ( v ) For the discussed families , the median number of Pfam domains is 1 . 06 while , for the metazoan proteins , this value is 1 . 7 ( P-value<1 . 0e-32 by KS test ) . Many metazoan proteins are multi-domain ( colored rectangle , Figure 5B ) . We tested whether the viral acquired sequences that belong to multi-domain proteins displayed a stronger tendency for a size reduction ( see scheme , Figure 5B ) . A reduction in length of viral proteins may be a reflection of reducing the number of domains ( Figure 5B , b–c ) , shortening the length of the linker sequences ( Figure 5B , a ) or even the trimming of the length of the domain itself . Among the 667 analyzed Pfam families , in 103 of them , the metazoan proteins contain at least 3 Pfam domains . In 85% of this set ( 88 families ) , the viral proteins are shorter ( Figure 5B , virus ) . Remarkably , the average length of these 103 metazoan proteins families is 912 amino acids relative to 503 amino acids for the viral proteins that belong to these families . Similarly , in this set of multi-domain proteins the viral proteins have an average of 2 . 9 domains , while the metazoan proteins have 4 . 6 domains on average ( paired t-test , p-value of 1 . 0e-11 ) . This shows that the tendency to reduce the protein length and the number of domains is stronger when the number of Pfam occurrence in the original host protein is higher . In order to reduce the risk of misclassification , we further restricted the analysis to Pfam families of viruses-metazoa ( with ≥3 Pfam domains ) that contain at least 2 viral proteins ( total of 50 families ) . The length of the viral proteins is significantly reduced . For 90% of these families ( above the reference line , Figure 5C ) , the viral proteins are shorter than their matched metazoan proteins . In order to determine whether the reduction in length is due to a reduction in the number or the properties of the domains , we repeated the analysis for the ratio of the number of distinct domains ( depicted by the different colored rectangles , Figure 5B ) in the viral and their relevant metazoan sequences ( Figure 5D ) . For 80% of the families ( families above the reference line ) , number of different Pfam domains that are associated with viral proteins is reduced . Note that by this measure , a short viral protein ( Figure 5B a–b ) still has a ratio of 1 . 0 . We show that the viral proteins are not only significantly shortened , but have also converged to a simpler domain composition . The length of the individual domains between the viral and the metazoan host proteins is identical ( Supportive data Figure S2 ) . Recall , that this observation may be mainly due to the definition of belonging to a Pfam domain family . The high statistical significance of these trends is consistent with a possibility that the short viral proteins have resulted from the acquisition of fragments from the host protein . Alternatively , it can be the result of a refinement of the acquired sequences during viral evolution . We separated each protein into three segments: ( i ) The Pfam domain ( s ) ; ( ii ) The tail linker ( TAIL ) that combines the amino acid extension towards the N- and the C-termini of the protein , beyond the boundary of the domain ( s ) ; ( iii ) The internal domain linker ( IDOL ) that comprises the sum of the amino acid spacers between domains . Clearly a single domain protein lacks IDOL . We performed a separate analysis for the TAIL and the IDOL sequences ( Figures 6A–6D ) . The study was performed on all the families that have at least 2 Pfam domains ( unique or repeated ) . The average TAIL in viral proteins is 14 amino acids while the metazoan protein TAIL length is 85 amino acids ( p-value<1 . 0e-150 , Figures 6A–6B ) . Trimming of protein tails at both termini often leads to a loss of cellular localization signals ( e . g . , KDEL , PDZ binding sites are found at C-termini ) [17] . Importantly , the average IDOL length of the viral proteins in the Pfam families is 30 amino acids , while , for the metazoan equivalent proteins , the length is 67 amino acids ( p-value<1 . 0e-150 , Figures 6C–6D ) . While a short TAIL may be explained by the viruses having acquired a fragmented sequence from their hosts , the same trend was found for the IDOL . Figure 6C shows that while only 54% of the metazoa IDOL have a length of <40 amino acids , in the viral proteins from the same Pfam families , 96% of the proteins have IDOL that are shorter than 40 amino acids . These results are consistent with an active trimming and refinement process throughout viral evolution . Short IDOL length is advantageous in suppressing protein misfolding , and hence , improving translation effectiveness [46] . Similarly to the finding of short IDOL sequences in viral proteins , we identified instances of an internal domain which is missing in the viral protein while the flanking domains are maintained in the same order in the eukaryotic homologous protein ( Figure 6E ) . The viral putative phosphatidylinositol kinase L615 ( UniProt: Q5UR69 ) is a 701 amino acid protein from the Acanthamoeba polyphaga Mimivirus ( APMV ) that infects Amoeba . It has two Pfam domains: FYVE ( PF01363 ) followed by PI3_PI4_kinase ( PF00454 ) . There are no other known proteins with identical domain architecture in the Amoebozoa kingdom ( there are 7 such proteins in other kingdom , e . g . , Stramenopiles and Excavata ) . However , there are 3 proteins from the genus Dictyosteliida ( slime molds ) that do belong to the Amoebozoa kingdom ( UniProt D3BQ22 , Q54UU9 and EGC34678 ) . In all 3 of these proteins , the architecture is composed of FYVE domain followed by PI3Ka and PI3_PI4_kinase . The missing domain of PI3Ka in the Mimivirus ( APMV ) provides an evidence for an active elimination of an internal domain based on parsimonious argument . The findings of shorter IDOL ( Figures 6C–6D ) or absence of internal domains ( Figure 6E ) are probably the result of the trimming and shortening of the sequences after their acquisition by the virus . The possibility of a domain insertion in eukaryotes cannot be excluded . The exhaustive search for sequences that were hijacked by viruses from their host allowed us to speculate on the underlying modes of mimicry . It was shown that once a mimicry function by a virus is established , the corresponding functional partner protein of the host undergoes a fast positive selection to overcome the deleterious effect of the viral mimicry [20] . According to these findings , the viral proteins that originated from the hosts are short versions of the full-length host proteins ( Figures 5–6 , Supplemental Figures S2 , S4 ) . Furthermore , these proteins are characterized by a substantial reduction in the architectures of the domains ( Figure 5 ) and the protein linkers ( Figure 6 ) . We classified these proteins into distinct ( yet not exclusive ) modes of action . For simplicity , we unified the viral acquired sequences from the cross-taxa families to 5 strategy modes ( Figure 7 ) . Mode A depicts a competition on a receptor binding by a viral ligand that replaces the natural one . Examples for this mode are the expression of the secreted IL-10 ( Figure 2 ) , IL-8 ( UniProt: Q98158 , Q98314 , D2E2Z5 ) and PDGF ( UniProtKB Q80GE8 , Q2F842 and D0VXD7 ) . These secreted mitogens are identified in class I and class VI viruses ( Table 2 ) . Viral proteins participate in a rich protein-protein interaction ( PPI ) network [47] . Mode B illustrates PPI , where the virus uses an acquired sequence for replacing a host partner protein or for interacting with a preexisting protein complex . The result is an alteration of the cells' function . Examples for viral proteins that interfere with the host PPI are the anti-apoptotic Bcl-2 sequences and Profilin ( Table 2 , for example UniProt: Q5IXM3 , P33828 , P68695 ) . Mammalian Semaphorins ( Sema7 ) and the Smallpox virus A39R protein ( Table 2 , UniProt: Q775N9 , B7SV99 , Q0N658 , A0ES13 ) share identical binding modes with a cross-reactivity towards common receptors [48] . Mode C depicts the role of protein modifications ( e . g . , phosphorylation ) . A viral protein can either mimic the host modifications ( Figure 7 , marked C1 ) . Alternatively , a modification occurs by a viral enzyme ( Figure 7 , marked C3 ) . Such mimicry can lead to a modification of the original site or at an entirely new site ( Figure 7 , marked C2 ) . Apparently , there are instances in which both the modifying enzyme and the target proteins are both sequences that were acquired from the host ( Figure 7 , marked C4 ) . This mode is dependent on the presence of active kinases ( or phosphatases ) . For example , human cytomegalovirus ( HCMV ) kinase introduces phosphorylation sites that perfectly mimic the function of the cellular CDK2 ( cyclin dependent kinase ) [49] . An evolutionary tree alignment for viral B1R protein kinase ( Supplemental Figure S3 ) supports the functional overlap and mimicry with the closely related cellular kinases . Mode D depicts the importance of nucleic acid regulation of transcription . In this mode , a viral protein mimics the host regulation by either competing for an existing transcription factor ( Figure 7 , marked D1 ) , or by modifying the transcription program following a DNA/RNA binding ( Figure 7 , marked D2 ) . For example , the Epstein-Barr virus ( EBV ) encodes an activator protein that is similar to Fos/Jun family ( bZIP_1 , PF00170 . For example , UniProt: Q80GR6 , Q8QQX9 , Q6USE5 , D2Y5S7 ) . The difference in specificity and the dimerization properties of the EBV activator allows the activation of an alternative transcription program [50] . Mode E collectively points to the generic strategies for damaging and deactivating the host proteins . It could be achieved by protein tagging ( i . e . , SUMO , ubiquitin ) , or the activation of viral proteases . Among the cross-taxa Pfam families , some families are associated with specialized proteases ( Table S3 ) . Mode E shows the various routes by which acquired sequences alter key cellular processes . Molecular mimicry in trafficking and the subcellular localization is common to many viruses . For example , Soluble N-ethylmaleimide sensitive factor Attachment Protein ( α-SNAP ) is a conserved protein among all eukaryotes . It was also found in Canarypox and Fowlpox viruses [51] . These proteins may alter the balance of the vesicular trafficking , docking and the membrane fusion machinery . In autophagy , viral proteins exploit processes such as membrane fusion and protein folding for the benefit of their replication [52] . We limit the discussion to the modes by which the shorter versions of the viral acquired proteins exhibit their impact on some cellular functions . The described modes ( A–E ) are effective in additional instances of molecular and functional mimicry [53] , [54] . Inspecting the viral proteome is challenging , as the majority of viral sequences are redundant and poorly annotated . Importantly , the rapid evolution and the high mutation rate in some viral classes often leads to the loss of a detectable sequence similarity and , therefore , additional cases of virus hijacking events cannot be detected based on sequence similarity search methods . Despite these drawbacks , we have traced hundreds of viral proteins with respect to their hosts . Only a small fraction of them shows high sequence similarity with corresponding host proteins . For the majority of the cases , the origin of the viral sequences and possible derivations from the host call for applying powerful models for remote homologues . We provided analysis for 670 homologous families ( according to the Pfam definition ) . For half of these families we provided support for sequence acquisition by the viruses from their hosts . The candidate sequences for a host to viral acquisition are useful in exploring the mechanisms by which viruses hijack and refine sequences . We found that most of the viral proteins that potentially originated from host sequences are significantly shorter and contain fewer domains . Furthermore , we propose that the sequence refinement by the virus is a dynamic process . The inter-domain linkers ( e . g . , sequences connecting domains , but excluding the amino- and carboxyl tails ) are significantly short , relative to other related proteins ( Figure 6 ) . The viral proteins act in the cell according to a finite number of strategies . The simpler domain composition of these viral proteins is sufficient for the utilization of functional mimicry . Currently , we are expanding the analysis by identifying short peptides in viral proteomes that serve as competition agents for neutralizing critical cellular functions . The collections of 187 UniRef90 clusters and the 667 Pfam cross-taxa families are available as interactive tables . These tables are available at: www . protonet . cs . huji . ac . il/virost/tables/UniRef90 . html www . protonet . cs . huji . ac . il/virost/tables/Pfam . html
UniProKB includes 990 , 049 sequences ( taxonomy-viruses ) . The viral proteins include ∼15 , 000 reviewed proteins ( UniProt/SwissProt ) . The rest of the proteins are from UniProt/TrEMBL . There are 430 . 6 K sequences after removal of HIV and HBV sequences . Only 241 . 8 K are full-length ( 56 . 1% ) , while the rest are denoted as ‘fragments’ . The percentage of full-length proteins in metazoa is 54% ( 1 . 191 M/2 . 2051 M ) . The pre-calculated classifications of UniRef90 ( i . e . , identity of >90% at the amino acid level ) reduce the UniProKB set to 175 , 236 clusters . Additional steps of filtrations are: ( i ) Considering only clusters with a minimal size of 2 proteins ( 62 , 129 clusters ) ; ( ii ) Clusters that also include the metazoan proteins ( 187 clusters ) . ViralZone is a database that manually assigns host-virus pairs ( http://www . expasy . ch/viralzone , coordinated by UniProt/SwissProt ) . ViralZone holds reference strains viruses that belong to 83 families and 330 genera . This is a high quality collection of ‘complete proteome’ . All viruses are classified into 7 disjoint classes ( Baltimore classification index ) : ( I ) Double stranded DNA viruses; ( II ) Single stranded DNA viruses; ( III ) Double-stranded RNA and Single-stranded RNA viruses with positive and negative sense , respectively ( IV , V ) ; ( VI ) Positive sense single stranded RNA viruses that replicate through a DNA intermediate; ( VII ) Double-stranded DNA viruses that replicate via a single-stranded RNA intermediate . Major families of viruses infecting vertebrates are listed in Supporting information , Table S1 . Pfam 24 . 0 ( 11 , 912 families ) [32] is a high quality resource for domains and families . A valid cross-taxa list was generated . Eukaryotes and viruses cross-taxa resulted in 1 , 165 Pfam entries . The following filtration steps were applied: ( i ) Pfam families with at least one viral protein and at least one metazoan protein ( taxid: 33208 ) , total of 859 Pfam families . ( ii ) Restricting the Pfam to families that have at least one metazoan protein and at least one metazoan-infecting virus resulted in 796 Pfam families . ( iii ) Pfam families with >95% viral proteins for structural element of the virus ( e . g . , Env , Coat , Capsid ) . ( iv ) Enzymes of the replication system were excluded , as these genes are the outcome of several events of genetic exchange [55] . Specifically , we excluded families of RNA/DNA polymerases ( 39 families ) , Exo/Endonuclease ( 16 families ) , Helicase ( 15 families ) , tRNA synthetase ( 8 families ) and Primase ( 8 families ) . We also manually eliminated the cluster represented by the GFP ( PF01353 ) that reflects the inevitable contamination from the extensive use of GFP as vectors in many molecular biology techniques . The filtered list includes 667 protein Pfam families ( Supplemental data Table S3 ) . We define linker sequences as TAILs ( Tail Linkers ) and IDOLs ( Inter Domain Linkers ) . The TAILs are all sequences at the two terminals external to the first and last domain in the protein . Each protein provides two entries . The IDOL is a collection of all inter-domain sequences ( excluding TAIL ) . Protein TAIL's length was defined as the mean of the two tail segments . In the same way , IDOL length was defined as the mean of the lengths of the inter domains linkers . We collected the Pfam data for all proteins having at least 2 domains ( i . e . , having at least one IDOL ) and one of the domains belong to the 667 Pfam domain families ( Table S3 ) . There are ∼57 , 000 such viral proteins and ∼98 , 000 metazoan proteins . Statistical tests were applied for the set of viral proteins in view of the host cellular protein for each cluster ( or Pfam family collection ) . We applied statistical confidence tests ( P-values ) based on the non-parametric Kolmogorov-Smirnov ( KS ) , Student t-test and the hypergeometric distribution tests . The KS test is based on the maximum distance between the two cumulative curves based on the separated viral and host proteins and viral and metazoan for the TAILs and IDOLs . Multiple sequence alignments ( MSA ) by ClustalW were used for constructing the Phylogenetic trees . Local alignment searches are from NCBI-BLAST . BLAST was activated with a ‘gap costs’ for Existence: 10 and for Extension: 1 . The resetting of the BLAST parameters was needed for systematic identification of missing domains detection scheme . The phylogenetic trees were built using the iTol [56] . | Many studies focused on the exchange of genetic material between viruses and cellular hosts . The diversity of viruses argues that , along the evolutionary history , viruses have shaped the host genomes . While most viruses have many opportunities to exchange genetic material with their hosts , tracing such events is challenging as the origin of the sequences is masked by the high mutation rate of many viruses . On the other end , for completing a successful infection cycle the viruses must cope with the cell machinery for entry , replication and translation while hiding from the host immune system . We collected evidence for instances of viral protein sequences that were most probably “stolen” from the hosts . Additionally , a shared ancestry with metazoa is associated with 670 Pfam domain families . For half of these families , the origin of the viral proteins from its host is supported . For about 75% of the cross virus-metazoa families , the viral proteins are significantly shorter than their counterpart host proteins . Most of these cross-taxa viral proteins are single domain proteins and proteins with a simple domain composition relative to the proteins of their hosts . These viral proteins provide insights on the overlooked intimacy of viruses and their multicellular hosts . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"genomics",
"virology",
"viral",
"classification",
"immunology",
"biology",
"computational",
"biology",
"evolutionary",
"biology",
"microbiology",
"evolutionary",
"processes",
"proteomics",
"genetics",
"and",
"genomics"
] | 2012 | Viral Proteins Acquired from a Host Converge to Simplified Domain Architectures |
In the tropics , the utilization of insecticides is still an important strategy for controlling Aedes aegypti , the principle vector of dengue , chikungunya and Zika viruses . However , increasing insecticide resistance in Ae . aegypti populations might hinder insecticide efficacy on a long-term basis . It will be important to understand the dynamics and evolution of insecticide resistance by assessing its frequency and the mechanisms by which it occurs . The insecticide resistance status of four Brazilian Ae . aegypti populations was monitored . Quantitative bioassays with the major insecticides employed in the country was performed: the adulticide deltamethrin ( a pyrethroid—PY ) and the larvicides , temephos ( an organophosphate ) and diflubenzuron ( a chitin synthesis inhibitor ) . Temephos resistance was detected in all populations although exhibiting a slight decrease over time probably due to the interruption of field use . All vector populations were susceptible to diflubenzuron , recently introduced in the country to control Ae . aegypti . Resistance against deltamethrin was extremely high in three populations . Molecular assays investigated substitutions in the voltage gated sodium channel ( NaV ) , the PY target site , at positions 1011 , 1016 and 1534 . Elevated frequencies of substitutions Val1016Ile and Phe1534Cys related to high PY resistance levels were identified . Biochemical assays detected alterations in the activities of two detoxifying enzyme classes related to metabolic resistance , glutathion-S-transferases and esterases . The results obtained were evaluated in the context of both recent insecticide use and the records of dengue incidence in each locality . The four Ae . aegypti populations evaluated were resistant to the neurotoxic insecticides , temephos and deltamethrin . However , they were still susceptible to diflubenzuron . A probable correlation between adult insect resistance to PY and the domestic application of insecticides is discussed , pointing to the need for awareness measures regarding the correct utilization by citizens . This work aims to contribute to the efficient and rational management of Ae . aegypti control of both larvae and adults .
The mosquito Aedes aegypti is the main vector of dengue virus , an arbovirus of major importance worldwide . Among the Americas , Brazil is the country most affected by this pathogen [1] , considered hyper-endemic , since the four serotypes , DENV-1 to DENV-4 , circulate [2] . Recently , two other arboviruses transmitted by Ae . aegypti have been spreading rapidly , chikungunya ( CHIKV ) and Zika ( ZIKV ) . CHIK was introduced to America through the Caribbean and its presence was confirmed in Brazil in June 2014 [3] . ZIKV was introduced to the American continent from Northeast Brazil [4] . ZIKV was associated with several neurological disorders , including cases of microcephaly and other developmental alterations in newborns from mothers infected during pregnancy . This suspicion , later confirmed , launched the announcement of a Public Health Emergency of International Concern [5] that persisted until the end of 2016 [6] . Vectorial transmission of arboviruses depends upon three components , the host ( in this case , humans ) , the virus and the vector . Despite intense efforts of biomedical research , when DENV , CHIK and ZIKV are considered , neither effective vaccines for large scale use nor specific drugs , able to block clinical manifestations , are yet available on the market . Thus , strategies that focus on the control of mosquito vectors are currently the main tools against these health problems [1] . From a formal point of view , ‘information , education and social communication’ are a key component of the Brazilian dengue vector control program [7] . However in practice , insecticides play a very important role regarding control actions , from the perspective of both public managers and general society [8] . Ae . aegypti control in Brazil employs insecticides against larvae and adults . Larvicides are applied in households , 4 to 6 times a year , during visits of control agents , ideally only in water containers that cannot be discarded . In contrast , ultra low volume applications of adulticides do not have a preventive function . In spite of indiscriminate domestic use , these products are employed by health personnel only to block outbreaks in epidemic seasons or at strategic points , aiming to reduce the adult populations . One must be aware that Brazil follows WHO guidelines in order to decide which insecticides are employed in public health . In addition , when larvicides are considered for Ae . aegypti control , only products approved for drinking water are allowed [9 , 10] . For a long time , the temephos organophosphate ( OP ) was the sole larvicide available against Ae . aegypti . In Brazil , use of this OP started in 1967 when the vector was reintroduced in the country [11] . Due to the dengue epidemic in 1986 [12] , its use was intensified [8] . From 2009 on , the spread of temephos resistant Ae . aegypti populations relegated this OP to a secondary choice of larvicide in the country [13] , the recommendation , by WHO Pesticide Evaluation Scheme ( WHOPES ) , of the availability of other products for use in drinking water containers [10] , also contributing to this decision . In 2009 , the substitution of OP larvicides by Insect Growth Regulators ( IGR ) began . The first IGR adopted on a national scale was diflubenzuron , a chitin synthesis inhibitor ( CSI ) [14 , 15] . Ultimately , it was agreed that a rotation scheme , in principle every four years , would be adopted for larvicides [16] . Until 2001 , in addition to the temephos larvicide function , other OPs were used in conjunction for the control of adults . The above mentioned resistance of larvae to temephos induced the adoption of a different approach , consisting of distinct class insecticides against both larvae and adults . The aim in this case was to delay resistance development by varying selection pressure , now exerted with different products in distinct stages of the mosquito life cycle . With this strategy the Brazilian Ministry of Health ( MoH ) expected to preserve the few products that were still effective . That year , the OPs were replaced by pyrethroids ( PY ) for adult control . However , a rapid spread of Ae . aegypti resistance to PY ensued , due mainly to mutations in the target site , the NaV [17 , 18] . Therefore since 2009 , the MoH initiated the implementation of the OP malathion , the only non-PY adulticide recommended by WHO [9] . Brazil is a country of continental dimensions and , despite general MoH guidelines , there are local decisions , deriving both from public managers and the private initiative , that exert distinct pressures on vector populations . In addition , taking into account the varied genetic backgrounds of Ae . aegypti from different locations , it is appropriate to assume that distinct resistance profiles and mechanisms can be selected in different geographic regions . This multiplicity of scenarios justifies the importance of monitoring the insecticide resistance dynamics of natural vector populations , whether they are exposed to the pressure of a given insecticide or to its interruption in the field . In the present study , we evaluated the dynamics of resistance of Ae . aegypti populations of four distinct Brazilian regions over the course of one year . The chief insecticides employed by the National Dengue Control Program ( PNCD ) were considered . Two of the major associated mechanisms , metabolic and target site resistance , were also investigated . In the first case , the activity of different classes of detoxifying enzymes was quantified . The target site for PY was analyzed with molecular assays . Several insecticide resistance mechanisms are potentially effective against distinct insecticides simultaneously , belonging or not to the same class , a phenomenon known as cross-resistance . Therefore , the identification of resistance mechanisms involved in each specific situation , together with the possibility of using alternative insecticides bearing different modes of action ( such as the use of a CSI to replace OP ) is important in vector control programs . Furthermore , comparison of insecticide resistance levels and resistance mechanisms with the local history of chemical control can contribute to a rational management of resistance and , consequently , preservation of the insecticides still available .
The use of anesthetized mice to blood feed mosquitoes was authorized by Fiocruz Ethical Committee for Animal Use ( CEUA P-0498/08 and CEUA L-0007/09 ) Dengue is endemic throughout Brazil , except for the southernmost region . Four midsized tropical cities , each one in a different region , all with representative climate regimes and significant chronicles of dengue cases , were chosen ( Fig 1 , more details in [19] ) : Santarém , Pará ( PA ) State , North Region . 2°26’35”S , 54°42’29”W . This city , located in the Amazon region , bears an elevated humidity , high precipitation indices and temperatures ranging from 23 to 33°C . Santarém has a demographic density of 12 . 9 inhabitants/km2 [20] . Parnamirim , Rio Grande do Norte ( RN ) State , Northeast Region . 5°54’56”S , 35°15’46”W . Parnamirim has a milder and dryer climate compared to Santarém . Temperatures vary from 22 to 29°C and the city has 1 , 858 inhabitants /km2 [21] . Duque de Caxias , Rio de Janeiro ( RJ ) State , Southeast Region . 22°47’08”S , 43°18’42”W . Similar to Parnamirim , it is a densely populated city , with 1 , 828 inhabitants/ km2 [22] . Temperature differences among seasons are more pronounced compared to the North and Northeast Regions . During summer , heavy precipitation and flooding often occur . Campo Grande , Mato Grosso do Sul ( MS ) State , Central-West Region . 20°26’34”S , 54°38’47”W . Of all cities in this study , Campo Grande is located most above sea level , roughly at 600 meters altitude . Among the four study areas here presented , the highest temperature and precipitation amplitudes were registered in Campo Grande . Winter is particularly dry and cold in this city , with a demographic density of 97 inhabitants/ km2 [23] . Ovitraps were used for monthly egg collection [24] , in three 1 km2 areas in each of the four municipalities ( Fig 1 ) . Egg collection was initiated in November 2009 ( Duque de Caxias ) , December 2009 ( Parnamirim and Campo Grande ) and March 2010 ( Santarém ) proceeding for 12 months . In each 1 km2 area , 120 ovitraps were installed . In the laboratory , eggs of the parental generation collected in the field were reared until the adult stage , and specimens were identified up to the species level . Ae . aegypti parental adults always corresponded to at least 90% of field samples , and these mosquitoes were used to search for kdr mutations ( see below ) . For each municipality , on four occasions roughly at three month intervals , F1 colonies were established in order to perform bioassays and biochemical tests . As depicted in S1 Table , the number of adult females starting the colonies was always over 500 . Except for Parnamirim , 18–20 months after the last egg collection , a new field sample was obtained for each locality , and the F1 derived specimens were submitted to a temephos dose-response assay , as indicated below . Synchronously reared F1 L3 instar larvae or 1–3 day old adult females were used for bioassays with , respectively , larvicide or adulticide compounds ( see below ) . Rearing was performed essentially as described by Bellinato et al . ( 2016 ) [25] . The Rockefeller ( “Rock” ) strain was adopted both as an internal quality control of all assays and an insecticide susceptible reference lineage [26] . For each insecticide and each field population , as well as for Rockefeller , effective doses ( ED50 and ED95 ) were obtained by probit analysis with the aid of Polo-PC software [27] . Resistance ratios ( RR ) were then calculated by dividing the ED of field populations by that of the corresponding Rock . RR95 was used to compare all bioassays in accordance with the Brazilian MoH guidelines [28] . Quantitative bioassays were employed to evaluate the susceptibility status of field Ae . aegypti populations against the OP temephos and the CSI diflubenzuron . Eight to ten insecticide concentrations , varying from 0 . 006 to 0 . 072 mg/L for temephos and 1 . 0 to 5 . 5 μg/L for diflubenzuron , were used per assay . For each insecticide concentration in each assay , four replicates , each one with 20 or 10 larvae were exposed to temephos or diflubenzuron , respectively . This corresponds to a total of 640–800 larvae per temephos assay and to 320–400 larvae in the case of diflubenzuron . Each assay was repeated at least three times on different days , mortality varying between 10 and 95% [29 , 30] . Results were registered 24 hours after temephos exposure . In the case of diflubenzuron , according to protocols standardized previously [25 , 31 , 32 , 33] the bioassays were followed until adult emergence of all control specimens , not exposed to the CSI . Quantification of adult resistance to the deltamethrin PY was also performed through dose-response assays , adhering to methodology adapted from the original WHO protocol , with insecticide impregnated papers [34] . Up to 10 different deltamethrin concentrations were used per assay , varying between 2 . 1 and 109 . 6 mg/m2 , depending upon the susceptibility status of the field sample under test . Assays were repeated at least three times on different days . In all cases , three replicates with 15 to 20 adult females each were used . Quantification of enzyme activities potentially involved in insecticide detoxification was performed in agreement with a standardized biochemical procedure [35 , 36] . In all cases , 80 to 120 non-blood-fed young females ( up to 24 hours after emergence ) , stored at -80°C , were individually analyzed . For each female , the following enzyme activities were quantified: glutathione-S-transferase ( GST ) , esterase ( EST ) and mixed function oxidase ( MFO ) . Three substrates were employed for EST: α- and β-naphtyl and ρ-nitrophenyl acetates , accounting respectively , for activities named α-EST , β-EST and ρNPA-EST . According to former protocols , the 99 percentile of the susceptible control strain Rockefeller ( p99Rock ) was calculated for each enzyme class . Field population data were classified as follows: enzyme activity of any given population was considered unaltered when 0–15% specimens remained beyond p99Rock; values between 15 and 50% and above 50% were classified as altered or highly altered , respectively [35 , 36] . Allele-specific PCR was applied to investigate the presence of the Ile1011Met , Val1016Ile and Phe1534Cys mutations in the PY target site , NaV . Adults of the parental generation derived from monthly field collected eggs were used for evaluation of 1016 allelic frequencies . The other positions , 1011 and 1534 , were investigated in the first and last months of the first year interval . In all cases , genomic DNA was extracted from 30 individual adult males of each field sample . If no substitutions were detected at the three positions in all samplings of a population , 30 additional specimens were submitted to evaluation . This was done in order to obtain more accurate measures of kdr frequencies . Males were recruited in order to avoid the risk of contamination with spermathecae in the case inseminated females were used . The methodology described elsewhere [37 , 38] was followed .
Biological assays identified resistance to temephos in all the Ae . aegypti populations evaluated throughout the study period ( Fig 2 , S2 Table ) . Duque de Caxias presented the highest RR95 levels ( between 9 . 8 and 16 . 3 ) and Campo Grande , the lowest ( 3 . 6 to 7 . 9 ) . Nevertheless , a temephos resistance decay trend was observed in all cases in the period ( 2009–2012 ) although the rate of decay was different among populations . In the course of the study , temephos RR95 decreased up to 50% in mosquitoes from Campo Grande , 40% in Duque de Caxias , 30% in Santarém and 15% in Parnamirim . This result is compatible with the withdrawal of temephos in the four municipalities , as reported by each Municipal Health Secretariat . Despite the decrease in temephos resistance , the RR95 always remained above 3 . 0 . This value corresponds to the threshold defined by the MoH , above which temephos interruption is recommended [28] . It was also evident that slopes obtained for field populations were always lower than those for Rockefeller ( S2 Table ) , pointing to a higher heterogeneity compared to the control strain . Emergence inhibition of adults ( EI ) was the parameter evaluated in the dose-response tests of larvae exposed to diflubenzuron . Subtle variations in the effective doses were noted throughout the analyses of all populations with no apparent trend ( Fig 3 , S3 Table ) . Diflubenzuron RR95 always remained below 3 . 0 when compared to the Rockefeller strain , indicating susceptibility of field populations to this IGR . Moreover , slopes of the evaluated populations were always higher than the Rockefeller strain ( S3 Table ) , suggesting , unlike results for temephos , a greater homogeneity of these field populations in relation to diflubenzuron susceptibility . The deltamethrin RR95 was extremely elevated in all municipalities , always higher than 10 ( S4 Table ) . Excluding Parnamirim , where deltamethrin RR95 ranged between 10 . 1 and 14 . 3 , all other municipalities were above 35 . It is noteworthy that in Campo Grande , for example , the lowest value obtained was 58 . 2 . No trend in RR was noted during the study period , neither a tendency for decrease nor increase . Adult bioassay results are separate for each population , simultaneously with the dengue incidence in the intervals evaluated ( Fig 4 ) . In two locations , Duque de Caxias and Campo Grande , the highest RR values were in the period corresponding to the highest dengue incidence . In particular , the numbers of dengue cases in Campo Grande were compatible with an explosive outbreak . Adulticide applications by the municipal health agents in each locality were also included in Fig 4 . In this case , only the applications carried out in the study areas ( and not in the entire municipality ) are shown . Both the intensity and frequency of adult chemical control by the Municipal Health Secretariats varied widely . The intense use of deltamethrin by health agents in Duque de Caxias ( much more than the amount recommended by the MoH , see Discussion ) should be noted , as well as the use of malathion , a non-PY adulticide , in Campo Grande precisely the municipality where the largest dengue incidence was concomitant with the highest recorded deltamethrin RR95 . In general , as was the case with temephos and deduced from the slope values , heterogeneity of field populations was higher than that of the Rockefeller strain regarding deltamethrin status . Adult females were submitted to biochemical assays , disclosing changes in all classes of detoxifying enzymes ( Table 1 ) . GST and EST were the most affected activities . However , concerning esterases , greater alterations were observed with the "ρNPA" substrate . The MFO enzymes were the least altered in all populations . Only DQC and PNM populations presented GST and EST altered activities in all evaluated samples . In Campo Grande , no ρNPA-EST alteration was detected , and Santarém was the population with the least changes in the detoxifying enzymes . The presence and frequency of the Val1016Ile mutation in the Ae . aegypti NaV was investigated monthly in all field populations ( Fig 5 ) . Additionally , quantification of two other AaNaV mutations , Phe1534Cys and Ile1011Met , was performed in samples collected in the first and last months of evaluation for each population ( Table 2 ) . While the relation of the Val1016Ile and Phe1534Cys mutations with PY resistance is well documented [18 , 37] , the status of the Ile1011Met substitution [39] is still controversial ( see Discussion ) . Substitutions at positions 1016 and 1534 are recessive , i . e . , resistance to PY is expressed only in homozygosis [40] . Regarding the third position , there is evidence that the Ile1011Met mutation can be used as a marker of an early duplication event in this species , occurring in the wild type and susceptible genotype . Hence , decrease in the rate of the Ile1011Met substitution should occur in parallel with an increase of the more recent kdr mutants in the 1534 and 1016 positions [38] . Fig 5 shows , for the four populations evaluated , the kdr 1016Ile allelic frequency and the genotypic frequency of the kdr homozygotes at this position ( 1016Ile/Ile ) . For two populations , Duque de Caxias and Campo Grande , rates of the 1016Ile substitution were high throughout the evaluation period , allelic frequencies always above 70% . In these localities , in contrast to the low frequencies in the Ile1011Met mutation , allelic frequencies of the Phe1534Cys substitution were also high ( Table 2 ) . The substantial dissemination of these kdr mutations in Duque de Caxias and Campo Grande endorsed the high levels of PY resistance previously detected ( Fig 5 ) . In contrast in Parnamirim , mutations at 1016 and 1534 sites are present , notwithstanding at a low frequency . Accordingly in this population , the highest Ile1011Met frequencies were present ( Table 2 , see Discussion ) . No homozygous kdr specimens at position 1016 were detected in this population ( Fig 5 , Table 2 ) , and allelic kdr 1016Ile rates were always below 10% . Accordingly , deltamethrin RR levels in Parnamirin , although high , were much lower than those observed in the other three populations ( Fig 5 ) . Regarding position 1534 , approximately 13% of kdr homozygotes were found , suggesting participation of the PY target site mechanism in the resistance of this population to deltamethrin . In spite of the high deltamethrin resistance levels in Santarém ( Fig 4 , S4 Table ) , the 1016Ile mutation was not apparent in any mosquito from this population ( Fig 5 ) . However , the kdr 1534Cys allelic frequencies were high , 94% in the first month of collection and 100% at the end of the work . In parallel , the Ile1011Met frequencies were the lowest , reaching zero in the last evaluation ( Table 2 ) .
In Brazil , the monitoring of insecticide resistance in Ae . aegypti populations assists the rational management of chemical control . We accompanied , over the course of one year , the dynamics of resistance of four field populations belonging to different geographical scenarios and with distinct vector control policies . The quantification of resistance levels together with major resistance mechanisms with respect to the temephos and diflubenzuron larvicides as well as the deltamethrin adulticide , all employed in the control of this vector on a national scale , was considered ( S5 Table ) . The results are discussed taking into account the previous insecticide use and dengue cases in each locality . In Brazil , since 1967 until recently , temephos was the only larvicide adopted by the public health services for the control of Ae . aegypti . We confirmed that all populations evaluated in the present study were resistant to this OP . Our results are in accordance with prior reports [13 , 25 , 41] and even with MoH data which in 2009 already pointed to temephos susceptibility alterations in 90% of the evaluated Brazilian populations [42] . Resistance to OPs has also been observed throughout Latin America , with reports in several countries such as Colombia , Mexico , Cuba , Martinique and Argentina [43–50] . As stated by the Brazilian MoH , recommendations of Ae . aegypti chemical control management in the country are based on RR95 values . In the case of temephos , suspension is indicated when RR95 is above 3 . 0 [28] . In our study , we detected RR95 for temephos between 3 . 6 and 16 . 3 , the highest values in Duque de Caxias and Santarém . Despite the widespread resistance to temephos , in all municipalities a tendency for resistance ratios to decrease was observed during the evaluation period , attributed to the interruption of temephos utilization in the studied areas . However , resistance levels decreased slowly and the temephos RR95 of mosquito populations from all localities remained above the susceptibility threshold value , preventing the reutilization of this OP . Parnamirim , one of the municipalities evaluated here , is located in the metropolitan region of Natal , the capital of RN in the Northeast Region . In accordance with our results , data from the MoH also point to a decrease in Ae . aegypti temephos resistance levels in Natal , after replacement by Bti in 2005 . The RR95 was 18 . 6 in 2004 [36] and was reduced to 8 . 2 in 2007 [28] . In addition , Lima et al . ( 2011 ) [51] observed a decrease in temephos resistance in Ae . aegypti from Juazeiro do Norte , State of Ceará , also in the Northeast Region , after its discontinuation , temephos RR95 declining around 30% in six years ( from 10 . 4 in 2003 to 7 . 4 in 2009 ) . This same work registered an increase of temephos RR95 in two locations , Crato and Barbalha , that maintained temephos during this same period , from 7 . 5 to 30 . 0 in Barbalha and from 9 . 0 to 192 . 7 in Crato . Wirth and Georghiou ( 1999 ) [52] also reported , in Ae . aegypti from Tortola , a small Caribbean island , a decrease in temephos resistance ten years after application interruption in the field . RR90 , was 46 . 8 in 1985 [53] and declined to 6 . 3 in 1995/6 [54] . The Ae . aegypti resistance status to the inhibitors of chitin synthesis ( CSI ) , another class of larvicides recently introduced in the country , was also quantified . Taking into account the same cutoff established for temephos ( RR95 = 3 . 0 ) , diflubenzuron data point to susceptibility of all evaluated populations , confirming previous results obtained in the country for the dengue vector [25 , 30 , 55] . The recent introduction in Brazil of CSI compounds against Ae . aegypti , together with their distinct mechanism of action regarding conventional insecticides , contributes to the low resistance rates in the evaluated populations . In Brazil , except for the State of São Paulo , PY has only been adopted as an adulticide since 2000–2001 , after dissemination of resistance to OP in Ae . aegypti populations was confirmed [56] . This decision was made as a management strategy to expose larva and adult stages to compounds with different action mechanisms . However , mosquito field samples collected shortly afterwards ( 2002–2003 ) already exhibited signs of PY resistance [57] . Since then , PY resistance has been detected in several regions of the country [18 , 25 , 51 , 58 , 59] . Resistance to deltamethrin was extremely high in all populations studied here . Parnamirim exhibited the lowest RR95 levels , despite the magnitude , between 10 . 1 and 14 . 3 . For the remaining populations , deltamethrin RR95 was always above 35 . In Brazil , adulticides are not enlisted by public health managers as infestation prevention tools . Such products are used in attempts to block outbreaks or at strategic points such as airports and other potential vector entry points . According to the MoH , ultralow volume applications of adulticides should not exceed 5–7 times in the course of a year , in general during epidemic periods and in very specific situations and places [7] . However , a survey of the insecticide spatial applications against Ae . aegypti in the studied localities ( Fig 4 ) revealed a great variation and even an uncontrolled use of these products by local public managers . Many differences were detected among municipalities in adulticide applications , both in frequency and number . In several situations MoH recommendations were far exceeded , and up to nine applications have been registered in one single month . Despite this , the PY resistance levels during the study could not be temporally correlated to the ‘public’ spatial applications of adulticides in each locality . In two populations , Campo Grande and Duque de Caxias , the highest deltamethrin RR95 levels were registered precisely during periods of intense dengue transmission . It is worth mentioning that the Brazilian MoH considers that incidence rates beyond 300 dengue cases / 100 , 000 inhabitants are high [60] . In Campo Grande , in particular , the greatest dengue epidemic ever faced occurred during the period of our study ( see Fig 4 ) , when it also presented the highest deltamethrin resistance levels detected throughout the study . However , in this locality deltamethrin was not applied during this outbreak , the adulticide employed being the malathion OP . Our hypothesis is that arbovirus outbreaks cause a collective panic in the local population with a consequent pursuit towards individual protection and control measures . As a result there is a great and uncontrolled rise in the domestic use of PY products that are commercially available . This situation has a direct effect on the elevated PY resistance during epidemic outbreaks [8 , 25 , 61] . Although the variety of insecticides for public health use is limited , chemical management is still a relevant component of vector control programs . This combination leads to the rapid selection of resistant populations together with the exhaustion of available insecticides , often resulting in control impairment . However , in many situations the suspension of one specific insecticide by a given vector control program does not necessarily lead to its real field interruption due to its intensified domestic use at each new outbreak as well as continued availability of some products in the retail marked . In addition , the lack of integration of the different vector borne disease control programs cannot be neglected . A significant participation of alterations in the NaV , the PY target site in the central nervous system , in the resistance to this class of insecticides was apparent . The Val1016Ile mutation has been previously related to PY resistance in Brazilian Ae . aegypti populations [18 , 37] as well as those of other Latin American countries [62–65] . For this reason its frequency was monitored monthly . Later , the 1534Cys allele was identified in several places throughout the country , even in the absence of the 1016Ile kdr allele [37] . We then opted to investigate its frequency together with substitutions in the 1011 position , also potentially interfering with the NaV . Besides Brazil , the Phe1534Cys mutation is related to PY resistance in several other localities , such as the Cayman Islands and Thailand [66–67] . In our present study , we always found the 1534Cys mutant allele frequencies higher than those of the 1016Ile kdr allele ( Table 2 ) , a situation that corroborates previous evidences that the Val1016Ile substitution takes place after the kdr mutation at position 1534 , in a genetic background already containing the 1534Cys mutation [37 , 65 , 66 , 67] . Regarding the Ile1011Met mutation , there are indications that this substitution can be used as a diagnostic of a NaV duplication event [38] with an unclear relationship with PY resistance . The higher frequencies of 1011Met were evident in populations where the kdr 1016Ile and 1534Cys were lower ( Table 2 ) . Duque de Caxias and Campo Grande displayed extremely high levels of resistance to deltamethrin and also very high frequencies of kdr mutations at positions 1016 and 1534 . In Nova Iguaçu , a municipality contiguous to Duque de Caxias , the 1016Ile mutation was not detected in 2003 . However in 2008 , the allelic frequency of 1016Ile was 62 . 5% , two years later peaking at 95% [18 , 38] . A rapid increase in the 1016Ile allele was also observed in Campo Grande , frequency of 31 . 8% in 2008 [18] increasing to values above 85% in 2010 ( Table 2 ) , a situation suggestive of a rapid spread of this mutation in the region . The extremely high levels of PY resistance in Duque de Caxias and Campo Grande probably result from the combined effect of both mutations , 1016Ile and 1534Cys , as already described elsewhere [25 , 37 , 66] . In contrast , as expected , the 1011Met mutation frequencies were low in both municipalities . In the populations of Duque de Caxias and Campo Grande , the rapid increase in the frequency of the kdr allele 1016Ile is indicative of a strong selective pressure . A fast increase in the 1016Ile mutation frequency was also observed in Mexico . Until 1999 this substitution had not been detected , however , it was already high in 2008 . In 2011 , frequencies above 90% of the 1016Ile allele were reported in three regions of the country , suggesting imminent fixation of this kdr allele [68 , 69] . There is evidence that identification of PY resistance through laboratory assays may indicate impairment of spatial applications in the field [49] . In addition , similar to what was observed in the field , laboratory selection with PY of six Mexican Ae . aegypti populations in the course of five generations resulted in the increase of up to three-fold in the frequencies of the 1016Ile allele [70] . Although Ae . aegypti from Santarém possessed high rates of deltamethrin resistance ( RR95 between 35 . 0 and 60 . 0 ) , the 1016Ile kdr mutation was not detected and the 1011Met mutation frequency was very low , reaching zero , in these mosquitoes . Notwithstanding , 90–100% specimens were homozygous for the 1534Cys kdr mutation . Parnamirim , in the Northeast Region , presented the lowest deltamethrin RR as well as the lowest kdr 1016Ile and 1534Cys frequencies . In particular , the 1016Ile allele remained below 10% throughout the study . It is worth mentioning that mutations in this position had not been detected in NE Brazil in surveys prior to 2010 [18] . Later identification of 1016Ile in mosquitoes from Crato and Juazeiro do Norte , both in the State of Ceará [51] , suggests their recent arrival in this Region . The restricted use of PY in the field by Parnamirim local managers , together with the absence of dengue outbreaks in the period with probable reduced domestic use of insecticides , are probably the basis of the comparatively lower PY resistance levels in this locality , as well as their limited variation throughout the study . It is noteworthy that the highest frequencies of the 1011Met mutation appeared in Parnamirim , which is in agreement with previous evidence linking this mutation to a susceptible NaV haplotype [38] . Our results agree with data reported recently , relating high levels of pyrethroid resistance to multiple NaV mutations , a common situation in Latin America Ae . aegypti populations [71 , 72] Evaluation of metabolic resistance was achieved with biochemical tests quantifying the activity of the main classes of detoxifying enzymes . Although this methodology applied in vector population monitoring routine has the potential to reflect the general dynamics of resistance , we learned with its known limitations that it is not always possible to establish precise correlations between biologic and biochemical assays for each evaluated population at a given moment [36] . Herein , we attempted to compare the metabolic changes of adult females mainly with resistance to the adulticide deltamethrin . As expected , although there was no strict temporal correlation between the levels of PY resistance and the intensity of the metabolic changes for each population , of the three enzyme classes evaluated , GST and EST ( and especially ρNPA-EST ) were strongly altered while MFO was the least affected class . Regarding MFO , our data , as well as those from other Ae . aegypti Brazilian populations , differ from other countries whose PY resistance levels tend to correlate with MFO profile alterations [25 , 36 , 73] . Still , these results corroborate previous studies that related PY resistance in Brazilian field populations with increased GST and ρNPA-EST activities [36 , 74] . Santarém was the population with the lowest contribution of metabolic mechanisms to resistance levels , while Duque de Caxias and Parnamirim were the most affected . Alterations of detoxifying enzymes were identified as the main mechanism of PY resistance in Parnamirim taking into account the low frequency of kdr mutations in this population . Different from the other three populations , deltamethrin resistance levels in Parnamirim were consistently lower , also corroborating the strong contribution of PY target site alterations to the intensity of resistance to this class of insecticides , a situation already reported previously [40] . The influence of mutations in the PY target site on elevated resistance levels was confirmed with Santarém mosquitoes , whose high resistance ratios are parallel to the high frequency of the 1534Cys kdr allele although the detoxifying enzymes in this population were the least altered in the study . In agreement with this situation , the population of Campo Grande also revealed high levels of PY resistance and high frequencies of kdr mutations , while persistent changes in metabolic resistance were only detected for GST enzymes . Regarding resistance mechanisms , data indicate that different vector populations find different solutions to counteract the challenge represented by insecticides . This is attributed to the multifactorial nature of metabolic resistance as well as the abovementioned limitations of the biochemical methodology employed ( which quantifies general activities and not molecular species ) [36] . According to Moyes et al . ( 2017 ) [72] , there is currently plenty of evidence of resistance to the two main classes of insecticides employed all over the world , PY and the OP temephos . In particular , as data related to Latin America abundantly show , high resistance levels are common in the continent . Nonetheless , our results present a trend towards a slow decrease in Ae . aegypti resistance to temephos since discontinuation of this OP larvicide in the field started in 2009 . The CSI susceptible levels are probably a consequence of the recent introduction in the Ae . aegypti control routine in the country . In contrast , extremely high and disseminated PY resistance levels were noted , indicating a significant participation of the domestic use of this class of compounds in the selection pressure of Brazilian vector populations ( S5 Table ) . The exacerbated domestic use of PY insecticides is seasonal , occurring mainly during outbreaks , and it can be accompanied by the seasonal elevation of resistance levels . Finally , it was possible to highlight the limitations of chemical control as the main methodology for Ae . aegypti control , taking into account both larvae and adults . There is growing evidence of the need for joint actions with other types of methodologies , social mobilization , mechanical control and biological complementary alternatives . We believe that the adoption of insecticides in a rational way is a strategy to complement other types of controls . | Among the pathogens transmitted by Aedes aegypti , dengue virus is the most important due to the number of people affected or at risk and the high rate of mortality worldwide . The confirmation that Ae . aegypti is also the vector of Zika , chikungunya and urban yellow fever poses serious consequences for public health , pointing to the need of reevaluating current vector control strategies . Although there is growing recognition of the importance of social participation and community engagement to prevent high levels of infestation , insecticides are considered important vector control tools . Nevertheless , the massive and indiscriminate adoption of insecticides to control larvae and adults contributes to resistance spread . In particular , the domestic use of adulticides , especially in epidemic seasons , is assumed to induce high levels of resistance in Ae . aegypti populations . However , the consequences of insecticide interruption upon the resistance of field populations has been less investigated . We evaluated , in four Brazilian regions over one year , the dynamics of dengue vector population resistance to the principal insecticides used in the country . The main resistance mechanisms were also investigated . Data are discussed taking into account the potential relationship among dengue outbreaks , public and private chemical control and insecticide resistance . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"larvicides",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"geographical",
"locations",
"enzymology",
"animals",
"developmental",
"biology",
"mutation",
"substitution",
"mutation",
"pest",
"control",
"enzyme",
"metabolism",
"infectious",
"disease",
"control",
... | 2018 | The impact of insecticide applications on the dynamics of resistance: The case of four Aedes aegypti populations from different Brazilian regions |
Podoconiosis is an environmental lymphoedema affecting people living and working barefoot on irritant red clay soil . Podoconiosis is relatively well described in southern Ethiopia , but remains neglected in other parts of the Ethiopian highlands . This study aimed to assess the burden of podoconiosis in rural communities in western Ethiopia . A cross-sectional study was conducted in Gulliso woreda ( district ) , west Ethiopia . A household survey in the 26 rural kebeles ( villages ) of this district was conducted to identify podoconiosis patients and to measure disease prevalence . A more detailed study was done in six randomly selected kebeles to describe clinical features of the disease , patients' experiences of foot hygiene , and shoe wearing practice . 1 , 935 cases of podoconiosis were registered , giving a prevalence of 2 . 8% . The prevalence was higher in those aged 15–64 years ( 5 . 2% ) and in females than males ( prevalence ratio 2 . 6∶1 ) . 90 . 3% of patients were in the 15–64 year age group . In the detailed study , 335 cases were interviewed and their feet assessed . The majority of patients were farmers , uneducated , and poor . Two-third of patients developed the disease before the age of thirty . Almost all patients ( 97 . 0% ) had experienced adenolymphangitis ( ALA - red , hot legs , swollen and painful groin ) at least once during the previous year . Patients experienced an average of 5 . 5 ALA episodes annually , each of average 4 . 4 days , thus 24 working days were lost annually . The incidence of ALA in podoconiosis patients was higher than that reported for filariasis in other countries . Shoe wearing was limited mainly due to financial problems . We have documented high podoconiosis prevalence , frequent adenolymphangitis and high disease-related morbidity in west Ethiopia . Interventions must be developed to prevent , treat and control podoconiosis , one of the core neglected tropical diseases in Ethiopia .
Podoconiosis ( endemic non-filarial elephantiasis ) is a non-infectious geochemical disease caused by exposure of bare feet to red clay soil derived from volcanic rock . It results in progressive bilateral swelling of the lower legs . Podoconiosis is mostly a disease of agrarian people who work barefoot , particularly on red clay soils of volcanic areas [1] , [2] , [3] , [4] . Mineral particles , absorbed through the skin of the foot , are taken up into macrophages in the lower limb lymphatics and are thought to induce an inflammatory response in the lymphatic vessels , leading to fibrosis and obstruction of the vessel lumen [5] . Podoconiosis is widespread in highland areas of tropical Africa , Central America and northern India . It is considered to be a considerable public health problem in more than ten African countries including Uganda [6] , Tanzania [7] , Kenya [8] , Rwanda , Burundi , Sudan , Ethiopia [9] , Cameroon [1] , [10] , and Equatorial Guinea [11] . It is estimated that the total number of cases per country is highest in Ethiopia [12] , [13] , [14] , [15] . In Ethiopia the basalt area covers more than 200 , 000 km2 which is approximately one-fifth of the land surface , and the fertility of the soil in such areas attracts an agricultural population of 20 . 5 million people [5] . Eleven million Ethiopians ( 18% of the population ) are at risk through exposure to the irritant soil , and estimate based on prevalence data from an endemic area in southern Ethiopia suggests that between 500 , 000 and 1 million people are affected . In Ethiopia most studies on podoconiosis have been conducted in Wolaita zone , southern Ethiopia . In Wolaita zone the prevalence of podoconiosis is over 5% [4] , [16] , and people with podoconiosis are half as productive as controls , costing the zone more than US$16million annually [17] . Furthermore , podoconiosis is one of the most stigmatizing health problems in the zone [18] , [19] , [20] . An epidemiological study in resettlement schemes of west Ethiopia showed that the prevalence of podoconiosis is higher in indigenous ( 9% ) people than in settlers ( 5% ) . People from non-endemic areas develop disease on average 9 years after being moved to an endemic area [21] . Apart from this , there are no recent studies on podoconiosis in west Ethiopia in general and in West Wollega zone in particular where podoconiosis is common . The present study aims to determine the magnitude of podoconiosis in Gulliso woreda of West Wollega zone in terms of prevalence , socio-economic impact , incidence/prevalence of associated morbidities , and to assess the experience and perspectives of patients regarding prevention and treatment .
Ethical approval for the study was granted by the Ethical Review Committee of the Oromia Regional Health Bureau . Informed verbal consent was obtained from the study participants before conducting the study . When children aged under 18 years ( the legal age for giving consent for research in Ethiopia ) were encountered , consent was obtained from their parents or guardians , and assent was obtained from children aged 12 or above . The use of verbal consent was approved by the ethical review commitee because the majority of the study participants cannot read and write . During and at the end of the survey , enumerators and health workers provided health education about foot hygiene and foot wear to prevent podoconiosis and treat early stages of the disease . Cases with advanced form of the disease were advised to visit the nearest health facility ( Ayira Hospital ) for advanced care and treatment . The study took place in Gulliso woreda ( a government administrative unit , equivalent to a district ) , West Wollega zone , Oromia region of Ethiopia . The Woreda is located 500 km west of Addis Ababa , the capital city of Ethiopia , and has an altitude of 1 , 500–1 , 800 m above sea level . The population of the woreda is 69 , 856 , of which 88 . 7% live in 26 rural kebeles and are subsistence farmers producing coffee as a cash crop . For many decades the study population has had access to well functioning health services though hospital and clinics run by non-governmental organizations . The population has also had above-average primary and secondary education resulting in a relatively higher literacy rate than the national average for rural areas in Ethiopia . The present study is a cross sectional quantitative study conducted in two phases . The first part of the study involved a survey of all households in the 26 rural kebeles ( the smallest administrative unit/village in Ethiopia ) of Gulliso woreda to identify podoconiosis patients . The head of the household was asked whether any member of the household had podoconiosis . When a case was reported in the household , his/her sex , age , and marital status were recorded , and both feet were clinically examined . Following this , the second part of the study was conducted in 6 kebeles selected using the simple random sampling method . Kebeles were allocated random numbers that were written in pieces of papers . All papers were rolled and shuffled in a hat , and six were drawn . The selected kebeles were Eka , Hawate Sutchi , Jarso Lalo , Sage Guji , Saka Jirbi , and Moga Kobera . All 335 patients identified in these kebeles were included in the study population . A structured questionnaire was administered to the identified patients to explore demographics of patients , features of disease , and previous experiences and future perspectives of patients on prevention and treatment methods of podoconiosis . The first part of the study was conducted by 26 health extension workers in government service in Gulliso woreda . The researchers trained the health extension workers on the nature , aetiology , treatment and prevention of podoconiosis . Practical training was also provided on clinical diagnosis and disease staging using a recently developed podoconiosis staging system [22] . A pre-test done in Gulliso town to test the skills of the health extension workers ensured reliability of their diagnostic skills . Following this , the health extension workers registered people with podoconiosis in their respective kebeles over a period of two weeks . In the second part of the study seven nurses led by a member of the research team administered a structured questionnaire to podoconiosis patients in the 6 selected kebeles . Clinical characteristics included ( i ) disease onset; ( ii ) adenolymphangitis ( ALA ) defined as painful inflammation of the foot and leg with swollen lymph nodes and fever [14] , and its consequence on patients' health seeking behaviour and morbidity; and ( iii ) clinical assessment of legs and feet of patients with regard to presence of mossy lesions ( fluid filled or papylomatous horny lesions giving the skin a rough appearance [22] ) and wounds . The largest circumference of the leg was measured using a tape to a precision level of the nearest centimeter between the level of the ankle and knee . Data were entered and analysed using the Statistical Package for Social Sciences ( SPSS ) software v . 10 . 0 . Descriptive statistics was done using summary statistics such as frequency , mean/median , and summary figures and tables . The overall prevalence of podoconiosis was calculated as the ratio of the number of patients with podoconiosis and the total population in the surveyed kebeles . Statistical significance was tested using the Chi square test . The level of significance was set at α of 0 . 05 .
Table 1 illustrates characteristics of podoconiosis affected individuals identified in the survey . Of 1 , 935 patients , 659 ( 34 . 1% ) were male , and the male to female prevalence ratio was 1∶2 . 6 , showing podoconiosis to be more than twice as prevalent among females as males . The age distribution of cases shows that 90 . 3% of the cases belonged to the economically productive age group ( χ2 = 31 . 7 , p<0 . 001 ) . The overall prevalence of podoconiosis was 2 . 8% ( 1 , 935/69 , 465 ) ( 95% CI = 2 . 1% to 3 . 5% ) . The prevalence was even higher among the economically active age group of 15–64 years ( 5 . 2% ) . The prevalence was also higher in kebeles with non-settler population than those with recent settler population ( 2 . 8% vs . 1 . 4% , χ2 = 14 . 79 , p<0 . 001 ) . The mean age of patients at the time of survey was 40 . 7 years . Of patients aged 18 years and above , 17 . 3% were never married . Of 1 , 935 patients identified in the survey , 335 ( 17 . 3% ) were approached for the detailed study in six randomly selected kebeles . Characteristics of the respondents are presented in Table 2 . The detailed study subjects consisted of 243 ( 72 . 5% ) females and 92 males ( 27 . 5% ) , giving a male to female ratio of 1∶2 . 6 . About 85% of the patients belonged to the economically active age group of 15–64 years . More than half of the patients had no formal education . The proportion of uneducated females was higher than that of males , and the difference was statistically significant ( χ2 = 9 . 08 , p = 0 . 003 ) . The majority ( 80 . 6% ) of the study participants were farmers . The average age of onset of podoconiosis was 26 years , and the average duration of illness between time of onset and time of interview was 17 years . Forty percent said the disease started before age 20 and two-thirds said it started before age 30 . There was no statistical difference between the sexes with regard to onset of podoconiosis . The majority of patients ( 325 , 97% ) had experienced adenolymphangitis at least once during the one year period preceding the date of interview . Patients reported an average of 5 . 5 such episodes/year and 96% said they had to stay in bed ( mean 4 . 4 days/episode ) ; meaning that on average each patient lost 24 days of activity per year ( range , 0–192 days per year ) . During episodes of adenolymphangitis , the majority of patients sought treatment either at a clinic ( 56% ) or a pharmacy ( 15% ) . The frequency and duration of ALA was higher in patients with larger circumference of the leg measured below the level of the knees ( χ2 = 5 . 94 , p<0 . 05 and χ2 = 6 . 67 , p<0 . 01 , respectively ) . Neither the frequency of attacks nor the duration of illness were significantly associated with age , sex , or age of onset of the illness . Table 3 shows parameters used for clinical characterization . Mossy lesions were observed in 175 ( 53% ) patients . It was more common among males ( 64% ) than females ( 49% ) , and the difference was statistically significant ( OR = 1 . 89 , 95% CI = 1 . 15–3 . 11 , p = 0 . 011 ) . Open wounds were present in 24 ( 7 . 2% ) patients and were more common among patients that had mossy lesions than those without ( 12 . 6% vs . 1 . 3% , respectively; χ2 = 15 . 6 , p<0 . 001 ) . Respondents were asked about their experience of and attitudes towards footwear and personal hygiene , which are central to prevent , treat and control disease progression . The majority of patients reported that they had no problem finding enough water and that it took an average of 10 minutes' walk to reach a water source . Two-thirds of patients reported washing their feet at least once per day , and 58% said they washed their feet with soap daily . Of the study participants in the detailed study , 303 ( 96% ) had worn shoes at least once in their life . The mean age when patients started wearing shoes was 23 years ( ±15 . 9 ) , a time often coinciding with the onset of signs and symptoms of podoconiosis . The experience of wearing shoes did not vary between males and females . However , the type and quality of shoe worn varied , more males than females wearing the better quality and more expensive leather shoes ( 19 . 6% vs . 8 . 7% , χ2 = 7 . 4 , p = 0 . 007 ) . The times when people walked barefoot were in the field ( 22% ) , during rainy season ( 13% ) and at home ( 11% ) ( Table 4 ) .
The present study showed that podoconiosis is a problem of public health importance in Gulliso woreda , west Ethiopia . The prevalence of podoconiosis in Gulliso woreda ( 2 . 8% ) is lower than reports from Wolaita zone , southern Ethiopia ( 5 . 5% ) [4] . One of the limitations of our study is that it reported only cases of podoconiosis with overtly swollen legs . In addition , cases were examined only when they were reported by the head of the household . Given the stigma associated with podoconiosis , a number of patients may not have shown themselves to the study team . This may have resulted in an underestimate of the prevalence of podoconiosis . Moreover , the study woreda has relatively better access to education and health care , and is not likely to represent the most remote woredas in the zone or region , where higher prevalence rates are expected . It is possible that disease incidence has declined with better education , more shoe wearing and access to health care when compared to the other study areas . Recent studies also indicated the role of genetic factors in podoconiosis [23] , and differences in disease prevalence among various population groups may be due to differences in frequency of genetic variants that confer susceptibility to podoconiosis . The prevalence of the disease in the woreda is still higher than that of all forms of tuberculosis ( estimated to be 579 per 100 , 000 population for Ethiopia ) [24] and HIV/AIDS ( 2 . 0% in Oromia region ) [25] . As reported in other studies , the disease starts in the second and third decades of life and its prevalence and severity increases up to the 6th decade of life , corroborating the suggestion that podoconiosis is a chronic condition that disables but rarely kills [16] . The prevalence of podoconiosis was much lower in villages dominated by settler populations . An epidemiological study in resettlement schemes of west Ethiopia showed that prevalence of podoconiosis is higher in indigenous people than in settlers ( 9 . 0% vs . 5 . 0% ) . People from non-endemic areas develop disease on average 9 years after being moved to an endemic area [21] . Moreover , observations in our survey showed that the kebeles with settler populations had higher use of footwear , better personal hygiene practices and a relatively advanced standard of living . Interestingly , the prevalence of podoconiosis was higher among females than males similar to the studies in Ocholo ( 1∶4 . 2 ) [26] , Wolaita ( 1∶1 . 4 ) [14] and Pawe ( 1∶1 . 4 ) [27] . An epidemiological study in settler populations in Keffa region [21] showed male predominance . Our survey also showed that in the two kebeles with settler populations the prevalence was slightly higher among males than females , but the difference was not statistically significant . In contrast , a more recent community based survey in Wolaita zone reported that the prevalence in males and females is similar [4] . The study also reported that 10% of women were not willing to be examined due to modesty , and this may have underestimated the prevalence in women . Our survey showed that early stages of podoconiosis were more common in women and severe forms of the disease and mossy lesions were more prevalent in men . This may be because females look for help earlier , as appearance constitutes a greater motivation for them . Moreover , women may have a better access to rinsing water during their routine activities in the household such as fetching water and washing clothes . This curbs progression of the disease to the advanced stages . Men may be indifferent to the condition initially , so do not look for care and do not look after the leg . Once the disease becomes severe they will also admit to having the condition . One of the most important findings of this study is the high frequency of acute episodes of adenolymphangitis in 97% of the study population . Most patients sought medical treatment for acute episodes . Many described the occurrence of monthly episodes as coming always during ‘chagino’ , the local term to describe the time when the moon is absent . The study documented frequent acute attacks among patients , probably more frequent than those of malaria which is currently declining in the woreda . The associated symptoms and the subsequent exhaustion that lasts for days [13] cause serious morbidity and long absence from productive work . This presents further evidence of the widely held opinion that podoconiosis aggravates poverty due to working days lost ( almost one month/year ) and expenditure on health care , often ineffective . The incidence of adenolymphangitis and the resulting incapacitation found in our study was higher than that reported for filariasis in southeast Tanzania [28] , south India [29] , and Papua New Guinea [30] . Acute attacks of ALA were significantly positively associated with leg circumference . The larger the swelling , the less likely the person to find adequate footwear , leaving the legs more exposed to soil and predisposed to acute attacks . The socio-economic impact of the disease is high . Of 10 patients , nine belonged to the economically active age group . Previous studies also showed the huge economic burden of podoconiosis [17] , and high prevalence of the disease among the economically active age group [4] . Over half of the study subjects in our survey were uneducated , the rate being even higher among females . Moreover , the proportion of unmarried women among all patients was significantly higher than the proportion for rural Oromia region ( 19 . 2% vs . 6 . 1% , χ2 = 12 . 5 , p<0 . 001 ) , indicating stigmatisation [31] . Previous studies demonstrated several manifestations of social stigma among podoconiosis endemic communities [18] , [32] . The disease leads to social exclusion of individuals and their families , e . g . school dropout , lack of marriage prospects , exclusion from community events . Social exclusion also leads to the invisibility of podoconiosis . The belief that effective treatment does not exist may discourage patients and health workers . These may be some of the factors that help explain the lack of attention paid to this important and common disease by researchers , policy makers , and health professionals . The World Health Organization defines access to drinking water as the availability of at least 20 litres of water per person per day within a round trip walking distance of 30 minutes [33] . Accordingly , the patients in our survey are at a reasonable distance from the source of water ( 20 minutes round trip ) , but may not have access to adequate water . The use of protective shoes , combined with consistent washing with soap and water was uncommon among study subjects . There were also gender disparities in the quality of shoes worn . Disproportionately fewer females wore the better quality leather shoes compared to males . In conclusion , we have documented high podoconiosis prevalence , frequent adenolymphangitis and high disease-related morbidity in west Ethiopia . This was evidenced by the high prevalence of the disease , incidence of adenolymphangitis and the associated socio-economic impact . Interventions must be developed to raise awareness of podoconiosis , to prevent , treat and control the disease . The higher prevalence among women implies that women's development and empowerment programs that operate in west Ethiopia must embrace podoconiosis prevention and control programs . Furthermore , podoconiosis control and prevention programs must give additional attention to women to treat and rehabilitate them so that they can resume important social and economic roles in the community . | Podoconiosis is a chronic non-infectious disease resulting in below-knee swelling of the legs in bare-footed people living in red clay soil areas . It is an important and yet neglected problem in tropical Africa , central and south America , and north India . Podoconiosis can be prevented by consistently wearing shoes and washing feet . We aimed to assess the burden of the disease , to characterize features of the disease , and to describe foot hygiene and shoe wearing practice of patients in west Ethiopia . First , we did a survey of the 26 rural villages . We identified 1 , 935 podoconiosis patients , giving a prevalence of 2 . 8% . Podoconiosis was twice as prevalent in females as males . Second , we did a more detailed study among 335 patients in six randomly selected villages . We found that the majority of patients were farmers , uneducated , and poor . The disease developed before the fourth decade of life and the majority of patients became bed-ridden because of frequent attacks of red , hot legs and swollen and painful groin . Shoe wearing was limited mainly due to lack of money . We conclude that podoconiosis imposes a huge burden in west Ethiopia , and recommend that interventions be developed to prevent , treat and control the disease . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"primary",
"care",
"epidemiology",
"global",
"health",
"public",
"health"
] | 2011 | Burden of Podoconiosis in Poor Rural Communities in Gulliso woreda, West Ethiopia |
The increase in availability of whole genome sequences makes it possible to search for evidence of adaptation at an unprecedented scale . Despite recent progress , our understanding of the adaptive process is still very limited due to the difficulties in linking adaptive mutations to their phenotypic effects . In this study , we integrated different levels of biological information to pinpoint the ecologically relevant fitness effects and the underlying molecular and biochemical mechanisms of a putatively adaptive TE insertion in Drosophila melanogaster: the pogo transposon FBti0019627 . We showed that other than being incorporated into Kmn1 transcript , FBti0019627 insertion also affects the polyadenylation signal choice of CG11699 gene . Consequently , only the short 3′UTR transcript of CG11699 gene is produced and the expression level of this gene is higher in flies with the insertion . Our results indicated that increased CG11699 expression leads to xenobiotic stress resistance through increased ALDH-III activity: flies with FBti0019627 insertion showed increased survival rate in response to benzaldehyde , a natural xenobiotic , and to carbofuran , a synthetic insecticide . Although differences in survival rate between flies with and without the insertion were not always significant , when they were , they were consistent with FBti0019627 mediating resistance to xenobiotics . Taken together , our results provide a plausible explanation for the increase in frequency of FBti0019627 in natural populations of D . melanogaster and add to the limited number of examples in which a natural genetic mutation has been linked to its ecologically relevant phenotype . Furthermore , the widespread distribution of TEs across the tree of life and conservation of stress response pathways across organisms make our results relevant not only for Drosophila , but for other organisms as well .
Understanding the functional consequences of naturally occurring mutations is one of the key challenges in modern biology . Recent years have seen an explosion in the availability of genomic data that have opened up the possibility of searching for adaptive mutations on an unprecedented scale [1] . Although there are some examples in which adaptive mutations have been connected to their phenotypic effects [2]–[5] , our knowledge of the functional consequences of particular genetic variants is still very limited . Mapping genotype to phenotype is a difficult task due to the large number of genes that contribute to some phenotypes , to the pervasiveness of genetic interactions , and to the complex environmental influences on the phenotypic outcome [6] , [7] . Current efforts in genotype-phenotype mapping include projects in several model organisms [8] . Among them , Drosophila melanogaster is one of the most promising cases due to the high quality gene annotation , deep understanding of developmental , physiological , and metabolic networks , and the availability of genetic resources . Because genes tend to work in evolutionarily conserved pathways , genotype-phenotype insights obtained in D . melanogaster provide valuable information that is relevant for other organisms as well [7] . Most ongoing projects in Drosophila focus on mapping SNP variants to a given set of phenotypic traits such as olfactory behavior or stress resistance [9]–[12] . While SNPs certainly contribute to ecologically relevant phenotypes , these efforts ignore other types of mutations , such as those caused by transposable element ( TE ) insertions . TEs have the ability to generate mutations of great variety and magnitude , ranging from subtle regulatory mutations to large genomic rearrangements that can have complex phenotypic effects . Additionally , TEs have been shown to be susceptible and responsive to environmental changes; as such , they might have an important role in environmental adaptation [13]–[15] . We have recently used TEs as a tool to identify putatively adaptive mutations to the out-of-Africa environments in D . melanogaster on a genome-wide scale [16] , [17] . We screened 763 TEs and identified 18 putatively adaptive TEs based on their population dynamics [16] , [18] . For a subset of the candidate TEs , we also demonstrated that they show signatures of selective sweeps [16] , [19] , evidence of population differentiation [17] , and two of them , FBti0019430 and FBti0018880 , have already been linked to adaptive fitness effects [20]–[22] . Thus , putatively adaptive TEs in this set are good candidates to perform follow-up experiments that should allow us to map genotype to phenotype and to identify the underlying mechanisms of adaptive mutations . In this study , we focused on mapping one of the previously identified putatively adaptive insertions , the 186 bp POGON1 element FBti0019627 , to its ecologically relevant phenotype . FBti0019627 is inserted in the 3′ UTR region of kinetochore Mis12-Ndc80 network component 1 ( Kmn1 ) gene , and it is closely located to CG11699 , a gene of unknown function ( Figure 1A ) [23] . Kmn1 and CG11699 genes partially overlap and encode cis-natural antisense transcripts [24] . FBti0019627 has recently increased in frequency in out-of-African populations most likely due to positive selection , as suggested by the signatures of a selective sweep in the flanking regions of this TE , including CG11699 and Kmn1 coding sequences [16] . Here , we used an integrative approach , that combines gene structure and gene expression analyses , protein modeling and docking simulations , enzymatic activity and stress resistance assays , to map genotype to phenotype while disentangling the molecular and biochemical mechanisms underlying the adaptive effect of FBti0019627 insertion . We show that , besides being incorporated into Kmn1 transcript , FBti0019627 affects the choice of polyadenylation signal of CG11699 and as a result , only the short 3′ UTR transcript of this gene is produced . These structural changes are associated with increased CG11699 expression in flies with the insertion , leading to xenobiotic resistance through increased ALDH-III activity . Xenobiotic resistance is an ecologically relevant phenotypic trait that provides a plausible explanation for the recent increase in the frequency of FBti0019627 insertion due to positive selection [16] , [17] .
To confirm that the TE is inserted in the Kmn1 transcript ( Figure 1A ) , we performed 3′ RACE experiments in flies from outbred-1 populations with and without FBti0019627 insertion ( see Material and Methods ) . As expected , we found that the TE is incorporated into the Kmn1 transcript in flies with the insertion , while flies without the insertion have a 188 bp shorter transcript due to the absence of the TE ( Figure 1B ) . Additionally , we discovered a previously unreported transcript with a much shorter 3′ UTR , only 73 bp long , that is present both in flies with and without the insertion ( Figure 1B ) . To check whether the TE also affects the structure of CG11699 , we carried out 3′ RACE experiments . We found that while flies with FBti0019627 insertion have only one transcript with a 110 bp long 3′ UTR , flies without the insertion have two transcripts that differ in the length of their 3′ UTRs: 110 bp long and 221 bp long ( Figure 1B ) . We further analyzed whether the difference in CG11699 transcripts present in flies with and without the insertion is due to FBti0019627 insertion . We identified the cleavage site of each transcript and performed a motif search analysis to identify the polyadenylation signals ( PASs ) and GU-rich downstream sequence element ( DSEs ) that are most likely being used to generate the short and the long 3′ UTR transcripts [25] . We found a weak proximal PAS and a strong distal PAS and their corresponding DSEs upstream and downstream respectively of the two cleavage sites ( Figure 1C ) . In flies with the insertion , the TE is inserted between the PAS and the distal cleavage site disrupting the DSE ( Figure 1D ) . In flies with the insertion , the distal cleavage site is not used; as a consequence , only the transcript with the short 3′ UTR is produced . Because flies with and without the insertion differ in CG11699 transcript isoforms , the length of the overlapping region between this gene and Kmn1 is also different: 28 bp in flies with the insertion and 140 bp in flies without the insertion ( Figure 1B ) . Overall , our results indicated that , besides being incorporated into Kmn1 transcript , FBti0019627 insertion affects the PAS choice of CG11699 . As a result , flies with and without this insertion differ in their CG11699 transcript isoforms and in the length of the overlap between CG11699 and Kmn1 . We decided to focus on CG11699 for further investigation . To confirm that only the short 3′ UTR transcript is produced in flies with the insertion , and to determine the relative abundance of the short and long 3′ UTR transcripts in flies without the insertion , we performed transcript-specific qRT-PCR in flies from outbred-1 populations ( see Material and Methods ) . We found that in flies with the insertion , the long 3′ UTR transcript is barely detectable . This confirmed that when the TE is present , only the short 3′ UTR transcript is produced ( t-test p-value = 0 . 003 and 0 . 004 male and female respectively , Figure S1 ) . On the other hand , in flies without the insertion ∼70% of the total CG11699 expression is due to the longer transcript ( t-test p-value = 0 . 008 and 0 . 013 male and female respectively , Figure S1 ) . These results are in accordance with the computational prediction of the distal PAS being stronger than the proximal PAS ( Figure 1C ) . Short 3′UTR transcript isoforms usually show increased relative expression levels compared to long 3′UTR transcript isoforms [26] , [27] . Thus , we expected that flies with the insertion would have a higher level of expression of CG11699 compared to flies without the insertion , because 100% of the CG11699 isoforms are short in flies with the insertion , while only 30% of the isoforms are short in flies without the insertion . Indeed , our results showed that flies with the insertion have an increased level of expression of CG11699: ∼2 . 6 fold ( t-test p-value = 0 . 011 ) in males and ∼2 . 3 in females ( t-test p-value = 0 . 029 ) ( Figure 2 ) . Thus , FBti0019627 insertion affects the relative abundance of the short and long 3′ UTR transcripts and it is also associated with an overall increased expression of CG11699 . CG11699 encodes a transmembrane protein of unknown function that physically interacts with Aldehyde dehydrogenase III ( ALDH-III ) [28] . It has been shown that over-expression of CG11699 increases ALDH-III activity in phosphorylated membrane extracts [29] . We hypothesized that flies with FBti0019627 insertion , which have increased CG11699 expression , would have increased ALDH-III activity in the membrane . To test this hypothesis , we measured ALDH activity using different concentrations of benzaldehyde , which is a highly reactive substrate of this enzyme ( see Material and Methods; [29] , [30] ) . We compared ALDH-III substrate-activity curves in flies with and without the insertion from the outbred-1 populations . In agreement with our expectations , we found that flies with the insertion have significantly higher ALDH-III enzymatic activity than flies without the insertion ( p-value = 0 . 0042 ) ( Figure 3 ) . Benzaldehyde is highly toxic when present at concentrations that are too high to be rapidly eliminated because it readily form adducts with DNA , RNA , and proteins [31] . Additionally , benzaldehyde generates reactive oxygen species ( ROS ) that induce lipid peroxidation in the membrane [32] . ALDH-III not only metabolizes exogenous aldehydes , such as benzaldehyde , but also plays a protective role against endogenous aldehydes generated as a result of lipid peroxidation [30] , [33] . Therefore , flies with FBti0019627 insertion that show increased ALDH-III activity ( Figure 3 ) should be more resistant to high doses of benzaldehyde . To test this hypothesis , we compared the survival rate of outbred-1 populations with and without FBti0019627 insertion after an acute exposure to benzaldehyde ( see Material and Methods ) . We analyzed 3 replicas of 50 flies each per sex and per strain for unstressed and stressed conditions ( 1 , 200 flies total ) . While there were no differences in survival rate between flies with and without the insertion in unstressed conditions , we found that flies with the insertion showed increased survival rate compared to flies without the insertion when exposed to high concentrations of benzaldehyde ( females t-test p-value = 0 . 0035 , odds-ratio ( 95% confidence intervals ) = 3 . 11 ( 1 . 48–6 . 50 ) , and males t-test p-value = 0 . 026 , odds-ratio = 4 . 48 ( 2 . 24–8 . 96 ) ; Figure 4 ) . We confirmed these results by replicating the experiment using a larger sample size ( 9 replicas of 50 flies each: 3 , 600 flies in total ) ( Figure 4 ) . Again , both male and female flies with the insertion were more resistant to benzaldehyde than flies without the insertion ( females t-test p-value≪0 . 001 , odds-ratio = 7 . 53 ( 5 . 42–10 . 46 ) and males t-test p-value = 0 . 0017 odds-ratio = 6 . 94 ( 5 . 17–9 . 31 ) ; Figure 4 ) . These results suggest that FBti0019627 mediates resistance to benzaldehyde , which is consistent with increased CG11699 expression ( Figure 2 ) and increased ALDH-III activity ( Figure 3 ) observed in flies with this insertion . To further confirm that resistance to benzaldehyde is due to the insertion and not to any other background mutation present in the outbred-1 populations , we performed the acute exposure to benzaldehyde experiment using flies with three different genetic backgrounds: two DGRP inbred strains , two introgressed strains , and outbred-2 populations ( see Material and Methods ) . We found that females with the insertion showed a significant increase in survival rate compared to females without the insertion in two of the three new backgrounds analyzed: the DGRP strains ( Mann-Whitney p-value = 0 . 001 odds ratio = 8 . 04 ( 4 . 42–14 . 63 ) ) and the outbred-2 populations ( t-test p-value≪0 . 001 odds ratio = 3 . 29 ( 2 . 14–5 . 05 ) ; Figure 4A ) . Both introgressed females with and without the TE were highly resistant to benzaldehyde and did not show statistically significant differences ( Mann-Whitney , p-value>0 . 05 ) ( Figure 4A ) . Similar results were obtained for males; DGRP males with the insertion showed a higher survival rate compared to males without the insertion , although the difference was not statistically significant ( t-test , p-value = 0 . 12 ) . Introgressed males with the insertion were more resistant to benzaldehyde than introgressed males without the insertion ( t-test , p-value≪0 . 001 odds-ratio = 6 . 94 ( 3 . 08–7 . 06 ) ) . Finally , outbred-2 males with and without the insertion were both highly sensitive to benzaldehyde ( t-test , p-value>0 . 05 ) . While the differences in survival rate between flies with and without the insertion were not always significant , when they were , they were consistent with our expectations . These results strongly suggest FBti0019627 insertion mediates resistance to benzaldehyde and that mutations other than the FBti0019627 insertion also affect this phenotype . Aldehydes are present in decomposing fruits , a common food source for D . melanogaster in nature [34] . However , it is not clear whether flies in nature are exposed to such high concentrations of aldehydes as we used in our acute exposure experiments [35] . We searched for other ecologically relevant compounds for D . melanogaster natural populations that could also interact with ALDH-III . Insecticides and herbicides such as carbamates and thiocarbamates are known to inhibit ALDH2 in humans and rats by covalent modification of the nucleophilic active site residue [36] , [37] . ALDH enzymes share a wide range of common physiological functions and substrates and are predicted to have very similar catalytic site structures [37] , [38] . It is thus possible that carbofuran , a carbamate insecticide , could also react with the active site of D . melanogaster ALDH-III inhibiting this enzyme . We built a homology-based model of this protein and we performed preliminary docking studies with aldi1 , a known ALDH-III inhibitor , and with carbofuran . We found that the size and the shape of carbofuran molecule fits in the catalytic funnel of ALDH-III ( Figure 5a ) . The aromatic rings and the oxo groups of both compounds are located in the same regions ( Figure 5b ) and the distance between the electrophilic group of carbofuran and the nucleophilic active site residue of ALDH-III is similar to the distance found for the known inhibitor ( Figure 5B ) . Therefore , these preliminary docking results are compatible with carbofuran being a possible ALDH-III inhibitor . Although we cannot conclude that carbofuran is an ALDH-III inhibitor , increased ALDH-III activity could also lead to increased carbofuran resistance because carbofuran is an electrophilic molecule that causes lipid peroxidation through the generation of reactive oxygen species ( ROS ) [39]–[41] . As we have previously mentioned , ALDH3 is known to efficiently metabolize lipid peroxidation derived aldehydes [30] and could therefore play a protective role against carbofuran toxic effects . We hypothesized that flies with the insertion , which show increased ALDH-III activity , could have increased resistance to this carbamate insecticide . In order to evaluate this hypothesis , we compared the survival curves of flies with and without the insertion ( 3 , 200 flies in total ) that were exposed to concentrations of carbofuran similar to those used in the field ( http://www . epa . gov/oppsrrd1/REDs/carbofuran_red . pdf ) . We used flies with four different genetic backgrounds: outbred-1 populations , DGRP strains and introgressed strains previously used for the benzaldehyde experiments , and two new DGRP strains ( see Material and Methods ) . We found that both males and females flies with the insertion were more resistant to carbofuran than flies without the insertion ( Figure 6 ) ( Table 1 ) . Only introgressed males with and without the insertion did not show differences in survival rate ( Figure 6B ) ( Table 1 ) . The magnitude of the effect varied across backgrounds: the effect size was bigger for outbred-1 and RAL-391/783 compared to introgressed and RAL-810/857 ( Table 1 ) strongly suggesting that mutations other than FBti0019627 influence this phenotype . Finally , we also expect that CG11699 mutant flies , which have been previously shown to be highly sensitive to high doses of benzaldehyde [29] , should also be highly sensitive to carbofuran . These mutant flies showed reduced or null CG11699 expression levels [42] , and thus reduced ALDH-III activity [29] . As expected , our results showed that all CG11699 mutant flies died in the first 7 hours of treatment confirming the predicted high sensitivity of these flies to the insecticide ( Figure 6 ) . Our results obtained from four different genetic backgrounds showed that FBti0019627 insertion mediates resistance to carbofuran insecticide , which is consistent with increased CG11699 expression ( Figure 2 ) leading to increased ALDH-III activity ( Figure 3 ) . Similar to the results obtained with benzaldehyde , we also found differences in the magnitude of the effect between backgrounds; this is most likely explained by the contribution of other mutations to this phenotype . Additionally , results obtained with CG11699 lab mutants further confirmed the association between CG11699 expression levels , ALDH-III activity levels , and xenobiotic resistance . In Drosophila , there is a common oxidative stress response and a specific oxidative stress response that varies depending on the oxidative stress-inducing agent [43] . Both benzaldehyde and carbofuran are lipophilic electrophiles that induce the generation of reactive oxygen species ( ROS ) leading to lipid peroxidation [31] , [39]–[41] . To test whether FBti0019627 insertion confers resistance to other oxidative stress-inducing agents with different physicochemical properties than carbofuran and benzaldehyde , we used H2O2 to induce oxidative stress . While both carbofuran and benzaldehyde are lipophilic compounds , H2O2 is a small polar molecule that is not expected to directly interact with membranes [44] . We compared the survival curves of outbred-1 populations and DGRP strains with and without the insertion by analyzing 20 replicas of 20 flies each per sex and per strain , for unstressed and stressed conditions ( 3 , 200 flies in total ) . Female outbred-1 flies with the insertion were more sensitive than females without the insertion ( log-rank p-value = 0 . 001 , odds-ratio = 1 . 4 ( 1–1 . 8 ) ) while males with the insertion were more resistant ( log-rank p-value = 0 . 019 , odds-ratio = 1 . 5 ( 1 . 1–2 ) ( Figure 7 ) . In both cases , the lower confidence interval of the odds-ratio was 1 or close to 1 indicating that these results barely reach statistical significance ( see Material and Methods ) . On the other hand , DGRP strains with the insertion were more sensitive to H2O2 than strains without the insertion ( log-rank p-value≪0 . 001 , both for male and female flies ) ( Figure 7 ) . However , this result is explained by the presence in RAL-783 of a TE insertion named Bari-Jheh that confers resistance to oxidative stress [22] . Flies with FBti0019627 insertion were equally or more sensitive to H2O2 compared to flies without the insertion , suggesting that FBti0019627 does not play a role in resistance to H2O2 ( Figure 7 ) . If the functional interplay of CG11699 and ALDH-III plays a role in general response to oxidative stress , we would expect CG11699 mutant flies to be highly sensitive to oxidative stress induced by H2O2 . However , after 138 hours of treatment , 50% of the mutant males and 90% of the mutant females were alive ( Figure 7C ) . These results contrast with the high sensitivity of CG11699 mutant flies to carbofuran: all flies were dead after only 7 hours of stress exposure ( Figure 6C ) . Taken together , our results indicate that resistance to benzaldehyde and carbofuran in flies with the insertion is due to a specific oxidative stress response induced by lipophilic electrophiles and mediated by ALDH-III .
In this study , we showed that FBti0019627 insertion mediates resistance to xenobiotics by increasing CG11699 expression leading to increased ALDH-III activity ( Figure 2 and Figure 3 ) . Flies with FBti0019627 insertion show increased survival in response to benzaldehyde ( Figure 4 ) and to carbofuran ( Figure 6 ) compared to flies without the insertion . Benzaldehyde is an aromatic aldehyde found in fruits in decomposition , and carbofuran is a carbamate insecticide that has been widely used in nature [41] . Thus , both fatty and aromatic aldehydes and carbamate insecticides found in D . melanogaster habitats are likely agents of selection driving the previously reported increase in FBti0019627 frequencies [16] , [17] . Note that other ALDH-III substrates present in natural D . melanogaster habitats could also be acting as agents of selection of this mutation . We confirmed that xenobiotic resistance is due to FBti0019627 insertion and not to any other background mutation by performing experiments using flies with five different genetic backgrounds: two pairs of outbred populations , two pairs of DGRP inbred strains , and one pair of introgressed strains . Although outbred populations , inbred strains , and introgressed strains differ in their patterns of linkage disequilibrium , in the composition and site frequency distribution of alleles , and in the presence/absence of heterozygous individuals , we consistently observed that flies with the insertion showed increased resistance to xenobiotics compared to flies without the insertion ( Figure 4 and Figure 6 ) . Differences in survival rate between flies with and without the insertion were not always significant . However , when they were , they were consistent with our expectations , suggesting that FBti0019627 mediates resistance to xenobiotics . The lack of consistent patterns among backgrounds when a different selective agent was used , i . e . oxidative stress induced by H2O2 , further reinforces the role of FBti0019627 in xenobiotic resistance . Effect size of the mutation also varied across backgrounds indicating that genes other than the one affected by the TE insertion are also contributing to the xenobiotic resistance phenotype . These results contrast with previous findings in which the putatively causative mutations of several quantitative traits could not be replicated between strains [12] . While epistatic interactions do not appear to dominate the effect of FBti0019627 , they probably play an important role . Although there are a few examples of TE insertions mediating insecticide resistance in Drosophila [20] , [22] , [45]–[48] , previous evidence linking TEs and resistance to natural xenobiotics was only indirect , i . e . based on the observation that TEs are enriched within or close to resistance genes [49] , [50] . Therefore , our results provide the first experimental evidence for a role of TEs in both natural and synthetic xenobiotic resistance in eukaryotes . Given the widespread distribution of TEs across the tree of life , the conservation of stress response pathways across organisms , and the ubiquitous presence of natural and/or synthetic xenobiotics in the environment , it is likely that TEs are involved in resistance to xenobiotic stress in organisms other than D . melanogaster . Our results indicate that the insertion of FBti0019627 interferes with the choice of CG11699 polyadenylation signal ( PAS ) . As a result , in flies with the insertion , only the short 3′ UTR transcript is produced . As expected , the change in the length of the 3′ UTR leads to increased CG11699 expression levels . Shorter 3′ UTRs isoforms are less likely to possess microRNA binding site and/or other regulatory sequences such as AU-rich elements; consequently , they produce higher levels of transcripts and of protein [26] , [27] , [51] . Alternative polyadenylation , which leads to transcripts with 3′ UTRs of different lengths , is emerging as a major player in controlling gene regulation [27] . Deciphering the mechanisms behind the choice of alternative polyadenylation sites is considered to be one of the most interesting questions that remains to be answered . Our results provide evidence for TEs playing a role in this selection . FBti0019627 insertion also affects the transcript length of Kmn1 , which could lead to a change in the level of expression of this gene . Further experiments should help elucidate the effect of FBti0019627 on Kmn1 , which would provide a more complete picture of the effect of this insertion . Besides elucidating the molecular mechanism underlying the adaptive effect of FBti0019627 insertion , in this analysis we also shed light on its biochemical underpinnings . We showed that increased CG11699 expression is associated with increased ALDH-III activity as was first proposed by Arthaud et al ( 2011 ) [29] . Flies with FBti0019627 insertion are more resistant to benzaldehyde ( Figure 4 ) and carbofuran ( Figure 6 ) but not to H2O2 ( Figure 7 ) suggesting that resistance to benzaldehyde and carbofuran is due to a specific stress response induced by lipophilic electrophiles and mediated by ALDH-III . We also found that a lab mutant strain with null or low levels of CG11699 expression , which has been previously shown to be sensitive to benzaldehyde [29] , is also highly sensitive to carbofuran but not to H2O2 . This result reinforces the functional interplay between CG11699 expression , ALDH-III activity , and xenobiotic resistance . Insecticide resistance is an ongoing challenge for pest management and our results add ALDH-III to the list of previously reported enzymes that play a role in this resistance [49] . Mapping genotype to phenotype is currently one of the key challenges in biology [52] . Our approach to genotype-phenotype mapping in D . melanogaster combines a genome-wide screen for adaptive TE insertions , in which we gathered several lines of evidence suggesting their adaptive role , with hypothesis-driven mechanistic and functional analyses of the identified TEs [16] , [17] , [20] , [22] . In this study , we further showed that this approach is able to identify true biological signals of selection and to provide a causal link between genotype and phenotype . By integrating results from gene structure and gene expression analyses , we were able to identify the molecular effect of the insertion ( Figure 8A ) . We combined these results with the wealth of genetic and biochemical information available for Drosophila to construct mechanistic models and experimentally verify their predictions ( Figure 8B ) . Our results provide a plausible explanation for the increase in frequency of FBti0019627 insertion in out-of-Africa populations ( Figure 8C ) , and adds to the limited number of examples in which a natural TE insertion has been linked to its ecologically relevant phenotypic effect [21] , [22] , [53] . Besides the other candidate adaptive TE insertions already identified [16] , [17] , the increasing availability of next generation sequencing data and of computational pipelines to estimate the frequency of TEs in populations should lead to the identification of a larger set of candidate adaptive TEs in the near future [54]–[56] . The in-depth individual analysis of these TEs is very promising , and it should help us obtain a general picture of the adaptive process .
Total RNA was extracted form 40 mg of embryos , 50 L3 larvae , 40 5-day-old males , 40 5-day-old females , and ovaries from 35 5-day-old females using Trizol and a PureLink RNA Mini kit ( Ambion ) . RNA was treated on-column with DNase I ( Invitrogen ) . Reverse transcription was carried out using 3 µg of total RNA for embryos , females , and larvae and 1 . 5 µg of total RNA for males and ovaries . cDNA was constructed using the SuperScript II RT First Strand Synthesis system for RT-PCR ( Invitrogene ) . We amplified the cDNA 3′ends of CG11699 and Kmn1 genes using a Universal Amplification primer and two nested gene specific primers for CG11699 ( 5′-AGCCGCACCGATTTCGAGAGTCT-3′ and 5′-CTGGCAGCCTGGAACGAGGAATA-3′ ) and Kmn1 ( 5′-CATGATGGAGCTGCAGTGCAATA-3′ and 5′-CCAACGGTGACCCTAAGCTATGC-3′ ) . The 3′ RACE products were cloned using TOPO TA Cloning Kit for Sequencing ( Invitrogene ) following the manufacturer's instructions . When there were several 3′ RACE products , DNA from each individual band was extracted from the agarose gel before the cloning reaction . Several clones per 3′ RACE reaction were sequenced in both directions using M13 forward and reverse primers . Position-specific scoring matrices were derived from the empirical analysis of D . melanogaster Polyadenylation Signal ( PAS ) and GU-Rich Downstream Element ( DSE ) motifs published in [25] . The log-likelihood matrix was computed assuming all nucleotides were equiprobable . Sliding windows of 6 bp were run along the 50 bp region upstream of the cleavage site to search for the occurrence of PAS motifs . Sliding windows of 7 bp were run along 50 bp region downstream of the cleavage site to search for occurrence of DSE motifs . We expected the PAS signal to be located between the nucleotide positions −26 to −12 upstream of the cleavage site and the DSE to be located between the positions +1 and +25 downstream of the cleavage site . The highest scoring motifs located in these regions were considered as the most probable PAS and DSE motifs being used . Total RNA was extracted from three biological samples of 50 adult males and 50 adult females ( 4–6 days posteclosion ) using Trizol reagent and PureLink RNA Mini kit ( Ambion ) . RNA was then treated on-column with DNase I ( Thermo ) during purification , and then treated once more after purification . Reverse transcription was carried out using 500 ng and 300 ng of total RNA for females and males respectively using Anchored-oligo ( dT ) primer and Transcriptor First Strand cDNA Synthesis Kit ( Roche ) . The resulting cDNA was used for qRT-PCR with SYBR green master-mix ( BioRad ) on an iQ5 Thermal cycler . Total expression was measured using a pair of primers specific to a 118 bp cDNA amplicon spanning the exon2/exon3 junction of CG11699 present in both transcripts ( 5′-CTGGAAGCTATCCGGAGCCAA-3′ and 5′-CGTGAGACTCTCGAAATCGGTGCG-3′ ) . Long 3′UTR isoform expression was measured using a pair of primers specific to a 91 bp cDNA amplicon located in the 3′ most region of CG11699 3′UTR and therefore , only present in the long transcript ( 5′-ACCAGAACATAAAACGAAACCTTTG-3′ and 5′-TGACCGAAACAAATGAAAACCG-3′ ) . In both cases , expression was normalized using Act5C as an endogenous control gene ( 5′-GCGCCCTTACTCTTTCACCA-3′ and 5′-ATGTCACGGACGATTTCACG-3′ ) . We used serial dilutions of plasmid DNA to derive standard curves for each amplicon . Each curve was then used to determine the quantity of the corresponding transcript relative to the reference gene taking into account the reaction efficiency of each primer pair in order to avoid spurious results caused by differences in the efficiency of the different primer pairs . Reaction efficiencies ranged between 91 , 4% and 99 . 7% ( r2 larger than 0 . 99 ) . Three replicates of 30 4-to-6 day old outbred females with and without the insertion were transferred to 1 . 5 ml microcentrifuge tubes under light CO2 anesthesia . Flies were homogenized with 1 ml of cold buffer ( 0 . 22 M sucrose , 0 . 12 M mannitol , 1 mM EDTA and 10 mM tricine , pH 7 . 2 ) [57] using a 2 ml glass tissue grinder on ice . The homogenate was briefly centrifuged at 4°C to pellet down whole cells and other debris . The supernatant was transferred to a clean microcentrifuge tube and centrifuged for 30 minutes at 13 , 000 rpm at 4°C in order to obtain a pellet enriched in membranes . The pellet was re-suspended in 1 ml of membrane-disrupting homogenization buffer containing 1% Triton X-100 , incubated for 15 minutes on ice , and centrifuged again for 30 minutes at 13 , 000 rpm at 4°C ( adapted from [58] , [59] ) . The supernatant , containing the solubilized proteins that were bound to the membrane , was transferred to a clean microcentrifuge tube and was immediately used for protein quantification and enzymatic activity determination . ALDH-III ( EC 1 . 2 . 1 . 5 ) oxidizes benzaldehyde to benzoic acid using NAD ( P ) + as an acceptor and producing NAD ( P ) H in the process . We measured ALDH-III activity by monitoring the increased in absorbance at 340 nm produced by the formation of NAD ( P ) H ( Figure S2A ) . To control for differences in overall protein abundance between samples , we quantified the total protein content for each homogenate using Quick Start Bradford Protein Assay ( BioRad ) following the manufacturer's instructions . Each enzymatic activity determination was performed by mixing approximately 100 µg of membrane protein extract ( 100–200 µl ) with 1 ml of reaction buffer ( 50 mM sodium phosphate , 1 mM NAD+ , 1 mM NADP , pH 8 and benzaldehyde at 0 . 01 , 0 . 05 , 0 . 1 , 0 . 5 or 1 mM ) in 1 . 5 ml disposable cuvettes . Negative controls without substrate and negative controls without NAD ( P ) + were run to ensure that the formation of NAD ( P ) H was specific for the assay conditions we wanted to test . We measured the formation of NAD ( P ) H every minute at 340 nm for 15 minutes using a UV-1700 PharmaSpec spectrophotometer ( Schimadzu ) . The slope of the linear increase in absorbance over the measurement time for each condition ( R2>0 . 98 ) was used as initial reaction rate . Once we confirmed that ALDH activity showed a linear relationship with the total amount of protein used in the assay ( Figure S2B ) , we normalized the enzymatic activity measures by the total amount of protein in each sample ( activity was expressed as mOD·min−1·mg−1 ) . In order to estimate Vmax in our samples , we fitted the Michaelis-Menten equation to our experimental data by least-squares method using GraphPad Prism version 6 . 0e for Mac OS X ( GraphPad Software , La Jolla California USA ) . We performed a replicates test for lack of the fit and obtained a p-value greater than 0 . 05 , indicating that there was no evidence to reject the Michaelis-Menten model . We estimated Km and Vmax and their corresponding 95% confidence intervals and performed a statistical analysis of the comparison between the three replicates of flies with and without the insertion . The D . melanogaster protein sequence of ALDH-III isoform Q ( NP_724562 . 2 ) was used to search for a structurally resolved closely related protein . The structure of human ALDH3A1 co-crystallized with aldi1 , a covalent inhibitor of ALDH3A1 , was selected as a template ( PDB ID: 3SZB ) . The alignment of these two protein sequences was built using the pfam hidden markov model of ALDH family ( Aldedh ) and the package hmmer3/b 3 . 0 . The final alignment had a 51 . 54% identity along 424 amino acids comprising the catalytic domain , the NAD binding domain , and part of the bridging domain of ALDH-III . The ALDH-III model was built using the automodel class in Modeller 9 . 7 [60] . Energy minimization was carried out using VMD [61] and the extensions Automatic PSF builder and NAMDgui [62] . Default parameters were used except for the dielectric constant , which was set at 80 to simulate an implicit water environment . The stereochemical properties of the template and the model were evaluated using PROCHECK [63] and their pseudoenergetic profiles and z-score were calculated and compared using PROSA-II [64] . Superimposition of the model with human ALDH3A1 and retrieval of the corresponding structural alignment was performed using STAMP [65] . For the regions were the pseudo-energy in the model was higher than in the template , the PSI-PRED [66] predicted secondary structure of the model was compared with the description obtained using dssp [67] . The docking simulations were performed for a known ALDH-III inhibitor , aldi1 , and for carbofuran [68] . The protein structures and ligands were prepared for docking using the Autodock plugin for PyMol [69] . The energy-scoring grid was prepared as a 20 Å×20 Å×20 Å box centered around the catalytic cysteine of the ALDH-III model . The obtained ligands and receptor were used as the input for vina with default parameters [70] . The docking results were visualized and evaluated using PyMol . Redocking of aldi1 with human ALDH3A1 was used to verify that the docking parameters specified for this docking study were correct . The 10 highest scoring poses of 5 docking simulations were evaluated by comparing their localization in the catalytic pocket with the localization of aldi1 in a structural superimposition of ALDH-III model and ALDH3A1 . The criteria that were used to select the best poses were: ( i ) co-planarity of the aromatic rings with Tyr-114 and aldi1; ( ii ) orientation of the carbamate group towards Cys-243; and ( iii ) distance between the carbamate electrophilic carbon and nucleophilic thiol group of the active site cysteine . | Given the predictions of future environmental fluctuations , it is crucial to understand how organisms adapt to changing environments . The fruit fly Drosophila melanogaster is an ideal model organism to study environmental adaptation because of our deep understanding of developmental , physiological , and metabolic networks , as well as the ease of experimental manipulation . In this study , we showed that a previously identified putatively adaptive mutation , the insertion of the transposable element FBti0019627 , mediates resistance to both natural and synthetic xenobiotics in Drosophila melanogaster . By combining experimental and computational approaches , we further elucidated the molecular and the biochemical mechanisms underlying this natural adaptive mutation . Our results should be relevant for other organisms as well since there are many similarities between species in the way cells respond to stress . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"biology",
"and",
"life",
"sciences",
"population",
"biology",
"evolutionary",
"biology",
"molecular",
"biology"
] | 2014 | A Transposable Element Insertion Confers Xenobiotic Resistance in Drosophila |
Apart from sharing common ancestry with chordates , sea cucumbers exhibit a unique morphology and exceptional regenerative capacity . Here we present the complete genome sequence of an economically important sea cucumber , A . japonicus , generated using Illumina and PacBio platforms , to achieve an assembly of approximately 805 Mb ( contig N50 of 190 Kb and scaffold N50 of 486 Kb ) , with 30 , 350 protein-coding genes and high continuity . We used this resource to explore key genetic mechanisms behind the unique biological characters of sea cucumbers . Phylogenetic and comparative genomic analyses revealed the presence of marker genes associated with notochord and gill slits , suggesting that these chordate features were present in ancestral echinoderms . The unique shape and weak mineralization of the sea cucumber adult body were also preliminarily explained by the contraction of biomineralization genes . Genome , transcriptome , and proteome analyses of organ regrowth after induced evisceration provided insight into the molecular underpinnings of visceral regeneration , including a specific tandem-duplicated prostatic secretory protein of 94 amino acids ( PSP94 ) -like gene family and a significantly expanded fibrinogen-related protein ( FREP ) gene family . This high-quality genome resource will provide a useful framework for future research into biological processes and evolution in deuterostomes , including remarkable regenerative abilities that could have medical applications . Moreover , the multiomics data will be of prime value for commercial sea cucumber breeding programs .
Echinodermata , an ancient phylum of marine invertebrates , comprises 5 extant classes , including Echinoidea ( sea urchins ) , Asteroidea ( sea stars ) , Holothuroidea ( sea cucumbers ) , Ophiuroidea ( brittle stars ) , and Crinoidea ( sea lilies ) . Together , the phyla Echinodermata , Hemichordata , and Chordata form the deuterostome clade , based on their closely shared developmental features . To date , 2 complete echinoderm genomes , that of the sea urchin Strongylocentrotus purpuratus and that of the sea star Acanthaster planci , have been successfully sequenced [1 , 2] . However , because sea cucumbers are unique among echinoderms , possessing many distinctive biological characteristics , their genome holds invaluable insight that can extend the scope and depth of molecular research in Echinodermata and Deuterostomia . Sea cucumber adults exhibit an elongated shape that belies their pentaradial symmetry , combined with weak calcification in the form of microscopic ossicles that contrasts with the solid calcified test of sea urchins . Exploring these features can help in exploring the evolution of mineralization in echinoderms , which remains poorly understood . Of even greater interest is the fact that sea cucumbers display a capacity to regrow body parts and internal organs [3] , which is much greater than that of sea stars and sea urchins , making them prime regeneration models . The use of A . japonicus in this field is facilitated by its natural ability to discard its internal organs , rapidly regenerate them , and restore normal functions within a few weeks , through a process that involves cell migration , proliferation , differentiation , and organ/tissue reconstruction [3–5] . Finally , sea cucumbers , like many echinoderms , can be extremely long-lived and somewhat immune to senescence [6 , 7] . Therefore , knowledge of the complete genome of a sea cucumber will provide a unique framework for studies that seek to understand cell and tissue regeneration , treat organ failure , and alleviate the symptoms of aging . Sea cucumbers are also widespread , occurring from the shore to the abyss , and can represent up to 80% of the whole biomass of benthic invertebrates in some areas . They are the target of important fisheries and represent the fastest-growing aquaculture sector worldwide [8] . However , overfishing and poor management of these valuable resources are a growing concern [8 , 9] . The sea cucumber A . japonicus is one of the most studied echinoderms; it is being cultivated commercially on a large scale in the western North Pacific Ocean and is one of the most valuable sea foods worldwide , due to its potent nutritional and medicinal properties [10–12] . In China alone , around 200 , 000 tons of sea cucumbers were produced in 2015 , with an estimated value of about 4 , 000 , 000 , 000 United States dollars [13] . Improving genomic knowledge of this sea cucumber may therefore benefit the seafood industry and concurrently yield pharmaceutical and biomedical breakthroughs . In early 2017 , a draft genome of A . japonicus was published that represented only about 80 . 5% of the estimated genome size ( 0 . 82 Gb ) , with scaffolds N50 value of 10 . 5 Kb [14] . These data provided an important resource for sea cucumber genomics , but their incompleteness and fragmentation limit applications for research . Here we present a high-quality reference genome of A . japonicus investigated through a multiomics approach , providing valuable insights into the molecular and genomic basis of crucial evolutionary traits in sea cucumbers and deuterostomes . Knowledge of the complete genome of a holothuroid offers a particularly useful framework for studies that seek to understand the mechanisms of cell and tissue/organ regeneration .
We present a high-quality sea cucumber reference genome from the commercially cultivated species A . japonicus , generated from Illumina and PacBio platforms . An integrated multiomics approach offered insight into the genome architecture of sea cucumbers and the genetic underpinnings of their unique biological traits . The sea cucumber constitutes a particularly promising model animal for regenerative medicine because of the convenient induction of organ regeneration and the newly available genomic data for A . japonicus . A genome resource of this completeness and quality also makes an important contribution to holothuroid and echinoderm research . The findings should facilitate our understanding of the requirements for sustainable utilization and effective breeding of echinoderms , in support of the high-value sea cucumber industry .
The animal material for genome sequencing and assembly was from a male A . japonicus captured off the coast of Laoshan , Qingdao , China . The sea cucumber was acclimated in sea water at 15 ± 1°C before experiments . Muscle , gonad , and respiratory tree tissues were collected and immediately frozen in liquid nitrogen and stored at −80°C . Genomic DNA was extracted using a TIANamp Marine Animal DNA Kit ( TIANGEN , Beijing , China ) according to the manufacturer’s instructions . For Illumina sequencing , short-insert paired-end ( PE ) ( 180 bp and 500 bp ) and long mate-paired ( MP ) ( 5 Kb , 10 Kb , and 20 Kb ) DNA libraries were constructed according to the manufacturer’s instructions ( Illumina , San Diego , California , US ) . Sequencing runs for the PE libraries were performed on the Illumina HiSeq2000 platform , and long MP libraries on the HiSeq2500 platform . To obtain long reads to promote genome assembly , Pacific Biosciences RS II ( Pacific Biosciences , Menlo Park , California , US ) was used as the sequencing platform . Five 10-Kb SMRTbell libraries were prepared and sequenced using the C4 sequencing chemistry and P6 polymerase . In order to get the best assembly results , we tried Short Oligonucleotide Analysis Package de novo assembly tool ( SOAPdenovo ) [64] , de Bruijn graph to Overlap-Layout-Consensus ( DBG2OLC ) ( https://sourceforge . net/projects/dbg2olc/ ) , FALCON ( https://github . com/PacificBiosciences/FALCON ) , and SMARTdenovo ( https://github . com/ruanjue/smartdenovo ) . The assembly result of SOAPdenovo was fragmented ( 1 , 165 , 887 contigs , N50 1 , 770 bp , and N90 230 bp ) . Compared with the SOAPdenovo assembly using only Illumina data , DBG2OLC , FALCON , and SMARTdenovo genome assemblies using PacBio data or a hybrid of PacBio and Illumina data gave more satisfactory results , all with contig N50 values above 110 Kb . This result confirmed the effectiveness of the approaches and the superiority of PacBio long reads for large genome assembly ( S2 Table ) . After comparing the 3 approaches of DBG2OLC , FALCON , and SMARTdenovo , genome assembly by FALCON gave higher continuity ( with a contig N50 of 190 Kb ) than the other methods . To test the amount of PacBio data sufficient for A . japonicus genome assembly , we performed a set of assemblies by FALCON with coverage from 30× to 70× of PacBio data . For all assemblies , default parameters were used . Genome size and N50 of assembled contigs were calculated and used for performance evaluation ( S1 Fig ) . On the basis of these assembly experiments , 70-fold coverage of PacBio data ( 64 Gb ) was sufficient and used for the A . japonicus genome assembly . The assembled contigs were corrected 8 times with Illumina PE reads using Quiver ( https://github . com/PacificBiosciences/GenomicConsensus ) . Finally , we generated scaffolds and performed gap filling with SSPACE 3 . 0 ( https://www . baseclear . com/genomics/bioinformatics/basetools/SSPACE ) using Illumina MP sequencing data . The assembly result was evaluated by remapping high-quality PE reads ( 180-bp libraries , totaling 128 , 784 , 478 paired reads ) to scaffolds using Bowtie with parameters of--rdg 3 , 1--rfg 3 , 1--gbar 2 [65] . Assembly completeness was examined by mapping 81 , 639 unigenes from transcriptomes using BLAST ( https://blast . ncbi . nlm . nih . gov/Blast . cgi ) , and the physical coverage of each gene was calculated with SOLAR [66] . The assembled genome was also validated by checking the coverage of 248 conserved core eukaryotic genes using the CEGMA program ( version v2 . 5 ) [67] . To locate scaffolds on chromosomes , 2 parents and 130 offspring were genotyped , and a high-density linkage map was constructed with Genotyping by Sequencing ( GBS ) technology [68] . Sequencing depth averaged 50-fold for parents and 10-fold for the 130 offspring . Clean reads were mapped to genome scaffolds by the Burrows-Wheeler Aligner [69] . Variants including SNPs were identified using SAMtools [70] . JoinMap 4 . 0 was applied for linkage analysis based on a maximum likelihood algorithm [71] . Recombination values were converted to genetic distances in centiMorgans based on the Kosambi mapping function . A total of 71 . 2% of the genome scaffolds anchored in the linkage map , and Circos64 was used for visualization [72] . To estimate the level of heterozygosity in the genome , we carried out k-mer ( where k represents the chosen length of substrings ) distribution approximation using simulated heterozygous genome sequences ( S2 Fig ) . Jellyfish were used to calculate k-mer depth distribution [73] . PE reads from short-insert libraries ( 180 bp ) were used for analysis . Adaptor and low-quality reads were trimmed using the NGS QC Toolkit ( v2 . 3 . 3 ) [74] . Reads for 26 . 01 Gb of clean data were split into k-mers , and k-mer depth was computed to obtain depth distribution . Two peaks of k-mer depth 14 and 28 were observed in 17-mer analysis . We found that the k-mer distribution was fitted best by a simulated k-mer distribution with 2% heterozygosity ( S3 Fig ) . To obtain an exact heterozygous rate , we used the software gce to analyze k-mer frequency data with parameters of -c unique_depth -H 1 ( ftp://ftp . genomics . org . cn/pub/gce ) . We used gce to estimate the k-mer heterozygous ratio ( KHR ) based on the following formula: KHR = a1 / 2 / ( 2 − a1 / 2 ) . We used the formula base heterozygous ratio ( BHR ) = KHR / Kmer ( assuming each SNP caused K new k-mers ) for BHR . An exact heterozygous rate of 1 . 59% was estimated , which was close to the evaluation by simulated k-mer distribution . We also investigated the level of heterozygosity by mapping all reads back to the assembled genome using BWA with default parameters . For reads with multiple mapping positions , only the single best hit was retained . SNPs and INDELs were called based on alignment results with SAMtools [70] . A number of SNPs ( 5 , 379 , 554 ) and INDELs ( 486 , 341 ) were detected on the assembly genome , suggesting heterozygosity of about 0 . 73% . Because the part of the genome with the lowest heterozygosity had the best assembly , 0 . 73% was considered an underestimation due to sampling bias . RepeatModeler ( http://www . repeatmasker . org/RepeatModeler/ ) was used for repeat family identification in a de novo approach . RepeatMasker was used to identify transposable elements by aligning genome sequences against RepBase ( RepBase21 . 04 ) and a local library generated by RepeatModeler with default parameters . Tandem repeats were analyzed with the Tandem Repeats Finder ( TRF ) program [75] . Transposable elements ( TEs ) in the A . japonicus genome were discovered by a combination of de novo-based and homology-based approaches . A local repeat library of 1 , 507 sequences was produced by RepeatModeler and was used for the following prediction [76]: All major classes of TEs were summarized and compared with S . purpuratus ( S7 Table ) . A total of 210 . 87 Mb ( 26 . 20% ) of the assembled genome was predicted to be TEs , slightly less than S . purpuratus ( 32 . 67% ) . Of these , 18 . 93% were classified as unknown repeats . Among the 7 . 10% annotated repeat families , retrotransposable element Long Interspersed Nuclear Elements ( LINEs ) ( 2 . 31% ) and DNA transposons ( 3 . 20% ) were 2 major TEs discovered in the assembled genome that were similar to S . purpuratus . However , S . purpuratus has more DNA transposable elements ( 8 . 36% ) than A . japonicus ( S7 Table and S4 Fig ) . Noncoding RNA genes were predicted for repeat-masked genome by sequence- and structure-based alignments with the Rfam noncoding RNA database ( http://xfam . org/ ) with an E-value cutoff at 0 . 01 , using Infernal [77] . Specifically , for miRNA identification , small RNA sequencing data were downloaded from the SRA database of NCBI , including sRNAs from respiratory tree ( SRX878425 ) , tubefoot ( SRX642039 ) , and longitudinal muscle ( SRX878452 ) . Adaptor and primer sequences were trimmed , and low-quality sequences removed . Clean sRNA reads were compared with the Rfam database ( http://xfam . org/ ) to exclude noncoding RNAs other than miRNAs . The remaining sRNA reads were subjected to miRNA identification by mapping to predicted pre-miRNA structures in the A . japonicus genome using miRDeep2 [78] and miReap ( http://sourceforge . net/projects/mireap/ ) . Finally , novel and conserved miRNAs were classified by searching against the miRBase ( http://mirbase . org/ ) . Three approaches were used to predict protein-coding genes: homology-based predictions , de novo predictions , and transcriptome-based predictions . Homologous proteins from 8 known whole genome sequences Homo sapiens , Danio rerio , B . floridae , S . kowalevskii , S . purpuratus , Daphnia pulex , C . gigas , and Hydra vulgaris were used for alignment to the repeat-masked A . japonicus genome using Exonerate ( version 2 . 2 . 0 ) [79] . Genewise ( version 2 . 2 . 0 ) was used to generate gene structures based on homology alignments of proteins to the genome [80 , 81] . For ab initio gene prediction , we used Augustus ( version 2 . 5 . 5 ) , Genescan ( version 1 . 0 ) , GlimmerHMM ( version 3 . 0 . 1 ) , and SNAP15 to predict coding genes . RNA-Seq data were used to improve gene annotation and mapped to the genome using Tophat ( version 2 . 0 . 8 ) [82] . Cufflinks ( version 2 . 1 . 1 ) ( http://cole-trapnell-lab . github . io/cufflinks/ ) was used to identify spliced transcripts in gene models [83] . All gene evidence predicted from the 3 approaches was combined by EVM into a weighted and nonredundant consensus of gene structures [84] . Gene models generated by EVM were filtered according to the following criteria: coding region lengths less than 150 bp and values of reads per kilobase of exon model per million mapped reads ( fragments per kilobase of transcript per million mapped reads [FPKM] ) < 5 when a predicted gene supported by ab initio methods only hit with the uniref 90 database [85] . For gene functional prediction , NCBI nr and the SwissProt database were used for gene blasts . All predicted genes were blasted against the 2 databases using BLASTP ( E-value ≤ 1E-10 ) . To understand the evolutionary relationship of A . japonicus with other metazoans , we performed systematic gene comparisons . Full protein-coding genes of 17 genomes , including human ( H . sapiens ) , lizard ( Anolis carolinensis ) , turtle ( Chrysemys picta ) , frog ( Xenopus tropicalis ) , zebrafish ( D . rerio ) , amphioxus ( B . floridae ) , acorn worm ( S . kowalevskii ) , sea urchin ( S . purpuratus ) , sea star ( A . planci ) , sea cucumber ( A . japonicus ) , water flea ( D . pulex ) , leech ( Helobdella robusta ) , oyster ( C . gigas ) , octopus ( Octopus bimaculoides ) , sea anemone ( Nematostella vectensis ) , trichoplax ( Trichoplax adhaerens ) , and sponge ( Amphimedon queenslandica ) , were used for comparisons . For greater insight into the evolutionary dynamics of the genes , we determined the expansion and contraction of the gene ortholog clusters among these 17 species . We used CAFE software for computational analysis of gene family evolution [86] and defined expansion and contraction by comparing cluster size differences between ancestors and each current species . Extinction and evolution of gene families were processed by CAFE , using a random birth and death process model to identify gene gain and loss along each lineage of the RAxML tree ( Fig 2 ) . For all species , expanded and contracted gene families ( compared to ancestors ) were compared with A . japonicus to identify gene families that expanded or contracted only in the sea cucumber . For genes exclusively present and gene families specifically expanded in A . japonicus , we conducted gene ontology ( GO ) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes ( KEGG ) pathway enrichment analysis using blast2go and KAAS [87 , 88] . Using Omicshare CloudTools ( http://www . omicshare . com/tools/ ? l=en-us ) , enriched GO terms and KEGG Orthology ( KO ) terms were calculated relative to the background of full protein-coding genes . Among the 30 , 350 protein-coding genes of A . japonicus , many-to-many orthologs ( 15 . 38% ) , patchy orthologs ( 36 . 44% ) , and species-specific genes ( 43 . 17% ) were 3 major gene model groups ( Fig 2C ) . Echinoderm-specific orthologs ( 1 . 94% ) were also detected in the A . japonicus genome . GO enrichment analyses of the A . japonicus-specific genes and significantly expanded gene families suggested that most genes were associated with GO terms in cellular process , metabolic process , and binding and catalytic activity ( S6 and S8 Figs ) . Based on the full protein-coding genes , A . japonicus-specific genes were predominately enriched in GO terms related to signal recognition and immunity ( S12 Table and S8 Fig ) . Similar results were found for GO enrichment analysis of significantly expanded gene families relative to full genes ( S13 Table , S8 Fig ) . KEGG pathway enrichment analyses indicated that A . japonicus-specific genes and gained genes were also enriched in pathways related to signal recognition and immunity ( S14 and S15 Tables , S8 Fig ) . Peptide sequences were clustered by the Markov clustering program orthoMCL [89] . These sequences were also searched against the nr database by an all-versus-all BLASTP with threshold E ≤ 1E-05 and then clustered by MCL with an inflation value of 1 . 5 . A total of 43 single-copy orthologous genes were clustered among 17 genomes . Ortholog alignments were produced using MUSCLE ( v3 . 6 ) and concatenated into a single multiple-sequence alignment by an in-house Perl script . A neighbor-joining phylogeny was reconstructed using MEGA ( v5 ) [90] . To examine the phylogeny of sea cucumbers , we used a maximum likelihood ( ML ) method for genome-wide phylogenetic analysis based on single-copy genes from the 10 deuterostome and 7 nondeuterostome genomes . Based on the gene clustering results from orthoMCL , 49 , 351 gene families were collected from the 17 species . Among them , 43 single-copy genes were used for phylogenetic tree construction . To understand the relationship of A . japonicus to other echinoderms , we constructed another phylogenetic tree of 5 echinoderms based on orthologous genes . To extend the taxonomic sampling , gene families were surveyed in transcriptome datasets in F . serratissima , Patiria miniata , and A . filiformis ( NCBI SRA database accession numbers SRR2454338 , SRR573710 , SRR573709 , SRR573708 , SRR573706 , SRR573707 , SRR573705 , SRR573675 , SRR1523743 , SRR1533125 , SRR794587 , SRR794568 , SRR789489 , and SRR3097584 ) . Transcriptome data for the 3 echinoderms were assembled into unigenes using Trinity and cap3 with default parameters [91 , 92] . Single-copy gene families were extracted from the full genes of S . kowalevskii ( outgroup ) , A . japonicus , and S . purpuratus . For each gene family , protein sequences from all represented sequenced genomes were searched using BLASTP ( E-value cutoff 1 . 00E-10 ) against unigenes from each species . The unigene with the best score was translated as the longest open reading frame ( ORF ) in the frame detected by BLASTP . Gene clustering analysis was performed on the unigenes . Finally , 2 , 066 orthologous genes were obtained for constructing a phylogenetic tree using the ML method . For ML tree construction , sequence alignments were performed using MUSCLE 3 . 6 [93] . The substitution models that best fit observed alignment data were estimated using the program jModelTest 2 [94] . Using PhyML [95] , we performed ML analysis with the substitution model WAG + gamma + Inv , and 1 , 000 bootstraps were conducted to produce the branch support values . Molecular clocks and divergence time were estimated by combining r8s and RAxML programs [96 , 97] . Maximum likelihood phylogeny and branch lengths were obtained by RaxML with 1 , 000 bootstrap replicates . The birth-death model with fossil calibrations was used for Bayesian estimation of species divergence times [94 , 98–105] . Fossil-derived timescales and evolutionary history were obtained from TIMETREE [106] . The divergence time of TEs was calculated using RepeatMasker ( http://www . repeatmasker . org/ ) . In order to better understand the regeneration mechanism , we conducted a set of genomic and multiomic analyses of intestinal regeneration in A . japonicus . Adults of A . japonicus ( 100–120 g ) were collected from the coast of Qingdao , Shandong Province . They were held in a laboratory for 1 week in sea water at 15–17°C and fed once a day . Evisceration was induced by injecting about 2 mL 0 . 35 molL-1 KCl into the coelom . Eleven individuals per stage were sampled at 0 . 5 hours , 2 hours , 6 hours , 3 dpe , 5 dpe , 7 dpe , 14 dpe , and 21 dpe , and intestinal tissues were collected for RNA isolation; noneviscerated sea cucumbers served as controls . Dissected intestines were frozen in liquid nitrogen and stored at −80°C until RNA extraction . Total RNA was extracted using the TRIzol extraction method ( Thermo Fischer Scientific , Germany ) according to the manufacturer’s protocol . Poly-A mRNA was isolated from 40 μg total RNA per sample using oligo-dT-coupled beads and sheared . Isolated RNA samples were used for first-strand cDNA synthesis using random hexamers and Superscript II reverse transcriptase . After end repair and addition of a 3′-dA overhang , cDNA was ligated to an Illumina paired-end adapter oligo mix and size-selected by gel purification to enrich for approximately 200 bp fragments . After 16 PCR cycles , transcriptomes of the entire intestinal regeneration process of 9 stages were sequenced using Illumina HiSeq-2000 and the PE end sequencing module . These stages corresponded to 6 key phases of intestinal regeneration: early response ( 0 to 6 hours post evisceration ) , wound healing ( 6 hours to 3 dpe ) , blastema formation ( 3 to 7 dpe ) , lumen formation ( 7 to 14 dpe ) , intestinal differentiation ( 14 to 21 dpe ) , and growth ( 21 dpe ) [57 , 109] . RNA expression analysis was based on the predicted genes of A . japonicus genome . Tophat was used to map mRNA reads to the genome , and Cufflinks was used to calculate expected FPKM as expression values for each transcript . Corresponding proteomic studies ( 3 , 5 , 7 , 14 , and 21 dpe ) of intestinal regeneration were analyzed using isobaric tags for relative and absolute quantitation ( iTRAQ ) coupled with mass spectrometry ( MS ) . Total protein was taken from sample solutions , and protein was digested with Trypsin Gold ( Promega , Madison , Wisconsin , US ) at a protein:trypsin ratio of 30:1 at 37°C for 16 hours . After digestion , peptides were dried by vacuum centrifugation , reconstituted in 0 . 5 mol/L TEAB , and processed according to the manufacturer’s protocol for 8-plex iTRAQ reagent ( Applied Biosystems , Foster City , California , US ) . The LC-ESI-MS/MS analysis was based on Triple TOF 5600 ( AB SCIEX , Ontario , Canada ) . Raw data files were acquired from Orbitrap and converted into MGF files by Proteome Discoverer 1 . 2 ( PD 1 . 2 , Thermo Fischer Scientific ) . Protein identification used the Mascot search engine ( Matrix Science , London , United Kingdom ) . | Echinoderms , ubiquitous in the marine environment , are important from evolutionary , ecological , and socioeconomic perspectives . Together with chordates and hemichordates , they form the deuterostome clade , making them a crucial node in the study of chordate ancestry . Within echinoderms , class Holothuroidea is unique; its members ( the sea cucumbers ) display remarkable regenerative abilities and play key roles as sediment bioturbators and symbiotic hosts , and many are prized in the seafood and pharmaceutical industries . The sea cucumber genome therefore has the potential to significantly contribute to our understanding of important evolutionary and biological processes and help enhance aquaculture programs . Here we present a high-quality genome sequence for the economically important species Apostichopus japonicus . Through comparative analyses , we identified 763 echinoderm-specific gene families enriched in genes encoding membrane proteins , ion channels , and signal transduction proteins . Marker genes associated with the notochord and gill slits were also found , providing valuable insight into the origin of chordates . The reduced number and low expression levels of biomineralization genes reflect the skeletal degeneration seen in sea cucumbers . Importantly , 2 gene families appeared to be expanded in A . japonicus and may play crucial roles in its heightened regenerative potential . Together , findings from the sea cucumber genome provide important and novel insights into echinoderm and deuterostome biology . | [
"Abstract",
"Introduction",
"Results",
"and",
"discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"methods",
"and",
"resources",
"genomic",
"library",
"construction",
"animals",
"invertebrate",
"genomics",
"genome",
"analysis",
"dna",
"construction",
"molecular",
"biology",
"techniques",
"echinoderms",
"starfish",
"research",
"and",
"analysis",
"metho... | 2017 | The sea cucumber genome provides insights into morphological evolution and visceral regeneration |
During meiosis , chromosomes undergo DNA double-strand breaks ( DSBs ) , which can be repaired using a homologous chromosome to produce crossovers . Meiotic recombination frequency is variable along chromosomes and tends to concentrate in narrow hotspots . We mapped crossover hotspots located in the Arabidopsis thaliana RAC1 and RPP13 disease resistance genes , using varying haplotypic combinations . We observed a negative non-linear relationship between interhomolog divergence and crossover frequency within the hotspots , consistent with polymorphism locally suppressing crossover repair of DSBs . The fancm , recq4a recq4b , figl1 and msh2 mutants , or lines with increased HEI10 dosage , are known to show increased crossovers throughout the genome . Surprisingly , RAC1 crossovers were either unchanged or decreased in these genetic backgrounds , showing that chromosome location and local chromatin environment are important for regulation of crossover activity . We employed deep sequencing of crossovers to examine recombination topology within RAC1 , in wild type , fancm , recq4a recq4b and fancm recq4a recq4b backgrounds . The RAC1 recombination landscape was broadly conserved in the anti-crossover mutants and showed a negative relationship with interhomolog divergence . However , crossovers at the RAC1 5′-end were relatively suppressed in recq4a recq4b backgrounds , further indicating that local context may influence recombination outcomes . Our results demonstrate the importance of interhomolog divergence in shaping recombination within plant disease resistance genes and crossover hotspots .
Meiosis is a specialized cell division that is central to sexual reproduction in eukaryotes [1 , 2] . It is characterized by a single round of DNA replication , followed by two successive rounds of chromosome segregation , generating four haploid gametes from a single diploid mother cell [1 , 2] . During prophase I , homologous chromosomes also pair and undergo reciprocal genetic exchange , termed crossover [3] . Crossovers ensure accurate chromosome segregation , by creating a physical link between homologous chromosomes that , together with chromosome cohesion , promote balanced segregation during the first meiotic division [1 , 2] . Importantly , meiotic crossovers also create genetic diversity by recombining linked variation [1 , 2 , 4] . Meiotic recombination thus impacts upon genetic adaptation in sexual populations , by combining independently arising mutations more rapidly than in asexual species [4] . Meiotic recombination initiates via DNA double-strand breaks ( DSBs ) generated by SPO11 topoisomerase VI-related transesterases [5–7] . In Arabidopsis ~100–200 meiotic DSBs form per meiosis , estimated from immunostained RAD51 , DMC1 , RPA1 and ƔH2A . X foci that occur along paired chromosomes at leptotene stage [8–10] . In budding yeast , endonuclease and exonuclease activities ( Mre11-Rad50-Xrs2 , Sae2 and Exo1 ) act at DSB sites to generate 3′-overhanging single-strand DNA ( ssDNA ) [11–14] , between 100s and 1000s of nucleotides in length [15 , 16] . Resected ssDNA is bound first by RPA1 and then RAD51 and DMC1 proteins , which together promote interhomolog invasion and formation of a displacement loop ( D-loop ) [17 , 18] . Stabilization of the D-loop likely involves template-driven DNA synthesis from the invading 3′-end [3 , 19] . Strand invasion intermediates may then undergo second-end capture to form double Holliday junctions ( dHJs ) , which can be resolved as a crossover or non-crossover , or dissolved [20 , 21] . The conserved ZMM pathway acts to promote meiotic DSB repair via dHJs and crossovers [2 , 3 , 22] . In Arabidopsis ~10 DSBs per meiosis are repaired as crossovers [23–26] . The majority ( ~90% ) of these crossovers are dependent on the ZMM pathway in Arabidopsis [2] . This pathway includes ZIP4 , the SHOC1 XPF endonuclease and its interacting partner PTD , the MER3 DNA helicase , the HEI10 E3 ligase , the MSH4/MSH5 MutS-related heterodimer and the MLH1/MLH3 MutL-related heterodimer [2 , 22] . ZMM factors are thought to stabilise interhomolog joint molecules , including dHJs , and promote crossover resolution [27] . ZMM-dependent crossovers ( also known as Class I ) also show the phenomenon of interference , meaning that they are more widely distributed than expected at random [2 , 22 , 28 , 29] . In plants and other eukaryotes a large excess of initiating meiotic DSBs proceed to resection and strand invasion , but are repaired as non-crossovers ( that may be detectable as gene conversions ) , or via inter-sister repair [2] . Disassociation of strand invasion events occurs via partially redundant anti-crossover pathways in Arabidopsis that include , ( i ) the FANCM helicase and its cofactors MHF1 and MHF2 [30–32] , ( ii ) the BTR complex: RECQ4A , RECQ4B , TOPOISOMERASE3a and RECQ4-MEDIATED INSTABILITY1 ( RMI1 ) [33–37] , and ( iii ) FIDGETIN-LIKE1 ( FIGL1 ) and FLIP1 [38 , 39] . Plants mutated in these anti-crossover pathways show increased non-interfering crossovers , which are also known as Class II events [2] . This likely occurs as a consequence of reduced disassociation of interhomolog strand invasion events , which are alternatively repaired by non-interfering crossover pathway ( s ) [30 , 34 , 38] , including via MUS81 [40 , 41] . Hence , alternative repair pathways act on SPO11-dependent DSBs during meiosis to balance crossover and non-crossover outcomes . Due to the formation of interhomolog joint molecules during meiotic recombination , sequence polymorphisms between chromosomes can result in mismatched base pairs [42] . During the mitotic cell cycle DNA mismatches , or short insertion-deletions ( indels ) , caused by base mis-incorporation during replication , or exogenous DNA damage , can be detected by MutS-related heterodimers [43] . MutS recognition of mismatches and the subsequent promotion of repair plays a major anti-mutagenic role in vivo [43] . MutS complexes also play anti-crossover roles during meiosis when heterozygosity leads to sequence mis-matches , following interhomolog strand invasion [44–47] . Accumulating evidence also indicates that Class I and II crossover repair pathways show differential sensitivity to levels of interhomolog polymorphism . For example , Arabidopsis fancm mutations show increased crossovers in inbred , but not in hybrid contexts , whereas figl1 and recq4a recq4b mutations are effective at increasing crossovers in both situations [34 , 38 , 48–51] . This implies that the non-interfering crossover repair pathways acting in these backgrounds are influenced differently by interhomolog polymorphism . Genome-wide mapping of crossovers in anti-crossover mutants , or backgrounds with additional copies of the ZMM gene HEI10 , have further shown that the resulting recombination increases are highly distalized towards the sub-telomeres , correlating with regions of lowest interhomolog polymorphism [49–51] . At larger physical scales ( e . g . kb to Mb ) structural rearrangements , such as translocations and inversions , are potently associated with crossover suppression [52 , 53] , and increased levels of divergence within the Arabidopsis 14a hotspot correlated with reduced crossover frequency [54] . Despite the suppressive effects of interhomolog polymorphism on recombination , at the chromosome scale wild type crossovers in Arabidopsis show a weak positive relationship with interhomolog diversity , i . e . heterozygosity [49 , 50] . Linkage disequilibrium ( LD ) based historical crossover estimates are also positively correlated with population diversity in Arabidopsis [48 , 55 , 56] . Furthermore , juxtaposition of megabase scale heterozygous and homozygous regions in Arabidopsis associates with increased crossover frequency in the heterozygous regions , which is dependent on the Class I repair pathway [48] . Therefore , the relationship between interhomolog polymorphism and meiotic crossover frequency is complex , with both negative and positive relationships , depending on the scale and region analysed . In this work we explore the influence of interhomolog polymorphism on meiotic recombination at the scale of crossover hotspots in Arabidopsis thaliana . Specifically , we map crossovers across the RAC1 and RPP13 disease resistance genes , which encode proteins that recognise effector proteins from the oomycete pathogens Albugo laibachii and Hylaoperonospora parasitica , respectively [57 , 58] . We observe a non-linear negative relationship between interhomolog polymorphism and crossover frequency within both RAC1 and RPP13 , supporting a local inhibitory effect of mismatches on crossover formation . This relationship was observed using different RAC1 haplotypic combinations , which vary in the density and pattern of polymorphism . Despite recombination rates increasing genome-wide in anti-crossover mutants and HEI10 transgenic lines , RAC1 crossover frequency was stable or significantly decreased in these backgrounds . The resistance of RAC1 to genome-wide crossover increases may relate to the high level of interhomolog polymorphism at this locus , the pericentromeric location or local chromatin environment . Using deep sequencing of RAC1 crossover molecules we show that the negative relationship between crossovers and interhomolog divergence is maintained in the fancm , recq4a recq4b and fancm recq4a recq4b anti-crossover mutants . However , crossover frequency at the 5′ end of RAC1 was relatively decreased in recq4a recq4b mutant backgrounds , indicating an influence of local context on recombination outcomes .
We previously identified the RESISTANCE TO ALBUGO CANDIDA1 ( RAC1 ) Arabidopsis disease resistance gene region as containing crossover hotspots , using both historical linkage disequilibrium ( LD ) based estimates and experimental pollen-typing in Col×Ler F1 hybrids [59 , 60] . RAC1 encodes a TIR-NBS-LRR domain resistance protein , which recognises effectors from the oomycete pathogens Albugo candida and Hylaoperonospora parasitica [57 , 61 , 62] . RAC1 exists as a singleton TIR-NBS-LRR gene in most accessions and shows high levels of population genetic diversity ( e . g . θ = 0 . 012–0 . 013; π = 0 . 043–0 . 054 ) [55 , 56 , 59 , 60] . We compared the RAC1 locus to a recombination map of 3 , 320 crossovers mapped by genotyping-by-sequencing ( GBS ) of 437 Col×Ler F2 individuals ( mean crossover resolution = 970 bp ) ( Fig 1A ) [50 , 60] . We also assessed levels of interhomolog polymorphism by measuring the density of Col/Ler SNPs per 100 kb [63] , in addition to levels of DNA methylation as an indication of heterochromatin ( Fig 1A ) [64] . Together this showed that RAC1 is located on the edge of pericentromeric heterochromatin , in a region of higher than average crossover frequency and interhomolog polymorphism ( Fig 1A ) . Using historical recombination maps generated by analysing the 1 , 001 Genomes Project SNP data , we identified RPP13 as a further NBS-LRR gene with higher than average historical crossover frequency ( 10 . 56–10 . 57 cM/Mb ) , and high levels of population SNP diversity ( θ = 0 . 011–0 . 013 , π = 0 . 044–0 . 045 ) [55 , 56 , 59 , 60] . These levels of diversity and recombination were comparable to those observed at RAC1 . RPP13 recognizes the Hylaoperonospora parasitica effector ATR13 to mediate disease resistance , and which together display co-evolutionary dynamics [58 , 65] . Similar to RAC1 , RPP13 is a singleton NBS-LRR gene located on the edge of pericentromeric heterochromatin , in a region of higher than average crossover frequency and interhomolog polymorphism ( Fig 1A ) . We examined the RAC1 and RPP13 regions using genome-wide maps of chromatin and meiotic recombination [60 , 64 , 66] . Nucleosome occupancy was assessed using MNase-seq data , which showed enrichment within the gene exons and was depleted within the promoter , intron and terminator regions ( Fig 1B ) . Arabidopsis SPO11-1-oligonucleotides mark meiotic DSB sites and show an inverse pattern to nucleosome occupancy [66] . Consistently , at RAC1 and RPP13 we observed higher levels of SPO11-1-oligonucleotides in the nucleosome-depleted intergenic regions ( Fig 1B ) . H3K4me3 ChIP-seq showed enrichment at the 5′-end of the genes , consistent with active transcription [67] , and we observed RAC1 and RPP13 transcription using RNA-seq data from stage 9 flowers ( Fig 1B ) [60 , 66] . Both RAC1 and RPP13 show low levels of DNA methylation , in contrast to the ATENSPM3 EnSpm/CACTA ( AT1TE36570 ) element located adjacent to RAC1 , which is heavily methylated , nucleosome-dense and suppressed for SPO11-1-oligos ( Fig 1B ) . The RAC1 promoter intergenic region contains short fragments of several transposable elements , including HELITRONY3 and ATREP15 Helitrons ( Fig 1B ) . Transposable elements in these families have relatively low DNA methylation , are nucleosome-depleted and show higher levels of SPO11-1-oligos , compared with other repeat families in Arabidopsis ( Fig 1B ) [66] . Therefore , despite the location of RAC1 and RPP13 on the edge of pericentromeric heterochromatin , these genes display euchromatic chromatin and recombination features ( Fig 1A and 1B ) [64 , 66] . In order to experimentally measure crossovers within RAC1 and RPP13 we used pollen-typing [68 , 69] . This method uses allele-specific PCR amplification from F1 hybrid genomic DNA extracted from gametes , in order to quantify and sequence crossover molecules ( Fig 2A and S1 Fig ) [68 , 69] . This method is directly analogous to mammalian sperm-typing methods [70–73] . Genomic DNA is extracted from pollen ( male gametophytes ) collected from individuals that are heterozygous over a known crossover hotspot ( Fig 2A ) . Allele-specific primers annealing to polymorphic sites flanking the region of interest are used to PCR amplify single crossover or parental molecules , using diluted DNA samples ( Fig 2A ) . Titration is used to estimate the concentrations of amplifiable crossover and parental molecules , which are used to calculate genetic distance ( cM = ( crossovers/ ( crossovers+parentals ) ) ×100 ) ( Fig 2A ) . Sanger sequencing of PCR products amplified from single crossover molecules is performed to identify internal recombination points , to the resolution of individual polymorphisms ( Fig 2A ) . Together this information describes the recombination rate ( cM/Mb ) topology throughout the PCR amplified region [68 , 69] . It is also possible to mass amplify crossover molecules , which may be pooled and sequenced using paired-end reads to identify crossover locations ( Fig 2A ) [68] . To investigate whether RPP13 was associated with crossover hotspots we designed and optimised Col/Ler allele-specific oligonucleotides ( ASOs ) flanking this resistance gene ( S1A Fig ) . The RPP13 ASO primers specifically amplified crossovers from pollen , and not leaf DNA , extracted from the same Col/Ler F1 plants ( S1B Fig ) . We performed DNA titrations to quantify crossover and parental molecules across RPP13 and observed a genetic distance of 0 . 055 cM , equivalent to 9 . 78 cM/Mb across the 5 , 626 bp amplicon ( S1 Table ) . When analysing crossovers we plot their frequency against panmolecules , where we include all bases from both accessions ( S2 Fig and S2–S5 Tables ) . For example , the RPP13 amplicon is 5 , 431 bp in Col , 5 , 526 bp in Ler and 5 , 626 bp in the Col/Ler panmolecule , with 195 inserted bases from Ler and 100 from Col ( S2 Fig and S5 Table ) . We sequenced 44 single crossover molecules and observed clustering of recombination events at the 5′-end of RPP13 , overlapping regions encoding the coiled-coil and NB-ARC domains ( Fig 2C ) . RPP13 shows a peak crossover rate of 125 cM/Mb ( S6 Table ) , compared to the genome average of 4 . 82 cM/Mb for male Col/Ler F1 hybrids [74] . Crossovers were also observed in the adjacent gene At3g46540 ( Fig 2C ) . A single crossover was observed in a 5 bp interval within At3g46540 , which results in a high recombination estimate ( 250 cM/Mb ) ( S6 Table ) . However , as a single crossover event is responsible for this recombination measurement , this may reflect sampling , rather than the presence of a high activity hotspot . The region of highest crossover activity within RPP13 overlaps with nucleosome-occupied , H3K4me3-modified exon sequences ( Fig 2D ) . In contrast , highest SPO11-1-oligos occur in flanking nucleosome-depleted intergenic regions ( Fig 2D ) . The RAC1 gene is located within a 9 , 482 bp ( Col/Ler ) pollen-typing PCR amplicon ( Fig 3 ) . We previously reported analysis of 181 single crossover molecules within the RAC1 amplicon [60] , which we combined with an additional 59 events here to give a total of 240 crossovers ( S7 and S8 Tables ) . We observed a peak recombination rate of 61 . 7 cM/Mb within RAC1 ( Fig 3A and S8 Table ) . An adjacent gene contained within the amplicon , At1g31550 ( GDSL ) , also showed intragenic crossovers ( Fig 3A ) [60] . Similar to RPP13 , elevated crossover frequency within RAC1 overlapped nucleosome-occupied and H3K4me3-enriched exon sequences ( Fig 3B ) [66] . Highest crossover frequency occurred within the RAC1 5′ exons encoding the NB-ARC and TIR domains ( Fig 3A ) . A further similarity with RPP13 , is that highest levels of SPO11-1-oligos are observed in the nucleosome-depleted intergenic regions flanking RAC1 , in addition to the largest intron ( Figs 2 and 3 ) . Hence , both RPP13 and RAC1 have highest crossover frequency within transcribed gene 5′ regions , despite higher levels of initiating SPO11-1 dependent DSBs occurring in the adjacent intergenic regions . RAC1 and RPP13 show high levels of interhomolog polymorphism between Col and Ler , with 27 . 4 and 34 . 5 SNPs/kb , respectively ( compared to the genome average of 3 . 85 SNPs/kb ) [63] . This is also reflected in high levels of population genetic diversity at RAC1 and RPP13 [55 , 56 , 59 , 60] . Therefore , we repeated RAC1 pollen-typing with crosses using different parental accessions , where the pattern of interhomolog divergence varied , in order to investigate its influence on crossover frequency ( Fig 3A ) . Pollen-typing relies on allele-specific primers that anneal at SNPs or indels [68 , 75] . Therefore , we used the 1 , 001 Genomes Project data to identify accessions sharing the Col/Ler allele-specific primer polymorphisms , but differing with respect to internal polymorphisms within the RAC1 amplicon ( Fig 3A and S2 Fig ) . This identified Mh-0 ( Mühlen , Poland ) and Wl ( Wildbad , Germany ) as meeting these criteria . Col×Wl and Col×Mh have 33 . 0 and 21 . 1 SNPs/kb within the RAC1 amplicon , respectively . Therefore , we extracted pollen genomic DNA from Col×Wl and Col×Mh F1 hybrids and amplified and sequenced 92 and 124 crossover molecules , respectively ( Fig 3A and S9 and S10 Tables ) . For Col×Ler and Col×Mh we performed DNA titration experiments and did not observe a significant difference in crossover frequency ( P = 0 . 309 ) ( S7 Table ) . Crossover topology within the RAC1 amplicon was conserved between the three haplotype combinations tested ( Fig 3A and S8–S10 Tables ) . For instance , by comparing crossovers in adjacent 500 bp windows ( counted against the Col reference sequence ) we observed significant positive correlations between the recombination maps ( Spearman’s Col×Ler vs Col×Wl r = 0 . 595 P = 9 . 14×10−3; Col×Ler vs Col×Mh r = 0 . 612 P = 6 . 91×10−3; Col×Wl vs Col×Mh r = 0 . 723 P = 6 . 96×10−4 ) . For each cross , highest crossover frequency was observed within the RAC1 and GDSL transcribed regions ( Fig 3A and S8–S10 Tables ) . In each case , we calculated the number of crossovers and polymorphisms in adjacent 500 bp windows ( Fig 4 and S11 Table ) , where SNPs were counted as one and indels were counted according to their length in base pairs . In all cases , a significant negative relationship was observed between crossovers and polymorphisms ( all RAC1 windows , Spearman’s r = -0 . 685 P = 1 . 11×10−8 ) ( Fig 4 and S11 Table ) . This was also observed when analysing the RPP13 Col×Ler data in the same manner ( Spearman’s r = -0 . 890 , P = 2 . 43×10−4 ) ( Fig 4E and S12 Table ) . We fitted a non-linear model to the data using the formula y = log ( a ) +b×x^ ( -c ) , where y is the number of crossovers , x is polymorphisms , a is the intercept and b and c are scale parameters . Together this shows a strong , negative non-linear relationship between interhomolog polymorphisms and crossover frequency within RAC1 and RPP13 . We previously found that at the RAC1 5′-end there is a strong CTT motif , which have been associated with high crossover frequency in Arabidopsis [23 , 59 , 60 , 76] . Ler and Wl share a SNP in this motif but this does not obviously associate with differences in recombination rate [Col/Mh: CTTCGTCATCTTCTTCT; Ler/Wl: CTTCTTCATCTTCTTCT] . Previous work has revealed an influence of interhomolog polymorphism on meiotic recombination pathways in Arabidopsis [38 , 46 , 48–50] . Therefore , we sought to investigate RAC1 crossover frequency in backgrounds with altered recombination pathways . Specifically we tested , ( i ) mutations in the anti-crossover genes recq4a recq4b , fancm and figl1 [30 , 33 , 34 , 38 , 49] , ( ii ) mutations in the msh2 MutS homolog [46] , and ( iii ) transgenic lines with additional HEI10 copies [51] . Each of these genotypes was available in Col and Ler backgrounds , which could be crossed together to generate Col/Ler F1 hybrids used for RAC1 pollen-typing . We measured RAC1 crossover frequency via DNA titration experiments ( Fig 5 and S13–S16 Tables ) . The mean number of crossovers and parental molecules per μl were used to test for significant differences , by constructing 2×2 contingency tables and performing Chi-square tests . We compared three biological replicates of wild type Col/Ler F1 hybrids using this method , which did not show significant differences ( Fig 5A–5C and S13–S16 Tables ) . Previous findings have demonstrated genome-wide crossover increases in hybrid recq4a recq4b and figl1 mutants [34 , 49 , 50] , whereas fancm increases strongly in inbred , but not in hybrid backgrounds [38 , 48] . Despite the crossover increases in these backgrounds , we observed that RAC1 genetic distance significantly decreased in the recq4a recq4b , fancm , figl1 , msh2 , recq4a recq4b fancm and figl1 fancm mutants ( Fig 5A–5C and S13–S16 Tables ) . Furthermore , when we compared wild type with lines containing additional HEI10 copies we did not observe a significant difference in RAC1 crossover frequency ( Fig 5D and S13–S16 Tables ) . Therefore , in backgrounds with either increased Class I ( HEI10 ) or Class II ( fancm , figl1 , recq4a recq4b ) crossover repair , the RAC1 hotspot is unexpectedly resistant to increasing recombination frequency . To analyse RAC1 crossover distributions in wild type versus fancm , recq4a recq4b and recq4a recq4b fancm anti-crossover mutants , we mass amplified crossovers and performed pollen-sequencing [60 , 68] . In this approach , allele-specific PCR amplification is performed using multiple independent reactions seeded with an estimated ~1–3 crossover molecules per reaction ( Fig 2A ) . Crossover concentrations are first estimated using titration experiments ( Fig 5 and S13 Table ) . Mass amplified allele-specific PCR products are then pooled , sonicated and used for sequencing library construction ( S3 Fig ) [60 , 68] . These libraries were subjected to paired-end 2×75 bp read sequencing ( S17 Table ) . The Col and Ler RAC1 haplotypes from our laboratory strains were Sanger sequenced , in order to generate templates for aligning sequence data to . Read pairs were split and aligned to Col and Ler haplotypes separately using Bowtie allowing no mismatches ( -v 0 ) , such that BAM files were obtained for the reads aligned to either Col or Ler [77] . We then filtered for read pairs where one member mapped distally to Col and the other member mapped proximally to Ler , on opposite strands . This mapping configuration was expected due to the allele-specific primer orientation used during pollen-typing amplification . Consistent with these read pairs representing crossover molecules , their width distributions are similar to that of the sonicated PCR amplification products , prior to adapter ligation ( S3 Fig ) . The crossover reads were then matched to the Col/Ler panmolecule , and counts were added to intervening sequences . These values were then normalized by the total number of crossover read pairs per library . Finally , the profiles were weighted by RAC1 genetic distance ( cM ) , measured previously via DNA titration ( S13 Table ) . For wild type , fancm , recq4a recq4b and fancm recq4a recq4b we generated two biologically independent libraries for each genotype , sampling either ~300 or ~1 , 000 crossovers and the recombination profiles were found to be similar ( Figs 6A and S4 ) . Therefore , for subsequent analysis the reads from the 300 and 1 , 000 crossover libraries were pooled for each genotype . The wild type 1 , 000 crossover dataset was previously reported [60] . Overall recombination topology was similar between wild type , fancm , recq4a recq4b , and recq4a recq4b fancm mutants ( Spearman’s wild type vs fancm r = 0 . 923 <2 . 2×10−16; wild type vs recq4a recq4b r = 0 . 902 <2 . 2×10−16; recq4a recq4b fancm r = 0 . 925 <2 . 2×10−16 ) ( Fig 6A , S4 and S5 Figs and S14 Table ) . Crossovers occurred predominantly within the gene transcribed regions and were reduced in the highly polymorphic intergenic regions , in all genotypes ( Fig 6A and S4 and S5 Figs ) . In wild type , highest crossover frequency was observed at the RAC1 5'-end , with distinct peaks associated with the first and second exons , in addition to elevated crossovers occurring within the last three LRR domain-encoding exons ( Fig 6A and S4 Fig ) . Crossovers were also evident at the 5′ and 3′ ends of the adjacent gene ( GDSL ) , although at a lower level to those observed in RAC1 ( Fig 6A and S4 Fig ) . In fancm the crossover profile was similar , except for in the first RAC1 exon where crossover frequency was reduced compared to wild type ( Fig 6A ) . In the recq4a recq4b and fancm recq4a recq4b mutants we observed that the RAC1 5′ crossover peaks in exons 1 and 2 were relatively reduced ( Fig 6A ) . The RAC1 LRR crossover peaks in recq4a recq4b and fancm recq4a recq4b backgrounds were also less broad and became focused towards the end of exon 5 ( Fig 6A ) . The 5′-end of GDSL was reduced in the recq4a recq4b and fancm recq4a recq4b mutants ( Fig 6A ) . The local changes to crossover frequency in these recombination mutant backgrounds may reflect differential interactions with interhomolog polymorphism or chromatin within the analysed region . To investigate the relationship between crossovers and polymorphisms , we calculated recombination ( crossover read pairs/cM ) and polymorphism values in adjacent 250 bp windows , against the RAC1 Col/Ler panmolecule . Consistent with our previous observations , all genotypes showed a significant negative correlation between crossovers and polymorphisms ( Spearman’s: WT r = -0 . 64 , fancm r = -0 . 58 , recq4a recq4b r = -0 . 57 , fancm recq4a reqc4b r = -0 . 57 ) ( S18 Table ) . As described above , a non-linear model fitted the data using the formula y = log ( a ) +b×x^ ( -c ) , where y is the crossovers , x is polymorphisms , a is the intercept and b and c are scale parameters ( Fig 6B ) . Hence , the suppressive effect of polymorphisms was observed within RAC1 in both wild type and anti-crossover mutants .
The concentration of meiotic DSBs and crossovers in narrow hotspots is widespread among eukaryotes , which has important implications for genetic diversity and adaptation [78–80] . For example , sequencing of SPO11-oligonucleotides has revealed meiotic DSB hotspots in fungi , animals and plants [66 , 81–83] . Varying genetic and epigenetic factors control DSB hotspot location in these species . SPO11-oligo hotspots in budding yeast and plants are highest in nucleosome-free regions associated with genes and transposons [66 , 81 , 84 , 85] , which demonstrates the importance of chromatin for initiation of meiotic recombination . Variation in nucleosome occupancy and SPO11-1-oligos in plants correlates with AT-sequence richness [66] , which is known to exclude nucleosomes [86] . In fission yeast SPO11-oligo hotspots are broader , located intergenically and do not show a clear association with nucleosome occupancy [83] . Mammalian meiotic DSB hotspots are directed to specific DNA sequences by binding of the PRDM9 KRAB-zinc finger protein [80 , 82 , 87] . PRDM9 possesses a histone methyltransferase SET domain which directs H3K4me3 and H3K36me3 histone modifications to nucleosomes flanking the bound DNA target sites [82 , 88–90] . Hence , depending on the species , chromatin and DNA sequence make varying contributions to the fine-scale distributions of meiotic DSBs . In many species , including budding yeast and plants , there is a positive correlation between meiotic DSB levels and crossover frequency at the chromosome scale [66 , 81] . However , extensive variation in the ratio of DSBs to crossovers is observed along chromosomes [66 , 81–83] . Equally , at the fine-scale there is a weak correlation between crossovers and DSB frequency within Arabidopsis hotspot regions [66] , as observed at RAC1 and RPP13 . An extreme example of similar variation occurs in fission yeast , where an inverse relationship is observed , with DSB hotspots occurring in regions of lowest crossover formation [83 , 91] . Variation in crossover:non-crossover ratios has also been observed between mammalian hotspots [71 , 72 , 78 , 92] . For example , crossover:non-crossover variation occurs at heterochiasmic mouse hotspots , where DSBs occur in both male and female meiosis , but crossovers only form in male meiosis [93] . Furthermore , data in budding yeast indicate that interhomolog joint molecules may be mobile [94] , and repeated rounds of strand invasion and dissolution may occur during repair [95 , 96] , which could cause local differences in the locations of the initiating DSB and final crossover resolution . Hence , the levels of initiating DBSs are important for crossover levels , but they are not the sole determinant of recombination outcomes . In plants , somatic homologous recombination has been analysed using ‘split GUS’ reporter systems [97] . Recombination between repeated GUS sequences located on the same or different reporter T-DNAs restores β-glucuronidase activity [97] . Increasing levels of polymorphism in the recombining GUS repeats was found to inhibit homologous recombination [98 , 99] . For example , 1 . 9% sequence divergence between the GUS repeats caused a 10-fold reduction in somatic recombination [98] . In a related study , a single mismatch in a 618 bp GUS repeat caused a 3-fold suppression of recombination , although addition of further SNPs had less effect , suggesting ‘divergence saturation’ in this system [99] . These data are consistent with genetic analysis in budding yeast where mitotic and meiotic recombination are inhibited by polymorphism , with similar kinetics [47 , 100] . For example , progressive addition of SNPs at the URA3 hotspot reduced meiotic crossovers , with a simultaneous increase in non-crossover repair [101] . Consistent with these previous studies , at RAC1 and RPP13 we observe a non-linear , negative relationship between interhomolog polymorphism and meiotic crossover formation . A likely mechanism for the suppressive effects of polymorphism on crossover repair during meiosis is via MutS related heterodimers , including MSH2 , which are capable of recognising mismatches and promoting disassociation of strand invasion events [44 , 45 , 102] . Indeed , evidence exists in Arabidopsis for MSH2 acting as a hybrid-specific anti-crossover factor at the megabase scale [46] . However , this relationship appears complex , as we observe a significantly decreased RAC1 crossover frequency in the msh2 mutant . Our observations may suggest regional changes in crossover distributions in msh2 , rather than a global increase . The inhibitory effect of interhomolog polymorphism on crossover formation may also account for discrepancies observed between SPO11-1-oligos and crossovers at the fine-scale [66] . It is possible that mismatches in interhomolog joint molecules could alter their mobility and further influence the location of crossover resolution . The phenomenon of crossover interference , which reduces the likelihood of adjacent DSBs being repaired as crossovers in the same meiosis , is also important to consider [29] . In addition to interhomolog polymorphism , chromatin marks may differentially influence meiotic recombination pathways and locally alter crossover:noncrossover ratios . For example , we observe that H3K4me3 is elevated at the 5′-ends of RAC1 , GDSL and RPP13 , which correlates with regions of high crossover activity . Although it is also notable that substantial 3′-crossovers occur in these genes , where H3K4me3 occurs at lower levels . Although H3K4me3 levels do not strongly correlate with SPO11-oligo levels in animals , fungi or plants [66 , 81 , 82 , 103] , this mark is spatially associated with recombination hotspots in multiple species [23 , 59 , 76 , 87 , 104] . In budding yeast the Spp1 subunit of the COMPASS methylase complex simultaneously interacts with H3K4me3 and the Mer2 meiotic chromosome axis component [105 , 106] , providing direct support for the tethered-loop/axis model for recombination [107] . Analogous interactions are observed between mouse COMPASS CXXC1 , PRDM9 and the IHO1 axis protein [108] . Hence , the presence of H3K4me3 at the 5′-ends of RPP13 and RAC1 may promote crossover formation via similar mechanisms , downstream of DSB formation . Heterochromatic modifications also show specific interactions with the meiotic recombination pathways . For example , in Arabidopsis loss of CG context DNA methylation via the met1 mutation , or loss of non-CG DNA methylation/H3K9me2 via cmt3 or kyp/suvh4 suvh5 suvh6 , both cause an increase in SPO11-1-oligos in pericentromeric heterochromatin [66 , 109] . However , the CG and non-CG mutants show increased and decreased pericentromeric crossovers , respectively [66 , 109] . This indicates that despite these heterochromatic mutants showing greater SPO11-1 DSB activity close to the centromeres , other chromatin features likely modify downstream repair choices . In this study we measured RAC1 crossover frequency in backgrounds with , ( i ) elevated HEI10 dosage and thereby increased Class I activity [51] , ( ii ) increased Class II crossovers via loss of function fancm , figl1 and recq4a recq4b anti-crossover mutations [30 , 33 , 34 , 38 , 49] , and ( iii ) loss of function mutants in the mismatch repair factor msh2 [46] . Despite these backgrounds showing elevations in crossover frequency elsewhere in the genome , RAC1 remained resistant to recombination increases or showed small but significant decreases . In this respect it is notable that genome-wide maps of crossovers in HEI10 , figl1 , fancm and recq4a recq4b backgrounds have shown that recombination increases are highly distalized towards the sub-telomeres , which are chromosome regions of lower interhomolog polymorphism [49–51] . Therefore , the location of RAC1 on the edge of the chromosome 1 pericentromere may make this locus relatively insensitive to distalized crossover increases . It is also possible that high polymorphism levels within RAC1 , in addition to the surrounding regions of heterochromatin , may contribute to maintenance of stable crossover frequency between wild type and the high recombination backgrounds tested . The local inhibitory relationship between polymorphism and crossovers that we observe has implications for the evolution of plant hotspots . Data from several species are consistent with meiotic recombination being mutagenic [110–112] . For example , this may occur as a result of DNA polymerase base misincorporation during DSB-repair associated DNA synthesis [110–112] , or mis-alignment during strand invasion causing insertions and deletions via unequal crossover [113] . Therefore , high levels of recombination over many generations may cause higher levels of heterozygosity at hotspots , which may then suppress further recombination in specific crosses . Crossover inhibition is likely to be particularly potent when unequal crossover generates large insertion-deletion polymorphisms , which are commonly observed at plant disease resistance loci and can contribute to functional diversity in pathogen recognition [60 , 113–115] . Despite the negative relationship that we observe between interhomolog polymorphism and crossovers at RAC1 and RPP13 , at the chromosome scale wild type crossovers in Arabidopsis show a weak positive relationship with interhomolog diversity [49 , 50] . Similarly , LD-based historical estimates are positively correlated with population diversity in Arabidopsis [48 , 55 , 56] . These population-scale relationships are likely to be partly explained by hitchhiking/background selection , causing more extensive reductions in diversity in regions of low recombination that are under selection [4] . However , other effects may also contribute . For example , in Arabidopsis juxtaposition of megabase scale heterozygous and homozygous regions increases crossover frequency in the heterozygous region , at the expense of the homozygous region [48] . This heterozygosity juxtaposition effect is dependent on the Class I interfering repair pathway [48] . Therefore , both positive and negative interactions are observed between polymorphism and recombination depending on whether hotspot versus chromosome scales are analysed , with significant additional effects caused by chromosome position and chromatin context .
Arabidopsis lines used in this study were the HEI10 line ‘C2’ [51] , recq4a-4 ( Col , N419423 ) recq4b-2 ( Col , N511130 ) [36] , recq4a ( Ler W387* ) [34] , fancm-1 ( Col , ‘roco1’ ) [30] , fancm ( Ler , ml20 ) , figl1-1 ( Col , ‘roco5’ ) [38] , figl1 ( Ler , ml80 ) and msh2-1 ( Col , SALK_002708 ) [116] . Genotyping of Col recq4a-4 , Col recq4b-2 , Ler recq4a and HEI10 T-DNA was performed as described previously [50] . Col and Ler wild type or mutant backgrounds were crossed to obtain F1 hybrids , on which pollen-typing was performed . The msh2-1 allele was introduced into Ler-0 background by six successive backcrosses . Genotyping of msh2-1 was performed by PCR amplification using msh2-F ( 5'-AGCGCAATTTGGGCATGTCT-3' ) and msh2-R ( 5'-CCTCCCATGTTAGGCCCTGTT-3' ) oligonucleotides for the wild type allele , and msh2-F and msh2-T-DNA ( 5'-ATTTTGCCGATTTCGGAAC-3' ) oligonucleotides for the msh2-1 allele . Pollen-typing was performed as described [68] . Genomic DNA was extracted from hybrid F1 pollen ( Col×Ler , Col×Wl or Col×Mh ) , and used for nested PCR amplifications using parental or crossover configurations of allele-specific oligonucleotide ( ASO ) primers ( S19 and S20 Tables ) . For each genotype replicate , ~140 plants were grown and used for pollen collection . The relative concentrations of parental ( non-recombinant ) and crossover ( recombinant ) molecules were estimated by titration [68–70] . Recombination rate was calculated using the formula cM = ( crossovers/ ( parentals+crossovers ) ) ×100 . Amplified single crossover molecules were treated with exonuclease I ( NEB , M0293 ) and shrimp alkaline phosphatase ( Amersham , E70092 ) , and then Sanger sequenced to identify recombination sites to the resolution of individual polymorphisms . For analysis we PCR amplified and sequenced the target regions from Col , Ler , Wl and Mh accessions , and used these data to generate Col×Ler , Col×Wl or Col×Mh panmolecules , which include all bases from both accessions ( S2 Fig ) . To analyse the relationship between crossovers and polymorphisms we used adjacent 500 bp windows along the panmolecules and assigned crossover and polymorphism counts , where SNPs were counted as 1 , and indels as their length in base pairs . When crossover events were detected in SNP intervals that overlapped window divisions the crossover number was divided by the proportional distance in each window . For example , if two crossovers were detected in a 150 bp interval , of which 50 bp were in window A and 100 bp in window B , we counted 2× ( 50/150 ) = 0 . 67 crossover in window A , and 2× ( 100/150 ) = 1 . 33 crossover in window B . A non-linear model was fitted to the data using the formula; y = log ( a ) +b×x^ ( -c ) , where y is the number of crossovers , x is polymorphisms , a is the intercept and b and c are scale parameters . Multiple independent RAC1 crossover PCR amplifications were performed , where each reaction was estimated to contain between 1–3 crossover molecules , based on previous titration experiments . RAC1 crossover amplification products were then pooled , concentrated by isopropanol precipitation and gel purified . 1–2 μg of DNA in 100 μl of TE was sonicated for each sample using a Bioruptor ( Diagenode ) ( high setting , 30 seconds ON , 30 seconds OFF for 15 minutes ) , and fragments of 300–400 bp were gel purified , end-repaired and used to generate sequencing libraries ( Tru-Seq , Illumina ) . The libraries were sequenced on an NextSeq instrument ( Illumina ) using paired-end 75 bp reads . Reads were aligned to the parental sequences ( Col and Ler ) using Bowtie , allowing only exact matches [77] . Reads were filtered for those that aligned to one parental sequence only . To identify crossover read pairs , we filtered for read pairs having a centromere-proximal match to Col and a centromere-distal match to Ler , on opposite strands , which is consistent with RAC1 pollen-typing amplification . Read pair coordinates were then converted into pancoordinates using the Col/Ler key table ( S2 Table ) . A value of 1 was assigned to all panmolecule coordinates between each crossover read pair . This process is repeated for all read pairs and values normalized by the total number of crossover read pairs , and finally weighted by genetic distance ( cM ) . The fastq files associated with RAC1 crossover sequencing have been uploaded to ArrayExpress accession E-MTAB-6333 “Meiotic crossover landscape within the RAC1 disease resistance gene” . | Sexually reproducing plants and animals produce gametes with half the number of chromosomes , which can participate in fertilization . A specialized cell division called meiosis generates gametes , where the chromosomes are copied once and segregated twice . A further key feature of meiosis is that chromosomes physically pair and undergo reciprocal exchanges , called crossovers . Due to independent chromosome segregation and crossovers , meiosis creates gametes that are genetically diverse , which has a major effect on patterns of sequence variation in populations . Interestingly , the frequency of crossover is also highly variable along the lengths of chromosomes and tends to be concentrated in narrow hotspots . Here we studied two crossover hotspots in detail that are located within disease resistance genes , using the model plant Arabidopsis . We show that within these hotspots , greater levels of genetic difference between the recombining chromosomes locally inhibits crossover formation . We also show that hotspots within one of these resistance genes are surprisingly resistant to genetic backgrounds that increase crossovers elsewhere in the genome . This indicates that patterns of polymorphism and hotspot location along the chromosome are both important for control of recombination activity . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"methods"
] | [
"meiosis",
"plant",
"anatomy",
"cell",
"cycle",
"and",
"cell",
"division",
"population",
"genetics",
"brassica",
"cell",
"processes",
"pollen",
"plant",
"science",
"model",
"organisms",
"experimental",
"organism",
"systems",
"dna",
"molecular",
"biology",
"techniques"... | 2018 | Interhomolog polymorphism shapes meiotic crossover within the Arabidopsis RAC1 and RPP13 disease resistance genes |
Although evolutionary transitions from sexual to asexual reproduction are frequent in eukaryotes , the genetic bases of these shifts remain largely elusive . Here , we used classic quantitative trait analysis , combined with genomic and transcriptomic information to dissect the genetic basis of asexual , parthenogenetic reproduction in the brown alga Ectocarpus . We found that parthenogenesis is controlled by the sex locus , together with two additional autosomal loci , highlighting the key role of the sex chromosome as a major regulator of asexual reproduction . We identify several negative effects of parthenogenesis on male fitness , and different fitness effects of parthenogenetic capacity depending on the life cycle generation . Although allele frequencies in natural populations are currently unknown , we discuss the possibility that parthenogenesis may be under both sex-specific selection and generation/ploidally-antagonistic selection , and/or that the action of fluctuating selection on this trait may contribute to the maintenance of polymorphisms in populations . Importantly , our data provide the first empirical illustration , to our knowledge , of a trade-off between the haploid and diploid stages of the life cycle , where distinct parthenogenesis alleles have opposing effects on sexual and asexual reproduction and may help maintain genetic variation . These types of fitness trade-offs have profound evolutionary implications in natural populations and may structure life history evolution in organisms with haploid-diploid life cycles .
Although sexual reproduction , involving meiosis and gamete fusion , is almost ubiquitous across eukaryotes , transitions to asexual reproduction have arisen remarkably frequently [1] . These asexual reproductive mechanisms include parthenogenesis , which involves the development of an embryo from an unfertilized gamete [1] . The genetic basis of parthenogenesis remains largely elusive in both plants and animals , although the factors triggering the transition to asexual reproduction have been most intensively studied in plants , motivated by the potential use of asexual multiplication in the production of agricultural crops ( e . g . [2 , 3] ) . In plants , parthenogenesis is a component of apomixis , which is the asexual formation of seeds , resulting in progeny that are genetically identical to the mother plant . In gametophytic apomixis , the embryo sac develops either from a megaspore mother cell without a reduction in ploidy ( diplospory ) or from a nearby nucellar cell ( apospory ) in a process termed apomeiosis . Apomeiosis is then followed by parthenogenesis , which leads to the development of the diploid egg cell into an embryo , in the absence of fertilization ( reviewed in [4] ) . In some apomictic plants , inheritance of parthenogenesis is strictly linked to an apomeiosis locus ( reviewed in [5] ) , whereas in other species the parthenogenesis locus segregates independently of apomeiosis [6–8] . For example , apomixis in Hieracium is controlled by two loci termed LOSS OF APOMEIOSIS ( LOA ) and LOSS OF PARTHENOGENESIS ( LOP ) , involved in apomeiosis and parthenogenesis , respectively [9] . A third locus ( AutE ) involved in autonomous endosperm formation , was shown to be tightly linked to the LOP locus [10] . In Pennisetum squamulatum , apomixis segregates as a single dominant locus , the apospory-specific genomic region ( ASGR ) , and recent work has highlighted a role for PsASGR-BABY BOOM-like , a member of the BBM-like subgroup of APETALA 2 transcription factors residing in the ASGR , in controlling parthenogenesis [11] . There has been very little interest in parthenogenesis in organisms outside the animal and plant lineages , despite the fact that this mode of reproduction is widespread across the eukaryotic tree [12–17] . These alternative systems could provide novel insights into the genetic basis of parthenogenesis . For example , parthenogenesis is a common phenomenon in the brown algae , a group of multicellular eukaryotes that has been evolving independently from animals and plants for more than a billion years [18] , and can be an integral component of a species life cycle ( [19–21] ) . Once released into the surrounding seawater , gametes of brown algae may fuse with a gamete of the opposite sex , to produce a zygote which will develop into a diploid heterozygous sporophyte . Alternatively , in some brown algae , gametes that do not find a partner will develop parthenogenically , as haploid ( partheno- ) sporophytes ( e . g . [22] ) . Parthenogenesis in brown algae can therefore be equated with gametophytic embryogenesis in plants , where embryos are produced from gametes [23] , but in the case of brown algae the parthenogenetic gamete is haploid . The brown algae are therefore excellent models to study the molecular basis of parthenogenesis because gametes are produced directly by mitosis from the multicellular haploid gametophyte , allowing parthenogenesis to be disentangled from apomeiosis . Although parthenogenesis has been described in several species of brown algae ( e . g . [19 , 24 , 25] ) , the genetic basis , the underlying mechanisms and the evolutionary drivers and consequences of this process remain obscure . The haploid-diploid life cycle of the model brown alga Ectocarpus involves an alternation between a haploid gametophyte and a diploid sporophyte . Superimposed on this sexual cycle , an asexual , parthenogenetic cycle has been described for some Ectocarpus strains ( Fig 1A ) [19 , 22] . In this parthenogenetic cycle , gametes that fail to meet a partner of the opposite sex develop into haploid partheno-sporophytes . These partheno-sporophytes are indistinguishable morphologically from diploid sporophytes [19] . Partheno-sporophytes can produce gametophyte progeny to return to the sexual cycle through two mechanisms: 1 ) endoreduplication during development to produce diploid cells that can undergo meiosis or 2 ) individuals that remain haploid can initiate apomeiosis [19 , 22] . Here , we used a quantitative trait locus ( QTL ) approach to investigate the genetic basis of parthenogenesis in Ectocarpus siliculosus . We show that parthenogenesis is a complex genetic trait under the control of three QTLs , one major QTL located on the sex chromosome , another on chromosome 18 , and a further minor QTL also on chromosome 18 . We used genomic and transcriptomic analyses to establish a list of 77 candidate genes within the QTL intervals . We detected significant sex-by-genotype interactions for parthenogenetic capacity , highlighting the critical role of the sex chromosome in the control of asexual reproduction . Moreover , we observed negative effects of parthenogenetic capacity on male fitness in sexual crosses indicating trade-offs between sexual and asexual reproduction during the life cycle of Ectocarpus . Antagonistic selection for different alleles at the same locus in males and females , or during the haploid and diploid life cycle phases , has been shown to be able to maintain stable genetic polymorphisms ( [26 , 27] ) . Likewise , temporal or spatial habitat heterogeneity may cause strong fluctuating selection on sex-specific traits , thereby contributing to the maintenance of genetic variation in populations [28] . Based on our observations , we discuss the potential roles of sexually-antagonistic , ploidally-antagonistic and fluctuating selection in the maintenance of polymorphism at the parthenogenesis QTLs .
To precisely quantify the parthenogenetic capacity of two strains of E . siliculosus , clonal cultures of male ( RB1 ) and female ( EA1 ) gametophytes , collected from a field population in Naples ( S1 Table ) , were induced to release gametes under strong light ( see methods ) and pools of male and female gametes were allowed to settle separately , without mixing of the two sexes , on coverslips . Development of the gametes was then followed for 16 days ( Fig 1B , S1 Table ) . After 5 days , both male and female gametes had started to germinate and went through the first cell divisions . After 16 days , 94% of the female gametes on average had grown into >10 cell filaments , whereas 96% of the male gametes remained at the 3–4 cell stage and cell death was observed after about 20 days . Strains were therefore scored as parthenogenetic if more than 4% of parthenotes grew beyond the ten-cell stage ( Fig 1B ) . In several brown algal species , unfused male and female gametes show different parthenogenetic capacities , and it is usually the female gametes that are capable of parthenogenesis whereas male gametes are non-parthenogenic ( e . g . [25 , 29] ) . To investigate if there was a link between parthenogenetic capacity and sex , we crossed the female ( EA1 ) P+ strain with the male ( RB1 ) P- strain described above ( S1 Fig , S1 Table ) . The diploid heterozygous zygote resulting from this cross ( strain Ec236 ) was used to generate a segregating family of 272 haploid gametophytes . These 272 siblings were sexed using molecular markers [30] and their gametes phenotyped for parthenogenetic capacity ( see above ) . The segregating population was composed of 144 females and 128 males , consistent with a 1:1 segregation pattern ( chi2 test; p-value = 0 . 33 , S2 Table ) . Phenotypic assessment of the parthenogenetic capacity of the gametes released by each gametophyte revealed a significant bias in the inheritance pattern , with 84 individuals presenting a P- phenotype and 188 a P+ phenotype ( Chi2 test; p-value = 2 . 86x10-10 ) ( S2 and S3 Tables ) . Strikingly , all female strains exhibited a P+ phenotype whereas , although most male stains exhibited a P- phenotype , 30% of males were P+ ( S2 Table ) . This result indicated the presence of a parthenogenesis locus ( or loci ) that was not fully linked to the sex locus , and suggested a complex relationship between gender and parthenogenetic capacity . A subset of the segregating family derived from the EA1 x RB1 cross was tested for phenotype stability . We cultivated two male P+ gametophytes , two male P- gametophytes and two female P+ gametophytes under different environmental conditions , varying light levels and temperature . After two weeks in culture , fertility was induced , and the parthenogenetic capacity of the gametes was scored ( S4 Table ) . The parthenogenetic phenotype of all strains was stably maintained regardless of the culture conditions . We also tested the stability of the parthenogenetic phenotype across generations: gametes of each of the three types ( male P+ , male P- and female P+ ) were allowed to develop into partheno-sporophytes . Note that this experiment is possible with P- males because a small proportion of male P- gametes ( less than 4% ) does not exhibit growth arrest and is able to grow to maturity . After two weeks in culture , gamete-derived partheno-sporophytes produced unilocular sporangia and released spores that developed into gametophytes . This second generation of gametophytes was again phenotyped for parthenogenetic capacity , and the results showed without exception that the parthenogenetic phenotype was stably maintained across generations ( S4 Table ) . To further investigate the inheritance of parthenogenetic capacity , a male P+ individual was crossed with a P+ female ( S1 Fig ) . A total of 23 gametophyte lines were produced from two heterozygous sporophytes resulting from this cross . Phenotyping for sex and parthenogenesis revealed that all gametophyte lines exhibited a P+ phenotype , regardless of their sex ( S5 Table ) . We concluded that parthenogenesis is controlled by ( a ) genetic factor ( s ) . To produce a genetic map based on the EA1 x RB1 cross , a ddRAD-seq library was generated using 152 lines of the segregating progeny ( S1 Fig ) and sequenced on an Illumina HiSeq 2500 platform . A total of 595 million raw reads were obtained , of which 508 million reads passed the quality filters with a Q30 of 74 . 1% . A catalogue of 8648 SNP loci was generated using filtered reads from the parental strains and the STACKS pipeline ( version 1 . 44 ) [31] . Twenty-eight individuals were removed due to excessive missing genotypes ( see Methods ) and highly distorted markers were also removed . The final map was constructed with 124 individuals and contained 5595 markers distributed across 31 linkage groups ( LGs ) and spanning 2947 . 5 centimorgans ( cM ) . The average spacing between two adjacent markers was 0 . 5 cM and the largest gap was 17 . 6 cM ( on LG23 ) . The lengths of the 31 LGs ranged from 174 cM with 397 markers to 13 cM with 31 markers ( Fig 2A , S6 Table ) . The 5595 markers included one non-SNP-based marker that was not derived from the RAD-seq library . This PCR-based marker was used to map the sex-determining region of the sex chromosome because the high level of divergence between male and female sequences in this region precludes the generation of co-dominant RAD-seq markers . This sex marker was mapped to position 51 cM on linkage group 2 , thereby identifying this linkage group as the sex chromosome . Note that the Peruvian Ectocarpus strain that was used to generate the reference genome sequence [32] was originally taxonomically classified as Ectocarpus siliculosus but subsequent analysis has demonstrated that this strain actually belongs to a distinct species within the Ectocarpus siliculosi group [33] . The genetic map generated here using bona fide Ectocarpus siliculosus strains is therefore for a novel species relative to the genetic maps generated for the Peruvian strain [34 , 35] . To decipher the genetic architecture of parthenogenesis in E . siliculosus , we applied an “all-or-none” phenotyping and a QTL mapping approach [36 , 37] , by considering P+ and P- as the two most ‘extreme’ phenotypes . We used the high-resolution genetic map to statistically associate markers with the P+ and P- phenotypes in the segregating family described above . QTL mapping and association analysis identified three QTLs for parthenogenesis: two large-effect QTLs ( r2 > 15% ) and one smaller-effect QTL ( r2 = 11 . 9% ) ( Fig 2A ) . Together , these three QTLs explained 44 . 8% of the phenotypic variance . The QTLs were located on two different LGs , LG2 and LG18 ( Fig 2A ) . One of the large effect parthenogenesis QTLs ( P1 ) , which is located on LG2 , co-localized with the sex marker , corresponding to the SDR of the sex chromosome ( Fig 2 ) . The P1 locus was detected at the highest significance level ( p-value <0 . 0001 ) with the Kruskal-Wallis statistical test ( K* = 20 . 392 ) . The other major effect locus , which we refer to as the P2 locus , was located on LG18 , and was also detected at the highest significance level with a Kruskal-Wallis statistical test ( p-value<0 . 0001 , K* = 19 . 993 ) ( S7 Table ) . A non-parametric interval mapping ( IM ) method detected both the P1 and P2 loci , and indicated a proportion of variance explained ( PVE ) of 16 . 6% for P1 and 16 . 3% for P2 . The P1 locus spanned 13 . 36 cM from 44 . 84 to 52 . 0 cM with a peak position at 51 . 0 cM whereas the P2 locus spanned 2 . 82 cM , from 92 . 77 to 95 . 59 cM with a peak position at 93 . 98 cM . The third QTL ( P3 ) , which was also located on LG18 , was detected only with the Kruskal-Wallis statistical test ( K* = 14 . 634 , p-value<0 . 0005 ) . The P3 QTL had a smaller effect than P1 and P2 , and explained 11 . 9% of the phenotypic variance ( Fig 2A and 2B; S7 Table ) . Note that the QTL mapping described above was implemented using all 152 progeny ( S1 Fig ) , which included both male and female strains . To investigate the contribution of the sex-specific , non-recombining region of the sex chromosome , we performed the same analysis using a subset of 93 male strains . The result showed that when females were excluded , the P1 and the P3 QTLs were not detected , and only the QTL located on LG18 ( P2 ) was significantly detected ( S7 Table ) . The absence of detection of the P1 QTL was not a result of reduced statistical power due to the small sample size , because the QTL was detected when a sub-sample of 93 male and female individuals with the same sex ratio as the full 124 samples was used ( S7 Table ) . The minor P3 QTL was at the limit of significance when the 93 sub-sampled individuals were used , suggesting that the reduced sample size prevented the detection of this minor QTL . To further characterise the cross between the two parthenogenetic strains Ec236-91 ( female P+ ) x Ec236-202 ( male P+ ) ( S1 Fig ) , both strains were genotyped to determine which alleles of the autosomal parthenogenesis QTLs they carried . The maternal strain Ec236-91 carried the B ( maternal ) allele at the P2 locus and the A ( paternal ) allele at the P3 locus , whereas the male P+ strain Ec236-202 had B alleles at both the P2 and P3 loci ( S8 Table ) . All the male progeny of this cross had a P+ phenotype ( S5 Table ) , indicating that a B allele at the P2 locus is sufficient to confer male gamete parthenogenesis . An epistasis analysis was carried out to detect potential interactions between the parthenogenesis QTLs . Two analyses were performed , using either all 152 male and female progeny ( ‘full dataset’ ) or the subset of all the 93 male individuals . We observed significant sex-by-genotype interactions for parthenogenetic capacity . The analysis of the full dataset identified an epistatic interaction between the P2 QTL and the P1 QTL ( Fig 3 ) . When the same analysis was carried out with only the males , this epistatic interaction was not detected ( S9 Table ) . In Fig 3 , the B allele was inherited from the female parent , and the A allele from the male parent . All females were parthenogenetic and therefore their parthenogenetic phenotype was independent of the allele carried at the P2 locus . In contrast , the phenotype of males depended on the allele carried at the P2 locus . An additional interaction was detected between the P2 QTL and the P3 QTL . In this case , the frequency of P+ individuals was higher when the maternal B allele was present at the P2 locus and the effect was strongest when the P3 locus carried the paternal A allele ( Fig 3B , S10 Table ) . Several additional interactions were detected between the P2 QTL and markers on several autosomes when the male-only dataset was analysed ( S9 Table ) . A total of 167–169 genes ( depending on whether we consider the U or the V , respectively , for P1 ) were identified within the three QTL intervals . Gene Ontology enrichment tools were used to test if some functional categories were over-represented in these genes . BLAST2GO analysis indicated a significant enrichment in processes related to signalling and cell communication ( p-value < 0 . 0001 ) ( S11 Table ) . We used several approaches to identify candidate parthenogenesis genes within the three QTL intervals . First , we reasoned that genes involved in parthenogenesis should be expressed in at least one of the gamete types , P+ or P- , where parthenogenesis is initiated . Strains EA1 and RB1 did not produce enough gametes for RNA extraction . We therefore generated RNA-seq data from P+ female and P- male strains from another species within the E . siliculosi group , Ectocarpus species 1 [38] ( see methods ) . We analysed the abundance of the transcripts of orthologs of the 167–169 genes within the three QTL intervals ( 42/44 genes , 30 genes and 95 genes in the P1 , P2 and P3 intervals respectively , S7 Table ) . Based on this analysis , 113/119 genes ( again , depending on whether we consider the U or the V , respectively ) were classed as being expressed in at least one of the gamete types ( S12 Table ) . Second , we looked for genes that were significantly differentially expressed between P+ and P- gametes , again using the data for Ectocarpus species 1 orthologues . Overall , 4902 orthologues were differentially expressed in P+ versus P- strains across the genome , of which 42 corresponded to genes located within the QTL intervals ( eight within the P1 , seven within the P2 and 27 within the P3 QTL intervals; Fig 2D , S12 Table ) . The QTL intervals were therefore significantly enriched in genes that we classed as being differentially expressed between P+ and P- strains ( Fisher exact test; p-value = 0 . 0212 ) . Third , we looked for polymorphisms with potential effects on the functions of the candidate genes . Comparison of the parental genomic sequences identified a total of 3046 indels and 9702 SNPs within the three QTL intervals , including 46 indels and 152 SNP located in exons ( S12 and S13 Tables ) . In total , 61 genes within the QTL intervals carried SNPs or indels that corresponded to non-synonymous modifications of the coding sequence and were therefore predicted to affect protein function . The male and female SDRs do not recombine [39] and have therefore diverged considerably over evolutionary time . This has included loss and gain of genes but also strong divergence of the genes that have been retained in both regions ( gametologs ) . All SDR genes were therefore retained as candidates ( S12 Table ) . We then combined the three approaches . The criteria we used were that genes involved in parthenogenesis must be expressed in gametes and they should have either differential expression in P+ versus P- gametes or carry a non-synonymous polymorphism . This reduced the number of candidates to 13/18 ( U/V chromosome ) genes in the P1 , nine genes in the P2 and 50 genes in the P3 QTL ( Fig 2D , S12 Table ) . Taking genes that were both differentially expressed in P+ versus P- gametes and that carried a non-synonymous polymorphism ( S12 Table , Fig 2D ) further reduced the list of candidate genes to 8/13 ( U/V ) , one and 15 candidates in P1 , P2 and P3 , respectively . It is not clear why some strains of Ectocarpus exhibit male gamete parthenogenesis whilst others do not . More specifically , bearing in mind that all strains tested so far exhibit parthenogenesis of female gametes , why are male gametes not parthenogenetic in some lineages ? To address this question , we investigated if there were differences in fitness between P- and P+ male gametes for parameters other than parthenogenetic growth . Specifically , we examined fertilisation success ( capacity to fuse with a female gamete ) and growth of the resulting diploid sporophyte . We tested twelve combinations of crosses between four P- or P+ males and five females ( S14 and S15 Tables ) . We fitted a generalized linear mixed model with a logistic regression to the data and observed that the male P- gametes showed a higher rate of fusion with female gametes than the P+ male gametes ( p = 0 . 0048 ) . However , the data showed overdispersion , which needed to be accounted for to avoid biased parameter estimates . Accounting for overdispersion led to a higher but still significant p-value ( p = 0 . 047 ) . Therefore , we concluded that , overall , male P- gametes fuse more efficiently with female gametes than do P+ male gametes . Moreover , embryos arising from P- male gametes also grew significantly more rapidly than embryos derived from fusion with male P+ gametes ( Fig 4B and 4C , Mann-Whitney u-test p<0 . 05 , S16 Table ) . The overall size of zygotes is expected to be correlated with zygotic and diploid fitness [40–42] . We therefore hypothesised that if P- male gametes are larger , fusion with a female gamete would generate larger ( and therefore fitter ) zygotes . Measurements of gamete size of P+ and P- strains revealed significant differences in gamete size between different strains ( Kruskal-Wallis test , Chi2 = 3452 . 395 , P<2 . 2e-16 , S17 Table , Fig 4D ) . However , there was no correlation between the parthenogenetic capacity of male gametes and their size , suggesting that the increased fitness of the zygotes was unlikely to be related to the size of the male gametes . Taken together , these analyses indicate that P+ male gametes exhibit overall reduced fitness in sexual crosses , both in terms of success of fusion with a female gamete and growth of the resulting embryo . We found no link between the size of the male gamete and the capacity to perform parthenogenesis , which excludes the possibility that the fitness decrease is due to the size of the male gamete .
In this study , we uncovered the genetic architecture of parthenogenesis in the brown alga E . siliculosus and demonstrated that this trait is controlled by two major and one minor QTL loci that , together , account for 44 . 8% of the phenotypic variation . The two main QTL loci were located at the SDR of the sex chromosome and on LG18 , respectively , and the minor QTL was also located on LG18 . Analysis of differential expression patterns and polymorphism for genes within the QTL intervals allowed the establishment of a list of 77 candidate parthenogenesis genes: 13/18 genes within the sex chromosome QTL interval ( in the U and V respectively ) , nine genes within the P2 locus and 50 within the interval of the minor P3 locus . Interestingly , within the major P2 QTL a strong candidate gene coded for a membrane-localized ankyrin repeat-domain palmitoyltransferase ( Ec-20_004890 ) . In S . cerevisiae , genes belonging to the same family are involved in the gamete pheromone response pathway , regulating the switching between vegetative and mating states [43 , 44] . Our results reveal a critical role for the sex chromosome in the control of parthenogenesis , with a major effect QTL being located within the SDR . In females , parthenogenesis occurs regardless of the alleles carried at the P2 and P3 loci , whereas males are parthenogenetic only when they have a P+ allele at least at one of the autosomal QTL . Accordingly , epistatic interactions between the SDR and the major P2 QTL locus were detected only when both males and females were included in the analysis . The epistatic effect could be due to the production of a repressor of parthenogenesis by the male V-specific region or to the production of an activator of parthenogenesis by the female U-specific region ( in either case the activator or repressor could be directly encoded by the SDR or produced indirectly as part of the male or female sex-differentiation programs ) . Interestingly , several genetically male kelp strains of Laminaria pallida and Macrocystis pyrifera were shown to produce unusual reproductive structures resembling eggs , which were capable of parthenogenesis ( Ingo Maier , D . Muller , pers . commun . ) . These observations suggest that , if an activator of parthenogenesis is produced in the female , then this probably occurs indirectly as part of the female sex-differentiation program rather than being directly encoded by the SDR , at least in these species ( because these kelp individuals lacked a U sex chromosome ) . Our results indicate that parthenogenetic capacity has a negative impact on male gamete fitness during sexual reproduction . Zygotes produced by fusion with P- male gametes grew significantly faster than those produced by fusion with P+ male gametes . We also observed that P- male gametes exhibited a higher percentage of successful fertilisations with female gametes than P+ male gametes . The underlying mechanistic basis for the decreased fitness of P+ compared with P- male gametes is currently elusive . P+ and P- alleles may have a direct phenotypic effect on the male gametes , although we have not detected any difference in parameters such as gamete size . A link between mitochondria metabolism and parthenogenetic capacity has been evoked in another brown algal species [25] . Moreover , parthenogenesis-promoting alleles at P2 and P3 may have an indirect effect on fitness because they favour asexual rather than sexual reproduction , leading to reduced recombination rates and hence reduced capacity to eliminate deleterious mutations due to Hill-Robertson interference [45 , 46] . It is not clear , however , whether the degree of asexual reproduction in field populations would be sufficient for this effect to be significant and note also that increased parthenogenesis is also expected to lead to an increase in haploid purifying selection , which would tend to counteract the effect of reduced recombination to some extent . Although we have not investigated allele frequencies in natural populations in this study , an earlier analysis of the Naples E . siliculosus population indicated that about 12 . 5% of male individuals were phenotypically P+ [47] , suggesting that P+ alleles of parthenogenesis QTLs may be relatively abundant in this population . If P+ males are expected to exhibit reduced fitness in sexually reproducing populations , and females are phenotypically P+ regardless of the allele at the P2 and P3 QTL , how can autosomal P+ alleles be preserved in the population ? In other words , how is polymorphism maintained at the parthenogenesis QTLs ? In the following paragraphs , we discuss several hypotheses that may explain the maintenance of this polymorphism . Note that it is likely that different evolutionary forces have been acting on the sex chromosome QTL ( P1 ) compared to the autosomal QTLs ( P2 and P3 ) and these differences are discussed where appropriate . One interesting possibility is that parthenogenesis is a sexually antagonistic trait ( or at least differentially selected in males versus females ) . Indeed , this is probably the most likely explanation for the existence of the fixed polymorphism between males and females at the P1 locus ( SDR QTL ) if we consider that parthenogenetic capacity will tend to be beneficial for females but less beneficial for males because of negative effects on male sexual reproduction . The incidence of parthenogenesis in males and females would reflect the costs and benefits of each mode of reproduction in each of the sexes . Parthenogenesis in males may be more costly because of potential mutations caused by reactive oxygen species production during active swimming [48] , which would be exposed during ( haploid ) parthenogenetic development . Moreover , the smaller size of male gametes may impede early development ( for example due to limited resources to invest in parthenogenetic germination ) . In females , whose gametes have a passive behaviour and are bigger , a ‘facultative’ parthenogenesis strategy could be favoured , specifically in conditions of sperm limitation . Although we could not measure the effect of parthenogenetic capacity on female gamete fitness , because all females were phenotypically P+ , sexual antagonism would be consistent with the pervasiveness of the female P+ phenotype and the differences in fitness between P+ and P- males . This phenomenon would be particularly relevant in spatially heterogeneous and/or unpredictable environments , where the P+ or P- allele ( s ) in males would be alternatively selected for or against , depending on female density . Another potential mechanism for the maintenance of genetic variation is opposing selection during the diploid and haploid stages of biphasic life cycles , also known as ploidally-antagonistic selection [26] . Parthenogenesis could be considered an example of a trait under ploidally/generation antagonistic selection because P- alleles transmitted by the male gamete are advantageous to the diploid ( sporophyte ) generation ( because zygotes grow faster if the father is P- ) but detrimental to the haploid ( partheno-sporophyte ) generation ( because if they do not find a female gamete , male gametes that carry a P- allele die ) . Ploidally-antagonistic selection has been proposed to have a significant impact on major evolutionary dynamics , including the maintenance of genetic variation and the rate of adaptation [26 , 49 , 50] . Note , however , that the detrimental effect on males will depend on the availability of female gametes so that , under conditions where sexual reproduction is favoured , P- alleles should be beneficial for males and P+ detrimental . Mathematical modelling [51] predicts that when selection differs between the sexes ( and in particular when the gametophyte-deleterious allele is neutral or slightly beneficial in one of the sexes ) , being close to or within the SDR expands the range of parameters allowing generation-antagonistic mutations to spread . Indeed , conflict arising from generation-antagonism or from differences in selection in gametophytes compared with the sporophyte generation is best resolved by complete linkage to the SDR [51] . The effect of ploidally-antagonistic selection is therefore likely to be strongest for the P1 locus but this effect could also influence allele frequencies at the P2 and P3 loci . Temporal or spatial changes in population density are extremely common and these are expected to cause strong fluctuating selection on life-history traits , thereby contributing to the maintenance of genetic polymorphism in populations ( e . g . [28] ) . Polymorphism can also be maintained when selection fluctuates as a result of developmental changes , for example when selection only occurs in one sex , resulting in the sheltering of alleles in the sex where they are not under selection [52] . In the case of E . siliculosus , for example , autosomal parthenogenesis alleles could be protected from selection because they have no phenotypic effect in females . In other words , if phenotypic expression of autosomal P- allele ( s ) is limited to males , fluctuating selection of this sex-limited trait could lead to the existence of a protected polymorphism , and contribute to the maintenance of genetic variance at the autosomal QTLs . The P+ allele would be maintained because it is advantageous in males when females are rare or when populations have low density . Note that this mechanism is not relevant for the P1 locus because alleles are systematically transmitted to individuals of the same sex . In the brown algae , the ancestral state appears to have been sexual reproduction through fusion of strongly dimorphic gametes ( oogamy ) , that were incapable of parthenogenesis ( [53] , reviewed in [29] ) . Gamete parthenogenesis evolved secondarily , and a challenge for understanding the adaptive nature of gamete parthenogenesis in these organisms would be to identify the conditions under which it occurs in nature . Brown algae exhibit a remarkable degree of reproductive plasticity during their life cycle [19 , 54] and it is possible that this plasticity is related to their capacity to adapt to new conditions , in particular low population densities or very fragmented habitats where finding a partner may be problematic . Our data supports the idea that P+ individuals may employ a bet-hedging strategy in the sense that their gametes are capable of reproducing either sexually or parthenogenically , thereby ensuring reproduction in heterogeneous and unpredictable environments . It has been predicted that in marginal populations , or other situations where mates are limited , parthenogenesis could be adaptive and thus selectively favoured [55] . In animals ( fish , Drosophila ) rapid transition between reproductive strategies were observed following the removal of the mate , supporting the hypothesis that parthenogenesis has a reproductive advantage under conditions of isolation from potential mates [56] . A recent study of Ectocarpus siliculosus populations in NW of France has shown that asexual populations are prevalent in the field , but gamete parthenogenesis does not appear to play a critical role in this population , and instead , sporophytes are produced mainly from the development of diploid , asexual spores [57] . However , parthenogenesis is an important process in field populations of other brown algal species , such as Scytosiphon lomentaria [20] . Additional population data are required , specifically parthenogenesis allele frequencies for natural populations where individuals are found at different densities , for marginal versus central populations and for different types of habitat , to further investigate whether there is an adaptive benefit to parthenogenesis .
Gametophytes of E . siliculosus ( S1 Table ) were maintained in culture as previously described [58] . E . siliculosus strains can be maintained in the gametophyte generation indefinitely , with weekly changes in culture media [58] . Clonal cultures of male and female gametophytes were subjected to strong light ( 100 μm photons/m2/s ) and low temperatures ( 10°C ) to induce fertility resulting in the release of large numbers of gametes ( >10e5 ) . Gametes were allowed to settle on coverslips and their development was monitored under an inverted microscope ( Olympus BX50 ) . The sex of the gametophytes was assessed using PCR-based sex markers ( FeScaf06_ex03 and 68_56_ex02 , which correspond to conserved regions within the U and V SDRs , respectively , [39 , 59] ) . Parthenogenetic capacity was evaluated after 15 days in culture using an inverted microscope ( Olympus CKX41 ) . Strains were scored as parthenogenetic if more than 4% of parthenotes grew beyond the ten-cell stage . A cross between a parthenogenetic female ( strain EA1 ) and a non-parthenogenetic male ( strain RB1 ) was carried out using a standard genetic cross protocols [60] and a diploid heterozygous sporophyte was isolated ( Ec236 ) ( Fig 1; S1 Table ) . At maturity , the sporophyte ( strain Ec236 ) produced unilocular sporangia , i . e , the reproductive structure where meiosis takes place ( Fig 1 ) . A total of 272 unilocular sporangia were isolated , and one gametophyte was isolated from each . The 272 strains of the EA1 x RB1 derived segregating population were cultivated in autoclaved sea water supplemented with half strength Provasoli solution [61] at 13°C , with a light dark cycle of 12:12 ( 20 μmol photon m-2 s-1 ) using daylight-type fluorescent tubes [58] . All manipulations were performed in a laminar flow hood under sterile conditions . We phenotyped the strains for parthenogenetic capacity ( P+ or P- ) and for sex ( male or female ) . Parthenogenetic capacity was was evaluated after 15 days in culture . Parthenogenesis in E . siliculosus is a binary trait . Strains were scored as parthenogenetic ( P+ ) if more than 4% of parthenotes grew beyond the ten-cell stage . In order to assess phenotype stability , gametophytes were sub-cultivated in different conditions for two weeks and then exposed to high intensity light to induce fertility . Parthenogenetic capacity was measured using the released gametes ( S3 Table ) . To test the stability of the phenotype across generations , we cultivated partheno-sporophytes and induced them to produce unilocular sporangia and release meio-spores to obtain a new generation of gametophytes . The parthenogenetic capacity of gametes derived from these second-generation gametophytes was then tested ( S3 Table ) . Note that this experiment is feasible in P- males because a very small proportion ( less than 4% ) of their gametes are nevertheless able to develop into mature partheno-sporophytes . Each of the 272 gametophytes of the EA1 x RB1 segregating family was frozen in liquid nitrogen in a well of a 96 well plate . After lyophilization , tissues were disrupted by grinding . DNA of each gametophyte was extracted using the NucleoSpin Plant II kit ( Macherey-Nagel , Germany ) according to the manufacturer’s instructions and stored at -80°C . Sexing of gametophytes was carried out using two PCR-based sex markers , one for each sex ( FeScaf06_ex03 forward: CGTGGTGGACTCATTGACTG; FeScaf06_ex03 reverse: AGCAGGAACATGTCCCAAAC; 68_56_ex02 forward: GGAACACCCTGCTGGAAC; 68_56_ex02 reverse: CGCTTTGCGCTGCTCTAT ) [39] . PCR was performed with the following reaction temperatures: 94°C 2min; 30 cycles of 94°C 40s , 60°C 40s and 72°C 40s; 72°C 5min , and with the following PCR mixture 2 μL DNA , 100 nM of each primers , 200 μM of dNTP mix , 1X of Go Taq green buffer , 2 mM of MgCl2 , 0 . 2 μL of powdered milk at 10% and 0 . 5 U of Taq polymerase ( Promega ) . A double digest RAD sequencing ( ddRAD-seq ) library was generated using 152 individuals from the EA1 x RB1 segregating population . Parthenogenetic individuals were selected ( 37 females and 36 males ) as well as non-parthenogenetic males ( 79 individuals ) . DNA extraction was performed for each individual ( Macherey-Nagel , NucleoSpin Plant II kit ( GmbH & Co . KG , Germany ) and DNA quantity was measured and standardized at 100 ng using a PicoGreen ( Fischer Scientific ) method for quantification . The DNA quality was checked on agarose gels . The ddRAD-seq library was constructed as in [62] using HhaI and SphI restriction enzymes ( New England Biolabs , https://www . neb . com/ ) . Those enzymes were selected based on an in silico digestion simulation of the Ec32 reference genome [18] using the R package SimRAD [63] . After digestion , samples were individually barcoded using unique adapters by ligation with T4 DNA ligase ( New England Biolabs , https://www . neb . com/ ) . Then , samples were cleaned with AMPure XP beads ( Beckman Coulter Genomics ) , and PCR was performed with the Q5 hot Start High-Fidelity DNA polymerase kit ( New England Biolabs , https://www . neb . com/ ) to increase the amount of DNA available for each individual and to add Illumina flowcell annealing sequences , multiplexing indices and sequencing primer annealing regions . After pooling the barcoded and indexed samples , PCR products of between 550 and 800 bp were selected using a Pippin-Prep kit ( Sage Science , Beverly , MA , USA ) , and the library was quantified using both an Agilent 2100 Bioanalyzer ( Agilent Technologies ) and qPCR . The library was sequenced on two Illumina HiSeq 2500 lanes ( Rapid Run Mode ) by UMR 8199 LIGAN-PM Genomics platform ( Lille , France ) , with paired-end 250 bp reads . The ddRAD-seq sequencing data was analysed with the Stacks pipeline ( version 1 . 44 , [31] ) . The raw sequence reads were filtered by removing reads lacking barcodes and restriction enzyme sites . Sequence quality was checked using a sliding window of 25% of the length of a read and reads with <90% base call accuracy were discarded . Using the program PEAR ( version 0 . 9 . 10 , [64] ) paired-end sequencing of short fragments generating overlapping reads were identified and treated to build single consensus sequences . These single consensus sequences were added to the singleton rem1 and rem2 sequences produced by Stacks forming a unique group of singleton sequences . For this study , paired-end reads and singleton sequences were then trimmed to 100 bp with the program TRIMMOMATIC [65] . The genome of the male parent of the population ( strain RB1 ) was recently sequenced to generate an assembly [66] guided by the Ectocarpus species 7 reference genome published in 2010 [67] . We performed a de novo analysis running the denovo_map . pl program of Stacks . Firstly , this program assembles loci in each individual de novo and calls SNPs in each assembled locus . In a second step , the program builds a catalog with the parental loci and in a third step , loci from each individual are matched against the catalogue to determine the allelic state at each locus in each individual . We then used BWA ( Li , H . Aligning sequence reads , clone sequences and assembly contigs with BWA-MEM . arXiv:1303 . 3997 ) to align the consensus sequence of the catalog loci to the reference genome and used the Python script “integrate_alignments . py” of the Stacks pipeline to integrate alignment information back into the original de novo map output files [68] . In a final step , SNPs were re-called for all individuals at every locus and exported as a vcf file . The vcf file obtained with the Stacks pipeline was first filtered to keep only loci with maximum of 10% of missing samples and samples with a maximum of 30% of missing data . The program Lep-MAP3 ( LP3 ) [69] was used to construct the genetic map ( S18 Table ) . LP3 is suitable to analyse low-coverage datasets and its algorithm reduces data filtering and curation on the data , yielding more markers in the final maps with less manual work . In order to obtain the expected AxB segregation type for this haploid population , the pedigree file was constructed by setting the parents as haploid grand-parents and two dummy individuals were introduced for parents . The module ParentCall2 of LP3 took as input the pedigree and the vcf files to call parental genotypes . The module SeparateChromosomes2 used the genotype call file to assign markers into linkage groups ( LGs ) . Several LOD score limits were tested to obtain an optimal LOD score of 8 giving a stable number of LGs . The module JoinSingles2All was then run to assign singular markers to existing LGs by computing LOD scores between each single marker and markers from the existing LGs . The module OrderMarkers2 then ordered the markers within each LG by maximizing the likelihood of the data given the order . Sex averaged map distances were computed and 10 runs were performed to select the best order for each LG , based on the best likelihood . This module was run with the parameters grandparentPhase = 1 and outputPhasedData = 1 in order to obtain phased data for QTL mapping . This phased data was converted to fully informative genotypic data using the script map2gentypes . awk distributed with the LP3 program . The PCR-based sex markers ( FeScaf06_ex03 and 68_56_ex02 , which correspond to conserved regions within the U and V SDRs , respectively ) were recoded as a unique sex marker , with the B allele for females and A allele for males . Identification and mapping of QTLs were carried out using R/qtl ( version 1 . 39–5 ) and MapQTL version 5 . Non-parametrical statistics were used because parthenogenetic capacity was phenotyped as a binary trait ( either 0 for non-parthenogenetic or 1 for parthenogenetic ) . In R/qtl , the scanone function was used with the “binary” model to perform non-parametrical interval mapping with the binary or Haley-Knott regression methods . The function refineqtl was used to obtain improved estimates of the locations of the QTLs . In MapQTL , the Kruskal-Wallis non-parametric method was used . To determine the statistical significance of the major QTL signal , the LOD significant threshold was determined by permutation . The small number of gametes released from Ectocarpus siliculosus strains did not allow RNA-seq data to be obtained from this species . To analyse gene expression in P- ( male ) and P+ ( female ) gametes , we therefore used two Ectocarpus species 1 strains belonging to the same Ectocarpus siliculosi group [33] , a P- male ( NZKU1_3 ) and a P+ female ( NZKU32-22-21 ) , which produce sufficient numbers of gametes for RNA extraction . Note that , although there are currently only three official species in the Ectocarpus genus , genetic analysis has shown that there are at least 15 species in the genus with most differences being cryptic [33] . Gametes of male and female Ectocarpus species 1 were concentrated after brief centrifugation , flash frozen and stored at -80°C until RNA extraction . RNA was extracted from duplicate samples using the Qiagen RNeasy plant mini kit ( www . qiagen . com ) with an on-column DNase I treatment . Between 69 and 80 million sequence reads were generated for each sample using Illumina HiSeq 2000 paired-end technology with a read length of 125 bp ( Fasteris , Switzerland ) ( S19 Table ) . Read quality was assessed with FastQC ( http://www . bioinformatics . babraham . ac . uk/projects/fastqc ) , and low quality bases and adapter sequences were trimmed using Trimmomatic ( leading and trailing bases with quality below 3 and the first 12 bases were removed , minimum read length 50bp ) [65] . High score reads were used for transcriptome assembly generated with the Trinity de novo assembler [70] with default parameters and normalized mode . RNA-seq reads were mapped to the assembled reference transcriptome using the Bowtie2 aligner [71] and the counts of mapped reads were obtained with HTSeq [72] . Expression values were represented as TPM and TPM<1 was applied as a filter to remove noise if both replicates of both samples exhibit it . Differential expression was analysed using the DESeq2 package ( Bioconductor; [73] ) using an adjusted p-value cut-off of 0 . 05 and a minimal fold-change of two . The reference transcripts were blasted to the reference genome Ec32 predicted proteins ( http://bioinformatics . psb . ugent . be/orcae/overview/EctsiV2 ) ( e-value cut-off = 10e-5 ) and the orthology relationship between Ectocarpus species 1 and Ec32 ( Ectocarpus species 7 ) was established based on the best reciprocal blast hits . We used two methods to identify putative candidate genes located in the QTL intervals . First , a marker-by-marker method , by mapping the sequences of the markers located within each QTL interval to the reference genome of the closely related reference species strain Ec32 ( Cock et al . , 2010 ) . When a sequence successfully mapped to the Ec32 genome , a coordinate was recorded for the marker , relative to its position on the physical map of Ec32 . The second method used the same approach but was based on the reference genome of the paternal strain of the population ( strain RB1 ) . There were some differences between the two lists obtained by the two methods , which are due to the following factors: ( a ) because the assembly of the RB1 genome was guided by the Ec32 reference genome and its annotation was based on Ec32 transcriptomic data , the RB1 genome potentially lacks some genes that would be due to loci such as genes that are unique to the species E . siliculosus ( RB1 strain ) being omitted during the guided assembly . Hence the list obtained with the first method ( using the Ec32 genome ) contains genes that are absent from the RB1 genome; ( b ) while the two species are closely related , they are not identical , and the E . siliculosus genetic map exhibited some rearrangements compared to Ec32 which placed some markers , along with associated genes , into the QTL intervals ( these missing markers were located elsewhere on the Ec32 genome ) . In summary , the list obtained with Ec32 genome contained some genes that are missing from the RB1 genome because of its imperfect guided assembly and the list obtained with the RB1 genome contained some genes absent from the corresponding intervals on Ec32 because of rearrangements . A final , conservative list of candidate genes was obtained by merging the two lists in order not to omit any gene that were potentially located within the intervals . Draft genomes sequences are available for the parent strains RB1 and EA1 [66] . Using Bowtie2 , we aligned the EA1 genome against the RB1 genome and generated an index with sorted positions . The program samtools mpileup [74] was used to extract the QTL intervals and call variants between the two genomes . The positions of variants between the two genomes were identified and filtered based on mapping and sequence quality using bcftools [75] . The annotation file generated for the RB1 genome was then used to select SNPs and indels located in exons of protein-coding genes for further study ( bcftool closest command ) . The effect of polymorphism on modification of protein products was assessed manually using GenomeView [76] , the RB1 genome annotation file ( gff3 ) and the vcf file for each QTL region . A Gene Ontology enrichment analysis was performed using two lists of genes: a predefined list that corresponded to genes from all three QTL intervals and a reference list including all putative genes in the mapped scaffolds based on the Ec32 reference genome and that had a GO term annotation . The analysis was carried out with the package TopGO for R software ( Adrian Alexa , Jörg Rahnenführer , 2016 , version 2 . 24 . 0 ) by comparing the two lists using a Fisher’s exact test based on gene counts . Epistasis analysis was carried out with the R package R/qtl ( version 3 . 3 . 1 ) . Two analyses were performed , one with the full data set ( female and male genotypes generated with RAD-seq method ) and the second with only the male individuals . For both analyses , the scantwo function from R/qtl were used with the model “binary” as the phenotypes of the individuals is either 1 ( P+ ) or 0 ( P- ) . Reproductive success was assessed in the segregating population by measuring the capacity of male P+ and P- gametes to fuse with female gametes and by measuring the length of the germinating sporophytes derived from these crosses . For this , we crossed males and females as described in [60] . Briefly , we mixed the same amount of male and female gametes ( app . 1x103 gametes ) in a suspending drop , and the proportion of gametes that succeeded in fusing was scored . Two different P+ males ( Ec236-34 and Ec236-245 ) and two different P- males ( Ec236-10 and Ec236-298 ) were crossed with five different females ( Ec236-39; -203; -65; -284 and Ec560 ) ( S14 Table ) . Between 18 and 254 germlings were counted for each cross . To determine whether there was a difference between the capacity of P+ and P- male gametes to fuse with female gametes , we applied a generalized linear mixed model with a logistic regression ( family = binomial ) using the glmer function from the R package lme4 ( details in S14 Table ) . The model was run with and without observation-level random effects to account for overdispersion [77] . The length of zygotes derived from three different crosses between female strains and either male P- or male P+ strains were measured after between 3h to 4 days of development using Image J 1 . 46r [78] . For all datasets , the assumption of normality ( Shapiro test ) and the homoscedasticity ( Bartlett’s test ) were checked . The latter’s assumptions were not met for zygote length , and consequently statistical significance differences at each time of development was tested with a non-parametrical test ( Mann Whitney U-test , α = 5% ) . Gamete size was measured for representative strains of each parthenogenetic phenotype found in the segregating population ( P+ and P- ) ( S3 Table ) . Synchronous release of gametes was induced by transferring each gametophyte to a humid chamber in the dark for approximately 14 hours at 13°C followed by the addition of fresh culture medium under strong light irradiation . Gametes were concentrated by phototaxis using unidirectional light , and collected in Eppendorf tubes . Gamete size was measured by impedance-based flow cytometry ( Cell Lab QuantaTM SC MPL , Beckman Coulter ) . A Kruskal-Wallis test ( α = 5% ) followed by a posthoc Dunn’s test for pairwise comparisons were performed using R software to compare female and male gamete size ( S16 Table ) . | Asexual reproduction is widespread among all major clades of eukaryotes . Parthenogenesis represents a specific mode of asexual reproduction , secondarily derived from sexual reproduction , and refers to the development of a multicellular organism from an unfertilised gamete . Parthenogenesis has evolved independently in a wide variety of groups , but the genetic basis and the evolutionary forces driving transitions from sexual to parthenogenetic reproduction remain elusive . Here , we explore genetic , genomic and transcriptomic data from the brown alga Ectocarpus to uncover the genetic architecture of parthenogenesis . The brown algae are a group of complex multicellular organisms that have been evolving independently from animals and plants for more than a billion years , and they have recently emerged as important models to study the evolution of reproductive modes . We show that parthenogenesis is a complex genetic trait under the control of the sex locus , together with two additional autosomal quantitative trait loci , highlighting the critical role for the sex chromosomes in the control of asexual reproduction in this organism . We identify several negative effects of parthenogenesis on male fitness and reveal evidence for trade-offs between sexual and asexual reproduction during the life cycle of Ectocarpus . Our results support the idea that parthenogenesis may be under both sex-specific selection and generation/ploidally-antagonistic selection , but the action of fluctuating selection on this trait may also contribute to the maintenance of polymorphisms in populations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"methods"
] | [
"quantitative",
"trait",
"loci",
"alleles",
"sequence",
"assembly",
"tools",
"developmental",
"biology",
"genome",
"analysis",
"molecular",
"biology",
"techniques",
"plants",
"research",
"and",
"analysis",
"methods",
"sex",
"chromosomes",
"genomics",
"gene",
"mapping",
... | 2019 | A key role for sex chromosomes in the regulation of parthenogenesis in the brown alga Ectocarpus |
Rates of random , spontaneous mutation can vary plastically , dependent upon the environment . Such plasticity affects evolutionary trajectories and may be adaptive . We recently identified an inverse plastic association between mutation rate and population density at 1 locus in 1 species of bacterium . It is unknown how widespread this association is , whether it varies among organisms , and what molecular mechanisms of mutagenesis or repair are required for this mutation-rate plasticity . Here , we address all 3 questions . We identify a strong negative association between mutation rate and population density across 70 years of published literature , comprising hundreds of mutation rates estimated using phenotypic markers of mutation ( fluctuation tests ) from all domains of life and viruses . We test this relationship experimentally , determining that there is indeed density-associated mutation-rate plasticity ( DAMP ) at multiple loci in both eukaryotes and bacteria , with up to 23-fold lower mutation rates at higher population densities . We find that the degree of plasticity varies , even among closely related organisms . Nonetheless , in each domain tested , DAMP requires proteins scavenging the mutagenic oxidised nucleotide 8-oxo-dGTP . This implies that phenotypic markers give a more precise view of mutation rate than previously believed: having accounted for other known factors affecting mutation rate , controlling for population density can reduce variation in mutation-rate estimates by 93% . Widespread DAMP , which we manipulate genetically in disparate organisms , also provides a novel trait to use in the fight against the evolution of antimicrobial resistance . Such a prevalent environmental association and conserved mechanism suggest that mutation has varied plastically with population density since the early origins of life .
The probability of spontaneous genetic mutations occurring during replication evolves among organisms [1] . This mutation rate can also vary at a particular site in a particular genotype , dependent upon the environment [2] . Specifically , mutation rate can increase with endogenous and exogenous factors [3] . Indeed , any factor that affects the balance between mutagenesis and DNA repair can modify the mutation rate . These include intracellular nucleotide pools [4] , organism age [5] , and factors affecting the expression [6] and stochastic presence or absence [7] of low copy number repair proteins . Where such mutation/repair-balance factors depend on the environment , the result is mutation-rate plasticity . Plastic mutation rates have been most thoroughly addressed for stress-induced mutagenesis . This may involve the induction of error-prone polymerases , for instance , in the E . coli SOS response [8] . We have recently identified a novel mode of mutation-rate plasticity in response to population density in E . coli . This plasticity does not have any very obvious association with stress—the densest populations , experiencing the most competition , show the lowest mutation rates [9] . Understanding mutation-rate plasticity is hampered by the difficulty of accurately measuring any mutation rate . Spontaneous mutation rates have long been estimated in microbes using counts of cells gaining a phenotypic marker of mutation in environments lacking selection for that marker: the “fluctuation test” created by Luria and Delbrück in 1943 [10] . Alternative approaches to measuring rates of mutation in the absence of selection , such as accumulation of mutations through many population bottlenecks [11] , directly comparing genome sequences of parents and offspring [12] , or targeted population sequencing [13] , are much more laborious and thus poorly suited to potentially dynamic responses . Therefore , well-conducted fluctuation tests [14] , remain the most appropriate tool to assay environmental dependence in mutation rates . Population density affects many traits , particularly in microbes [15] . Its association with mutation rate has great potential to affect evolutionary trajectories [16] in ways relevant to the evolution of antimicrobial resistance [9] . However , thus far , this plasticity associated with population density is poorly understood: Its prevalence across domains of life is unknown . Whether it varies among organisms , enabling its evolution , remains to be tested . A little is known about the relationship of mutation rate with density perception in 1 organism [17] , but the required downstream mechanisms of mutation or repair remain uncharacterised . Here , we address each of these issues . We demonstrate that there is indeed density-associated mutation-rate plasticity ( DAMP ) across domains of life: high population density is associated with low mutation rates . DAMP differs between closely related organisms , indicating that this trait does indeed evolve . Strikingly , the same mutation avoidance mechanism is required to modulate mutation rate in response to population density in both prokaryotes and eukaryotes .
In order to test the nature and prevalence of mutation-rate plasticity , we considered over 70 years of published mutation rates estimated by fluctuation test . We collated 474 individual mutation-rate estimates from 68 independent studies that conducted fluctuation tests in organisms of 26 species from all domains of life and viruses , in which the final population density ( D ) an organism’s population reached was either reported or could be obtained from the original authors ( Fig 1 and S1 Table ) . There is a clear negative association between mutation rate and D ( Spearman’s ρ = −0 . 66 ) , spanning more than 6 orders of magnitude in both measures . This association potentially involves both between- and within-organism variation in mutation rate . We therefore analysed this relationship using a linear mixed-effects model ( Model S-I in S1 Text ) , accounting for various features of the original experiments ( organism , culture media , phenotypic marker , publication , and phylogenetic relationships among organisms ) . We find that , having taken all these factors into account , changes in D explain 93% of variation in published mutation-rate estimates ( N = 474 , LR1 = 22 , P = 2 . 9×10−6; Model S-I in S1 Text ) . The model identifies substantial variation between-organisms in a negative within-organism association of mutation rate with D ( slope varies from −0 . 46 to −0 . 98; S2 Fig ) . The average slope across organisms is −0 . 67 ( −0 . 89 to −0 . 48 CI ) , meaning that mutation rate doubles with a 64% reduction in D ( 54% to 76% CI ) , quantitatively similar to the plasticity reported for E . coli B strains [9] , in which the figure is 77% ( 61% to 96% CI ) . Despite the striking association shown in Fig 1 , the relationship could originate in various processes , including , but not limited to , mutation-rate plasticity . We consider several hypotheses: 1 . Technical bias: ( 1a ) The same estimate of final population size ( Nt ) is typically used to calculate both D and the mutation rate . Therefore , any error in Nt could itself lead to a negative association between mutation rate and D . ( 1b ) Fluctuation tests typically assume that the phenotypic markers used are selectively neutral; however , in practice , this is not always the case . If cultures grown to higher D also , typically , go through more generations , a systematic tendency towards phenotypic markers being costly could lead to underestimated mutation rates , specifically at high D . 2 . Reporting bias: there could be an underrepresentation of reported low mutation rates at low D and high mutation rates at high D . This is expected because standard volume microbial cultures with low D may not have sufficient mutational events at marker loci to achieve good estimates of low mutation rates . Similarly , in dense populations , high mutation rates can produce more mutants than it is practical to count . 3 . DAMP: the relationship in Fig 1 is consistent with DAMP across domains of life . However , the data in Fig 1 comes from diverse studies using very different experimental and analytical set-ups . It remains for us to test the association at different marker loci in different organisms within a single experimental and analytical framework . To test these hypotheses , we focused on the 2 most-diverged genetic model organisms in Fig 1: the bacterium E . coli ( strain MG1655 ) and the eukaryotic yeast S . cerevisiae ( strains S288C , BY4742 and Sigma1278b ) . Using fluctuation tests , we estimated mutation rates in batch cultures at 2 marker loci in each organism: rpoB and gyrA in E . coli and 25S ribosomal proteins and URA3 in S . cerevisiae ( see Materials and Methods ) . These confer resistance to antibiotics rifampicin , nalidixic acid , hygromycin B , and 5-Fluoro-orotic acid ( 5-FOA ) , respectively . In each case , we varied culture volume and added nutrients to give different D ( see Materials and Methods ) . To test hypothesis 1a ( correlation of errors ) , we estimated Nt by 2 independent methods for each organism: an ATP-based luminescence assay and colony-forming units ( CFU ) for bacteria , haemocytometer cell counts ( CC ) and CFU for yeast . Using CFU to estimate mutation rate ( typical in Fig 1 ) and any of the 3 methods to estimate D ( Fig 2 and S4 Fig ) , we find significant variation of mutation rates with changing D: mutation rate varies from 5-fold to 23-fold across both loci in all organisms , where mutation rate is lower at high D ( Fig 2 ) . This refutes hypothesis 1a that the broad association between population density and mutation rate ( Fig 1 ) is caused by a correlation of errors . To test hypothesis 1b ( costs of marker mutations ) for the association between population density and mutation rate , we first considered the fitness effects of resistance mutations . Others have found small negative effects ( 12% on average in Pseudomonas aeruginosa [18] ) and sometimes positive effects on fitness of mutations at the rpoB locus considered in Fig 2A [19 , 20] . Consistent with this , the fitness effects of mutations in our experiments are , on average , close to neutral ( S5 Fig ) and , where present , are similar across population densities ( S6 Fig ) . Secondly , we reanalysed the data in Fig 2 assuming different average fitness effects of resistance mutations . We find that , even assuming large fitness costs ( >50% ) , there is only a small flattening of the negative slope ( S5 Fig ) . Thus , it is highly unlikely that the differential growth of resistant strains is responsible for the slopes seen in Fig 2 , refuting hypothesis 1b that the association in Fig 1 may be accounted for by technical bias driven by differential growth of mutant strains . Typically , mutation rates at which the estimated number of mutational events per culture ( m ) is either too low ( <0 . 3 ) or too high ( >30 ) are excluded and may not be reported [14] . To test hypothesis 2 ( reporting bias ) we kept a close account of all estimates . Those 34 estimates that would typically be excluded are shown as open symbols in Fig 2 . As expected , these points fall in either the lower left or upper right of the data . Nonetheless , the association between D and mutation rate is similar and highly significant with or without these data ( −0 . 68 [−0 . 80 to −0 . 56 CI] versus −0 . 68 [−0 . 79 to −0 . 56 CI] in E . coli and , on average , −0 . 65 [−0 . 79 to −0 . 52 CI] versus −0 . 48 [−0 . 63 to −0 . 33] in S . cerevisiae ) , refuting hypothesis 2—that the negative association between population density and mutation rate in Fig 1 is caused by reporting bias . In contrast , our findings from different loci and different organisms within a single experimental and analytical framework are consistent with the pattern across the literature ( Fig 1 and S7 Fig ) , strongly supporting hypothesis 3—that there is negative DAMP across the domains of life . Our assays in S . cerevisiae identify significant variation in DAMP slope among strains ( Fig 2B ) : the slope for S288C and Sigma1278b is −0 . 98 ( −1 . 2 to −0 . 73 CI ) , whereas in BY4742 , it is −0 . 32 ( −0 . 40 to −0 . 25 CI ) . This limited difference in plasticity suggests that this trait may evolve . To investigate the extent of interorganism variation in DAMP , we tested another model organism , much more closely related to E . coli: P . aeruginosa PAO1 ( both are gram-negative gamma Proteobacteria ) . We find P . aeruginosa has a greatly reduced DAMP slope relative to E . coli , not significantly different from 0 ( Fig 2A and S4A Fig , slope estimate −0 . 15 [−0 . 40 to +0 . 095 CI] , Model S-II in S1 Text ) . This indicates that , while DAMP is very widespread , it has evolved among closely related organisms . Diverse mechanisms , some broadly conserved , could in principle be modulated to give DAMP . These include polymerases used for DNA replication and repair , some of which are more error-prone than others [21] , and systems that repair mutational mismatches [22] or remove mutagenic nucleotides before they can be incorporated into DNA [23] . To identify mechanisms by which the observed DAMP occurs , we tested E . coli strains deleted for genes involved in these processes ( S2 Table ) . A strain lacking the error-prone polymerase Pol IV ( encoded by dinB ) , implicated in stress-induced mutagenesis [8] , displays DAMP ( Fig 3; −1 . 1 [−1 . 3 to −1 . 00 CI] ) . E . coli’s methyl-directed DNA mismatch repair ( MMR ) system was hypothesised to be involved in DAMP [9] . Despite >100-fold increases in mutation rates , strains lacking MMR proteins ( MutH , MutL , and MutS ) still display DAMP , albeit with a less steep slope ( Fig 3; −0 . 20 [−0 . 30 to −0 . 11 CI] ) . We also considered other systems potentially associated with DAMP: a strain lacking MetI ( Fig 3 ) , a transporter protein responsible for feeding methionine to the activated methyl cycle , which was previously implicated in DAMP [9] . Despite growing over only a relatively narrow range of densities , the metI deletant shows DAMP with a slope indistinguishable from the dinB deletant , as does a strain lacking Dam methylase , required for MMR to identify the ‘correct’ DNA strand ( Fig 3 ) . Finally , a strain lacking Endonuclease VIII ( encoded by nei in E . coli ) also shows DAMP , with the greatest slope among all these strains ( Fig 3 , −1 . 5 [−1 . 8 to −1 . 3 CI] ) . In contrast to the systems tested above , mutT deletion removed the dependence of mutation rate on D ( Fig 4A and S8A Fig , likelihood ratio test of slope N = 63 , LR1 = 1 . 5; P = 0 . 22 , Model S-VIII in S1 Text ) . MutT is the mutation-avoidance component of the 8-Hydroxyguanine ( GO ) system , protecting cells from mutagenic effects of damaged nucleotides [23] . Specifically , the MutT Nudix hydrolase removes 8-oxo-dGTP from the free-nucleotide pool . This prevents AT to GC transversions , which occur when 8-oxo-dGTP is incorporated in place of a thymidine triphosphate nucleotide during DNA synthesis [25] . The GO system also includes mutation-correction proteins MutM and MutY , which target 8-oxo-dGTP once incorporated in DNA [23] . Strains lacking either protein display DAMP ( Fig 4B and S8B Fig; Model S-X in S1 Text ) . Thus , while neither error-prone polymerase Pol IV nor the MMR system with Dam methylase is necessary ( Fig 3 ) , DAMP in E . coli ( Fig 2A ) requires scavenging the oxidised nucleotide 8-oxo-dGTP from the cellular pool . Eukaryotes possess more , and more diverse , genes and systems for mutation avoidance and correction than bacteria . Nonetheless , the mutation-avoidance mechanism required for DAMP in E . coli is conserved across domains of life [26] . PCD1 encodes the yeast 8-oxo-dGTPase functionally homologous to bacterial MutT [27] . We tested PCD1-Δ strains in 2 different S . cerevisiae backgrounds ( BY4742 and Sigma1278b ) . As in E . coli ΔmutT , mutation rates are greatly increased in PCD1-Δ , with no evidence of DAMP ( Fig 4C and S8C Fig likelihood ratio test of slope N = 42 , LR1 = 0 . 18 , P = 0 . 67 , Model S-XII in S1 Text ) . In contrast , MLH1 , a yeast gene homologous to E . coli MMR gene mutL , displays DAMP ( Fig 4D and S8D Fig; Model S-XIV in S1 Text ) . Therefore , DAMP not only occurs across domains of life , but , in the bacteria and eukaryotes tested , requires the same specific mutation-avoidance mechanism .
The negative association between published mutation rates and population density ( D , Fig 1 ) is remarkably tight . Much of the between-organism variation in microbial mutation rate is associated with variation in genome size , in which larger genomes tend to have smaller per-base-pair mutation rates [28] . We account for this in our analysis of the data in Fig 1 by allowing mutation rates to vary among organisms and accounting for their phylogeny ( explicitly including a typical genome size for each organism makes little difference to this analysis , Model S-I in S1 Text ) . However , the within-organism variation in mutation rate associated with D is strong enough to explain radical differences in mutation-rate estimates for the same organism within and between laboratories , without assuming any inconsistency in the fluctuation test itself . For instance , the estimates for Salmonella at the bottom of Fig 1 vary by over an order of magnitude but diverge little from a negatively sloping straight line . This suggests that , once population density is taken into account , fluctuation tests can give a more precise estimate of mutation rate than previously believed . Fluctuation tests , however , have drawbacks , such as the possibility of unanticipated selection on mutant cells in a supposedly nonselective environment [19] . We avoided that specific issue here by using short incubation times and making estimates at multiple loci . The more general drawback of using fluctuation tests for considering environmental correlates ( shared with most other methods for estimating mutation rates ) is that they average across the time-varying environment of batch culture . Thus almost any environmental variable , including population density , has no fixed value and is itself associated with many other characteristics of the culture , both environmental and organismal ( e . g . , times spent in different phases of the culture cycle ) . Thus , while we have demonstrated an association of mutation rate with final population density D ( Fig 2 ) and the dependence of that association on a downstream mechanism ( Fig 4 ) , the link between 8-oxo-dGTPase and D remains unclear . Nonetheless , 2 important things can be said about this link . First , DAMP is observed across a wide range of environmental conditions and organisms ( Fig 1 and Fig 2 ) , which argues that features of particular environments , such as the starting nutrient concentration , as manipulated here , are unlikely to provide a general link between D and mutation rate . Second , we have previously demonstrated that in 1 organism , E . coli , cell–cell interactions are involved in DAMP and deletion of a quorum-sensing gene ( luxS ) breaks the association between D and mutation rate [9] , demonstrating that population interactions can be important for DAMP . Such quorum-sensing mechanisms occur widely ( even in some phages [29] ) . Future work , therefore , needs to focus on asking whether DAMP is associated with particular environmental molecules . Mutation rate is a central population genetic parameter . Across organisms , it is negatively associated with the other central population genetic parameter , effective population size ( Ne ) [1] . In our experiments , despite 2 orders of magnitude variation in each measure , we find no consistent association between mutation rate and Ne ( S9 Fig ) . This is unsurprising because the proposed reason for the negative association across organisms is that selection for replication fidelity is more efficient at higher Ne , meaning that , over the long term , average mutation rates evolve to be lower at higher Ne [1] . In our short-term experiments , there is little opportunity for such evolutionary change to occur , so we do not see this association . Nonetheless , this reinforces the clear distinction between within-organism plasticity and among-organism variation in mutation rate . Both have shaped mutation rates in the published literature ( Fig 1 ) , and we are able to separate the 2 , both statistically and by focused experiments ( Fig 2 ) . There may be links between the causes of among-organism variation and within-organism plasticity in mutation rate , for instance , in the differing opportunities for selection on replication fidelity in polymerases expressed in common or rare environmental conditions [30] . However , the evolutionary causes and effects of within-organism plasticity in mutation rate in general , and DAMP in particular , need further investigation . The evolutionary causes of plasticity in mutation rates need not be adaptive [30] . Nonetheless , mutation is an evolutionary mechanism , so any plastic variation in mutation rates will have consequences for evolutionary trajectories [31] . What the evolutionary consequences might be depend on how mutation rate associates with the environment . For evolutionary computing , in which mutation rate is controlled , understanding the effect of that control is an important area of research [32] . In biology , constitutively high mutation rates can evolve under specific circumstances [33 , 34] , but incur the costs of many , typically deleterious , mutations . If plasticity is such that mutation rate is inversely related to absolute organismal fitness , then organisms may benefit from a high mutation-supply rate without paying the full evolutionary cost of a constitutively raised mutation rate , as seen in mathematical studies of evolutionary systems [16] and population genetic models [35] . In some circumstances , DAMP can result in such a negative association of mutation rate with fitness [9] , but the evolutionary effects of this remain to be tested . The probability of a particular mutational event occurring ( e . g . , the emergence of spontaneous antibiotic resistance ) might be expected to increase with D , as denser populations , containing more cells , will have had more opportunity for mutation . But for the DAMP described here , this increase is offset by a reduction in the mutation rate . This offsetting means that for organisms with DAMP , numbers of mutational events per space and time vary much less with Nt than expected from a fixed mutation rate per generation ( S10 Fig ) . Population genetic models typically consider mutations per replication to be constant for an organism . However , we find that the approximate constant is the number of mutational events per space and time ( S10 Fig ) . This is consistent with observations of invariant numbers of mutations per time in Mycobacterium infections [36] and , indeed , in human somatic [37] and germ cells [6] . Both the occurrence across domains of life ( Fig 2 ) and the conserved mutation avoidance mechanism required ( Fig 4 ) point to an ancient evolutionary origin for DAMP . Furthermore , Fig 1 suggests that DAMP also occurs in viruses and bacteriophage . Any variation in mutation rates in viruses and phage lacking mutation-avoidance or -correction mechanisms must be mediated by the host environment . Consistent with this , we see different mutation rates , but similar DAMP , for the same RNA virus in different host cells ( S11 Fig reanalysed from [38] ) . DAMP itself , therefore , seems closely related to basic processes of replication common to all organisms . Nonetheless , our findings are limited to organisms in which it is possible to assay mutation rate by fluctuation tests . This excludes multicellular eukaryotes , so how our findings might apply to them is unclear . Recent findings of variation in mitochondrial mutation rates at different population sizes and densities of the nematode Caenorhabditis elegans highlight the challenge of separating out what population density could mean at the organism , tissue , cellular , and subcellular ( e . g . , mitochondrial ) levels [39] . Even so , if it were possible to manipulate microbial DAMP clinically as well as genetically ( Fig 4 ) , for instance as a strategy to slow the rate at which antibiotic resistance arises [40] , that could be applicable across the breadth of microbes , including pathogenic viruses .
See S2 Table . We used MilliQ water for all media . Tetrazolium arabinose agar ( TA ) , Davis minimal medium ( DM ) , and M9 minimal medium were prepared according to [41] . Luria-Bertani medium ( LB ) , yeast extract peptone medium ( YP ) , and yeast nitrogen base ( YNB ) were prepared according to manufacturers’ instructions . Magnesium sulphate heptahydrate , thiamine hydrochloride , carbon source ( 3 g/l L-arabinose or various concentrations of D-glucose ) , and 2 , 3 , 5-triphenyltetrazolium chloride ( Sigma T8877 ) were sterile filtered and added to a cooled medium . Selective TA medium was supplemented with freshly prepared rifampicin ( 50 μg/ml ) or nalidixic acid ( 30 μg/ml ) . Selective YP medium was supplemented with freshly prepared 5-FOA ( 1 , 000 μg/ml ) or hygromycin B ( 300 μg/ml ) . For all cell dilutions , sterile saline ( 8 . 5 g/l NaCl ) was used . All media were solidified as necessary with 15 g/l of agar ( Difco ) . We did fluctuation tests with E . coli and P . aeruginosa as explained in [9] . In short , strains were first inoculated from frozen stock and grown in liquid LB medium at 37°C and then transferred to nonselective liquid DM ( for E . coli ) or M9 ( for P . aeruginosa ) , supplemented with a particular concentration of glucose ( 25–300 mgl−1 ) , and allowed to grow overnight shaking at 37°C . E . coli and P . aeruginosa were again diluted into fresh DM or M9 medium , respectively , giving the initial population size ( N0 ) of around 10 , 000 ( range 2 . 5×102 to 1 . 3×105 ) and 5 , 000 ( range 2 . 5×103 to 1 . 2×104 ) , respectively . Various volumes ( 0 . 5–10 ml ) of parallel cultures were grown to saturation for 24 hours at 37°C in 96-deep-well plates or 50 ml falcon tubes . The position of each culture on a 96-well plate was chosen randomly . Nt of each culture was determined by 2 independent techniques . Nt was determined by CFU in which appropriate dilution was plated on a solid nonselective TA medium . Estimates of Nt using net luminescence were determined using a Promega GloMax luminometer and the Promega Bac-Titer Glo kit , according to manufacturer's instructions . We measured the luminescence of each culture 0 . 5 seconds and 510 seconds after adding the Bac-Titer Glo reagent and calculated net luminescence as LUM = LUM510s − LUM0 . 5s . Each estimate of Nt is an average of 3 cultures . Evaporation ( routinely monitored by weighing the plate before and after 24 hours of incubation ) was accounted for in the Nt value determined by CFU and was also used in statistical modelling as a variance covariate . We obtained the observed number of mutants resistant to rifampicin or nalidixic acid , r , by plating the entirety of remaining cultures onto solid selective TA medium that allows spontaneous mutants to form colonies . Plates were incubated at 37°C , and mutants were counted at the earliest possible time after plating . For rifampicin plates , this was 44–48 hours , when nalidixic acid was used , the incubation time was 68–72 hours . We did fluctuation tests with yeast in a similar way to fluctuation tests with bacteria ( see above ) . Strains were inoculated from frozen stock in liquid YP medium with 20 mg/ml of glucose at 30°C ( 200 rpm ) and then transferred to nonselective liquid YNB medium supplemented with a particular percentage of YP ( v/v ) and glucose , except for S288C in which YP was not added . We then allowed cultures to grow overnight at 30°C ( 200 rpm ) . Overnight cultures were again diluted into fresh medium giving N0 of around 5 , 000 per parallel culture ( range 5×102 to 5 . 1×104 ) . Various volumes of parallel cultures ( 0 . 35–10 ml ) were grown in yeast nitrogen base with 25–8 , 000 mgl−1 glucose and 0%–7% v/v YP in 96-deep-well plates or in 50 ml falcon tubes to saturation for 48 or 72 hours at 30°C ( 200 rpm ) . We positioned each culture on the plate randomly . Nt was determined by CFU , in which an appropriate dilution was plated on solid nonselective YP medium . Nt determined with haemocytometer ( Cellometer Auto M10 –Nexcelom ) ( CC ) was done according to manufacturer’s instructions . Nt was calculated with 3 cultures per mutation-rate estimate , in which for each culture , CFU and CC were determined . Evaporation was accounted for in the Nt value determined by CFU and also used in statistical modelling as a variance covariate . We obtained the observed number of mutants resistant to 5-FOA or hygromycin B , r , by plating the entirety of remaining cultures onto solid selective YP medium . Plates were incubated at 30°C , and mutants were counted at the earliest possible time after plating , for both markers that was 68–72 hours . For Figs 2A , 2B , 3 , 4A , 4B , 4C and 4D , we used 21 , 14 , 14 , 8 , 5 , 5 , and 3 independent experimental blocks , respectively , carried out on different days . Within an experimental block , multiple 96-well plates , or groups of falcon tubes , were used . Any individual mutation-rate estimate requires multiple parallel cultures , which were all carried out on a particular plate , or group of falcon tubes . For Figs 2A , 2B , 3 , 4A , 4B , 4C and 4D , the median number of parallel cultures used ( with interquartile range ) was 16 ( 15–16 ) , 16 ( 16–16 ) , 16 ( 16–16 ) , 16 ( 16–16 ) , 16 ( 16–16 ) , 16 ( 16–16 ) , and 16 ( 15–16 ) , respectively . To calculate m from the observed number of mutants , we employed the Ma-Sandri-Sarkar maximum-likelihood method implemented by the FALCOR web tool [42] or rSalvador [43 , 44] . The mutation rate per cell per generation is calculated as m divided by Nt . The median ( with interquartile range ) of the coefficient of variation for Nt estimated with CFU and ATP-based luminescence assay is 15 . 9% ( 9 . 6%–24 . 8% ) and 10 . 9% ( 6 . 9%–18 . 4% ) , respectively . Ne is calculated as the harmonic mean of the population size across generations . All statistical analysis was executed in R v3 . 2 . 4 and v3 . 3 . 1 , respectively , when using spaMM v 1 . 7 . 2 ( Model S-I , S1 Text ) and nlme v3 . 1 ( Model S-II to S-XVIII , S1 Text ) packages for linear mixed effects modelling [45 , 46] . This enabled the inclusion within the same model of experimental factors ( fixed effects ) , blocking effects ( random effects ) , and factors affecting variance ( heteroscedasticity ) as described in S1 Text . In all cases , the log2 mutation rate was used . Raw data ( S1 Data , described in S4 Table ) and R code ( S1 Code ) are provided , sufficient to reproduce Figs 1–4 , S1 , S3 and S4 Figs , S6 Fig and S8–S11 Figs . E . coli genomes were sequenced with the Illumina HiSeq2500 platform using 2 x 250 bp paired-end reads . Sequencing and initial read quality checking were provided by MicrobesNG ( http://www . microbesng . uk ) and deposited at the European Nucleotide Archive ( accession number ERP024110 , http://www . ebi . ac . uk/ena/data/view/ERP024110 ) . Strains derived from MG1655 and from the Keio collection were aligned to the E . coli str . K-12 substr . MG1655 ( NC_000913 . 3 ) and E . coli BW25113 ( NZ_CP009273 . 1 ) genomes , respectively . Mutations ( i . e . , single nucleotide substitutions , small and large indels , and copy number variants ) were predicted using breseq-0 . 27 . 2 using the default settings [47] . We identified studies that used Luria-Delbrück fluctuation tests for estimating mutation rates . We considered all papers citing the original reference [10] and further searched the Google Scholar and Web of Science databases with keywords “mutation rate” , “Luria Delbruck” , “fluctuation test” , and “fluctuation assay”; we also considered papers cited by papers identified in this way . We collected mutation rate estimations from studies spanning over 70 years , starting with Luria and Delbrück’s pioneering paper in 1943 outlining the fluctuation test [10] . In all , we collected 474 mutation rate estimations from 68 separate studies , covering 26 different organisms from across domains of life ( Archaea , Bacteria , Eukaryota ) and viruses . From these studies , we recorded the mutation rate estimation , the estimator used for calculating mutation rate , the D of parallel cultures , identity of the nonselective medium , the organism studied , the selective marker used and its concentration , and the study the estimate came from . We excluded estimates that ( i ) involved microorganisms cultured in intentionally selective conditions , ( ii ) used genetically manipulated or mutator strains , or ( iii ) did not plate the entire culture volume onto the selective media . Any of this information that was not included in the published article was collected via a direct communication with the corresponding author . Where the number of observed mutants per plate was available and the estimator was not the Ma-Sandri-Sarkar maximum-likelihood method , we recalculated the mutation rate using this method implemented by the FALCOR web tool [42] . When only the proportion of plates without mutants was available , we recalculated the mutation rate using the P0 method [14] , implemented by equation −ln ( P0 ) /Nt in which P0 is the proportion of plates containing no mutant colonies . For viral mutation rates , we recorded the mutation rate as substitutions per strand copying . Where the published mutation rates were not in this format , these were converted using equation 10 in [48] . All these points are highlighted in the column ‘recalculation’ as ‘yes’ in S1 Data ( see S4 Table ) . To take account of the fact that organisms may show similarity ( including in mutation rates ) through common ancestry , we accounted for phylogenetic relatedness using a correlation matrix within our Model S-I ( S1 Text ) . This matrix was derived from a phylogeny ( S1 Fig ) constructed from a combination of 3 published phylogenies . All the bacteria and archaea came from the ‘All-Species Living Tree’ Project ( LTP ) [49] version LTPs123 , and this phylogeny was used to add other organisms . S . cerevisiae was taken from [50] , and viruses were taken from [51] . Branch lengths for S . cerevisiae and the viruses were scaled to correspond to those in the LTP tree as follows . For S . cerevisiae , average branch lengths from the tips to the last common ancestor of the archaea Halobacterium and Sulfolobus were compared for the LTP and Lane and Darst [50] trees . The ratio between them was applied to the branch length of S . cerevisiae in Lane and Darst phylogeny [50] before adding it to the LTP tree at the branch point of the bacteria and archaea . For the viruses , 4 common tips from both trees ( P . aeruginosa , Thermus thermophilus , S . cerevisiae , and Halobacterium ) were selected , and the distance of each to a shared common ancestor ( P . aeruginosa with T . thermophilus and S . cerevisiae with Halobacterium ) were plotted against each other . A straight line was then fitted through the origin , and the gradient of that line was used to correct the branch length of the viruses in the Nasir and Caetano-Anollés tree [51] before being added to the combined LTP/S . cerevisiae tree at the branch point of the 3 domains . Some organisms in our analysis were not present in these phylogenies . These were as follows: separate serovars of Salmonella enterica ( serovars Typhimurium and Enteritidis ) , which we treated as subspecies of S . enterica ( indica and enterica , present in the tree ) ; Vesicular stomatitis virus and Measles virus , which were combined into their common order of Mononegavirales; and Bacteriophage ΦX174 , which was positioned at Bacteriophage M13 . | Spontaneous mutations fuel evolution , but the rate at which they occur can vary for a particular organism depending on its environment—a phenomenon known as mutation-rate plasticity . For microbes growing in liquid , the density to which a population can grow is a key feature of the environment . We find that organisms’ mutation rates are associated with the density to which they grow , such that higher population densities are associated with lower mutation rates . Initially we identify this density-associated mutation-rate plasticity ( DAMP ) in data culled from the published literature: beyond well-known patterns of mutation-rate variation among organisms , we see substantial variation within diverse organisms , the large majority of which is associated with population densities . We test this association in the laboratory , finding DAMP at different sites in the genomes of both bacteria ( Escherichia coli ) and eukaryotes ( the yeast , Saccharomyces cerevisiae ) . In each case , DAMP requires a protein that avoids mutation by cleaning cells of oxidatively damaged guanine nucleotides ( MutT in E . coli and Pcd1 in yeast ) . In our assays , DAMP results in a lower probability of seeing the evolution of antibiotic resistance at higher population densities . We anticipate that DAMP affects the course of evolution more generally and that understanding its causes and effects will help us to understand and control evolutionary trajectories . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"taxonomy",
"organismal",
"evolution",
"microbial",
"mutation",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"microbiology",
"pseudomonas",
"aeruginosa",
"fungi",
"phylogenetics",
"model",
"organisms",
"data",
"manage... | 2017 | Spontaneous mutation rate is a plastic trait associated with population density across domains of life |
Axon pathfinding and synapse formation rely on precise spatiotemporal localization of guidance receptors . However , little is known about the neuron-specific intracellular trafficking mechanisms that underlie the sorting and activity of these receptors . Here we show that loss of the neuron-specific v-ATPase subunit a1 leads to progressive endosomal guidance receptor accumulations after neuronal differentiation . In the embryo and in adult photoreceptors , these accumulations occur after axon pathfinding and synapse formation is complete . In contrast , receptor missorting occurs sufficiently early in neurons of the adult central nervous system to cause connectivity defects . An increase of guidance receptors , but not of membrane proteins without signaling function , causes specific gain-of-function phenotypes . A point mutant that promotes sorting but prevents degradation reveals spatiotemporally specific guidance receptor turnover and accelerates developmental defects in photoreceptors and embryonic motor neurons . Our findings indicate that a neuron-specific endolysosomal degradation mechanism is part of the cell biological machinery that regulates guidance receptor turnover and signaling .
Axon guidance , target selection , and synapse formation determine the neuronal connectivity of the brain and rely on the spatially and temporally controlled localization of guidance receptors [1] , [2] . The dynamic localization of these receptors is at least partly regulated at the level of vesicular membrane trafficking through the secretory pathway , endosomal recycling , and endolysosomal degradation [3] , [4] . Endosomal routing is also a means of receptor activation and inactivation: receptors may signal from the plasma membrane or endosomal compartments , and receptor signaling can be turned off by endolysosomal degradation [4] , [5] . A growth cone may reuse a number of guidance receptors through cycles of endo- and exocytosis . Alternatively , constitutive synthesis and degradation may provide a constant stream of receptors that can be sorted to exert spatiotemporally defined roles . However , for most cell types it is unknown which mode of receptor trafficking prevails to regulate receptor ( de ) activation during development and function . Similarly , surprisingly little is known about the neuron-specific molecular mechanisms that underlie guidance receptor trafficking for either strategy during brain wiring . The Drosophila nervous system has proven to be a powerful system for the characterization of the molecules that guide axons along their pathways and enable correct target selection [6]–[9] . The visual system has been particularly useful , because both photoreceptors and visual interneurons are dispensable for viability and are easily genetically manipulated in otherwise wild-type flies . Genetic screens based on methods that generate mutant visual neurons in heterozygous flies led to the discovery of numerous important secreted and membrane-associated guidance molecules and receptors , their regulators , and signal-transducing proteins [10]–[12] . Amongst the many known guidance molecules and receptors implicated in visual system development are the cadherins N-Cadherin ( N-Cad ) and Flamingo ( Fmi ) [10] , [13] , [14] , the tyrosine phosphatases DPTP69D and Dlar [15]–[17] , and the immunoglobulin superfamily cell adhesion molecules Fasciclin 2 ( Fas2 ) and Roughest ( Rst ) [18]–[20] . Although spatiotemporally dynamic expression has been shown for most of these receptors , almost nothing is known about their intracellular trafficking , activation , turnover , and degradation . Genetic mosaic screens in the Drosophila visual system have also led to the discovery of numerous mutants with membrane and organelle trafficking defects [21]–[23] . The Drosophila gene v100 was originally identified in a screen for mutants that affect synapse formation , specification , or function [23] , [24] . v100 encodes subunit a1 of the V0 complex , the membrane-bound sector of the two-sector vesicular ( v- ) ATPase [25] , [26] . V100 is a neuron-specific subunit of the v-ATPase that is required for neurotransmitter release [24] and provides a neuronal degradation mechanism in photoreceptors . This degradation mechanism is created by a dual function: V100 sorts vesicles into endosomal compartments and subsequently acidifies degradative compartments as part of the v-ATPase holoenzyme . Loss of v100-dependent degradation leads to adult-onset degeneration , but no developmental or synaptic specification defects in photoreceptors [27] . Similarly , v100 mutant embryos exhibit normal nervous system morphology [24] . In this study , we report that a neuron-specific , v100-dependent membrane sorting and degradation mechanism is required for brain wiring in Drosophila . Loss of v100 results in missorting and intracellular accumulation of guidance receptors at the time and place where they are subject to active turnover . These accumulations precede axon mistargeting . We further show that guidance receptors aggregate on endolysosomal compartments and cause exacerbated gain-of-function phenotypes in v100 mutant photoreceptors as well as in the embryonic nervous system . Our findings suggest that continuous receptor turnover and degradation by a neuron-specific mechanism is a general mode of guidance receptor trafficking . Our data further suggest that a v100-dependent neuronal degradation mechanism underlies a regulatory strategy that depends on a constant turnover of receptors that can be sorted to exert spatiotemporally defined roles .
Several genetic mosaic methods have been developed that render visual system neurons homozygous mutant in heterozygous flies [11] , [12] , [28] . In our previous studies of v100 function in photoreceptors , we used the “ey3 . 5Flp” system developed by Salecker and colleagues , which renders only photoreceptors mutant [28] , [29] . Our studies on v100 in photoreceptors uncovered defects in neurotransmission [24] and neurodegeneration [27] , but no developmental defects . In contrast to this photoreceptor-specific method , the original eyFLP system [11] generates thousands of homozygous mutant neurons in the central nervous system ( CNS ) in addition to photoreceptors . Importantly , eyFLP affects only CNS neurons of the visual and olfactory systems that are not required for viability of the organism under laboratory conditions and thereby allows the investigation of v100 mutant central brain neurons in a living fly ( Figure S1 ) [29] . Surprisingly , we found severe axon pathfinding and targeting defects in these eyFLP v100 brains that were not present in our previous experiments when only photoreceptors were mutant ( Figure 1A–1D ) . Further analysis of the eyFLP v100 brain with the active zone marker Brp ( nc82 ) revealed severe structural defects in the arrangement of synaptic neuropils resulting from defective axon pathfinding during pupal development prior to synaptogenesis ( Figure 1E and 1F ) . For clarity we will hereafter refer to the photoreceptor-specific system as eyFLPPRonly and the original eyFLP system that additionally renders CNS neurons mutant as eyFLPCNS . Neuron-specific expression of v100 with the elavc155-Gal4 driver is sufficient to rescue viability in Drosophila [24] . However , Peri and Nusslein-Volhard recently reported a function for the zebrafish ortholog of v100 in phagosomal/lysosomal fusion in microglial cells [30] . The zebrafish v100 ( atp6v0a1 ) is a true ortholog because the protein is more closely related to Drosophila V100 ( 61% identical ) and the human subunit a1 ( 82% identical ) than it is to the closest paralog in zebrafish ( V0 subunit a2 , 54% identical ) . We therefore wondered whether the developmental CNS defects described here could be attributed to a non-neuronal cell type . We analyzed v100 mutant brains rescued with only neuronal v100 expression . As shown in Figure 1D , neuronal expression of v100 rescues the wiring defect of an eyFLPCNS brain . Hence , v100 is required in CNS neurons for brain wiring in Drosophila . We have recently shown that V100 is expressed in the pupal and adult visual system [27] . To determine the onset of V100 expression in the developing CNS , we performed co-labeling experiments with the developing synapse marker N-Cad and the active synapse marker Synaptotagmin ( Syt ) . As shown in Figure 1G–1I , anti-V100 labeling of a larval brain hemisphere reveals strong enrichment in the synaptic neuropils of the functional larval brain ( arrows ) . In contrast , regions of neuronal and glial differentiation are labeled at only background levels , suggesting no prominent role during early brain development . However , at the time of axon targeting at 20% pupal development ( P+20% ) , V100 is strongly enriched in the developing first synaptic neuropil in the optic lobe , the lamina plexus , where axon terminals are actively sorting to generate a precise visual map ( arrows in Figure 1J–1L ) . V100 labeling at this time is most prominent in the lamina plexus , but increases in all neuropils throughout development ( Figure 1M–1O ) . Note that V100 labeling , although enriched in the synaptic neuropils , appears distinctly different from that of the synaptic vesicle marker Syt ( Figure 1M–1O ) . These data show that V100 is enriched in specific synaptic regions of the visual system prior to synaptogenesis . Taken together our data indicate that v100 plays a hitherto unrecognized developmental role in CNS neurons of the adult brain . Next , we asked whether the observed brain wiring defects are caused by early cell death . Immunolabeling of activated Caspase-3 [31] in eyFLPCNS brains reveals no significant difference in the number of cells undergoing programmed cell death between mutant and wild type during development ( Figure S2A and S2B; eyFLPCNS v100: 34±10 apoptotic cells per confocal optic lobe section; control: 31±9 ) or in 10-d-old optic lobes ( mutant neurons marked in green , control unmarked; Figure S2C ) . This is consistent with the previously documented finding of slow adult-onset degeneration in photoreceptors , which causes cells to become unhealthy long after development is complete [27] . These data indicate that the brain wiring defects are not the result of premature cell death . Our previous characterization of V100 function revealed roles in synaptic vesicle exocytosis [24] and endolysosomal degradation in neurons [27] . The brain wiring defects described in this study are unlikely to be caused by defects in neurotransmitter release , since we and others have previously shown that neuronal activity , including synaptic vesicle release , is not required for photoreceptor or optic lobe development [23] , [32] . In contrast , v100's role in neuronal endolysosomal degradation could potentially be required for development since many signaling molecules are regulated through the endolysosomal pathway . This idea raises the question how a defect in endolysosomal trafficking could lead specifically to neuronal connectivity defects in the brain without affecting earlier stages of neuronal development . Cell adhesion molecules that function as guidance receptors are key proteins directing axon pathfinding and targeting . To investigate a possible link between v100 and the observed brain wiring defects , we analyzed several guidance receptors known to play roles during optic lobe development and visual map formation in the Drosophila brain . First , we investigated the localization patterns of the five guidance receptors Dlar , N-Cad , Fmi , Fas2 , and Rst [10] , [13] , [17] , [33] , [34] in 1-d-old control and eyFLPCNS brains . All five guidance receptors exhibit a similar phenotype of aberrant accumulations in synaptic neuropils and cell bodies of the eyFLPCNS optic lobe ( Figure 2 ) . This phenotype is most pronounced for Dlar and N-Cad , whose wild-type expression patterns in the optic lobe are restricted to synaptic neuropils ( Figure 2A , 2B , 2I , and 2J ) [13] , [17] . While these findings show that guidance receptor localization is indeed affected in v100 mutant neurons , they also indicate that the underlying intracellular trafficking defect is not specific to a particular guidance receptor . Hence , our results suggest that the developmental defects are a cumulative effect of the mislocalization of many receptors . v100 mutant photoreceptors exhibit a slow accumulation of endolysosomal compartments [27] . Intracellular accumulation of guidance receptors might cause developmental defects in CNS neurons by at least two mechanisms . First , receptors might fail to be transported to the plasma membrane , leading to loss-of-function phenotypes . Second , receptors might fail to endocytose or accumulate on signaling-active endosomal compartments , leading to gain-of-function phenotypes . In contrast to such loss- or gain-of-function effects , accumulation of receptors in signaling-inactive lysosomal compartments should not lead to any receptor-specific defects . We therefore analyzed guidance receptor localization on intracellular compartments using an eyFLPCNS-based approach where only mutant cells are fluorescently marked ( mosaic analysis with a repressible cell marker [MARCM]; [35] ) . First , we confirmed that v100 mutant CNS neurons have the same endolysosomal accumulations previously described for photoreceptors [27] . As shown in Figure 3A and 3B , both the early endosomal marker 2xFYVE-GFP and the late endosomal marker Rab7 accumulate in v100 mutant neurons . 2xFYVE-GFP is a cytosolic probe that predominantly marks early endosomal compartments by associating with PI ( 3 ) P-rich membranes [36] . Note that in this experiment only 50% of CNS neurons are mutant and these cells are marked with 2xFYVE-GFP expression . As shown in Figure 3A , 2xFYVE-GFP exhibits only low levels of labeling in wild-type clones . In contrast , v100 mutant CNS neurons exhibit substantial accumulations ( arrows in Figure 3B ) . Rab7-positive compartments exhibit similar accumulations . However , PI ( 3 ) P-rich endosomal accumulations are even more apparent than Rab7 accumulations ( compare green and red labeling in Figure 3B ) . These results are consistent with our previous characterization of endolysosomal accumulations in photoreceptors and indicate that v100 mutant CNS neurons exhibit the same endolysosomal trafficking problem . As in photoreceptors , early endosomal markers are upregulated strongest [27] . Next , we analyzed subcellular guidance receptor localization . As shown in high-resolution confocal images in Figure 3C , the receptor Dlar accumulates in highly heterogeneous compartments in CNS neuronal cell bodies . In this experiment , we marked the mutant cells with the lysosomal marker lamp-GFP , a transmembrane protein that traffics through the endolysosomal pathway and is quickly degraded in wild type [37] . Co-labeling with the early endosomal marker Syx7 reveals that 68 . 7% of all Dlar accumulations are Syx7-positive ( arrows ) , but only 29 . 4% are lamp-GFP-positive . We observed similar results for all guidance receptors ( Figure 3D and 3E and data not shown ) . Amongst these receptors , Fas2 exhibited the strongest colocalization with the early endosomal marker Syx7 both in cell bodies of CNS neurons in the medulla cortex ( arrows in Figure 3D ) and at photoreceptor synapses in the lamina ( arrows in Figure 3E ) . Interestingly , our high-resolution analyses of subcellular localization revealed a pattern of increased guidance receptor accumulations on the outside of large ( up to 5 µm ) Syx7-positive compartments , as shown in Figure 3E for Fas2 . We made similar observations for guidance receptor accumulations using two additional genetic manipulations , namely , increased sorting into endosomal compartments and receptor overexpression in v100 mutant neurons , as described below . In summary , our findings indicate that guidance receptors accumulate after endocytosis in the compartments most prominently labeled by early endosomal markers . Why do eyFLPPRonly adult photoreceptors lack a developmental defect ? Photoreceptors conclude axon pathfinding less than 48 h after differentiation , while many adult CNS neurons adopt the neural fate many days before brain connectivity is established [38] . v100 mutant neurons exhibit a progressive increase of intracellular accumulations because of lack of degradation [27] . We reasoned that in CNS neurons , disruptive intracellular accumulations might occur sufficiently early during neuronal development to cause developmental defects . In contrast , in photoreceptors such defects might occur only after the critical developmental time periods of axon pathfinding and target recognition . To compare the time course of intracellular trafficking defects in eyFLPPRonly and eyFLPCNS , we investigated guidance receptor localization in developing and adult brains . As shown in Figure 4A and 4B , optic lobes of eyFLPCNS v100 brains at P+30% exhibit Dlar accumulations that are absent in eyFLPCNS control brains . To identify even small changes of Dlar levels in photoreceptors , we analyzed mutant and neighboring control terminals in MARCM clones . As shown in Figure 4C , mutant photoreceptors exhibit Dlar levels indistinguishable from control . In 1-d-old adult eyFLPCNS optic lobes , Dlar accumulations are further increased ( Figure 4D and 4E ) , while mutant photoreceptor terminals are just beginning to show receptor accumulations ( Figure 4F ) . We observed similar temporal dynamics for the other guidance receptors , although with varying onset , localization , and severity of accumulations , as discussed in the next section . Our data show that guidance receptor accumulations occur in both CNS neurons and photoreceptors . However , the photoreceptor defects are delayed and seem to occur sufficiently late to allow normal development . These observations are consistent with the idea that cell-specific axonal targeting defects depend on the dynamics of a progressive degradation and intracellular accumulation defect . Next , we tested whether the differential onset of Dlar accumulations in CNS neurons versus photoreceptors reflects a general degradation problem of transmembrane proteins that traffic through the endolysosomal system . We analyzed the time course of lamp-GFP accumulations in developing optic lobe neurons ( eyFLPCNS ) . In late third instar larvae , lamp-GFP exhibits a prominent degradation and accumulation phenotype in the v100 mutant CNS ( Figure 4G and 4H ) , while mutant photoreceptors show almost no lamp-GFP accumulations at this stage ( Figure 4I ) . However , accumulations do become apparent in photoreceptors at P+30% , i . e . , even before Dlar accumulations become discernible ( Figure 4J; compare to Figure 4C ) . In summary , accumulations of lamp-GFP , like accumulations of Dlar , reveal a progressive intracellular degradation defect that occurs earlier in mutant CNS neurons than in photoreceptors . We have previously shown that v100 mutant photoreceptors have an acidification defect as evidenced by Lysotracker labeling experiments [27] . Lysotracker is a membrane-permeable dye that accumulates in highly acidified compartments in cells , i . e . , lysosomes , late endosomes , and autophagosomes . Larval eye discs show no difference in Lysotracker uptake in mutant versus control cells , while pupal eye discs show a 50% reduction in Lysotracker signal in mutant cells [27] . To characterize the onset of acidification defects in optic lobe CNS neurons , we generated GFP-labeled v100 mutant clones as before ( eyFLPCNS MARCM ) . In control experiments , we used the same approach , except both the marked and unmarked cells were wild type . As shown in Figure 4K–4M , we found a significant reduction of Lysotracker signal in mutant CNS neurons of the third instar larva , i . e . , at the same time as when photoreceptors are differentiating and do not yet exhibit Lysotracker defects . Furthermore , the strength of the larval CNS defect is reminiscent of photoreceptors at P+40% , i . e . , approximately 2 d later . As was the case for Dlar and lamp-GFP accumulations , the observed reduction of Lysotracker-positive compartments in optic lobe CNS neurons is sufficiently early to account for the brain wiring defects . In contrast , a similar reduction of strongly acidified compartments in photoreceptors is observed only after axon pathfinding and visual map formation are concluded . Our data show that both the accumulation of membrane proteins and the loss of Lysotracker-positive degradative compartments precede the onset of developmental defects in CNS neurons . In contrast , our results argue that v100 mutant photoreceptors lack a developmental defect because endolysosomal defects are delayed . To test the causality of this correlation , we designed an experiment to accelerate v100 endolysosomal trafficking defects and assay the effect on photoreceptor development . We have previously generated a mutant version of v100 that accelerates and thereby exacerbates null mutant phenotypes by selectively rescuing the sorting of cargo into degradation-incompetent compartments . Selective rescue of the endosomal sorting function but not the acidification function of v100 is achieved by expressing the mutant v100R755A in v100 null mutant neurons . In contrast , v100R755A expression in wild-type neurons has almost no effect since the wild-type protein is present to acidify degradative compartments [27] . As shown in Figure 5A and 5B , v100R755A expression in v100 mutant photoreceptors ( eyFLPPRonly ) leads to axon targeting defects ( arrows in Figure 5A ) that are completely absent when v100R755A is expressed in wild-type neurons , consistent with our previous report that v100R755A does not act as a dominant-negative [27] . Similarly , these developmental defects are not observed in v100 null mutant photoreceptors or in mutant photoreceptors that are rescued with wild-type v100 ( Figure 1B and 1D ) . In addition , large amounts of the photoreceptor-specific transmembrane protein Chaoptin accumulate when v100R755A is expressed in mutant neurons ( arrowhead in Figure 5A ) . These data suggest that v100R755A expression in mutant neurons accelerates intracellular accumulations and causes developmental defects in photoreceptors . Next , we assessed the effect of v100R755A expression on the wiring defect in v100 mutant optic lobe CNS neurons ( Figure 5E ) . Strikingly , v100R755A causes a dramatically worse wiring defect than the null mutant and effects a total loss of recognizable neuropil structure ( Figure 5C and 5E ) . Neuronal expression of wild-type v100 fully rescues this defect ( Figure 5D ) , and no such defect is observed when v100R755A is expressed in wild-type neurons ( Figure 5F ) . Where do guidance receptors accumulate in neurons with developmental defects caused by v100R755A-accelerated sorting ? Figure 5G and 5H shows cross-sections through photoreceptor cell bodies of 1-d-old eyes in which the cells on the right side of the clonal boundaries are v100 mutant and express v100R755A , while the neighboring clones on the left side are wild type . In this experiment the mutant cells are marked with synapto-pHluorin ( green , MARCM ) , which accumulates in endosomal compartments [27] . As shown for Dlar and Fas2 in Figure 5G and 5H , guidance receptors exhibit strong accumulations in mutant photoreceptor cell bodies containing synapto-pHluorin aggregates . Interestingly , a substantial amount of both Dlar and Fas2 encircles Syx7 labeling and is found on the plasma membrane of the dramatically enlarged cell bodies ( arrow heads in Figure 5G and 5H ) as well as on Syx7-positive compartments ( arrows ) . Very similar cell body membrane accumulations are observed for the other guidance receptors ( data for Dlar , Rst , and Fas2 in Figure S3 . These observations suggest an endocytic defect of membrane receptors . While it is at this point unclear whether these endocytic defects are primary or secondary to an accelerated clog-up or recycling problem in the endocytic pathway , these observations clearly show that guidance receptors do not accumulate only in signaling-incompetent lysosomal compartments . In summary , our data indicate that accelerated endolysosomal sorting into degradation-incompetent compartments causes guidance receptor accumulations on plasma and/or endosomal membranes and accelerates the onset and severity of developmental defects in both photoreceptors and CNS neurons . Our observations suggest that accelerated sorting by v100R755A expression in v100 mutant neurons during brain wiring accelerates the accumulation of guidance receptors . To test this idea we analyzed Dlar , N-Cad , Fmi , Fas2 , and Rst in v100 mutant photoreceptors ( eyFLPPRonly ) with or without v100R755A expression at P+30% . As shown in Figure 6 , none of the guidance receptors exhibit obvious receptor accumulations either in the developing eye or at photoreceptor synapses at this developmental time point in the v100 null mutant ( also compare Figure 4C ) . Very mild increases are only just discernible for Rst in the developing eye and for Fmi at synapses ( arrows in Figure 6J' and 6M' ) . In comparison , accelerated sorting into degradation-incompetent compartments ( v100R755A in v100 ) leads to increased accumulations with highly variable severity and in different parts of the neuron for these five receptors at P+30% . As shown in Figure 6O , Rst accumulations are strongly increased in the eye , while N-Cad and Fas2 exhibit comparably mild increases ( Figure 6G and 6S ) and Dlar and Fmi are apparently unaffected in cell bodies in the eye ( Figure 6C and 6K ) . In contrast , at photoreceptor synapses in the same brains , Fmi is strongly increased ( Figure 6L ) , whereas Rst , Dlar , N-Cad , and Fas2 show mild or no increased accumulations ( Figure 6D , 6H , 6P , and 6T ) . Since all five guidance receptors analyzed here as well as other transmembrane proteins including lamp-GFP and CD8-GFP accumulate in v100 mutant neurons over time ( compare Figures 2 and 4 ) , we conclude that only receptors that are in the endolysosmal system at a given time in the cell body or at the synapse are subject to v100R755A-accelerated sorting and v100-dependent degradation . This interpretation is consistent with the two strongest effects shown here: Rst plays a key role in membrane sorting during eye development at P+30% , but is not yet strongly expressed at synapses [20] , [39] , whereas Fmi plays a key role in photoreceptor targeting at P+30% [13] . Co-labeling of the v100R755A-accelerated accumulations of Rst in the eye and Fmi at synapses with Syx7 reveals many colocalizing accumulations ( Figure S4 ) . The colocalization with the early endosomal marker is consistent with the findings for both v100 mutant photoreceptors and CNS neurons ( Figures 3C–3E , 5G , and 5H ) . In summary , specifically restoring the sorting function of v100 accelerates the rate of guidance receptor accumulation in developing neurons and reveals the spatiotemporal dynamics of guidance receptor turnover . Our findings in both photoreceptors and CNS neurons indicate that guidance receptors accumulate on membranes where they could potentially exert increased signaling . In particular , the v100R755A-accelerated sorting leads to accumulations of receptors both on endosomal compartments and on the plasma membrane . These findings are not consistent with the idea of accumulations in signaling-incompetent lysosomal compartments or failed exocytic membrane delivery . Rather , our data strongly suggest defects along the endocytic pathway . To directly test the activity of missorted guidance receptors in v100 mutant neurons , we designed an experiment to challenge v100 mutant photoreceptors ( eyFLPPRonly ) with overexpression of guidance receptors and other transmembrane cargo . We reasoned that increased numbers of guidance receptors should lead to receptor-specific gain-of-function phenotypes that are exacerbated when v100-dependent sorting and degradation are removed . In contrast , increased numbers of membrane proteins without signaling function should not cause developmental defects , even though they may still accumulate in the same intracellular compartments . As control transmembrane cargo , we selected lamp-GFP and myristoylated RFP ( myrRFP ) . Overexpression of both lamp-GFP and myrRFP leads to pronounced accumulations in synaptic terminals of eyFLPPRonly v100 mutants , but not in the synaptic terminals of wild-type photoreceptor neurons ( Figure 7A and 7B ) . However , even co-overexpression of both transmembrane-anchored fluorescent probes in v100 mutant photoreceptors causes no appreciable developmental defects ( Figure 7C and 7D ) . We conclude that accumulations of membrane proteins without signaling function are not sufficient to cause developmental defects . In contrast , overexpression of Rst , Fas2 , or N-Cad in v100 mutant photoreceptors causes well-defined , strong axon pathfinding or visual map formation defects . Specifically , overexpression of Rst causes distinct axon fasciculation and pathfinding defects , a phenotype that is dramatically worsened in a v100 mutant background ( Figure 7E and 7F ) . In contrast , overexpression of N-Cad in wild-type photoreceptors does not cause any appreciable developmental defect , whereas overexpression of N-Cad in v100 mutant photoreceptors causes distinct defects in visual map formation in the lamina ( Figure 7G–7J ) . This phenotype is very different from the misrouted axon bundles caused by increased Rst function . Whereas the Rst-specific fasciculation and pathfinding defects are best shown in the 3-D visualizations of axon projections in the brain , the visual map formation defect in the lamina is best demonstrated by the lamina cross-sections shown in Figure 7G–7J . Similarly , overexpression of Fas2 in wild-type photoreceptors causes no pathfinding defects , but a highly specific sorting defect of synaptic terminals in the lamina ( i . e . , a specific visual map formation defect; 11% of synaptic cartridges contain more than eight or less than four terminals , compared to <1% in wild type ) , and this Fas2-dependent phenotype is substantially worsened in a v100 mutant background ( 23% of synaptic cartridges contain less than four or more than eight terminals; Figure 7K and 7L ) . In contrast , loss of Fas2 in photoreceptors causes no obvious defects in axon targeting or visual map formation ( data not shown ) . These results show that overexpression of guidance receptors , but not membrane-tagged fluorescent probes without signaling function , causes specific developmental defects that strongly suggest exacerbated gain-of-function phenotypes . Our findings are consistent with the idea that both guidance receptor overexpression and increased receptor sorting into degradation-incompetent compartments lead to developmental defects because of increased guidance receptor activity . Indeed , both genetic manipulations lead to increased colocalization of guidance receptors with the early endosomal Syx7 , as shown for Fas2 in Figure S5 . Taken together with the finding of early accumulations of guidance receptors on endosomal compartments in v100 mutant CNS neurons , our findings support the idea that brain wiring defects in the adult CNS result at least partially from increased guidance receptor activity . To further test the idea that v100-dependent accumulations of guidance receptors lead to increased receptor signaling we turned to the Drosophila embryonic nervous system . Drosophila embryonic motor axons have long provided a simple in vivo model for characterizing axon guidance molecules [7] , [40] , since individual axons can be followed to their targets and phenotypes that result from increased signaling can often be differentiated from loss-of-function defects ( e . g . , [19] , [33] , [41] , [42] ) . The discovery of the progressive v100-dependent neuronal degradation mechanism makes clear predictions for guidance receptor sorting in the embryonic nervous system . Specifically , we propose that 24 h of embryonic development is not sufficient to lead to aberrant receptor function . However , both accelerated sorting into degradation-incompetent compartments ( v100R755A in v100 ) as well as guidance receptor overexpression in v100 mutant neurons should accelerate the occurrence of receptor-specific phenotypes similar to the effects shown for photoreceptors . To test this hypothesis , we analyzed axon pathfinding and guidance receptor sorting in the embryo . The guidance receptor Fas2 not only plays a critical role in axon pathfinding , but also is one of the most commonly used markers to analyze pathfinding , branching , and fasciculation defects in the embryonic nervous system [40] . Furthermore , the Drosophila embryonic nervous system has been used as a model to differentiate the effects of increased versus decreased Fas2 signaling [19] , [33] , [43] . As shown in Figure 8A and 8B , Fas2-positive ISNb axons reveal no statistically significant guidance defects in null mutant embryos ( blue bar in Figure 8D ) . In contrast , accelerated sorting into degradation-incompetent compartments ( v100R755A in v100 ) leads to statistically significant axon guidance defects ( Figure 8C; red bar in Figure 8D ) . Interestingly , these phenotypes are indicative of increased axon-axon fasciculation , a phenotype that is known to result from increased Fas2 signaling in axons [33] . As shown in Figure 8H , Fas2 immunolabeling is significantly increased in v100R755A-“rescued” embryos . Furthermore , co-labeling with the early endosomal marker Syx7 reveals increased accumulations of Fas2 in degradation-incompetent compartments ( v100R755A in v100 ) of embryonic neurons ( Figure S6A–S6C , S6G , and S6H ) . Similar to v100R755A expression in v100 mutant neurons , overexpression of Fas2 in v100 mutant neurons leads to significantly enhanced gain-of-function fasciculation defects compared to Fas2 overexpression in v100 heterozygous or wild-type neurons ( Figure 8E–8G ) . These results reveal that the V100-dependent degradation pathway regulates the levels of Fas2 in neurons in both the Drosophila visual and embryonic systems , and strongly argue that these v100-dependent accumulations lead to increased Fas2 signaling . Interestingly , certain aspects of embryonic nervous system development and axon pathfinding remain largely unaffected . For example , midline crossing , which is partly regulated by the Slit-Robo system [44] , is mostly resistant to v100R755A-accelerated receptor sorting ( with only low-penetrance defects ) . Similarly , v100R755A-accelerated receptor sorting does not enhance Sema-1a/PlexA–mediated repulsive signaling at the midline ( data not shown; [45] ) . However , analysis of Robo1 receptor expression reveals mild accumulations in the ventral ganglion that are increased by v100R755A expression ( Figure S6D–S6F , S6I , and S6J ) . These findings are consistent with our observation that v100R755A expression in v100 mutant neurons reveals spatiotemporally specific turnover rates of guidance receptors . A straightforward explanation for the lack of midline crossing defects is that loss of degradation does not lead to aberrant Robo signaling within the time frame of embryonic development . Finally , the embryonic nervous system allows us to directly test in the v100 mutant whether guidance receptors are successfully trafficked to the membrane surface through the secretory pathway . We made embryonic filet preparations in which axons are directly accessible to antibody washing solutions in the absence of detergent . Since the Fas2 immunohistochemistry antibody is specific to the intracellular domain , we tested this idea with an antibody against the extracellular domain of the guidance receptor DPTP69D , which functions in ISNb axon pathfinding similarly to Fas2 , at the same time and place [46] . As shown in Figures 8I and S7 , this receptor exhibits slightly increased levels of expression on the axon membrane surface in v100 mutant and v100R755A-“rescued” neurons compared to control . Taken together , our analysis of v100-dependent receptor sorting in the embryonic nervous system fully supports our results in photoreceptors and adult brain CNS neurons . Specifically , these results highlight that numerous guidance receptors are subject to the v100-dependent “sort-and-degrade” mechanism , that receptor trafficking defects are downstream of receptor secretion in the endosomal pathway , and that increased levels of guidance receptors lead to exacerbated gain-of-function defects .
Membrane trafficking underlies the growth and remodeling of axonal and dendritic branches . However , the loss of v100-dependent endolysosomal trafficking presented here has no apparent effect on membrane addition and remodeling . Instead , we identified a role for v100 in intracellular receptor trafficking . Intracellular trafficking and the v-ATPase are known to play critical roles in the dynamic localization and signaling of a plethora of transmembrane receptors [47] , [48] . Receptors may signal from the plasma membrane or may be endocytosed to exert a signaling function [5] . A prominent example in neuronal development is the regulation of cellular differentiation by endocytosis of the Notch ligand Delta [49] . However , loss of v100 causes no early developmental defects , and v100 is therefore not required for the regulation of receptor-mediated signaling that governs cellular differentiation and early tissue patterning . In contrast , we report that CNS neurons of the developing adult brain exhibit axon pathfinding and synaptic specification defects . Our findings indicate that V100 has a specialized task in neurons and has no function in the essential endolysosomal machinery required for early development . In contrast , the loss of key subunits of the V1 complex of the v-ATPase ( which is probably required for all v-ATPase function ) cause cell lethality . Specifically , eyFLP vha55 and eyFLP vha68 lead , in stark contrast to eyFLP v100 , to an abolishment of the eye ( P . R . H , unpublished data ) . The v100 mutant phenotypes are most similar to those of two other intracellular trafficking mutants that we have described before , n-syb mutants and sec15 mutants . Loss of n-syb , the gene that encodes the vesicle SNARE neuronal Synaptobrevin , leads to guidance receptor accumulations and synaptic specificity defects in the Drosophila visual system [50] . sec15 encodes a component of the Exocyst complex required for neuronal targeting or secretion functions other than neurotransmitter release . Similar to loss of n-syb , loss of sec15 leads to mislocalization of guidance receptors and photoreceptor targeting defects [29] . These findings represent mounting evidence for the employment of neuronal intracellular trafficking machinery during brain wiring . However , the neuronal degradation function presented here for v100 differs from the earlier findings for n-syb and sec15 , in that loss of v100 does not lead to targeting or “tiling” defects in the photoreceptor terminal field . Curiously , the guidance receptors most prominently affected by loss of either n-syb or sec15 are Fas2 and Dlar , while N-Cad and Fmiare not affected in sec15 mutant photoreceptors [29] . In contrast , all these guidance receptors are affected by loss of v100 in CNS neurons . We interpret these differences in the context of differing molecular functions: while loss of sec15 may lead to targeting defects of a subpopulation of neuronal vesicles required for guidance receptor localization , loss of v100 disrupts general receptor turnover downstream of the secretory pathway in neurons . This disruption could be partly due to “clog-up” of the endolysosomal pathway or due to defective endosomal recycling . Our challenge experiments using guidance receptor overexpression in v100 mutant photoreceptors and embryonic motor neurons are similar to Wingless overexpression experiments in intracellular degradation mutants . Dubois et al . [51] showed that Wingless is targeted to lysosomes and is continuously and specifically degraded posterior to each stripe of Wingless transcription . Disruption of lysosomal degradation leads to Wingless accumulations and , together with Wingless overexpression , ectopic signaling [51] . Similarly , we find that guidance receptors undergo constant turnover ( see discussion in the next section ) and that their overexpression in v100 mutant neurons leads to ectopic signaling . For example , N-Cad overexpression in wild-type photoreceptors , analogous to the Wingless experiments , does not cause obvious defects . In contrast , N-Cad overexpression in v100 mutant photoreceptors causes gain-of-function phenotypes . Similarly , overexpression of low levels ( one copy ) of Fas2 causes only very mild fasciculation defects in embryonic motor neurons [33] . In contrast , the same level of Fas2 overexpression in v100 mutant neurons causes a phenotype very similar to high levels of Fas2 overexpression ( two copies ) [33] . These observations strongly suggest increased gain-of-function phenotypes and are not consistent with loss-of-function phenotypes for these receptors . However , these findings do not exclude the possibility that parts of the compound brain wiring defects in eyFLPCNS v100 mutants are due to loss of function for other proteins affected by v100-dependent sorting . Importantly , v100 is a neuron-specific gene , and its loss does not lead to hallmark phenotypes of general lysosomal degradation mutants , including autofluorescent lipofuscin or ceroid accumulations or aberrant multilamellar lysosomal organelles [27] , [52] , [53] . Hence , V100 provides a neuronal degradation mechanism specifically required after differentiation for late brain development and neuronal maintenance . How guidance receptors are dynamically localized is unknown for most receptors . Several guidance receptors are known to be regulated by intracellular trafficking . Sema3A-induced endocytosis of Neuropilin-1 has been shown to be required for growth cone collapse during axon guidance [54] . Similarly , internalization of UNC-5A prevents UNC-5A-mediated growth cone collapse in hippocampal axon guidance [55] . One of the best characterized examples of intracellular dynamic sorting is the guidance receptor Robo [44] , [56] , [57] . During embryonic nervous system development certain axons are prevented from crossing the midline by a repellent guidance cue that binds to the Robo receptor . During a short time window , Robo is removed from the plasma membrane and the axon crosses the midline exactly once . Thereafter , Robo receptors return to the membrane and prevent the axon from crossing again . Remarkably , this dynamic relocalization of the Robo receptor is achieved by diverting a continuous supply of receptors from the endoplasmic reticulum/Golgi temporarily into the endolysosomal pathway for degradation by means of the intracellular sorting receptor Comm . Hence , the dynamic membrane presentation of Robo receptors on the growth cone is not regulated by endo- and exocytosis of a fixed amount of receptors . Instead , the regulation occurs via an intracellular sorting receptor , revealing a strategy that relies on constitutive synthesis and degradation of receptors that can be sorted to exert spatiotemporally defined functions . Notably , the proposed diversion of Robo receptors into degradative compartments is only very short . Indeed , we observe a mild increase of Robo accumulations in embryonic neurons . However , the lack of developmental defects suggests that these accumulations are not sufficient to cause aberrant signaling . We propose that loss of v100-dependent degradation leads to only a slow build-up of undegraded receptors , and 24 h of embryo development is not sufficient to lead to neuronal connectivity defects . The role of v100 in guidance receptor turnover is most strikingly highlighted by the selective rescue of v100-dependent sorting into degradation-incompetent compartments . Rescue of the sorting function , without rescue of acidification-dependent degradation , leads to a dramatically accelerated accumulation of endogenously expressed guidance receptors . Interestingly , these accumulations are increased compared to the v100 null mutant . Hence , V100 actively promotes vesicle sorting into endosomal compartments destined for degradation . In addition , we observe accumulations of guidance receptors on the plasma membrane . While we cannot exclude a primary defect in endocytosis , a secondary effect due to clog-up of the endolysosomal system or endosomal recycling defects seems more likely . In either case , these observations clearly show that guidance receptors do not exclusively accumulate in signaling-incompetent compartments . In addition , the absence of early developmental defects indicates , and our staining of DPTP69D in the embryo demonstrates , functional guidance receptor exocytosis . Our findings reveal several key features of v100-dependent “sort-and-degrade . ” First , in the complete absence of v100-dependent sorting and degradation , this turnover is at least partially taken over by a v100-independent degradation pathway . This interpretation is consistent with our previous model , in which V100 acts in parallel to an essential endolysosomal pathway that ensures cellular differentiation and viability [27] . Second , the progressive nature of the “sort-and-degrade” mechanism is similar in all different types of neurons analyzed here . We conclude that the occurrence of neuronal connectivity defects is a function of the duration between neuronal differentiation and synaptic specification . Third , these experiments reveal that guidance receptors are subject to a constant turnover . Indeed , combined measurement of guidance receptor accumulation in v100 mutant neurons and v100R755A-“rescued” neurons is a tool to assess the turnover rate of different guidance receptors . The idea that there is constant turnover is supported by the observation of different accumulation kinetics for several guidance receptors investigated here . For example , our experiments at P+30% reveal high Rst turnover in the developing eye but not at synapses , high Fmi turnover at synapses but not in the eye , and very little Dlar turnover at this developmental time point . Taken together , our findings suggest that v100-dependent “sort-and-degrade” is required for guidance receptor turnover , and its manipulation is a method to assess receptor turnover at different time points .
y w; P ( ry+ = neo FRT82B ) isogenized flies were used as control animals . v100 null mutant and overexpression lines have previously been described [24] . Allele v1004 was the mutant allele used in all experiments . All further fly strains are described in detail below . Flies were reared at room temperature , except for pupal staging experiments , where flies were reared at 25°C ( P+100% corresponds to 103 h ) . For photoreceptor-specific mosaics ( eyFLPPRonly ) [28] , [29] the base genotype is ey3 . 5FLP;;FRT82B , v100/FRT82B , cl , w+ . For optic lobe CNS neuron clones ( eyFLPCNS ) [11] the base genotype is eyFLP;;FRT82B , v100/FRT82B , cl , w+ . In order to express different reporters in either photoreceptors or all neurons , the following flies were generated . ( 1 ) For eyFLPPRonly: ey3 . 5FLP;GMR-Gal4 , ( X* ) ; FRT82B , v100/FRT82B , cl . ( 2 ) For eyFLPCNS: eyFLP , elav-Gal4; ( X* ) ; FRT82B , v100/FRT82B , cl . ( X* ) stands for one of the following UAS constructs: UAS-myr-RFP , UAS-Lamp-GFP , UAS-N-Cad , UAS-Fas2 , UAS-Rst , UAS-v100 , or UAS-v100R755A . In addition , we generated a chromosome that contains both UAS-myrRFP and UAS-Lamp-GFP . v100R755A overexpression and control experiments were done at 18°C , because higher levels of v100R755A expression in v100 mutant neurons cause cell death [27] . We used several variations of the MARCM technique [35] to generate positively marked clones with or without the expression of additional reporters or rescue constructs . In these flies , the FRT82B , cl , w+ was replaced with FRT82B , tub-Gal80 . The following flies were generated . ( 1 ) For eyFLPCNS: eyFLP , elav-Gal4; ( X* ) ;FRT82B , tub-Gal80/FRT82B , v100 . ( 2 ) For eyFLPPRonly: ey3 . 5FLP; ( Y* ) ; FRT82B , tub-Gal80/FRT82B , v100 . ( X* ) stands for one of the following UAS constructs: UAS-Lamp-GFP [37] , UAS-pHluorin [58] , or UAS-2xFYVE-GFP [36]; ( Y* ) stands for recombined chromosomes containing GMR-Gal4 and UAS-Lamp-GFP , UAS-pHluorin , or UAS-2xFYVE-GFP . The following genotype was used to negatively mark clones with RFP: ey3 . 5FLP;GMR-Gal4 , UAS-Lamp; FRT82B , UAS-RFP/FRT82B , v100 . For Lysotracker experiments , brains were removed from third instar lavae and were immobilized on a Sylgard-coated microscope slide using glue stitch . The membrane surrounding the optic lobe was carefully torn so that Lysotracker could enter . Lysotracker Red was added to HL3 at 50 nM . Then 200 µl of this solution was placed onto the prepared tissue , and an image was acquired within 5 min , as recommended by the manufacturer to prevent alkalizing effects . Live imaging was performed as described previously [59] . Dissections were performed as described previously [59] . Brains were fixed in phosphate buffered saline ( PBS ) with 3 . 5% formaldehyde for 40–50 min and washed in PBS with 0 . 4% Triton X-100 . High-resolution light microscopy was performed using the a Leica SP5 resonance scanning confocal microscope . Imaging data were processed and quantified using Amira 5 . 2 ( Indeed ) and Adobe Photoshop CS4 . Fluorescence data were quantified using GraphPad Prism 4 . The following antibodies were used at 1∶1 , 000 dilution: anti-activated Caspase-3 , Dlg , Syx7/Avl , Rab7 , Sun/CD63 , and Syt . Brp ( mAb nc82 ) , Chaoptin ( mAb 24B10 ) , N-Cad ( mAb DNEx8 ) , Rst ( mAb 24A5 ) , Flamingo ( mAb #74 ) , and Fas2 ( mAb 1D4 ) were used at 1∶50 . Guinea pig anti-V100 was used at 1∶2 , 000 . All embryonic immunostaining and assessment of motor axon guidance was done using standard approaches [42] such that whole-mount embryos were fixed , washed in PBS containing 0 . 1% Triton X-100 , and incubated in antibodies to Fas2 ( 1∶4 , 1D4 supernatant , [60] ) . Brightfield and DIC visualization and imaging were done using a Zeiss Axioimager upright microscope , and images were captured using a Zeiss Axiocam HR camera and Zeiss Axiovision software . | Brain wiring is determined by genetic and environmental factors , nature and nurture . The Drosophila brain is a model for the genetic basis of brain wiring . The fly visual system in particular is thought to be “hard-wired , ” i . e . , encoded solely by a genetic program . Some key genes encode the guidance receptors that serve as “wiring” and synaptic connectivity signals . However , it is poorly understood how guidance receptors are spatiotemporally regulated to serve as meaningful synapse formation signals . Indeed , many genes required for brain wiring do not encode the guidance receptors themselves , but rather encode parts of the cell biological machinery that governs their spatiotemporal signaling dynamics . For example , the vesicular ATPase is an intracellular sorting and acidification complex involved in regulating guidance receptor turnover and signaling . The protein V100 is a member of this v-ATPase complex , and in this study we show that mutations in the v100 gene cause brain wiring defects specifically in the adult brain . We further describe a V100-dependent intracellular “sort-and-degrade” mechanism that is required in neurons , and find that when this mechanism is perturbed , it leads to progressive build-up of and aberrant signaling by guidance receptors . These data suggest that a v100-dependent neuronal degradation mechanism provides a cell biological basis for guidance receptor turnover and spatiotemporally controlled dynamics during neural circuit formation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neuroscience/neuronal",
"and",
"glial",
"cell",
"biology"
] | 2010 | Guidance Receptor Degradation Is Required for Neuronal Connectivity in the Drosophila Nervous System |
Accurate gene or protein function prediction is a key challenge in the post-genome era . Most current methods perform well on molecular function prediction , but struggle to provide useful annotations relating to biological process functions due to the limited power of sequence-based features in that functional domain . In this work , we systematically evaluate the predictive power of temporal transcription expression profiles for protein function prediction in Drosophila melanogaster . Our results show significantly better performance on predicting protein function when transcription expression profile-based features are integrated with sequence-derived features , compared with the sequence-derived features alone . We also observe that the combination of expression-based and sequence-based features leads to further improvement of accuracy on predicting all three domains of gene function . Based on the optimal feature combinations , we then propose a novel multi-classifier-based function prediction method for Drosophila melanogaster proteins , FFPred-fly+ . Interpreting our machine learning models also allows us to identify some of the underlying links between biological processes and developmental stages of Drosophila melanogaster .
Protein or gene function prediction is a difficult computational challenge which has received increasing attention in the previous decade , with one major goal being to assist experimental biologists in making testable hypotheses about the role of uncharacterised proteins in biological systems . Ab initio prediction of gene function using in-silico methods has made great strides in the recent years , with the best methods typically making use of various protein sequence-based features in a Machine Learning framework [1–4] . FFPred is one such method , as the main component method used by the Jones-UCL team , consistently ranked near the top in independent benchmark challenges [5 , 6] . The most common method for predicting protein function is to rely on simple homology-based transfer , where function annotations are transferred from a well characterised protein to a target protein on the basis of clear common ancestry between the two . In contrast to methods exploiting direct homology-based information , FFPred predicts protein function using intrinsic features directly derived from protein sequence , such as amino acid composition , intrinsically disordered regions , signal peptides and so on . By using a wide variety of sequence features , FFPred shows better performance when making functional predictions for proteins where direct homology information provides little or no predictive power . However , while FFPred and other similar methods tend to perform well in molecular function prediction , the prediction accuracy for biological process function is frequently poorer . To assist in predicting biological process terms , it may be useful to integrate data that go beyond the features that can be derived solely from the protein sequence , such as RNA-seq data . Drosophila melanogaster is a well-studied organism that is a common model used to investigate the complex biological mechanisms of development , such as cell migration , nervous system development and so forth . Therefore , there is value in developing a protein function prediction method , which aims to not only accurately predict protein function , but also be able to identify key biological processes associated with each developmental stage . To the best of our knowledge , there is no published work which systematically studies Drosophila-specific protein function prediction , except one relevant work done by Costello , et al . ( 2009 ) [7] . The authors proposed predicting Drosophila gene function by relying on gene networks that are constructed by integrating different data sources , such as microarray expression data , genetic interaction and protein-protein interactions . However , the authors did not study the additional predictive power of sequence information , which is the main data source in protein function prediction . They also only discuss the prediction of biological process terms , rather than terms from all three domains of function covered in this work . In addition , although some other existing protein function prediction methods , e . g . [8–10] , show capacity to predict Drosophila protein function , all of them use models trained by integrating other species’ data , rather than specifically focusing on Drosophila , and none of them investigates the relationship between protein function and various developmental stages of the organism . High coverage temporal transcription expression profile data already exists for Drosophila through modENCODE [11 , 12] . The data include the time-course RNA expression information during the whole life-cycle of Drosophila . In contrast to the majority of tissue-specific [13 , 14] or certain developmental stage-specific ( such as embryo stage-specific [15 , 16] ) microarray gene expression data , this type of RNA-seq data provides the most complete gene expression information to help investigate the role of proteins played during the life-cycle of Drosophila . Using these datasets , we show we can improve the performance of protein function prediction in Drosophila and further discover informative links between protein function itself and Drosophila development . In this work , we systematically evaluate the predictive power of temporal transcription expression profile data for protein function prediction . We firstly create FFPred-fly by re-training our standard FFPred model using Drosophila-specific sequence information , and then show how FFPred-fly can be combined with an RNA-seq dataset to significantly boost its performance in biological process function prediction . We choose Drosophila development as our exemplar RNA-seq dataset , so as to focus on a well-characterised developmental system , the results of which can be more readily interpreted . However , the framework we present is quite generic and could be easily extended to integrate FFPred with any organism specific RNA-seq dataset .
We firstly evaluate the predictive power of individual types of expression-based features by comparing with the sequence-based features . Tables 1 and 2 report the mean MCC and AUROC values obtained by predicting different domains of GO terms using four different classification algorithms and additionally the Opt-Classifier . The bold-type figures denote the highest mean MCC or AUROC value for each column . In general , for predicting biological process function , expression-based features give the higher accuracy compared with the sequence-based features . Three of the four classification algorithms obtain the higher mean MCC values by adopting expression-based features ( i . e . RF with Ave , AdaBoost with Num+Ave+Main and KNN with Ave+Main ) , while all four types of classification algorithms obtain the higher mean AUROC values by using Num+Ave features . In addition , the highest result obtained by the Opt-Classifier ( i . e . 0 . 268 of the mean MCC value obtained by Ave and 0 . 712 of the mean AUROC value obtained by Num+Ave ) also suggests the better predictive performance of expression-based features . In terms of predicting molecular function and cellular component terms , sequence-based features give higher mean MCC values and AUROC values , when using all four types of classification algorithms . This fact is further confirmed with the results obtained by the Opt-Classifier , i . e . 0 . 519 and 0 . 411 of the mean MCC value , 0 . 855 and 0 . 817 of the mean AUROC value , respectively for MF and CC terms . We then report the MCC and AUROC values obtained by all different types of features when predicting all 301 GO terms . The scatter plots in Fig 2 . a , 2 . b , 2 . c , 2 . d , 2 . e and 2 . f respectively display the MCC and AUROC values obtained by the Opt-Classifier for predicting three individual domains of protein function . In each scatter plot , the x axis represents the MCC or AUROC values obtained by the sequence-based features , while the y axis represents the MCC or AUROC values obtained by different types of expression-based features . The red diagonal indicates the case when the MCC or AUROC values obtained by the sequence-based features and individual type of expression-based features are equal . The different colours of dots denote the pairs of MCC or AUROC values obtained by different types of expression-based features and the sequence-based features . In terms of predicting biological process function , as shown in Fig 2 . a , the dots in different colours ( except blue ) drop on both sides of the diagonal , while the blue dots almost all fall into the area below the diagonal . In Fig 2 . d , the majority of dots in different colors ( except blue ) fall on the area above the diagonal . This fact indicates that all types of expression-based features ( except Num ) obtain better performance than the sequence-based features . For predicting molecular function , as shown in Fig 2 . b and 2 . e , almost all dots drop in the area below the diagonal , indicating the consistent fact that all types of expression-based features perform worse than the sequence-based features . For predicting cellular component function , as shown in Fig 2 . c and 2 . f , all dots in different colours ( except blue ) drop on the both sides of the diagonal . It suggests that expression-based features ( except the Num ) show similar predictive performance against the sequence-based features . We conduct the Wilcoxon signed-rank test ( two-tailed at 5% of significance level ) on each pair of the MCC or AUROC values obtained by the individual types of expression-based features and sequence-based features . The results are included in Tables A and B in S1 Text . Overall , the significance test results confirm the findings . To begin with , one type of expression-based features—Num , performs significantly worse than the sequence-based features when predicting all three domains of protein function . Moreover , for predicting biological process function , all types of expression-based features ( except Num ) significantly outperform the sequence-based features , with one exception of the Num+Main features obtaining non-significantly better MCC values than the sequence-based features . Furthermore , for predicting molecular function , all types of expression-based features perform worse than sequence-based features . For predicting cellular component function , almost all types of expression-based features ( except Num ) perform non-significantly differently to sequence-based features , with exceptions of the Main and Num+Main features . The former obtains a significantly lower MCC value , while the latter obtains significantly lower MCC and AUROC values , compared with the Seq type of features . We further evaluate the predictive power of the combination of expression-based and sequence-based features . We report the mean MCC and AUROC values obtained by different types of feature combinations in Tables 3 and 4 . Overall , the combinations of expression-based and sequence-based features obtain higher mean MCC and AUROC values when predicting all three domains of protein function . Almost all of the four classification algorithms obtain higher mean MCC and AUROC values by exploiting the feature combinations for predicting the three domains of protein function , except the KNN classification algorithm , which obtains better results by only adopting sequence-based features for molecular function and cellular component function prediction . The Opt-Classifier also obtains the highest mean MCC and AUROC values for predicting the three domains of function by exploiting the combinations of feature types ( i . e . MCC of 0 . 287 and AUROC of 0 . 726 for predicting BP terms with Seq+Num+Ave+Main features; MCC of 0 . 530 with Seq+Ave+Main features and AUROC of 0 . 862 with Seq+Num+Ave features for predicting MF terms; MCC of 0 . 463 with Seq+Ave+Main features and AUROC of 0 . 848 with either Seq+Ave+Main or Seq+Num+Ave+Main features for predicting CC terms ) . We also report the MCC and AUROC values obtained by predicting all GO terms with the Opt-Classifier . Analogously to Fig 2 , the scatter plots in Fig 3 display the comparison of MCC and AUROC values obtained by the sequence-based features and its combination with different types of expression-based features . For predicting biological process function , as shown in Fig 3 . a and 3 . d , the majority of dots drop in the area above the diagonal . This suggests the fact that merging expression-based features with sequence-based features improves the predictive performance , compared with only adopting the sequence-based features . For predicting molecular function , as shown in Fig 3 . b and 3 . e , almost all dots drop on both sides of the diagonal , indicating similar predictive power of sequence-based features and its combinations with expression-based features . For predicting cellular component function , the combinations of expression-based and sequence-based features outperform sequence-based features , since the majority of plots drop in the area above the diagonal . The Wilcoxon signed-rank test results also confirm that all combinations of expression-based and sequence-based features ( except Seq+Num ) obtain significantly higher accuracy than only sequence-based features for predicting both BP and CC domains of protein function . In the case of predicting molecular function , almost all combinations of expression-based and sequence-based features show non-significant differences except feature combinations Seq+Num+Ave and Seq+Ave+Main which both show significantly higher MCC values , while all feature combinations obtain significantly higher AUROC values except Seq+Num and Seq+Num+Ave+Main . We further compare the predictive accuracy obtained by all 15 different types of features over the cross validation procedure . The results are shown in the boxplots in Figures A and B in S1 Text . Overall , the Seq+Num+Ave+Main type of features obtains the best accuracy ( also obtains the best ranking by considering both MCC and AUROC values , as shown in Table C in S1 Text ) for predicting biological process function , whereas both the Seq+Num+Ave and Seq+Ave+Main features performs best for predicting molecular function terms . Seq+Ave+Main features also performs the best for predicting cellular component function . We then further compare the predictive performance of different types of features using a larger training protein set , i . e . adopting the whole 70% of the protein set for training , then testing on the remaining 30% of protein set . The results show that Seq+Num+Ave+Main features perform best for predicting BP and MF domains of protein function , while the Seq+Ave+Main features obtain the highest accuracy on predicting cellular component function , since Seq+Num+Ave+Main and Seq+Ave+Main features respectively obtain the best ranking for predicting corresponding domains of protein function , by considering both MCC and AUROC values , as shown in Table D in S1 Text . We also compare the predictive performance of different classification algorithms . The pie-charts in Figures C and D in S1 Text display the proportion of GO terms for which individual classification algorithm obtains the best performance . In general , the RF is the best performing classification algorithm . For predicting biological process function , KNN and RF are the best performing classification algorithms , but RF outperforms other classification algorithms on predicting other two domains of function . We further propose a new Drosophila melanogaster-specific protein function prediction method , namely FFPred-fly+ , by exploiting the optimal combination of expression-based and sequence-based features w . r . t . corresponding domain of protein function . According to the results discussed in the previous section , we use Seq+Num+Ave+Main features for predicting biological process function and molecular function , and Seq+Ave+Main features for predicting cellular component function . FFPred-fly+ considers 4 different candidate classification algorithms ( i . e . RF , AdaBoost , KNN and LDA ) . It firstly selects the single best classification algorithm for each GO term according to the predictive performance on cross validation with varying numbers of splits of the training set , depending on the number of proteins with that GO term ( see [2] for details ) . Then the selected algorithm is trained on the whole protein training set . The performance of trained classifier is evaluated by conducting the prediction on the independent protein test set . Note that , in [2] , the performance of FFPred-fly was evaluated by testing on a 30% split test set . Hence , in this work , we evaluate the relative performance of FFPred-fly+ by testing on the same 30% test set , while conducting the classification algorithm selection and training process on a 70% split as the training set . In other words , the training data is used for both algorithm selection and training , but not for final testing . The results are shown in the scatter plots in Fig 4 , where the MCC and AUROC values obtained by both methods are displayed . In each figure , the values on x-axis denotes the MCC or AUROC values obtained by the FFPred-fly approach , while the values on y-axis denote the MCC or AUROC values obtained by the FFPred-fly+ approach; the diagonal indicates the case when the MCC or AUROC values for the prediction on same function obtained by two approaches are equal; the plots in blue indicate the MCC or AUROC values obtained by the FFPred-fly+ are greater than the ones obtained by the FFPred-fly . Two dashed lines on both of sides of diagonal indicate the value of difference on MCC or AUROC values obtained by two approaches is 0 . 1 . Overall , for predicting all 301 GO terms on the three different domains of protein function , as shown in Fig 4 . a and 4 . b , FFPred-fly+ outperforms FFPred-fly , since more dots drop in the area above the diagonal . In detail , 187 out of 301 GO terms obtain higher MCC values by using FFPred-fly+ . Among those 187 function , the difference in MCC values obtained by two approaches is greater than 0 . 1 for 56 GO terms . Among those 114 of functions obtained higher MCC values by FFPred-fly , the difference in MCC values obtained by two approaches is greater than 0 . 1 for merely 27 GO terms . Analogously , 230 out of 301 GO terms obtain higher AUROC values by using FFPred-fly+ . 40 GO terms obtain 0 . 1 higher AUROC values with FFPred-fly+ , whereas only 3 GO terms obtain 0 . 1 higher AUROC values with FFPred-fly . FFPred-fly+ performs better on predicting biological process function , as shown in Fig 4 . c and 4 . d , most of the dots in blue drop on the area above the diagonal . In detail , FFPred-fly+ obtains higher MCC values for 135 out of 196 BP terms , while 47 of them have 0 . 1 higher MCC values than the ones obtained by FFPred-fly . 157 out of 196 BP terms obtained higher AUROC values by FFPred-fly+ , while 40 of them have 0 . 1 higher than the ones obtained by FFPred-fly . For predicting molecular function , as shown in Fig 4 . e and 4 . f , FFPred-fly+ and FFPred-fly show comparable predictive performance , since the numbers of blue and red dots are similar . The latter obtains higher MCC values on slightly more terms ( i . e . 37 out of 68 ) , whereas the former obtains higher AUROC values on more terms ( i . e . 45 out of 68 ) . Among those 37 MF terms with higher MCC values , 10 terms’ MCC values are 0 . 1 higher than the ones obtained by FFPred-fly+ . Among those 31 of GO terms with higher MCC values obtained by FFPred-fly+ , 4 terms’ MCC values are 0 . 1 higher than the ones obtained by FFPred-fly . For predicting cellular component function , as shown in Fig 4 . g and 4 . h , FFPred-fly+ shows better predictive performance . FFPred-fly+ obtains higher MCC values for 22 out of 37 GO terms and higher AUROC values for 28 out of 37 terms . 5 of 22 terms are 0 . 1 higher MCC values to the ones obtained by FFPred-fly . 3 of 15 terms obtain 0 . 1 higher MCC values by FFPred-fly , and 2 out of 9 terms higher AUROC values , compared with the ones obtained by FFPred-fly+ . We also conduct the Wilcoxon signed-ranked test ( two-tailed at 5% significance level ) on MCC and AUROC values obtained by the two approaches . The results of significance test on MCC values suggest that FFPred-fly+ significantly outperforms FFPred-fly when predicting 196 BP function ( p-value ( BP ) = 9 . 6e-08 ) , whereas no significant difference between the performance of two approaches when predicting MF and CC function . Conversely , FFPred-fly+ obtains significantly better AUROC values when predicting all GO terms ( p-value ( All ) <2 . 2e-16 ) , but also for predicting the individual domains of GO , i . e . p-value ( BP ) <2 . 2e-16 , p-value ( MF ) = 9 . 0e-03 and p-value ( CC ) = 2 . 8e-02 . Table 5 displays 15 GO terms ( 5 for each domain of protein function ) that obtain the biggest improvement on predictive performance by FFPred-fly+ , compared with FFPred-fly . The meanings of the first 7 columns are self-explanatory , while the rightmost column of ( M C C × I n c ) is a normalised value by simultaneously considering the actual MCC value and the increase on MCC obtained by FFPred-fly+ . All GO terms in Table 5 in each domain are ranked in descending order according to the values of ( M C C × I n c . Generally , for predicting BP terms , GO:0007051 obtains the highest ( M C C × I n c ) value . The MCC value obtained for GO:0007051 reaches 0 . 489 ( with Random Forests ) with an increase of 0 . 256 , while the MCC value obtained by FFPred-fly is 0 . 233 . For predicting MF terms , the MCC value for the top-ranked term , i . e . GO:0003735 , reaches 0 . 930 ( which is also the highest among all MCC values obtained by predicting MF terms shown in this table ) , with an increase of 0 . 209 . For predicting CC terms , the MCC value for the top-ranked term GO:0005840 reaches 0 . 924 ( which is also the highest among all MCC values obtained by predicting CC domain of terms shown in this table ) , with an increase of 0 . 262 . We then further evaluate the reliability of prediction confidence score estimation by FFPred-fly+ . Here we define the prediction confidence score as the posterior probability of the predicted GO term annotation for each protein . The higher the confidence score , the higher the likelihood that the annotation is correct . We calculate the correlation coefficient between the varying confidence score thresholds and the corresponding precision values . In more detail , we define a positive prediction if the prediction’s confidence score is greater than the given confidence score threshold . Then the precision value is calculated by T P T P + F P , where TP denotes the numbers of corrected predicted annotations and TP + FP denotes the numbers of all predictions , where the confidence score is greater than the threshold . As shown in Fig 5 , the confidence score for predicting all three domains of GO terms shows a positive correlation with the precision value ( the r values are nearly equal to 1 . 00 ) . For example , the precision values are all greater than 0 . 8 for predicting BP terms when the threshold is greater than 0 . 8 . This further confirms that FFPred-fly+ is able to make good confidence score estimates for the predicted GO term annotations . We then further verify the predicted protein-GO term annotations by screening the false positive predicted proteins by FFPred-fly+ . The new protein-GO term annotations are generated by adopting new versions of the data ( GOA . gaf 03-10-2016 & go . obo 08-10-2016 ) . Table 6 displays 5 examples of predicted proteins included in Swiss-Prot and annotated with the GO terms by merely experimental evidence codes ( e . g . IMP ) . The rightmost column denotes the number of ancestor layers in the GO hierarchy for individual GO terms . Note that , according to the GO hierarchy , the more ancestor layers the GO term has , the more specific the protein function that GO term defines . Our system successfully detects proteins’ annotations of GO terms with 6 or 7 ancestor layers . For example , protein RE32936p is successfully predicted as relating to “generation of neurons” ( GO:0048699 ) ; “protein Box A-binding factor” is successfully predicted as involved with the process of animal organ morphogenesis ( GO:0009887 ) . The InterPro database [23] includes functional information on families , domains and other descriptive information about each protein . In this work , we compare our FFPred-fly+ method with an InterPro-based method for predicting those same 301 GO terms . We firstly obtain the domain , families and other protein descriptive information for 39855 Drosophila proteins by accessing the match_complete . xml file ( release date 09/07/2014 ) . Then we assign the GO term annotations to each protein through the interpro2go file ( release date 04/07/2014 ) . Overall , FFPred-fly+ performs better , especially when predicting biological process and cellular component function terms , whereas InterPro performs better for predicting molecular function terms . The results are shown in Fig 6 , where the x-axis denotes the MCC values obtained by using InterPro , and the y-axis denotes the MCC values obtained by FFPred-fly+ . In detail , when predicting all 301 GO terms , FFPred-fly+ obtains higher MCC values for 197 terms , while 158 of them exceed the MCC score obtained using InterPro by more than 0 . 1 . Note that , 72 terms obtain zero MCC values by InterPro , due to the fact that those GO terms were not assigned to any Drosophila proteins or functional descriptions in the database . For predicting biological process terms , FFPred-fly+ obtains higher MCC values for 150 out of 196 terms ( 131 of them by more than 0 . 1 ) and 69 terms are not assigned to any Drosophila protein . For predicting molecular function terms , the InterPro-based method obtains higher MCC values for 50 out of 68 terms ( 44 by more than 0 . 1 ) . For predicting cellular component terms , FFPred-fly+ obtains higher MCC values for 29 out of 37 terms ( 20 by more than 0 . 1 ) and InterPro has a zero MCC for 3 of them . The statistical significance test results further confirm that FFPred-fly+ performs better when predicting all 301 terms with p-value ( All ) = 4 . 4e-12 , biological process terms with p-value ( BP ) <2 . 2e-16 and cellular component terms with p-value ( CC ) = 2 . 1e-05 , while InterPro-based method performs better when predicting molecular function terms with p-value ( MF ) = 1 . 4e-06 . To gauge the usefulness of the approach , we have looked at the novel GO term assignments made by FFPred-fly+ for proteins that currently have no meaningful biological process function annotations according to the latest UniProt-GOA database ( version 10-04-2017 ) . In detail , there are 4359 out of 19834 proteins ( 22% ) that have only biological process root term annotations ( i . e . GO:0008150 ) , which convey no useful information , or don’t have any biological process term annotations at all . FFPred-fly+ assigned biological process annotations to 2964 ( 68% ) of the unannotated proteins , with a high confidence ( i . e . greater than 80% likelihood ) . For example , we report the prediction of GO:0030154 annotation for protein M9NEB4 with 90% confidence . All of our predictions are accessible via the following url: http://bioinfadmin . cs . ucl . ac . uk/ffpred/fly/ .
The importance of features indicates the power of the features for predicting target classes by the given classification algorithm . For Random Forests , the feature importance is evaluated by bootstrap analysis of the trained model . As shown in Eq 1 , the feature importance FIf is defined as the decrease on impurity ( the mean decrease on Gini index between parent node to direct descendent nodes ) among all contained trees ( DecImp f t ) times the proportion of instances Prop ( Instf ) used to construct the trees during the bootstrapping process , then normalised by the total number of trees ( No . Tree ) in the Random Forests . F I f = ∑ D e c I m p f τ × P r o p ( I n s t f ) N o . T r e e ( 1 ) R F I f = F I f F I M a x _ g l o b a l × F I f F I M a x _ l o c a l ( 2 ) Note that , the best-performing types of features used for predicting BP , MF and CC terms are combinations of expression and sequence-based features . Therefore , we discuss the feature importance of expression-based features by considering two factors , i . e . the relative importance w . r . t . the feature having the global maximum importance among all expression-based and sequence-based features , and the relative importance w . r . t . the feature having the local maximum importance only within the expression-based features . As shown in Eq 2 , we calculate the relative importance by obtaining the square root of the product of the proportion of individual feature importance against the global maximum feature importance and the local maximum feature importance respectively . The maximum value of RFI is 1 . 0 , indicating that the feature has both globally and locally maximum importance value . In addition , recall that , the optimal types of feature combinations consist of three or two types of expression-based features ( i . e . Seq+Num+Ave+Main for BP and MF terms , Seq+Ave+Main for CC terms ) . Therefore , for an individual feature ( time-point ) , we only choose the feature having the maximum related importance . For example , in Seq+Num+Ave+Main types of feature , if the time-point 1 has 0 . 2 of importance in features Num , 0 . 5 of importance in features Ave , and 0 . 4 of importance in features Main , we select 0 . 5 as the importance value for time-point 1 feature of Seq+Num+Ave+Main . We then further group all 30 time-point features , by selecting the maximum feature importance among individual main developmental stages , i . e . time-points 0—12 for embryo , time-points 13—18 for larva , time-points 19—24 for pupa , and time-points 25—30 for adult . We focus on the GO terms that obtain the highest accuracy by Random Forests when evaluating on the 30% test protein set ( i . e . 100 BP terms , 51 MF terms , and 20 CC terms ) . The distribution of RFI values for all those GO terms is shown in Fig 7 , where each colour of dots indicates the maximum relative feature importance value for each main developmental stage of Drosophila melanogaster . Generally , the majority of expression-based features indicate high importance when predicting biological process functions . As shown in Fig 7 . a , 61 out of 100 BP terms are with the RFI value being greater than the 0 . 7 of threshold at least one developmental stage . On the contrary , only a few expression-based features show high importance when predicting MF and CC terms . As shown in Fig 7 . b and 7 . c , almost all MF and CC terms have RFI<0 . 7 on all main developmental stages . These results further confirm the findings discussed in the previous sections , i . e . transcription expression profile-based features only show the relevance to predict the biological process function . Therefore , hereafter , we only further discuss the expression-based features’ relative importance on predicting biological process function . We display the relative feature importance values of 30 time-point features for predicting 10 types of specific development-associated biological process function in the heatmap shown in Fig 8 . a . The reddest colour indicates the highest RFI value ( i . e . 1 . 0 ) , while the bluest colour denotes the lowest RFI value ( i . e . 0 . 0 ) . We also further conduct hierarchical clustering analysis on those 10 functions , according to their RFI values over 30 time-point features . Note that , we show the distribution of RFI values for all other 51 BP terms in the S1 Table . Those terms are either ancestor terms for those 10 specific terms or located in the Gene Ontology hierarchy with less than three ancestor layers ( only denoting relative generic protein function ) . In general , the analysis of RFI values of expression-based features successfully identifies the association between developmental processes and certain developmental stages . It is obvious that those clustered 4 functions in the top of the heatmap are all relevant to Drosophila melanogaster’s nervous system development . According to the heatmap , those functions indicate their high relevance with more than one developmental stage of Drosophila melanogaster . Regulation of nervous system development ( GO:0051960 ) shows an active role at the embryonic and larval stage of Drosophila melanogaster development . In Fig 8 . b , the first peak of RFI value reaches 0 . 92 at T7-Embryo , along with the highest RFI value being obtained by T13-Larva . It is known that the formation of the central nervous system ( CNS ) starts at the early embryonic stage , in which the region of neuroectodermal cells are determined in order to form the neuroblasts with the delamination process from the epithelium at later embryonic stage [24–26] . At larval stage , the CNS vigorously develops as a process of neuronal morphogenesis and many adult-specific neurons are also produced with the neuroblasts persisting into larval life [27 , 28] . The peripheral nervous system ( PNS ) is also active during embryonic and larval stages , in which the sensory neurons are stereotyped positioned , associating with many biological processes , such as asymmetric cell divisions and neighbouring neurons interactions [29] . Eye morphogenesis ( GO:0048592 ) is another developmental process that is relevant to more than one main developmental stage . According to Fig 8 . c , the RFI value reaches its peak value at T7-Embryo , and raises up again at T22-Pupa . It indicates the high relevance of eye morphogenesis on the embryonic and pupal stages . It has been known that the morphogenesis of Drosophila eyes commences with the development of eye anlage at the embryonic stage . The eye anlage is generated after the partition on dorsal head neuroectoderm [30] , and then progresses to be the visual primordium that further derives to the eye-antennal imaginal disc [31] . The formation of adult eye is mainly progressed during the pupal stage , including the events related with the progression of morphogenetic furrow and pattern formation [32] . Note that , the pupal stage is also an especially important stage for the tissue morphogenesis of Drosophila , such as wing and leg morphogenesis [33 , 34] . As shown in Fig 8 . d , the peak of RFI value for predicting cell morphogenesis ( GO:0000902 ) is obtained by the time-point T22-Pupa . The distribution of RFI values for other functions listed at the bottom of the heatmap shows their relevance to unique developmental stages of Drosophila , such as cell migration , male gamete generation and so on . Cell migration ( GO:0016477 ) is one function that is detected by our system as a typical biological process happening on the embryonic stage of Drosophila . As shown in Fig 8 . e , it is clear that the RFI value reaches the peak of 1 . 0 at T5-Embryo , and it is obviously higher than the average RFI value of 0 . 20 . Actually , cell migration is a well-studied area in biology , especially for the Drosophila melanogaster system , since it was found as a complex phenomenon during embryonic stage associates with the body plan of Drosophila [35–37] . This indicates significant cell movement activities , such as primordial germ cells ( PGCs ) , phagocytic cells , cells of the tracheal system , etc . PGCs is a well-known group of cells that shape the gonads of Drosophila . During the germband extension process of the embryo , the PGCs are moved along the dorsal and then towards the center . After contacting with the somatic cells , the PGCs form the gonad on either side of embryo , during the retraction process of the germband . In this study , we evaluate the predictive power of temporal transcription expression profiles for the task of novel Drosophila melanogaster protein function prediction . We conclude that the features generated based on expression profiles obtain better performance on predicting biological process function , whilst also indicating similar performance for predicting other domains of protein function , compared with conventional sequence-based features . Furthermore , by combining expression-based and sequence-based features , the performance for predicting all three domains of protein function can be further improved . Based on the optimal types of feature combinations , our newly-proposed protein function prediction approach also significantly outperforms FFPred-fly . With the help of our machine learning models , we further illustrate the capacity of our system to help highlight some of the links between protein function and developmental stages of Drosophila melanogaster .
As our reference data set , we make use of all available Drosophila melanogaster-specific proteins and their corresponding Gene Ontology ( GO ) annotation data as described below . The Gene Ontology terms are categorised into the usual three domains of protein function , i . e . Biological Process ( BP ) , Molecular Function ( MF ) , and Cellular Component ( CC ) . We firstly generate the protein sets for GO terms by adopting the consistent procedure described in [2] . In general , for each individual GO term ( a . k . a . protein function ) , a set of proteins is grouped as one dataset , according to the annotation information . In that dataset , an instance ( protein ) is characterised by a set of features , and assigned to a class , i . e . either annotated with that GO term or non-annotated with that GO term . The total 258 features generated from the protein sequence information are the same as the ones used by [2] , such as amino acid composition , transmembrane segments and so on ( the full list of sequence-based features is included in Table E in S1 Text ) . Then the whole protein set is randomly divided into two subsets , with a proportion of 7:3 . For cross-validation , 70% of proteins are used for classifier training , and the remaining 30% used for evaluating the performance of the trained classifier . The same 30% test set is also to compare the performance of our newly proposed Drosophila melanogaster-specific protein function prediction method with our previous sequence-based method [2] retrained on Drosophila melanogaster . The annotations for Drosophila melanogaster proteins were retrieved from the Gene Ontology Annotation ( GOA ) for Drosophila [38] database ( version 01-09-2014 ) . The hierarchical dependency information between terms was retrieved from the Gene Ontology database [39] ( version 12-09-2014 ) . The amino acid sequences information was obtained from the UniProt Knowledgebase [40] ( version 2014_08 ) . In total , 10519 proteins are assigned to 301 GO terms , including 196 Biological Process terms , 68 Molecular Function terms , and 37 Cellular Component terms ( see S2 Table ) . We evaluate the predictive performance of transcription expression profile-based features in two ways . Firstly , we evaluate the predictive power of the newly-generated expression-based features by comparing with the predictive power of the conventional protein sequence-based features ( i . e . Seq ) , due to its proven success on predicting different eukaryotic organisms’ protein-GO term annotations [2] . We further generate four more types of expression-based features , as the different combinations of those three generated types of expression-based features , i . e . Num+Ave , Num+Main , Ave+Main or Num+Ave+Main . All those 7 types of expression-based features plus Seq type of features are compared . The number of features in those 7 different types of features ranges from 30 to 258 . For example , the number of features in Ave type of features is 30 , denoting the average expression profile over 30 different time-points , and the total number of the conventional Seq type of features is 258 . Moreover , we further evaluate the predictive power of features that simultaneously consist of both expression-based and sequence-based features . We combine the Seq type of features with all other types of expression-based features ( i . e . Seq+Num , Seq+Ave , Seq+Main , Seq+Num+Ave , Seq+Num+Main , Seq+Ave+Main , Seq+Num+Ave+Main ) , while still choosing the predictive performance of Seq features as the benchmark . The number of features in those types of features ranges from 258 to 348 . For example , the Seq+Num+Ave+Main type of features includes 258 sequence-based features plus 3 types of expression-based features , while each type of those expression-based features consisting of 30 individual features . We evaluate the predictive performance of this newly proposed Drosophila melanogaster-specific protein function prediction method by benchmarking with the standard FFPred 2 . 0 method [2] . In this work , we re-train FFPred 2 . 0 ( hereafter denoting FFPred-fly ) by training on the same Drosophila melanogaster-specific source data described in the previous section . In detail , we re-train the SVM classifiers for GO terms by adopting the 70%-split Drosophila melanogaster protein training set , then evaluate the performance on predicting the GO terms annotation for the remaining 30% . Note that , the grid-search hyper-parameter optimisation and the backwards feature group elimination processes are also included during the classifier training . In this work , we choose the well-known Matthews Correlation Coefficient ( MCC ) ( Eq 3 ) and the Area Under Receiver Operating Characteristics Curve ( AUROC ) as the metrics of predictive performance [2 , 3] . The value of MCC ranges between -1 to 1 , where value 0 indicates that the predictive performance is not better than random prediction . The value of AURCO ranges between 0 . 5 to 1 . 0 , where 0 . 5 denotes the random predictive performance and 1 . 0 denotes the perfect predictive performance . MCC = T P × T N - F P × F N ( T P + F P ) ( T P + F N ) ( T N + F P ) ( T N + F N ) ( 3 ) | Despite painstaking experimental efforts and the extensive sequence similarity based annotation transfers , less than a half of the fruit fly protein sequences in UniProtKB have some functional annotation . To help fill in this gap , we test the usefulness of publicly available temporal gene expression profiles and their combination with many biophysical attributes that can be effectively derived from the corresponding protein sequence . We find that such an integrative function prediction method provides more accurate predictions than using sequence data alone and we expect these predictions to help narrow down the number of experimental assays required to characterise fly protein function . We demonstrate by highlighting correlations between predicted biological process functions and known facts about fly developmental stages . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [] | 2017 | Analysis of temporal transcription expression profiles reveal links between protein function and developmental stages of Drosophila melanogaster |
The Gram-positive , spore-forming bacterium Paenibacillus larvae is the etiological agent of American Foulbrood ( AFB ) , a globally occurring , deathly epizootic of honey bee brood . AFB outbreaks are predominantly caused by two genotypes of P . larvae , ERIC I and ERIC II , with P . larvae ERIC II being the more virulent genotype on larval level . Recently , comparative proteome analyses have revealed that P . larvae ERIC II but not ERIC I might harbour a functional S-layer protein , named SplA . We here determine the genomic sequence of splA in both genotypes and demonstrate by in vitro self-assembly studies of recombinant and purified SplA protein in combination with electron-microscopy that SplA is a true S-layer protein self-assembling into a square 2D lattice . The existence of a functional S-layer protein is novel for this bacterial species . For elucidating the biological function of P . larvae SplA , a genetic system for disruption of gene expression in this important honey bee pathogen was developed . Subsequent analyses of in vivo biological functions of SplA were based on comparing a wild-type strain of P . larvae ERIC II with the newly constructed splA-knockout mutant of this strain . Differences in cell and colony morphology suggest that SplA is a shape-determining factor . Marked differences between P . larvae ERIC II wild-type and mutant cells with regard to ( i ) adhesion to primary pupal midgut cells and ( ii ) larval mortality as measured in exposure bioassays corroborate the assumption that the S-layer of P . larvae ERIC II is an important virulence factor . Since SplA is the first functionally proven virulence factor for this species , our data extend the knowledge of the molecular differences between these two genotypes of P . larvae and contribute to explaining the observed differences in virulence . These results present an immense advancement in our understanding of P . larvae pathogenesis .
Due to their role as pollinators in natural and agricultural ecosystems , managed honey bees ( Apis mellifera ) are among the most important productive livestock [1] , [2] . Therefore , pathogens and parasites attacking honey bees and causing fatal infectious diseases in individual bees which eventually might lead to the collapse of entire colonies have implications which reach far beyond apiculture [3] , [4] . Paenibacillus larvae ( P . larvae ) , a Gram-positive , rod-shaped , spore-forming bacterium , is the most devastating bacterial pathogen of honey bees . It is the etiological agent of the epizootic American Foulbrood ( AFB ) , a non-rare , globally occurring brood disease which is classified as notifiable disease in most countries [5] . Only the spores of P . larvae are infectious and they can successfully establish an infection only in first instar larvae younger than 24–36 hours . Larvae become infected by consuming spore contaminated larval diet . Spores germinate in the midgut lumen where they massively proliferate before breaching the epithelium and invading the haemocoel [6] . Dead larvae are decomposed by P . larvae to a ropy mass . Lack of nutrients might then trigger sporulation of P . larvae . In the end , the ropy mass dries down to a hard scale which consists of billions of spores . These spores drive disease transmission within the colony as well as between colonies when either fed to young larvae or when distributed by contaminated adult bees . Hence , P . larvae is an obligate killer because transmission depends on the death of the host ( for recent reviews see [7]–[9] ) . Recently , different genotypes of P . larvae ( P . larvae ERIC I–IV ) have been described based on repetitive-element PCR using enterobacterial repetitive intergenic consensus ( ERIC ) primers [10] and it has been established that these genotypes differ phenotypically [5] , [11] , [12] . The most important phenotypic difference is the difference in virulence on larval level [13] which translates into virulence differences on colony level [14] . Infection bioassays with different strains of P . larvae representing the four genotypes ERIC I–IV revealed that P . larvae ERIC I strains needed about 13 days to kill all infected larvae ( LT100 of about 13 days ) while the other three genotypes ERIC II–IV turned out to be rather fast killers with an LT100 of about 5–7 days . Therefore , genotype ERIC I can be considered less virulent and the other three genotypes can be considered highly virulent on larval level . Several epidemiological studies revealed that only ERIC I and ERIC II are frequently isolated from AFB outbreaks worldwide [11] , [15]–[17] . Therefore , these two genotypes are the most important genotypes in clinical honey bee pathology . P . larvae undoubtedly is an important honey bee pathogen , therefore , it is enigmatic that the molecular pathogenesis of AFB still remains elusive . Proteases have been implicated as virulence factors since the early days of AFB research when it became evident that P . larvae secretes an astounding level of proteolytic activity [18] , [19] . However , so far the exact role of proteases in the disease process could not be established and no specific protease could be identified as being involved in pathogenesis [20] , [21] . Recently , enolase , an enzyme involved in sugar metabolism , has been presented as putative virulence factor although the possible mechanism of action remained elusive [22] . The existence of two different P . larvae genotypes ( ERIC I and II ) with opposing virulence and the availability of a draft genome sequence of P . larvae [23] enabled us to search for putative virulence factors via comparative genomics [24] and proteomics [25] . The latter approach led to the identification of a putative surface- ( S- ) layer protein which is expressed by P . larvae ERIC II but is missing from the proteome of P . larvae ERIC I [25] . S-layer ( glyco ) proteins , in general , are water-insoluble proteins endowed with an intrinsic capability to self-assemble into two-dimensional crystalline arrays in vivo , completely covering bacterial cell surfaces during all stages of the growth cycle , as well as in vitro , either in suspension or on several supports . With up to 15% of the total protein biosynthesis effort of a bacterial cell being devoted to S-layer protein synthesis , it is conceivable that the S-layer plays a pivotal role for the bacterium in its natural environment through providing a protective coat against external host or natural environmental forces as well as an adhesion and surface recognition mechanism [26] , [27] . While discrete functions of S-layers still remain largely elusive , it is its predicted involvement in cell surface-associated phenomena that has triggered investigations on the role of S-layers in bacterial pathogenesis . Bacillus anthracis , for instance , synthesizes two S-layer proteins ( Sap and EA1 ) [28] containing a ligase domain and an adhesin domain [29] , which both are proposed to be of relevance for the establishment of anthrax [30] . Clostridium difficile , the etiological agent of antibiotic-associated diarrhea and pseudo-membraneous colitis [31] , expresses two superimposed S-layers , which were shown to be involved in the mechanism of gut colonization and in the process of adhesion to the intestinal mucosa [32] , [33] as well as in modulation of the function of human monocytes and dendritic cells [34] . Other examples for S-layers acting as virulence factors concern the human pathogens Tannerella forsythia [35] , [36] , Campylobacter rectus [37] or Campylobacter fetus [38] . Among bacteria not affecting humans is Aeromonas salmonicida causing furunculosis in fish and Bacillus thuringiensis , an entomopathogenic member of the B . cereus group , with a broad host range among invertebrates . For both bacteria the contribution of the S-layer protein to pathogenesis could be demonstrated [39] , [40] . The present study deals with the assessment of the virulence potential of a putative S-layer on the honey bee pathogen P . larvae ERIC II . Based on the availability of the naturally occurring S-layer deficient genotype ERIC I of P . larvae and the concomitant construction of a P . larvae ERIC II slpA-knockout ( ko ) mutant , a functional analysis of the SlpA S-layer was performed . More specifically , this study addresses the questions ( i ) whether a functional splA gene exists in the genome of ERIC II but not of ERIC I , by using comparative genome analysis; ( ii ) whether the SplA protein of P . larvae ERIC II is a true S-layer protein , by performing in vitro self-assembly studies of purified , recombinant protein followed by electron microscopy analysis , and ( iii ) whether the putative S-layer protein SplA of P . larvae ERIC II has biological functions in pathogenesis . The latter question was addressed by comparing an splA-ko-mutant with wild-type bacteria with respect to adhesion to larval midgut cells and mortality of infected larvae . These data demonstrated that SplA is a key virulence factor of P . larvae ERIC II . Furthermore , it is the first functionally proven virulence factor for the species of P . larvae , presenting an immense advancement in our understanding of P . larvae pathogenesis .
Several P . larvae field isolates [13] representing the P . larvae genotypes ERIC I and ERIC II [5] , were used in the course of this study ( Table 1 ) . Most strains had been extensively characterized in previous studies [5] , [13] , [14] , [25] . Genetic manipulation of P . larvae ERIC II resulting in the generation of S-layer knockout clones was performed with the field strain P . larvae 04-309 ( DSM 25430 ) , which had been shown to be a rather virulent strain on larval level with an LC50 of less than 100 cfu/ml for exposed larvae in previous studies [13] . Cultivation of the non-manipulated wild-type bacteria was performed as described previously [11] , [12] . P . larvae 04-309 manipulated knockout clones were cultivated on MYPGP-agar plates [41] supplemented with 5 µg/ml chloramphenicol and incubated at 37°C for 2–3 days . For pre-cultures , a single bacterial colony was inoculated into 3 ml of MYPGP broth [41] supplemented with 5 µg/ml chloramphenicol and cells were grown for 16 h at 37°C with shaking ( 200 rpm ) . Subsequently , 10 ml of MYPGP broth supplemented with 5 µg/ml chloramphenicol were inoculated with a maximum of 300 µl of the pre-culture to adjust an OD600 = 0 . 01 and incubated at 37°C with shaking ( 200 rpm ) until the early exponential phase was reached . Escherichia coli DH5α cells ( Invitrogen ) transformed with plasmids pTT_wsfA243 [42] or pTT_splA101 ( see below ) were cultivated in selective Luria Bertani ( LB ) media ( agar and broth ) supplemented with 30 µg/ml chloramphenicol . Plasmid DNA preparations were carried out following the manufacturer's protocols ( QIAprep Spin Miniprep kit , Qiagen ) . Concentration and purity of the plasmid preparations were analyzed by photometric analysis ( Nanodrop ) and agarose gelelectrophoresis . Preparation of spore suspensions for adhesion assays and exposure bioassays was performed as described previously [5] , [13] , [14] . Briefly , for sporulation of P . larvae , 100 P . larvae colonies resuspended in 300 µl brain heart infusion broth ( Oxoid , Germany ) were used to inoculate the liquid part of Columbia sheep blood agar slants followed by incubation at 37°C for 10 days . Subsequently , the liquid part was analyzed by phase contrast microscopy ( VWR IT 400 ) for the absence of vegetative cells . Spore concentrations were determined by cultivating serial dilutions on Columbia sheep blood agar plates as described previously [11] , [12] . Cell morphology of the S-layer knockout mutant P . larvae 04-309 ΔsplA and the parent wild-type strain P . larvae 04-309 was examined by scanning electron microscopy ( SEM ) ( Phillips SEM 515 ) . For this purpose , cells were grown in MYPGP broth at 37°C with shaking until the exponential phase was reached , harvested by centrifugation , and washed three times with sterile double-distilled water . Bacterial suspensions were deposited on an SEM grid and dried at room temperature . Dry grids were coated with gold in a sputter coater and analyzed using a scanning electron microscope operated at 15 kV . For determining the genomic sequence of the recently identified , putative S-layer protein of P . larvae ERIC II , we performed a TBLASTN analysis [43] with the obtained putative S-layer protein sequence of P . larvae 04-309 ( ERIC II ) [25] against the sequence of P . larvae BRL 230010 [23] accessed through Baylor Paenibacillus larvae Data ( http://www . hgsc . bcm . tmc . edu/bcm/blast/microbialblast . cgi ? organism=Plarvae ) . The best hit was with an open reading frame located in contig 240 and showing some homologies ( e-value 9e-45 , 47% ) to the Sap ( surface array protein ) precursor of B . anthracis ( ZP_02394181 . 1 ) but lacking SLH ( S-layer homology ) -domains . Next , we selected a primer pair ( Table 2 ) with one primer ( SplA-R5 ) located in the vicinity of the 3′-end of the sequence putatively coding for the S-layer SAP precursor and the other primer ( SplA-F4 ) located upstream of the 5′-end of a neighbouring open reading frame exhibiting putative SLH domain sequences . The obtained amplicons of about 3166 bp in P . larvae 04-309 ( ERIC II ) and P . larvae 03-189 ( ERIC I; DSM 25429 ) were sequenced ( Eurofins MWG GmbH , Ebersberg , Germany ) and analyzed . To verify that the genomic sequences obtained from P . larvae 04-309 ( ERIC II ) and P . larvae 03-189 ( ERIC I ) were not strain- but genotype-specific we sequenced the putative splA-gene in another ten strains per genotype ( Table 1 ) . All sequence analyses were performed at Eurofins MWG GmbH ( Ebersberg , Germany ) and DNA alignments were performed by using the software Vector NTI ( Invitrogen ) . The sequence encoding the S-layer protein SplA of P . larvae 04-309 ( ERIC II ) was amplified from genomic DNA by PCR with the primer pair Slp773_NcoI-for and Slp773_XhoI-re ( Table 2 ) ( Invitrogen ) using Phusion High-Fidelity DNA Polymerase ( Fermentas ) with annealing at 58°C and extension at 72°C for 80 s . PCR was performed in a thermal cycler My Cycler apparatus ( Bio-Rad ) . The splA amplification product was purified using the GeneJET Gel Extraction Kit ( Fermentas ) , followed by digestion with NcoI/XhoI and insertion into the linearized and dephosphorylated expression vector pET28a ( + ) ( Novagen ) . The resulting plasmid for was named pET28a-Spl773 . For overexpression of SplA , E . coli BL21Star ( DE3 ) cells were transformed with pET28a-Spl773 by electroporation according to the manufacturer's instructions ( Invitrogen ) . Subsequently , freshly transformed cells were grown in LB medium at 37°C and 200 rpm [44] to the mid exponential growth phase ( corresponding to an OD600 of ∼0 . 6 ) , protein expression was induced with a final concentration of 0 . 5 mM isopropyl-β-D-thiogalactopyranosid , and cultures were grown for additional 4 h at 37°C and 200 rpm , with kanamycin added to the medium at a final concentration of 50 µg/ml . E . coli BL21 cells expressing SplA were harvested by centrifugation at 5 , 000×g ( Beckman J2-Hs centrifuge and JA-10 rotor ) at 4°C for 15 min . The pellet was resuspended in 10 ml of incubation buffer ( 50 mM Tris/HCl pH 7 . 2 , 10 mM MgCl2 , 0 . 5% Triton ) per gram of wet pellet . Subsequently , 0 . 8 mg/ml of lysozyme ( Sigma Aldrich ) and 500 units of benzonase ( Merck ) per g of wet pellet were added and the suspension was incubated at 37°C for 30 min . 0 . 72 g of urea were added per ml of suspension , and buffer was added to obtain a final concentration of 6 M urea . Extraction was performed at 25°C with shaking for 30 min . Cellular debris and insoluble components were removed by centrifugation at 5 , 000×g ( 4°C , 15 min ) followed by ultracentrifugation at 250 , 000×g ( Beckman Coulter Optima L-100 XP Ultracentrifuge and 70Ti rotor ) at 4°C for one hour . The supernatant was concentrated using Amicon Ultra centrifugal filters ( Millipore ) and subsequently applied to chromatography on a Superdex 200 prep grade XK16 FPLC-column ( 1 . 6×60 cm; GE-Healthcare , Uppsala , Sweden ) using 50 mM Tris/HCl pH 7 . 2 , containing 10 mM MgCl2 and 6 M urea , as eluent , and collecting 1-ml fractions at a flow rate of 1 ml/min . Fractions containing the SplA protein according to Coomassie-stained SDS-PAGE were pooled and dialyzed at 4°C against distilled water containing 10 mM CaCl2 to promote self-assembly . An approximately 1 mg/ml-suspension of SplA protein was applied to Formvar and carbon coated 300-mesh TEM grids ( Agar Scientific ) that were rendered hydrophilic upon glow discharge using a Pelco easiGlow apparatus ( Ted Pella ) . The grids were incubated for 3 min face down on a drop of the protein suspension . Samples were fixed with 2 . 5% glutaraldehyde , washed three times with MilliQ water , and stained with 1% uranyl acetate solution ( pH 4 . 2 ) for 150 s [45] . Samples were investigated using a Tecnai G2 20 Twin transmission electron microscope ( TEM; FEI ) , operating at 80 keV . Pictures were taken with an FEI Eagle 4 k CCD camera ( 4096×4096 pixels ) . The magnification was calibrated by using negatively stained catalase crystals [46] . Image processing and lattice refinement were done with software developed in house based on Fourier domain techniques [47] , [48] . To specifically create a vector for the knockout of splA in P . larvae 04-309 ( ERIC II ) , we used vector pTT_wsfA243 [42] . This vector derived from vector pTT_wsfP1176 [49] , targeted for intron insertion at position 243/244 from the start codon of wsfA in Paenibacillus alvei . WsfA and WsfP both play a role in S-layer glycosylation reactions in P . alvei [42] . The vector is a fusion of the commercially available targetron vector pJIR750ai ( Sigma ) offering the bacterial mobile group II intron LI . LtrB sequence of Lactococcus lactis , the Geobacillus-Bacillus-E . coli shuttle vector pNW33N , and the sgsE S-layer gene promoter of G . stearothermophilus NRS 2004/3a [50] upstream of the intron cassette . Specific disruption of the splA-gene in P . larvae 04-309 ( ERIC II ) was performed following a recently described strategy [49] . The LI . LtrB targetron of vector pTT_wsfA243 was retargeted prior to transformation into P . larvae 04-309 ( ERIC II ) . For this purpose , identification of potential insertion sites of the intron in the splA open reading frame and design of PCR primers for the modification of the intron RNA was accomplished by a computer algorithm ( http://www . sigma-genosys . com/targetron ) . Modified intron RNA sequences ( EBS1 , EBS2 , δ ) specifically base pair with the insertion site target sequences in splA ( EBS1 , EBS2 , δ′ ) , leading to intron insertion and disruption of the gene . The algorithm predicted 25 possible insertion sites , with position 101 ( position 101 from the initial start codon of splA ) having the highest predicted insertion efficiency ( E-value 0 . 004 ) . Three primers ( IBS splA 101 , EBS1d splA 101 , and EBS2 splA 101 , Table 2 ) were necessary to retarget the intron to base pair with the splA target site sequence in the P . larvae chromosome . The retargeted Ll . LtrB targetron ( 353 bp ) was subsequently digested with restriction enzymes HindIII and Bsp1407I ( FastDigest , Fermentas ) and ligated into pTT_wsfA243 digested with the same restriction enzymes , thereby replacing the wsfA targetron and generating the plasmid pTT_splA101 . The mutant P . larvae 04-309 strain was analyzed for the presence of intron DNA in splA and presence/absence of SplA in 2D-SDS-PA gel electrophoresis as described recently [25] . Electrocompetent P . larvae 04-309 cells ( ERIC II ) were prepared as described [51] and 1 µg of plasmid pTT_splA101 was transformed by electroporation as recently established [52] . Cells were recovered in MYPGP broth for 16 h , plated out on MYPGP-agar containing 5 µg/ml chloramphenicol and incubated for 3 days . Clones were screened for intron insertion by PCR analysis using primers flanking the SplA intron insertion position 101 of the ORF: SplA-F and SplA-R ( Table 2 ) . SplA knockout was further verified by sequencing the PCR product of P . larvae 04-309 ΔsplA . Colony morphology of the knockout mutant P . larvae 04-309 ΔsplA and the parent wild-type strain P . larvae 04-309 was evaluated on both , CSA and MYPGP agar plates after cultivation at 37°C for 3 days . Growth rates of both strains were assessed in liquid broth ( MYPGP ) by measuring the OD600 until the stationary phase was reached . Cell shapes of both strains were also examined by using a scanning electron microscopy ( see above ) . For adhesion assays , pupal gut cells were isolated from 10 days-old pupae . Pupae were shortly immersed in H2O2 and washed with 1×PBS ( 137 mM NaCl , 2 . 7 mM KCl , 10 mM Na2HPO4 , 2 mM KH2PO4; pH 7 . 4 ) . Heads were cut off; thorax and abdomen were fixed on a Sylgard coated dish and carefully opened . Cold L15 medium ( Sigma-Aldrich ) containing 10% antibiotic-antimycotic solution ( Sigma-Aldrich ) was prepared as described before ( Kreissl and Bicker , 1992 ) and carefully added on the dish . The guts were extracted and transferred into a well of a 24-well plate with fresh L15 medium pre-warmed at 37°C; up to ten guts were put into one well . The medium was carefully removed and 1 ml of enzyme solution was added ( 0 . 05% trypsin and 0 . 5% collagenase/dispase ) . The plate was incubated with gentle shaking at 4°C , 30°C and 4°C for 1 hour each . Subsequently , guts in medium were transferred to a 1 . 5 ml-reaction tube and centrifuged for 3 minutes with 300×g . The supernatant was discarded and the pellet was gently homogenized with 40 µl of L15 medium ( pre-warmed at 37°C ) per gut . 40 µl of cell suspension were dispensed per well of a 96-well plate , and after 20 minutes of incubation 60 µl of pre-warmed ( 37°C ) BM3 medium [53] were added per well . After 24 hours incubation at 33°C , the medium was discarded and replaced with 100 µl fresh BM3 medium . Cells were cultured for 6 or 7 days before used in adhesion assays . Medium was exchanged every second day . Vitality of the cells was checked using Invitrogen's Mitotracker Red FM protocol to specifically label active mitochondria in live cells following the manufacturer's instructions: Gut cells were incubated in chamber slides , washed with BM3 medium and adhered cells were incubated with 100 µl BM3 medium supplemented with Mitotracker Red FM in a final working concentration of 300 nM at 33°C for 1 h . Subsequently , the supernatant was removed and cells were washed three times with 100 µl of 1×PBS . Supernatant was removed and cells were fixed with 100 µl Roti Histofix ( Roth ) for 20 minutes at room temperature . Subsequently , cells were washed with 1×PBS . For visualisation of the nuclei , cells were additionally treated with 1 µg/ml DAPI solution ( in methanol , Roche ) for 10 minutes . Supernatant was removed and cells were intensely washed with 1×PBS . The chamber was removed and cells were embedded with ProLong gold antifade reagent ( Invitrogen ) before fluorescence activity detection ( Nikon Ti-E Inverted Microscope ) . Pupal gut cells were analyzed microscopically using DIC for light microscopy , the TexRed filter for detection of active mitochondria and the DAPI filter for nucleus visualisation . Adhesion assays were performed according to a modified protocol developed for Streptococcus pyogenes [54] . Briefly , bacterial cells were grown as 5 ml-cultures in MYPGP medium to the early exponential phase ( OD600 0 . 3 ) . Bacterial cultures were centrifuged at 4°C for 10 min at 3214×g and the pellet of one 5 ml-culture was resuspended in 5 ml of pre-warmed ( 37°C ) BM3 cell medium . A 96 well plate containing 6 to 7 day old monolayers of pupal gut cells was inoculated with bacterial suspensions . 100 µl of bacterial suspension was added to each well and incubated at 37°C for 1 hour . Wells were extensively washed 3 times with 1×PBS to remove unattached bacteria . For counting colony forming units ( CFU ) , eukaryotic cells were detached and lysed for 10 min at 37°C with 50 µl of 0 . 5% trypsin per well . Subsequently , cell-adherent bacteria were plated for enumeration . Three dilutions of each well were plated and incubated for 3 days at 37°C . The average number of bacteria recovered per well was determined from three independent wells . Experiments were repeated three times and the percentage of adherent bacteria was calculated in relation to P . larvae 04-309 wild-type adherent bacteria ( 100% ) in each experiment . Data were analyzed by one-way analysis of variance ( ANOVA ) followed by the Bonferroni post-hoc test using GraphPad Prism 5 software ( ***p-value<0 . 001 ) . To analyse the functional role of SplA during pathogenesis , exposure bioassays were performed essentially as already described [5] , [13] . First instar larvae of the infection groups were exposed to larval diet ( 66% royal jelly ( v/v ) , 33% glucose ( w/v ) and 33% fructose ( w/v ) ) containing spores of P . larvae 04-309 wild-type ( ERIC II ) and of the SplA deficient knockout mutant P . larvae 04-309 ΔsplA at a final concentration of 100 cfu/ml ( corresponding to about LC80 of the wild-type strain to ensure that observation of both , decrease and increase in mortality would be possible ) . A concentration of 100 cfu/ml will result in about 1 cfu theoretically taken up by each larva assuming that first instar larvae consume about 10 µl larval diet during the experimental time window for infection ( 24 hours ) . An infection dose of 1 cfu per larva is the lowest reasonable dose and , hence , concentrations of less than 100 cfu/ml were not tested . Control larvae were fed with normal larval diet throughout the entire experiments . An experiment consisted of three replicates of 30 larvae each: one infected with the knockout mutant P . larvae 04-309 ΔsplA , one infected with the parent wild-type strain P . larvae 04-309 , and one non-infected control . Each experiment was performed three times ( n = 3 ) and each time larval health status and mortality were monitored daily over 15 days . Dead animals were classified as dead from AFB only when vegetative P . larvae could be cultivated from the larval remains . Surviving animals were also analyzed for the presence of vegetative P . larvae . On no occasion was P . larvae cultivated from remains of dead control animals or from survivors ( pupae at day 15 post infection ) of any of the three replicates . SplA knockout stability was proven by PCR with the gene specific primer pair SplA-F and SplA-R , flanking the intron insertion site 101 . PCR amplicons were analyzed by gel electrophoresis on a 1% agarose gel , stained with ethidium bromide and visualised by UV light . Total mortality of the P . larvae 04-309 wild-type and the P . larvae ΔsplA mutant was calculated for each replicate as proportion of larvae that died from AFB compared to the total number of exposed larvae . Data represent mean values ± SEM . To obtain the time course of infection and to calculate the LT100 ( lethal time , i . e . , time it takes the pathogen to kill all infected animals , [55] ) , the cumulative proportion of AFB-dead larvae per day ( 100% equals all AFB-dead larvae in this replicate ) was calculated for each replicate and plotted against time [5] , [13] . Survivors were excluded from this calculation because the LT is a measure for the proportion of dead animals only [55] . Data represent mean values ± SD of three independent infection assays , each with three replicates and 30 larvae per replicate; data were statistically analyzed with an unpaired Student's T Test using GraphPad Prism 5 software . The GenBank accession numbers for splA of P . larvae ERIC I and ERIC II are JQ353715 and JQ353714 , respectively .
By comparative proteome analysis between P . larvae ERIC I and ERIC II we recently demonstrated the expression of a putative S-layer protein SplA in ERIC II but not in ERIC I strains of P . larvae [25] . To verify this difference on genome level we first determined the sequence of the corresponding gene in several ERIC I and ERIC II strains ( Table 1 ) . Sequence analysis revealed that both genotypes harbor a gene putatively coding for an S-layer protein . However , in P . larvae ERIC I strains , the gene was interrupted by a frameshift mutation due to an inserted adenine ( A ) resulting in a premature stop of translation due to a stop codon TAA; the insertion and resulting frameshift mutation was missing in P . larvae ERIC II strains ( Fig . 1A ) . Therefore , only in P . larvae ERIC II a functional gene for a putative S-layer protein could be identified . Protein BLAST alignment of the translated ERIC II sequences with translated sequences of S-layer protein genes from other Bacillaceae revealed that the P . larvae ERIC II protein harbors all conserved domains characteristic of S-layer proteins . Using the Conserved Domain Database ( CDD; NCBI ) two regions in the N-terminal part of the protein coding for SLH ( S-layer homology ) domains belonging to SLH superfamilies [56]–[58] were identified . These two domains are predicted to range from residues 117–164 ( SLH1 ) and 188–220 ( SLH2 ) ( Fig . 1B ) . To support the S-layer protein status of the putative P . larvae ERIC II SplA protein self-assembly studies were performed with the recombinant protein . SplA was expressed from pET28a-Spl773 in E . coli BL21 cells and subsequently purified from 4-h cultures by gel permeation chromatography with 6 M urea contained in the eluent , which was necessary to keep the proteins disintegrated ( i . e . , in the monomeric state ) . Fractions 26–29 containing the S-layer protein according to SDS-PAGE ( not shown ) were pooled . The molecular mass of rSplA on the SDS-PA gel corresponded to the calculated molecular mass of 113 . 5 kDa ( not shown ) . After 3 hours of dialysis of the S-layer pool against 10 mM CaCl2 solution , typical self-assembly products were visible in the dialysis tube . The dialysate was diluted to a protein concentration of approximately 1 mg/ml and subjected to negative-staining . The TEM micrographs of negatively-stained rSplA clearly revealed the formation of monolayered , cylindrical self-assembly products with dimensions of approximately 2 µm length and 180 nm width ( Fig . 2A ) . Within these cylinders , the S-layer protein species are aligned in a 2D lattice of good long-range order ( Fig . 2B–D ) exhibiting lattice parameters of approximately 10 . 0 nm×15 . 4 nm and γ = 90° . These lattice parameters are in good agreement with those of the S-layer lattices from diverse members of the Bacillaceae family [59] . To be able to functionally analyze SplA , P . larvae ERIC II mutants deficient in the expression of SplA were constructed . This was accomplished by using the commercially available Targetron System ( Sigma-Aldrich ) with a modified vector [49] which leads to gene disruption due to the insertion of an intron into the target gene . This may be a less elegant method then , e . g . , homologous recombination leading to gene deletion , however , it is the first method ever developed for knocking out gene expression in P . larvae and it was the only method that worked with this bacterium . Successful insertion of the splA-specific retargeted intron ( 915 bp ) into the target gene splA of the P . larvae ERIC II strain 04-309 was demonstrated by PCR-analysis of the corresponding genomic region . While the wild-type amplicon had the expected size of 1530 bp , the mutant amplicon carrying the insertion migrated at 2445 bp ( Fig . 3A ) . This insertion resulted in loss of expression of SplA in the knockout-strain , designated P . larvae 04-309 ΔsplA , as demonstrated by 2D-SDS-PAGE analysis ( Fig . 3B ) . The dominant spot for SplA in the SDS-PA gel of the parent wild-type strain is totally absent from the gel of the knockout mutant strain . Further analysis showed that the mutant strain P . larvae 04-309 ΔsplA clearly differed from the wild-type strain in colony morphology both on Columbia sheepblood agar ( CSA ) plates as well as on MYPGP agar plates ( Fig . 4A ) . On CSA plates the mutant P . larvae 04-309 ΔsplA no longer showed the usual circular colony form but , instead , the colonies grew with an irregular shape indicative of irregular growth on these plates . Such irregularities were not observed on MYPGP-plates . On these plates the colonies of the mutant strain just showed a different color and opacity indicating that growth of the mutant bacteria was not generally impaired . Accordingly , growth of the mutant P . larvae 04-309 ΔsplA and the parent wild-type strain in liquid broth did not differ significantly ( p-value = 0 . 878; data were analyzed with an unpaired Student's t-test using GraphPad Prism 5 software ) ( Fig . 4B ) and germination and sporulation of both strains were also similar ( data not shown ) . By scanning electron microscopy ( SEM ) analysis , we observed that the vegetative cells of 04-309 ΔsplA were elongated and resembled more the cells of P . larvae ERIC I ( data not shown ) while the wild-type strain P . larvae 04-309 exhibited a rather short and stubby cell morphology ( Fig . 4C ) . We hypothesized that SplA might be involved in or mediate bacterial adhesion to midgut cells . To test this we performed cell adhesion assays with P . larvae 04-309 ΔsplA and the parent wild-type strain P . larvae 04-309 as well as P . larvae 03-189 , representing the naturally S-layer deficient genotype P . larvae ERIC I [25] using primary pupal midgut cells ( Fig . 5A ) . Comparing the cell adhesion capacity of the two wild-type strains P . larvae 04-309 and P . larvae 03-189 representing the two different genotypes ERIC II and ERIC I , respectively , revealed a significantly lower adhesion of the ERIC I strain versus the ERIC II strain: The cell adhesion capacity of P . larvae 03-189 was only 10 . 59%±5 . 61% ( mean ± SEM ) of the adhesion capacity of P . larvae 04-309 ( p-value 0 . 001 ) . Comparing the wild-type strain P . larvae 04-309 with the mutant strain P . larvae 04-309 ΔsplA gave a similar result: The splA-knockout mutant had a residual adhesion of as little as 5 . 36%±0 . 86% ( mean ± SEM ) compared to the parent wild-type strain ( Fig . 5B ) ( p-value 0 . 001 ) . These results strongly support a function of SplA in bacterial adhesion to the larval midgut . Successful adhesion to the larval midgut might be an important step in pathogenesis . We therefore performed exposure bioassays in the laboratory with P . larvae 04-309 ΔsplA and P . larvae 04-309 wild-type . Groups of first instar larvae were either fed with spores of the wild-type strain or with spores of the mutant strain both at a concentration of 100 cfu/ml . This spore concentration resulted in ∼80% mortality in the P . larvae wild-type infected larvae corroborating our previous results which showed an LC50 of less than 100 cfu/ml for this strain [13] . Control larvae were mock infected and used as internal quality control of the exposure bioassays [13] . Each dead larva of the infected groups was analyzed and only those larvae which died from P . larvae infection were included in the calculation . Comparing larval mortality in the groups infected with wild-type bacteria and mutant bacteria revealed a significant ∼55% decrease in larval mortality in the larvae infected with P . larvae 04-309 ΔsplA: While the wild-type strain killed 78 . 9%±1 . 13% of the larvae , the mutant strain killed only 34 . 4%±5 . 55% of the exposed larvae ( Fig . 6A ) . Because P . larvae 04-309 ΔsplA was still able to kill larvae , SplA does not appear to be an essential factor for pathogenicity . However , the virulence of the mutant was markedly reduced demonstrating that SplA is an important virulence factor . Representatives of the P . larvae genotype ERIC II are characterized by killing the larvae rather fast with an LT100 of about 6–7 days as opposed to the LT100 of ERIC I strains which is about 12 days [7] , [8] , [13] . To unravel whether SplA is involved in influencing the time course of disease we determined the LT100 of the mutant and the wild-type parent strain and analyzed the time course of disease by cumulatively calculating the proportion of larvae that died from AFB in both groups ( Fig . 6B ) . The time course of infection did not differ significantly between the parent and the mutant strain ( p-value = 0 . 6767; data were analyzed with an unpaired Student's t-test using GraphPad Prism 5 software ) . The LT100 of the wild-type strain was 7 . 0±1 . 0 ( mean ± SD ) days confirming previous data [13] . The LT100 of the mutant strain did not differ significantly with 10 . 3±2 . 5 ( mean ± SD ) days ( p-value = 0 . 214 ) . In both groups about 90% of the infected animals died before the onset of metamorphosis , i . e . on day 6 to 7 post infection in the exposure bioassays ( Abb . 6B ) [7] , [8] . These results indicate that SplA is not involved in determining the time course of disease or the rather low LT100 of P . larvae ERIC II when compared to P . larvae ERIC I . All bacteria isolated from dead larvae were analyzed via PCR to verify their identity as P . larvae . Bacteria isolated from P . larvae which died from 04-309 ΔsplA infection were additionally analyzed via PCR and it was confirmed that they still carried the insertion ( Fig . 6C ) .
Despite being one of the most important honey bee pathogens , the pathogenesis of P . larvae infections is poorly understood hampering the development of sustainable control or curative measures . The existence of different genotypes of P . larvae which differ in virulence [5] , [13] opened the possibility to explore the virulence mechanisms by simply comparing these genotypes using different –omics approaches . Comparative genomics using suppression subtractive hybridization ( SSH ) [60] allowed us to identify several putative virulence factors including toxins and secondary metabolites [24] . Recently , comparative proteome analysis led to the identification of an S-layer protein SplA expressed only by the highly virulent genotype P . larvae ERIC II but missing in representatives of P . larvae ERIC I [25] . Considering the role of S-layer proteins in several pathogenic bacteria it was self-evident to hypothesize that P . larvae SplA is involved in the infection process . However , studies on the role of distinct proteins in the molecular pathogenesis of P . larvae infections in honey bee larvae have been hampered in the past by the lack of a genetic system allowing the targeted disruption of gene expression in this pathogen . Recently , a gene knockout system for application in Paenibacillus alvei has been published [49] which proved to be applicable to P . larvae , too . Using this system we were able to knock out the SplA gene in P . larvae and to analyze the biological function of SplA because P . larvae knockout-mutants for SplA turned out to be viable and could be used for functional studies , especially infection studies . Thus , we here provide a most valuable molecular tool and previously unavailable means for studying P . larvae , especially P . larvae pathogenesis . Analyzing the in vivo-function of specific bacterial proteins will now be possible which will greatly enhance our understanding of the interaction between P . larvae and honey bee larvae . In addition , P . larvae can serve as a model system to analyze the in vivo function of an S-layer protein . The P . larvae ERIC II S-layer protein SplA possesses all characteristics of a typical , functional S-layer protein . The protein stretches over 1008 amino acids , including a predicted signal sequence comprised of 49 amino acid residues at the N-terminal part of the protein . The N-terminal signal sequence is proposed to be cleaved off after position 49 ( predicted by SignalP 4 . 0 server: [61] ) . The mature protein without the signal sequence has a predicted molecular weight of around 103 . 5 kDa and an isoelectric point of 4 . 97 . SplA possesses two S-layer homology ( SLH ) domains as predicted by CDD ( NCBI: [56] , [57] ) . These features predict this protein to target the bacterial surface via the secretory pathway during cell proliferation [59] , [62]–[64] . SplA exhibits strong homology to an S-layer protein Sap precursor of B . anthracis and it also has homology to S-layer proteins from other bacteria , like B . weihenstephanensis ( Sap: ADQ08578 , 74% query coverage , e-value 1e-67 ) and Geobacillus sp . WCH70 ( S-layer protein: YP_002951037 , 81% query coverage , e-value 9e-84 ) . In addition , studies with the recombinant SplA of P . larvae confirmed that the proposed SplA protein is a true S-layer protein fully covering the bacterial cell surface and capable of self-assembling . These features are essential in the context of analyzing the functional role of the S-layer of P . larvae . Pathogenesis can be divided into three steps: After encountering a host , firstly , the pathogen needs to enter the host , secondly , it has to establish infection and thirdly , it will cause disease [55] . None of these steps is remotely understood for P . larvae . Recently , it was shown that the life cycle of P . larvae in infected larvae starts with a non-invasive phase in which P . larvae massively proliferates in the midgut lumen obviously without affecting the midgut epithelium [6] . During this stage of infection P . larvae has not yet entered the host because the midgut lumen is still considered ‘outside’ . Entering the host is accomplished not until the bacteria breach the epithelium . For breaching the epithelium bacteria might need to adhere to the host cells followed by invasion into host tissue . Many bacterial structures attached to or protruding from the cell surface mediate host cell adhesion as prerequisite for invasion . S-layer proteins are known bacterial adhesins and it was this capacity to mediate adhesion to host cells that led to research into the role of S-layer proteins in pathogenesis . And indeed , in distinct pathogenic bacteria it could be demonstrated that S-layer proteins or proteins containing SLH-domains are key factors in pathogenesis . For instance , for the S-layer proteins of Tannerella forsythia , an important bacterium in periodontitis , it was recently shown that they are responsible for adhesion to host cells . Impaired adhesion in the S-layer deletion mutants resulted in reduced invasion of the host cells . However , no difference in virulence between the wild-type strain and the deletion mutants could be observed [35] . In the present work , we demonstrated that the S-layer protein SplA of P . larvae mediated adhesion to host cells . Loss of SplA in the knockout-mutant resulted in loss of bacterial adhesion to larval midgut cells . Adhesion of P . larvae to the midgut epithelial cells might be a prerequisite for the subsequent breaching of the epithelium and invasion of the haemocoel which was shown to be accompanied by larval death [6] . We , therefore , speculated that larval mortality would be lower in the knockout-mutant due to impaired adhesion capacity . And indeed , in exposure bioassay a significant reduction of larval mortality in the larvae infected by the S-layer knockout-mutant compared to the parent wild-type strain could be observed . These results suggest that SplA mediated adhesion of P . larvae ERIC II to the larval midgut epithelium plays an important role in the adhesion to and breaching of the epithelium , which are both considered key steps in the pathogenesis of P . larvae infections [7] . Although the factors involved in the subsequent invasion process still remain elusive we have shed some light onto the first step in the molecular pathogenesis of P . larvae ERIC II infections . An additional role of SplA in the recognition by P . larvae immune receptors or in escaping the immune system of the larvae cannot be excluded and will be addressed in future studies . Interestingly , P . larvae ERIC I does not possess a functional splA-gene and does not express an S-layer protein although it is as lethal as P . larvae ERIC II albeit less virulent [13] . This suggests that P . larvae ERIC I has developed a different strategy to accomplish invasion of the haemocoel and killing the larva . It is noteworthy in this context that by suppression subtractive hybridization we could identify several toxins in ERIC I but not in ERIC II [24] also suggesting that these two genotypes do not only differ in virulence but presumably also differ in their pathogenicity mechanisms . Obviously there are several ways to kill a larva in the course of AFB and it will be most interesting to elucidate the different strategies . In addition to their proposed role in virulence , S-layer proteins were shown to play an important role in maintaining cell shape and structure [65] . Comparing colony morphology and cell shape between the parent and the mutant P . larvae strain revealed that lack of expression of SplA in 04-309 ΔsplA indeed led to a change in cell and colony morphology ( Fig . 4 ) . Therefore , SplA not only is an important factor during infection , it also determines cells shape and colony morphology of P . larvae . In B . anthracis , similar morphological differences between S-layer deficient mutants and wild-type parent strains have been observed [66]: B . anthracis S-layer knockout mutants show a lengthened cell shape and a change in colony morphology . The present work provides convincing evidence that SplA is a key virulence factor of P . larvae ERIC II involved in the early steps of pathogenesis , i . e . in the adhesion of the bacteria to larval midgut cells . These findings are a major progress in our understanding of the molecular pathogenesis of P . larvae . In addition , this study clearly advances our understanding of in vivo functions of S-layers . Interestingly , S-layer knockout constructions have not been described for Lactobacilli and Paenibacilli so far , since in these bacterial species S-layer proteins obviously play an obligate role in cell maintenance [67] , [68] . Since P . larvae is one of the few examples where SplA knockout-mutants are viable , we propose P . larvae as a model organism for in vivo-studies of S-layer protein function . | Paenibacillus larvae is the most devastating bacterial pathogen of honey bees . However , the molecular interactions between infected larvae and P . larvae are poorly understood and little more than speculation exist concerning virulence factors . Recently , a putative S-layer protein has been identified in P . larvae . We here demonstrate that only representatives of P . larvae genotype ERIC II harbor a functional splA-gene and that SplA is a true S-layer protein with self-assembly capability . The presence of a functional S-layer protein is novel for P . larvae . When elucidating the biological function of SplA we broke new ground by establishing primary cell culture for pupal gut cells and by developing a genetic system for disruption of gene expression in this important honey bee pathogen . By using these novel methods we were able to prove that SplA serves as a shape-determining factor , mediates adhesion to host cells , and is a key virulence factor of P . larvae ERIC II . These results present an immense advancement in our understanding of P . larvae pathogenesis . Furthermore , we propose P . larvae as a model system for the analysis of the in vivo functions of S-layer proteins because P . larvae SlpA knockout-mutants retain viability and are thus suitable for functional studies . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"veterinary",
"diseases",
"model",
"organisms",
"veterinary",
"microbiology",
"biology",
"microbiology",
"veterinary",
"science"
] | 2012 | Identification and Functional Analysis of the S-Layer Protein SplA of Paenibacillus larvae, the Causative Agent of American Foulbrood of Honey Bees |
In recent years , an increased focus has been placed upon the possibility of the elimination of soil-transmitted helminth ( STH ) transmission using various interventions including mass drug administration . The primary diagnostic tool recommended by the WHO is the detection of STH eggs in stool using the Kato-Katz ( KK ) method . However , detecting infected individuals using this method becomes increasingly difficult as the intensity of infection decreases . Newer techniques , such as qPCR , have been shown to have greater sensitivity than KK , especially at low prevalence . However , the impact of using qPCR on elimination thresholds is yet to be investigated . In this paper , we aim to quantify how the sensitivity of these two diagnostic tools affects the optimal prevalence threshold at which to declare the interruption of transmission with a defined level of confidence . A stochastic , individual-based STH transmission model was used in this study to simulate the transmission dynamics of Ascaris and hookworm . Data from a Kenyan deworming study were used to parameterize the diagnostic model which was based on egg detection probabilities . The positive and negative predictive values ( PPV and NPV ) were calculated to assess the quality of any given threshold , with the optimal threshold value taken to be that at which both were maximised . The threshold prevalence of infection values for declaring elimination of Ascaris transmission were 6% and 12% for KK and qPCR respectively . For hookworm , these threshold values are lower at 0 . 5% and 2% respectively . Diagnostic tests with greater sensitivity are becoming increasingly important as we approach the elimination of STH transmission in some regions of the world . For declaring the elimination of transmission , using qPCR to diagnose STH infection results in the definition of a higher prevalence , than when KK is used .
Since the launch of the London Declaration of Neglected Tropical Diseases in 2012 , much progress has been made to reduce both the prevalence and associated morbidity of the neglected tropical diseases ( NTDs ) , including lymphatic filariasis ( LF ) , schistosomiasis , onchocerciasis and the soil-transmitted helminths ( STHs ) [1] . STHs are the most prevalent NTD globally and are often co-endemic with other helminth infections . STH species include hookworm ( Necator americanus , Ancylostoma duodenale and Ancylostoma ceylanicum ) , roundworm ( Ascaris lumbricoides ) and whipworm ( Trichuris trichiura ) . These species remain endemic in large parts of the world , mainly affecting the poorest regions of sub-Saharan Africa and South-East Asia [2 , 3] . Although infection with STH rarely results in death , high worm burdens in children are thought to affect growth , impair intellectual development and cause anaemia , resulting in an estimated 5 million disability-adjusted life years lost ( DALYs ) [2 , 4] . Currently , the World Health Organisation ( WHO ) aims to treat 75% of pre-school-aged children ( pre-SAC: 2–4 years ) and school-aged children ( SAC: 5–14 years ) in high-risk areas by 2020 [5] . However , this target has not yet been achieved in many areas [3] . With morbidity occurring more frequently amongst children than adults ( excepting hookworm infection in pregnant women ) , the latter are normally excluded from mass drug administration ( MDA ) programmes . Moreover , SAC are easy to reach when they attend school and , with limited resources for public health available in many STH endemic regions , it is a cost-effective method to reach a large proportion of the population . However , if adults remain untreated , they will continue to contribute to the reservoir of infectious eggs or larvae , resulting in children being continually re-infected form this infectious reservoir[6–8] . Targeted treatment of SAC-only is , therefore , unlikely to result in transmission interruption . Continued treatment with MDA will remain a necessity to control the worm burden amongst SAC unless major improvements to water , sanitation and hygiene ( WASH ) infrastructure occur through economic growth that benefits the rural poor as well as the middle classes in urban areas [9 , 10] . Anthelmintic drugs are currently donated by the major pharmaceutical companies ( GlaxoSmithKline and Johnson & Johnson ) as promised in the London Declaration . However , the continuation of large scale drug donations beyond 2020 remains uncertain [1] . Therefore , initiatives that explore the feasibility of interrupting transmission are important at present [11] . Two ongoing studies investigating the feasibility of interruption of STH transmission by MDA only are the TUMIKIA project [12] and the DeWorm3 project [13] . The countries that are involved in these projects ( Kenya for TUMIKIA; Benin , Malawi , and India for DeWorm3 ) have run successful MDA elimination programmes for LF and therefore have a strong foundation upon which to build an STH elimination programme . LF is treated by albendazole or mebendazole in combination with either ivermectin or diethylcarbamazine . Both albendazole and mebendazole are also effective in reducing the prevalence and infection intensity of STH [13 , 14] . Moreover , MDA programmes for the elimination of LF are provided to the whole community and therefore offer a unique opportunity to upscale STH MDA programmes from targeting pre-SAC and SAC to community-wide initiatives . The DeWorm3 project was specifically designed to leverage the successes of the LF programmes to increase the likelihood of STH elimination . STH parasites are dioecious and reproduce sexually within the human host . Consequently , both sexes need to be present within a single host to produce fertilized eggs . Hosts subsequently expel the ( fertilized ) eggs into the environment , where ( re- ) infection of other hosts can occur [15] . As both STH prevalence and worm burden decrease , the likelihood of both sexes residing within a single human host declines . Therefore , critical prevalence and average intensity thresholds exist at which the number of human hosts with both male and female worms falls below a critical fraction of the population , resulting in a breakpoint of transmission [15] . When this breakpoint is crossed , the parasites cannot reproduce frequently enough to sustain transmission and the population of worms will decay to extinction , even in the absence of further treatment . Various studies have investigated the optimal prevalence threshold , under defined MDA regimes , at which to reliably detect ( with a defined level of confidence ) the interruption of transmission for hookworm , and the time post-cessation of treatment at which prevalence should be measured [9] . Truscott et al . suggested that 1–2% prevalence , as determined by two independent stool samples analysed through the McMaster technique and measured one to two years after the last round of MDA , is a good indicator of STH transmission interruption [9] . However , detecting infected individuals using standard diagnostic tools is challenging at low prevalence levels . Low worm burdens and limited egg output complicate the detection of STH species , as the number of eggs in the stool reduces when the number of worms in the host decreases [16] . Studies show , through repeated sampling of a single stool , that many false negatives are recorded at low prevalence levels [17 , 18] . Multiple diagnostic tools are available to test for the presence of STH . Kato-Katz ( KK ) is one of the most commonly applied methods , as it is relatively cheap and easy to apply in resource-limited settings [19] . Moreover , KK is also applied to diagnose other helminth infections such as the schistosomes [20] . KK is a measure of the egg output in stool and a moderate-to-high correlation exists between egg output and worm burden [16] . However , due to the density dependence of worm fecundity , egg count is a poor proxy measure for worm burden . Previous studies have shown great heterogeneity in epg counts across multiple stool samples due to the aggregated distribution of eggs with the stool , stool composition and day-to-day variability within a given host [18] . Further , as worm burden reduces with prevalence , so too does the sensitivity of KK [21] . Multiple slides and/or stool samples collected over multiple days can increase the sensitivity of KK [22] , as the egg output per host is highly variable [23] . However , the availability of skilled technicians plus economic considerations often limits the number of slides that can be examined , especially in the design of Monitoring and Evaluation ( M&E ) programmes in large national programmes or trials . New techniques ( for helminthology ) , such as the quantitative polymerase chain reaction ( qPCR ) , are a promising alternative to KK . Such techniques have been in use for other infectious agents for some time but have been slow to be introduced in helminthology . With qPCR , the DNA of helminth eggs in faeces is amplified and , unlike in standard PCR , this can be translated to a quantitative measure of the infection intensity . qPCR has been found to be more sensitive than KK , particularly in low prevalence settings [16] . This is of particular importance in the context of elimination , due to the low prevalence and reduced worm burdens expected as the transmission breakpoint is approached . Due to the expected low worm burdens in these settings , KK would have lower sensitivity whilst qPCR is more likely to detect the very lowest of intensities . Recently , qPCR has also been shown to exhibit less measurement error ( less variation between readings ) compared with KK [17] . Molecular techniques have also been shown to have a higher specificity as they are able to distinguish between hookworm species [24 , 25] . KK remains the standard diagnostic tool implemented in STH morbidity control programmes due to its continued presence in WHO guidelines despite its low sensitivity . This is in part due to the low cost of the technique . Nevertheless , with the shift in focus from morbidity control in SAC to elimination through interruption of transmission in entire communities , there is a need for more sensitive diagnostic tools . At present , however , the impact of qPCR on defining the elimination of transmission has yet to be investigated . In this paper , we aim to quantify the impact of using qPCR on measured prevalence as opposed to KK , in a population approaching elimination . We employ epidemiological data and mathematical models of transmission and MDA impact in our analyses . We focus these analyses on the impact of the sensitivity of different diagnostic tools on measuring the optimal prevalence threshold at which to declare the interruption of transmission .
In this paper , data are utilised from an STH epidemiology study performed in five villages in western Kenya [16] . The objective of this study was to measure the sensitivity of modern molecular tools ( qPCR ) to examine the intensity of infection and prevalence of STH ( all three of the major infections ) and to compare this with standard KK analysis [16] . All inhabitants of the five villages were approached to participate in this study , though the number of households targeted for stool sample collection was reduced , and those included were selected at random . At the start of the study , stool samples were collected from 1551 enrolled individuals and were examined with KK ( duplicate smears on two different days ) . All the participants with positive KK results were treated with 400 mg of albendazole and their stool was collected from days 2 to 7 post-treatment . Worms ( Ascaris ) were collected from the stool and the number of worms per individual was recorded . It is estimated that approximately 80% of worms were expelled from each individual in this period , based on a pilot study done by the research group and previous work by others [16 , 26 , 27] . All positive individuals plus an additional 10% of participants who were not found to be positive , and represented the control group , were included in the first Ascaris expulsion study ( n = 4773 ) . All remaining , willing members of the five study villages ( n = 3687 ) were then treated with 400 mg of albendazole . Seventy-four participants with a positive A . lumbricoides egg count by KK provided stool samples 3 weeks post-treatment to test treatment efficacy . All participants had negative KK results at this time point , suggesting that treatment was successful . A follow-up study was performed three months post-treatment during which stools were collected from 1227 individuals , and worm expulsion was performed on 66 individuals . Samples from both the baseline and follow-up study were examined with KK and qPCR ( n = 1884 samples from 796 people with measurements by both KK and qPCR ) . We implemented a stochastic , individual-based model of parasite transmission and treatment , details of which can be found in past publications [28–30] . The model simulates the worm burden of individual hosts within a community ( i . e . village ) undergoing MDA and its structure is founded on a deterministic model based on a set of partial differential equations [6 , 8 , 30] . The model parameter values that were adopted in this study are shown in S1 Table . S1 Fig compares the stochastic individual-based model with the deterministic model as described in Truscott et al . , 2014 and 2016 [6 , 8 , 30] . With non-linear systems , the mean of a stochastic formulation is not necessarily equal to the deterministic prediction . But the non-linearity in this helminth model formulation are not so severe that this pattern arises . As can be seen in S1 Fig , the mean of the stochastic model converges to the deterministic result as the number of simulations rises . Hosts become infected when exposed to the infectious reservoir of eggs in the habitat , to which infected hosts contribute both fertilized and unfertilized eggs . These contributions are age-dependent and the precise pattern of this dependency varies between the STH species [6 , 31] . In the case of Ascaris , children dominate the contribution to the infectious reservoir whilst , in contrast , the highest hookworm burdens are found amongst adults [32] . The age profile of infection of Ascaris is fitted to data collected during a study performed in the 1980s in southern India [30 , 33] , while the hookworm data are fitted to epidemiological measurements collected in Vellore , India in 2013–15 [34 , 35] . The worm burdens in hosts ( and egg output ) follow a negative binomial probability distribution , which is typically observed for all helminth species in humans and other mammalian hosts [18 , 30 , 36–38] . This assumption means that a large proportion of people will have few worms , and a small proportion of people suffer from very high worm burdens . The negative binomial parameter k , which varies inversely with the degree of worm aggregation , is used to describe the degree of clumping . The relationship ( from the negative binomial probability generating function ) between the proportion of infected individuals ( P ) and mean worm burden ( M ) is given by the following equation: P=1− ( 1+Mk ) −k In this study , we set k as 0 . 285 ( S2 Fig ) and 0 . 35 [35] for Ascaris and hookworm , respectively . For Ascaris , data from Easton et al . [16] was applied to derive estimates of the aggregation parameter k . We included records from individuals who participated with the worm expulsion study ( n = 205 ) . However , individuals with an egg count>0 but no expelled worms were excluded from this analysis ( n = 38 ) to fit a negative binomial distribution ( S1 R Code ) . Due to the breakpoint phenomenon , it is not necessary to eliminate all worms . Breakpoints for transmission as measured by the mean worm burden have been estimated for LF , onchocerciasis and , more recently , for STH and schistosomiasis [9 , 39 , 40] . The performance of prevalence thresholds can be quantified through stochastic simulation models or in real field settings within studies . The positive predictive value ( PPV ) and negative predictive value ( NPV ) have been shown to be good measures for quantifying the quality of a given threshold in previous studies [9 , 10] . The PPV is the proportion of simulations ( or villages or clusters in a field trial ) achieving elimination , and that are identified as achieving elimination , divided by all the simulations ( villages or clusters ) that do achieve elimination based on the prevalence achieved two years post-MDA . The NPV is defined as the proportion of simulations that bounce back to the endemic levels recorded or defined prior to treatment , that are identified as bounce-backs , divided by the simulations that are identified as bounce-backs . For the purpose of this study , we assume that these communities behave as independent units , consequently migration was not included in this study . Migration patterns are thought to be important as individuals with STH infections can contaminate the environmental reservoir with open defecation . A modelling study showed that the likelihood of elimination reduces substantially when the number of immigrants increases [10] . In the context of the endgame , molecular data could be useful in investigating who is infecting whom ( i . e . form with a village population of from another village ) . However , real data describing migration patterns in the context of STH transmission is currently lacking . It is certainly likely to prolong transmission in areas , particularly if there is migration of infected individuals ( such as migrant labourers ) who are not treated . A practical solution would be to recommend migrants to receive treatment of albendazole upon arrival in their home village [10] . We ran 1000 simulations , and for each simulation we construct a village of 1000 people . Age-dependent births and deaths of the human hosts are included , and follow the demography of a typical low-income country [41] . The simulations are run for 50 years and include 10 endemic years ( assuming no treatment ) , four years of LF MDA community-wide treatment and three years of STH community-wide treatment . We varied the coverage and frequency of the treatments for Ascaris and hookworm . To establish the optimal prevalence threshold to declare elimination as reflected as the point where the PPV and NPV intersect , we must ensure that not all simulations result in elimination . Each species requires a different MDA coverage and frequency of treatment programme to get near to the breakpoint . To a large extent , this depends on parasite life expectancy ( easier for longer lived species ) and the magnitude of the basic reproductive number R0 ( see S1 Table for an overview of parameter values selected for this study ) . For Ascaris , we applied annual treatment with a coverage of 70% in pre-SAC and SAC and 35% in adults . For hookworm , we modelled biannual treatment with a coverage of 70% for pre-SAC and SAC , and 60% for adults . Table 1 shows an overview of the study including timings and frequency of treatment . Individuals are chosen at random at each time point of MDA delivery ( we assume there is no systematic non-adherence ) stratified by population size in each age group . We assume that no treatment is provided after the last round of MDA for STH . At low prevalences , the dynamics of transmission are dominated by chance events ( stochastic processes ) and it often takes many years for some populations to bounce back or finally eliminate their parasites . Therefore , we run each simulation for 50 years to allow for sufficient time to observe any bounce-backs . At year 50 , we examine if the prevalence has bounced back to endemic levels or if transmission was truly interrupted . Fig 1 illustrates the effects of the STH MDA treatment ( as defined in the simulations ) on the true prevalence of Ascaris ( Fig 1A ) and hookworm ( Fig 1B ) . In this paper , true prevalence is defined as the proportion of individuals with at least one worm ( either male or female ) . We present the time series of the STH elimination programme , including the endemic period ( 10 years ) , LF community-wide treatment , and intensified community-wide STH treatment followed by a period of no treatment . The results plotted in Fig 1 illustrate three simulations that achieved interruption of transmission and three simulations that did not achieve this goal , subsequently bouncing back to endemic prevalence levels . The diagnostic tools examined in this study are based on the detection of helminth eggs in a faecal sample from patients . The production of fertilized eggs is only possible when a male and female worm are within the same host . However , female Ascaris worms produce ( unfertilized ) eggs even in the absence of male worms [30 , 42 , 43] . In contrast , female hookworms are thought to only release eggs fertilized after mating [30] . These factors clearly affect measured prevalence at low worm burden intensities , as the likelihood of having both male and female worms in the same host decreases . To investigate the prevalence in a population , 250 random individuals from the whole community were selected . In this study , we are only interested in the absence or presence of worms within a human host , and not the intensity of infections in individuals . Therefore , we adopt a simple approach for detection , and use an egg detection probability based on the data collected by Easton et al . , [16] . A host has a probability of 0 . 98 to be found positive with qPCR if they have at least one female worm ( Ascaris ) or have both a male and female worm ( hookworm ) . The data collected by Easton et al . , [16 , 17] was also designed to investigate measurement error in the readings . Multiple readers/technicians were used to analyse different KK slides prepared using samples from the same hosts , and these samples were tested repeatedly by qPCR . As a result , there are up to 8 KK slides and up to 20 qPCR readings per host ( i . e . per faecal sample ) per time point ( baseline and follow-up ) . To test for KK positivity , an individual found positive with qPCR was selected at random . For that chosen individual , two of their KK results were selected at random ( without replacement ) . If at least one these samples had one or more eggs , the individual was identified as positive . Based on previous studies , we assume that the specificity is 100% for both KK and qPCR ( i . e . there are no false positives ) [24 , 44 , 45] . As the sensitivity of the diagnostic tools differ both across species and between studies , an analysis was performed to investigate whether this has an impact on the value of the prevalence threshold to declare transmission elimination as suggested by high PPV and NPV values .
To investigate the best prevalence threshold for declaring that the parasite population has crossed the breakpoint in transmission , and hence transmission has been interrupted , it was important to ensure that not all simulations achieve elimination . In the case of Ascaris 64% of the simulations achieved elimination whilst for hookworm this value was 47% . In this study , the negative binomial aggregation parameter , k , was set to 0 . 285 for Ascaris and 0 . 35 for hookworm . The transmission parameter , R0 , was fixed at 2 . 12 for Ascaris and 2 . 2 for hookworm . Fig 1 shows how the prevalence changes during the four phases of the programme . The mean ( true ) prevalence during the endemic phase is approximately 52% for Ascaris ( Fig 1A ) and 60% for hookworm ( Fig 1B ) . If interruption of transmission is not achieved , the infection levels bounce-back to endemic levels ( Fig 1C and 1D ) . To achieve interruption of transmission , the number of individuals harbouring worms of both sexes needs to become so low that the likelihood of successful mating falls below the level required to maintain the parasite population . Fig 2 shows the combination of individuals with only male worms , only female worms , or worms of both sexes at four key time points , from our simulations of worm loads in individuals . These time points include the endemic state , post-LF treatment , post-STH treatment and two years after the last round of MDA ( Table 1 ) . In the simulations that bounce back to endemic levels , both the proportion and absolute number of hosts in which both sexes reside are higher in value compared with simulations that do achieve interruption of transmission ( Fig 2 ) . For example , in the Ascaris simulations which do not achieve interruption of transmission , the proportion of hosts in which both sexes reside reduces from 34% in the endemic phase to 20% during the LF phase and 9% in the STH phase , but then increases to 13% two years after the last round of MDA is provided ( Fig 2A ) . However , for the Ascaris simulations that do achieve interruption of transmission , the proportion of hosts in which both sexes reside is 33% , 14% and 2% for the endemic phase , LF phase and STH phase . As prevalence also decreases when a village moves towards parasite elimination , the proportion of hosts that harbour both sexes , and thus can produce fertilized eggs , is reduced to <1% ( Fig 2A ) . Similar results were obtained from the hookworm simulations ( Fig 2B ) . If interruption of transmission is achieved for hookworm , then on average 0 . 4% of the hosts in the population are predicted to excrete fertile helminth eggs into the environmental reservoir from the simulations ( Fig 2B ) . Fig 3 . shows the relationship between mean DNA concentration ( ng/μL ) and the mean egg count + 1 for Ascaris ( A ) and hookworm ( B ) from the results recorded in Easton et al . [16] . The red square highlights the hosts that had zero egg count but were found positive with qPCR . All available samples from each host and each timepoint were included in this analysis , and the means were calculated . For Ascaris , 54 individuals were found positive with both methods , 74 were found positive with qPCR and negative with KK , and 3 individuals were negative with qPCR and positive with KK . For hookworm , 42 individuals were found positive with both methods , 145 were found positive with qPCR and negative with KK , and 3 individuals were negative with qPCR and positive with KK . The egg counts from KK and the DNA quality provided by qPCR show a strong positive relationship , we apply these data in this study to investigate the importance of diagnostic sensitivity on declaring interruption of transmission . Determining the threshold to declare interruption of transmission is not straightforward as there is no clear separation between the distribution of simulated endline prevalences for the entire community ( 2 years post-treatment ) of simulations that achieve interruption of transmission and for simulations that bounce-back to the endemic state ( Fig 1A and 1B ) . The PPV and NPV values can help to determine the optimal threshold since in an ideal world the values of both should be maximised . Fig 4 shows the distribution of prevalences measured two years post-MDA for Ascaris ( Fig 4A , 4C and 4E ) and hookworm ( Fig 4B , 4D and 4F ) . If true prevalence ( the proportion of individuals harbouring either one male worm , one female worm of worms of both sexes ) could be measured , the optimal threshold for declaring elimination would be 20% for Ascaris and 9% for hookworm ( Tables 2 and 3 ) . However , as both KK and qPCR can only detect egg-producing female Ascaris and fertilized eggs for hookworm , the prevalence thresholds need to be substantially lower due to the lower sensitivity . Thus , the threshold value for interruption of transmission of Ascaris is 6% when KK is applied and 12% with qPCR is used ( Table 2 ) . For hookworm , these threshold values are lower , at 0 . 5% for KK and 2% for qPCR ( Table 3 ) . It is important to note that the PPV values are considerably lower for hookworm compared with Ascaris , especially for the KK method . This reflects the higher KK sensitivities for Ascaris found by Easton et al . [16] , who found a KK sensitivity of 32% for hookworm and 70% for Ascaris . A higher threshold is beneficial as the sample size required to detect a threshold decreases as the threshold increases ( S3 Fig ) [46] . Fig 5 . shows the results of the sensitivity analysis , where the sensitivity levels of qPCR tools were varied from 70%-95% . The sensitivity of the KK results was kept relative to the qPCR result based on the data of Easton et al . [16] as described above . These results indicate that the PPV values for both species are more dependent on the prevalence threshold rather than the sensitivity of the diagnostic test ( Fig 5A and 5B ) . For Ascaris , the PPV remains > 0 . 95 for qPCR sensitivity levels up to 70% if the prevalence thresholds < 10% . Whilst the PPV values for KK are > 0 . 8 for qPCR sensitivity levels > 75% ( Fig 5A ) . The PPV values for hookworm remain > 0 . 85 if the prevalence threshold is 2% or lower and the qPCR sensitivity > 75% . The KK PPV values drop to 0 . 54 when the qPCR sensitivity is assumed to be 70% and the prevalence threshold is 2% . For hookworm , the NPV values are always > 0 . 95 if the prevalence threshold is 2% or higher for both qPCR and KK ( Fig 5D ) . Whilst for Ascaris the NPV values are substantially lower for qPCR results compared with KK results ( Fig 5C ) .
In this study , we have focused on the absence or presence of parasites in the human host and not the intensity of infection . However , the current WHO guidelines are based on morbidity control in SAC , and this is reflected within the guidelines by a series of classes for the intensity of infection , as measured by the number of eggs per gram ( epg ) of faeces , to reflect low , medium , and high intensities . Some published studies have suggested different methods to mimic the measuring of STH infection [30 , 47] . As mentioned previously , epg is not a direct measure of the number of female worms present in the host , as the number of eggs produced per female worm is known to exhibit strong density-dependence . Consequently , the total number of eggs produced per worm declines as worm burden rises [38 , 48 , 49] . This non-linearity can be accounted for within diagnostic models as illustrated by Coffeng et al . and Truscott et al . [30 , 47] . These models require worm expulsion data to estimate the two fecundity parameters , the density dependent parameter ( γ ) , which measures the severity of the non-linearity , and the mean number of eggs produced per female worm in the absence of density dependent constraints ( λ ) [49] . Fitting these two model parameters requires acquiring worm expulsion data in combination with epg data from stools , such as KK , and to link epg output to the number of female and male worms within a host plus recording per capita egg output per female worm . When non-linear density dependent functions ( power or exponential ) are fitted to such data they show much variability due to errors in measurement in both epg and worms expelled [30] . Worm expulsion studies are highly demanding and time-consuming . Moreover , the results of the worm count are unreliable as it is difficult to collect all stools produced over a period of several days and , in the case of hookworm , find all the worms present in stool . To collect between 80% and 97% of the worms present within a host , stool needs to be collected from day 2 to at least day 7 post-MDA for all individuals [16 , 26 , 27] . However , compliance of stool donation varies between individuals and is negatively correlated with the length of the collection period . Therefore , the number of worms collected per individual is typically an underestimation of the true number of worms that are present in a host [27] . The individuals who were included in the worm study by Easton et al . ( 2016 ) had Ascaris worm counts ranging from zero up to 54 parasites per individual [16] . This is higher than the worm counts in the simulation data , therefore , it may be possible that the sensitivity reported in Easton et al . ( 2016 ) is overestimated at very low worm burdens . However , our sensitivity analysis indicates that our findings are robust for different qPCR sensitivity levels assumed . For Ascaris , the sensitivity of qPCR has no effect for qPCR sensitivity levels greater than 75% when the prevalence threshold to declare interruption of transmission is less than or equal to 7 . 5% . Whilst for hookworm , PPV values remain over 0 . 85 if the sensitivity of qPCR is 75% for hookworm with a 1–2% prevalence threshold to declare elimination . The hookworm expulsion data was only performed during the follow-up study and was therefore not very successful ( only 11 individuals with hookworm were detected ) [16] . Expulsion data from areas which are close to the interruption of transmission are not currently available , but some studies have recently collected , or are in the process of collecting , such information ( i . e TUMIKIA and DeWorm3 ) . There is a clear need to acquire more data regarding the sensitivity of diagnostic tools and qPCR particularly in low-intensity settings . Animal models may be a good alternative as worm counts are more reliable post-mortem [50] . One of the main disadvantages of qPCR is the costs involved . However , there has been continued progression in the development of molecular techniques , resulting in lower economic costs per test [51] . The per test direct costs should not be considered in isolation , since indirect costs such as labour should also be considered [52] . In addition , the sensitivity of KK , particularly for hookworm , also depends on the time between stool collection and processing of the samples , as hookworm eggs/larvae degrade quickly over time [53] . Moreover , qPCR results show less between-sample variation than KK , making it possible to use just one sample , thus perhaps making qPCR more cost-effective [17] . The likelihood of achieving interruption and the ability to detect this depends on many factors , such as the baseline prevalence ( i . e . the intrinsic transmission intensity in a given location ) . In the simulations described in this paper , the transmission parameter ( R0 ) , severity of density dependence in fecundity , age-related changes in exposure to infection and the value of the aggregation parameter k were fixed . Clustered randomized trials ( such as DeWorm3 and TUMIKIA ) are often based on clusters of villages . These villages are likely to exhibit variability in the intrinsic or pristine transmission intensity and in STH species mix . By grouping these villages into clusters , it becomes more difficult not only to achieve elimination , but also to measure its occurrence village by village . The PPV value drops substantially as the number of villages per cluster increases [9] and a lower prevalence threshold may be needed to increase the PPV and NPV values and hence improve the certainty in declaring elimination . Therefore , the optimal threshold may vary in different spatial and country settings . One specific weakness in our simulation studies is the assumption that the aggregation parameter k is independent of the prevailing prevalence of infection . If the value of k reduces ( worms become more aggregated as prevalence falls ) perhaps due to persistent non-compliers to treatment , different threshold prevalence may be appropriate in defining an elimination threshold . This aspect is the subject of further study . Countries which move from morbidity control programmes towards transmission elimination should carefully consider the difference in prevalence estimates due to increased sampling and the higher sensitivity gained from the new diagnostic tools . In determining the impact of MDA when moving towards elimination goals , it is important to sample community-wide and not just in SAC . For example , for hookworm , intensity and prevalence are normally found to be higher amongst adults [32] . However , Ascaris infections may be lower when measured community-wide as the highest infections are typically found in SAC . In such measures , due note must be taken of community demography . Due to the intensified sampling and more sensitive diagnostics , the measured prevalence will be higher than currently measured during LF transmission assessment surveys ( TAS ) , which are commonly based on one slide of KK and only recorded in SAC . The higher-than-expected prevalence may alarm programme managers involved in elimination trials and result in a higher coverage or frequency of MDA than specified in the programme design specifications , which consequently could harm the experimental design . Good communication will be necessary to tackle these issues . Temperature and humidity affect the survival of the free-living stages in the environment and this can be important for the timing of MDA . However , the impact may differ between settings and the data needed to quantify the impact are very limited at present [54] . The number of eggs released per female worm differs between the helminth species . Ascaris female worms are thought to release approximately 200 , 000 eggs per day [55] , whilst female hookworms are thought to produce approximately 30 , 000 eggs per day [56] . Given the large number of eggs produced , we do not expect worm debris to contribute significantly to the DNA detected by PCR techniques . Hence , any detected DNA should mainly derive from eggs produced by the female worms . However , the sensitivity of qPCR in cases where no eggs are produced is currently unknown . qPCR techniques may produce positive results when worms in a host do not produce eggs . For example , this may be the case for Ascaris when a host has only male worms , or for hookworm when both sexes do not reside within a host . Recognising that such events may occur is much easier than presenting solutions to negate such measurement errors . Serological assays may be an alternative to qPCR . Serology assays , measuring specific antibody responses to defined worm antigens , can overcome the problem of identifying immature parasitic infections [57 , 58] . These methods , including enzyme-linked immunosorbent assay ( ELISA ) , can provide quantitative measurements of worm infection , through specific antibody titres , of both mature and immature worms in the intestine , as well as circulating larvae . At present , however , little is understood about whether such antibody titres reflect past or current infection , or both . They obviously can detect circulating antibodies produced by previous , as well as current infections . In the absence of reinfection these antibodies are generally thought to have a short half-life of up to six months but much uncertainty surrounds such figures as does the role of memory cells in stimulating new antibody production [59 , 60] . Serology assays may become very valuable tools in the future when testing for elimination of infection in a community , one or more years after the end of an MDA programme . At present , however , too much uncertainty surrounds the question of what antibody titres precisely measure and the variability in response patient-by-patient to a defined antigenic exposure . In conclusion , diagnostic tests with improved sensitivity are increasingly important as we move towards a goal of transmission interruption for the helminth NTDs . However , there are currently no “gold standards” for the detection of STH infections aside from the now very old faecal egg detection methods by microscopy or floatation [16 , 30 , 42 , 51] . The use of qPCR as a diagnostic tool for STH is a very recent development . At present there is a need for standards plus protocols to be agreed on internationally ( e . g . by WHO ) and clearly defined for the DNA to be detected plus sample preparation and replication . More research is needed to investigate whether qPCR can detect juvenile infections . If this is the case , and it is thought to be unlikely , the elimination threshold should be higher than specified . However , even though the use of qPCR presents a number of important challenges , its high sensitivity makes it a very valuable tool in working towards the elimination of STH species . | Soil-transmitted helminths are categorised as a neglected tropical disease and comprise four dominant species ( two hookworms , Trichuris trichuria & Ascaris lumbricoides ) that affect the poorest people in the world . The World Health Organisation ( WHO ) has made great strides in reducing the morbidity induced by STH infections in pre-school aged and school-aged children through mass drug administration . Many countries are now considering moving from morbidity reduction in school-aged children to community-wide treatment with the aim of transmission elimination . These helminths reproduce sexually within a human host and therefore both male and female worms must be present to produce fertilized eggs . The density of female and male worms below which mating success is too low to sustain parasite populations is defined as the ‘breakpoint’ in transmission . Both the prevalence and intensity of infection are very low as this breakpoint is approached when worm numbers are highly aggregated in their distribution within the human host population . Consequently , it becomes increasingly challenging to identify infected individuals using standard microscopic diagnostic tools ( such as Kato-Katz ) . New and more sensitive molecular diagnostics tools , such as qPCR , are a necessity in settings where communities are moving towards the interruption of transmission . This paper demonstrates that the threshold to declare interruption of transmission is 50% lower when microscopic techniques are applied compared with molecular techniques . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"helminths",
"tropical",
"diseases",
"hookworms",
"parasitic",
"diseases",
"animals",
"health",
"care",
"simulation",
"and",
"modeling",
"ascaris",
"neglected",
"tropical",
"disease... | 2018 | Testing for soil-transmitted helminth transmission elimination: Analysing the impact of the sensitivity of different diagnostic tools |
In many biological settings , two or more cells come into physical contact to form a cell-cell interface . In some cases , the cell-cell contact must be transient , forming on timescales of seconds . One example is offered by the T cell , an immune cell which must attach to the surface of other cells in order to decipher information about disease . The aspect ratio of these interfaces ( tens of nanometers thick and tens of micrometers in diameter ) puts them into the thin-layer limit , or “lubrication limit” , of fluid dynamics . A key question is how the receptors and ligands on opposing cells come into contact . What are the relative roles of thermal undulations of the plasma membrane and deterministic forces from active filopodia ? We use a computational fluid dynamics algorithm capable of simulating 10-nanometer-scale fluid-structure interactions with thermal fluctuations up to seconds- and microns-scales . We use this to simulate two opposing membranes , variously including thermal fluctuations , active forces , and membrane permeability . In some regimes dominated by thermal fluctuations , proximity is a rare event , which we capture by computing mean first-passage times using a Weighted Ensemble rare-event computational method . Our results demonstrate a parameter regime in which the time it takes for an active force to drive local contact actually increases if the cells are being held closer together ( e . g . , by nonspecific adhesion ) , a phenomenon we attribute to the thin-layer effect . This leads to an optimal initial cell-cell separation for fastest receptor-ligand binding , which could have relevance for the role of cellular protrusions like microvilli . We reproduce a previous experimental observation that fluctuation spatial scales are largely unaffected , but timescales are dramatically slowed , by the thin-layer effect . We also find that membrane permeability would need to be above physiological levels to abrogate the thin-layer effect .
In many biological processes , two or more cells come into physical contact to form a cell-cell interface . These include cell-cell contacts like those in the epithelium [1 , 2] that change on timescales of hours , and also transient contacts that form on seconds timescales , including those formed by lymphocytes and other immune cells that must interrogate many cells rapidly [3 , 4] . A fundamental question for all cell-cell interfaces is how receptors and ligands come into contact , despite being separated by extracellular fluid , various large surface molecules like ectodomains of membrane proteins , and other structures in the negatively-charged glycocalyx . The contribution of large surface molecules has received most attention , for example producing spatial pattern formation based on molecular size [5–9] of the T cell receptor ( TCR ) and the immunotherapy target PD-1 [10] . In this work , we focus on the role of the fluid [11–14] . To highlight the potential importance of the hydrodynamics of extracellular fluid at an interface , we perform a preliminary calculation ( unrealistically ) assuming cells are rigid , impermeable spheres of radius rcell . In order to bring these cells into close contact , a force F pushes them together , as shown in Fig 1A . This fluid dynamics problem can be solved analytically for the separation distance z , yielding [15 , 16] d z d t = - 1 6 π η r cell ( z r cell ) F ( 1 ) where η is the extracellular fluid viscosity . This equation is reminiscent of the Stokes drag formula for a sphere in free fluid , but modified by a factor ( z/rcell ) ∼ ( 10 μm/10 nm ) ∼ 103 . In other words , the force required to move two cells together is increased by a thousand-fold , a strikingly large correction . This observation , known as the “lubrication limit” , “confinement effect” or “thin-layer effect” [11 , 15 , 16] , heuristically arises because a small change in z requires incompressible fluid to move a large distance to outside the interface . The cell surface is not a rigid sphere , but a deformable membrane subject to thermal undulations , active forces , and hydraulic permeability due largely to membrane inclusions like aquaporins . In this context , we ask , what is the role of the fluid in close-contact formation ? Are thermal undulations sufficient for receptor proximity ? Are typical F-actin filopodial forces , ∼10 picoNewtons [17 , 18] , sufficient for receptor proximity ? And how much force is required for rapid proximity ( <1 second ) ? If there is a significant thin-layer effect , the force required will increase for smaller cell-cell distances , but larger distances require longer protrusions , suggesting the possibility of an optimal “attack range” which might explain the biological benefit of filopodia . If the membrane is permeable to extracellular fluid [19] , how much permeability is required for rapid proximity ? Factors that influence permeability , such as aquaporins , are under regulation [20] , differentially localized , and impact cell processes including cancer angiogenesis [21] , raising the possibility that cell-cell contact can be regulated in this way . In contrast to previous theoretical studies of cell-cell interfaces , many of which capture membrane and molecular dynamics but exclude hydrodynamics , or exploit equilibrium statistical physics and therefore omit dynamics , studying the influence of active forces requires a full fluid dynamics model . Such models have been studied using both analytical methods [22 , 23] and computational methods [24] , reviewed in [25] . We have developed a computational fluid dynamics algorithm capable of simulating fluid-structure interactions with thermal fluctuations on seconds- and microns-scales [26] based on the Stochastic Immersed Boundary Method [24 , 27–30] . Here , we use this to simulate two opposing membranes , variously including thermal fluctuations , active forces , and membrane permeability . We find that the thermal fluctuations are significantly modified by the thin-layer effect for a range of assumptions about molecular sizes . Active forces are sufficient to drive proximity . The thin-layer effect has the consequence of introducing two timescales ( milliseconds and microseconds ) in response to the two length scales inherent in the system . We find that membrane hydraulic permeability overcomes the thin-layer effects , but only for values larger than previous physiological estimates .
Receptor-ligand contact for the TCR occurs when the membranes are separated by Δ z 0 ⋆ ≈ 13 nm . For the remainder of the manuscript , we refer to “membrane proximity” or “receptor-ligand contact” , defined as membrane configurations with separation Δ z 0 ≤ Δ z 0 ⋆ . Other parts of the membranes are separated by a distance Δz∞ , where estimates range from 22 nm to 150 nm [7 , 31–37] for ectodomains of signaling molecules like CD45 , non-specific binding pairs like LFA-ICAM and cadherins , and the glycocalyx . At the same time , cells themselves are rcell ∼ 2 μm for the smallest T cells [3] . To explore the consequences of the thin layer geometry , plus the incompressibility of fluid , we are required to simulate a 3D system with a resolution of receptor-ligand size Δ z 0 ⋆ in a domain larger than the cell , which has radius rcell . ( The analogous system in 2D would be insufficient since the opportunity for evacuating from the interface is fundamentally dependent on dimensionality of the boundary . ) The two cell surfaces are represented by elastic disks , as shown in Fig 1B , subject to bending resistance and approximate inextensibility . These disks are held by boundary tension σ0 in their plane , and separated by approximately inextensible nonspecific molecules of size Δz∞ , which we refer to as the far-field separation . These non-specific adhesions are absent from a region of radius rfree around the center of the disk , which we identify as the site of the receptor . We assume both intracellular and extracellular fluids are Newtonian with viscosity of water , η = 10−3 Pas . At the small length scales in our simulation , of ∼nm , the viscosity of the cytosol can be one or two orders of magnitude larger [38] , and at large length scales in our simulation , the viscosity is even larger . The variability of viscosity , and its dependence on length scale of observation , is an active area of research and is attributed to the heterogeneous content of the cytoplasm and ordering of water [38] . ( We do not a priori include here the concept of “effective viscosity” to describe the phenomenon of slower timescales due to fluid confinement [22] , since we anticipate these emerge from the dynamics naturally . ) Thus , all times we report are underestimates , and dynamics at more realistic viscosity and cell separation are expected to be slower . Due to the linear nature of the fluid dynamics equations we use , all times scale linearly with viscosity . To simulate this model , we use an implementation of the Stochastic Immersed Boundary framework . We largely overcome the numerical challenges mentioned above , allowing us to simulate with parameters within the order-of-magnitude of experimentally estimated values , shown in Table 1 . This framework , discussed in more detail in Methods and in the Supplemental Material , numerically approximates the fluid in an Eulerian representation , discretized in a rectangular Cartesian grid , while approximating the structure ( in this case , the one or two membranes ) in a Lagrangian representation , discretized as a triangulated mesh . As a control , we simulate a single membrane with thermal undulations , being held in place by adhesion molecules attached to fixed points in the fluid , as if it were attached to a “ghost” membrane . This simulation could be identified , for example , with a situation in which a cell is adhered to a highly permeable surface like a sparse network of extracellular matrix [43 , 44] that provides minimal hydrodynamic confinement . A snapshot top view is shown in Fig 2B . We find that the position of the receptor fluctuates as a Gaussian with standard deviation σ = 3 . 12nm ( 95% confidence interval [3 . 10 , 3 . 15]nm ) and an autocorrelation well-described by a single exponential decay with timescale τ = 1 . 05 × 10−6 s ( 95% confidence interval [1 . 02 , 1 . 08] × 10−6 s ) . Dynamics ( both deterministic and stochastic ) of a single membrane can be decomposed into modes , each with a timescale that , in some geometries , can be solved for explicitly [23 , 25] . The timescale of the nth mode associated with tension scales as ηrfree/σ0 n ∼ 10−6 s/n , and that the nth mode associated with bending scales as ηrfree/Bn3 ∼ 10−8 s/n3 , although both with significant prefactors . Thus , our computational results are broadly consistent with the dominant mode being the first mode associated with tension . We next simulate the interface with two membranes , as shown in Fig 3D . The membranes are held at Δz∞ = 60nm outside the free radius . We run simulations for 1 s . We observe a stationary probability with mean separation 〈Δz〉 = 70 . 0nm . This blistering by 10nm is due to an entropic repulsive pressure arising from thermal fluctuations [23 , 45 , 46] and is not observed in simulations where thermal fluctuations are removed ( shown below ) . We observe a relatively small change in the amplitude of fluctuation compared to the single-membrane case , from 3 . 2nm to 4 . 6nm ( 95% confidence interval [4 . 5 , 4 . 6]nm ) . The autocorrelation of Δz0 does not fit a single exponential , but rather fits a two-timescale decay ( black curve , Eq 21 ) with a fast timescale τfast = 5 . 3 × 10−7 s ( 95% confidence interval [5 . 0 − 5 . 4] × 10−7 s ) comparable to the single-membrane autocorrelation above , but also a slow timescale τslow = 8 . 2 × 10−5 s ( 95% confidence interval [8 . 1 , 8 . 3] × 10−5 s ) . The fraction of the autocorrelation function described by the slow process is c2 = 0 . 48 ( 95% confidence interval [0 . 43 , 0 . 53] ) . The double exponential equation we use to fit the autocorrelation is not a perfect fit , reflecting the inherent complexity of this process and the need for such computational modeling . This finding is in agreement with previous experimental work [11 , 23] showing that spatial amplitudes are not changed significantly , but fluctuation timescales are significantly altered by confinement . We again compare the timescale with previous estimates for a similar case that has been previously studied: a membrane near a wall [23] . The timescale of the nth mode associated with tension scales as η r free 4 / σ 0 ( Δ z ∞ ) 3 n 4 ∼ 10 - 4 s / n 4 ( see Eq . 2 . 18 in [23] ) . Again , our computational results are broadly consistent with the dominant mode being the first mode associated with tension . Since the rate of receptor triggering is determined by the timescale of proximity ( i . e . , sufficient for close contact between receptor and ligand ) , we next want to use the fluid dynamics simulations to estimate the mean first-passage time ( MFPT ) to proximity . Since these simulations include the target ligand only implicitly , we can infer the mean time to proximity for several values of Δ z 0 ⋆ . For the simulations with Δz∞ = 60nm , we ran fluid dynamics simulations for 1 second . For Δ z 0 ⋆ = 13 nm , proximity of Δ z 0 < Δ z 0 ⋆ was not observed , suggesting it is a rare event in the sense that it occurs on a timescale much larger than the fluctuation timescale . Computational expense prohibits us from simulating significantly longer times . To overcome this computational challenge of observing such rare close contacts , in this section , we develop an approximation based on Ornstein-Ulhenbeck ( OU ) processes [47] , and then use the Weighted Ensemble [48 , 49] computational method to find the mean first time to a particular state of the system , here defined as the first time for the membranes to be within a distance of Δ z 0 ⋆ of each other . Full details are in Methods and S1 Appendix . For the interface , we find that membranes will displace by 20 nm ( i . e . , the separation distance deviated from its mean of 70 nm down to 50 nm ) in approximately 10−2 s . The time until a displacement larger than this grows super-exponentially: for a displacement of 25 nm ( i . e . , down to separation Δ z 0 ⋆ = 45 nm ) , it takes ∼1 s . For the single membrane case , an analytical approximation exists for the single-component OU [50] , solid black line in Fig 4 , allowing us to confirm our computational method ( further validation is provided in S1 Appendix ) . In Fig 4 , the interface case apparently has a larger ( i . e . , slower ) MFPT for the single membrane . However , we note that these numbers are not directly comparable . The single-component OU describes the position of a single membrane , which has standard deviation σ1 = 3 . 1 nm , while the two-component OU describes the distance between two membranes , which has a standard deviation σ2 = 4 . 6 nm . A more direct comparison would be a hypothetical simulation in which two “single” membranes were held at a distance of 70 nm , but did not interact via fluid therefore would fluctuate independently . In such a case , the separation between these membranes would be σ indep = 2 σ 1 2 ≈ 4 . 4 nm , approximately the same as the interface standard deviation . Cells , including the T cell , continuously extend active processes driven by F-actin like filopodia and microvilli [42 , 51] that facilitate receptor binding [37 , 52] . To explore the effect of hydrodynamics on active processes at an interface , we simulate a force F at the receptor site , spread over a disk of radius a = 10 nm 1B . In Fig 5 , we find that a force of F = 20 pN is sufficient to drive proximity from a far-field separation of Δz∞ = 50 nm for both single membranes and interfaces . We perform deterministic simulations with thermal forces omitted ( black curves in Fig 5 ) . The dynamics are quantitatively similar , and the simulations are much less computationally taxing . For this reason , for the remainder of this section we perform simulations without thermal fluctuations . Note in Fig 5 the stochastic and deterministic simulations approach equilibrium on approximately the same slow timescale , but the equilibrium separation is larger when thermal forces are included due to the entropic repulsion discussed above . The shape of the protrusion is shown in Fig 5C and 5D . Membrane profiles are reminiscent of micrographs of microvilli in T cells ( see , e . g . , [37] Fig . 3G ) : The edges are rounded due to membrane bending resistance , and closest proximity is at the tip , with cell separation distance gradually tapering off . To isolate the influence of the thin-layer effect , we perform identical simulations with and without a second membrane , for various active forces , in Fig 6A . For a single membrane , the distance approaches a new equilibrium rapidly , ∼10−5 s . For an interface , Fig 6A demonstrates a rapid initial movement , ∼10−5 s , followed by a slower approach to the same equilibrium separation . In Fig 6B we explore this further by plotting the position of both membranes for F = 10 pN: We find that there is an initial rapid movement of the top membrane , i . e , the driven membrane ( blue curve in B ) ∼10−5 s , however this is accompanied by a rapid depression of the bottom membrane , i . e . , the passive ( red curve ) . Then , on a slower timescale ∼10−3 s , the passive membrane returns . We attribute the rapid depression to the incompressibility of the extracellular fluid , and the slow timescale to the thin-layer timescale identified above , as the excess fluid must drain from the interface . The plasma membrane is under tension , maintained by hydrostatic pressure and regulation of exocytosis , endocytosis and membrane ruffles [40 , 53] and is in the range of 3 − 300 pN/ μm [40 , 41] and is spatially nonuniform [54] . We apply membrane tension in our simulation as a boundary surface tension with magnitude σ0 . We find that higher surface tension necessitates more force for the equivalent equilibrium displacement , as shown in Fig 7A . This demonstrates that the system is above the critical length scale below which surface tension is insignificant compared to membrane bending [17 , 55] . Note that these these data show the equilibrium position in response to a constant force , therefore there is no effect of fluid dynamics , and thus no thin-layer effect . The results we report are sensitive to the properties of the large surface molecule , such as the size of the region rfree around the receptor that is free of these molecules . To put this parameter in a form more readily comparable with molecular surface densities , we define the parameter ρ = 1 / r free 2 , which has units of nm−2 . We previously estimated that receptor-ligand contact occurs in depletion zones with rfree ∼ 100 nm [6] . In Fig 7B , we explore the equilibrium separation as a function of force for various surface densities , rfree = 80 nm ( ρ = 1 . 6 × 10−4 nm−2 ) , rfree = 150 nm ( ρ = 4 . 4 × 10−5 nm−2 , rfree = 300 nm ( ρ = 1 . 1 × 10−5 nm−2 . As expected , higher density of surface molecules reduces the equilibrium displacement . The results we report are also sensitive to the molecular size Δz∞ of the large surface molecule . The large range of estimates for Δz∞ arises from the uncertainty about which molecules dominate the process of keeping the membranes apart . Molecules like CD45 may sterically maintain membrane separation by as little as 22nm . Non-specific adhesion molecules like LFA-ICAM and cadherins are estimated to span a range from 28nm [31] to 43nm [32 , 33] . Estimates for the thickness of the glycocalyx range from 40 − 50 nm [34 , 35] to 150 nm [36 , 37] . So , we explore receptor proximity driven by active forces , varying the far-field separation Δz∞ from 30nm to 120nm in Fig 8 . To clarity , as with previous simulations , the majority of the membranes ( everything outside the radius rfree ) are held apart at a fixed distance Δz∞ . From this fixed distance , an active protrusion is extended from one membrane towards the other . We examine the time dynamics of the protrusion . As the starting distance is increased , the time before proximity Δ z 0 < Δ z 0 ⋆ increases , shown in Fig 7B . However , we find that above a critical far-field separation Δz∞ ≈ 80 nm , membrane deformation speed increases , even though force is kept constant at F = 10 pN . The effect is modest but sufficient so that , over a large range of far-field separations ∼60 − 130 nm , the time to proximity does not increase for increasing separation , Fig 7C . Heuristically , this plateau arises because of a significant thin-layer effect dominates motion . Since this effect depends sensitively on the thickness of the thin layer , increasing the thickness reduces the effect , and the active protrusion can push more easily through the free fluid . On the other hand , although speed increases , the distance to the target cell also increases . Thus there is an optimal “attack distance” from which to extend a protrusion . If a membrane were perfectly water permeable ψ = ∞ there would be no thin-layer effect . Biological membranes are sufficiently permeable that , in fast motile cells , fluid velocity appears stationary in the lab frame of reference when myosin contractility is inhibited [19] . Therefore , it is a priori reasonable to expect that there is a magnitude of permeability above which the slow-timescale behavior of the interface is removed , leaving only fast dynamics . We repeat the active force simulations at F = 20 pN , and explore permeabilities at each order of magnitude , in Fig 9 . We find that the first significant deviation from impermeability ( ψ = 0 ) occurs at ψ ∼ 102 nm/sPa ( red dotted ) . By ψ = 104 nm/sPa , the top membrane time series is comparable to the single-membrane case ( Fig 6B ) , i . e . , very little thin-layer effect is observed . The transition to fast-only dynamics occurs through a reduction in timescale , and only a weak reduction in amplitude ( green curve is compressed horizontally ) . Previous theoretical studies [22 , 23] have estimated the parameter regimes in which permeation ( i . e . , flow across membranes ) will dominate parallel flows ( i . e . , flow parallel to the interface ) . Specifically , [22] finds that the critical scale of permeability , above which permeation is dominant , is ψ ∼ ( Δ z ∞ ) 3 η ( r free ) 2 ∼ 10 4 nm / s Pa ( 2 ) where we have changed their equation to our notation , and assumed that the dominant wavelength is ∼ ( 1/rfree ) . This scaling is in agreement with our simulations . Moreover , this is orders of magnitude larger than the permeability estimates 10−2 nm/sPa [38] . The largest indirect estimates from motile epithelial keratocytes gives ∼101 nm/sPa [19] . So , taken together , our results suggest the thin-layer effect cannot be abrogated by physiological levels of permeability . | The elastohydrodynamics of water in and around cells is playing an increasingly recognized role in biology . In this work , we investigate the flow of extracellular fluid in between cells during the formation of a cell-cell contact , to determine whether its necessary evacuation as the cells approach is a rate-limiting step before molecules on either cell can interact . To overcome the computational challenges associated with simulating fluid in this mechanically soft , stochastic and high-aspect-ratio environment , we extend a computational framework where the cell plasma membranes are treated as immersed boundaries in the fluid , and combine this with computational methods for simulating stochastic rare events in which an ensemble of simulations are given weights according to their probability . We find that the membranes fluctuate independently with a characteristic timescale of approximately microseconds , but that as the cells approach , a new , slower timescale of approximately milliseconds is introduced . Thermal undulations nor typical amounts of membrane permeability can overcome the timescale , but active forces , e . g . , from the cytoskeleton , can . Our results suggest an explanation for differences in molecular interactions in live cells compared to in vitro reconstitution experiments . | [
"Abstract",
"Introduction",
"Results"
] | [
"classical",
"mechanics",
"fluid",
"mechanics",
"permeability",
"viscosity",
"developmental",
"biology",
"membrane",
"receptor",
"signaling",
"materials",
"science",
"molecular",
"development",
"materials",
"physics",
"cellular",
"structures",
"and",
"organelles",
"chemical... | 2019 | Hydrodynamics of transient cell-cell contact: The role of membrane permeability and active protrusion length |
Periplasmic binding proteins ( PBPs ) are a large family of molecular transporters that play a key role in nutrient uptake and chemotaxis in Gram-negative bacteria . All PBPs have characteristic two-domain architecture with a central interdomain ligand-binding cleft . Upon binding to their respective ligands , PBPs undergo a large conformational change that effectively closes the binding cleft . This conformational change is traditionally viewed as a ligand induced-fit process; however , the intrinsic dynamics of the protein may also be crucial for ligand recognition . Recent NMR paramagnetic relaxation enhancement ( PRE ) experiments have shown that the maltose binding protein ( MBP ) - a prototypical member of the PBP superfamily - exists in a rapidly exchanging ( ns to µs regime ) mixture comprising an open state ( approx 95% ) , and a minor partially closed state ( approx 5% ) . Here we describe accelerated MD simulations that provide a detailed picture of the transition between the open and partially closed states , and confirm the existence of a dynamical equilibrium between these two states in apo MBP . We find that a flexible part of the protein called the balancing interface motif ( residues 175–184 ) is displaced during the transformation . Continuum electrostatic calculations indicate that the repacking of non-polar residues near the hinge region plays an important role in driving the conformational change . Oscillations between open and partially closed states create variations in the shape and size of the binding site . The study provides a detailed description of the conformational space available to ligand-free MBP , and has implications for understanding ligand recognition and allostery in related proteins .
Periplasmic Binding Proteins ( PBPs ) are major components of the bacterial cell envelope that are involved in nutrient uptake and chemotaxis [1] , [2] . Gram-negative bacteria use PBPs to transport ligands into the cytosol by association with a membrane-bound ATP-binding cassette ( ABC ) transporter [3] . Gram-positive bacteria differ in that they employ a slightly different design , in which the PBPs motif is directly attached to a membrane-anchored receptor . In addition , several mammalian receptors contain extracellular ligand binding domains that are homologous to PBPs . These include glutamate/glycine-gated ion channels such as the NMDA receptor; G protein-coupled receptors , including metabotropic glutamate , GABA-B , calcium sensing , and pheromone receptors; and atrial natriuretic peptide-guanylate cyclase receptors . Many of these receptors are promising drug targets [4] . The structures of PBPs ( ∼100 X-ray structures ) have been called a ‘gold mine’ for studying the general mechanisms of protein-ligand recognition [2] , as PBPs have been identified that can transport a large variety of substrates , including: carbohydrates , amino acids , vitamins , peptides , or metal ions [5] . The affinity of PBPs for diverse substrates also make them ideal templates for the design of diverse in vitro and in vivo biosensors with tailored properties [6] . Maltose binding protein ( MBP ) is a part of the maltose/maltodextrin system of Escherichia coli , which is responsible for the uptake and efficient catabolism of maltodextrins . MBP is the prototypical member of the PBP superfamily . It has been the subject of extensive study due to its importance in various biological pathways [2] , [6] , and its utility as an affinity tag for protein expression and purification [7] . The protein folds into two domains of roughly equal size: the C terminal domain ( CTD ) , and the N terminal domain ( NTD ) ( Fig . 1 ) . The two domains are connected via a short helix and a two-stranded β-sheet that form an interdomain hinge region . Like other PBPs , the binding site of MBP is located on the interdomain cleft between domains . X-ray structures of MBP solved in the presence and absence of ligand indicate that the protein undergoes an important conformational change from an open to a closed state in the presence of the ligand , the effect of which is to better stabilize the ligand by reducing the size of the cleft [2] . The conformational change has been dubbed the “Venus Fly-Trap Mechanism” [8] due to its resemblance to the traps on the carnivorous plant that closes only when stimulated by prey . An ‘induced-fit’ mechanism [9] is often invoked to describe the ligand recognition process . In this scenario , the ligand participates in remodeling the binding site by interacting directly with the protein . Alternatively , it is also possible that the apo protein already exists in a mixture of open and closed conformations . In which case , the ligand would play a more passive role shifting the equilibrium toward the closed state , a mechanism traditionally described as ‘conformational selection’ , or ‘population shift’ [10] . Computer simulations and NMR studies are often needed to distinguish between the two scenarios , as X-ray structures typically do not provide detailed information about the ensemble of conformations available to the ligand-free protein [11] . Until recently , the bulk of our understanding of the mechanism of substrate recognition in MBP came from crystallographic studies that indicated only two possible conformations , a ligand-free open , and a ligand bound closed structure . In 2007 , Tang et al . [12] reported the first NMR paramagnetic relaxation enhancement ( PRE ) measurements on apo MBP . By attaching a spin label ( TEMPO ) on the NTD and CTD of the apo protein , domain hinge-bending motions could be studied . These measurements indicated the existence of a dynamic equilibrium between a major open state , and a minor partially closed state . Because experimental PRE rates for MBP could not be explained either by the X-ray crystal structure of the apo-state ( open conformation ) , nor by the ligand-bound closed-state , it was possible to postulate that a partially closed structure exists . The transition between an open , and a partially closed state , was determined to involve a rotation around the hinge region . The best agreement between computed and experimental PRE and Residual Dipolar Coupling ( RDC ) data , was obtained by considering that substrate-free MBP exists in equilibrium between a major , open state ( 95% ) and a minor , semi-closed state , populated ∼ 5% of the time , which corresponds to a very small energy difference between the two states ( ∼2 kcal/mol ) . The time-scale of the exchange between states was estimated to be between 20 ns to 20 µs [12] . From a theoretical point of view , it is understood that a pre-existing equilibrium between different PBP conformations could play an important role in facilitating ligand recognition [13] . However , it remains a considerable challenge to access , using fully atomistic MD simulations , a detailed statistical analysis of slow conformational dynamics in proteins mediated by such hinge-bending motions . In the past decade , the development of increasingly efficient simulation algorithms has led to a large number of theoretical studies using Molecular Dynamics ( MD ) simulations to probe the intrinsic dynamics of PBPs [14] , [15] , [16] , [17] , [18] , [19] , [20] , [21] , [22] . In 2003 , Pang et al . [14] studied the glutamine binding protein ( GlnBP ) using ∼5 ns MD simulations . They observed large vibrations in the apo protein in the direction of a closed structure , and found that the open apo structure was more flexible than the closed structure . Subsequently , Pang et al . [16] confirmed that this is a general result by performing a comparative study of different PBPs , which also showed that different PBPs could display slightly different dynamical properties . The authors also observed that the opening and closure rate in the presence of a substrate could be fast ( nanosecond time-scale ) , even though they also noted that obtaining converged sampling for the opening and closure events was challenging on the nanosecond time-scale . In 2006 , Kandt et al . [17] performed longer MD simulations on BtuF , a protein involved in vitamin B12 uptake . Using 12 simulations of 30–50 ns each , they were able to observe the initiation of opening and closing motions in both apo and holo simulations , with larger motions in the apo simulations . This behavior of the protein was interpreted to be compatible to the Venus Fly-trap model . The observation of enhanced molecular flexibility in the open state was confirmed by other groups for similar PBPs , such as the iron binding proteins FhuD [18] , and FitE [19] , and the heme binding proteins , ShuT , and PhuT [20] . In 2009 , Loeffler and Kitao [21] studied GlnBP in the open liganded form , and reported closing events occurring during the simulations . Taken together , these MD studies on PBPs have helped characterize the intrinsic flexibility of ligand-free PBPs on the nanosecond time-scale . The consensus of opinion from these calculations is that the ligand recognition proceeds through a Venus flytrap mechanism , and that the apo PBPs structure is very flexible , with a tendency to oscillate along the modes that lead from the open to the closed structure . In 2005 , a simulation study of the MBP protein was carried out by Stockner et al . [22] . Using 4 MD simulations of 30 ns , started from both open and closed states , with and without substrate , the authors could show that the ligand-free MBP structure ( open form ) naturally evolves toward a closed state in the presence of a substrate . Similarly , the closed state was found to evolve toward an open state when the substrate was removed . The rapid time-scale of this conformational change was consistent with experimental rate constant for sugar binding ∼1–2×107 M−1 s−1 [23] , which suggests a rate of closure around 30–50 ns . However , the time-scale of the simulations was too short to observe any pre-existing equilibrium in apo MBP between an open and a partially closed conformer . This can be explained by the presumed slow exchange rate ( 20 ns to 20 µs ) [12] between the two conformations of apo MBP . In this paper , we have used accelerated Molecular Dynamics ( aMD ) simulations [24] of MBP to show that the apo protein exists in a dynamical equilibrium between an open and a semi-closed conformation . A number of methods have been developed to enhance the sampling of slow conformational changes in proteins , including targeted MD [25] , and conformational flooding [26] . However , within the framework of this study , a key advantage of aMD is that it allows us to study the conformational behavior and dynamics of the protein without using a pre-defined reaction coordinate . In previous studies , aMD has been successfully employed to study slow time-scale dynamics in proteins , such as HIV-protease [27] , ubiquitin [28] , IKBA [29] and H-Ras [30] . The enhanced conformational space sampled by aMD has also been shown to significantly improve the theoretical prediction of experimental NMR observables , such as residual dipolar couplings , scalar J-couplings [28] and chemical shifts [29] , which are sensitive to dynamic averaging on the micro- to millisecond time-scale . In this paper , we show that aMD simulations successfully allow the study of the transition from the open state of apo MBP to the hidden semi-closed conformation . This provides the first atomistic view of the transition between open and partially closed states in a PBP . NMR parameters computed from the simulations agree well with experiments . Free energy calculations , and continuum electrostatics calculations are used to provide new insights into the mechanism and energetics of the exchange between the open and semi-closed states of apo MBP .
In the framework of the Solomon-Bloembergen Model-Free formalism ( SBMF ) [47] , the transverse paramagnetic relaxation enhancement rate , Γ2 , between the electron located at the paramagnetic center associated with the spin-label and a 1H nucleus is given by: ( 1 ) where fSBMF is defined as: ( 2 ) In the above equations , r is the distance between the electron and 1H nucleus , s is the electron spin quantum number , g is the electron g-factor , γI is the proton gyromagnetic ratio , μ0 is the permeability of a vacuum , μB is the magnetic moment of the free electron , S2 is the order parameter associated with the interaction vector between the electron and the 1H nucleus and ωI/2π is the Lamor frequency of the proton . The reduced generalized spectral density function , JSBMF ( ω ) is given by: ( 3 ) The associated correlation time , τ , is either τc or τt , which are defined as: ( 4 ) and ( 5 ) where τr is the rotation correlation time of the molecule , and τi is the correlation time for internal motion of the interaction vector between the electron and the 1H nucleus assuming that the internal motion is not coupled to the overall tumbling of the molecule . τs is the effective electron relaxation time , which in the present work is ignored as it is assumed to be substantially longer than the rotation diffusion time of the molecule and so has a negligible effect on the associated spectral density function [48] . The reader is referred to Ref . 47 for complete derivations and discussion of the above equations . The accelerated molecular dynamics simulations identified two states for apo-MBP ( see Results ) : A major open state and a minor semi-closed state . For each state , six structures were obtained from the aMD simulations and in each case , the Asp41 residue was modified to a non-standard CYS-TEMPO spin-label residue . The force-field for this non-standard residue was generated using the AMBER force-field ( GAFF ) [49] . A simulated annealing protocol was performed with Cartesian restraints applied to all solute atoms other than those associated with the modified CYS-TEMPO residue , in order to generate different configurations for the CYS-TEMPO group . Four different CYS-TEMPO configurations were generated for each of the six initial starting structures in each state . In this way , 48 systems were generated ( 24 systems in each state ) . Each system was placed at the center of a pre-solvated water box and brought to equilibrium at 300K , 1-bar pressure . A 20-ns production MD run was performed for each system and the average of the inverse distances to the power six , <r−6> , the associated order parameters , S2 , and internal correlation times , τi , for the interaction vector between the electron located at the nitroxide group of the TEMPO spin-label and all amino-1H nuclei were calculated . Transverse paramagnetic relaxation rates for each state ( open and semi-closed ) were calculated from each simulation using equation ( 1 ) and averaged over the 24 simulations . Assuming the relative population of the open and semi-closed states to be 95%-5% ( see MM/PBSA analysis ) the final theoretical free-energy weighted combined open/semi-closed PRE data was obtained .
Conventional and accelerated MD simulations were performed to study the conformational behavior and dynamics of MBP . When started in the open state ( Sim 1 and Sim 3 ) , conventional simulations ( 50 ns ) were unable to explore a new conformation , as shown by the limited range of interdomain angles observed during theses simulations ( Table 1 ) . The MBP structure remained open throughout the simulation at an interdomain angle , θ constant between 154 and 161° ( Fig . 2 ( a ) ) . In comparison , accelerated MD simulations , initiated from the open state ( Sim2 , and Sim4 ) , transition to a semi-closed state , indicated by a change of ∼ 3 Å in the Cα RMSD , followed by a sharp decrease in θ from ∼ 160° to ∼ 145° . These simulations were repeated 5 times , and in all cases were shown to visit a partially closed state and to eventually come back to the open state . The interdomain angle during a typical aMD trajectory is drawn in Fig . 2 ( a ) , as a function of the number of aMD steps . The number of aMD steps is used in this plot , since the ‘real’ time-scale in aMD is non-linear , and difficult to assess accurately . However , previous work on proteins [27] , [28] , [29] , [30] has suggested that the level of acceleration used in this study corresponds to sampling the protein motion on the micro- to millisecond time-scale . NAMD2 simulations with CHARMM force field were performed for comparison , and an identical transition to a semi-closed structure was observed . Standard MD simulations were then started from the semi-closed conformation , to show that this new conformational state is indeed stable . These additional simulations displayed a stable interdomain angle ( at average values of θ = 146 ( 3 ) ° , and 148 ( 3 ) ° , for sim5 , and sim6 , respectively ) . Interestingly , these θ values are exactly intermediates between the interdomain angles of the X-ray structures in the fully open ( θ = 161° ) and fully closed ( θ = 137° ) conformations . Moreover , the structures obtained with CHARMM and AMBER were almost identical when superimposed , with a backbone RMSD <1 . 5 Å . This should be compared to the larger RMS deviations observed with respect to the open ( 3 . 0–3 . 5 Å ) , and closed forms ( 2 . 0–2 . 5 Å ) . To further characterize the conformational space sampled in the simulations , and show that the identified semi-closed state occupies a local energy minimum on the PES , the trajectories were analyzed in terms of their principal components of atomic fluctuations [50] . All publicly available X-ray structures of MBPs were used to perform a principal components ( PC ) analysis ( see Ref . [51] for details ) supplementary material for details ) . The first two PCs were found to capture >95% of the variance associated with the open and closed structures , and were therefore used to further analyze the trajectories . As can be seen in Fig . 2 ( b ) , the X-ray structures display two well-defined clusters on the PC-space that correspond to the open ( ligand-free ) and closed ( with bound ligand ) conformers . A representative MD simulation initiated in the open state ( sim1 ) is shown to explore only the local vicinity of the open X-ray structures , whereas a representative aMD simulation ( sim2 ) is able to explore a region of the PC-space that is intermediate between the open and closed form . In addition , when starting a MD simulation from the semi-closed conformer ( sim5 ) , the simulation remains in the basin corresponding to the semi-closed conformer , which indicates that a stable minimum on the potential energy surface exists for this structure , between the fully open and fully closed states . After establishing the existence of a hidden semi-closed state of MBP , the trajectories were further analyzed to gain new insight into the pre-existing equilibrium dynamics between the open and semi-closed states . Visual inspection of the trajectories reveals that a flexible part of the protein , called the balancing interface ( Fig . 1 ) , is displaced during the transformation . The balancing interface is a beta-hairpin motif that belongs to the CTD , and which interacts only through weak vdW interactions with the NTD . Prior to the conformational change , the contact between the tip of the balancing interface and the NTD is lost . This event is generally followed by a rotation of the two globular domains of ∼ 15° around the hinge β-sheet residues . To establish the role of the balancing interface in the conformational change , the ABF method was used to introduce an external force that artificially pulls the balancing interface away from the NTD . 5 ABF simulations were carried out for 50 ns , starting from the open conformer . In each of these simulations , once the balancing interface was no longer interacting with the CTD , a spontaneous conformational change occurred leading to the semi-closed state . In addition , 5 ABF simulations were started from the semi-closed conformer . In these simulations , as the distance between the tip of the balancing interface and the NTD was decreased , the structures reverted back into the open conformer , showing that the process is reversible . These simulations also provide information about the free energy barrier for the displacement of the balancing interface ( ∼2 kcal/mol ) ( Fig . 3 ( a ) ) . The involvement of the balancing interface in the transition to the semi-closed state is consistent with two mutational studies that have previously suggested that it plays a role in the conformational change of MBP [52] , [53] , and described the balancing interface as a “spring” that maintains the protein open . Moreover , X-ray crystal structures of both the apo ( open ) and holo ( closed ) states of MBP show that the contact between the NTD and the balancing interface is lost only in the closed state [54] , [55] , [56] . Our simulations are consistent with this representation , and indicate that the balancing interface acts as a “molecular switch” that initiates the conformational change when displaced . To provide insight into the energetics of the transformation , we have analyzed the relative free energy changes for each residue as a function of the conformational state . This was done by performing an MM/GBSA per residue free energy decomposition analysis for both the open and semi-closed states . The results , presented in Fig . 3 , indicated as expected that residues near the tip of the balancing interface help stabilize the open conformation . In contrast , several non-polar residues near the hinge region were found to help stabilize the semi-closed conformation , by re-packing and dramatically reducing their solvent-accessibility during the conformational change . In particular , residues Ile178 and Leu311 were found to be solvent-exposed in the open state , but buried inside the protein in the semi-closed state ( Fig . 3 ( a ) ) . Ile178 was able to interact with another non-polar residue , Ile333 , in the semi-closed state . The composition of the balancing interface is interesting in this respect , because it contains several hydrophobic residues that are all located on one side of the beta hairpin motif ( Phe169 , Ile178 , Val181 , Val183 ) , while the other side contains side-chains that are hydrophilic ( Lys170 , Glu172 , Lys175 , Asp177 , Lys179 , Asp180 , Asp184 ) . Visual inspection of the trajectories suggests that the opening of the flap momentarily increases the solvent accessibility of non-polar residues in the hinge region ( Ile178 , Val181 , Ile329 , Ile333 ) . This most likely destabilizes the open structure and helps drive the conformational transition to the semi-closed state . The proposed mechanism is consistent with the mutational study of Telmer and Shilton [52] , which has previously shown that certain mutations in MBP far away from the sugar binding pocket can increase the affinity for maltose and reduce the off-rate . These mutations ( Met321 , Gln325 , and residues 171–177 ) were found to occur in residues that are important for protecting non-polar regions of the protein from the solvent , in particular residues Ile178 , and Leu311 . The picture that emerges is one in which the improved binding affinity for maltose in the mutants could be explained by the fact that these mutants are more likely to find themselves in a semi-closed state to avoid a large exposure of hydrophobic regions of the protein . To learn more about the different energy terms , the various enthalpic and entropic contributions to the total free energy in the open and semi-closed states of MBP were further analyzed using MM-PBSA calculations . The results are shown in Table 2 , ( force field terms for the protein , polar and non-polar solvation , and entropy term derived from a normal mode analysis ) . In agreement with the experimental study of Tang et al . [12] , the two conformations were found to have roughly the same total energy , and the open structure was found to be slightly more stable ( ΔG ∼ −2 kcal/mol ) . Interestingly , the protein electrostatic energy was found to help stabilize the open conformational state , which as discussed above , is likely to be due to the favorable interaction between the tip of the balancing interface and the NTD . In contrast , solvation effects were found to favor the semi-closed state to a similar degree . This analysis suggests that the open conformation is more stable because the favorable electrostatic interactions in the balancing interface are slightly superior to the destabilizing hydrophobic effect . However , as the balancing interface moves into solution , the repacking of non-polar residues leads to a conformational change into the slightly more compact semi-closed state . Interestingly , close-range interactions between the two domains were found to be very limited . In a previous study , Stockner et al . [22] found that , during the transformation into the closed state in the presence of a substrate , a salt-bridge between the Glu111 ( hinge region ) and Lys15 ( NTD ) residues can assist the conformational change in holo MBP transition into the closed state . The authors also reported a temporary interaction between Tyr155 ( CTD ) and the backbone carbonyl of Glu111 . Here , we observed a similar interaction between Glu111 and Lys15; however , we also find that the salt-bridge is not always formed in the semi-closed state . The interaction between Tyr155 and Glu111 is also observed and appears to help stabilize the semi-closed conformation ( Fig . 4 ) . In addition , a second stabilizing hydrogen bond was discovered between Asp65 ( NTD ) and Trp340 ( CTD ) . These two interactions , shown in Fig . 4 , help stabilize the semi-closed state . As previously discussed , nuclear magnetic resonance-based PRE data provides very accurate information about the conformational behavior of proteins . Having identified a pre-existing dynamic equilibrium for apo MBP between an open and semi-closed state from the aMD simulations and calculated the relative free energies of these states using an MM/PBSA analysis , the validity and accuracy of our theoretical results can be determined by back-calculating the experimental data . To confirm that the simulations can sample accurately the conformational space of MBP , we have computed the PRE rates directly from structures obtained in MD trajectories . In Fig . 5 , we compare the PRE calculated for several simulations started from the open conformations , and for the case where the back-calculated data is the product of averaging over all the MD simulations ( see PRE Methods section for details ) . The agreement with experiment was significantly improved in that case ( Fig . 5 ) . In addition to the NMR PRE data , there is a very high degree of structural similarity between the semi-closed state that we obtain from our aMD simulation study and distances reported in the study of Tang et al . For example , the distance between residues Glu153 ( CTD domain ) and Asp14 ( NTD domain ) was reported to be ∼ 13 . 8 Å , in the semi-closed state . Excellent agreement was obtained here in our simulations in the semi-closed state ( 14 . 5±0 . 9 Å ) . Moreover , this value is quite different from the average distance measured in X-ray structures corresponding to the open , or closed conformations ( 17 . 1–19 . 3 Å , and 9 . 5–10 . 2 Å , for open , and closed structures , respectively ) , which further demonstrates that the semi-closed conformation is a distinct state .
Our computer simulations confirm the results of Tang et al . and solidly establish the existence of a preexisting dynamical exchange in MBP between an open , and a semi-closed conformational state . In order to rapidly exchange between states , the protein must exist in two conformations with similar free energies that are connected by a relatively flat free energy pathway . In principle , this could be achieved by using a variety of residues to assist the transformation , such as , “hook-and-eye residues” that can connect different regions of the protein during an exploration phase , or “anchor-and-latch residues” that may be important in a later stage to lock the protein into a stable conformation . In MBP , we have found that large hydrophobic residues are important because they can be used to help stabilize the partially closed conformation . This may be a principle used by other hinge-bending proteins to increase compactness , as the solvent accessibility of hydrophobic side-chains is easier to reduce in compact structures [57] . Interestingly , in MBP the open state is slightly more stable than the semi-closed state , which raises the question of how this is achieved . To answer this question , we performed continuum electrostatic calculations , and found that electrostatic interactions between the balancing interface and the NTD help to maintain the open structure . Similarly , the displacement of the balancing interface into the solvent was generally followed by a spontaneous conformational transition into the semi-closed state . Thus , the picture that emerges is one in which the re-packing of non-polar residues >20 Å away from the ligand binding site can affect its properties . This observation is consistent with a large number of mutational studies showing the important structural role played by non-polar residues near the hinge region [52] , [53] , [58] , [59] , [60] . From a design perspective , the important role of hydrophobic residues in facilitating the hinge-bending motion could be seen as counter-intuitive , as one could expect short-range electrostatic interactions between the two globular domains to be more effective than solvation effects . For example , electrostatic interactions are known to be very effective in protein-protein recognition [61] , [62] , enzymatic catalysis [63] , ion channel selectivity [64] , and protein-ligand interactions [2] , [65] . In contrast , the simulations suggest that close-range electrostatic interactions do not contribute significantly to the hinge bending motion , as the associated contacts between the NT and CT domains are very limited . Several intra-domain salt-bridges near the binding site were seen to break and reform during the simulations ( such as Glu153 and Arg344 , Glu45 and Arg66 , Arg367 and Asp363 ) , but the formation of salt-bridges between the two domains did not occur . Interestingly , a similar conclusion was reached by Sinha et al . [66] , who analyzed the role of salt-bridges from a database of 36 hinge-bending proteins . They found that close-range electrostatic interactions are largely absent between moving domains . Instead , large non-polar buried surface areas were often found to be used to create a smooth energy landscape where thermal fluctuations can drive hinge-bending motions . Our simulations of the apo protein indicate that the fully closed state is never reached in the absence of a ligand . This suggests that an induced-fit step may still be required to reach the fully closed state . The existence of a pre-existing equilibrium dynamics in MBP between open and semi-closed conformations is most consistent with a two-step mechanism for ligand-recognition , in which the first step involves the formation of an encounter complex in the semi-closed state via an population shift mechanism , followed by a fast induced fit rearrangement , resulting in the fully closed state . Interestingly , the pre-existing dynamical equilibrium between the open and semi-closed states may not be shared by all PBPs [67] . Studies by Bermejo et al . of GluBP indicate that this system does not seem to achieve a ligand-free open to partially closed transition [67] . The existence of two conformations may be advantageous in MBP in order to lower the substrate selectivity , and facilitate the recognition of a large variety of maltooligosaccharides , with up seven glucose units [54] . Future studies will be needed to show if the existence of a dynamical equilibrium between multiple stable conformations really provides a functional advantage in hinge-bending proteins . It is also possible that the existence of multiple low energy states facilitate allosteric effects . For example , we found that a displacement of the balancing interface , 20 Å away from the ligand-binding site , had a direct influence on the shape and size of the binding site . In conclusions , in this paper we presented a combination of conventional and accelerated MD simulations that have been used to gain insight into the pre-existing dynamical equilibrium of conformational states of apo-MBP . The simulations showed that: ( 1 ) The previously unobserved semi-closed state is stable on the nanosecond time-scale . Moreover , excellent agreement was obtained between computed PRE rates and experiments , when considering the relative population of the two states at room temperature . ( 2 ) Visual inspection of the trajectories and free energy calculations indicated that the balancing interface is displaced during the transformation and acts as a switch that mediates the open- to semi-closed conformational transition . ( 3 ) The sharp transition between the open and semi-closed forms is consistent with a model in which the apo protein is in equilibrium between only two , well-defined conformational states . Solvation effects and the packing of non-polar side chains were found to assist the transformation . This interpretation is consistent with a large number of mutational and NMR studies showing an increased affinity for maltose when large non-polar residues are introduced in the hinge region [53] , [58] . Because the semi-closed state has a much lower population in apo MBP , it has not been observed previously in X-ray crystal structures . However , the size and shape of the binding site in the semi-closed state appears to be ideally suited for binding to maltodextrins . Therefore , it is possible that the semi-closed conformation , although less populated , is highly relevant for substrate recognition . Ongoing work in our group is focusing on exploring the affinity of both the open and the semi-closed conformation of MBP for different substrates . The simulations presented here provide a structural basis that can be used in future studies to explore the mechanism of substrate recognition in MBP , and in other PBPs . Future studies will reveal if other family members can exchange between open and partially closed conformations , and how to best utilize this phenomenon in drug design protocols . | Maltose-binding protein ( MBP ) is a bacterial protein involved in nutrient uptake . Crystallographic studies have revealed two stable conformations: a ligand-free open form and a liganded closed form . The interconversion between the 2 forms has been traditionally viewed as a ligand “induced” process . However , recent NMR experiments have suggested that the ligand-free protein is flexible enough to visit partially closed states . Because these states may display higher affinity for the ligand than the open state , ligand binding may proceed via a “conformational selection” of the most geometrically and chemically favored conformer . Mechanistic details of this process remain elusive as atomic structures of relevant intermediate conformations are currently lacking . In this study , we use atomistic simulations to characterize the flexibility of the MBP protein , and confirm the existence of a hidden semi-closed state . The relative stabilities of the two states is predicted and corroborated with existing experimental data . A key finding of the study is that the closed form of the protein – adopted by the ligand-bound form – is not observed in the simulations of the ligand-free protein . This has implications for understanding ligand binding mechanisms in MBP and related proteins . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"biotechnology",
"biomacromolecule-ligand",
"interactions",
"molecular",
"dynamics",
"classical",
"mechanics",
"statistical",
"mechanics",
"gibbs",
"free",
"energy",
"enthalpy",
"biological",
"systems",
"engineering",
"newton's",
"laws",
"of",
"motion",
"protein",
"structur... | 2011 | Accessing a Hidden Conformation of the Maltose Binding Protein Using Accelerated Molecular Dynamics |
Paramyxoviruses can establish persistent infections both in vitro and in vivo , some of which lead to chronic disease . However , little is known about the molecular events that contribute to the establishment of persistent infections by RNA viruses . Using parainfluenza virus type 5 ( PIV5 ) as a model we show that phosphorylation of the P protein , which is a key component of the viral RNA polymerase complex , determines whether or not viral transcription and replication becomes repressed at late times after infection . If the virus becomes repressed , persistence is established , but if not , the infected cells die . We found that single amino acid changes at various positions within the P protein switched the infection phenotype from lytic to persistent . Lytic variants replicated to higher titres in mice than persistent variants and caused greater infiltration of immune cells into infected lungs but were cleared more rapidly . We propose that during the acute phases of viral infection in vivo , lytic variants of PIV5 will be selected but , as the adaptive immune response develops , variants in which viral replication can be repressed will be selected , leading to the establishment of prolonged , persistent infections . We suggest that similar selection processes may operate for other RNA viruses .
Paramyxoviruses are primarily known for the acute infections and associated diseases they cause , such as mumps , measles and respiratory illnesses . However , under certain conditions they can establish persistent ( or prolonged ) infections [1] , which can be considered as infections that continue for longer than would be expected from a prototypical acute infection ( 2–3 weeks ) . For example , both immunocompromised and immunocompetent patients have been shown to shed parainfluenza virus ( PIV ) 2 , 3 and 4 for weeks and even years after infection [1] , whilst dogs infected with canine distemper virus can shed virus for months after initial infection [2] . Prolonged/persistent infections can lead to chronic diseases , for instance subacute sclerosing panencephalitis ( SSPE ) is associated with measles [3] , postviral olfactory dysfunction is associated with PIV3 [4] and chronic kidney disease is associated with feline paramyxovirus [5 , 6] . Little is understood of the molecular mechanisms by which paramyxoviruses establish and maintain persistent infections . Like other viruses , they must avoid elimination by the host immune response , while maintaining their genomes in at least some infected cells . Since it is highly probable that continual high-level viral replication in a cell will either directly kill the cell or lead to viral recognition and killing by innate and adaptive immune responses , it is likely that viral replication must be at least partially repressed in some cells in order for a virus to establish a persistent infection [7] . The interferon ( IFN ) response , and the ability of viruses to circumvent it , is also likely to play an important role in the establishment and maintenance of persistent infections . Thus , patients who have a defective ability to respond to type I IFN can become persistently infected with measles , mumps and rubella viruses following MMR vaccination with serious sequelae [8 , 9] . On the other hand , we have suggested that the IFN response may also dampen down viral replication in some cells , thereby facilitating persistence [10 , 11] . Using parainfluenza virus type 5 ( PIV5 ) as a model , we show here that viral replication at late times after infection can also be repressed in an IFN-independent mechanism , thereby leading to the establishment of persistent infections . PIV5 ( previously known as simian virus 5; species Mammalian rubulavirus 5 , genus Rubulavirus , family Paramyxoviridae ) has been isolated from numerous mammals including , humans , primates , pigs , ungulates ( cattle ) , dogs [12 , 13] and lesser pandas [14] . There is also some evidence that cats , hamsters , rabbits and guinea pigs can be infected [15 , 16] . The association of PIV5 with acute disease is often not obvious , although the virus causes kennel cough in canines [17] and may , or may not , cause acute respiratory symptoms in pigs [18 , 19] and calves [20] . PIV5 can also cause unsuspected persistent infections of tissue culture cells , including the AGS line [21 , 22] , guinea pig kidney and mouse fibroblast cells [23] , and is likely to establish persistent infections in vivo [24–26] . PIV5 has a non-segmented , negative-sense RNA genome of 15 , 246 nucleotides ( nt ) containing seven tandemly arranged genes , that encode eight proteins , flanked by 3’-leader and 5’-trailer sequences at the genome ends . From the 3’-leader sequence , the genome encodes the nucleocapsid protein ( NP ) , V protein ( V ) , phosphoprotein ( P ) , matrix protein ( M ) , fusion protein ( F ) , small hydrophobic protein ( SH ) , haemagglutinin-neuraminidase ( HN ) , and the large protein ( L ) . The genomic RNA is encapsidated by NP , forming a flexible helical nucleocapsid complex that is associated with the viral RNA-dependent RNA polymerase complex ( vRdRP ) consisting of L and P ( extensively reviewed in ref [27] ) . Previous work has highlighted the importance of the phosphorylation of the P protein in regulating the activity of the vRdRP and influencing viral growth [28] . Furthermore , P plays a role in limiting the induction of host cell responses by influencing the fidelity of viral RNA synthesis [29] . The use of mass spectrometry has identified multiple sites on the P protein that can be phosphorylated , including serine residues at positions 36 , 126 , and 157 and a threonine residue at position 286 [30] . In addition , it has been shown that host cell Polo-like kinase 1 ( PLK1 ) can phosphorylate a serine residue at position 308 [31] . Mutation of the serine residues to alanine residues at either position 157 or 308 , thereby preventing phosphorylation at these residues , significantly enhanced the activity of the vRdRP in mini-genome assays and the replication of recombinant viruses that bear these mutations [31] . In contrast , phosphorylation of the threonine residue at position 286 may enhance viral replication , since mutating it to an alanine residue reduced vRdRP activity and viral growth [30] . Here we show that amino acid substitutions , found in natural isolates of PIV5 , at residues 157 and 308 , as well as at other sites , including some which cannot be phosphorylated , also influence the activity of the vRdRP in mini-genome assays and determine whether replication is or is not repressed at later times post infection leading to lytic or persistent phenotypes respectively . As only single nucleotide changes in all the wild-type isolates of PIV5 sequenced are predicted to convert them from lytic to persistent phenotypes , or visa versa , we propose that the selection of lytic or persistent variants , from a quasispecies population ( defined as the mutant distributions that are generated upon replication of viruses in infected cells and organisms [32] ) , at early and late times after infection , respectively , may be a mechanism that PIV5 , and some other RNA viruses , have evolved to increase their success of transmission .
In contrast to infection with most strains of PIV5 held within our laboratory ( see below ) , a high multiplicity of infection ( moi ) of A549 , MRC5 or Vero cells with the PIV5-W3 isolate led to >90% of the cells surviving to become persistently infected; these infected cell-lines can be readily passaged ( S1 Fig ) . We therefore decided to investigate the molecular events that lead to the establishment and maintenance of PIV5-W3 persistence in the expectation that this may lead to a better understanding of paramyxovirus persistence in vivo . Initially , we monitored the synthesis of viral proteins and the levels of viral mRNA and genomic RNAs following a high moi of A549 cells ( Fig 1 ) . Ongoing viral protein synthesis at various times post-infection ( p . i . ) was visualised by metabolically labelling A549 cells that had been infected with PIV5-W3 at a high moi , with [35S]-L-methionine for 1h ( Fig 1 , panel a ) . At 24h post infection ( p . i . ) NP and M were synthesised at sufficiently high levels to be clearly detectable above the background of cellular protein synthesis , which is not significantly repressed in PIV5-W3-infected cells . However , by 48 and 96h p . i . , synthesis of NP and M had fallen below the levels that could be detected above the background of cellular protein synthesis . Immune precipitation analysis revealed that even by 36h p . i . there was an obvious reduction in the synthesis of all viral proteins ( S2 Fig ) . In contrast , immunoblot analysis of the same samples showed that the relative levels of ( accumulated ) NP were slightly higher at 96h p . i . than at 24h ( Fig 1 , panel b ) , even though at 96h p . i . there was little , if any , de novo virus protein synthesis ( Fig 1 , panel a ) . It was possible that the decrease in viral protein synthesis observed between 24 and 48h p . i . was due to an IFN-induced antiviral state within the infected cells . Although , this appeared unlikely because MxA , an IFN-induced protein , was not induced in the infected A549 cells ( Fig 1 , panel b ) , we monitored the kinetics of viral protein synthesis in cells that were infected in the presence of ruxolitinib ( an inhibitor of JAK1 that blocks IFN signalling [33] ) . There were no changes in the switch-off kinetics in the presence of ruxolitinib ( Fig 1 , panel a ) . Furthermore , viral protein synthesis was also switched off with the same kinetics in A549/Npro cells which cannot produce IFN because BVDV-Npro targets IRF3 for proteasome-mediated degradation ( [34]; S3 Fig ) , confirming that the decrease in virus protein synthesis observed was independent of the IFN response . To investigate whether the observed switch-off of viral protein synthesis was due to inhibition of viral transcription , we used high-throughput sequencing ( HTS ) to quantitate the levels of viral mRNA and viral genomic RNA during infection . Maximal levels of viral transcription were observed between 12 and 18h p . i . , at which times the amount of viral mRNA comprised almost 5% of total cellular mRNA ( Fig 1 panel c ) . Thereafter , the amount of viral mRNA slowly declined such that by 96h p . i . it amounted to less than 0 . 2% of total cellular mRNA , thus indicating that it was the reduction in viral mRNA that was responsible for the observed switch-off of viral protein synthesis ( note: although viral mRNA and antigenome sequences cannot be distinguished by directional sequencing , antigenome sequences contributed <2% of the total viral mRNA and antigenome reads , see Fig 1C legend ) . Levels of viral genomic RNA continued to increase until 48h p . i . ( Fig 1 , panel c ) . Strikingly , high levels of genomic RNA were also present at 96h p . i . , by which time there was very little viral transcription occurring . Defective virus genomes ( DVGs ) were not detected by HTS at 96h p . i . ( or in persistently infected cultures ) suggesting that they do not play a role in the switch-off of viral transcription and protein synthesis ( discussed in greater detail below ) . The switch-off of viral protein synthesis could also be inferred from immunofluorescence studies ( Fig 2 , panel a ) aimed at visualising the presence within infected cells of HN ( which has a half-life of 2 . 5 hours [35] ) and NP ( which has a half-life of days ) . At 24h p . i . all cells were strongly positive for both NP and HN . However , by 96h p . i . , while all the cells remained positive for NP , less than 50% of the cells were also positive for HN , and many of those were only weakly positive ( Fig 2 , panel a ) . Since HN possesses neuraminidase activity , the levels of HN expression at later times p . i . could also be inferred by staining cells for the presence of sialic acid . At 24h p . i . none of the infected cells expressed detectable amounts of sialic acid on their surface ( Fig 2 , panel b ) . However , by 72h p . i . some cells were positive for sialic acid . The fact that a high proportion of cells were negative for HN , and positive for sialic acid , strongly suggests that there was little or no ongoing viral protein synthesis in these cells at late times p . i . . These results demonstrate a degree of cellular asynchrony in the relative levels of viral gene expression at late times p . i . . Cells persistently infected with PIV5-W3 grew slightly slower than uninfected cells for the first couple of passages but by passage 3 ( p3 ) replicated as fast as uninfected cells and showed few visual signs of being infected . Immunostaining of persistently infected cells at p3 showed that whilst all the cells were infected there was heterogeneity in expression of the HN protein ( Fig 2 , panel b ) . Thus , some cells were positive for HN and negative for sialic acid , and others were negative for HN and positive for sialic acid . All cells were positive for NP and P expression , although the amount of NP and P present in the cells varied considerably ( Fig 2 , panels b and c ) . In general , cells that were strongly positive for NP ( and P ) were negative for sialic acid , and those that were weakly positive for NP ( and P ) were positive for sialic acid . HTS of the cells at 96 h p . i . , and of persistently infected cells , showed that viral mRNA constituted less than 0 . 2% of the total RNA ( Fig 1 , panel c ) . This level of viral mRNA characterised the population of cells as a whole but , given the heterogeneity of virus expression , the levels of viral mRNA must have varied considerably among cells within the persistently infected population . These results therefore strongly suggest that within the persistently infected population as a whole , active viral transcription was occurring in some cells ( HN-positive cells ) but was largely , if not completely , repressed in others ( HN-negative cells ) . We next investigated whether infectious virus could be rescued from cells in which viral transcription and replication were repressed . Persistently infected cells at passage 3 were stained for surface expression of the HN protein , and FACS was used to sort HN-positive and HN-negative cells into individual wells of 96 well plates ( Fig 2 , panel d ) . Colonies from each population were grown out in the presence , or absence , of a pool of PIV5-neutralizing antibodies to prevent viral spread between cells . Immunofluorescence showed that all the colonies remained infected ( Fig 2 , panel e ) , regardless of whether the sorted cells were derived from HN-positive or HN-negative cells , and whether or not they had been cultured in the presence or absence of neutralizing antibody . Furthermore , upon removal of the medium containing neutralizing antibody , infectious virus was recovered from all colonies tested . The fact that all the cells remained positive for NP in the presence of high titres of neutralizing antibody demonstrates that the cells remained infected as they divided , and that the production of infectious virus was not required for the maintenance of persistence within the colonies . Previous work has shown that phosphorylation of PIV5 P regulates the activity of the vRdRP and that phosphorylation of the serine ( S ) residue at position 157 ( S157 ) of the P protein plays a role in down-regulating viral gene expression [31 , 36] . To determine whether S157 plays a critical role in the switch-off of viral transcription and replication , and in the establishment of persistence , we generated a recombinant virus , rPIV5-W3:P ( F157 ) , in which the serine residue at position 157 was replaced by a phenylalanine ( F ) residue in the PIV5-W3 backbone . The F substitution was chosen because several strains of PIV5 have this residue at position 157 . Sequence analysis of rPIV5-W3:P ( F157 ) confirmed that this was the only amino acid substitution in the recombinant virus . Following infection with rPIV5-W3:P ( S157 ) at high moi , switch-off of viral protein synthesis occurred 24 and 72h p . i . , and >95% of the cells survived the infection ( Fig 3 , panels a and b ) . In striking contrast , no detectable switch-off of viral transcription or replication occurred in cells infected with rPIV5-W3:P ( F157 ) , and by 72 h p . i . ~90% of infected cells had died ( Fig 3 , panels a , b and c ) . Furthermore , rPIV5-W3:P ( S157 ) generated poorly defined plaques , which needed to be immunostained for clear visualisation , whereas plaques produced by rPIV5-W3:P ( F157 ) were easily visualised by crystal violet staining because the cells within the plaques died , leaving obvious holes in the monolayer ( Fig 3 , panel d ) . Significantly , in single-step growth curves , rPIV5-W3:P ( F157 ) grew more rapidly and to higher titres than rPIV5-W3:P ( S157 ) ( Fig 3 , panel e ) . It has been shown previously that a cellular kinase , Polo-like kinase 1 ( PLK1 ) , interacts with the P protein through the phosphorylated S157 residue and phosphorylates other sites on the P protein , including S308; phosphorylation at either of these sites reduces virus transcription [31] . We used mass spectrometry to compare the phosphorylation of the P protein in cells infected with rPIV5-W3:P ( S157 ) and rPIV5-W3:P ( F157 ) ( S4 Fig ) . These results confirmed that S157 and S308 were phosphorylated in cells infected with rPIV5-W3:P ( S157 ) . Despite identifying multiple other phosphorylation sites on P in this way , we did not identify any sites , other than S157 , that were phosphorylated on rPIV5-W3:P ( S157 ) but not on rPIV5-W3:P ( F157 ) , and vice versa , including S308 ( S4 Fig ) . However , we could not rule out the possibility that the relative levels of phosphorylation at the different sites did not vary significantly between rPIV5-W3:P ( S157 ) and rPIV5-W3:P ( F157 ) . In a previous study , Sun et al [31] , showed that a PLK1 kinase inhibitor ( BI2536 ) increased PIV5-W3 gene expression in infected cells at 18–20h p . i . . Therefore , we tested whether BI2536 treatment prevented or delayed the switch-off of rPIV5-W3:P ( S157 ) protein synthesis ( S5 Fig ) . BI2536 had no discernible effect on the switch-off of viral protein synthesis , or the ability of rPIV5-W3:P ( S157 ) to establish a persistent infection , suggesting that PLK1 is not the only cellular kinase that can phosphorylate serine-157 and inhibit viral gene expression . Having established that both transcription and replication of the W3 strain of PIV5 are significantly down-regulated within 48h , we next investigated whether this was the case for other PIV5 strains . A549 cells were infected at high multiplicity with the W3 , CPI+ , MEL , LN , SER and H221 strains of PIV5 and were metabolically labelled at various times p . i . with [35S]-L-methionine . Expression of NP and M proteins was repressed with time in cells infected with PIV5-W3 , but no obvious switch-off of viral protein synthesis was observed for the other strains of PIV5 ( Fig 4 , panel a ) . HTS confirmed that high levels of virus protein synthesis at late times p . i . in CPI+-infected cells were because virus transcription was not repressed ( S6 Fig , panel a ) . These studies also showed that the maximal levels of viral mRNA were significantly higher in CPI+ and rPIV5-W3:P ( F157 ) , approximately 17% and 13% respectively ( S6 Fig , panel a and b ) , than in rPIV5-W3:P ( S157 ) -infected cells ( approximately 5%; Fig 1 , panel b ) . A comparison of PIV5 P protein sequences published in GenBank strains revealed that CPI+ , MEL and LN have F157 ( Fig 5 ) , consistent with their failure to shut-down expression as observed with rPIV5-W3 ( F157 ) . However , we had expected that viral protein synthesis would be repressed at late times p . i . with H221 and SER because they have a serine at residue 157 , but it was not . There are three amino acid sequence differences in P between PIV5-W3 and PIV5-SER ( S69L , T155P and T293K ) [13]; strikingly , T155P is in close proximity to S157 . There are four amino acid differences in P between PIV5-W3 and PIV5-H221 ( V226M , T293K , N306K and I381D ) ; N306K is in close proximity to S308 . We next checked whether P155 , K306 , and other amino acid changes around S157 , have a direct effect on vRdRP activity by using a minigenome system . We initially compared the ability of P from PIV5-W3 ( S157 ) and PIV5-CPI+ to stimulate viral RNA synthesis ( Fig 4 , panel b ) . In agreement with previously published data [36] , P from PIV5-CPI+ was considerably more active than that derived from PIV5-W3 . As predicted , substituting S for F at position 157 stimulated vRdRP activity ( Fig 4 , panel b ) . We next examined the phenotypes of other changes in the W3 P protein . The T155P substitution ( as observed in PIV5-SER ) significantly enhanced the activity of PIV5-W3 P in the minigenome assay ( Fig 4 , panel b ) . We also noted stimulatory effects of amino acid substitutions at positions 156 and 159 . The N306K substitution ( observed in PIV5-H221 ) and a substitution at amino acid 308 ( S308A ) also significantly enhanced P activity . In contrast , the T293K substitution ( observed in both PIV5-SER and -H221 ) had only a small effect . These data show directly that the potential for phosphorylation in two motifs TSSPI ( residues 155–159 ) and NDS ( residues 306–308 ) represent targets for the negative regulation of P activity . To further test the effect of changes in P on the gene expression profile of W3 , we generated recombinant viruses with a T to P substitution at position 155 , rPIV5-W3:P ( P155 ) , or an N to K substitution at position 306 , rPIV5-W3:P ( K306 ) . Strikingly , both rPIV5-W3:P ( P155 ) and rPIV5-W3:P ( K306 ) behaved similarly to rPIV5-W3:P ( F157 ) , in that viral protein synthesis was not inhibited at late times p . i . ( Fig 4 , panel c ) , and viral infection resulted in increased cell death . In contrast , substituting a lysine for an arginine residue at position 254 ( rPIV5-W3:P ( R254 ) ) , which is part of a putative sumoylation site [37] , had no effect on the switch-off of PIV5-W3 transcription , neither did deletion of the SH gene ( Fig 4 , panel c ) . These results demonstrate that single amino acid substitutions at multiple sites within P ( observed in natural isolates of PIV5 ) can switch PIV5 from a virus with a persistent phenotype to one with a lytic phenotype . Having shown that the repression of viral transcription and replication , and the establishment of persistence , depends on the integrity of the TSSPI motif at residues 155–159 in PIV5-W3 , we compared all the PIV5 P protein and gene sequences available in the GenBank database ( Fig 5 ) . It was striking that in all strains residue 155 was either threonine or proline , residue 156 was either serine or asparagine , residue 157 was either serine or phenylalanine , and residue 159 was either threonine or isoleucine . Residue 158 ( proline ) was invariant . No strain had more than one difference from W3 in this region ( e . g . none had both a proline at residue 155 and a phenylalanine at residue 157 ) , and codon redundancy at these residues was such that a single nucleotide substitution was sufficient to change the virus from one with a predicted lytic to a W3-like persistent phenotype . Similarly to PIV5-W3 , some strains had threonine at residue 155 and serine at residue 157 but differed from PIV-W3 sequence at neighbouring residues that increased vRdRP activity in the minigenome assays ( e . g . residue 156 could be serine or asparagine , and residue 159 isoleucine or threonine; Fig 4 , panel b ) . However , again only one nucleotide substitution was required to convert the sequence of P in these strains to that of PIV5-W3 . It is also notable that PIV5-H221 was the only strain with lysine instead of asparagine at residue 306 , and again codon redundancy ensured that a single nucleotide substitution was sufficient to convert it to a PIV-W3 phenotype . Defective virus genomes ( DVGs ) have been shown to play a role in the establishment and maintenance of persistent infections of tissue culture cells by many positive and negative sense RNA viruses [38] . We therefore used HTS to determine whether DVGs may play a role in the establishment of persistence with PIV5-W3 . Firstly , we showed that HTS both of purified nucleocapsids and of total cell RNA ( following the physical removal of ribosomal and mitochondrial RNA ) could be used to successfully detect the presence of DVGs ( S7 Fig ) . From this analysis we determined that we would detect any DVGs if their breakpoint sequences contributed more than 0 . 02% of genomic sequences , or if the DVG contributed more than 0 . 2% ( or even less , see below ) of the total DVG and non-defective genomes . Using this approach , no DVGs were detected , either in purified nucleocapsids or in total cell RNA , isolated from passage 3 persistently infected PIV5-W3 cells . HTS both of nucleocapsid RNA and total cell RNA extracted from p3 persistently infected cells also revealed that there were no changes in the consensus sequence of PIV5-W3 in the persistently infected cells . Although , approximately 80% of cells die by 3 days p . i . following infection with CPI+ ( as determined by measuring cell viability using PrestoBlue as described in Fig 3 ) , some cells survive and , with difficulty , it is possible to establish a persistently infected cell-line from these surviving cells . This necessitated that the surviving cells be cultured for many weeks without sub-culturing , replacing the culture medium regularly . Eventually the surviving cells began to grow and could be passaged . However , even then they continued to grow poorly and showed obvious signs of virus cytopathic effects within the monolayers . High levels of DVGs ( the ratio of DVGs to non-defective genomes was 1 . 7:1 , S7 Fig ) were detected in cells persistently infected with CPI+ , suggesting that they may play a role in the establishment of persistent infections under these conditions . Also , in marked contrast to cells persistently infected with PIV5-W3 in which the amount of viral mRNA was less than 0 . 2% of total cellular mRNA , the levels of CPI+ mRNA in persistently infected cells was significantly higher , approximately 6% of total cellular mRNA . Furthermore , several polymorphic mutations were identified in the CPI+ persistently infected cell-lines but all of these , except for one , were synonymous mutations . The exception was located at position 13093 of the genome , an A to T change , resulting in a phenylalanine to leucine substitution in L , that was present in 17% of the reads , but the biological significance of this is unclear . We wished to investigate whether the phenotypic differences observed between rPIV5-W3:P ( S157 ) and rPIV5-W3:P ( F157 ) were reflected in differences in their biological properties in a mouse model system . However , first , and in agreement with our previously published data [39] , we demonstrated that , as observed in A549 cells , rPIV5-W3:P ( S157 ) protein synthesis was switched off in BALB/c fibroblasts and that the cells survived the infection . In contrast , rPIV5-W3:P ( F157 ) protein synthesis was not switched off in murine fibroblasts by 72 h p . i . and most of the cells died following infection ( Fig 6 , panels a and b ) . To determine whether rPIV5-W3:P ( S157 ) and rPIV5-W3:P ( F157 ) had different biological properties in vivo , BALB/c mice were infected intranasally with rPIV5-W3:P ( S157 ) or rPIV5-W3:P ( F157 ) and sacrificed at 1 , 2 , and 7 days p . i . , and the amount of virus present in the lungs was estimated by quantitative PCR ( Fig 6 , panel c ) . In addition , the amount of inflammation at the time of sacrifice was assessed by measuring the number of cells in the lungs ( Fig 6 , panel c ) [40] . These results showed that , although rPIV5-W3:P ( F157 ) had replicated to higher titres than rPIV5-W3:P ( S157 ) by 2 days p . i . , it was cleared more rapidly . Thus , by 7 days p . i . there was significantly less rPIV5-W3:P ( F157 ) present in the lungs than rPIV5-W3:P ( S157 ) ( p<0 . 001 ) . Interestingly , the amount of rPIV5-W3:P ( S157 ) present in the lungs at 7 days p . i . was similar to that observed at 1 and 2 days p . i . . There were significantly more cells in the lungs after rPIV5-W3:P ( F157 ) infection at days 2 ( p<0 . 05 ) and 7 ( p<0 . 01 ) p . i . than rPIV5-W3:P ( S157 ) , indicating greater inflammation [40] . However , neither virus caused overt disease as measured by weight loss ( Fig 6 , panel c ) .
Within-host RNA viral persistence has many potential consequences for both virus and the host [7] . For example , persistently infected individuals may act as viral reservoirs within host communities , and persistent infections may be important in the development of long-lasting protective immunity . However , little is known about the molecular mechanisms by which RNA viruses establish such infections . In part this may be because , unlike the better understood situations with DNA viruses or retroviruses , and despite the examples of hepatitis C virus and bornaviruses , it is often difficult to determine whether certain RNA viruses have evolved specific molecular mechanisms to establish and maintain persistent infections . To establish such infections within a host following lytic infection , it is likely that viral replication must be repressed in at least some cells in order either to prevent viral replication from killing the cell or to avoid the infected cell being eliminated by innate and adaptive immune responses [7] . In the case of paramyxoviruses and other members of the order Mononegavirales , and given their general mode of replication , it is not obvious how viral replication could be specifically repressed in order to facilitate virus persistence . Using PIV5 as a model , we report that viral transcription and replication can be repressed by phosphorylation of P , resulting in the establishment of persistently infected cell cultures ( without the need for the presence of DVGs ) in which the virus can flux between active and repressed states within individual cells . Since the consensus genome sequence of PIV5-W3 does not change in persistently infected cells we suggest that the amount of P ( Fig 2 , panel c ) , as well as its level of phosphorylation , varies heterogeneously over time within persistently infected cells and it is this which determines whether the virus is active or repressed within individually infected cells . We also speculate that in vivo , depending on the status of the adaptive immune response , variants with lytic or persistent phenotypes will be selected for or against . Since the quasispecies nature of RNA viruses will generate a cloud of virus mutants in vivo [32 , 41 , 42] as the virus spreads from cell to cell , during acute phases of infection rapidly replicating PIV5 variants will be selected for . However , eventually cells that continuously synthesize high levels of viral proteins will be efficiently killed by cytotoxic T cells and possibly ADCC ( antibody dependent cell cytotoxicity ) . Consequently , as the adaptive immune response develops , variants whose replication may be repressed , thus avoiding cell killing , will be selected , leading to the establishment of persistence . Since virus can be reactivated from cells in which it has been repressed , it is likely that small amounts of infectious virus will continuously be produced in persistently infected individuals perhaps as the immune response waxes and wanes . If such variants are transmitted to a new host again initially rapidly replicating variants will have a selective advantage , setting up a cycle of alternative selection of acute vs persistent variants in at least some infected individuals . Given that single amino acid ( nucleotide ) changes can determine whether a particular PIV5 variant has a lytic or persistent phenotype this mechanism may have evolved to allow PIV5 to establish both productive acute infections as well as persistent infections , thereby potentially increasing its chances of transmission [43] . For similar reasons , it is possible that other RNA viruses may have evolved analogous mechanisms in which single amino acid ( nucleotide ) changes can determine whether a particular variant has a lytic or persistent phenotype . Our results support and extend those published by the He group showing that phosphorylation of S157 and S308 on P results in repression of viral RNA synthesis [31] . However , our mass spectroscopy analysis showed phosphorylation of S308 in peptides from cells infected with rPIV5-W3:P ( S157 ) and from cells infected with rPIV5-W3:P ( F157 ) ( although we could not quantitate the degree of phosphorylation at this , or other , residues which might be quite heterogeneous ) , and yet rPIV5-W3:P ( F157 ) replication was not inhibited at late times p . i . . Furthermore , BI2536 , an inhibitor of PLK1 , did not influence the observed switch-off of PIV5-W3 protein synthesis nor did it prevent the establishment of persistence ( S5 Fig ) . This strongly suggests that cellular kinases other than PLK1 are ( also ) responsible for phosphorylating P . Although , we do not know the other kinase ( s ) responsible , the fact that PIV5-W3 transcribes and replicates its genome efficiently until 12-18h p . i . suggests that they are unlikely to be constitutively expressed cellular kinases but may be induced kinases , for example the ER stress response kinases activated by PIV5 infection [44] . If the latter , then perhaps this may also be a mechanism that the virus has evolved to help establish persistence . We note that P is also highly phosphorylated in other paramyxoviruses ( as are the phosphoproteins of other members of the order Mononegavirales , including Ebola virus ) and that the level of phosphorylation influences viral transcription [28 , 45 , 46] . Furthermore , between different strains/isolates there can be substitutions of amino acids , including serine , that may be phosphorylated , opening up the possibility that lytic and persistent variants of these viruses may be selected in vivo . Work presented here and elsewhere [31] clearly shows that phosphorylation of residues within the TSPPI motif ( amino acids 155–159 ) and the NDS motif ( amino acids 306–308 ) strongly influence the activity of the vRdRP . Residues 155–159 and 306–308 are outside the known N-terminal and C-terminal binding sites on P for NP [47 , 48] and are outside its predicted oligomerisation domain [49]; no binding site for L has yet been mapped . It is interesting to note that a structural prediction for the P protein of the closely-related rubulavirus , PIV2 [50] places the 155–159 and 306–308 motifs in long non-structured regions that flank the predicted oligomerisation domain; it is tempting to speculate that phosphorylation alters the conformation of the non-structured regions thereby influencing the properties of the P protein . In this regard , it is of note that the threonine to proline substitution at residue 155 enhanced the vRdRP activity more than the serine to phenylalanine substitution at residue 157 . Indeed , P155 led to the most active vRdRP in minigenome assays , even more active than A308 ( and K306 ) . The reason for this is unclear but suggests a model in which active and repressed states of vRdRP are in equilibrium , and that phosphorylation of P at residues 157 and 308 moves this equilibrium towards the repressed state . In this case , substituting threonine with proline at position 155 may cause a major structural change that locks vRdRP in the active state . DVGs may also play a role in the establishment of persistent infections with PIV5 , at least in vitro , as observed with other RNA viruses [38] . We show here that , with difficulty , persistently infected cell-lines can be established following a high moi of A549 cells with CPI+ lytic isolate of PIV5 . In contrast to cells persistently infected with PIV5-W3 , the CPI+ persistently infected cell-line had high levels of DVGs ( S7 Fig ) , suggesting under these circumstances DVGs may play a role in CPI+ virus persistence . The characteristics of this persistently CPI+-infected cell-line was very different from that established by PIV5-W3 . Ongoing virus transcription was much higher , the cells grew much more slowly and there were clear signs of a virus cytopathic effect . There was no evidence from HTS for selection of variants of CPI+ ( e . g . S157 ) that would be predicted to have a persistence phenotype . However , this would be expected as the cells were initially infected at a high moi making it unlikely that such variants could be selected . Rather , as discussed above , we speculate that the selection of virus variants with a persistence phenotype , such as PIV5-W3 , would likely occur following low moi infections in vivo in the presence of an ongoing adaptive immune response . Indeed , a critical point we are making here is that PIV5 , and thus potentially other paramyxoviruses , may have evolved specific molecular mechanisms for the establishment of persistent infections which do not rely on the production of DVGs . Although we have highlighted the importance of the phosphorylation status of P in determining whether or not a particular variant can establish persistence , it is possible that other single amino acid ( nucleotide ) changes in other genes , including L , may also play a role . We have also previously suggested that the interferon response may play an important role in repressing viral replication in some cells , thereby facilitating the establishment of persistence , and that there may be alternating selection of IFN-resistant and IFN-sensitive viruses during the acute and persistent phases of infection [11] . Interestingly , single amino acid ( nucleotide ) substitutions in the V protein , which is the viral IFN antagonist , can also determine whether a variant is IFN-sensitive or IFN-resistant [51] . Given that P and V are encoded by the same gene and share their N-terminal sequences , this gene may have evolved in such a manner as to facilitate the establishment of persistent infections . In this regard , it is of interest that PIV5 and other paramyxoviruses block the IFN response in such a way as not to cause cell death , which is a pre-requisite for establishing persistence . PIV5 has been isolated on numerous occasions from a variety of host species but its association with disease is often tentative and unclear . Of possible relevance is that the disease potential of lytic and persistent variants of PIV5 is likely to be different . Thus , lytic variants may cause more cell death and spread more rapidly in vivo than persistent variants . Although PIV5 does not replicate to high titres in , or naturally infect , mice , this idea is supported by the observation that , rPIV5-W3:P ( F157 ) replicated better than rPIV5-W3:P ( S157 ) in mice and induced more cellular infiltration into the lungs of infected mice , which is a clear sign of greater pathology . Also if there are mixed populations of persistent and lytic variants in vivo , then the balance of the two may also influence disease outcomes . Interestingly , both lytic and persistent variants were detected by HTS in our stocks of PIV5-H221 , which was isolated from a dog with kennel cough , but which had only been passaged a limited number of times in tissue culture cells following its initial isolation [13] . Thus , although the consensus sequence at position 157 of P was serine , phenylalanine was predicted in 4% of the viral population , and although the consensus at position 306 was a lysine , 5% of the sequenced population encoded asparagine [13] . It is of note that all the PIV5 strains sequenced , apart from the W3 strain , are predicted to have a lytic phenotype , which argues against the suggestion that viral transcription and replication are reduced at late times p . i . in order to limit the production of viral PAMPs and hence the induction of antiviral cytokines [31] . However , as lytic strains are more likely to induce a cytopathic effect than persistent variants , their selection may be favoured during clinical isolation . Furthermore , lytic variants , which give an obvious cytopathic effect in tissue culture cells , may evolve during the isolation of PIV5 from clinical material . This may have occurred during the isolation of PIV5 from human bone marrow cells , which were co-cultured with either MRC5 or Vero cells , as immunofluorescence was initially used to detect PIV5 during virus isolation as there was often an absence of a clear virus induced cytopathic effect [24 , 52] . Similarly , during the isolation of cryptovirus ( a strain of PIV5 ) , human lymphocytes from a patient with SSPE ( there is no suggestion that PIV5 can cause SSPE ) were cultured with AV3 ( continuous human amnion ) cells , but the first clear signs of cytopathic effect only became visible after 20 passages [26] . On the other hand , tissue culture cell-lines can be persistently infected with PIV5 with no overt signs of infection , for example AGS cells which are commercially available from ATCC and ECACC [21] . Understanding the mechanisms by which paramyxoviruses , and other RNA viruses , can establish persistence in vivo is important for both fundamental and practical reasons . It may lead to a more complete view of viral epidemiology , and thus to better control measures . In addition , if the induction of long-lasting immunity is enhanced by viral persistence , then understanding the mechanisms by which viruses can establish such infections may lead to improved vaccine design .
Vero , 293 and A549 cells ( all from the European Collection of Authenticated Cell Cultures; ECACC ) and derivatives were grown at 37°C as monolayers in 25 cm2 or 75 cm2 cell culture flasks , in Dulbecco's modified Eagle's medium ( DMEM ) supplemented with 10% ( v/v ) foetal bovine serum at 37°C . Stocks of PIV5 strains W3 , LN , MEL , H221 , SER and CPI+ ( described in [13] ) were grown and titrated in Vero cells . Commercial cell-viability assays ( PrestoBlue ( ThermoFisher Scientific ) were performed according to manufacturer instructions . Infected or uninfected cells were metabolically labelled for 1h with [35S]-L-methionine ( 500Ci/mmol , MP Biomedical , USA ) at various times p . i . as indicated in the text . After labelling , cells were lysed in disruption buffer , sonicated and heated for 5 min at 100°C and proteins were analysed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis ( SDS-PAGE ) . The gels were fixed , stained , dried , and resolved radiolabelled bands visualized by phosphorimager analysis . Procedures for immunoprecipitation , immunoblotting and immunofluorescence have been described previously [53 , 54] . The antibodies used included monoclonal antibodies ( mAbs ) against PIV5 HN , P and NP [55] and against cellular MxA and β-actin ( Sigma , A5441 ) . Sialic acid on the surface of cells was visualised by staining with a recombinant protein in which green fluorescent protein ( GFP ) had been fused to two carbohydrate-binding modules derived from Vibrio cholerae [56] . Viral plaques were immunostained with a pool of mAbs against PIV5 followed by alkaline phosphatase-conjugated goat anti-mouse immunoglobulin G ( Abcam , ab97020 ) , and plaques were visualised with SigmaFast BCIP/NBT . For FAC-sorting , cells were prepared as a single-cell suspension by trypsinisation and immunostained with a pool of mAbs against HN . Single cells were sorted into individual wells of 96-well microtiter plates on the basis of whether or not they were positive for HN using a Becton Dickinson FACSJazz instrument . The changes in the P gene of the PIV5 W3 genome [T155P , S157F , K254R , N306K and S308A] were generated by primer-mediated mutagenesis using oligonucleotides purchased from Sigma and the modified fragments inserted into the rPIV5-W3 backbone plasmid , pBH276 [57] using standard molecular biology approaches . Base changes were confirmed by DNA sequencing . The pBH276-derived template plasmids ( 1μg ) were transfected together with pCAGGS-based helper plasmids directing the synthesis of PIV5-NP ( 100ng ) , PIV5-P ( 100ng ) and PIV5-L ( 500ng ) into 6-well dishes containing ~106 BSRT7 cells per well using linear polyethyleneimine ( PEI ) of molecular weight 25 , 000 ( Polysciences Inc . , Warrington PA , USA ) , or Fugene , under standard conditions . Successful recovery was confirmed by immunofluorescence screening using a monoclonal antibody ( PIV5-Pk ) conjugated to a FITC fluorophore , which recognises PIV5 V and P [55] . Working stocks of virus were produced from positive wells by two successive passages at low multiplicity of infection ( moi ) in Vero cells , and stocks were harvested , clarified by centrifugation , and flash frozen in liquid nitrogen . Confluent monolayers of A549 cells , grown in 25 cm2 flasks , were infected at a high moi with either rPIV5-W3:P ( S157 ) or rPIV5-W3:P ( F157 ) and at 24h p . i . were lysed in disruption buffer and submitted to the FingerPrints Proteomics Facility ( University of Dundee , UK ) for SDS-PAGE gel analysis . The samples were run on a 4–12 Bis-Tris gel with MOPS running buffer ( Thermo Fisher Scientific ) and the gel stained with SimplyBlue SafeStain ( Thermo Fisher Scientific ) . PIV5 P bands ( ~45kDa ) were excised for in-gel processing and trypsin digestion prior to analysis by mass spectrometry using a RSLCnano UHPLC system coupled to a LTQ Orbitrap Velos Pro mass spectrometer ( Thermo Scientific ) . The resultant data were analysed using the Mascot Search engine ( Version 2 . 4 . 1 ) using the Sprot Human database and the sites of phosphorylation annotated using the Mascot delta score . Infected cells in 25cm2 flasks were lysed in 1 ml of Trizol and RNA was extracted using a Direct-zol RNA miniprep kit ( Zymo ) . A directional sequencing library was prepared from rRNA-depleted RNA using a TruSeq stranded total RNA library prep kit ( Illumina , U . K . ) . Quality control and quantification of the cDNA library were monitored using DNA-specific 1000 or 5000 chips on a Bioanalyzer 2100 ( Agilent Technologies ) and a Qubit fluorometer ( Invitrogen ) . Individual libraries were pooled at 10 nM each and sequenced on the MiSeq platform ( Illumina ) . Abundances of genome and antigenome/mRNA reads were calculated relative to total read numbers ( including cellular reads ) from which residual rRNA and mitochondrial RNA reads had been removed . These reads were identified by aligning the trimmed , filtered data to reference genomes for human 18S , 28S , 5S and 5 . 8S rRNA and mitochondrial DNA ( accession numbers NR_003286 . 2 , NR_003287 , X51545 , J01866 , NC_012920 ) , and then removed . The presence of defective virus genomes ( DVGs ) was assessed using ViReMa ( Routh and Johnson 2013 ) . ViReMa detects potential recombination by identifying reads that contain sequences mapping to different regions of the genome , and thus facilitates the identification and quantification of DVG populations . Using standard techniques the Renilla luciferase gene in pSMG-RL [36] was replaced by a gene encoding firefly luciferase . An additional modification to reduce transcriptional readthrough from cryptic promoters within the vector backbone was made by incorporating two copies of a 237 bp fragment from SV40 ( coordinates 2533–2770 ) that includes the bidirectional polyadenylation site and transcriptional terminator site downstream from the T7 RNA polymerase terminator sequence . To determine the activity of the minigenome , 25ng of the resulting plasmid ( pPIV5MG-Fluc . ter ) , was transfected into 293 cells together with pCAGGS-based helper plasmids directing the synthesis of PIV5-NP ( 100ng ) , PIV5-P ( 100ng ) and PIV5-L ( 500ng ) , 500ng of a pCAGGS-based plasmid directing the synthesis of T7 RNA polymerase ( codon-optimised for expression in human cells ) , and 50ng of a β-galactosidase-expressing transfection control plasmid , pCATlac . Transient transfections used PEI and were left for 40 h before harvesting . Luciferase and β-galactosidase activity assays were carried out and normalised as previously described [58] . Variants of the P gene were generated by primer-mediated mutagenesis as described above . Female BALB/c mice were obtained from Charles River ( Bath ) at 7–9 weeks of age and housed in accordance with the United Kingdom Home Office guidelines . All work was conducted with approval from the Animal Welfare and Ethical Review Board of Imperial College London . Mice were infected intranasally with 2 x 106 plaque-forming units ( pfu ) of virus in 100 μl . Mice were provided with food and water ad libitum and monitored daily for signs of illness . Statistical comparisons of mouse data were as described in Figure legends were performed using Prism 6 ( GraphPad Software Inc . , La Jolla , CA , USA ) . Viral load in lung tissue was assessed by quantitative PCR ( qPCR ) of bulk PIV5 M gene RNA . RNA was extracted from frozen lung tissue using Trizol extraction after homogenisation in a TissueLyzer ( Qiagen ) and converted into cDNA using random primers ( GoScript , Promega ) . qPCR of PIV5 M gene was carried out using SYBRselect master mix and 250 nM forward ( 5’-TCATGAGCCACTGGTGACAT-3’ ) and reverse ( 5’-TGGAATTCCCTCAGTTGTCC-3’ ) primers on a Stratagene Mx3005p instrument ( Agilent Technologies ) . In order to normalise M gene levels , levels of cellular Gapdh mRNA were measured using forward ( 5’-AGGTCGGTGTGAACGGATTTG-3’ ) and reverse ( 5’-TGTAGACCATGTAGTTGAGGTCA-3’ ) primers . Lung tissue was homogenised through a 100μm cell strainer and centrifuged at 500 x g for 5 minutes , as described previously [59] . Supernatants were removed , and red blood cell lysis buffer ( ACK lysing buffer , ThermoFisher ) was added to the cell pellet and mixed for 5 min before a further centrifugation at 500 x g for 5 minutes . Remaining cells were resuspended in DMEM and viable numbers were quantified by trypan blue exclusion . All animal experiments were performed in accordance with the United Kingdom’s Home Office guidelines under PPL P4EE85DED and all work was approved by the Animal Welfare and Ethical Review board ( AWERB ) at Imperial College London . Studies followed the ARRIVE guidelines and all animal infections and infectious work was carried out in biosafety level two facilities . | As well as causing acute infections that result in mild to serious disease , many RNA viruses can establish prolonged or persistent infections in some infected individuals , that occasionally lead to chronic or reactive disease . Little is known about the molecular mechanisms involved in the establishment of such infections . Using parainfluenza virus type 5 ( PIV5 ) as a model , we show how lytic and persistent variants of the virus can be selected on the basis of single amino acid substitutions and propose that the selection of persistent variants as the adaptive immune response develops following an acute infection might be a mechanism these viruses have evolved to enhance their transmission rates . As well as being of fundamental interest , understanding the molecular basis by which RNA viruses establish persistent infections may improve our understanding of virus epidemiology ( and hence improve the control of virus infections ) and of virus:host interactions that influence the relationship between virus persistence and chronic/relapsing disease . Furthermore , the knowledge of how RNA viruses , such as PIV5 , establish persistent infections may lead to improve vaccine design since vectors which can establish persistent infections may induce longer-lasting more robust immunity . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"phosphorylation",
"sequencing",
"techniques",
"medicine",
"and",
"health",
"sciences",
"respiratory",
"infections",
"rna",
"extraction",
"microbiology",
"pulmonology",
"molecular",
"biology",
"techniques",
"rna",
"sequencing",
"extraction",
"techniques",
"microbial",
"geno... | 2019 | The switch between acute and persistent paramyxovirus infection caused by single amino acid substitutions in the RNA polymerase P subunit |
The Government of Senegal has embarked several years ago on a project that aims to eradicate Glossina palpalis gambiensis from the Niayes area . The removal of the animal trypanosomosis would allow the development more efficient livestock production systems . The project was implemented using an area-wide integrated pest management strategy including a sterile insect technique ( SIT ) component . The released sterile male flies originated from a colony from Burkina Faso . Monitoring the efficacy of the sterile male releases requires the discrimination between wild and sterile male G . p . gambiensis that are sampled in monitoring traps . Before being released , sterile male flies were marked with a fluorescent dye powder . The marking was however not infallible with some sterile flies only slightly marked or some wild flies contaminated with a few dye particles in the monitoring traps . Trapped flies can also be damaged due to predation by ants , making it difficult to discriminate between wild and sterile males using a fluorescence camera and / or a fluorescence microscope . We developed a molecular technique based on the determination of cytochrome oxidase haplotypes of G . p . gambiensis to discriminate between wild and sterile males . DNA was isolated from the head of flies and a portion of the 5’ end of the mitochondrial gene cytochrome oxidase I was amplified to be finally sequenced . Our results indicated that all the sterile males from the Burkina Faso colony displayed the same haplotype and systematically differed from wild male flies trapped in Senegal and Burkina Faso . This allowed 100% discrimination between sterile and wild male G . p . gambiensis . This tool might be useful for other tsetse control campaigns with a SIT component in the framework of the Pan-African Tsetse and Trypanosomosis Eradication Campaign ( PATTEC ) and , more generally , for other vector or insect pest control programs .
Tsetse flies ( Glossinidae ) transmit trypanosomes which cause human African trypanosomosis ( HAT ) and African animal trypanosomosis ( AAT ) , a debilitating disease of humans ( sleeping sickness ) and livestock ( nagana ) , respectively [1–4] . The economic cost of AAT in Africa has been estimated at USD 4 . 75 billion per year [5 , 6] . For decades , several approaches have been used to manage trypanosomosis , either targeting the parasite using chemotherapy and/or targeting the vector through the use of insecticides . However , less than 2% of the infested area ( estimated at around 10 million km² ) have been freed of tsetse flies [7] . One of the main reasons for these limited successes has been the reliance on a single control tactic , rather than integrating several control tactics in an area-wide approach [7 , 8] . There are four methods environmentally and economically acceptable that are currently used in a context of area-wide integrated pest management ( AW-IPM ) approaches to manage populations of tsetse flies: artificial baits ( insecticide-treated traps/targets or ITT ) , insecticide-treated cattle ( ITC ) , aerial spraying using the sequential aerosol technique ( SAT ) and the sterile insect technique ( SIT ) [7 , 8] . The African Heads of State and Government decided in 2000 to increase efforts to address the tsetse and trypanosomosis problem on the African continent and created the Pan-African Tsetse and Trypanosomiasis Eradication Campaign ( PATTEC ) [9] . Under this umbrella , the Government of Senegal initiated a tsetse eradication program in the Niayes area that integrated the SIT with other control tactics such as IIT’s and ITC [10 , 11] . For the SIT component the program used a strain of Glossina palpalis gambiensis that was originally colonized at the Centre International de Recherche-Développement sur l’Elevage en zone Subhumide ( CIRDES ) , Bobo-Dioulasso , Burkina Faso . The G . p . gambiensis pupae were mass-reared at the CIRDES and the Slovak Academy of Sciences ( SAS ) , Bratislava , Slovakia and supplemented with excess material from a colony maintained at the FAO/IAEA Insect Pest Control Laboratory , Seibersdorf , Austria and transported as irradiated male pupae by air to Dakar , Senegal [12–14] . The pupae were transferred to an insectary in Dakar for emergence and further processing . The sterile males were marked with a fluorescent dye powder ( DayGlo ) that was mixed with sand that was put on top of the pupae . Emerging male flies would pick up the powder whilst crawling through the sand , especially in the ptilinum that would later be retracted into the head capsule . The marking of the sterile male flies is required to allow discrimination of sterile from wild flies in the monitoring traps [13] to assess program progress [15] . All flies trapped during the monitoring were transferred to the laboratory and the head capsules examined using a fluorescence camera and / or a fluorescence microscope . In this way , released sterile flies could be distinguished from indigenous flies and the ratio of sterile over wild flies calculated , which is an important parameter for monitoring the efficiency of a SIT campaign [15–17] . However , marking flies with fluorescent dye powder is not infallible , and in some situations some flies can be poorly marked ( i . e . with just a few powder particles ) or conversely , wild flies in the monitoring traps might become contaminated with a few grains of powder . In addition , tsetse flies in the monitoring traps might be predated upon by ants and lose their head , which makes it difficult to differentiate released sterile flies from wild flies . Any doubt must be removed on the origin ( mass-reared sterile flies or wild flies ) of the trapped flies in the field especially in the final eradication phase . A single erroneously classed fly ( through poor marking or through contamination ) can result in wrong decisions made by the programme managers , i . e . to either continue or stop the eradication process , with potentially large financial losses as a result . Absence of trapped wild flies in continuous monitoring can indeed be considered as an evidence that the population has been eliminated from the target area [18 , 19] . A more accurate method that removes any doubt on the origin of caught flies would therefore be very useful , and in this paper we present the development of a molecular tool that is based on the mitochondrial gene COI ( cytochrome oxidase I ) to discriminate sterile from wild males .
Forty eight G . p . gambiensis flies were used to create the reference database of COI sequences i . e . to test the basic hypothesis that the COI sequences of the released G . p . gambiensis flies from the CIRDES colony were different from wild flies in the target areas in Senegal and Burkina Faso . Thirty one of these males originated from the CIRDES colony , seven flies ( three males and four females ) were collected in Pout ( 55 km East of Dakar , Senegal ) in September 2012 before the release of sterile males had been initiated in this area [12] , ten flies were collected in Burkina Faso , of which eight males were from Seguere ( 50 km North of Bobo-Dioulasso ) in December 2013 and two males were from Guinguette ( 15 km West of Bobo-Dioulasso ) in June 2015 ( Table 1 ) . The COI sequences of the CIRDES flies were compared with one another and then with those from Pout , Seguere and Guinguette . The validation of the method was done with twenty flies trapped in the field during the release operations in Senegal: seventeen flies ( eleven males , five females and one unidentified sex ) collected in Pout in December 2014 and January 2015 , and three males collected in Kayar ( 85 North-East of Dakar , Senegal ) in July 2012 ( Table 1 ) . Among the seventeen from Pout , ten were analyzed blind at the molecular level while their status ( wild or sterile ) was known ( UV fluorescence reading undoubted ) ( Fig 1 ) . The remaining seven flies from Pout and the three from Kayar were doubtful using the UV camera ( i . e . it was not possible to class them as wild or sterile ) . The COI sequences of these flies were compared to the reference data i . e . to the sequences of the CIRDES colony flies . Female flies were added in this comparative study because the method used to separate male and female pupae in the mass-rearing insectary is not completely accurate i . e . about 3% female pupae remain in male batch after sex separation and are therefore released together with the sterile males during the release operations [13 , 14] . Flies from the CIRDES colony were killed in a freezer and kept in 70% alcohol at room temperature ( around 25°C ) . The head of the flies collected in the field was removed and stuck on paper and kept at room temperature . We used heads for the genetic analysis because the discrimination with the fluorescence camera is done on the heads and we used those flies , which gave an unequivocal result with the camera , for discrimination with molecular tools . DNA was extracted from the head of each tsetse fly using cetyl trimethyl ammonium bromide ( CTAB ) and chloroform:isoamyl alcohol 24:1 followed by precipitation with isopropanol . Polymerase chain reaction ( PCR ) was used to amplify partial sequences of cytochrome oxidase ( COI ) , using the primer pair CI-J-2195/CULR [20] . PCR conditions: 34 . 9 μl of double distilled water containing 5 μl of 10X PCR buffer , 1 μl of 10 mM dNTP ( final concentration 0 . 2 mM ) , 1 μl of each 10 μM primer , 6 μl of 25 mM MgCl2 ( final concentration 3 mM ) were incubated with 0 . 5U of Taq DNA polymerase ( MP Biomedicals ) and 1 μl of template DNA . The temperature cycles were: 5 min at 95°C , 40 cycles of 93°C for 1 min , 55°C for 1 min and 72°C for 2 min , then 72°C for 7 min . PCR products were purified using QIAquick PCR purification kit ( Qiagen , Valencia , CA , USA ) using the manufacturer’s instructions , then sent for sequencing to GATC biotech . For each PCR product sequenced , forward and reverse sequences were aligned and traces examined using CodonCode Aligner ( CodonCodeCorporation ) . Sequences were aligned and trimmed using Blast in GenBank ( https://blast . ncbi . nlm . nih . gov/Blast . cgi ? PAGE_TYPE=BlastSearch ) . The study was conducted in the framework of the tsetse control program in Senegal , implemented by the Division of Veterinary Services , Ministry of Livestock . This project received official approval from the Ministry of Environment of Senegal , under the permit N°0874/MEPN/DE/DEIE/mbf .
A total of 48 flies were sequenced to set up the molecular reference data . Sequence data for the mitochondrial gene COI ( 888 nucleotides ) were aligned from the 31 individuals of the CIRDES colony . Among these , 29 individuals displayed 100% identity . A substitution ( nucleotide T instead of C ) was found at position 790 for one individual and one individual showed a sequence of 889 nucleotides including an inserted T after position 822 ( Table 2 ) . Of the seven wild individuals that were collected in Pout before the sterile male release operations had started , the sequence of six flies were 100% identical , irrespective of the sex , and one fly ( a female ) was different from the others by one substitution at position 247 ( G instead of A ) . The comparison between sequences of wild flies from Pout and those of sterile flies from the CIRDES colony resulted in substitutions in seven or eight positions ( Table 2 ) . The sequences of the wild male flies collected in Seguere ( eight flies ) and Guinguette ( two flies ) , Burkina Faso , all displayed at least one and up to five substitutions compared to the sequences of the sterile males from the CIRDES colony ( Table 2 ) . The detail of the comparison of the COI sequences can be seen in S1 Table . The efficiency of the molecular method was tested using flies collected in the field during the eradication operations in Senegal . A total of twenty flies were caught in the field of which seventeen flies ( eleven males , five females and one unidentified ) originated from Pout and three male flies from Kayar . Four male individuals trapped in Pout displayed 100% homology with the haplotype of the individuals from the CIRDES colony while thirteen ( seven males , five females and one unidentified sex ) showed between seven and eight substitutions ( almost always the same ) along the 888 nucleotides sequence ( Table 2 ) . For the three males from Kayar , two individuals displayed 100% identity with the haplotype of the individuals from the CIRDES colony and one differed by eleven substitutions ( see S1 Table for the detail of the comparison of COI sequences ) . This means that the four and two males from Pout and Kayar , respectively were sterile males i . e . from the CIRDES colony while the other individuals were wild flies ( Table 2 ) . The results showed that all the ten flies from Pout , of which the status was determined by the UV camera but which were analyzed blindly with the molecular method , were identified correctly ( Table 3 ) . The substitute bases and their positions are presented in S1 Dataset .
The tsetse eradication project in the Niayes of Senegal adopted a rolling carpet approach and the entire target area is scheduled to be cleared from G . p . gambiensis by the end of 2017 . The preliminary data showed excellent progress in the eradication campaign [31–33] . In this context , the confirmation of the status of “tsetse-free area” in one block before stopping the release operations was very important . The molecular tool developed in this study allowed discrimination with high accuracy between the released sterile males from the CIRDES colony and their wild counterparts . This tool might be useful for other tsetse control campaigns with a SIT component in the framework of the PATTEC . In particular , eight isolated population of G . p . gambienis have been identified recently in West Africa and may be targeted soon by new eradication efforts [28] . More generally , the same tool might be developed for other vector or insect pest control programs . | The Government of Senegal has embarked since several years on a project that aims to create a tsetse-free area in the Niayes . The project was implemented using an area-wide integrated pest management ( AW-IPM ) strategy where the sterile flies used for the sterile insect technique ( SIT ) component were derived from a colony originating from Burkina Faso . Monitoring the efficacy of the sterile male releases requires the discrimination between wild and sterile males that are sampled in monitoring traps . Before being released , sterile males were marked with a fluorescent dye powder . The marking was however not infallible with some sterile flies only slightly marked or some wild flies contaminated with a few dye particles in the monitoring traps , making it difficult to discriminate between wild and sterile males using a UV camera . We developed a molecular technique based on the cytochrome oxidase gene that efficiently discriminates between wild and sterile males . This tool might be useful for other tsetse control campaigns with a SIT component or for other vector or insect pest control programs . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"sequencing",
"techniques",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"geographical",
"locations",
"pigments",
"animals",
"glossina",
"organisms",
"insect",
"pests",
"materials",
"science",
"pest",
"control",
"molecular",
"biology",
"techniques",
"tsetse",... | 2016 | A Molecular Method to Discriminate between Mass-Reared Sterile and Wild Tsetse Flies during Eradication Programmes That Have a Sterile Insect Technique Component |
Experimental studies show that human pain sensitivity varies across the 24-hour day , with the lowest sensitivity usually occurring during the afternoon . Patients suffering from neuropathic pain , or nerve damage , experience an inversion in the daily modulation of pain sensitivity , with the highest sensitivity usually occurring during the early afternoon . Processing of painful stimulation occurs in the dorsal horn ( DH ) , an area of the spinal cord that receives input from peripheral tissues via several types of primary afferent nerve fibers . The DH circuit is composed of different populations of neurons , including excitatory and inhibitory interneurons , and projection neurons , which constitute the majority of the output from the DH to the brain . In this work , we develop a mathematical model of the dorsal horn neural circuit to investigate mechanisms for the daily modulation of pain sensitivity . The model describes average firing rates of excitatory and inhibitory interneuron populations and projection neurons , whose activity is directly correlated with experienced pain . Response in afferent fibers to peripheral stimulation is simulated by a Poisson process generating nerve fiber spike trains at variable firing rates . Model parameters for fiber response to stimulation and the excitability properties of neuronal populations are constrained by experimental results found in the literature , leading to qualitative agreement between modeled responses to pain and experimental observations . We validate our model by reproducing the wind-up of pain response to repeated stimulation . We apply the model to investigate daily modulatory effects on pain inhibition , in which response to painful stimuli is reduced by subsequent non-painful stimuli . Finally , we use the model to propose a mechanism for the observed inversion of the daily rhythmicity of pain sensation under neuropathic pain conditions . Underlying mechanisms for the shift in rhythmicity have not been identified experimentally , but our model results predict that experimentally-observed dysregulation of inhibition within the DH neural circuit may be responsible . The model provides an accessible , biophysical framework that will be valuable for experimental and clinical investigations of diverse physiological processes modulating pain processing in humans .
The processing of pain engages a wide variety of neural circuits across the nervous system including those in the spinal cord , brainstem , thalamus , and cortex . More specifically , it is thought that the dorsal horn ( DH ) , an area of the spinal cord , serves as the initial processing center for incoming nociceptive , or painful signals , with the midbrain and cortex providing top-down modulation to that circuitry [1] . As a result , there is a tradition of modeling pain processing by focusing exclusively on spinal cord circuitry . This circuitry receives information about stimulation of peripheral tissues from several types of primary afferent nerve fibers . These afferents have their cell bodies in the dorsal root ganglia ( DRG ) , a cluster of nerve cell bodies located exterior to the spinal cord , and their axons ( or fibers ) target the DH [2] . Responses to innocuous stimulation are carried by rapidly conducting Aβ-fibers [3] , whereas nociceptors ( i . e . , nerve fibers that detect painful stimuli ) are only activated when a stimulus exceeds a specific threshold . There are two major classes of nociceptive fibers: fast conducting Aδ-fibers that mediate localized , fast pain and small-diameter C-fibers that mediate diffused , slow pain . Among the neuronal populations in the DH , the projection neurons ( PNs ) receive input from all fibers and constitute the majority of the output from the dorsal horn circuit up to the brain . In this article , we introduce a biophysically-based , mathematical model of the nociception-processing neural circuit in the DH , which expands on our earlier work [4] . We are particularly interested in using the model to investigate mechanisms for daily ( i . e . , diurnal ) modulation of pain sensitivity . In many clinical conditions , pain sensitivity follows a daily cycle [5] , that is , it exhibits a trough in the late afternoon and a peak sometime after midnight for humans [6] , but it is currently unclear how much of that rhythmicity is derived from daily fluctuation in the underlying causes of the pain versus rhythmicity in the neural processing of pain . Within the experimental pain literature , rhythmic influences on pain sensation occur regardless of whether pain responses are measured subjectively or objectively [7–10] , suggesting that the rhythmic modulation of pain responses occurs at the level of basic nociceptive processing . This rhythmic modulation of pain sensitivity also increases with pain intensity [9 , 11 , 12] . Furthermore , rhythmic influences on pain sensitivity are detectable in experiments involving a variety of different kinds of nociceptive stimuli , including cold , heat , electric current , pressure , and ischemia ( see Tables 1–2 in [6] ) . Interestingly , experimental studies have also shown daily rhythmicity in tactile discrimination in nearly opposite phase to pain sensitivity , namely highest tactile sensitivity occurring in the late afternoon and lowest in the morning [13] . There are several hypotheses for the source of the daily rhythm in pain sensitivity , including central nervous system , spinal , and peripheral mechanisms [5 , 14–18] . Recent studies show that cells in the DRG rhythmically express the primary genes responsible for generating an intrinsic 24-hour , or circadian , rhythmicity of other physiological processes , including Bmal1 , Clock , Per1 and Per2 [15 , 16] . In addition , the rhythm in behavioral nociception followed the gene expression rhythm [15] and disruption of their expression affected behavioral pain responses [16] . These findings motivate our use of a spinal cord model to test questions regarding daily influences on pain processing . As such , the model assumes that the daily modulation occurs at the level of primary afferent input to the spinal cord circuitry . Additionally , we specifically model the portion of experienced pain that arises from nociceptive input to the spinal cord and ignore any potential sources of top-down modulation . As concerns the connections between neuron populations in the DH , there are several proposed circuitries for the processing of touch , nociception , and itch ( see , e . g . , [19 , 20] ) . In this work , we take an approach similar to previous models of spinal cord nociception processing ( e . g . , [21] ) and employ the network architecture in the DH proposed by the gate control theory of pain [22] . In doing so ( and when we introduce daily modulation ) , we note that the aim of our work here is focused on the processing of painful , noxious stimulation , not mechanical , non-noxious stimulation , which we acknowledge may have a different circuitry ( for a review of circuitries for mechanical pain and itch , see [23] ) . The gate control theory of pain [22 , 24] posits that the neural circuitry in the DH exhibits a gating mechanism that is modulated by activity in the Aβ- and C-fibers [25] . Specifically , nociceptive C-fiber-facilitated activity in the DH circuit is inhibited by Aβ-fiber activity . When the amount of painful stimuli ( i . e . , activity in the C-fibers ) outweighs the inhibition from the Aβ-fibers , the “gate opens” and activates the PNs ( and the experience of pain ) . Although the gate control theory of pain [22] is a simplification and not a complete representation of the physiological underpinnings of pain processing [25] , it has been a productive starting point for several mathematical and computational models of pain [21 , 26–28] . For our model of the DH circuit , we implement a neuronal population firing-rate model formalism [29 , 30] to describe the population activity of projection , inhibitory , and excitatory neurons in the DH . Our choice of this commonly-used model formalism is based on the large number of afferent fibers and neurons in the DH , and the assumption that the majority of information flow in the DH circuit is through firing rates of neural populations rather than in specific spike timing within the populations [30 , 31] . An advantage of this formalism is its biophysical basis and relative simplicity , thus making our model an accessible theoretical framework for experimental and clinical investigations of diverse physiological processes modulating pain processing in humans . The rest of the paper is organized as follows . In the Methods section , we formulate the equation system of the neural circuit for pain processing in the DH , describing the time evolution of the average firing rates of the excitatory and inhibitory interneuron , and PN populations in response to input on the afferent nerve fibers . The model includes NMDA-mediated synaptic input from the C-fibers to the PNs that depends on postsynaptic activity . We also describe the use of a Poisson process to simulate neural spikes on the afferent fibers that represent the input from the DRG to the DH and the interactions between the afferent fibers incorporated in our model , respectively . In the Results section , we present validation studies for our model including reproduction of the wind-up phenomenon . With our principal aim to investigate the daily rhythmicity of pain sensation , we apply the model to predict the daily modulation of the pain inhibition phenomena . As a novel application of the model , we investigate effects of experimentally-observed dysregulation of inhibition within the DH circuit under neuropathic pain conditions ( i . e . , a chronic condition with persistent pain experience associated , e . g . , with peripheral nerve damage ) on the daily modulation of pain sensitivity . We find that dysregulation of Aβ-fiber dependent presynaptic inhibition of C-fiber signaling can account for it . Finally , we discuss limitations and future modifications , as well as importance and application , of our model in the Discussion .
We construct a model describing the spinal processing of nociceptive stimuli in humans by considering the average firing rate of three populations of neurons in the DH: the PNs ( P ) , inhibitory ( I ) interneurons , and excitatory ( E ) interneurons , in response to the average firing rate of the Aβ- , Aδ- , and C-afferent fibers ( see Fig 1 ) . In this work , we expand on a model developed in [4] , which follows the modeling approach similar to [26] with the exception that our model predictions are in terms of average firing rates of neuron populations [29] instead of average membrane potentials . In contrast to our previous model in [4] , the new elements of the model introduced in this work consist of including i ) Poisson processes to generate spiking activity on the input nerve fibers , ii ) NMDA receptor-mediated synaptic interactions and iii ) an additional inhibitory interneuron population I2 , and iv ) removal of the connection to the midbrain . These four modifications allow us to i ) represent a biologically realistic fiber input to which the model is robust , ii ) reproduce experimentally observed frequency effects during wind-up , iii ) expand the model parameter range that replicates patterns seen in experiments on neuropathy , and vi ) focus solely on modeling spinal-cord processing of pain , respectively . As concerns the general structure of connections between the neuronal populations in the circuit ( dashed rectangle in Fig 1 ) , we follow previous models of pain and use the circuitry presented , e . g . , in [21] . Briefly , PNs receive direct synaptic input from the three afferent fiber types , Aβ-fibers excite inhibitory interneurons and C-fibers excite excitatory interneurons . Both interneuron populations synapse onto the PNs and the inhibitory interneurons inhibit the excitatory interneurons . We also include Aβ-dependent presynaptic inhibition of C-fiber activity mediated through an additional inhibitory interneuron population ( I2 ) that is modeled indirectly [see Eq ( 5 ) ] . We assume that the input to our model circuit is a stimulation of the afferent fibers that has been pre-processed in the DRG . Based on fiber input and the connections between the neuron populations in the DH , our model computes the activity of the PNs , ( P in Fig 1 ) , whose output directly corresponds to the amount of pain experienced [32] . We note that there are many nuances in the perception of pain , including those originating in the cortex; however , we model the portion of pain that stems from nociceptive input to the spinal cord since it has been shown that pain perception correlates strongly with the firing rate of the PNs in the spinal cord [32 , 33] . According to the formalism of firing-rate models , e . g . , [29] , we assume that the rate of change of the average firing rate in spikes per second ( Hz ) of the projection , inhibitory , and excitatory neuron populations , fP , fI , and fE , respectively , is determined by a nonlinear response function ( see Fig 2A ) . These response functions determine the average firing rate response of a neuron population to a combination of external inputs ( i . e . , stimulations of the afferent fibers pre-processed in the DRG ) and the firing rates of the presynaptic neuron populations ( see Fig 1 ) . In the absence of input from other neuron populations and afferent fibers , the average firing rate of the neuron population decays exponentially . These assumptions yield the following set of equations for the average firing rate of each population: d f P d t = P ∞ ( g A β P f A β ( t ) + g A δ P f A δ ( t ) + ( g CP + g NMDA ) f C ( t ) + g EP f E - g IP f I ) - f P τ P , d f E d t = E ∞ ( g CE f C ( t ) - g IE f I ) - f E τ E , d f I d t = I ∞ [ g A β I f A β ( t ) ] - f I τ I , ( 1 ) where t is time in seconds , τP = 0 . 001 s , τE = 0 . 01 s , and τI = 0 . 02 s are the intrinsic time scales of the projection , excitatory , and inhibitory neuron populations , respectively . Weights gij denote the strength of the external input or connections from presynaptic neuron populations i ( i = Aβ , Aδ , C , P , E , I ) to neuron population j ( j = P , E , I ) . We indicate inhibitory synaptic input with a negative sign and excitatory synaptic input with a positive sign . We define the functions ( of time t ) for the external inputs , fAβ ( t ) , fAδ ( t ) , and fC ( t ) in the next section . The model includes N-methyl-D-aspartate ( NMDA ) type synapses from the C-fibers to the P population in the following way: to represent postsynaptic voltage-dependent removal of the magnesium ( Mg+ ) block on NMDA receptors , we assume that the synaptic weight , gNMDA , depends on the average firing rate of the P population ( fP ) , and thus , we consider gNMDA as a variable that changes as a function of time ( see Fig 2B ) , similar to [34]: d g NMDA d t = M ∞ ( f P ) - g NMDA τ NMDA , ( 2 ) where τNMDA = 1 s is the intrinsic time scale of the synaptic weight , gNMDA . We assume a sigmoidal shape for the monotonically increasing firing rate response functions of the neuronal populations P∞ , E∞ , I∞ , and the synaptic weight response function M∞ , and use hyperbolic tangent functions to represent them as follows: P ∞ ( x ) = max P 1 2 [ 1 + tanh ( 1 α P ( x - β P ) ) ] , E ∞ ( x ) = max E 1 2 [ 1 + tanh ( 1 α E ( x - β E ) ) ] , I ∞ ( x ) = max I 1 2 [ 1 + tanh ( 1 α I ( x - β I ) ) ] , M ∞ ( x ) = max M 1 2 [ 1 + tanh ( 1 α M ( x - β M ) ) ] , ( 3 ) where maxP , maxE , maxI , and maxM are the maximum firing rates of the projection , excitatory , and inhibitory populations , and the maximum synaptic strength of the NMDA-mediated input , respectively . In Eq ( 3 ) , the shape of the response functions is determined by the input x at which the average firing rate of the projection , excitatory , and inhibitory neuron population reaches half of its maximum value , x = βP , x = βE , and x = βI , respectively ( see Fig 2A ) . The slope of the transition from non-firing to firing in the projection , excitatory , and inhibitory neuron population is given by 1/αP , 1/αE , and 1/αI , respectively . The activation of the NMDA synapse , M∞ ( fP ) , is modeled as an increasing function of the firing rate of the projection neurons , representing the resulting increase in synaptic strength as postsynaptic membrane potentials depolarize and the magnesium block of the NMDA receptors is released [34] . We choose parameter values for the response functions in such a way that the input-output curve of the projection , excitatory , and inhibitory neuron populations agrees qualitatively with experimental observations . Hence , we assume the inhibitory interneuron population has a nonzero resting firing rate , as has been reported in [1 , 2] , and a higher maximum firing rate than that of the projection and excitatory interneuron populations , as has been assumed in a biophysically detailed model of the DH circuit [21] . In our model assumptions for the response functions , we mimic the model predictions of [21] that agree with data from experimental observations in [35 , 36] . As concerns the NMDA activation , we assume a similar sigmoidal shape but with a very slow rise time modeling the slow removal of magnesium ions from blocking the NMDA receptors with increase in cell activity . As the magnesium blockage is removed , the NMDA channels are clear to be activated and further depolarize the cell , resulting in an increase in the firing rate of the PNs . The function M∞ ( fP ) models this activation of the NMDA channels resulting from the removal of the magnesium ions ( see Fig 2B ) . All values of the parameters discussed above that we use in the numerical simulations of our model are listed in S1 Table . To model input from the DRG , we simulate 1000 afferent fibers of three types that project to the DH . We note that our choice of 1000 fibers is based on the number of afferent fibers experimentally observed in one nerve bundle that projects to a skeletal muscle in the rat [37] , which is on the order of 1000 [37 , 38] . These three afferent fiber types differ not only in diameter sizes but also in the level of myelination . As a result , impulses are transmitted at different speeds in the three fiber types . The majority ( 82% ) of these fibers are slow C-fibers ( with an average conduction velocity of 0 . 5-2 m/s ) , 9% are Aδ-fibers ( with an average conduction velocity of 5-30 m/s ) , and 9% are Aβ-fibers ( with an average conduction velocity of 30-70 m/s ) [3 , 38] . We assume that the times of initiation of activity in each of these fibers in response to nociceptive stimulation are roughly equivalent , resulting in the distribution of arrival times to the DH that has been experimentally observed , e . g . , in Fig 1 of [39] . We aim to model nerve fiber activity from a brief nociceptive stimulus at the periphery ( see Fig 1a in [39] ) . To do this , we use a Poisson process to simulate spike trains in the afferent fibers at a given firing rate . The activity of the afferent fibers in response to a brief nociceptive stimulus at t = 0 . 5 s can be seen in the raster plots in Fig 3A , where each small bar represents one spike/action potential and each row represents the activity in one afferent fiber over the course of 1 second . We consider the activity in 90 Aβ- , 90 Aδ- and 820 C-fibers with baseline frequency of 1 Hz and stimulus response frequency of 40 , 20 , and 20 Hz , respectively . Each fiber has an increased firing rate for a set amount of time ( 10 ms for both Aβ- and Aδ-fibers and 210 ms for C-fibers ) chosen to replicate the response in the PNs as measured experimentally in [39] . We choose these increased firing rates for the afferent fibers to simulate a response to a nociceptive stimulus ( see [33] and [40] for spiking dynamics of afferent fibers in response to varying levels of nociceptive stimuli ) and a low background drive to simulate spontaneous activity of the fibers [41] . To compute the average firing rate in each of the three fiber groups , we compute an instantaneous firing rate by counting the number of spikes in a one-millisecond window of time , and then use a moving average with a time window of 10 ms to create a smooth firing rate function . As a result , our simulated input to the spinal cord on fibers with different conductance speeds reproduces the observed pattern [39] of fast , brief Aβ- and Aδ-fiber activity ( i . e . , first pain ) followed by delayed , longer lasting C-fiber activity ( i . e . , second pain ) . When simulating our model , we use these smoothed average firing rates ( see Fig 3B ) representing the response in the three fiber groups to a brief nociceptive stimulus as input to the DH circuit model . Pain sensitivity follows a daily cycle in many clinical conditions [5] . There is strong evidence supporting rhythmicity in response to acute nociceptive stimuli [8 , 11–13 , 42] . In experiments where a rhythm in pain sensitivity was detected , its pattern is remarkably consistent , with pain sensitivity peaking during the hours when there is no daylight ( and when humans are typically asleep ) , that is , from midnight to 5 AM [5] . In previous work , we analyzed experimental data reporting on the daily rhythm in human pain sensitivity from four studies investigating: 1 ) the threshold for forearm pain in response to heat ( n = 39 , [40] ) , 2 ) the threshold for tooth pain in response to cold ( n = 79 , [13] ) , 3 ) the threshold for tooth pain in response to electrical stimulation ( n = 56 , [13] ) , and 4 ) the threshold for nociceptive pain in response to electrical current ( n = 5 , 8] ) . The data points from these studies are shown in Fig 4A and details on the derivation of these data points can be found in [6] . We note here that we aligned the data to the subject’s typical or scheduled wake time ( i . e . , 0 hours after wake ) and thus , clearly , this data represents a daily rhythm in pain sensitivity that includes sleep-wake-cycle effects that cannot be uncoupled from an endogenous circadian rhythm . The result is that , in this work , we discuss pain sensitivity as a function of hours since morning wake time to align our results with these data sources . The data strongly suggest a sinusoidal profile , and thus we fit a sinusoidal function to the data using Matlab’s [43] curve fitting scheme ( cftool ) ( see solid curve in Fig 4A ) . We hypothesize that this best-fit sinusoid ( R2 = 0 . 73 and root mean-square error of 4 . 69 ) represents a prototypical daily rhythm in pain sensitivity for humans , with a sharp peak in pain sensitivity occurring close to midnight ( following 18 hours of waking ) , and that then decreases during the night to reach a minimum in pain sensitivity in the afternoon ( following 9 hours of wake , or approximately 4pm ) . Experimental work also suggests a daily rhythm in the sensitivity of touch ( see Figs 1 and 2 in [13] ) with the highest sensitivity for tactile discrimination occurring in the late afternoon and the lowest sensitivity in the late morning [13] . Since cells in the DRG ( that contains the cell bodies of the afferent fibers ) rhythmically express clock genes responsible for generating rhythmicity of other physiological processes [15] , we assume in our model that daily modulation occurs at the level of primary afferent input to the spinal cord . Furthermore , these experimental observations motivate us to introduce rhythmicity in the model input from Aβ-fibers that exhibits nearly a 12-hr shift from the rhythm of the C-fiber-model inputs . We note here that although we consider rhythmicity in the Aβ-fibers [13] , our modeling work focuses on describing processing of nociceptive stimuli . Thus , our model does not simulate processing of strictly mechanical stimuli which may use different circuitry from that of nociceptive stimuli . We use the sinusoidal curve obtained from fitting the experimental data in Fig 4A , with the slight modification of making the period exactly 24 hours , to modulate the Aβ- and C-fiber activity as a function of the time of day in hours since typical morning wake time . We implement daily rhythmicity in the firing rates of the Aβ- and C-fibers by varying their stimulation response frequencies , R A β ( t ^ ) and R C ( t ^ ) , respectively , with approximately opposite phases . The average firing rates of the fibers ( 40 Hz for Aβ- fibers and 21 Hz for C-fibers ) were estimated from experiments of receptor activity in the human hand [40] . This yields equations for the firing rates of the fibers over the day as follows: R A β ( t ^ ) = 6 sin ( π 12 t ^ ) + 40 , R C ( t ^ ) = 1 2 sin ( π 12 t ^ + 2 . 8 ) + 21 , ( 4 ) where t ^ denotes time , in hours since morning wake time ( see blue and green curves in Fig 4B ) . The amplitudes of the daily modulation of response frequencies ( ±6 Hz for Aβ-fibers and ±0 . 5 Hz for C-fibers ) were chosen to fit the model’s simulated pain signal , namely the firing rate of the projection neuron population , to the experimental measurements of pain sensitivity , as described below . To model the effects of the Aβ-dependent presynaptic inhibition of C-fiber activity mediated through an additional inhibitory interneuron population ( I2 ) , we assume that the I2 population is only activated by high , stimulus-induced activity of the Aβ-fibers and that its activity tracks the daily modulation of R A β ( t ^ ) but at a lower firing rate . As a result , presynaptic inhibition lowers the stimulus response frequency of C-fiber activity , R C ( t ^ ) , as follows: R C eff ( t ^ ) = R C ( t ^ ) - g A β C ( R A β ( t ^ ) - 30 ) , ( 5 ) where gAβC scales the effects of the presumed I2 activity ( see black dashed curve in Fig 4B ) , and the -30 mimics the lower I2 firing rate . This presumed level of I2 activity maintains effective C-fiber activity on the same scale as the original C-fiber activity , see blue solid and black dashed lines in Fig 4B . We note that while the daily modulation of the stimulus response frequencies governing spikes on the afferent fibers is on the order of hours , our model output changes on the order of fractions of seconds ( e . g . , τP = 0 . 001 s ) . Because of such a difference in time scales , there is only a small change in the stimulus frequencies R A β ( t ^ ) and R C ( t ^ ) during the response to a brief nociceptive stimulus . Hence , we consider specific time points at a constant t ^ in a 24-hour period ( see Fig 4B ) when generating the ( daily modulated ) response of afferent fibers to stimulation . We compare the 24-hour rhythm in pain sensitivity computed by our model with the sinusoidal curve representing the human daily pain sensitivity fitted to experimental data in Fig 4A . Introducing the rhythmicity of fiber responses described above , we simulate our model equations at 7 time points over the 24-hour day , recording our model output ( firing rate of the projection ( P ) neuron population ) for each time point . To compare with the experimental curve , we compute variation as a percent of the mean by calculating the mean of the average response firing rates of the P population to stimuli given over the whole day , and comparing the firing rate at each time point during the day to that mean firing rate . Fig 5A shows the model pain sensitivity as a percent of the average over the day ( blue curve ) as compared to the experimental pain sensitivity ( black dashed curve ) . Notice that the average firing rate of the P population , as shown in Fig 5B , is above 25 Hz which can be considered as a threshold for pain ( see [44] ) . Furthermore , for the daily rhythmicity of pain sensitivity , the model output represented in terms of percent of mean ( firing rate of the P population ) closely follows experimental results ( see Fig 5A ) .
To simulate response to a brief painful stimulus at the periphery , we construct average firing rate functions for activity of the Aβ- , Aδ- and C-fibers based on the time of day as input to the DH , and calculate the resulting behavior of the PNs as described by the equations in ( 1 ) . Fig 6 displays the average firing rate of the P population in response to nociceptive stimuli at two time points during the 24-hr day . Our model reproduces the average firing-rate pattern of the populations of neurons in the DH when the three afferent fibers differ in their conductance speeds , as noted by three distinct activations of the PNs in Fig 6 . We follow [44] , and interpret the painful response as the firing rate of the PNs crossing a threshold of 25 Hz . The average firing rate of the P population is qualitatively similar to that seen experimentally ( e . g . , see Fig 1a in [39] ) and agrees with the daily variation in pain as reported in [13] ( lower sensitivity in the afternoon and higher sensitivity at night ) . Note here that we are only considering nociceptive stimulation of the afferent fibers as mechanical stimulation may follow a different circuit within the DH or more complicated activation of the different afferent fibers . To quantify the amount of pain experienced from the stimulation of the afferent fibers , we take the average firing rate of the PNs over the period of time when the C-fibers’ response has reached the DH ( see blue rectangle in Fig 6 ) . Note that the amount of time that the C-fiber response is activated is constant across the day and we consider the average firing rate above 25 Hz as painful [44] . The parameters for this model were chosen to give painful responses ( i . e . , firing rate of the P population above 25 Hz ) , but also to allow the neuron populations to reach their maximal firing rates during times of day with highest pain sensitivity . We note that the input from the spinal cord is only one component to the overall experience of pain . The P population reaching a maximum represents the maximum possible nociceptive response from this portion of the spinal cord . Thus , ( and as concerns all of the simulations of our model ) a maximal firing rate of the P population does not necessarily correspond to the maximal pain experience . Additionally , the chosen parameter set allows our model to sufficiently capture experimentally-observed phenomena such as wind-up and pain inhibition , but we recognize that this is not the only set of parameters that would yield these results . For a complete description of the parameter value choices , see S1 Table . In addition to the example model output in Fig 6 , we further validate our DH circuit model by showing that it reproduces wind-up —that is , increased ( and frequency-dependent ) excitability of the neurons in the spinal cord due to repetitive stimulation of afferent C-fibers [45] . Wind-up serves as an important tool for studying the role of the spinal cord in nociception and has often been used as an example phenomenon to validate single neuron models of the DH ( see [21 , 27 , 28] , for example ) . However , both the physiological meaning and the generation of wind-up remain unclear ( see [46] for a review ) . There are several possible molecular mechanisms proposed for the generation of wind-up [46] . Earlier work on single neuron models suggests that wind-up is generated by a combination of long-lasting responses to NMDA-receptor-mediated synaptic currents and membrane calcium currents providing for cumulative depolarization of the PNs [27] . Indeed , calcium conductances and NMDA receptors of the projection/deep dorsal horn neurons are included in all previous models of the DH circuit [21 , 27 , 28] . In addition , the study done in [28] emphasizes the effect ( direct or via influencing the dependence of the deep dorsal horn neurons on their intrinsic calcium currents ) NMDA and inhibitory synaptic conductances have on the extent of wind-up in the deep DH neurons [28] . As noted in the Methods section , we incorporate NMDA synapses into our model for the DH circuit by taking into account that the dynamics of the synaptic weight of the connection from the C-fibers to the PNs , gNMDA , depends on the average firing rate of the P neuron population [see Eq ( 2 ) ] . We assume that the dynamics of gNMDA are much slower than those of the neuron populations ( τNMDA = 1 s while , e . g . , τP = 0 . 001 s ) . As a result , in response to a repeated stimulus ( i . e . , when the model input as shown in Fig 3 is presented to the DH circuit at a frequency of 2 Hz ) , the average firing rate of the P population during the C-response increases ( see top panel in Fig 7A ) and the synaptic weight gNMDA exhibits slowly increasing dynamics in response to the increased activity in the P population ( see bottom panel in Fig 7A ) . For a repeated stimulus at 2 Hz , the latency , which we consider as the time from the start of the stimulus ( t = 0 . 5 s ) to the time when the average firing rate of P exceeds 25 Hz ( i . e . , considered as painful ) , decreases with the stimulus index ( i . e . , index 1 denotes the first stimulus in the repeated sequence ) , see Fig 7B , as seen in experiments [47] . However , the increase in the average firing rate of the P population depends on the frequency of the repeated stimulation , with optimal effects seen experimentally at stimulation frequencies between 1-3 Hz [46] . Our model captures the phenomenon of wind-up , as well as the frequency dependency . For example , when the model input is repeated at a frequency of 2 Hz , the mean of the average firing rate of the P population during the C-response ( see blue box on bottom of Fig 6 ) increases from about 25 Hz during the first stimulus to about 50 Hz during the fifth stimulus similar to previous modeling results [21] , while in the case of a stimulus repeated at 0 . 5 Hz , the mean P firing rate during the C-response does not change as a function of the stimulus index ( Fig 8A , yellow curve vs blue curve ) . We note that we simulate frequencies up to 3 . 22 Hz as this is the highest frequency we can model without an overlap in the P neuron responses ( see Fig 7A , top ) . We include it here to show the general trend of wind-up in response to an increase in frequency . It’s clear to see that as the frequency increases , and the responses are allowed to interact , the result would be a yet faster rise in the firing rate to its maximum due to the additive nature of the NMDA weight ( see Fig 7A , bottom ) . We also show that the latency time decreases with increasing frequency , ( see Fig 8B ) , with maximal effects seen for stimulation frequencies of 2-3 Hz and minimal effects seen for 0 . 5 Hz , as observed experimentally . It has been experimentally observed that stimulation of A-fiber afferents can lead to inhibition of the activity of the PNs that typically follows from stimulation of C-fiber afferents [21] . This is related to the idea that when you stub your toe , you immediately apply pressure on the toe and feel some lessening of pain . To capture this phenomenon in our model , we simulate a brief painful stimulus at the periphery that activates all three fibers ( stubbing of the toe ) and then deliver a second brief stimulation to the Aβ-fibers a short time thereafter ( pressure applied to toe ) , shown in Fig 9 by the red arrows . The arrival time of the second pulse to the Aβ-fibers is increased by 50 ms in each simulation , and the response in the projection neurons is shown in blue . For comparison with experimental data in [47] and model simulations in [21] , we visualize the average firing rates of P predicted by our model ( Fig 9A ) with a spike raster plot in Fig 9B . That is , we derive firing times from the numerically computed average firing rates of the P population , as explained in [48] . As the timing of the second pulse gets closer to the arrival of the C-fiber stimulation at the DH , there is a brief period of excitation followed by a longer period of inhibition , as seen in experiments [47] . While only qualitative descriptions of pain inhibition are reported in [47] , we quantify the amount the painful response is suppressed by the second activation of the Aβ-fibers by comparing the average firing rate of the P population during the C-response in each panel of Fig 9A ( thick curves ) to that in the bottom panel in Fig 9A where the secondary Aβ activation has no effect on the C-fiber response ( defining a baseline firing rate ) . The percent of this baseline firing rate is plotted in Fig 10A as a function of the delay time of the second Aβ pulse ( relative to the time of the original nociceptive stimulus ) . Note that the pain response decreases as the delay of the second Aβ pulse increases and its arrival time coincides with the C-response , as reported in [47] . We use our model of pain sensitivity to investigate the daily rhythmic effects on the phenomenon of pain inhibition . Fig 10B demonstrates changes in the percent of baseline firing rate of the P population during the C-response as a function of the delay in the second Aβ pulse , for each time of day ( i . e . , in hours since morning wake time ) . Our model predicts that pain inhibition is most effective during early afternoon ( 4-8 hours after wake ) , when the Aβ-fibers are the most sensitive to external stimulus ( i . e . , their stimulation frequency is at its highest daily value ) and the C-fibers are the least sensitive to external stimulation . This can be seen in the color plot in Fig 10B by the dark horizontal band around 4-8 hours after wake ( middle of the afternoon ) for all delays . Notice that for 16-20 hours after wake ( middle of the night ) , the pain percentage is very high and there is little change in the percent of pain as a function of delay time , indicating that pain inhibition is not very effective at these times . In addition to predicting the time of day that pain inhibition is most effective ( mid-afternoon ) , our model also predicts that a delay from 0 . 05 to 0 . 15 seconds after the original painful stimulus is ideal for the optimal lessening of pain experienced , as can be seen in both plots of Fig 10 by these particular delay times showing the lowest percent of pain response for all times of day . Neuropathic pain occurs due to various conditions involving the brain , spinal cord , and nerve fibers . It is distinguished from inflammatory conditions like arthritis in that it often appears in body parts that are otherwise normal under inspection and imaging , and is also characterized by pain being evoked by a light touch . Experiments on pain sensitivity in neuropathic patients suggest that neuropathic pain has a daily rhythm as well [15 , 49–52] , having its peak in the afternoon [53] . An afternoon peak in pain sensitivity is the reverse of the daily rhythm in pain sensitivity under normal conditions [6] . Nerve injury can cause a dysregulation of chloride ion transporters that control intracellular chloride concentration in DH neurons ( reviewed in [54] ) . Maintenance of a low intracellular chloride concentration is important for the functioning of inhibitory neurotransmission . Under typical conditions , the binding of the neurotransmitter GABA on postsynaptic receptors produces an inhibition of postsynaptic activity by allowing negatively-charged chloride ions to flow into the postsynaptic neuron , thus producing hyperpolarization ( or decrease in membrane voltage ) . If intracellular chloride concentrations stay semi-permanently elevated , chloride ions may flow out of the cell in response to GABA receptor activity producing excitatory rather than inhibitory effects . Several authors have hypothesized that dysregulation of inhibition in spinal pain processing circuits could explain the development of pain sensation in response to non-noxious stimuli under neuropathic conditions [54 , 55] . Specifically , several authors [54 , 56 , 57] implicated a switch in presynaptic inhibition to presynaptic excitation in the DH as one culprit for eliciting neuropathic pain phenomena . As a result , we set out to determine if a switch from presynaptic inhibition to presynaptic excitation in our model is sufficient to replicate the experimentally-observed 8-12 hour change in the phasing of daily rhythms in pain sensitivity under neuropathic conditions . We show that our model can capture such an inversion of the rhythmicity of the firing rate of the PNs with a change from inhibition ( normal conditions ) to excitation ( neuropathic conditions ) in the presynaptic influence of the Aβ-fibers on C-fiber synaptic signaling . The location where the change from inhibition to excitation occurs is denoted in our model diagram by the two asterisks in Fig 1 . Thus , we assume that under neuropathic conditions , the connection from I2 to the synaptic terminals of E and P is excitatory instead of inhibitory . Recall from the Methods section that we model presynaptic inhibition as an Aβ-dependent decrease in the stimulus response firing rate of the C-fibers [see Eq ( 5 ) ] . The assumption that presynaptic inhibition turns to excitation results instead in an Aβ-dependent increase in the C-fiber stimulus response firing rate represented by the following equation R C eff neuro ( t ^ ) = R C ( t ^ ) + g A β C neuro ( R A β ( t ^ ) - 30 ) , ( 6 ) where g A β C neuro is the strength of the effect of Aβ-fiber activity on C-fiber activity under neuropathic conditions ( see red curve in lower panel of Fig 11A ) . This daily variation in the stimulus response frequency of C-fiber activity results in the desired inversion of projection neuron population firing rate response to a brief nociceptive stimulus ( Fig 11C ) , and thus pain sensitivity ( Fig 11B ) , across the day . Thus , under normal conditions , the pain sensitivity rhythm follows the daily rhythm of the C-fibers ( compare blue curves in all panels ) but mimics the rhythm in the Aβ-fibers under neuropathic conditions ( compare red curves in B and C with green curve in A ) . In our model , we obtain this inversion of rhythm in pain sensitivity by assuming that Aβ-dependent presynaptic excitation under neuropathic conditions has a larger magnitude than presynaptic inhibition under normal conditions . Specifically , the weighting factor g A β C neuro = 0 . 25 under neuropathic conditions is larger than gAβC = 0 . 05 under normal conditions . This can be interpreted as an increase in firing rates of the Aβ-fibers under neuropathic conditions that results in increased excitation of the I2 inhibitory population , and thus larger magnitude of presynaptic excitation compared to presynaptic inhibition under normal conditions . There are several proposed mechanisms for the many types of neuropathic pain , some of which show increased activity of the Aβ-fibers [41] . To investigate the dependence of the magnitude of Aβ-dependent presynaptic excitation on the inverted daily rhythm , we simulate the model response to brief nociceptive stimuli across the day for different values of the weighting parameter g A β C neuro ( see Fig 12 ) . Results show that weak presynaptic excitation ( black and blue curves ) reduces the amplitude of daily variation in P population firing rates and does not induce an inverted rhythm . For larger values of g A β C neuro , the correct rhythmicity is obtained and amplitude increases but eventually saturates . Larger magnitudes of presynaptic excitation only serve to increase the firing rate over the entire day , thus increasing the average over the entire day and not affecting the variation in percent of the mean . Note that the amplitude of the rhythmicity of pain sensitivity under neuropathic conditions is small , about 5% as compared to 15% under normal conditions . There are few experimental studies that measure the amplitude of modulation of pain sensitivity under neuropathic conditions; however , one study shows neuropathic pain sensitivity to have a similar amplitude to , if not slightly larger than , acute pain [51] . Our model proposes that the rhythm is intrinsic to the afferent fibers , however many believe that there may also be daily rhythms within the top-down inhibitory modulation of many of the neuronal populations in the pain-processing circuit [58] . With this initial hypothesis of rhythmicity in the fiber input , our model replicates the overall increase in pain sensitivity under neuropathic conditions , reflected by increased firing rates of the PNs . Indeed , our model simulations suggest that inhibition turned excitation at the level of the fibers is a possible mechanistic explanation for the inversion of pain sensitivity rhythms seen under neuropathic conditions . Modulation by daily rhythms could also be explored in alternative parts of the pain processing circuit , including the output of the projection neurons and its propagation along the spinal cord . These additional mechanisms in combination with disinhibition may enhance the modulation of the daily rhythm of pain sensitivity under neuropathic conditions .
We have developed a firing-rate model for the processing of nociceptive stimuli in the DH of the spinal cord , with a particular interest in investigating the daily rhythmicity of pain sensitivity . Our model follows the formalism of many neuron firing-rate-based models , but to our knowledge , it is novel for pain processing in the spinal cord . In addition to accounting for typical pain phenomena such as wind-up and pain inhibition , our model captures the rhythmicity in pain sensitivity over the 24-hour day mediated by intrinsic rhythmicity of afferent fiber activity . We include experimentally-justified presynaptic inhibition from the Aβ-fibers to the C-fibers , and show how disinhibition of this pathway under neuropathic conditions is sufficient to induce the experimentally-observed inversion of the rhythmicity of pain sensitivity . Our minimalistic model is based on physiology and thus provides an accessible theoretical framework for experimental and clinical investigations of diverse physiological processes modulating pain processing in humans . In contrast to a detailed biophysical model of a single neuron [21 , 27 , 28] or a large-scale network of individual neurons [44] , we construct equations to describe the population activity of projection , inhibitory , and excitatory neurons in the DH . As a result , we work with average firing rates for each of the three neuron populations according to the formalism developed in [29] . Therefore , our modeling approach is similar to [26] but our model predictions are in terms of average firing rates of neuron populations instead of potentials of individual cells . In our choice of model formalism , we assume that the neurons in each population behave similarly , i . e . , they receive similar inputs and respond similarly to those inputs , such that we can consider the average behavior over all neurons in each population as the primary mode of information transfer in the circuit . This is a limitation in the sense that often interesting phenomena in neuroscience arises from the nonlinear interactions between neuron spike timings and their differences in interpreting incoming stimuli . However , results from other modeling approaches that replicate spiking behavior [21 , 44] have not indicated that discrimination of spike timings contributes substantially to spinal pain processing . Additionally , some parameters in this model formalism cannot be easily obtained from experiments . For example , the weights with which one population influences another [see gPrePost in Eq ( 1 ) ] , represent an average synaptic strength from all neurons in one population to all neurons in another , which cannot be measured experimentally . We choose parameter values for these weights in order to replicate experimental data on the response of the PNs under different conditions . Finally , although there is experimental evidence to show that an increase in the activity of the PNs correlates with an increase in pain sensation [32] , the choice of instituting a threshold of 25 Hz on the PNs above which the model output is considered painful is somewhat arbitrary . However , we follow this convention used in [44] because to our knowledge a more physiologically accurate approximation has not yet been determined . The circuitry of our DH model is based on the gate control theory of pain [22 , 24] , similar to previous mathematical models for spinal nociception processing [21 , 26–28] . While using different model formalisms , circuit activity in these models centers around inhibition of PN responses to C-fiber input by Aβ-fiber activity . In this way , Aβ-fiber activity gates responses to nociceptive stimuli . More recent results have called into question gate control theory [25] . In particular , a large-scale network model of spinal cord neural circuitry has been constructed [44] that includes numerous known cell types , their laminar distribution , and their modes of connectivity . This model has been used to investigate the mechanisms of pain relief through dorsal column stimulation ( DCS ) , a procedure to treat neuropathic pain . The results shown in [44] identify limitations of the gate control theory and propose alternate circuitry that more accurately accounts for the effects of DCS on nociceptive and neuropathic pain . As concerns our model predictions for neuropathy , the low amplitude of the neuropathic pain rhythm in the model output may suggest that a simple spinal cord model is not sufficient to completely describe the phenomenon of an inversion in the rhythm of pain modulation under neuropathic vs normal conditions . Indeed , the daily rhythm that we use in the model is likely to reflect both the influences of circadian rhythms and sleep homeostasis , of which the sleep homeostatic component presumably increases throughout the evening , and therefore , would potentially amplify the peak in the neuropathic pain rhythm that occurs during that time . Furthermore , our current model does not include top-down modulation of spinal pain processing from the brain for which there is experimental evidence in support of circadian regulation of top-down inhibition [5 , 59] . In this study , we do not consider the neuropathic property in which patients experience pain in response to a non-noxious , mechanical stimuli . Instead , we restrict our attention to the response to nociceptive stimuli since mechanical stimulation may be processed by different pathways . Nonetheless , our model predicts that neuropathic conditions can , in part , be explained by Aβ-dependent presynaptic excitation of C-fiber synaptic signaling that is of a larger magnitude than the presynaptic inhibition that occurs under normal conditions . Specifically , to obtain the experimentally-observed inversion in the rhythmicity of pain sensitivity experienced by neuropathic patients , our model predicted an increase in the Aβ-dependence on C-fiber stimulus response [compare g A β C neuro in Eq ( 6 ) to gAβC in Eq ( 5 ) ] . This increase could potentially be due to increased response firing rates of Aβ-fibers , as well as by increased efficacy of the excitatory effects of the secondary inhibitory population I2 . These effects cause an increase in firing rates of PNs in response to brief nociceptive stimuli , but could also contribute to increases in PN responses in mechanical stimuli processing pathways . Additional studies on the interaction of the pathways processing non-noxious and nociceptive stimuli , and their properties under neuropathic conditions are needed to fully understand this phenomenon . Often it is difficult , if not impossible , to experimentally measure properties of individual neurons in vivo , and in response to all possible nociceptive ( and mechanical ) stimuli . Due to this lack of knowledge , it is often impractical to build detailed models of DH neurons in which many parameters would need to be determined from biological data . In this respect , simpler population firing-rate models , like the one presented here , have an advantage in that there are significantly fewer parameters and they are constrained by measurements of more accessible macroscopic properties of the circuit . We have developed a novel firing-rate model for the neural circuit in the DH that processes nociceptive stimuli and we have shown that it can capture the same experimentally-observed phenomena as more detailed models . Additionally , we were able to clearly propose and test a mechanism for the daily rhythm in pain sensitivity and modulations of that rhythmicity under neuropathic conditions . Given its accessibility compared to more detailed or larger biophysically-based models , our model is suitable for including experimental results , e . g . , on the activity of the afferent fibers , and appropriate for experimental and clinical investigations of diverse physiological influences on pain processing , such as the effects of sleep deprivation on pain sensitivity [60] or the mechanisms underlying the efficacy of spinal cord stimulation for treatment of chronic pain conditions . | Human pain sensitivity follows a daily ( ∼24 hour ) rhythm . In particular , humans experience the highest sensitivity to pain in the middle of night and lowest in the afternoon . Patients suffering from neuropathy , a disease resulting from nerve damage leading to an increase in pain sensitivity , experience an approximately 12-hour shift in their rhythmicity such that the highest sensitivity occurs in the afternoon . Neuropathy is a difficult condition to treat since it is often unfeasible to locate the damaged nerve and it is also unclear how this damage causes a shift in rhythmicity and an increase in pain . Understanding the mechanism underlying the shift in rhythmicity may lead to improvements in the knowledge of the transmission of pain from the damaged nerve to the pain-processing center in the spinal cord , and thus better treatment protocols . We have built a population-based model to describe this transmission with a particular focus on daily rhythms . We show that our model reproduces experimentally-observed rhythmicity of both normal pain responses , as well as neuropathic pain . Our model predicts that a potential mechanism underlying the shift in rhythmicity for neuropathic pain is a change in the interaction of the nerve fibers from inhibition to excitation . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"somatosensory",
"system",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"nervous",
"system",
"neuroscience",
"signs",
"and",
"symptoms",
"nerve",
"fibers",
"interneurons",
"neuropathic",
"pain",
"sensory",
"physiology",
"spinal"... | 2019 | Modeling the daily rhythm of human pain processing in the dorsal horn |
The blood cancer T cell large granular lymphocyte ( T-LGL ) leukemia is a chronic disease characterized by a clonal proliferation of cytotoxic T cells . As no curative therapy is yet known for this disease , identification of potential therapeutic targets is of immense importance . In this paper , we perform a comprehensive dynamical and structural analysis of a network model of this disease . By employing a network reduction technique , we identify the stationary states ( fixed points ) of the system , representing normal and diseased ( T-LGL ) behavior , and analyze their precursor states ( basins of attraction ) using an asynchronous Boolean dynamic framework . This analysis identifies the T-LGL states of 54 components of the network , out of which 36 ( 67% ) are corroborated by previous experimental evidence and the rest are novel predictions . We further test and validate one of these newly identified states experimentally . Specifically , we verify the prediction that the node SMAD is over-active in leukemic T-LGL by demonstrating the predominant phosphorylation of the SMAD family members Smad2 and Smad3 . Our systematic perturbation analysis using dynamical and structural methods leads to the identification of 19 potential therapeutic targets , 68% of which are corroborated by experimental evidence . The novel therapeutic targets provide valuable guidance for wet-bench experiments . In addition , we successfully identify two new candidates for engineering long-lived T cells necessary for the delivery of virus and cancer vaccines . Overall , this study provides a bird's-eye-view of the avenues available for identification of therapeutic targets for similar diseases through perturbation of the underlying signal transduction network .
Living cells perceive and respond to environmental perturbations in order to maintain their functional capabilities , such as growth , survival , and apoptosis . This process is carried out through a cascade of interactions forming complex signaling networks . Dysregulation ( abnormal expression or activity ) of some components in these signaling networks affects the efficacy of signal transduction and may eventually trigger a transition from the normal physiological state to a dysfunctional system [1] manifested as diseases such as diabetes [2] , [3] , developmental disorders [4] , autoimmunity [5] and cancer [4] , [6] . For example , the blood cancer T-cell large granular lymphocyte ( T-LGL ) leukemia exhibits an abnormal proliferation of mature cytotoxic T lymphocytes ( CTLs ) . Normal CTLs are generated to eliminate cells infected by a virus , but unlike normal CTLs which undergo activation-induced cell death after they successfully fight the virus , leukemic T-LGL cells remain long-term competent [7] . The cause of this abnormal behavior has been identified as dysregulation of a few components of the signal transduction network responsible for activation-induced cell death in T cells [8] . Network representation , wherein the system's components are denoted as nodes and their interactions as edges , provides a powerful tool for analyzing many complex systems [9] , [10] , [11] . In particular , network modeling has recently found ever-increasing applications in understanding the dynamic behavior of intracellular biological systems in response to environmental stimuli and internal perturbations [12] , [13] , [14] . The paucity of knowledge on the biochemical kinetic parameters required for continuous models has called for alternative dynamic approaches . Among the most successful approaches are discrete dynamic models in which each component is assumed to have a finite number of qualitative states , and the regulatory interactions are described by logical functions [15] . The simplest discrete dynamic models are the so-called Boolean models that assume only two states ( ON or OFF ) for each component . These models were originally introduced by S . Kauffman and R . Thomas to provide a coarse-grained description of gene regulatory networks [16] , [17] . A Boolean network model of T cell survival signaling in the context of T-LGL leukemia was previously constructed by Zhang et al [18] through performing an extensive literature search . This network consists of 60 components , including proteins , mRNAs , and small molecules ( see Figure 1 ) . The main input to the network is “Stimuli” , which represents virus or antigen stimulation , and the main output node is “Apoptosis” , which denotes programmed cell death . Based on a random order asynchronous Boolean dynamic model of the assembled network , Zhang et al identified a minimal number of dysregulations that can cause the T-LGL survival state , namely overabundance or overactivity of the proteins platelet-derived growth factor ( PDGF ) and interleukin 15 ( IL15 ) . Zhang et al carried out a preliminary analysis of the network's dynamics by performing numerical simulations starting from one specific initial condition ( corresponding to resting T cells receiving antigen stimulation and over-abundance of the two proteins PDGF and IL15 ) . Once the known deregulations in T-LGL leukemia were reproduced , each of these deregulations was interrupted individually , by setting the node's status to the opposite state , to predict key mediators of the disease . Yet , a complete dynamic analysis of the system , including identification of the attractors ( e . g . steady states ) of the system and their corresponding basin of attraction ( precursor states ) , as well as a thorough perturbation analysis of the system considering all possible initial states , is lacking . Performing this analysis can provide deeper insights into unknown aspects of T-LGL leukemia . Stuck-at-ON/OFF fault is a very common dysregulation of biomolecules in various cancer diseases [19] . For example , stuck-at-ON ( constitutive activation ) of the RAS protein in the mitogen-activated protein kinase pathways leads to aberrant cell proliferation and cancer [19] , [20] . Thus identifying components whose stuck-at values result in the clearance , or alternatively , the persistence of a disease is extremely beneficial for the design of intervention strategies . As there is no known curative therapy for T-LGL leukemia , identification of potential therapeutic targets is of utmost importance [21] . In this paper , we carry out a detailed analysis of the T-LGL signaling network by considering all possible initial states to probe the long-term behavior of the underlying disease . We employ an asynchronous Boolean dynamic framework and a network reduction method , which we previously proposed [22] , to identify the attractors of the system and analyze their basins of attraction . This analysis allows us to confirm or predict the T-LGL states of 54 components of the network . The predicted state of one of the components ( SMAD ) is validated by new wet-bench experiments . We then perform node perturbation analysis using the dynamic approach and a structural method proposed in [23] to study to what extent does each component contribute to T-LGL leukemia . Both methods give consistent results and together identify 19 key components whose disruption can reverse the abnormal state of the signaling network , thereby uncovering potential therapeutic targets for this disease , some of which are also corroborated by experimental evidence .
Boolean models belong to the class of discrete dynamic models in which each node of the network is characterized by an ON ( 1 ) or OFF ( 0 ) state and usually the time variable t is also considered to be discrete , i . e . it takes nonnegative integer values [24] , [25] . The future state of each node vi is determined by the current states of the nodes regulating it according to a Boolean transfer function , where ki is the number of regulators of vi . Each Boolean function ( rule ) represents the regulatory relationships between the components and is usually expressed via the logical operators AND , OR and NOT . The state of the system at each time step is denoted by a vector whose ith component represents the state of node vi at that time step . The discrete state space of a system can be represented by a state transition graph whose nodes are states of the system and edges are allowed transitions among the states . By updating the nodes' states at each time step , the state of the system evolves over time and following a trajectory of states it eventually settles down into an attractor . An attractor can be in the form of either a fixed point , in which the state of the system does not change , or a complex attractor , where the system oscillates ( regularly or irregularly ) among a set of states . The set of states leading to a specific attractor is called the basin of attraction of that attractor . In order to evaluate the state of each node at a given time instant , synchronous as well as asynchronous updating strategies have been proposed [24] , [25] . In the synchronous method all nodes of the network are updated simultaneously at multiples of a common time step . The underlying assumption of this update method is that the timescales of all the processes occurring in a system are similar . This is a quite strong and potentially unrealistic assumption , which in particular may not be suited for intracellular biological processes due to the variety of timescales associated with transcription , translation and post-translational mechanisms [26] . To overcome this limitation , various asynchronous methods have been proposed wherein the nodes are updated based on individual timescales [25] , [27] , [28] , [29] , [30] , including deterministic methods with fixed node timescales and stochastic methods such as random order asynchronous method [27] wherein the nodes are updated in random permutations . In a previous work [22] , we carried out a comparative study of three different asynchronous methods applied to the same biological system . That study suggested that the general asynchronous ( GA ) method , wherein a randomly selected node is updated at each time step , is the most efficient and informative asynchronous updating strategy . This is because deterministic asynchronous [22] or autonomous [30] Boolean models require kinetic or timing knowledge , which is usually missing , and random order asynchronous models [27] are not computationally efficient compared to the GA models . In addition , the superiority of the GA approach has been corroborated by other researchers [29] and the method has been used in other studies as well [31] , [32] . We thus chose to employ the GA method in this work , and we implemented it using the open-source software library BooleanNet [33] . It is important to note that the stochasticity inherent to this method may cause each state to have multiple successors , and thus the basins of attraction of different attractors may overlap . For systems with multiple fixed-point attractors , the absorption probabilities to each fixed point can be computed through the analysis of the Markov chain and transition matrix associated with the state transition graph of the system [34] . Given a fixed point , node perturbations can be performed by reversing the state of the nodes i . e . by knocking out the nodes that stabilize in an ON state in the fixed point or over-expressing the ones that stabilize in an OFF state . A Boolean network with n nodes has a total of 2n states . This exponential dependence makes it computationally intractable to map the state transition graphs of even relatively small networks . This calls for developing efficient network reduction approaches . Recent efforts towards addressing this challenge consists of iteratively removing single nodes that do not regulate their own function and simplifying the redundant transfer functions using Boolean algebra [35] , [36] . Naldi et al [35] proved that this approach preserves the fixed points of the system and that for each ( irregular ) complex attractor in the original asynchronous model there is at least one complex attractor in the reduced model ( i . e . network reduction may create spurious oscillations ) . Boolean networks often contain nodes whose states stabilize in an attracting state after a transient period , regardless of updating strategy or initial conditions . The attracting states of these nodes can be readily identified by inspection of their Boolean functions . In a previous work [22] we proposed a method of network simplification by ( i ) pinpointing and eliminating these stabilized nodes and ( ii ) iteratively removing a simple mediator node ( e . g . a node that has one incoming edge and one outgoing edge ) and connecting its input ( s ) to its target ( s ) . Our simplification method shares similarities with the method proposed in [35] , [36] , with the difference that we only remove stabilized nodes ( which have the same state on every attractor ) and simple mediator nodes rather than eliminating each node without a self loop . Thus their proof regarding the preservation of the steady states by the reduction method holds true in our case . We employed this simplification method for the analysis of a signal transduction network in plants and verified by using numerical simulations that it preserves the attractors of that system . In this work , we employ this reduction method to simplify the T-LGL leukemia signal transduction network synthesized by Zhang et al [18] , thereby facilitating its dynamical analysis . We also note that the first step of our simplification method is similar to the logical steady state analysis implemented in the software tool CellNetAnalyzer [37] , [38] . We thus refer to this step as logical steady state analysis throughout the paper . It should be noted that the fixed points of a Boolean network are the same for both synchronous and asynchronous methods . In order to obtain the fixed points of a system one can solve the set of Boolean equations independent of time . To this end , we first fix the state of the source nodes . We then determine the nodes whose rules depend on the source nodes and will either stabilize in an attracting state after a time delay or otherwise their rules can be simplified significantly by plugging in the state of the source nodes . Iteratively inserting the states of stabilized nodes in the rules ( i . e . employing logical steady state analysis ) will result in either the fixed point ( s ) of the system , or the partial fixed point ( s ) and a remaining set of equations to be solved . In the latter case , if the remaining set of equations is too large to obtain its fixed point ( s ) analytically , we take advantage of the second step of our reduction method [22] to simplify the resulting network and to determine a simpler set of Boolean rules . By solving this simpler set of equations ( or performing numerical simulations , if necessary ) and plugging the solutions into the original rules , we can then find the states of the removed nodes and determine the attractors of the whole system accordingly . For the analysis of basins of attraction of the attractors , we perform numerical simulations using the GA update method . The topology ( structure ) and the function of biological networks are closely related . Therefore , structural analysis of biological networks provides an alternative way to understand their function [39] , [40] . We have recently proposed an integrative method to identify the essential components of any given signal transduction network [23] . The starting point of the method is to represent the combinatorial relationship of multiple regulatory interactions converging on a node v by a Boolean rule:where uij's are regulators of node v . The method consists of two main steps . The first step is the expansion of a signaling network to a new representation by incorporating the sign of the interactions as well as the combinatorial nature of multiple converging interactions . This is achieved by introducing a complementary node for each component that plays a role in negative regulations ( NOT operation ) as well as introducing a composite node to denote conditionality among two or more edges ( AND operation ) . This step eliminates the distinction of the edge signs; that is , all directed edges in the expanded network denote activation . In addition , the AND and OR operators can be readily distinguished in the expanded network , i . e . , multiple edges ending at composite nodes are added by the AND operator , while multiple edges ending at original or complementary nodes are cumulated by the OR operator . The second step is to model the cascading effects following the loss of a node by an iterative process that identifies and removes nodes that have lost their indispensable regulators . These two steps allow ranking of the nodes by the effects of their loss on the connectivity between the network's input ( s ) and output ( s ) . We proposed two connectivity measures in [23] , namely the simple path ( SP ) measure , which counts the number of all simple paths from inputs to outputs , and a graph measure based on elementary signaling modes ( ESMs ) , defined as a minimal set of components that can perform signal transduction from initial signals to cellular responses . We found that the combinatorial aspects of ESMs pose a substantial obstacle to counting them in large networks and that the SP measure has a similar performance as the ESM measure since both measures incorporate the cascading effects of a node's removal arising from the synergistic relations between multiple interactions . Therefore , we employ the SP measure and define the importance value of a component v as:where NSP ( Gexp ) and NSP ( GΔv ) denote the total number of simple paths from the input ( s ) to the output ( s ) in the original expanded network Gexp and the damaged network GΔv upon disruption of node v , respectively . This essentiality measure takes values in the interval [0 , 1] , with 1 indicating a node whose loss causes the disruption of all paths between the input and output node ( s ) . In this paper , we also make use of this structural method to identify essential components of the T-LGL leukemia signaling network . We then relate the importance value of nodes to the effects of their knockout ( sustained OFF state ) in the dynamic model and the importance value of complementary nodes to the effects of their original nodes' constitutive activation ( sustained ON state ) in the dynamic model .
The T-LGL signaling network reconstructed by Zhang et al [18] contains 60 nodes and 142 regulatory edges . Zhang et al used a two-step process: they first synthesized a network containing 128 nodes and 287 edges by extensive literature search , then simplified it with the software NET-SYNTHESIS [42] , which constructs the sparsest network that maintains all of the causal ( upstream-downstream ) effects incorporated in a redundant starting network . In this study , we work with the 60-node T-LGL signaling network reported in [18] , which is redrawn in Figure 1 . The Boolean rules for the components of the network were constructed in [18] by synthesizing experimental observations and for convenience are given in Table S1 as well . The description of the node names and abbreviations are provided in Table S2 . To reduce the computational burden associated with the large state space ( more than 1018 states for 60 nodes ) , we simplified the T-LGL network using the reduction method proposed in [22] ( see Materials and Methods ) . We fixed the six source nodes in the states given in [18] , i . e . Stimuli , IL15 , and PDGF were fixed at ON and Stimuli2 , CD45 , and TAX were fixed at OFF . We used the Boolean rules constructed in [18] , with one notable difference . The Boolean rules for all the nodes in [18] , except Apoptosis , contain the expression “AND NOT Apoptosis” , meaning that if Apoptosis is ON , the cell dies and correspondingly all other nodes are turned OFF . To focus on the trajectory leading to the initial turning on of the Apoptosis node , we removed the “AND NOT Apoptosis” from all the logical rules . This allows us to determine the stationary states of the nodes in a live cell . We determined which nodes' states stabilize using the first step of our simplification method , i . e . logical steady state analysis ( see Materials and Methods ) . Our analysis revealed that 36 nodes of the network stabilize in either an ON or OFF state . In particular , Proliferation and Cytoskeleton signaling , two output nodes of the network , stabilize in the OFF and ON state , respectively . Low proliferation in leukemic LGL has been observed experimentally [43] , which supports our finding of a long-term OFF state for this output node . The ON state of Cytoskeleton signaling may not be biologically relevant as this node represents the ability of T cells to attach and move which is expected to be reduced in leukemic T-LGL compared to normal T cells . The nodes whose stabilized states cannot be readily obtained by inspection of their Boolean rules form the sub-network represented in Figure 2A . The Boolean rules of these nodes are listed in Table S3 wherein we put back the “AND NOT Apoptosis” expression into the rules . Next , we identified the attractors ( long-term behavior ) of the sub-network represented in Figure 2A ( see Materials and Methods ) . We found that upon activation of Apoptosis all other nodes stabilize at OFF , forming the normal fixed point of the system , which represents the normal behavior of programmed cell death . When Apoptosis is stabilized at OFF , the two nodes in the top sub-graph oscillate while all the nodes in the bottom sub-graph are stabilized at either ON or OFF . As shown in Figure 3 , the state space of the two oscillatory nodes , TCR and CTLA4 , forms a complex attractor in which the average fraction of ON states for either node is 0 . 5 . Given that these two nodes have no effect on any other node under the conditions studied here ( i . e . stable states of the source nodes ) , their behavior can be separated from the rest of the network . The bottom sub-graph exhibits the normal fixed point , as well as two T-LGL ( disease ) fixed points in which Apoptosis is OFF . The only difference between the two T-LGL fixed points is that the node P2 is ON in one fixed point and OFF in the other , which was expected due to the presence of a self-loop on P2 in Figure 2A . P2 is a virtual node introduced to mediate the inhibition of interferon-γ translation in the case of sustained activity of the interferon-γ protein ( IFNG in Figure 2A ) . The node IFNG is also inhibited by the node SMAD which stabilizes in the ON state in both T-LGL fixed points . Therefore IFNG stabilizes at OFF , irrespective of the state of P2 , as supported by experimental evidence [44] . Thus the biological difference between the two fixed points is essentially a memory effect , i . e . the ON state of P2 indicates that IFNG was transiently ON before stabilizing in the OFF state . In the two T-LGL fixed points for the bottom sub-graph of Figure 2A , the nodes sFas , GPCR , S1P , SMAD , MCL1 , FLIP , and IAP are ON and the other nodes are OFF . We found by numerical simulations using the GA method ( see Materials and Methods ) that out of 65 , 536 total states in the state transition graph , 53% are in the exclusive basin of attraction of the normal fixed point , 0 . 24% are in the exclusive basin of attraction of the T-LGL fixed point wherein P2 is ON and 0 . 03% are in the exclusive basin of attraction of the T-LGL fixed point wherein P2 is OFF . Interestingly , there is a significant overlap among the basins of attraction of all the three fixed points . The large basin of attraction of the normal fixed point is partly due to the fact that all the states having Apoptosis in the ON state ( that is , half of the total number of states ) belong to the exclusive basin of the normal fixed point . These states are not biologically relevant initial conditions but they represent potential intermediary states toward programmed cell death and as such they need to be included in the state transition graph . Since the state transition graph of the bottom sub-graph given in Figure 2A is too large to represent and to further analyze ( e . g . to obtain the probabilities of reaching each of the fixed points ) , we applied the second step of the network reduction method proposed in [22] . This step preserves the fixed points of the system ( see Materials and Methods ) , and since the only attractors of this sub-graph are fixed points , the state space of the reduced network is expected to reflect the properties of the full state space . Correspondingly , the nodes having in-degree and out-degree of one ( or less ) in the sub-graph on Figure 2A , such as sFas , MCL1 , IAP , GPCR , SMAD , and CREB , can be safely removed without losing any significant information as such nodes at most introduce a delay in the signal propagation . In addition , we note that although the node P2 has a self-loop and generates a new T-LGL fixed point as described before , it can also be removed from the network since the two fixed points differ only in the state of P2 and thus correspond to biologically equivalent disease states . We revisit this node when enumerating the attractors of the original network . In the resulting simplified network , the nodes BID , Caspase , and IFNG would also have in-degree and out-degree of one ( or less ) and thus can be safely removed as well . This reduction procedure results in a simple sub-network represented in Figure 2B with the Boolean rules given in Table 1 . Our attractor analysis revealed that this sub-network has two fixed points , namely 000001 and 110000 ( the digits from left to right represent the state of the nodes in the order as listed from top to bottom in Table 1 ) . The first fixed point represents the normal state , that is , the apoptosis of CTL cells . Note that the OFF state of other nodes in this fixed point was expected because of the presence of “AND NOT Apoptosis” in all the Boolean rules . The second fixed point is the T-LGL ( disease ) one as Apoptosis is stabilized in the OFF state . We note that the sub-network depicted in Figure 2B contains a backbone of activations from Fas to Apoptosis and two nodes ( S1P and FLIP ) which both have a mutual inhibitory relationship with the backbone . If activation reaches Apoptosis , the system converges to the normal fixed point . In the T-LGL fixed point , on the other hand , the backbone is inactive while S1P and FLIP are active . We found by simulations that for the simplified network of Figure 2B , 56% of the states of the state transition graph ( represented in Figure 4 ) are in the exclusive basin of attraction of the normal fixed point while 5% of the states form the exclusive basin of attraction of the T-LGL fixed point . Again , the half of state space that has the ON state of Apoptosis belongs to the exclusive basin of attraction of the normal fixed point . Notably , there is a significant overlap between the basins of attraction of the two fixed points , which is illustrated by a gray color in Figure 4 . The probabilities of reaching each of the two fixed points starting from these gray-colored states , found by analysis of the corresponding Markov chain ( see Materials and Methods ) , are given in Figure 5 . As this figure represents , for the majority of cases the probability of reaching the normal fixed point is higher than that of the T-LGL fixed point . The three states whose probabilities to reach the T-LGL fixed point are greater than or equal to 0 . 7 are one step away either from the T-LGL fixed point or from the states in its exclusive basin of attraction . In two of them , the backbone of the network in Figure 2B is inactive , and in the third one the backbone is partially inactive and most likely will remain inactive due to the ON state of S1P ( one of the two nodes having mutual inhibition with the backbone ) . Based on the sub-network analysis and considering the states of the nodes that stabilized at the beginning based on the logical steady state analysis , we conclude that the whole T-LGL network has three attractors , namely the normal fixed point wherein Apoptosis is ON and all other nodes are OFF , representing the normal physiological state , and two T-LGL attractors in which all nodes except two , i . e . TCR and CTLA4 , are in a steady state , representing the disease state . These T-LGL attractors are given in the second column of Table 2 , which presents the predicted T-LGL states of 54 components of the network ( all but the six source nodes whose state is indicated at the beginning of the Results section ) . We note that the two T-LGL attractors essentially represent the same disease state since they only differ in the state of the virtual node P2 . Moreover , this disease state can be considered as a fixed point since only two nodes oscillate in the T-LGL attractors . For this reason we will refer to this state as the T-LGL fixed point . It is expected that the basins of attraction of the fixed points have similar features as those of the simplified networks . Experimental evidence exists for the deregulated states of 36 ( 67% ) components out of the 54 predicted T-LGL states as summarized in the third column of Table 2 . For example , the stable ON state of MEK , ERK , JAK , and STAT3 indicates that the MAPK and JAK-STAT pathways are activated . The OFF state of BID is corroborated by recent evidence that it is down-regulated both in natural killer ( NK ) and in T cell LGL leukemia [45] . In addition , the node RAS was found to be constitutively active in NK-LGL leukemia [41] , which indirectly supports our result on the predicted ON state of this node . For three other components , namely , GPCR , DISC , and IFNG , which were classified as being deregulated without clear evidence of either up-regulation or down-regulation in [18] , we found that they eventually stabilize at ON , OFF , and OFF , respectively . The OFF state of IFNG and DISC is indeed supported by experimental evidence [44] , [46] . In the second column of Table 2 , we indicated with an asterisk the stabilized state of 17 components that were experimentally undocumented before and thus are predictions of our steady state analysis ( P2 was not included as it is a virtual node ) . We note that ten of these cases were also predicted in [18] by simulations . The predicted T-LGL states of these 17 components can guide targeted experimental follow-up studies . As an example of this approach , we tested the predicted over-activity of the node SMAD ( see Materials and Methods ) . As described in [18] the SMAD node represents a merger of SMAD family members Smad 2 , 3 , and 4 . Smad 2 and 3 are receptor-regulated signaling proteins which are phosphorylated and activated by type I receptor kinases while Smad4 is an unregulated co-mediator [47] . Phosphorylated Smad2 and/or Smad3 form heterotrimeric complexes with Smad4 and these complexes translocate to the nucleus and regulate gene expression . Thus an ON state of SMAD in the model is a representation of the predominance of phosphorylated Smad2 and/or phosphorylated Smad3 in T-LGL cells . In relative terms as compared to normal ( resting or activated ) T cells , the predicted ON state implies a higher level of phosphorylated Smad2/3 in T-LGL cells as compared to normal T cells . Indeed , as shown in Figure 6 , T cells of T-LGL patients tend to have high levels of phosphorylated Smad2/3 , while normal activated T cells have essentially no phosphorylated Smad2/3 . Thus our experiments validate the theoretical prediction . A question of immense biological importance is which manipulations of the T-LGL network can result in consistent activation-induced cell death and the elimination of the dysregulated ( diseased ) behavior . We can rephrase and specify this question as which node perturbations ( knockouts or constitutive activations ) lead to a system that has only the normal fixed point . These perturbations can serve as candidates for potential therapeutic interventions . To this end , we performed node perturbation analysis using both structural and dynamic methods .
In this paper we presented a comprehensive analysis of the T-LGL survival signaling network to unravel the unknown facets of this disease . By using a reduction technique , we first identified the fixed points of the system , namely the normal and T-LGL fixed points , which represent the healthy and disease states , respectively . This analysis identified the T-LGL states of 54 components of the network , out of which 36 ( 67% ) are corroborated by previous experimental evidence and the rest are novel predictions . These new predictions include RAS , PLCG1 , IAP , TNF , NFAT , GRB2 , FYN , SMAD , P27 , and Cytoskeleton signaling , which are predicted to stabilize at ON in T-LGL leukemia and GAP , SOCS , TRADD , ZAP70 , and CREB which are predicted to stabilize at OFF . In addition , we found that the node P2 can stabilize in either the ON or OFF state , whereas two nodes , TCR and CTLA4 , oscillate . We have experimentally validated the prediction that the node SMAD is over-active in leukemic T-LGL by demonstrating the predominant phosphorylation of the SMAD family members Smad2 and Smad3 . The predicted T-LGL states of other nodes provide valuable guidance for targeted experimental follow-up studies of T-LGL leukemia . Among the predicted states , the ON state of Cytoskeleton signaling may not be biologically relevant as this node represents the ability of T cells to attach and move which is expected to be reduced in leukemic T-LGL compared to normal T cells . This discrepancy may be due to the fact that the network contains insufficient detail regarding the regulation of the cytoskeleton , as there is only one node , FYN , upstream of Cytoskeleton signaling in the network . While the network is able to successfully capture survival signaling without necessarily capturing the cytoskeleton signaling , this discrepancy suggests that follow-up experimental studies should be conducted to determine the relationship between cytoskeleton signaling and survival signaling in the T-LGL network . We note that in the case of perturbation of TBET , PI3K , NFκB , JAK , or SOCS , the node Cytoskeleton signaling exhibits oscillatory behavior induced by oscillations in TCR . At present it is not known whether this predicted behavior is relevant . Using the general asynchronous ( GA ) Boolean dynamic approach , we analyzed the basins of attraction of the fixed points . We found that the basin of attraction of the normal fixed point is larger than that of the T-LGL fixed point . The trajectories starting from each initial state toward the T-LGL fixed point ( Figure 4 ) may be indicative of the accumulating deregulations that lead to the disease-associated stable survival state . Although the fixed points , being time independent , are the same for all update methods or implementations of time , the update method may affect the structure of the state transition graph of the system and the basins of attraction of the fixed points . We note that the GA method assumes that each node has an equal chance of being updated . If quantitative or kinetic information becomes available in this system , unequal probabilities may be implemented by grouping the nodes into several “priority classes” and assigning a weight to each class where higher weights indicate more probable transitions [51] . Incorporating such information into the state space may prune the allowed trajectories and give further insights into the accumulation of deregulations . We took one step further by performing a perturbation analysis using dynamical and structural methods to identify the interventions leading to the disappearance of the disease fixed point . We note that our study has a dramatically larger scope than the previous key mediator analysis of Zhang et al [18] . For the dynamical analysis , we employed the GA approach instead of the random order asynchronous method and considered all possible initial conditions as opposed to performing numerical simulations using a specific initial condition . Zhang et al only focused on the node Apoptosis , and identified as “key mediators” the nodes whose altered state increases the frequency of ON state of Apoptosis . An increase in Apoptosis' ON state does not necessarily imply that apoptosis is the only possible final outcome of the system . In this work , after finding the fixed points , which completely describe the state of the whole system , we performed dynamic perturbation analysis by fixing the state of each node to its opposite state in the T-LGL fixed point and determining which fixed points were obtained and what their basins of attraction were . This way we were able to identify and distinguish the key mediators whose altered state completely eliminates the leukemic outcome , and those whose altered state reduces the basin of attraction of the leukemic outcome . Moreover , numerical simulations , as done in [18] , may not be able to thoroughly sample different timing . In this study , using a reduction technique , we found the cases when timing does not matter with certainty ( where there is only one fixed point ) , and also the cases in which timing and initial conditions may matter ( where there are two reachable fixed points ) . For the perturbation analysis using the structural method , we used the simple path ( SP ) measure to identify important mediators of the disease outcome and observed consistent results with the dynamic analysis . Our dynamical and structural analysis led to the identification of 19 therapeutic targets ( the first 19 nodes in the first column of Table 2 ) , 53% of which are supported by direct experimental evidence and 15% of which are supported by indirect evidence . Multi-stability ( having multiple steady states ) is an intrinsic dynamic property of many disease networks [52] , [53] , which is related to the presence of feedback loops in the network . In a graph-theoretical sense , a feedback loop is a directed cycle whose sign depends upon the parity of the number of negative interactions in the cycle . A positive/negative feedback loop has an even/odd number of negative interactions . It was conjectured that the presence of positive feedback loops in the network is necessary for multi-stability whereas the existence of negative feedback loops is required for having sustained oscillations [54] . From a biological point of view , the former dynamical property is associated with multiple cell types after differentiation while the latter is related to stable periodic behaviors such as circadian rhythms [55] . We note that the T-LGL signaling network consists of both positive and negative feedbacks and thus has a potential for both multi-stability and oscillations . Indeed , the negative feedback in the top sub-graph of Figure 2A causes the complex attractor shown in Figure 3 . In contrast , the negative feedback on the node P2 of the bottom sub-graph is counteracted by the positive self-loop on the same node , thus no complex attractor is possible for the bottom sub-graph of Figure 2A . The two mutual inhibition-type positive feedback loops present in the bottom sub-graph and the self-loop on P2 generate the three fixed points , while the positive self-loop on Apoptosis maintains the normal fixed point once Apoptosis is turned ON . Negative feedback loops can be a source of oscillations [56] , homeostasis [56] , or excitation-adaptation behavior [57] . Especially , when the activation is slower than the inhibitory interaction in the negative feedback , it can lead to sustained oscillations [56] . In the T-LGL network , the negative feedback loop between the T cell receptor TCR and CTLA4 modulates stimulus-induced activation of the receptor in such a way that CTLA4 is indirectly activated after prolonged TCR activation , whereas the inhibition of TCR by CTLA4 is a direct interaction [58] . That is , activation is slower than inhibition in the negative feedback and thus an oscillatory behavior reminiscent of that obtained by our asynchronous Boolean model would also be observed in continuous modeling frameworks as well . Although no time-measurements of the T cell receptor activity in T-LGL exist , it has been reported that there is variability for TCR activation in different patients ( [43] and unpublished observation by T . P . Loughran ) , supporting the absence of a steady state behavior . Our study revealed that both structural and dynamic analysis methods can be employed to identify therapeutic targets of a disease , however , they differ in implementation efficiency as well as the scope and applicability of the results . The structural analysis does not require mapping of the state space and thus is less computationally intensive and is more feasible for large network analysis , but it may not capture all the initial states and thus may miss or inaccurately identify some important features . The dynamic analysis method , while computationally intensive , yields a comprehensive picture of the state transition graph , including all possible fixed points of the system , their corresponding basins of attraction , as well as the relative frequency of trajectories leading to each fixed point . We demonstrated that the limitations related to the vast state space of large networks can be overcome by judicious use of the network reduction technique that we developed in our previous study [22] . We conclude that the structural method incorporating the cascading effects of node disruptions is best employed for quick exploratory analysis , and dynamic analysis should be performed to get a thorough and detailed insight into the behavior of a system . Overall , the combined analysis presented in this study opens a promising avenue to predict dysregulated components and identify potential therapeutic targets , and it is versatile enough to be successfully applied to a large variety of signal transduction and regulatory networks related to diseases . | T-LGL leukemia is a blood cancer characterized by an abnormal increase in the abundance of a type of white blood cell called T cell . Since there is no known curative therapy for this disease , identification of potential therapeutic targets is of utmost importance . Experimental identification of manipulations capable of reversing the disease condition is usually a long , arduous process . Mathematical modeling can aid this process by identifying potential therapeutic interventions . In this work , we carry out a systematic analysis of a network model of T cell survival in T-LGL leukemia to get a deeper insight into the unknown facets of the disease . We identify the T-LGL status of 54 components of the system , out of which 36 ( 67% ) are corroborated by previous experimental evidence and the rest are novel predictions , one of which we validate by follow-up experiments . By deciphering the structure and dynamics of the underlying network , we identify component perturbations that lead to programmed cell death , thereby suggesting several novel candidate therapeutic targets for future experiments . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"biology",
"computational",
"biology",
"signaling",
"networks"
] | 2011 | Dynamical and Structural Analysis of a T Cell Survival Network Identifies Novel Candidate Therapeutic Targets for Large Granular Lymphocyte Leukemia |
Typhoid fever is endemic in Fiji , with high reported annual incidence . We sought to identify the sources and modes of transmission of typhoid fever in Fiji with the aim to inform disease control . We identified and surveyed patients with blood culture-confirmed typhoid fever from January 2014 through January 2017 . For each typhoid fever case we matched two controls by age interval , gender , ethnicity , and residential area . Univariable and multivariable analysis were used to evaluate associations between exposures and risk for typhoid fever . We enrolled 175 patients with typhoid fever and 349 controls . Of the cases , the median ( range ) age was 29 ( 2–67 ) years , 86 ( 49% ) were male , and 84 ( 48% ) lived in a rural area . On multivariable analysis , interrupted water availability ( odds ratio [OR] = 2 . 17; 95% confidence interval [CI] 1 . 18–4 . 00 ) , drinking surface water in the last 2 weeks ( OR = 3 . 61; 95% CI 1 . 44–9 . 06 ) , eating unwashed produce ( OR = 2 . 69; 95% CI 1 . 48–4 . 91 ) , and having an unimproved or damaged sanitation facility ( OR = 4 . 30; 95% CI 1 . 14–16 . 21 ) were significantly associated with typhoid fever . Frequent handwashing after defecating ( OR = 0 . 57; 95% CI 0 . 35–0 . 93 ) and using soap for handwashing ( OR = 0 . 61; 95% CI 0 . 37–0 . 95 ) were independently associated with a lower odds of typhoid fever . Poor sanitation facilities appear to be a major source of Salmonella Typhi in Fiji , with transmission by drinking contaminated surface water and consuming unwashed produce . Improved sanitation facilities and protection of surface water sources and produce from contamination by human feces are likely to contribute to typhoid control in Fiji .
Typhoid fever remains a substantial cause of morbidity and mortality in many low- and middle-income countries , with an estimated 17 . 8 million new episodes annually [1] . By 2015 , Oceania had fallen behind both Asia and sub-Saharan Africa to become the region with the lowest coverage of improved drinking water and improved sanitation [2 , 3] . Pacific island nations including Fiji [4] , Nauru [5] , and Papua New Guinea [6] report high case counts of typhoid fever and frequent large outbreaks of the disease . Despite this apparent high incidence , studies to investigate sources and modes of transmission of typhoid fever in the Pacific , where unique socio-demographic , behavioral , and environmental conditions may exist , have been rare [4 , 7] . A detailed understanding of local risk factors for typhoid fever is necessary to inform non-vaccine control measures . Furthermore , a robust understanding of the epidemiology of typhoid fever in Fiji is needed to inform decisions about the introduction of recently recommended [8] and prequalified typhoid conjugate vaccine . Typhoid fever is endemic in Fiji with disease occurring among both rural and urban residents [4 , 7] . Blood culture-confirmed typhoid fever infections have increased since the 1990s , rising rapidly since 2005 [9] . By 2010 , the incidence of typhoid fever in Fiji , identified by passive surveillance , was 52 per 100 000 population [10] . However , given the limited sensitivity of blood cultures , patterns of health seeking behavior among Fijians , and low access to blood cultures services , the actual incidence of typhoid fever in Fiji is likely to exceed this rate [10] . Previous typhoid fever case-control studies in endemic countries have demonstrated considerable variation in major sources and modes of transmission by location [11 , 12] . This variation underscores the importance of local research , especially in distinct environments such as those that exist in the smaller island states of Oceania . To identify risk factors for typhoid fever relevant to the distinctive island ecology of Oceania , we conducted a case-control study to inform typhoid control efforts in Fiji’s Central Division .
The Republic of Fiji , located in the southern Pacific Ocean , consists of 332 islands . Our study was undertaken in the Central Division with a population >370 , 000 people , representing 43% of the national population , and including the capital city Suva . At the time of the study , Central Division residents were comprised of 56% iTaukei ( indigenous Fijians ) , 38% Fijians of Indian descent , and 6% who identified as being of another ethnicity [13] . The majority of Central Division’s population resided in Suva , and the remainder lived in small rural villages and settlements near major waterways [7] . All public health services in Fiji are provided at the divisional and sub-divisional level . The Colonial War Memorial Hospital ( CWMH ) in Suva is the largest public referral hospital in the country providing clinical and laboratory services to all of the Central Division . We conducted a neighborhood , ethnicity , and age interval matched case-control study from 27 January 2014 through 31 January 2017 in Central Division , Fiji . Cases were defined as patients residing in Central Division who sought care at any Central Division public health facility , and had Salmonella enterica serovar Typhi ( Salmonella Typhi ) isolated from blood culture . Those aged >18 years were eligible for enrollment until 1 May 2014 when regulatory approval was received to extend enrollment to all ages . For persons aged <12 years , we interviewed the parent or guardian of the patient . If more than one case was detected in a household , we only enrolled the first case . Two neighborhood controls were selected for each case , one from a near neighborhood and the other from a more distant neighborhood from where the case arose . We sought near-neighborhood controls 100 paces in a random direction from the case household using a pen spin method . We sought distant-neighborhood controls in the same random direction as a near-neighborhood control , from the next closest river basin in rural areas and from the next closest sub-division in urban and peri-urban areas . We sought potential age-matched controls in the intervals <4 years , 5–14 years , 15–24 years , 25–34 years , 35–44 years , 45–54 years , 55–64 years , 65–74 years , and >75 years of age . We excluded potential controls that had experienced fever within the past one month . We administered a standardized questionnaire to all participants . The questionnaire sought information on basic demographics and focused on modifiable typhoid fever risk factors and exposures occurring during the two-week period prior to the onset of symptoms for cases and prior to the date of recruitment for controls . Domains of questions included those related to water and food consumption , including kava , a local drink of Fiji made of water infused with Piper methysticum , and behavioral practices such as attendance at community gatherings and handwashing . We investigated longer-term environmental factors including the occurrence of floods , droughts , and tropical storms two months prior to the onset of symptoms or interview . We recorded observations on the type and condition of water source and sanitation facility within the bounds of the household and also on distal environmental conditions . Blood cultures were collected from febrile patients at the clinicians’ discretion [14] . Blood was collected in BacT/ALERT standard aerobic and anaerobic blood culture bottles for adults ( 5–10mL ) and pediatric FAN bottles for children ( 2–5 mL ) ( bioMérieux , Marcy L’Etoile , France ) . Bottles were incubated for 5–7 days at 35°C in the BacT/Alert system . Broth from positive blood culture bottles was subcultured on blood , chocolate , and MacConkey agar . Non-lactose fermenting colonies were identified biochemically as probable Salmonella Typhi using Microbact identification system , Triple Sugar Iron ( TSI ) , and Lysine Indole Motility ( LIM ) media . Salmonella Typhi were confirmed by slide agglutination , using antibody reagents specific for serogroup O9 and Vi ( Difco Salmonella Antiserum , Becton Dickinson , Franklin Lakes , NJ USA ) . A three-level socioeconomic status ( SES ) index was created by principal component analysis of education , employment , household conditions , and asset variables [15] . Categorical variables were created for water source and sanitation facilities , with improved municipal water source and undamaged , improved municipal sewerage used as the referent category for each variable . Handwashing behavior questions at pre-specified times ( before eating , before cooking , and after defecation ) were recoded as an ordinal variable ranging from 0 ( never ) to 2 ( always ) . A knowledge-based dimension reduction strategy , which considered the causal pathway through directed acyclic graphs , was used to guide variable selection for model building and interpretation of multivariable analysis [16] . Odds ratios ( OR ) with exact 95% confidence interval ( CI ) were measured in a univariable analysis using conditional logistic regression for the selected variables . A multivariable conditional logistic regression model with all variables with a p-value <0 . 1 in the univariable analysis was applied and variables with p-value >0 . 05 were removed from the model in a step-wise manner . Population attributable fractions ( PAF ) for categorical , potentially modifiable risk factors were estimated from the prevalence of exposure in case patients and adjusted OR from the multivariable conditional logistic regression model [17 , 18] . The proportion of participants with missing data for variables selected for model building ranged from 0–4% . Missing values were imputed using all available data for each participant . Specifically , five complete datasets were created using the multivariate imputation by chained equations ( MICE ) method of imputation in STATA [19] . Results obtained from imputed datasets were not significantly different from non-imputed data , hence we chose to present our unimputed results . Analysis was done comparing cases to near-neighborhood and distant-neighborhood controls separately and together . Matched ORs with control groups separately were not significantly different to ORs with controls together . Therefore , we chose to present results for the combined control groups . Details on sample size estimation ( S1 File ) , questionnaire used for response collection ( S2 File ) , directed acyclic graph ( S1 Fig ) , and univariable results with analysis of control groups separately ( S1 and S2 Tables ) are provided in the supporting information . Analyses were conducted with Stata/SE 14 . 0 for Windows ( Stata , TX , USA ) . Ethics approvals were obtained from the Fiji National Health Research Committee , the Human Ethics Committee of the University of Otago , and the Human Research Ethics Committee of Edith Cowan University . Verbal and written details of the study were provided in Fijian , Hindi , or English as necessary and written informed consent was obtained from all participants or their guardians .
Of 39 , 775 blood cultures performed , 279 ( 0 . 7% ) patients with blood culture-confirmed typhoid fever were identified . No Salmonella Paratyphi A , B , or C were isolated during the study period . Of those blood culture-confirmed typhoid fever cases , 175 ( 62 . 7% ) were enrolled in the case-control study . Reasons for non-enrolment are shown in Fig 1 . A total of 349 controls were enrolled in the study . Baseline characteristics of enrolled cases and controls are described in Table 1 . Of enrolled cases 84 ( 48 . 0% ) were from a rural area , while 59 ( 33 . 7% ) and 32 ( 18 . 3% ) were from urban and peri-urban areas respectively . In the univariable analysis ( Table 2 ) , the odds of a case being from a low SES index household was almost 3 times that of controls ( matched OR 2 . 98 and 95% CI 1 . 64–5 . 44 ) . Cases more frequently accessed their main water source from outside the house ( OR = 2 . 96 , 95% CI 1 . 20–7 . 29 ) , had interrupted water availability ( OR = 2 . 40 , 95% CI 1 . 39–4 . 12 ) , and drank water from an untreated source in the last two weeks ( OR = 1 . 80 , 95% CI 1 . 07–3 . 03 ) . Among participants who reported drinking water from a source other than their main household water source , cases were more likely to have drunk water from a surface water source than controls ( OR = 3 . 04 , 95% CI 1 . 33–6 . 92 ) . No difference in kava consumption was observed between cases and controls . Compared to controls , cases were more likely to eat unwashed produce ( OR = 3 . 48 , 95% CI 2 . 06–5 . 89 ) . Cases were also more likely to have eaten outside of the house in the past two weeks ( OR = 1 . 56 , 95% CI 1 . 04–2 . 34 ) , and have attended a mass gathering in the past two weeks ( OR = 1 . 59 , 95% CI 1 . 05–2 . 40 ) . When categorized by latrine type , cases were more likely to have either an unimproved pit latrine ( OR = 26 . 62 , 95% CI 5 . 20–136 . 37 ) , or no toilet or latrine in the household ( OR = 14 . 17 , 95% CI 1 . 97–102 . 14 ) , or a damaged , improved municipal sewerage ( OR = 5 . 28 , 95% CI 1 . 23–22 . 73 ) , or an improved pit latrine ( OR = 4 . 40 , 95% CI 1 . 46–13 . 26 ) compared to controls . Furthermore , cases were more likely than controls to report having a latrine built by someone within their own household ( OR = 1 . 53 , 95% CI 1 . 12–2 . 30 ) . With respect to hygiene behaviors , a higher handwashing score after defecating ( OR = 0 . 41 , 95% CI 0 . 27–0 . 62 ) and using soap for handwashing ( OR = 0 . 38 , 95% CI 0 . 24–0 . 58 ) were associated with lower odds of typhoid fever . In terms of distal environmental factors , compared to controls , cases were more likely to have experienced flooding of their nearest river or stream in the past two months ( OR = 2 . 42 , 95% CI 1 . 13–5 . 12 ) . Cases were also more likely to have a dam located upstream from their closest river basin ( OR = 2 . 18 , 95% CI 1 . 22–3 . 91 ) than controls . On multivariable analysis ten exposures remained independently associated with typhoid fever ( Table 3 ) including drinking water from an alternative surface water source in the last two weeks ( OR = 3 . 61 , 95% CI 1 . 44–9 . 06 ) , not having constant water availability ( OR = 2 . 17 , 95% CI 1 . 18–4 . 00 ) , and eating unwashed produce ( OR = 2 . 69 , 95% CI 1 . 48–4 . 91 ) . The population attributable fraction ( PAF ) for these exposures was 7% , 15% , and 19% , respectively . Having any unimproved sewerage system or a damaged improved sewerage system was associated with increased odds of typhoid fever ( OR = 4 . 30 , 95% CI 1 . 14–16 . 21 ) . The summary PAF for these poor sanitation facilities was 72% . Frequent handwashing after defecating was associated with lower odds of Salmonella Typhi infection ( OR = 0 . 57; 95% CI 0 . 35–0 . 93 ) , as was using soap for handwashing ( OR = 0 . 61 , 95% CI 0 . 37–0 . 95 ) .
To our knowledge , this is the first case-control study to investigate sources and modes of transmission for typhoid fever in Fiji . We demonstrate that unimproved sanitation facilities are a likely major source of Salmonella Typhi , with transmission occurring through drinking contaminated surface water and eating unwashed produce . We also show typhoid to be common among both rural and urban populations in Fiji ( Table 1 ) . Improving sanitation facilities and increasing access to safe water and clean produce for rural populations presents a major challenge in Fiji and more broadly for other island nations of Oceania . The median age of cases in our study was 29 years , which is higher than the median age reported in several other typhoid endemic areas [11 , 12] . While this may suggest a distinct typhoid epidemiology in Fiji , under-ascertainment by blood culture of typhoid fever among younger age groups is also a possibility as younger patients in Fiji have been reported to receive early empiric treatment for fever without blood culture [4] . Of cases in our study , 95% were among iTaukei . This finding is in contrast to a sero-epidemiologic survey conducted in Fiji in 2013 , which found both iTaukei and Fijians of Indian ancestry to have similar sero-prevalence of antibodies against the Vi antigen of Salmonella Typhi [20] . Besides Vi serology being an inaccurate measure of Salmonella Typhi infection , a possible explanation for this difference is under-ascertainment by blood culture of typhoid fever among Fijians of Indian ancestry , who have been shown to preferentially present to private general practitioners and are again likely to receive early and empiric treatment for fever without blood culture [21] . Our results suggest that unimproved or damaged sanitation is a major source of Salmonella Typhi in Fiji . We found that people without access to improved sanitation facilities or with damaged improved sewerage systems were at particular risk . Those with typhoid fever were more likely than controls to have someone within their household build their toilet ( Table 2 ) . Others have shown latrines built by persons without expertise to be poorly constructed , built into permeable soil , and subject to flooding [7 , 22] . Notably , households with improved pit latrines had greater odds of typhoid fever in our study , compared to undamaged , improved municipal sanitation ( Table 3 ) . In Fiji , a common ‘improvement’ is use of buried steel drums as the receptacle for sewerage [23] . Such receptacles are subject to flooding , corrosion , and leakage [23] leading to contamination of surface water and crops by human feces [24] . Interestingly , eating unwashed produce was an independent risk factor for typhoid in our study . Related research has shown that gardens of patients with typhoid fever were more often positioned closer to the household toilet or septic tank than in control households and the majority of cases propagated vegetables directly on or below the toilet drainage area [25] . In contrast to water-related factors , poor sanitation has seldom been reported as a risk factor for typhoid fever [11 , 12 , 26] , highlighting the value of our local research . Contaminated drinking water is commonly identified as a risk factor for typhoid fever in case-control studies [26] . We showed that cases were more likely to report having poor water availability than controls ( Table 3 ) . Intermittent access to water is a common problem for households in many low-resource countries [27] and can lead to increased risk of typhoid fever by a number of mechanisms . First , in Fiji , as in many other locations , poor water availability from a primary source results in households shifting periodically to alternative water sources , including unsafe surface water [28 , 29] . Second , during periods of reduced water supply , households may rely on stored drinking water that is not disinfected . Although we did not demonstrate consumption of stored water as a risk factor for typhoid fever in Fiji , related research showed a significantly higher concentration of E . coli in stored drinking water in typhoid case households compared to control households , but not in source water [25] . Third , pressure drops associated with regular interruption in reticulated water supplies can result in negative pressure situations which , when combined with leaks in the distribution system , can result in inflow of environmental material [30] . While our study was not designed to confirm this mechanism , we hypothesize that this may occur in Fiji and warrants further investigation . Finally , poor water availability can affect the quality of sanitation facilities that rely on water as well as affect personal hygiene [31 , 32] . However , our multivariable analysis showed poor water availability as associated with increased odds of having typhoid fever even after adjustment for unimproved sanitation and handwashing frequency . Factors related to hygiene were also independently associated with typhoid in our study setting ( Table 3 ) . Frequent handwashing after defecating and using soap for handwashing were associated with lower odds of typhoid fever . Both of these handwashing behaviors have been associated with reduced risk of typhoid in other studies [12 , 28] . Distal conditions within the water catchment have yet to be thoroughly evaluated in case-control studies of typhoid fever . However , geospatial studies are beginning to examine such relationships on broad spatial scales [7 , 33 , 34] . Although no distal environmental conditions were statistically associated with typhoid fever in our multivariable model , experiencing flooding of the nearest river or stream in the past two months and reporting dams upstream in the river basin were significant in univariable analysis . Descriptive accounts of dam construction and bursts have been linked to an increase in typhoid cases in Nigeria [35] . However , the basis for these risks is not well understood and could be the subject of future research . At a sub-catchment scale , typhoid infection and disease in the Fijian setting has been linked with forest fragmentation , increased erosion , rainfall , and flood risk [7 , 34] . It is likely that such environmental factors result in overflow and damage to already poor sanitation facilities leading to contamination of produce and drinking water sources . Our study had a number of limitations . First , recall bias may influence the reliability of potential exposures over the long incubation period of typhoid fever [36] and social desirability bias is a common concern for sanitation and hygiene questions [37] . We sought to control the former by making observations of sanitation facilities . However , observations and more objective measures of handwashing practices were not possible . Therefore , our findings on handwashing should be interpreted with caution . Second , since the interviewers in this study knew the case status of the participant , it is possible that they may have acquired differential exposure information from cases . However , our questionnaire was standardized and interviewers were trained to ensure cases and controls were questioned in the same way . Third , the relative homogeneity of sampled environments and the collinear relationship of many factors may have masked the detection of potential risk factors . However , by recruiting a second control in a distant neighborhood and by obtaining a large sample size we expected to address power concerns of ‘over-matching’ and have adequate statistical power to identify minor associations . Finally , since case detection was by passive surveillance at public healthcare facilities , we are unlikely to have identified all typhoid fever illnesses and we may have missed cases that were more likely to be treated empirically or preferentially access private healthcare services . This could have also resulted in us missing exposures associated with increased risk of typhoid fever in these populations . Future research should include case finding at private health care facilities . In conclusion , our study demonstrates that unimproved and damaged sanitation facilities are an important source of Salmonella Typhi in Fiji . Transmission appears to be by drinking contaminated surface water and consumption of unwashed produce , and is common in both rural and urban populations . Poor hygiene practices also appear to increase odds for typhoid fever . Although not detected in this study , landscape factors contributing to flooding of poor sanitation facilities may also contribute to enhanced risk [7 , 34] . The situation in Fiji may reflect sources and modes of transmission predominant elsewhere in Oceania , where typhoid incidence is also high [1] and similar socio-demographic and environmental circumstances prevail . Meeting the 2030 Sustainable Development Goals goals [2] to improve sanitation facilities and protect surface water and produce from contamination by human feces are likely to contribute to typhoid control in Fiji . Central or household based water disinfection would also help to render fecally contaminated water safe for consumption . Such long-term socioeconomic , land and water management , and sanitation infrastructure developments together with uptake of typhoid conjugate vaccination in the interim are likely to result in effective typhoid control in Fiji and Oceania . | Modeling suggests that Oceania has surpassed Asia and sub-Saharan Africa as the region with the highest typhoid fever incidence . While Pacific Islands are often neglected due to small population sizes , there is an urgent need to understand the epidemiology of typhoid fever in the region . Fiji , an upper-middle income country in Oceania , has reported an increase in typhoid fever notifications over the last decade . However , the epidemiology of typhoid fever in Fiji is incompletely understood due to gaps in surveillance and lack of epidemiological research on local risk factors . We conducted a case-control study in the Central Division of Fiji to help inform prevention and control strategies . We found unimproved sanitation facilities to be major source of typhoid fever in Fiji , with transmission by drinking contaminated surface water and consumption of unwashed produce . We also found an association between poor water availability and poor hygiene with typhoid fever . Improvements in sanitation facilities to protect surface water and produce from contamination are likely to contribute to improved typhoid control in Fiji . Because of the distinct socio-demographic and environmental conditions found in Oceania , our findings may reflect sources and modes of transmission predominant elsewhere in the region . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"surface",
"water",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"geographical",
"locations",
"microbiology",
"salmonella",
"typhi",
"health",
"care",
"bacterial",
"diseases",
"signs",
"and",
"sy... | 2018 | Epidemiology and risk factors for typhoid fever in Central Division, Fiji, 2014–2017: A case-control study |
After germination , plants enter juvenile vegetative phase and then transition to an adult vegetative phase before producing reproductive structures . The character and timing of the juvenile-to-adult transition vary widely between species . In annual plants , this transition occurs soon after germination and usually involves relatively minor morphological changes , whereas in trees and other perennial woody plants it occurs after months or years and can involve major changes in shoot architecture . Whether this transition is controlled by the same mechanism in annual and perennial plants is unknown . In the annual forb Arabidopsis thaliana and in maize ( Zea mays ) , vegetative phase change is controlled by the sequential activity of microRNAs miR156 and miR172 . miR156 is highly abundant in seedlings and decreases during the juvenile-to-adult transition , while miR172 has an opposite expression pattern . We observed similar changes in the expression of these genes in woody species with highly differentiated , well-characterized juvenile and adult phases ( Acacia confusa , Acacia colei , Eucalyptus globulus , Hedera helix , Quercus acutissima ) , as well as in the tree Populus x canadensis , where vegetative phase change is marked by relatively minor changes in leaf morphology and internode length . Overexpression of miR156 in transgenic P . x canadensis reduced the expression of miR156-targeted SPL genes and miR172 , and it drastically prolonged the juvenile phase . Our results indicate that miR156 is an evolutionarily conserved regulator of vegetative phase change in both annual herbaceous plants and perennial trees .
Plants produce different types of leaves , buds , and internodes at different times in their development . Although many traits vary continuously , other traits are expressed in discontinuous pattern that allows shoot development to be divided into discrete juvenile , adult , and reproductive phases [1]–[5] . These transitions involve changes in many different traits that must be temporally and spatially coordinated if the plant is to survive and reproduce . This problem is particularly important in perennial species , which encounter numerous biotic and abiotic stresses during their long life cycles . Recent studies have begun to reveal the molecular mechanism of these phase transitions in the annual species Arabidopsis and maize , but the molecular mechanism of phase change in perennial woody species is still largely unknown . In the model annual forb , Arabidopsis thaliana , the major morphological difference between the juvenile and the adult phase of vegetative development is in leaf morphology . Adult leaves have serrations on their leaf margins and trichomes on the abaxial surface , which are lacking in juvenile leaves [6]–[8] . In maize , juvenile leaves lack trichomes but possess epicuticular wax , whereas adult leaves have the opposite traits [4] . These differences are mediated by two miRNAs , miR156 and miR172 , both of which target DNA-binding transcription factors . miR156 is highly abundant in seedlings , and decreases during subsequent development , while miR172 has an opposite expression pattern . Overexpression of miR156—which negatively regulates several SQUAMOSA PROMOTER BINDING PROTEIN-LIKE ( SPL ) genes—delays both the juvenile-to-adult and adult-to-reproductive phase transitions . Conversely , increasing the levels of different SPLs can in the most extreme case completely eliminate the juvenile phase [9]–[16] . miR172 targets several transcription factors related to the eponymous APETALA2 ( AP2 ) protein , including TARGET OF EAT1 ( TOE1 ) , TOE2 , TOE3 , SCHLAFMÜTZE ( SMZ ) , and SCHNARCHZAPFEN ( SNZ ) [17]–[20] in Arabidopsis , and Glossy15 in maize [15] . MIR172b is a direct target of SPL9 in Arabidopsis and its level gradually rises after germination in response to increasing SPL activity . Overexpression of the miR172-regulated genesTOE1 [13] and Glossy15 [15] delays the juvenile-to-adult vegetative transition . The hierarchical action of miR156 and miR172 and their SPL and AP2 targets in the control of vegetative phase change and flowering is conserved in the annual grasses rice and maize [15] , [16] , [21]–[25] . It is unknown , however , whether the juvenile-to-adult phase transition in woody perennial plants is controlled by the same factors as it is in annual species . First , the differences between juvenile and adult phases are often much more obvious in shrubs and trees , and are usually more stably expressed in woody plants than in herbaceous species . Second , juvenile and adult vegetative phases are quite brief in herbaceous plants such as A . thaliana and maize , but can last for many years in trees [26] . Here , we show that levels of miR156 and miR172 are closely correlated with the juvenile and adult phases of several woody species that have long been used in studies of vegetative phase change . We also demonstrate that miR156 expression varies with the age and morphology of the shoot in the poplar hybrid Populus x canadensis , and that miR156 overexpression dramatically delays phase change in this tree .
In A . thaliana , miR156 and miR172 control both the juvenile-to-adult vegetative transition and the adult-to-reproductive transition . miR156 expression is highest after germination and declines within two weeks , whereas miR172 shows the converse pattern [9] , [10] , [13] , [14] , [18] . To explore the possibility that these miRNAs also regulate phase change in woody plants , we examined their expression in several species with distinct , well-characterized juvenile and adult phases . Juvenile and adult phases of vegetative development were first described in Acacia species native to Australia [27] , [28] , where these phases are characterized by dramatic differences in leaf morphology . Early in shoot development , these species produces horizontally oriented , bipinnately compound leaves . The transition to the adult phase is marked by the production of phyllodes—vertically-oriented , simple leaves , in which adaxial cell types are present on both surface of the leaf blade [29] . This transition takes place at different nodes in different Acacia species , and is often accompanied by the production of transition leaves in which both leaf types are present in a single leaf ( Figure 1B ) . Juvenile and adult stages of vegetative development are also well differentiated in many species of Eucalyptus , including E . globulus , where juvenile leaves are horizontally oriented , ovate to acuminate in shape , lack a petiole , are covered with epicuticular wax , and produce palisade mesophyll solely on the adaxial surface of the leaf blade; in contrast , adult leaves are vertically oriented , lanceolate , petiolate , waxless , and have palisade mesophyll on both the surfaces of the leaf blade [30] ( Figure 1C and 1F ) . English ivy ( Hedera helix ) is a classic system for the analysis of vegetative phase change [31] . Juvenile and adult phases of shoot growth in this woody vine differ in leaf shape , phyllotaxis , the orientation of shoot growth , adventitious root production , growth rate , and anthocyanin production [32] ( Figure 1D ) . In the sawtooth oak , Quercus acutissima , juvenile leaves are ovate in shape and have a relatively short petiole , whereas adult leaves have an acute leaf tip and a longer petiole ( Figure 1E ) . Juvenile and adult phases of shoot development in this and other species of oak can also be readily differentiated by their pattern of leaf abscission: adult branches drop their leaves in the Fall , whereas juvenile branches retain their leaves until Spring ( Figure 1G ) . The levels of miR156 and miR172 were measured by northern blot analyses of the RNAs isolated from fully expanded leaves of A . confusa , A . colei , E . globulus , and Q . acutissima , and shoot apices of H . helix ( Figure 1H ) . Analyses were conducted using juvenile and adult leaves from the same plant in order to control for genetic variation between samples; at least two plants were examined for each species , and RNA levels were quantified by densitometry ( Table S1 ) . miR156 was expressed at a significantly higher level in juvenile leaves than in adult leaves , whereas miR172 had the opposite pattern ( Student's t test , p<0 . 0001 , n = 12 ) . This relationship was particularly striking in A . confusa and A . colei , where variation in the levels of miR156 and miR172 were correlated with node-to-node changes in leaf shape ( Figure 1A , 1B , 1H ) . The observation that this change in expression occurs at different nodes in these two species ( node 2 in A . confusa and node 6 or 7 in A . colei ) provides additional evidence that the expression of these miRNAs is associated with vegetative phase change rather than some other feature of shoot development , such as the distance of a leaf from the root system or the overall size of the shoot . The miR156 probe hybridized to 20 and 21 nt transcripts in A . confusa , A . colei and Q . acutissima , to a single 21 nt transcript in E . globulus , and a single 20 nt transcript in H . helix . Deep sequencing of small RNAs has revealed 20 nt miR156 transcripts in species ranging from moss to flowering plants ( www . mirbase . org ) . Many species also produce a closely-related miRNA that is 21 nt in length and differs from miR156 at three positions ( www . mirbase . org ) ; in Arabidopsis , this miRNA has been named miR157 [33] . Hybridization with a probe complementary to miR157 revealed a single 21 nt band in all five species we examined . This miRNA was expressed at the same , or higher , level than miR156 , and in the same developmental pattern ( Figure 1H ) . The observation that miR157 probe did not hybridize to a 20 nt fragment in A . confusa , A . colei , H . helix and Q . acutissima , and that the miR156 probe did not hybridize to a 21 nt fragment in H . helix , indicates that these probes do not cross-hybridize; thus , the 21 nt band observed on miR156 blots is unlikely to represent miR157 . miR156 transcripts with one additional 3′ or 5′ nucleotide ( i . e . , 21 nt miR156 transcripts ) have been observed by deep sequencing in several plants , including Arabidopsis , rice , and Populus [34]–[36] . It remains to be determined if these size variants are the sole product of specific miR156 loci , or are produced along with 20 nt forms by the imprecise processing of miR156 precursors . To determine if the variation in miR156 expression is functionally significant , we identified homologs of AtSPL3 and AtSPL9 in the recently completed genome sequence of Eucalyptus grandis ( DOE Joint Genome Institute and the Eucalyptus Genome Network; http://www . phytozome . net/eucalyptus . php ) , and used this sequence information to amplify the related transcripts from adult leaves of E . globulus . Quantitative RT-PCR ( qRT-PCR ) of juvenile and adult leaves from three different E . globulus trees demonstrated that transcripts of EglSPL3 and EglSPL9 were present at approximately 2-fold higher levels in adult leaves than juvenile leaves ( Figure 1I ) , consistent with the relative abundance of miR156 in these leaves ( Figure 1H; Table S1 ) . The expression pattern of these direct targets of miR156 supports the conclusion that miR156 plays an important role in vegetative phase change in E . globulus . As a further test of the hypothesis that that miR156 promotes juvenile development in trees , we took advantage of P . x canadensis cv . Guangzhao Yang , a hybrid of P . deltoides and P . nigra that is readily transformable . All of our studies were performed on clonal shoots regenerated from tissue culture . Although we were unable to examine the morphology of plants grown from seeds , regeneration typically induces rejuvenation in woody plants [37] , so it is reasonable to assume that the changes we observed in these regenerated plants mimic the changes that occur in seed-derived plants . This conclusion is supported by the observation that the leaf morphology of one-month old regenerated shoots of the clone used in this study closely resembled the juvenile leaves of P . trichocarpa , as described by [38] . There was a significant difference in the morphology of the leaves of 1-month- and 1-year-old regenerated shoots . 1-month-old plants had small , oval leaves , while the leaves of 1-year-old trees were larger and deltoid in shape ( Figure 2A; Table 1 ) . Leaf shape did not change further in older trees . In addition , 1-month-old plants had round petioles with only one vascular bundle , whereas the petioles of 1-year-old trees were flattened in an adaxial-abaxial plane and had three major vascular bundles ( Figure 2B ) . The internodes of 1-month-old plants were also significantly shorter than those of 6-month or 1-year-old trees ( Table 1 ) . The expression of miR156 was initially examined in fully expanded leaves from 1-month- , 1-year- , 4-year- , and 10-year-old trees . miR156 was highly expressed in leaves from 1-month-old plants , and was expressed at much lower levels in older trees ( Figure 2C ) . This expression difference is likely to be functionally significant because two miR156 targets , PcSPL3 and PcSPL9 , were expressed in the opposite pattern ( Figure 2D ) . The expression of miR172 was similar in 1-month- , 1-year- , and 4-year-old trees , but was elevated in 10-year-old trees ( Figure 2C ) . To characterize the expression pattern of miR156 and miR172 in more detail , we examined the levels of these miRNAs in 2 cm leaf primordia from the shoot apex of trees of different ages ( 0 . 5 m , 2 m , and 4 m tall ) , and in fully expanded leaves and leaf primordia of branches located at different positions on the primary shoot ( Figure 3A ) . miR156 was most highly expressed in leaf primordia from the primary shoot of 0 . 5 m and 2 m trees , and was expressed at a lower level in 4-m-tall shoots , while miR172 showed the opposite trend ( Figure 3B ) . Leaves on branches produced at the base of main stem ( 0 . 5 m , branch 1 ) recapitulate the change in leaf shape that occurs during the growth of the main stem . The first leaves on these branches resemble juvenile leaves , and leaf size and shape change gradually until the 10th node , by which time leaves have acquired the size and shape of adult leaves ( Figure 3C ) . Analyses of gene expression in these branch leaves revealed that miR156 was expressed at high levels in basal , juvenile-like leaves , and at lower levels in successively more apical leaves , whereas miR172 was expressed in the opposite pattern ( Figure 3D , branch 1 ) . Consistent with their similarity in size and shape , the first few leaves of these basal branches produced approximately as much miR156 as the leaves on the main shoot of 1-month-old plants ( compare Figure 2C and Figure 3D ) . The phase identity of a lateral branch typically matches the identity of the primary node from which it originated [1]–[3] . To determine if the expression of miR156 shows a similar pattern , we examined the level of miR156 in leaves from branches located 2 . 5 m and 4 meters from the base of the shoot . These positions were chosen to correspond to the height of the primary shoots examined in Figure 3B . Consistent with the expression of miR156 in leaf primordia from 2 m and 4 m tall primary shoots ( Figure 3B ) , the leaves of branches located 2 . 5 m from the base of the shoot ( branch 2 ) had relatively high levels of miR156 , whereas leaves on branches located 4 m from the base of the shoot ( branch 3 ) had much lower levels miR156 ( Figure 3E ) . This result provides additional evidence that miR156 regulates vegetative phase change in P . x canadensis . To determine if miR156 regulates phase change through its SPL targets , we over-expressed miR156 in P . x canadensis . Ten independent lines were generated; 9 lines had similar phenotypes , and three of these were analyzed in detail . PCR analysis using primers to the 35S promoter confirmed that these 3 lines were indeed transgenic ( Figure S1 ) . We confirmed by qRT-PCR that PcSPL3 and PcSPL9 were down-regulated in the most severe line , #1 ( Figure 4A ) . As a control , we regenerated 5 wild-type plants , and plants over-expressing β-GLUCURONIDASE ( 35S::GUS ) . 35S::GUS plants were indistinguishable from wild type clones ( Figure S2 ) . The most obvious phenotype of 35S::MIR156 plants was a change in plant height and leaf shape ( Figure 4B and Table 1 ) . Compared to wild-type plants , 35S::MIR156 plants were shorter and produced small , pale-green leaves ( Figure 4B and 4C and Table 1 ) . The severity of the phenotype of the three 35S::MIR156 lines ( #1–3 ) was correlated with their miR156 levels , with #1 and #2 having higher miR156 expression and a more severe phenotype than line # 3 ( Figure 4C , 4D ) . At six months of age , 35S::MIR156 plants resembled 1-month-old wild-type plants . Like these juvenile plants , 35S::MIR156 plants had leaves with an oval lamina ( compare Figure 4C to Figure 2B; Table 1 ) and round petioles , containing a single vascular bundle ( Figure 4E ) . Transverse sections of the lamina revealed that 35S::MIR156 plants had only a single layer of palisade mesophyll cells , in contrast to the leaves of 6-month-old wild-type plants , which had two palisade cell layers ( Figure 4F ) . In addition , 35S::MIR156 plants had shorter internodes and a faster rate of leaf initiation than 6-month-old wild type plants , and formed side branches at every node ( Table 1 and Figure 4G ) . These later traits are also characteristic of Arabidopsis and maize plants that over-express miR156 [11] , [16] , [20] , [36] , [39] . The changes in leaf and shoot morphology were paralleled by altered expression of SPL genes ( Figure 4A ) and corresponding changes in the expression of genes that are direct targets of SPL in Arabidopsis [9]–[11] , [13] , [14]—in particular , a homolog of FRUITFULL ( PcFUL ) ( Figure 4A ) and miR172 ( Figure 4D ) .
The morphology and physiology of a plant shoot change during its development . The most recognizable example of this is the transition from vegetative to reproductive growth , which is marked by the production of specialized structures , such as flowers or cones . The juvenile-to-adult transition is more difficult to recognize because it is usually accompanied by relatively subtle , species-specific changes . This has created considerable confusion about the nature of vegetative phase change . Because there is no common morphological marker for juvenile and adult phases of vegetative development , it is difficult to know whether temporal variation in particular vegetative traits in different species represent the same , or different , developmental processes . The identification of miR156 as a regulator of vegetative phase change in Arabidopsis and maize [10] , [13] , [16] , and the results presented here , resolve this long-standing problem . The expression patterns of miR156 and miR172 in woody plants with well-differentiated juvenile and adult phases , and the evidence that over-expressing miR156 delays vegetative phase change in P . x canadensis , strongly suggest that miR156 regulates vegetative phase change in many , if not all , flowering plants . miR156 is present in all major plant taxa , including bryophytes [40] , so it would not be surprising if it regulates vegetative phase change throughout the plant kingdom . This result has many important implications . Most importantly , it demonstrates the fundamental similarity between processes that overtly appear to be quite different: it is remarkable that the subtle changes in leaf morphology described as phase change in maize [22] and Arabidopsis [8] correspond to the much more dramatic changes in shoot architecture observed in Acacia , Eucalyptus , or Hedera . There was no a priori evidence that these events actually represent the same developmental process . Our results therefore validate the use of Arabidopsis and maize for the analysis of vegetative phase change , and suggest that the insights gained from these experimentally tractable species are likely to have broad applicability . ‘The evidence that vegetative phase change is mediated by a decrease in the expression of miR156 begs the question of how this decrease is regulated . miR156 plays a critical role in vegetative phase change , but control of this process resides with the factor or factors that control the expression of this miRNA . A recent study of vegetative phase change in Arabidopsis , maize and Nicotiana benthamiana indicates that the decline in miR156 is mediated by a signal produced by leaf primordia; neither the root system nor cotyledons appear to be important for this event [41] . This result suggests that the timing of vegetative phase change could be regulated by leaf number: assuming that all leaves are capable of producing a hypothetical phase change signal , then the switch from juvenile to adult development might occur when leaf number exceeds a certain threshold number . However , this simple model does not account for the tremendous variability in the timing of vegetative phase change in trees . For example , phase change occurs after 1 node in A . confusa ( Figure 1A ) , but 30 or more nodes in A . koa [42] . Similarly , in E . globulus , vegetative phase change occurs between 1 and 5 years after germination [43] , [44] . This variability suggests that the juvenile-to-adult transition is only weakly related ( if at all ) to the overall size of the shoot . Identifying the factors that regulate the expression of miR156 is an important goal for future research . The results presented here also have important practical implications . Many traits change during shoot development in trees , and the extent to which various traits are controlled by the same or different mechanisms is largely unknown [45] , [46] , [47] . Correlating changes in the expression of miR156 and miR172 with changes in various heteroblastic traits should make it possible to distinguish traits that are potentially regulated by these miRNAs from traits that are controlled by some other mechanism . It will be particularly interesting to learn if age-related changes in economically important traits—such as adventitious root production—are correlated with changes in miR156 expression , as this may open new avenues for the manipulation of these traits . Using miR156 expression as a marker for vegetative identity also makes possible to study the effects of various factors on phase change in situations in which this is otherwise difficult to do—for example , in species that do not undergo major morphological changes during vegetative development , or in short-term experimental situations that do not permit the development of fully formed leaves or shoots . This will facilitate the integration of information about vegetative phase change across species , and should help to accelerate research on this important but poorly understood developmental process .
Seeds of A . colei were obtained from the Australian Tree Seed Center ( Canberra , Australia ) , while seeds A . confusa were obtained from the Desert Legume Program of the U . of Arizona ( Tucson , AZ ) . These species were grown in Farfard #52 soil in the U . of Pennsylvania greenhouse , with supplemental illumination to extend the day length to 16 hours . Fully expanded juvenile and adult leaves were harvested from these plants when they were 2 months old . Juvenile and adult shoots of H . helix were harvested from single vines , or clones propagated from single vines . Analyses were conducted with plants growing outdoors in Media , Pennsylvania and the U . of Pennsylvania's Kasky garden , and with shoot apices of juvenile and adult clones grown in a growth chamber in short days ( 10 hrs light∶ 14 hours dark; 26°C∶21°C day∶night temperature ) to prevent flowering ( 31 ) . Fully expanded juvenile and adult leaves of Q . acutissima were harvested from trees growing on the campus of the U . of Pennsylvania . Juvenile and adult branches of these trees were identified during winter on the basis on the presence ( juvenile ) or absence ( adult ) of attached leaves , and newly expanded leaves from these branches were harvested in May , 2010 . Juvenile and adult leaves of E . globulus were harvested in October , 2010 from trees of different ages growing at three sites within the Presidio Trust in San Francisco , California . Leaves of 1-year- , 4-year- and 10-year-old P . x canadensis clones growing within 100 meters of each other at a field site in Shanghai were sampled in June , 2010 . Leaves of 1-month- and 6-month-old wild type and transgenic P . x canadensis clones were sampled in the greenhouse in Tübingen . The leaves or leaf primordia from lateral branches were harvested from 1-year-old clones grown in the greenhouse . Fully expanded leaves were detached , measured , and photographed . For leaf anatomy , leaves 1 . 5 cm in length and petioles were fixed , embedded and sectioned as previously described [39] . The rate of leaf initiation was determined from the number of the leaves produced within one week . Leaves or shoot apices from A . confusa . A . colei , H . helix , E . globulus , and Q . acutissima , were frozen in liquid nitrogen , and total RNA was extracted following a protocol modified from [48] . Small RNA was isolated and analyzed using the methods described in [49] . In brief , 1–2 grams of tissue was ground to make fine powder , pre-warmed ( at 65°C ) RNA extraction buffer ( 2% CTAB , 2% PVP40 , 100 mM Tris-HCl , 25 mM EDTA , 2 M NaCl , 0 . 5 g/L spermidine , 2% β-mercaptoethanol , pH 8 . 0 ) was added , and the mixture was incubated for 20 min at 65°C . RNA was extracted by treating the slurry twice with an equal volume of chloroform/isoamyl alcohol ( 24∶1 ) , and then precipitated with LiCl at a final concentration of 2 . 5 M . The pellet was dissolved in STE buffer ( 1 M NaCl , 10 mM Tris-HCl , 1 mM EDTA , pH 8 . 0 ) and extracted one more time with chloroform/isoamyl alcohol . RNA was precipitated with ethanol and used for RNA gel blots , as described in [49] . Densitometry of digitized images of these blots was performed using Image J ( http://rsbweb . nih . gov/ij/ ) . E . grandis homologs of AtSPL3 and AtSPL9 were identified by performing tblastn searches of the E . grandis genome ( http://www . phytozome . net/eucalyptus . php ) . PCR primers based on the E . grandis sequence ( Table S2 ) were used to amplify the corresponding genes from cDNA of fully expanded adult leaves of E . globulus , and the resulting PCR products—EglSPL3 ( HQ450389 ) , EglSPL9 ( HQ450390 ) , and EglEIF4 ( HQ450391 ) —were sequenced . qRT-PCR was performed on RNA isolated by the method described above . Reverse transcription was performed with SuperScript™II reverse transcriptase ( Invitrogen ) using an oligo ( dT ) primer . qRT-PCR reactions were performed using the EglSPL3 , EglSPL9 and EglEIF4 primers listed in Table S2 and the Power SYBR Green Master Mix ( Applied Biosystems ) . Reactions were monitored and analyzed using StepOneTM Software v2 . 0 . 1 ( Applied Biosystems ) , and were normalized to the quantity of EgEIF4 . Three technical replicates were performed for samples harvested from three trees , yielding a total of 9 reactions per leaf type . In the case of P . x canadensis , total RNA was extracted from leaves with Trizol reagent ( Invitrogen GmbH , Germany ) . One µg of total RNA was DNase I-treated and used for cDNA synthesis with oligo ( dT ) primer and Superscript reverse transcriptase ( Invitrogen ) . qRT-PCR was performed with SYBR-Green PCR Mastermix ( Invitrogen ) and amplification was real-time monitored on an MJR Opticon Continuous Fluorescence Detection System ( Biorad , Hercules , CA ) and analyzed using the software provided by the manufacturer . Two biological replicates ( each with three technical replicates ) were performed . The oligos for PcSPL3 , PcSPL9 , PcFUL , and PcACT were designed based on the homologous genes of P . trichocarpa [50]: PtSPL3 ( XM_002329758 ) , PtSPL9 ( XM_002322642 . 1 ) , PtFUL ( XM_002317909 . 1 ) , and PtACT ( XM_002298674 ) ( Table S2 ) . P . x canadensis cv . Guangzhao Yang plants , were grown at 23°C in 16 hours long days . The 35S::MIR156 [11] and p35S::GUS constructs were introduced into Agrobacterium tumefaciens ( strain GV3101 [pMP90] ) and used for plant transformation . An overnight A . tumefaciens culture was pelleted and resuspended in infection medium ( 1/2 MS , 45 g/L sucrose , 200 µM acetosyringone ) . Leaves were infected for 30 min and then transferred to co-culture medium ( MS , 0 . 25 mg/L 6-benzyl aminopurine , 0 . 25 mg/L kinetin , 0 . 25 mg/L trans-zeatin , 0 . 25 mg/L naphthalene acetic acid , 100 µM acetosyringone ) . After incubation at 24°C for 3 days , leaves were transferred to selective differentiation medium ( MS , 0 . 25 mg/L 6-benzyl aminopurine , 0 . 25 mg/L kinetin , 0 . 25 mg/L trans-zeatin , 0 . 25 mg/L naphthalene acetic acid , 500 mg/L carbenicillin , 50 mg/L kanamycin ) . Three weeks later , the explants were transferred to selective elongation medium ( MS , 0 . 1 mg/L 6-benzyl aminopurine , 300 mg/L carbenicillin , 100 mg/L kanamycin ) . This was repeated once . Kanamycin-resistant shoots were transferred into induction medium ( MS , 0 . 2 mg/L indole-3- butyric acid , 200 mg/L carbenicillin , 50 mg/L kanamycin ) for root induction . Wild-type plants were regenerated on plates without kanamycin selection . | The existence of discrete juvenile and adult phases of vegetative development in plants was first recognized in trees , in which these phases are usually prolonged and quite stable . Annual plants also undergo changes in vegetative morphology during shoot development , but the relationship between this process and vegetative phase change in trees is unclear . This is because both the timing and the nature of the morphological changes that mark these transitions are different in these groups of plants . Here we show that the expression pattern of miR156—a master regulator of vegetative phase change in Arabidopsis and maize—is conserved in woody plants with well-defined juvenile and adult phases , and we show that over-expression of this microRNA prolongs the expression of the juvenile phase in the tree Populus x canadensis . Our results indicate that the mechanism of the juvenile-to-adult transition is likely conserved throughout flowering plants . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"developmental",
"biology",
"plant",
"science",
"plant",
"biology",
"genetics",
"biology",
"evolutionary",
"biology",
"genetics",
"and",
"genomics"
] | 2011 | MiRNA Control of Vegetative Phase Change in Trees |
The activity of trans-membrane proteins such as ion channels is the essence of neuronal transmission . The currently most accurate method for determining ion channel kinetic mechanisms is single-channel recording and analysis . Yet , the limitations and complexities in interpreting single-channel recordings discourage many physiologists from using them . Here we show that a genetic search algorithm in combination with a gradient descent algorithm can be used to fit whole-cell voltage-clamp data to kinetic models with a high degree of accuracy . Previously , ion channel stimulation traces were analyzed one at a time , the results of these analyses being combined to produce a picture of channel kinetics . Here the entire set of traces from all stimulation protocols are analysed simultaneously . The algorithm was initially tested on simulated current traces produced by several Hodgkin-Huxley–like and Markov chain models of voltage-gated potassium and sodium channels . Currents were also produced by simulating levels of noise expected from actual patch recordings . Finally , the algorithm was used for finding the kinetic parameters of several voltage-gated sodium and potassium channels models by matching its results to data recorded from layer 5 pyramidal neurons of the rat cortex in the nucleated outside-out patch configuration . The minimization scheme gives electrophysiologists a tool for reproducing and simulating voltage-gated ion channel kinetics at the cellular level .
Ion channels are trans-membrane proteins that close and open in reaction to changes in membrane potential , among other factors , thus leading to a change in ion flow across the membrane . Membrane potential may be the most significant factor affecting the activity of ion channels , for not only do the kinetics of many channels depend on membrane potential but also changes in membrane potential are the main instigator of neuronal activity [1 , 2] . The kinetics of these voltage-dependent ion channels are complex , requiring the construction of intricate kinetic models to understand ion channel behaviour . The dominant paradigm for ion transport over the past 50 years is based on the seminal experiments of Hodgkin and Huxley [3–8] . Their detailed kinetic models derived from the giant axon of Loligo are still extremely useful in studies of ion channels and of neuronal physiology . However , a much more detailed picture of the mechanisms underlying membrane excitation has emerged over the years , emphasizing several disagreements with the Hodgkin-Huxley model . These include their proposed lack of connectivity between the activating and inactivating “gates” of the voltage-dependent sodium channel [9] and their premise that the inactivation gate voltage dependence is due to its coupling to the activation process; i . e . , the inactivation gate can close only after the activation gate opens [10 , 11] . Nevertheless , the Hodgkin-Huxley model is still predominant in many simulations of neuronal physiology , mostly due to its ease of use and conceptualization combined with the relative small number of free parameters needing estimation for the quantification of channel behavior . Most models proposed to replace the Hodgkin-Huxley model still use the same kinetic formalism and are only satisfactory in explaining certain aspects of channel behaviour but fail in others . For example , the classical Hodgkin-Huxley paradigm does not consider interactions between varying kinetic states , particularly that between activation and inactivation . This failure leads to an erroneous estimation of the kinetic parameters and thus of the predicted channel dynamics [12] . The best method , so far , for discovering the kinetics of ionic channels is by analysis of single-channel recordings [13] . However , single-channel recording and analysis suffer from several problems . One needs to accurately subtract the capacity of the electrodes , and the analysis of first latencies is extremely difficult [13–18] . Most data on neuronal voltage-gated conductances obtained in studies of cellular physiology have thus been collected when many channels were activated simultaneously , either in the whole-cell mode or in excised patches containing many channels . That is , most models of voltage-gated channels that aim to explain the physiological function of the channels are generated by analysing simultaneous activity in many channels . Here we propose a method for analyzing whole-cell recordings of voltage-gated channels . Our working hypothesis is that it is possible to verify the viability of voltage-dependent ion channel models using a genetic optimization algorithm concurrently with a full-trace fit of experimental data to the model . We therefore scanned several of the better-known models of voltage-gated sodium and potassium channels to examine their accuracy in predicting and reproducing measured currents . Though the data provided for this paper derive from somata of L5 pyramidal neurons of the rat cortex , the method suggested here is applicable to all types of voltage-clamp recordings from different neuron classes . Note that our data were obtained using the nucleated patch configuration; thus , the models proposed here are not as detailed as those that may be obtained using single-channel recording . However , they are functionally significant models which allow us ( and hopefully future researchers ) to predict , simulate , and analyze neuronal physiology .
The model and the genetic algorithm ( GA ) were programmed using NEURON 5 . 7 and 5 . 8 [19] . We parallelized the process using a cluster of ten Pentium 4 computers with a 3 GHz clock speed sharing the same network file system ( NFS ) . One of the machines functioned as a master , submitting and managing the jobs using a Parallel Virtual Machine ( PVM ) , while the rest were slaves , reading and writing information from a shared directory in the network file system . Ion channel models were implemented using the NMODL extension of NEURON [20] . Results were analyzed using custom procedures written in IgorPro 5 . 01 ( Wavemetrics , http://www . wavemetrics . com/ ) . A GA is a search algorithm based on the mechanisms of Darwinian evolution . It uses random mutation , crossover , and selection operators to breed better models or solutions ( individuals ) from an originally random starting population [21] . In this study , we started each search with a random population that was at least 20 times larger than the number of free parameters in the fitted model . Each individual in the population described a parameter set and the model was evaluated for each one of them . A search space was defined for each parameter; this avoided parameter combinations causing instability to the set of differential equations , while covering most of the physiological range expected for the parameters . Thus , for rate constants ( k ) the range was set from zero to 2 , 000 s−1 , for voltage-dependence parameters ( z ) from zero to 2000 V−1 , and zero to 100 pS for the conductance . Only in model C , where a parameter describing a voltage shift was required , was the range set from zero to +100 mV , disregarding the negative range of this parameter , since the expected value for this parameter was positive in both the simulated and the real data . The population was sorted according to the value of the cost function ( Equation 1 ) of each individual , and a new generation was created using selection , crossover , and mutation as operators . Selection used a tournament in which two pairs of individuals were randomly selected and the individual with the better score from each pair was transferred to the next generation . This procedure was repeated N/2 times ( where N is the size of the population ) until the new population was full . The one exception to this selection process ( and later to the crossover and mutation operators ) was the best individual that was transferred unchanged to the next generation to prevent a genetic drift . Each pair selected for transfer to the new population was subjected to a one-point crossover operator with a probability of 0 . 5 . After the new population was created , each parameter value in the new population was subjected to mutation with a probability of 0 . 01 . This allowed , albeit with a low occurrence frequency , the creation of double and even triple mutations to the same individual , thus increasing the variability in the new population . As detailed in Figures 1 and 2 , two types of mutation operators were used . The first was a substitution of the parameter value with a random value drawn from a flat random number distribution that spanned the entire search space of the parameter . The second mutation operator , depicted in Figure 2 , was a relative operator which changed the value of a parameter relative to its current value using a random number drawn from a Gaussian distribution centred on the current value of the parameter with a relative variance of 5% . We tested several values of mutation and crossover probabilities and found these values to be the optimal for the current project . Ideally , the termination criterion should be that some cost function reaches a value of zero . In practice this is not possible since the run time of the process is limited and reaching this score can take a long time . The simulations ranged from less than an hour for a simple model with a small amount of data ( see also the demonstration code in Text S1 , which takes two hours on a single Pentium 4 with a 3 GHz clock speed and less than 20 minutes on our cluster and the animation of the convergence of the GA to a model of a voltage-gated ion channel in Video S1 ) to more than a week for a 20-parameter model fitting many experimental points . Therefore , the process was terminated when the value of the best individual had not changed for several hundred generations . Depending on the complexity of the model , this occurred after 1 , 000–30 , 000 generations . The cost function calculated root mean distance between the target and the test ionic current: where T represents the target data set and t the test dataset . N was the total number of points in each simulated ionic current trace and M the number of voltage-clamp sweeps simulated in the model . To rank the ability of various models to fit the data , we used the Log Error Ratio ( LER ) : where χA and χB are the sum of squared errors for fitting the data to models A and B , respectively [22] . Equation 2 applies in theory to models containing a similar number of parameters . This can be corrected for by using the asymptotic information criterion AIC = 2 · ( NPA − NPB ) / n [23] , where NPA and NPB are the number of free parameters in each model and n is the number of data points . In this study a large dataset with 15 , 000–30 , 000 data points was used for fitting . Therefore , the AIC correction was small and not applied in the calculations . Only LER values are reported . The models used to generate simulated currents are described below . Some minor changes to these models were made when voltage-gated K+ currents recorded from nucleated patches were analyzed . The modifications are noted in Table 1 . Moreover , many published models contain mathematical expressions that hinder the proper use of minimization algorithms . For example , a common expression of a rate constant in a model for a voltage-gated channel can often be seen to assume the general form , k = Aexp ( −z ( V − V1/2 ) ) . However , the expression exp ( zV1/2 ) can also be expressed as part of the pre-exponential value leading to a simpler equation k = A′exp ( −zV ) . Using the former expression in a minimization scheme invariably leads the algorithm astray due to the interchangeability of the pre-exponential and the fixed parameters in the exponent . Therefore , in all the models we have taken from the modeling literature and that appear below we verified that such interchangeability was eliminated by modifying the formal description of the model . Slices ( sagittal , 300 μm thick ) were prepared from the somatosensory cortex of 13–15 days old Wistar rats that were killed by rapid decapitation , according to the guidelines of the Bar-Ilan University animal welfare committee . Slice preparation followed Stuart [27] . Slices were perfused throughout the experiment with an oxygenated artificial cerebrospinal fluid ( ACSF ) containing: ( mM ) 125 NaCl , 25 NaHCO3 , 2 . 5 KCl , 1 . 25 NaH2PO4 , 1 MgCl2 , 2 CaCl2 , and 25 Glucose ( pH 7 . 4 with 5% CO2 ) . All experiments were carried out at room temperature ( 20–22 °C ) . Pyramidal neurones from L5 in the somatosensory cortex were visually identified using infrared differential interference contrast ( IR-DIC ) videomicroscopy [27] . To record voltage-gated K+ currents , the standard pipette solution contained ( mM ) : 125 K-gluconate , 20 KCl , 10 HEPES , 4 MgATP , 10 Na-phosphocreatine , 0 . 5 EGTA , and 0 . 3 GTP ( pH 7 . 2 with KOH ) . 10 mM 4-AP was included in the bath solution to reduce the amplitude of the A-type K+ conductance [28] . The pipette solution for recording voltage-gated Na+ currents contained ( mM ) 120 Cs-gluconate , 20 CsCl , 10 HEPES , 4 MgATP , 10 Na-phosphocreatine , 1 EGTA , and 0 . 3 GTP ( pH 7 . 2 with CsOH ) . In addition , 30 mM TEA was added to the bath solution to reduce residual K+ current amplitude . A similar amount of NaCl was removed from the bath solution to maintain constant osmolarity . Nucleated outside-out patches [29] were extracted from the soma of L5 pyramidal neurons . Briefly , negative pressure ( 180–230 mbar ) was applied when recording in the whole-cell configuration , and the pipette was slowly retracted . Gentle and continuous retraction created large patches of membrane engulfing the nucleus of the neuron . Following extraction of the patch , the pressure was reduced to 20–40 mBar for the duration of the experiment . All measurements from nucleated and cell-attached patches were carried out with the Axopatch-200B amplifier ( Axon Instruments , http://www . axon . com ) . The capacitive compensation circuit of the amplifier reduced capacitive transients . Nucleated patches were held at −60 mV unless otherwise stated . Linear leak and capacitive currents were subtracted off-line by scaling of 20–30 average pulses measured during hyperpolarization ( −80 to −100 mV ) . Currents were filtered at 5–10 kHz and sampled at rates two to ten times higher than the filtering frequency . The reference electrode was an Ag-AgCl pellet placed in the pipette solution and connected to the experimental chamber via an agar bridge containing 150 mM KCl . Under these conditions , the total voltage offset due to electrode and liquid junction potentials [30] was 2 mV . Membrane potential was not corrected for this potential difference . Recordings were made with fire-polished Sylgard-coated ( General Electric , RTV615 ) pipettes ( 5–8 MΩ ) .
One of the main issues concerning GAs ( or any minimizing algorithm ) is computing power and , consequently , execution times . One of the main goals , when dealing with minimization algorithms , is therefore to produce algorithms which are non-cumbersome , while retaining their ability to efficiently sample the parameter space . A common method for achieving this goal is to limit the parameter search space , for example by adaptive trimming down of the ranges for parameter search [31] . These methods did not reduce the complexity and time detriments of our algorithm . Instead , while running and analyzing the performance of the GA , we observed a recurring behavior . In 14 preliminary runs , the GA demonstrated a comparable pattern of convergence , where after several hundred generations the best individual in each generation quickly converged to a region in the parameter space . This is illustrated in Figure 1 , which describes the steps taken by each parameter of the best individual in each generation in a nine-parameter model . The steps toward each target value are displayed as the percentage change from that parameter value in the previous generation . Figure 1A displays the steps taken by the parameters over the first two hundred generations of the GA progress . During the first fifty generations , large changes were observed in the values of some of the parameters in the best individual vector . In the further generations , parameter values in the best individual changed only in small steps . Figure 1B shows a detailed view of the algorithm convergence presented above , showing clearly that after several hundred generations the parameters have mutated to only a few percent from their values in the previous generation . Since the parameter search space remained constant , the small changes in value as parameters converged became more insignificant in comparison with the full range of the parameter search space . We solved this problem by defining a new range for each parameter mutation search space using a Gaussian centered on the best value of that parameter obtained in the previous generation with a relative variance of 5% . The randomly picked value was then multiplied by the current value of each individual found by the GA , providing a new value limited to ∼±22% of the current value of each individual ( Figure 2A ) . The Gaussian range was implemented only after the GA went through at least 500 generations , taking advantage of the fast convergence observed in the initial stages of the run . This new adaptive range for parameter mutation search space proved more efficient than the previously attempted fixed ranges . It increases the probability of the algorithm searching for better individuals around the best value of the individual found so far , as was indeed observed in the preliminary runs . At the same time the possibility of a better individual existing farther away is not dismissed ( as the Gaussian is infinite on both ends ) . A comparison between the GA with and without the adaptive Gaussian range revealed that , while the original algorithm gradually converged and reached a score of 0 . 1 after 10 , 000 generations ( Figure 2B , smooth line ) for the nine-parameter Model A described in the Methods , the new GA converged after 1 , 000 generations to a score of 10−4 ( Figure 2B , dashed line ) . Subsequently , we simulated potassium currents using several basic kinetic models for potassium channels ( see the Channel Models section in Methods ) . The potassium current data were saved and used as a reference for the GA convergence . We initially tested a nine-parameter model ( Model A ) with equal value parameters . This did not prevent the resulting simulated currents from resembling experimentally recorded ones . Figure 3A displays the activation of the conductance in response to 20 mV voltages steps from −80 to +60 mV . Figure 3B shows the deactivation of the conductance after 20 ms activation to +60 mV from potentials ranging from −120 mV to −20 mV in steps of 10 mV . This entire set of data was used as the target dataset for calculating the score function ( Equation 1 ) for each individual generated by the GA . After approximately 5 , 000 generations , the GA converged all of the nine parameters to within 40% of their true target ( Figure 3C ) . After the GA run was stopped , the data were run through a hill-climbing Principle Axis algorithm ( PrAxis ) [32] , which is part of the NEURON simulation environment . This converged the remaining parameters to 1%–2% of the target values ( Figure 3C ) . To determine the efficiency of the PrAxis routine alone , we generated 100 random parameter vectors and used them as starting guesses for this routine . In none of the cases was the PrAxis routine able to provide even a rough fit of the data . Thus , it is only the combination of the GA followed by the PrAxis that produced a good fit . We further simulated potassium currents using a thirteen-parameter model ( Model B ) . Again , the potassium current data for activation ( Figure 4A ) and deactivation ( Figure 4B ) was saved and used as the target dataset for the GA convergence . As in the previous experiment , after ∼5 , 000 generations the GA converged all but two parameters to within 40% of their target values ( the remaining two reaching values less than 200% of their target value ) . Once more , ( being independent of membrane potential ) converged to within a few hundredths of a percent from its target value . As before , the GA data were run through PrAxis , producing a fit that was only a few percent from the target value for all parameters ( Figure 4C ) . Similar results were obtained for several different target parameters sets . Using the dataset produced by Models A and B , we also investigated the ability of simulated annealing ( SA ) and random sampling to replace the GA scheme we present here . The SA algorithm performed as efficiently as the GA in some cases and much worse in others ( simulations not shown ) . Analysis of the simulations revealed that , unlike the GA , the cooling scheme of the SA algorithm had to be repeatedly fine-tuned to properly constrain the parameters of different models . We also tested the ability of extensive random sampling as a substitute for the GA . After 1 , 000 , 000 iterations , the best score obtained for Model B was ∼6 , 000 , at least two orders of magnitude bigger than the score obtained by the GA after a similar number of iterations . Since our analysis uses a large amount of data , we tested its accuracy in converging on the target parameters using varying amounts of data . We compared the accuracy of the combined convergence of the GA and PrAxis for an increasing number of stimulation sweeps ( Figure 9 ) previously used with Model B to produce the activation traces in Figure 4A . The score function produced a seemingly near-perfect fit when using a small number of stimulation sweeps , but its divergence from zero increased as the number of sweeps grew ( Figure 7A ) . However , an opposing trend was observed when comparing the mean deviation of the best parameter values reached by the GA and PrAxis from their target values ( Figure 7B ) . We therefore conclude that an increasing number of data sweeps is needed for an accurate estimate of the kinetic parameters of a model and does not lead to overfitting of the model . This conclusion was further supported by simulations run on Model C , which proved that an increasing number of stimulation protocols , expressing varying aspects of channel kinetics , also add to the accuracy of the process ( unpublished data ) . As our aim was to use our GA in fitting a model to currents recorded from neurons , we needed to test our GA's ability to converge in the presence of noise . We therefore created several simulations consisting of randomly generated noise of varying amplitudes . We first tested a 14-parameter model with equal parameters in the presence of random noise whose amplitude was 5% of the current value . ( This model was identical to the 13-parameter Model B plus a maximal conduction density for the deactivation protocol , simulating variability between consecutive recordings or , alternatively , data obtained from two different recordings . ) The final convergence of both GA and PrAxis resulted in a fit of an average of 5% from the true values ( fit not shown ) . We then ran three similar experiments with constant amplitude noise varying between ±10 , ±20 , and ±30 pA in each experiment . The results illustrate the GA's ability to converge in spite of the noisy data ( Figure 8 ) . The average distance of the parameters from their target values was 1 . 4% , 2 . 5% , and 1 . 4% for the three noise levels respectively . The response of voltage-gated channels to changes in membrane potential is traditionally measured using a step change in the membrane potential . This is the essence of the voltage-clamp method , which stems from the relative ease of analytically solving the differential equations describing channel relaxation following step activation . Here we have used numerical methods to solve the equations describing channel activation . Therefore , it occurred to us that the GA may also be able to estimate the parameters of a model using a set of data obtained with not-so-standard voltage-clamp protocols . One such protocol is the voltage ramp , which is appealing mainly from the experimental point of view . The voltage ramp is a very useful protocol since it allows the experimentalist a glimpse of the full voltage dependence of a channel using one fast protocol . Furthermore , of all voltage-clamp protocols , the voltage ramp is unique in that during the ramp the contribution of the capacitance to the current is constant . This allows simpler and cleaner leak subtraction than when a square pulse is applied and the capacitive current approaches infinity at the onset of the pulse . Figure 9A displays the activation ( top traces ) of a nine-parameter model of a voltage-gated K+ channel ( Model A ) in response to ramps of varying duration with potential ranging from −100 mV to +50 mV ( Figure 9A , bottom traces ) . Figure 9B displays the response of the same model to deactivating voltage ramps from +50 mV to −100 mV ( following a 50-ms step to +50 mV to fully activate the conductance ) . The dataset displayed in Figure 9A and 9B was used as target dataset to the GA to determine whether it can be used to constrain the parameters of Model A . After 5 , 000 generations , the error in the parameters , relative to the target parameters , ranged between 0 . 01 to 42% ( Figure 9C ) . When this set of parameters was used as an initial guess for the PrAxis hill-climbing algorithm , the error range was reduced to between 10−4% to 5% ( Figure 9C ) . Similar results were obtained when Model B was used to generate the target dataset . Thus , this set of simulations demonstrated that the GA could locate the global minimum even when the target dataset was generated using non-classical voltage-clamp protocols . Following our successful simulations of potassium channel models , we proceeded to more complex simulations of sodium channels . One model tested was based on the sodium current recorded from retinal ganglion cells [24] . The small changes we made to the model ( Model C in Methods ) were necessary to eliminate redundancy in the original model , which would have prevented the GA from constraining the parameters to the appropriate values . When considering a voltage-gated channel that displays both activation and inactivation , the repertoire of voltage protocols increases substantially . The five basic voltage protocols routinely applied in such cases are activation ( Figure 5A ) , deactivation ( Figure 5B ) , steady-state inactivation ( Figure 5C ) , pulse inactivation ( Figure 5D ) , and recovery from inactivation ( Figure 5E ) . Using this target dataset , the GA , after ∼11 , 000 generations , generated a parameter set that deviated from the target parameter set by 10%–200% ( Figure 5F ) . Using this set of parameters as an initial guess for the PrAxis hill-climbing algorithm reduced the error range from 10−4% to 10−2% , practically a perfect fit ( Figure 5F ) . Next we tested the GA on voltage-gated K+ currents recorded in nucleated outside-out patches extracted from layer 5 neocortical pyramidal neurons . These neurons contain several voltage-gated K+ channels [28 , 33] . To reduce the number of channels , we blocked the A-type K+ channel with 10 mM 4-AP . The residual current , the slow voltage-gated K+ channel [28 , 33] , was activated by voltage steps from −80 to +40 mV ( Figure 10A ) . All the traces in Figure 10 were recorded with an extracellular K+ concentration of 6 . 5 mM to generate larger deactivating tail currents . The activation and deactivation data traces were used as the target dataset for the GA , and the fitness of several models was tested . Following convergence of the GA , the best parameter set was used as an initial guess for the PrAxis routine . The results of these minimizations are summarized in Table 1 where models are compared using their LER [22] . The best fit was obtained for a model with three closed states and one open state in which the rate constants were exponential . However , the fitness of this model , containing 21 free parameters , did not differ substantially from a model with two closed states and only 11 parameters . Thus , to avoid overfitting , we decided that the latter model produced the best fit for this dataset . This fit is shown in Figure 10A and 10B as red lines . Similar results were obtained in two more patches . Next , we investigated whether the model obtained in this way could predict the response of the conductance to other , less traditional , voltage-clamp protocols . Responses to activation ramps from −100 mV to +40 mV , with varying durations starting from 40 ms and increasing in steps of 10 ms ( blue lines ) , were simulated . These were compared with currents recorded from the patch used in Figure 10A and 10B using identical voltage-clamp protocols ( Figure 10C ) . Simulations of responses to deactivation ramps from +40 mV to −80 mV with durations increasing in 5-ms steps from 2 ms ( blue lines ) were similarly compared with recorded current traces ( Figure 10D ) . Figure 10E and 10F depict a further comparison of data recorded from the patch and data simulated on the best model using the stimulation protocols shown in Figure 10A and 10B . The voltage-clamp protocols in Figure 10E and 10F were sine waves from −70 mV to +70 mV with a frequency of 50 Hz and 100 Hz , respectively . Again , the blue lines depict the current produced by Model A which had the best fit to the activation and deactivation data in Figure 10A and 10B . Subsequently , we used the GA on voltage-gated Na+ currents recorded in nucleated outside-out patches extracted from layer 5 neocortical pyramidal neurons . Data traces from activation , steady-state inactivation , pulse inactivation , and recovery from inactivation were used as the target dataset for the GA , and the fitness of several models was tested . The activation , steady-state inactivation , and pulse inactivation currents of the voltage-gated Na+ channel are shown in Figure 6 . The activation currents were produced in response to voltage steps from −30 to +35 mV in steps of 5 mV from a holding potential of −110 mV . The steady-state inactivation currents were produced in response to a voltage step to 0 from holding potentials varying from −105 mV to −5 mV in 10-mV increments . The pulse inactivation currents were produced in response to a voltage step to −50 mV ( from a holding potential of −110 mV ) for durations varying from 0 ms to 45 ms in increments of 5 ms followed by another voltage step to 0 ms for 30 ms . After convergence of the GA , the best parameter set was used as an initial guess for the PrAxis routine . The results of these minimizations are summarized in Table 2 where models are compared using their LER . The best fit was obtained for a model containing two closed states , an open state , and two inactivated states ( Model D , see Figure 6 ) . Examination of the visual fit reveals that Model D has a more accurate description of the channel inactivation , while Model E ( Figure 6 ) is slightly better in depicting the channel activation .
Our study describes a method for analyzing voltage-dependent ionic currents . We tested and affirmed the ability of the GA presented here to fit current traces using the full-trace analysis method . The GA was then used to test the fit of various previously published and new voltage-dependent ion channel models ( Figures 3 , 4 , 7–9 ) . The models were further used to produce currents equivalent to the potassium ( Figure 10 ) and sodium currents ( Figure 6 ) measured in patch-clamp experiments . We conclude that it is possible to verify the viability of voltage-dependent ion channel models using a genetic optimization algorithm concurrently with a full-trace fit of experimental data to the model . The advantage of our scheme over the more commonly used gradient descent algorithms is that GAs do not require an initial guess of the parameters [12 , 34 , 35] . With a large enough dataset , it is possible to arrive at the global minimum even from a random starting point . This is very important since , given wrong starting parameters , gradient descent algorithms may arrive at a local minimum . However , although this study obtained good results in almost all cases , this does not constitute a proof that this method will work each time , since there may be a random combination of parameters that defy the GA . Still , our results show that a useable and functionally significant ion channel model describing whole-cell currents may be produced using a GA with a dataset containing a “complete” whole-cell activation of the channel as its input . In contrast to the disjoint conventional method , the approach to data analysis and model fitting used here has been designated global curve fitting or the full-trace method [12 , 34] . However , the analysis presented here shows several important differences from previously suggested full-trace analyses . The most obvious difference is the use of multiple stimulation protocols; previous full-trace analyses used only activation and deactivation protocols for their data and fit [12 , 34 , 35] . Furthermore , while using the full current trace , previous analyses still made the same basic assumptions as their predecessors using the disjoint method , namely the number of gates in a model , the values of m0 and h∞ in relation to time constants and peak conductance ( when dealing with Hodgkin-Huxley–like channel models ) . Additionally , either due to experimental constraints or ease of use , the same description for rate functions employed by the disjoint method has been used in previous full-trace analysis experiments . Also , independent time constants were used for each potential step in full-trace fits , much as they are commonly used by the disjoint method . The analysis presented here made no assumptions similar to those used by the disjoint method , thus negating any possible predispositions toward certain time constant–voltage dependencies , stimulation protocols , or other rate-function–limiting conditions . In addition , it is not necessary to consider each voltage step on its own when calculating the time constant dependence on voltage . As all the voltage steps are simulated and fit at once , the resulting time constants should and do fit the target data at all voltage steps . Also , not only does the full-trace analysis presented here take advantage of the full range of stimulation protocols used for voltage-dependent ion channels ( Figure 5 ) , but diverse methods of stimulation can also be considered ( Figure 9 ) . The more complex the channel model ( i . e . , the more parameters it contained ) , the longer it took and the more difficult it was for the GA to fit the data to the model within a similar time frame . Markov models took longer for the algorithm to locate and converge on the global minimum area in the parameter space . It is also worth noticing that though our experiments show decent fits within reasonable time frames for models with up to 21 parameters , it would probably be prudent not to use such methods for models consisting of more than fifteen parameters when limited by time and computational overhead ( especially considering that neither the maximal conductances nor the reversal potential are voltage-dependent and that the search for their global minimum in the parameter space was much quicker than for the corresponding area of the voltage-dependent parameters ) . Moreover , it is probably best not to try and fit Markov models with many sequential closed states . The kinetic time constants of closed states that do not contribute to the actual whole-cell current will probably be hard to estimate . To investigate such complex models , it is clear that single-channel analysis is more suitable than the approach we have used [13] . Even though we present here a relatively good fit of several models , these models are phenomenological and may not fully describe the gating of the channels . A much more detailed investigation is required to produce a more accurate model of any channel . Such investigation will most probably entail both whole-cell and single-channel recording and analysis . Moreover , sequencing of the channel cDNA and introducing deletions or mutations to the sequence may be required in order to fully determine the structure–function relationship . However , our approach may be used for rapid evaluation of models , which may speed the production of physiologically relevant models . One forte of the presented approach is most visible when comparing the fit of two models to the same dataset ( Figure 6 ) . Both Model D and Model E fit the activation of the channels reasonably well . However , Model D also fits the steady-state inactivation and pulse-inactivation data , whereas Model E fails to fit this portion of the data . Had we used only a dataset containing the activation in the fit , these differences between the two models would not have been exposed . This emphasizes that no fitting routine can replace a rigorous investigation of channel kinetics . Such investigation is crucially required to determine the extent of the dataset required to faithfully represent all the kinetic behavior of the channel . For example , using only activation and deactivation protocols would do injustice to complex processes such as C-type inactivation of voltage-gated K+ channels [36 , 37] , the slow inactivation of T-type Ca2+ channels [38] , the dependence of channel gating on the concentration of ions in the solution [39 , 40] , and many other unique kinetic properties . Our approach also allows the use of nonstandard voltage-clamp protocols . As recently noted [15] , some parameters that may appear in models of voltage-gated channels may be better constrained when the fit is carried out on currents recorded using action-potential trains as a voltage-clamp command instead of the standard voltage-clamp protocol . Admittedly , the standard voltage-clamp protocols facilitate intuitive analysis of the data . Thus , a combination of standard voltage-clamp protocols , used for initial identification of the investigated conductance , with less-standard protocols , used for constraining a model for that conductance , may prove a beneficial approach for future investigations of voltage-gated conductances . Finally , our analysis technique is automatic . Thus , while the GA searches parameter space for the best combination , the investigator is free to perform more experiments or to design additional models . This is an improvement over the previous disjoint method where every trace had to be manually analyzed , which is inefficient and needlessly time-consuming . | Voltage-gated ion channels affect neuronal integration of information . Some neurons express more than ten different types of voltage-gated ion channels , making information processing a highly convoluted process . Kinetic modelling of ion channels is an important method for unravelling the role of each channel type in neuronal function . However , the most commonly used analysis techniques suffer from shortcomings that limit the ability of researchers to rapidly produce physiologically relevant models of voltage-gated ion channels and of neuronal physiology . We show that conjugating a stochastic search algorithm with ionic currents measured using multiple voltage-clamp protocols enables the semi-automatic production of models of voltage-gated ion channels . Once fully automated , this approach may be used for high throughput analysis of voltage-gated currents . This in turn will greatly shorten the time required for building models of neuronal physiology to facilitate our understanding of neuronal behaviour . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"physiology",
"neuroscience",
"rattus",
"(rat)",
"computational",
"biology"
] | 2007 | A Numerical Approach to Ion Channel Modelling Using Whole-Cell Voltage-Clamp Recordings and a Genetic Algorithm |
One goal of modern day neuroscience is the establishment of molecular maps that assign unique features to individual neuron types . Such maps provide important starting points for neuron classification , for functional analysis , and for developmental studies aimed at defining the molecular mechanisms of neuron identity acquisition and neuron identity diversification . In this resource paper , we describe a nervous system-wide map of the potential expression sites of 244 members of the largest gene family in the C . elegans genome , rhodopsin-like ( class A ) G-protein-coupled receptor ( GPCR ) chemoreceptors , using classic gfp reporter gene technology . We cover representatives of all sequence families of chemoreceptor GPCRs , some of which were previously entirely uncharacterized . Most reporters are expressed in a very restricted number of cells , often just in single cells . We assign GPCR reporter expression to all but two of the 37 sensory neuron classes of the sex-shared , core nervous system . Some sensory neurons express a very small number of receptors , while others , particularly nociceptive neurons , coexpress several dozen GPCR reporter genes . GPCR reporters are also expressed in a wide range of inter- and motorneurons , as well as non-neuronal cells , suggesting that GPCRs may constitute receptors not just for environmental signals , but also for internal cues . We observe only one notable , frequent association of coexpression patterns , namely in one nociceptive amphid ( ASH ) and two nociceptive phasmid sensory neurons ( PHA , PHB ) . We identified GPCRs with sexually dimorphic expression and several GPCR reporters that are expressed in a left/right asymmetric manner . We identified a substantial degree of GPCR expression plasticity; particularly in the context of the environmentally-induced dauer diapause stage when one third of all tested GPCRs alter the cellular specificity of their expression within and outside the nervous system . Intriguingly , in a number of cases , the dauer-specific alterations of GPCR reporter expression in specific neuron classes are maintained during postdauer life and in some case new patterns are induced post-dauer , demonstrating that GPCR gene expression may serve as traits of life history . Taken together , our resource provides an entry point for functional studies and also offers a host of molecular markers for studying molecular patterning and plasticity of the nervous system .
Molecular markers selectively expressed in individual neuron types represent invaluable tools to understand how cellular diversity in a nervous system is genetically encoded . Molecular markers that are constitutively and invariably expressed throughout the life of a specific neuron type provide static views of neuronal identity and hence provide entry points to study how invariable identity features are acquired during neuronal differentiation [1] . In contrast , some molecular features of a neuron display a remarkable plasticity in that their expression may be regulated by neuronal activity or in response to specific environmental cues . Such genes serve as markers to understand the nature of the gene regulatory programs that govern such dynamic features of a neuron . We reasoned that a significant expansion of the expression analysis of chemosensory G-protein-coupled receptors ( GPCRs ) , initiated more than 20 years ago [2] using gfp-based reporter gene technology [3] , may yield a significantly expanded resource of molecular markers that may label various aspects of neuronal identity and neuronal plasticity in the C . elegans nervous system . Animal genomes encode five major classes of GPCRs , of which the rhodopsin class ( or “class A” ) is the largest class [4 , 5] ( Table 1 ) . Rhodopsin class GPCRs can be subdivided into phylogenetically deeply conserved neurotransmitter receptors ( neuropeptides , acetylcholine , biogenic amines ) as well as non-conserved , chemosensory-type GPCRs ( csGPCRs ) ( Table 1 ) . The csGPCRs have independently expanded in distinct animal phyla where they serve to respond to diverse , physiologically relevant external and , supposedly , internal cues [4 , 6 , 7] . The genome of the nematode C . elegans encodes an exceptionally large battery of csGPCRs composed of 1 , 341 protein-coding genes ( Table 2 ) [2 , 7 , 8] , a remarkable number given the small size of its nervous system ( 302 neurons constituting 118 anatomically defined neuron classes ) [9] . These csGPCRs have been subdivided by sequence into families and super-families , as summarized in Table 2 [2 , 7] . Wormbase contains expression data for 131 csGPCRs , however for only 76 of them the expression site has been defined with single cell resolution ( S1 Table ) . The majority of these 76 reporters revealed expression in chemosensory neurons [2] . Functional studies have linked a small subset of these receptors to the sensation of specific environmental or pheromonal cues [12–21] , but in the absence of concerted de-orphanization efforts like those seen in other organisms [22 , 23] , the number of receptors with assigned ligands is still remarkably low . Intriguingly , a subset of the previously characterized csGPCR genes were also expressed in non-sensory neurons [2 , 24–28] , suggesting that these csGPCRs may also function as receptors of internal ligands of unknown identity . Providing some hints to the identity of these ligands , one csGPCR subclass , encoded by the srw genes , displays sequence similarities to peptide receptors [11 , 29] . The expression of csGPCRs in interneurons also prompted efforts to identify the function of some of these genes . Even though its ligand remains unknown , AIY-expressed sra-11 was found to be involved in the associative learning paradigm , olfactory imprinting [30] , while sra-13 acts in the vulva to control vulval development , which is affected by food signals [26] . In spite of the relative paucity of known ligands , the previously published expression patterns of csGPCRs provided molecular indicators for a number of intriguing and generally very poorly understood nervous system features: ( 1 ) the expression pattern of the GPCR gene str-2 revealed a left/right asymmetry in the two AWC olfactory neurons [31]; this lateralization phenomenon was later found to be required for olfactory discrimination [32] and spurred a host of studies aimed at revealing how this left/right asymmetry is developmentally programmed [33] . ( 2 ) The expression of several csGPCRs revealed a remarkable plasticity in response to changes in the environment . For example , expression of srd-1 , str-2 and str-3 changes in ASI neurons in response to dauer pheromones [34] , and expression of srh-34 and srh-234 in ADL is different in fed versus starved animals [35] . Using these dynamic reporter gene patterns , mechanisms controlling csGPCR plasticity have been elucidated [35 , 36] . ( 3 ) The csGPCR genes srd-1 , srj-54 , and odr-10 have been found to be expressed in a sexually dimorphic manner in sex-shared sensory neurons , suggesting that sexual identity impinges on sensory perception [2 , 37 , 38] . In this resource paper , we examined the expression of 244 reporter transgenes that monitor expression of previously uncharacterized csGPCR genes . Our explicit goal in this analysis was to ( 1 ) generate more neuronal identity markers , ( 2 ) test the hypothesis that many more sensory neurons may be lateralized , ( 3 ) identify more markers of neuronal plasticity , ( 4 ) identify more markers of sexual dimorphism , and ( 5 ) examine the extent of expression in non-sensory and non-neuronal cells ( suggesting roles as receivers of internal signals ) . Based on the molecular classification of csGPCRs into defined families , we were also interested in determining whether the expression of specific subfamilies—particularly those whose expression has not previously been examined—may reveal specific common themes ( i . e . , patterns of coexpression or expression in specific cells ) that may provide a hint to their function . We synthesize our findings with those of previous expression pattern analyses to carve out a number of general features of csGPCR expression patterns .
Strains were maintained by standard methods [39] . Mutant alleles used in this study were: pha-1 ( e2123 ) [40] , him-5 ( e1490 ) [41] , unc-43 ( n1186lf ) [42] , unc-43 ( n498gf ) [43] , and nsy-5 ( ky634 ) [44] . GFP reporters were generated using a PCR fusion approach [45] and injected without being subcloned . Genomic fragments were fused to the GFP coding sequence , which was followed by the unc-54 3′ untranslated region . A list of primers for all constructs can be found in the Supplementary Methods . Amplicons were injected at 50 ng/μl with the pha-1 rescuing plasmid ( pBX ) as a co-injection marker ( 50 ng/μl ) . Reporters were injected into a pha-1 ( e2123 ) or pha-1 ( e2123 ) ;him-5 ( e1490 ) mutant background strain [40] , resulting in transgenic arrays with little mosaicism . For each construct , two independent lines were scored . Reporter strains provided by the Vancouver Consortium were generated as described [46] . Further details and primer sequences used by the Vancouver Consortium can be found at http://www . gfpworm . org . A list of all reporter strains generated by us or provided by the Vancouver Consortium can be found in the Supplementary Methods . Worms were anesthetized using 100 mM sodium azide ( NaN3 ) and mounted on 5% agarose on glass slides . Images were acquired using an automated fluorescence microscope ( Zeiss , AXIO Imager Z . 2 ) . Acquisition of several z-stack images ( each approximately 1 μm thick ) was performed with the ZEN 2 pro software . Representative images are shown following max-projection of Z-stacks using the maximum intensity projection type . Image reconstruction was performed using Fiji software [47] . Neurons were identified either by labeling subsets of sensory neurons with DiD ( Thermo Fisher Scientific ) or by crossing reporter transgenes with landmark reporter strains in which known neuron classes are labeled with a red fluorescent reporter . For dye filling , worms were washed with M9 , incubated with DiD ( 1:500 ) in M9 for 1 hour at room temperature , washed 3 times with M9 , and plated on agar plates coated with food for 1–3 hours before imaging . Red fluorescent reporter strains used for cell identification are: otIs263[ceh-36p::TagRFP , rol-6 ( su1006 ) ] , vyIs51[str-2p::2xnls::TagRFP; ofm-1p::DsRed] [48] , otIs518[eat-4Fosmid::sl2::mCherry::h2b] [49] , otIs544[cho-1Fosmid::sl2::mCherry::h2b] [50] , otIs564[unc-47Fosmid::sl2::mCherry::h2b] [51] , otIs612[flp-18p::NLG-1::GFP11 , gpa-6p::NLG-1:::GFP1-10 , flp-18p::mCherry , nlp-1p::mCherry] , hdIs30[glr-1p::DsRed] , otIs521[eat-4prom8::tagRFP; ttx-3::gfp] . Clustering was performed on binary expression data from 272 neuron-expressed GPCR reporters for which we had cell ID information . Expression data was from our own analysis and available data from wormbase . org [52] . Only positive neuronal cell ID information per GPCR reporter was included in the binary expression matrix with no distinction between the absence of expression and unknown expression per neuron . Data were clustered using the R pvclust package ( https://cran . r-project . org/web/packages/pvclust/pvclust . pdf ) [53] using the euclidean distance metric with average linkage , bootstrap 1000 , and relative sample size ranging from a proportion of 0 . 5 to 1 . 4 of the original sample size . The relative proportion was incremented by 0 . 1 for each bootstrap resampling . Bootstrap probability ( BP ) value and approximately unbiased p-values ( AUs ) are derived from the multiscale-multistep bootstrap resampling . AU support values >95 indicate well-supported clusters and should be considered when evaluating dendrogram cluster relationships . Alternative distance and linkage methods showed clustering of the PHA , PHB , and ASH neurons in all cases ( 42 out of 84 cases had strong support with AU/BP >95 ) . GPCR upstream intergenic regions and intron lengths were extracted from C . elegans exon coordinates , version WS220 using a python script . Non-coding RNA exons were excluded from the intergenic distance calculations so that intergenic distances represent the nucleotide sequence distance between coding genes . The average intron length per gene was calculated by summing the intron sequence lengths for each gene and dividing by the total number of introns . Average intron lengths for genes with multiple isoforms were calculated for each isoform and then averaged , resulting in 1 average intron length per gene . To analyze GPCR reporter gene expression in dauers , mixed populations of respective strains were allowed to exhaust food for 5–7 days at 20°C . Dauers were isolated from starved plates by treatment with 1% SDS for 30 minutes and imaged within 1–2 hours of isolation . The cellular identity of expression changes in dauers was confirmed with red landmark strains , as mentioned above .
We chose to examine csGPCR expression patterns using gfp-based reporter gene technology , the standard tool of gene expression analysis in C . elegans [3 , 54] . The obvious shortcoming of this technology is that reporter genes may not capture the full cis-regulatory content of the respective GPCR-encoding locus , but as we will describe in more detail below , most GPCR-encoding loci are compact with small intergenic regions and introns . We emphasize that our approach is not necessarily aimed at identifying the complete set of cells expressing a GPCR , but , following ample precedent , is rather aimed at identifying novel and informative patterns of expression , as incomplete as these patterns may be . We utilized two sources of csGPCR reporters . A consortium at the University of British Columbia ( Vancouver ) has generated a valuable , large panel of reporters for 1886 genes in the C . elegans genome [46] . However , the site of expression of these reporters has not been determined with single cell resolution in the nervous system . We obtained 100 reporters from this collection that targeted GPCR loci , and for every reporter that produced a stable pattern of expression we undertook a detailed analysis of their sites of expression in the nervous system . In addition to these 100 reporter genes , we generated 144 of our own reporter genes . We adhered to the following principles in the choice of genes and design of reporters: first , we aimed to cover all 23 classes of chemoreceptor genes defined by Thomas and Robertson [7] ( Table 2 ) . Using phylogenetic trees assembled by Thomas and Robertson , we sampled each gene family evenly , generally avoiding the examination of close sequence paralogues , which we anticipated to reveal similar expression patterns . Our own reporters mostly contain all 5’ intergenic regions fused to gfp and contain , at most , 4 kb of sequence . The rationale behind this choice lies in the overall organization of GPCR loci ( summarized in S1 Fig ) . Eighty-nine percent of the approximately 1 , 300 csGPCR loci contain 5’ intergenic regions of less than 4 kb . We chose all of our samples from this pool , and , hence , all the reporters generated by us capture the full intergenic region . The reporters from the Vancouver consortium contain about 3 kb of 5’ intergenic region , at most [46] . Furthermore , csGPCR loci tend to have small introns ( average size 432 base pairs; almost half of them <200 base pairs; S1 Fig ) , indicating that relatively little cis-regulatory information resides in these introns , which provided the basis for our focus on intergenic regions . For some genes with very short upstream intergenic regions ( less than 500 bp ) we included the first intron ( if this was 300 base pairs or larger ) in order to increase the regulatory space contained in the reporters . The coordinates for all reporter constructs can be found in the Supplementary Material . Sites of expression within the nervous system were determined mainly for those reporters with the most robust expression patterns and was based on stereotyped cell position , cellular and process morphology , and co-labeling with either DiD ( which labels a subset of sensory neurons ) or by crossing with landmark strains in which specific neuron classes are labeled with a red fluorescent protein ( see Material and methods ) . All cell identification was initially done in young adult hermaphrodite animals . As we will describe in detail later , a number of these reporter strains were also subjected to analysis at different stages , under different conditions , and in the two different sexes . In our ensuing description of expression patterns of reporter genes , we summarize the expression observed with the previously described reporters , as well as the additional reporters analyzed by us . All of our expression analysis is summarized in a tabular form in S1 Table . Three overall features of the 375 csGPCR reporters are immediately apparent ( Fig 1 ) : first , 92% of analyzed reporters are expressed in the nervous system; second , expression is not restricted to the nervous system: 33% of the reporters are expressed both within and outside the nervous system and 8% are expressed exclusively in non-neuronal cells; and third , the vast majority of csGPCR reporters are expressed in very restricted numbers of cells ( Fig 1A and 1B ) . Of the neuronally-expressed reporters , 24% are expressed in single neuron pairs , 27% in 2 neuron pairs , 26% in 3–4 neuron pairs , 19% in 5–10 neuron pairs , and the remaining 4% in more than 10 neuron pairs . Expression outside the nervous system will be described in a later section . Within the nervous system , expression is most prominent in sensory neurons ( Fig 1C ) . 84% of the reporters are expressed in amphid sensory neurons ( which are made up of 12 pairs of neurons ) , 20% in phasmid sensory neurons ( made up of 2 pairs of neurons , PHA and PHB ) , and 17% in other sensory neurons . We find that every sensory neuron , except for URY and ADE neurons , expresses at least 1 GPCR ( Fig 1D; S2 Table ) . The number of GPCRs expressed in a given neuron class shows a striking range . The ASI neuron class expresses an impressive 99 GPCR reporters . After ASI , the nociceptive neurons ADL and ASH together with the phasmid neurons PHA and PHB are the sensory neuron classes with higher numbers of GPCRs , expressing 72 , 51 , 51 , and 49 reporters , respectively . Outside the amphid and phasmid neurons , the number of reporters expressed in sensory neurons dramatically drops , with all other sensory neurons expressing less than 10 GPCRs , in some cases only a single GPCR ( Fig 1; S2 Table ) . Of course , it needs to be kept in mind that we only consider expression of a fraction of the csGPCR loci , and hence each of these total numbers is expected to increase by several fold once all csGPCR expression patterns are identified . Twenty-four percent of the GPCR reporters for which we have information about neuron numbers are exclusively expressed in a single neuron class , and in all these cases the neuron class is a sensory neuron class ( Fig 2; S3 Table ) . In total , however , only 9 sensory neuron classes express single-neuron-specific GPCRs . The most striking one of them is the ADL nociceptive neuron class , which expresses 23 single-neuron–specific GPCR reporters ( and an additional 49 GPCR reporters expressed in additional neurons ) . The ADL-expressed , single-neuron–specific GPCRs do not fall into a specific GPCR subfamily but rather cover 7 distinct families . A small subset of the single neuron type-specific GPCRs are expressed outside the nervous system as well ( genes with asterisk in Fig 2A ) . This may indicate that these receptors do not detect external cues , but rather sense internal signals . Notably , expression of the csGPCR reporter collection is clearly not restricted to sensory neurons . A striking 35% of the csGPCR reporters are expressed in inter- and motorneurons ( Figs 1 and 3; Table 3; S1 Table ) . There is no unifying feature of the inter- or motorneurons that express GPCR reporters . They range from ventral cord motor neurons to head interneurons , and to command interneurons in the ventral cord . One interneuron , PVT , displays a very large number of expressed csGPCR reporters ( 57 different reporters ) ; however , PVT expression is generally observed in an unusually large amount of reporter genes and may , like posterior gut expression , be a reporter gene artifact that relies on cryptic regulatory elements in the reporter gene construct . Ninety-seven percent of inter- and/or motorneuron-expressed csGPCR reporters are also expressed in sensory neurons so only 3% of them show expression exclusively in inter- or motorneurons . In light of the inter-/motorneuron expression of csGPCR reporters , we can hypothesize that csGPCR reporters that are expressed in sensory neurons may actually not function as receptors for external sensory cues , but may rather function as they likely do in inter-/motorneurons , i . e . , as receptors of internal signals . We asked whether csGPCR expression profiles cluster by neuron class . To this end , we undertook unsupervised hierarchical clustering of expression profiles . The BP value for most associations was very weak with two exceptions: csGPCR reporters are often coexpressed in the two tail phasmid neuron classes PHA and PHB ( AU/BP > 95 ) , and expression in either or both of the phasmid neurons is associated with the expression in the head neuron ASH ( AU/BP > 95 ) ( Fig 4 ) . The ASH , PHA and PHB neuron classes are not closely related by lineage but all of these three neuron classes are nociceptive neurons that respond to similar cues and integrate sensory inputs from the head and tail [55 , 56] and that directly innervate command interneurons involved in reversal behavior [9] . While csGPCRs expressed in these neurons are likely involved in sensing nociceptive cues , it is notable that these coexpressed csGPCRs came from widely distinct csGPCR families ( Fig 4 ) . One major motivation for undertaking the csGPCR reporter analysis was to identify more lateralized neuron pairs in the nervous system . In vertebrates , there is a striking dearth of molecular correlates for widespread functional lateralization of the brain . In C . elegans , the chance discovery of left/right asymmetric sensory receptor expression has opened up new vistas on lateralization of the C . elegans nervous system [58] . Specifically , the lateralized expression of several csGPCRs in the AWC olfactory neuron pair [31] and guanylyl cyclase receptors in the gustatory ASE neuron pair [59] revealed a common theme of lateralization , providing means of sensory discrimination [32 , 60 , 61] . Since lateralization provides an elegant , straight-forward means for sensory discrimination , we speculated that such lateralization may be widespread in the nervous system and therefore took particular care in examining whether csGPCR reporters that we analyzed are expressed in a left/right asymmetric manner . We indeed identified 8 csGPCR reporters with left/right asymmetric gene expression in an otherwise bilaterally symmetric neuron pair . However , this laterality was only observed in the context of the AWC sensory neuron pair , which was previously known to express several csGPCRs in a left/right asymmetric manner [31 , 62] . Using previously described sets of mutants , we found that the asymmetry of these GPCR reporters is controlled by the same calcium-dependent signaling pathway [33] that controls all other previously known asymmetric GPCRs in the AWC neurons ( Fig 5 ) . Of course , our limited analysis does not exclude the existence of left/right asymmetrically expressed GPCR genes in other neuron classes , but it may not be as widespread as we initially hypothesized . Apart from brain lateralization , another domain of nervous system research displays a striking dearth of molecular markers . While the existence of sex-specific neurons is widely appreciated in the nervous system of most animals , including C . elegans [64] , it is much less clear to what extent neurons that are shared by the two sexes of a given species display molecular differences . Recent anatomical work in C . elegans revealed intriguing synaptic wiring differences between sex-shared neurons in the two sexes [65] , but even in C . elegans there is a dearth of sexually dimorphic molecular markers of sex-shared neurons . Given the distinct priorities that males and hermaphrodites display toward food and mate searching [66] , and given that a number of sex-shared sensory neurons are known to respond to different cues in a sex-specific manner [49 , 67] , we hypothesized that we may discover a multitude of sex-specifically expressed GPCRs . We indeed identified several GPCRs that are expressed in hermaphrodite-specific neurons ( HSNs , VC motor neurons ) or in several male-specific neurons ( Fig 6 ) ; however , we did not detect differences in GPCR expression in sex-shared neurons . We emphasize here , however , that we did not systematically analyze all 244 reporters that we analyzed in the hermaphrodite for differences in expression in the male , but rather focused on those GPCRs that show expression in 1–3 pairs of neurons in the hermaphrodites . Moving outside the nervous system , we found expression of individual GPCRs in essentially all tissue types ( Fig 7 shows examples; summarized in Table 4 ) . As we already mentioned above , the non-neuronal expression is often quite specific and there are only a few GPCRs that are expressed broadly in many different cell types ( e . g . , srbc-58 , srr-4 ) . Specific sites of non-neuronal expression include subsets of muscle cells , hypodermal cells , specialized epithelial cells , cells of the somatic gonad ( distal tip cells ) , individual cells of the excretory system , glial cells , and others ( Fig 7 , Table 4 ) . There are no obvious , specific associations of non-neuronal expression with expression in a specific set of neuron types . Also , non-neuronally expressed GPCR receptors are not biased toward a single subfamily . GPCRs expressed in non-neuronal tissues that are exposed to the environment , e . g . , epidermis , could be involved in sensing external cues , but other non-neuronal cells will rather respond to internal signals . As a cautionary note , we can not presently exclude that non-neuronal expression may be the result of lack of repressor elements in the reporter constructs , but we note that in C . elegans there is presently little evidence for non-neuronal repressor mechanisms restricting gene expression to the nervous system ( e . g . , [68] ) . Do any of the patterns described above cluster with sequence similarity ( i . e . , family membership ) of the receptors ? As described above , specific features of csGPCR expression patterns do not correlate with family membership , but we wanted to pursue this issue further via a more comprehensive analysis of entire chemoreceptor gene families . As defined by sequence analysis [7] , chemoreceptor gene families have very different sizes , ranging from a single gene per family ( srn family ) to 223 genes per family ( srh family ) ( Table 2 ) . We analyzed reporter gene expression patterns of all members of two small families to examine whether there are common themes in their expression patterns , their genomic location , and cis-regulatory control regions . We also analyzed the expression of the one family , the srn family , which only has a single member and is highly conserved in other Caenorhabditis species , to assess whether it may show an unusual expression pattern . However , we find the srn-1 reporter gene to be mainly expressed in amphid sensory neurons , like many other csGPCRs ( Fig 8 ) . The two small families for which we generated and analyzed reporter genes for all family members are the previously uncharacterized srm ( six members ) and srr ( nine members ) . Five out of the six srm family genes are syntenic to other family members ( Fig 8 ) . As these direct genomic adjacencies suggest local gene duplication , we could ask the question whether such local duplications also resulted in duplication of the 5’ cis-regulatory control regions and to what extent such duplicated cis-regulatory control regions retained similar expression profiles . We find that the adjacent srm-1 and srm-2 genes are expressed in a small set of mostly sensory neurons; some of these neurons are the same , others are different . The same theme applies to the adjacent srm-4 , srm-5 , and srm-6 genes . Their 5’ upstream regions direct expression to distinct , but partially overlapping sets of neurons . The srr gene family is composed of nine members . Reporter genes for all members displayed expression in diverse sets of neuron types with no common theme emerging . Outside the nervous system , it is notable that half of the family members are expressed in distinct cell types of the pharynx ( Fig 8 ) , suggesting a role for these genes in sensing food . We also sought to examine dynamic aspects of csGPCR expression . We focused on dynamics that relate to developmental timing and the response to harsh environmental conditions . To facilitate the identification of changes in expression , we focused our analysis on GPCRs that are robustly expressed in the adult in a small number of neurons ( in most cases not more than 1–3 neuron pairs in the head and/or 1–2 neuron pairs in the tail ) . At the first larval ( L1 ) stage , we did not detect any differences in expression in 79 out of 82 examined reporters , compared to adults . Due to the limitations of multicopy array-based fluorescent reporters , moderate intensity changes within a cell type might be difficult to notice and could have been missed . Three reporter genes , srh-11 , sru-48 , and sra-28 , show striking differences in L1 versus adult stages: all three reporter genes show expression in the ASK neuron at the L1 stage , but not at the adult stage ( Fig 9 ) . Additionally , srh-11 is expressed brightly in the ASI neuron at the L1 stage but dimly at the adult stage ( Fig 9 ) . Furthermore , dim expression of srh-11 and sra-28 reporter genes in the tail phasmid PHB and PHA neurons , respectively , is only observed at the L1 stage but not at the adult stage ( Fig 9 ) . We found that a substantial number of csGPCR reporter genes were dynamically expressed when animals enter the dauer stage , an environmentally controlled diapause arrest stage that is accompanied by substantial cell , tissue , and behavioral remodeling [69 , 70] . Initially again focusing on reporters that are expressed in a restricted number of neurons under well-fed conditions , we found that 16 out of 46 examined reporters show a diverse set of changes in animals that were sent into the dauer stage via a standard starvation/crowding protocol ( see Experimental Procedures ) . Many of the changes entail striking changes in the cellular specificity of GPCR reporter expression ( Fig 10 , Table 5 ) . The vast majority of differences are observed in the nervous system , but some changes also occur outside the nervous system . Changes in GPCR reporter expression in the dauer stage have previously been described for two GPCR reporters [34] ( summarized with our novel patterns in Table 5 ) , but the patterns we observe here are much broader and more complex . They can be summarized as follows: Do reporter expression changes observed in dauers recover upon re-feeding to the pattern observed in control-fed animals ? Examining csGPCR reporter expression in well-fed adult animals that had passed through the dauer arrest stage during larval development , we found that the expression of 11 of the 18 reporters , which showed dauer-specific gene expression changes , recovers to that of the fed state , i . e . , in these 11 cases , expression in the adult is independent on whether the animals had passed through the dauer stage or not . For 7 csGPCR reporters we discovered intriguing , cell-type–specific alterations in animals that have passed through the dauer stage ( Fig 11 , Table 5 ) . We observed three types of changes: Note that all of the reporters for which we observe changes in post-dauer recovery do recover their “fed patterns” in other neuron classes ( these could be considered as internal controls that argue against the changes in expression being a reporter gene artefact ) . Taken together , adult animals show neuron-class specific differences in the expression of csGPCR reporters depending on whether they have passed through periods of distress . csGPCR reporters therefore serve as reporters of life history traits . We tested 5 of the 16 csGPCR reporters that displayed changes in the dauer stage for whether their expression also changes in another starvation-induced arrest stage , the starvation-induced L1 arrest stage . Comparing expression in 2 day-starved L1 ( egg prep into M9 medium ) to fed L1 , we find that two reporters ( str-114 and sra-25 ) show the same changes as observed in dauer animals ( Fig 12 ) . In contrast , two reporters ( str-84 and srg-32 ) that change their expression in dauers , do not show changes in starved L1 versus fed L1 ( Fig 12 ) . One reporter , srh-15 , in addition to dauer-specific expression in ASK , is also expressed in ASI in starved L1 . Hence , the response of csGPCR expression to arrest conditions is diverse .
Most csGPCRs show a very restricted expression in few cell types . Many are expressed in single neuron classes . Those csGPCRs that express in multiple neuron classes do not display a coherent set of coexpressing neurons , with one notable exception: the nociceptive ASH , PHA , and PHB neurons express similar ( but not identical ) sets of csGPCRs . Some neurons coexpress a remarkably large number of GPCRs . The ASI neuron displays the most csGPCR genes at 99 , followed by many distinct types of nociceptive neurons . While csGPCRs have been found for all but two sensory neurons ( URY and ADE ) , there is a striking disparity in the number of csGPCRs coexpressed in sensory neuron classes . Amphid sensory neurons clearly coexpress the largest number of csGPCRs , while other sensory neurons express many fewer csGPCRs . The nociceptive ADL stands out in the list of amphid neurons , as it is the neuron expressing the most single-neuron–specific csGPCR reporters . While expression of csGPCRs clearly predominates in sensory neurons , they are also expressed in inter- and motorneurons and in a diverse set of non-neuronal cells . In most cases , each GPCR is restrictively expressed , suggesting that many different cell types in an organism show very distinct and cell-type–specific responses , possibly to internal signals . The similarity of one GPCR family , the srw family , to peptide receptors of other animal species provides hints to the nature of these ligands [11 , 29] . The expression of many members of the srr family in pharyngeal tissues suggests another source of ligands; perhaps these receptors respond to cues from ingested bacteria . In vertebrates , chemosensory GPCRs are now also becoming increasingly appreciated as being expressed in non-neuronal cells [6] . csGPCRs were detected in sensory neurons that are known to express distinct types of sensory receptors and engage in non-chemosensory behavior , e . g . , in gas-sensing neurons or different types of mechanosensory neurons . The expression of csGPCRs in these neuron classes may hint toward these neurons perceiving different sensory inputs , i . e . , they are likely polymodal . However , as discussed above , csGPCRs expressed in these neurons may not be involved in detecting external sensory cues , but measuring internal states . The absence of any overarching expression theme within gene families is striking . We did not observe that the expression of family members clusters in specific neuron classes or share any other specific expression features . Specifically: ( a ) left/right asymmetrically-expressed csGPCRs in the AWC neurons do not fall into the same family; ( b ) csGPCR reporters that are differentially regulated in larval stages or in the dauer stage do not come from a single family; ( c ) csGPCRs that share specific expression pattern themes ( e . g . , coexpression in the nociceptive ASH , PHA , and PHB neurons ) do not derive from specific families; ( d ) non-sensory neuron-expressed or non-neuronal expressed csGPCRs do not fall into a specific family . The only glimpse of perhaps some common function is observed in the small srr family ( nine genes ) , half of which appear to be expressed in non-neuronal pharyngeal tissue . An important note of caution is that these conclusions are based only on reporter genes . However , the substantial sample size on which these conclusions are based lends some credence to these conclusions . csGPCRs generally act as homo- or heterodimers [71] , thereby hugely increasing the amount of distinct sensory receptor complexes expressed in a cell . This combinatorial activity also makes prediction of function of a given csGPCR very difficult in that a csGPCR may have one function expressed in one cell ( in combination with another csGPCR ) , while it may have a very different function in another cell ( in combination with yet another csGPCR ) . While we recovered novel csGPCR genes expressed in a left/right asymmetric manner in the AWC neuron pair , we were surprised to find no other obvious left/right asymmetries in other sensory neuron pairs . Of course , such asymmetries may still be found with currently not analyzed csGPCR genes , but the number of AWC asymmetries recovered suggests that AWC neurons may be exceptional in their extent of lateralization . The only other asymmetry that we found revealed itself not in a sensory , but an interneuron , and only in a non-anticipated context . The sri-9 reporter transgene becomes induced in dauer animals in PVPL , but not PVPR , and PVPL expression is retained in postdauer animals . Molecular asymmetries in PVP neurons have not previously been reported but can perhaps be inferred by the fact that PVPL and PVPR are innervated in a left/right asymmetric manner by unilateral neurons . Specifically , PVPL , but not PVPR , is innervated by the unilateral DVB neuron . Perhaps sri-9 may play a role in this synaptic signaling context , but why this should be dauer-specific is unclear . One notable feature of our analysis was the extent of plasticity that csGPCR reporters show in the context of the dauer stage . Dauer animals are thought to remodel most tissue types and significantly alter behavioral patterns . Changes in csGPCR expression , and hence changes in the external and internal signal perception , fit very well into the mold of organismal plasticity and illustrate the plasticity of many different tissue types ( note , for example , the changes in csGPCR expression in muscle ) . We find it particularly intriguing that several csGPCRs represent markers of life history . Some of the changes in csGPCR reporter gene expression in dauers is retained in post-dauer animals and some csGPCR reporters turn on only in postdauer animals . Animal-wide expression transcriptomic analysis has previously identified large cohorts of transcripts that , like our csGPCR reporters , serve as markers of dauer life history , i . e . , transcripts change in dauers and some of these transcript changes persist in post-dauer animals [72] . However , due to the whole animal nature of this analysis , this previous study lacked cellular resolution . Our findings add a novel spatial component to these previous findings , since we find the life history traits to be strikingly neuron class-specific . The expression of the TRP channel gene osm-9 has also previously been shown to be modulated during dauer and post-dauer stages in a neuron class-specific manner; in this case , osm-9 expression is down-regulated in the ADL ( but not AWA ) chemosensory neurons and the repression is retained post-dauer , using RNAi and chromatin-based mechanisms [73] . In all except one case that we report here , we observe the opposite post-dauer effect; reporters that are turned on in dauers persist in non-dauers . The mechanistic basis of this may hence be distinct from the osm-9 case . It is important to remember here that the life history trait observations are based on transcriptional reporter genes which , on the one hand , may not accurately reflect expression of the endogenous locus , but , on the other hand , clearly provide a definitive molecular “read out” of changes in the “regulatory state” of specific neuron types , depending on whether they have passed through the dauer stage or not . Moreover , our transcriptional reporters also argue that the life history regulatory phenomenon must be transcriptional in nature . These csGPCR reporters will therefore provide excellent starting points to analyze the molecular mechanisms controlling this plasticity . The csGPCR reporter atlas can be put to a number of future uses . The sites of expression of specific csGPCRs point to potential functions of the csGCPRs , guiding the future analysis of csGPCR knockout strains . For example , csGPCRs expressed in the polymodal nociceptive ASH , ADL and phasmid neurons may be mediating the response to a number of distinct sensory cues processed by these neurons [56 , 74] . csGPCR expression patterns point to perhaps unexpected cellular sites of internal signal perception that warrant further investigation . For example , the excretory canal cell expresses at least six csGPCRs reporters ( considering that we only examined reporters for approximately 20% of GPCR loci , this number may increase several fold ) . The relevance of this expression could be tested through the excretory cell-specific expression of dominant negative versions of G-protein downstream signaling components . Similarly , the cellular dynamics in csGPCR expression patterns point to specific cells undergoing changes that warrant future characterization . For example , the induction or suppression of csGPCR reporter expression during the dauer stage in specific sensory and interneurons that were not previously associated with dauer-specific functions may warrant a closer examination of other molecular and functional changes of these neurons during the dauer stage . Because csGPCR reporter fusions also link precisely delineated sequences ( used for reporter construction ) to specific cellular sites of gene expression , patterns of coexpression of GPCRs can be used to extract cis-regulatory information , which in turn may point to trans-acting factors involved in controlling GPCR gene expression . A proof of principle for this type of analysis has already been conducted , pointing to a critical function of , for example , a basic helix-loop-helix ( bHLH ) factor in controlling csGPCR expression in the ADL nociceptive neuron [28] , and with now substantially more expression information available can be further extended to additional cell types . Lastly , green fluorescent protein ( GFP ) reporter transgenes have generally served as invaluable starting points for genetic mutant screens in which the genetic control of specific biological processes can be investigated . The csGPCR reporter collection provides a multitude of entry points . For example , the post-dauer expression of multiple reporter genes can be used to screen for mutants in which these life history traits fail to be properly expressed . GFP reporter genes have also served as invaluable cellular identity markers and here again the csGPCR reporter collection can be used to assess how the identity of specific cell types is genetically controlled . | Maps of gene expression patterns in nervous systems provide an important resource for neuron classification , for functional analysis , and for developmental studies that ask how different neurons acquire their unique identities . By analyzing transgenic GFP reporter strains , we describe here the expression pattern of 244 putative chemosensory receptor-encoding genes , which constitute the largest gene family in Caenorhabditis elegans . As expected , chemoreceptor expression is enriched in chemosensory neurons . Putative chemoreceptors are also expressed in a wide range of interneurons and motor neurons , as well as non-neuronal cells , suggesting that these receptors may sense internal cues in addition to environmental signals . Each chemoreceptor is expressed sparsely , often in just one neuron type , but each neuron type can express many chemoreceptors . Chemoreceptor expression is remarkably plastic , particularly in the context of the environmentally induced dauer diapause stage . Taken together , this molecular atlas of chemosensory receptors provides an entry point for functional studies and offers a host of markers for studying neuronal patterning and plasticity . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"methods",
"and",
"resources",
"nervous",
"system",
"caenorhabditis",
"social",
"sciences",
"reporter",
"genes",
"neuroscience",
"animals",
"animal",
"models",
"caenorhabditis",
"elegans",
"model",
"organisms",
"... | 2018 | An atlas of Caenorhabditis elegans chemoreceptor expression |
During meiosis , the rapid movement of telomeres along the nuclear envelope ( NE ) facilitates pairing/synapsis of homologous chromosomes . In mammals , the mechanical properties of chromosome movement and the cytoskeletal structures responsible for it remain poorly understood . Here , applying an in vivo electroporation ( EP ) technique in live mouse testis , we achieved the quick visualization of telomere , chromosome axis and microtubule organizing center ( MTOC ) movements . For the first time , we defined prophase sub-stages of live spermatocytes morphologically according to GFP-TRF1 and GFP-SCP3 signals . We show that rapid telomere movement and subsequent nuclear rotation persist from leptotene/zygotene to pachytene , and then decline in diplotene stage concomitant with the liberation of SUN1 from telomeres . Further , during bouquet stage , telomeres are constrained near the MTOC , resulting in the transient suppression of telomere mobility and nuclear rotation . MTs are responsible for these movements by forming cable-like structures on the NE , and , probably , by facilitating the rail-tacking movements of telomeres on the MT cables . In contrast , actin regulates the oscillatory changes in nuclear shape . Our data provide the mechanical scheme for meiotic chromosome movement throughout prophase I in mammals .
Meiosis is a specialized cell division for gametogenesis that involves unique chromosomal regulations , such as pairing/synapsis and recombination of homologous chromosomes . These processes are ensured by the dynamic chromosomal rearrangements that occur during meiotic prophase I , as have been extensively characterized in model systems involving Saccharomyces cerevisiae , Schizosaccharomyces pombe and Caenorhabditis elegans [1] , [2] . In these organisms , chromosomes move within the nucleus during meiotic prophase I , which facilitates the juxtaposition of homologous chromosomes and also may dissolve unfavorable entanglements between non homologous chromosomes . To this end , telomeres ( or pairing centers in worm ) are tethered to the nuclear envelope ( NE ) and assemble a conserved transmembrane-protein complex , the LINC-complex ( Linker of Nucleoskeleton and Cytoskeletone ) . The LINC-complex is connected to the cytoskeleton via actin cables in S . cerevisiae and microtubules ( MTs ) in S . pombe and C . elegans , which then facilitates telomere ( or pairing center ) mediated chromosome movements along the NE [3] , [4] , [5] , [6] , [7] , [8] , [9] , [10] , [11] , [12] , [13] . In mammals , early studies reported the characteristic rotational movements of spermatocyte nuclei in rodent species [14] , [15] , [16] . The rotary movements observed in mammalian meiosis had been thought to be the consequence of telomere movements along the NE , as is the case in lower eukaryotes . Indeed , later studies clarified the requirement of the mammalian LINC-complex protein , SUN1 , for normal meiotic progressions in mice [17] . In a previous study , we established an efficient DNA electroporation ( EP ) technique for live mouse testis , which enables rapid genetic manipulations for spermatocytes without the need for genetically engineered mice [18] . 3D time-lapse imaging of pachytene mouse spermatocytes , with visualizations of axial elements and telomeres by EPs of GFP-Scp3 and GFP-Trf1 transgenes respectively , not only confirmed the rotary nuclear movements , but also , for the first time , characterized the concomitant rapid telomere movements on the NE [19] . These superimposed-types of chromosome movements depend totally on MT polymerization activities and also the accumulation of the mammalian LINC-complex , SUN1-KASH5 , to the telomeres under the regulation of a meiosis-specific telomere binding protein , TERB1 [19] , [20] . These results highlight the existence of dynamic chromosome movements driven by telomeres in mammalian meiosis as well , while their mechanical and molecular properties , or stage-specific properties , remain largely unclear . In this study , we optimized the EP conditions and used them to dissect chromosome and microtubule organizing centers ( MTOCs ) movements in each meiotic sub-stage , leptotene/zygotene , bouquet , pachytene and diplotene . Further , we reveal that two cytoskeletal elements , MTs and actin , play different roles in meiotic nuclear dynamics . A combination of live-imaging and fixed cell observations provides insights into MT and MTOC dynamics for the regulation of rapid chromosome movements , supplying the mechanical scheme for chromosome movements during meiotic prophase I in mammals .
We have recently established an efficient DNA EP method for live mouse testes ( Fig . 1A; detailed procedures in S1 , S2 Figures ) [18] , [19] . To optimize the EP efficiency , testes from mice of various ages , 17 , 30 and 60 dpp ( day post-partum ) , were subjected to EP of a Green Fluorescent Protein ( GFP ) expression vector harboring the full-length cDNA of SYCE3 ( synaptonemal complex central element ) ( Fig . 1B–D ) [21] . The majority of germ cells underwent the first wave of spermatogenesis at 17 dpp , and completed meiosis at 30 dpp and spermatogenesis at 60 dpp ( Fig . 1C ) . We obtained reproducibly high EP efficiencies at 17 dpp ( 31% ) and 30 dpp ( 22% ) , but not at 60 dpp ( 3 . 5% ) , although the profiles of the sub-stage distribution of meiotic prophase cells were similar at each point ( Fig . 1D ) . We could detect GFP-SYCE3 expression not only in spermatocytes , but also in mitotic and haploid cells ( spermatogonia , round and elongated spermatids ) as seen in histological sections of testes after EPs ( S3 Figure ) . Transgene expression was increased when the lag time between DNA injection and EP was lengthened to at least 60 min ( Fig . 1E ) . Further , transgene expression was detected as early as 6 hr after EP , and had already peaked at 12 hr ( Fig . 1F ) . Finally , we noticed that the efficiency of transgene expression estimated by immunofluorescence ( IF ) largely depends on the DNA concentration used for injection , with 5 µg being close to saturation ( Fig . 1G ) . Thus , we optimized the in vivo EP method applicable for shot-term transgene expression into mouse testes . To explore the variation in the EP efficiencies of several proteins , we further examined MIS12 ( kinetochore ) , KASH5 ( telomere/NE ) , SCP3 ( axial element ) and RAD21L ( cohesin/axial element ) , and compared the results with those of SYCE3 [18] , [22] , [23] , [24] , [25] , [26] , [27] , [28] . As a result , GFP signals for each protein were detected on specific chromosomal parts as endogenous proteins ( Fig . 1G ) , although the efficiencies varied among the inserted genes . To explore this variation , transgene expressions were analyzed by Western blot ( WB ) ( Fig . 1H ) . Consistent with the results of IF , expressions of GFP-SYCE3 and MIS12 were detected by WB . In contrast , GFP-RAD21L expression was hardly detectable by either WB or IF , suggesting the protein expression/stability is rate-limiting in this case . The expression of GFP-SCP3 , however , was comparable to that of GFP-MIS12 in WB , while the IF of GFP-SCP3 was much weaker . A number of cells with nucleoplasmic GFP-SCP3 signals without axial element localization were observed ( Fig . 1I ) , suggesting that the turnover efficiencies of endogenous proteins might also be important for proper localization . Collectively , these data suggest that in vivo EP is applicable for various transgenes , and that the expression efficiency varies dependent , most likely , on protein stability and turnover . We next examined the functionality of the transgenes in testis . The inner nuclear membrane protein SUN1 is required for the nuclear peripheral distribution of meiotic telomeres , and the disruption of Sun1 prevents homologous pairing/synapsis due to the loss of meiotic chromosome movements [17] , [19] . We confirmed that telomeres , represented by TRF1 foci , were partly detached from the NE in spermatocytes from Sun1−/− mice ( 15 internal telomeres/cell ) ( Fig . 2A ) . Strikingly , the exogenous expression of SUN1-MYC protein by in vivo EP largely restored the telomere attachment defects in Sun1−/− testes ( 1–2 internal telomeres/cell ) ( Fig . 2A ) . To assess the homolog pairing/synapsis states , we quantified the number of TRF1 foci in Sun1−/− spermatocytes . Theoretically , there are 80 and 40 TRF1 foci before and after homolog pairing/synapsis , respectively ( the number of chromosome in mice is 2n = 40 ) . While WT spermatocytes gradually achieved homolog pairing/synapsis from leptotene to pachytene stage , Sun1−/− spermatocytes largely failed to complete homolog pairing/synapsis even in the proceeding zygotene-like stage ( >70 TRF1 foci ) ( Fig . 2B ) . Again , the exogenous expression of the SUN1-MYC protein partly rescued the telomere-pairing defects of Sun1−/− spermatocytes ( 57 TRF1 foci ) ( Fig . 2B ) . Consistently , the subsequent homolog synapsis process , assessed by the loading of a synaptonemal complex protein SCP1 , was also restored by the expression of the SUN1-MYC protein ( Fig . 2C ) . We also performed complementation assays using another meiotic mutant mouse , Rad21l−/− , and confirmed that the expression of GFP-RAD21L restored the pairing/synapsis defects and telomere aggregation phenotypes observed in Rad21l−/− ( S4 Figure ) [25] , [29] . Notably , however , GFP-Sun1 EP failed to rescue Sun1−/− phenotypes at all , even though GFP-SUN1 itself successfully targeted to NE proximal telomeres ( Fig . 2A–B ) . It is assumed that because the N-terminal domain of SUN1 interacts with telomeres through binding to TERB1 [19] , the N terminus GFP-tagging of SUN1 might impair its meiotic functions . Collectively , complementation assays are useful to validate the functionality of tagged proteins and further molecular analysis in vivo . To dissect the meiotic chromosome movements in live spermatocytes , we subjected simultaneous EPs of GFP-Scp3 and GFP-Trf1 transgenes to wild type testis ( 20 dpp ) to visualize chromosome axes and telomeres , respectively , as we demonstrated in the previous study [19] . At 24 hr after EPs , cell suspensions were diluted in Hoechst 33342-containing medium to visualize DNA , the cells were attached to the dishes with Cell-Tak ( BD Bioscience ) to avoid cell movements , and the cells were then subjected to time-lapse analysis . Consistent with our previous results , we can reproducibly observe the rapid movement of chromosomes within pachytene nuclei ( Fig . 3A ) , that comprise two superimposed types of chromosome movement , random telomere movement and unidirectional rotation of the entire nucleus ( rotation is highlighted with trajectories in Fig . 3A ) . Both of these movements were again significantly suppressed by the addition to the medium of nocodazole , an MT-destabilizing drug , ( Fig . 3A , bottom , and S1 , S2 Movies ) , confirming the previously established notion that meiotic chromosome movements in mammals depend totally on the MT polymerization activity , as is the case in S . pombe , C . elegans and perhaps some plant species , but not in S . cerevisiae [12] , [30] , [31] , [32] , [33] . To examine the stage specific properties of chromosome movement throughout meiotic prophase I , we prepared spermatocytes from 14 dpp ( dominantly in leptotene to zygotene ) and 21 dpp ( dominantly in pachytene to diplotene ) male mice after subjecting them to GFP-Trf1 and GFP-Scp3 EPs . Because endogenous TRF1 and SCP3 were intact and spermatocytes expressing GFP-TRF1 and GFP-SCP3 developed normally at least until diplotene stage ( see below ) , we reasoned that the transgenic spermatocytes might behave as wild-type . At 14 dpp , most of the GFP-positive spermatocytes showed faint chromosome axis signals ( represented by GFP-SCP3 ) and more than 40 telomeres ( represented by GFP-TRF1 ) , suggesting that these spermatocytes are in leptotene/zygotene stages ( Fig . 3B , top ) . In contrast , at 21 dpp , there are two types of GFP-positive spermatocytes . One showed about 40 GFP-TRF1 foci with intense GFP-SCP3 signals along the chromosome axes , corresponding to pachytene spermatocytes ( Fig . 3B , middle ) , while the other showed GFP-SCP3 signals less on the chromosome axes ( because of desynapsis ) but more accumulated at the edge of chromosome axes , corresponding to diplotene spermatocytes ( Fig . 3B , bottom ) . Diplotene spermatocytes are also distinguishable by their expanded and rather distorted nuclear shape compared to the earlier stage ( Fig . 3B , bottom ) ( Whole images in S5 Figure and S3 , S4 , S5 Movies ) . First , we observed overall chromosome movement by tracing pericentromeric heterochromatin , regions intensely stained by Hoechst 33342 , and found the presence of unidirectional rotations not only in pachytene but also leptotene/zygotene spermatocytes ( n>10 cells for each stages ) ( Fig . 3B , top and middle ) . In contrast , the rotary chromosome movements almost ceased in diplotene spermatocytes ( n>10 cells ) ( Fig . 3B , bottom ) . Further detailed observations of time-lapse images taken at short intervals ( 7 sec intervals ) allowed us to track identical GFP-TRF1 foci for several continuous time-points and to quantitatively calculate the 3-dimensional telomere velocities ( Fig . 3C and S5 Figure and S3 , S4 , S5 Movie ) . As a result , it is estimated that the average velocities of telomeres are almost the same in leptotene/zygotene ( 0 . 12 µm/sec ) and in pachytene ( 0 . 13 µm/sec ) spermatocytes , while those in diplotene spermatocytes are drastically reduced ( 0 . 083 µm/sec ) ( Fig . 3D ) . The trajectories of GFP-TRF1 foci also confirmed the rapid telomere movement accompanying rotary motion in leptotene/zygotene and pachytene and its diminishment in diplotene stage ( Fig . 3E ) . These observations suggest that both telomere movements and subsequent rotary chromosome movements persist throughout early meiotic prophase I , and are significantly down-regulated in diplotene spermatocytes . While an old observation of rat spermatocytes also hinted at the cessation of nuclear rotation in late prophase I [16] , our results , with GFP-SCP3 and GFP-TRF1 EPs , have , for the first time , defined the precise meiotic sub-stages in live-cells , and verified that chromosome movements operate in a stage-specific manner . To provide mechanistic insight into this phenomenon , we examined the spatiotemporal localization of SUN1 in fixed spermatocytes ( Fig . 3F ) because SUN1 accumulation to meiotic telomeres , regulated by TERB1 , is essential for MT-dependent chromosome movements in mice [17] , [19] . As reported , SUN1 is accumulated to telomeres throughout meiotic prophase I as punctate signals . However , specifically in late pachytene to diplotene stages , SUN1 gradually diminishes from telomeres and aggregates on the nuclear surface . Notably , SUN1 aggregation always occurs near γ-Tubulin signals , representing the position of the MTOC/centrosome ( Fig . 3F ) . This SUN1 aggregation is sensitive to hypotonic and detergent treatments , suggesting it is chromatin-unbound unstable population . Thus , telomere-freed SUN1 might be polarized to the MT-minus end likely through the activity of minus end directed motors such as the dynein-dynactin complex . The down-regulation of telomeric SUN1 pools may explain the reduced chromosome movements observed in late meiotic prophase I in our time-lapse observation . Our time-lapse imaging of diplotene spermatocytes showed that some chromosome ends accompanied intense GFP accumulations and formed clusters on pericentromeric heterochromatin , regions intensely stained by Hoechst 33342 ( Fig . 3B bottom ) . These sites may correspond to centromeres , because previous fixed cell observations also reported the accumulation of SCP3 at short-arm ends , centromeres , and the inter-centromere connections in late diplotene stage [34] , [35] . In our case , spermatocytes express both GFP-SCP3 and GFP-TRF1 , and mouse chromosome is telocentric , the intense GFP signals associating with heterochromatin in diplotene stage may compose both centromere-specific accumulation of GFP-SCP3 and the proximal short-arm GFP-TRF1 foci ( hereafter we refer these signals as GFP-TRF1 for simplicity ) . To dissect the different nature of short-arm and long-arm telomeres in diplotene stage , we traced a pair of GFP-TRF1 foci co-localizing with pericentromeric heterochromatin ( at short-arm ) and not co-localizing with pericentromeric heterochromatin ( at long-arm ) in the same diplotene cells ( Fig . 4A ) . We found that these pairs of GFP-TRF1 foci move in coordination maintaining a distance within 4 . 5 µm , implying that these pairs of GFP-TRF1 foci correspond to telomeres of homologous chromosomes , which are physically connected by a chiasmata ( Fig . 4A , B , C ) . The tracing of GFP-TRF1 foci revealed that a pair of heterochromatin-associated telomeres draws parallel trajectories while heterochromatin-unassociated telomeres draws rather disordered ones , suggesting that the heterochromatin-associated telomeres are more tightly connected and move in greater coordination than the heterochromatin-unassociated ones ( Fig . 4A , B and original images in S5D and S6 Figures ) . The quantification of 3-dimensional distances between a pair of GFP-TRF1 foci further confirmed the stable and close association between heterochromatin-associated telomeres ( <1 µm ) and a rather loose association between heterochromatin-unassociated telomeres ( <4 . 5 µm ) ( Fig . 4C ) . The tight association of telomeres on heterochromatin observed in our time-lapse imaging likely reflects the presence of the inter-centromere connection seen in fixed diplotene cells , which might facilitate the correct co-orientation of homolog centromeres and the subsequent disjunction of homologous chromosomes in anaphase I [34] , [35] . During the course of time-lapse imaging , telomeres and heterochromatin gathered in a limited region on the NE , forming a “bouquet-like” arrangement ( Fig . 5A and S6 Movie ) . The mobilities of clustered telomeres were apparently diminished , while cluster-free telomeres still moved rapidly in the same cell ( Fig . 5A , right ) ( S5B Figure and S8 Movie ) suggesting that telomere movements are locally constrained . Further , the polarized asymmetric distribution of telomeres and heterochromatin in bouquet stage persists throughout the time-lapse analysis , suggesting that the overall rotational movement is also largely suppressed ( Fig . 5A , S5B Figure and S6 and S8 Movies ) . These results contrast with those for S . pombe , in which all telomeres are clustered near the spindle pole body ( SPB/centrosome ) forming bouquet-like arrangements , and the entire nucleus moves rapidly accompanying the bouquet arrangements throughout meiotic prophase I [3] , [31] . To examine the spatiotemporal distribution of the mouse microtubule organizing center ( MTOC ) /centrosome , a cell organelle corresponding to yeast SPB , we carried out immunostaining for γ-Tubulin using fixed spermatocyte samples . As a result , γ-Tubulin appeared as punctate signals localizing to the cytoplasmic side of the nuclear periphery throughout meiotic prophase I and at spindle poles in metaphase I ( Fig . 3F and Fig . 5B ) . In line with observations reported in other model systems [33] , [36] , [37] , , but in contrast to an atypical case in S . pombe [3] , telomeres appear to be randomly distributed on the NE regardless of γ-Tubulin position and transiently assembled near the γ-Tubulin signal only in bouquet stage ( Fig . 5B ) . The abolishment of coordination between MTOC and telomere position in most stages , other than bouquet stage is likely for dynamic chromosome movements accompanying rotary movements of the entire nucleus . In S . pombe , the movement of SPB/MTOC drives the chromosome movements throughout prophase I [3] with accompanying drastic deformation of the entire nuclear shape . In contrast , a study in budding yeast visualized the stable positioning of SPB/MTOC during rapid chromosome movements [33] . To verify the mammalian case in which characteristic rotary movement of the whole nucleus dominates , we visualized MTOC in live pachytene spermatocytes by GFP-γ-Tubulin EP to testes of 17 dpp mice . At 17 dpp , a majority of spermatocytes are in pachytene stage , showing clustered and peripherally distributed heterochromatin ( Fig . 1D and Fig . 5C ) . By following identical heterochromatin during time-lapse imaging ( asterisk in Fig . 5C ) , we traced the unidirectional rotation of chromosomes . GFP-γ-Tubulin showed intense punctate signals at the nuclear surface as observed in fixed cells , suggesting that GFP-γ-Tubulin assembles to the centrosome in the endogenous form ( arrowhead in Fig . 5C ) . Clearly , the position of GFP-γ-Tubulin remains fairly stable during rapid rotational movement of chromosomes ( Fig . 5C and S7 Movie ) , suggesting that the rotational movement is not driven by MTOC movements themselves . This reinforces the aforementioned conclusion that the coordination between MTOC and telomeres is lost during rapid chromosome movements other than at bouquet stage . Although the inhibition of MT polymerization by Nocodazole leads to the almost complete cessation of rapid chromosome movements in mouse spermatocytes ( Fig . 3A ) , there remains a possibility that this is an indirect effect caused by the global destruction of cytoskeletal networks including that of actin followed by MT depolymerization . To address this possibility , we made the same time-lapse observations in the presence of cytochalasin D , an actin depolymerization drug , in the medium . As a result , in contrast to nocodazole addition , we still observed the rapid chromosome movements , including rotary motion of heterochromatin ( Fig . 6A ) and rapid telomere motion ( Fig . 6B ) even in the presence of cytochalasin D , suggesting that the rapid chromosome movements in mammals depends solely on MTs , but not actin ( S9 , S10 , S11 , S12 Movies ) . However , we noticed that the nuclear shape , as visualized by the 2D projective image of Hoechst 33342 signals , was oscillating during the time course even in the presence of nocodazole , but that this was largely suppressed in the presence of cytochalasin D ( Fig . 6A ) . We captured this nuclear oscillation in a quantitative manner as followed: 1 ) convert the nuclear 2D projection into a binary image; 2 ) calculate the binary area at each time point ( AT ) ; 3 ) calculate the change in the binary area at each time ( XOR ( AT , AT+1 ) ) ; 4 ) divide XOR ( AT , AT+1 ) by AT . These quantifications represent the relative change in the nuclear area and demonstrate a drastic reduction in nuclear oscillation in the presence of cytochalasin D , less reduction in the presence of nocodazole , and the further synthetic reduction in the presence of both cytochalasin D and nocodazole , where both nuclear oscillation and rapid chromosome movements are lost ( Fig . 6C ) . These data suggest that actin is required for the oscillation of nuclear shape , but not for the rapid chromosome movements driven by telomeres , for which MTs are largely responsible . We then examined the localization of actin in mouse testis . Fluorescently labeled phalloidin , a chemical that binds specifically to fibrous actin ( F-actin ) , demonstrated F-actin localization on the cell cortex uniformly in zygotene and in a rather polarized manner in pachytene spermatocytes in testis histological sections ( Fig . 6D ) . Staining spread spermatocytes confirmed the same F-actin localization patterns on the cell cortex ( Fig . 6E ) . After hypotonic and permeabilization treatments , a cloudy F-actin signal was also visible in the cytoplasm surrounding the nucleus ( Fig . 6F ) . These actin structures might somehow regulate the oscillation of nuclear shape during meiotic chromosome movements ( see Discussion ) . To obtain mechanical insight into the MT structures responsible for the rapid chromosome movement in mammalian meiosis , we examined the distribution of α-Tubulin by immunostaining fixed-pachytene spermatocytes . Under native staining conditions , α-Tubulin was observed on the cell membrane as densely networked filaments ( Fig . 7A , top ) . After hypotonic treatment and permeabilization with Triton-X100 , we could observe α-Tubulin filaments in the cytoplasm and also at the nuclear periphery ( Fig . 7A , middle ) . In extensively permeabilized cells ( see Materials and Methods ) , the cytoplasmic MTs were mostly washed out while the dense MT cables surrounding the NE remained ( Fig . 7A , bottom ) . These dense MT cables were observed throughout meiotic prophase I ( S7 Figure ) . Intriguingly , telomeres were frequently placed on this MT structure ( Fig . 7B ) . Since SUN1 is required for the stable association between telomeres and the NE ( Fig . 2 ) and the accumulation of MT-dependent cytoplasmic motors to meiotic telomeres [19] , [28] , we hypothesized that the telomere/MT cable interaction might also be mediated by SUN1 . To test this possibility , we examined the distribution of MTs and telomeres in Sun1−/− spermatocytes . Although the cable-like MT structures were present in Sun1−/− spermatocytes as in wild type spermatocytes and a portion of the telomeres still located on the NE , the colocalization of telomeres and MT cables was detectably impaired in the absence of SUN1 ( 74% in WT and 45% in Sun1−/− , Fig . 7B ) . These data suggest that the rapid chromosome movements during meiotic prophase are mediated by the rail tracking movement of telomeres along MT cables , and that they are connected in part by SUN1 ( Fig . 7C ) .
The development of mammalian meiocytes requires the help of supporting somatic cells so that , if meiocytes are isolated , it is technically difficult to culture them long term in vitro [44] , . Most genetic manipulations are suitable for cultured cells and not easily applicable to tissues in vivo because manipulations in tissues require laborious processes , namely , raising genetically engineered mice as done in the current convention . In other biological fields such as neuroscience , however , in vivo gene expression systems by electroporation ( EP ) have been established , and these work as powerful tools for in vivo genetic analysis [46] . In the field of mammalian meiosis , there are a few reports of success in reporter GFP expression , but this expression has never been applied to any practical analysis because of its low efficiency [47] , [48] . To overcome these difficulties , we have established a highly efficient DNA EP method for live mouse testis [18] , [19] . In this study , we further optimized EP efficiency ( Fig . 1 and S1 , S2 Figures ) , and succeeded in demonstrating the short-term expressions of various transgenes , such as GFP-SYCE3 , MIS12 , KASH5 , SCP3 and RAD21L , all of which localize to specific chromosomal parts as endogenous proteins . Our comprehensive assays suggest that DNA concentration , mouse age , and the lag time from DNA injection to EP are critical factors for efficient transgene expression ( Fig . 1 ) . We also demonstrate that transgenes are , in principle , sufficiently functional ( Fig . 2 and S4 Figures ) , suggesting that the EP technique is useful for molecular analysis in vivo . The regulation of chromosome movements in prophase I is one of the most intriguing aspects of meiosis . For the sake of physical access to the correct homolog partners , meiotic chromosomes tethered to the NE through telomeres ( or pairing centers in nematodes ) move rapidly along the NE . The responsible driving forces are diverse among organisms , such as actin in S . cerevisiae , and MTs in S . pombe , C . elegans and probably plants [12] , [13] , [31] , [32] , [33] . Recent studies in mice demonstrated that MTs are responsible for telomere-driven chromosome movement , although the mechanical properties have been poorly understood [18] , [19] , [30] . Here , we applied an in vivo electroporation technique to visualize and comprehensively dissect chromosome movements in different sub-stages during meiotic prophase I in mice ( Fig . 3 ) . We defined leptotene/zygotene , pachytene and diplotene spermatocytes according to GFP-TRF1 and GFP-SCP3 signals in live spermatocytes . Our assays reveal that the rapid chromosome movements that accompany random telomere motion and unidirectional chromosome movement persist through leptotene/zygotene to pachytene , and then diminish in diplotene spermatocytes . This observation is consistent with the earlier findings , which depicted rotational motion of rat spermatocytes and its cessation in late prophase [14] , [15] , [16] . Our current study goes beyond the previous studies by precisely defining meiotic sub-stages in live cells according to GFP-TRF1 and GFP-SCP3 signals , and quantitatively determining telomere motions by tracing GFP-TRF1 signals . Further , we discovered the gradual liberation of the SUN1 protein from telomeres in late prophase , which might explain the reduction in chromosome movements in this stage ( Fig . 3F ) . It is reported that the telomere localization of the meiosis-specific telomere binding protein TERB1 , which is responsible for SUN1 accumulation at telomeres , is down-regulated as well during diplotene , presumably by CDK1-Cyclin B dependent phosphorylation [19] . Thus , the temporal release of SUN1 from meiotic telomeres during diplotene might be the consequence of the loss of TERB1 from the telomeres . Notably , also in C . elegans , chromosome movements drastically decline in late prophase concomitant with the dispersal of the SUN-1 protein from pairing centers , uncovering the conserved molecular regulations defining the stage-specific properties of chromosome movements in different eukaryotic species [12] . In the diplotene stage , homologous chromosomes are physically connected by chiasmata , and the large chromosomal arm regions , including telomeres at long-arms , lose their tight association mediated by the synaptonemal complex . However , earlier studies using fixed cells found that centromeric regions are tightly connected even in late diplotene stage , suggesting there might be some chiasmata-independent homolog associations at centromeres in this stage [34] , [35] . In the current study , we carried out time-lapse imaging of diplotene cells to show that the homolog telomeres at short-arms are more tightly connected and move with greater coordination than at long-arms reflecting the presence of inter-centromere connections in live-cells as well ( Fig . 4 ) . Furthermore , our time-lapse imaging also implies that there might be physical connections between pericentromeric heterochromatin because the short-arm telomeres form several clusters on pericentromeric heterochromatin , and these clusters also seems to move in coordination . The molecular mechanism of the inter-centromere connection remains controversial . Because a portion of the synaptonemal complex is retained specifically at centromeres even after desynapsis in early diplotene stage , the synaptonemal complex may play a role [34] , [35] . However , because the synaptonemal complex is finally dissociated , if not completely , from centromeres in late-diplotene stage , concomitant with the gradual accumulation of axial elements ( SCP3 ) at centromeres , there might be some unknown mechanism to keep homolog centromeres together in this stage . Our time-lapse observations of pachytene spermatocytes revealed unexpected outputs following the inhibition of actin polymerization during chromosome movement ( Fig . 6 ) . Our data suggest that the rapid chromosome movement is essentially regulated by MTs , but that actin also contributes to the oscillation of nuclear shape during movement . Indeed , our cytological observations of testis sections and spread cells identified F-actin localization on both the cell cortex and around the nucleus of spermatocytes; probably either structure or both is responsible for actin-dependent nuclear shape oscillations . Notably , the analogous phenomenon , the reduction in nuclear shape change after actin depolymerization during meiotic chromosome movement , was also found in S . cerevisiae , although in this case the shape change might be an indirect consequence of the reduction in telomere-mediated chromosome movements , which is also under the regulation of actin in this model system [33] , [49] . Though we cannot address the role of actin mediated nuclear oscillation in meiotic progression or chromosomal regulation in this study , further detailed dissection of chromosome movements or analysis of the terminal meiotic phenotype after actin depolymerization will answer this question in the future . Our cytological observations further confirmed that the positions of MTOC and telomeres are mostly irrelevant , and that they are in transient juxtaposition on the nuclear surface only during the bouquet stage ( Fig . 5 ) , which might be a general feature of eukaryotic meiosis other than an atypical case reported in S . pombe [3] , [31] . Because the dynein-dynactin complex localizes to meiotic telomeres in a SUN1-KASH5 dependent manner [19] , [20] , [28] , dynein-dynactin dependent telomere movement toward the minus-end of MTs might drive the transient juxtaposition of telomeres and MTOC in the bouquet stage . However , since the dynein-dynactin complex continuously localizes to meiotic telomeres throughout prophase I , it is still unclear how telomeres are gathered to the MTOC only in the bouquet stage and then released thereafter . It is plausible that counteracting motors , such as MT plus-end directed motors , might also regulate the prophase chromosome movements as in S . pombe [50] , and that the temporal changes in the force balances between these counteracting motors may define the stage specific distribution of meiotic telomeres and chromosome movements . Also , the contribution of the chromosome axis , SCP3 , to the proper exit from bouquet stage is implicated [51] . It is also noteworthy that a transient bouquet configuration has also been observed in plant meiosis where a defined MTOC “centrosome” is absent . Probably , alternative regulations may ensure the bouquet formation without centrosomes , such as the temporal polarization and clustering of multiple MTOCs or the asymmetric redistributions of the NE component as discussed extensively in previous studies [32] , [52] , [53] , [54] . Our live-observations , combined with GFP-γ-Tubulin EP , further demonstrate that the position of GFP-γ-Tubulin ( MTOC ) remains stable even during rapid rotational chromosome movement in pachytene spermatocytes ( Fig . 5C ) . This observation is in line with the case in budding yeast , where a fluorescent-tagged SPB component , SPC42 , shows stable positioning during the chromosome movements [33] . Our assays further reveal that MTs are densely located on the nuclear surface as a cable-like fibrous structure , and that telomeres are frequently placed on the MT cables in a partially SUN1-dependent manner , suggesting that the dynamic chromosome movements are produced by a rail-racking mechanism ( Fig . 7 ) . The presence of MT around meiotic nuclei and the MT organizing activity on the NE surface have also been reported in plants [40] , [54] , [55] , although the cable-like structures were not observed in these reports . In budding yeast , actin instead of MTs forms an intense cable-like structure surrounding meiotic nuclei [33] . These issues implicate the analogous reformations of cytoskeletal networks taking place on the NE for meiotic chromosome movements in multiple eukaryotic species . Collectively , our results obtained using a cutting-edge EPs technique describe for the first time prophase chromosome movements and their responsible cytoskeletal structures comprehensively in mammals , and show that these largely fit the general models observed in a variety of eukaryotic species ( Fig . 7C ) .
Animal experiments were approved by the Institutional Animal Care and Use Committee ( approval #2512 , #2608 ) . The following antibodies were used: rabbit polyclonal antibodies against GFP ( Invitrogen ) , SCP1 ( Abcam ) , SCP3 ( Abcam ) , α-Tubulin ( Abcam ) and γ-Tubulin ( Abcam ) ; mouse polyclonal antibodies against MYC ( MBL ) , SCP3 , TRF1 , SCP1 [19]; and rat polyclonal antibody against SCP3 [19] . F-actin was stained by phalloidin ( life technologies; A12379 ) . Plasmid DNAs ( CAG or CMV promoter ) were extracted from E . coli DH5a by alkaline lysis ( MAXI-KIT; QIAGEN ) and solubilized in HBS buffer ( 2% HEPES , 0 . 8% NaCl , 5 mM KCl , 0 . 7 mM Na2HPO4 , 0 . 1% glucose ) . For efficient electroporation , we prepared a 5 µg/µl DNA solution , which was stored until use at −80°C . For injection ( testis from 17 dpp ICR mice ) , 9 µl of DNA solution was mixed with 1 µl of 0 . 1% FastGREEN ( SIGMA: #F7258 ) . Glass capillaries ( NARISHIGE: Model GD-1 , 1×90 mm 500 pcs ) ( S1A Figure ) were heated and pulled under gravity with a PC-10 puller ( NARISHIGE ) ( S1B Figure ) . To obtain the appropriate sharpness , the tip of the glass capillary was cut ( S1C Figure ) . The appropriate sharpness was about 0 . 05–0 . 1 mm ( S1 Figure ) . Male mice of various ages were anesthetized by an intraperitoneal injection of 0 . 5% pentobarbital sodium salt solution ( NACALAI #26427-14 ) ( S2A Figure ) . The solution was stored at room temperature . The appropriate volume of 0 . 5% pentobarbital sodium salt solution was experimentally estimated to be 12 µl/weight ( g ) . Mice were anesthetized for 1 to 5 hr depending on age . The testes of male mice under anesthesia were pulled from the abdominal cavity ( S2B Figure ) . 9 µl of plasmid DNA solution ( 5 µg/µl ) was injected into the rete testis using a mouth pipette equipped with a glass capillary by breath pressure under a stereomicroscope ( Leica; M165C ) ( S2C-D Figure ) . 1 hr after injection , the testes were held between a pair of tweezer-type electrodes ( BEX: LF650P5 ) ( S2E Figure ) , and electric pulses were applied four times , and then four times in the reverse direction at 30 V for 50 ms at 950 ms intervals per pulse using an electroporator ( BEX: CUY 21 EDIY-TYPE ) ( S2F Figure ) . The testes were then returned to the abdominal cavity , and the abdominal wall and skin were closed with sutures . Immunostaining or time lapse analysis was performed 24 hr after electroporation . In complementation assays using Sun1−/− and Rad21l−/− mice , immunostaining was performed 72 hr after electroporation . For live-imaging of spermatocytes , pCAG-GFP plasmids harboring γ-Tubulin cDNA or a mixture of pCAG-GFP plasmids harboring Trf1 cDNA and Scp3 cDNA were electroporated into wild type testes . After 24 hr , GFP-positive cells were imaged in phenol red-free Leibovitz's L-15 medium ( Gibco ) supplemented with 400 ng/ml Hoechst 33342 ( Wako Chemicals USA ) at 33°C with or without 5 µM nocodazole and 10 µM cytochalasin D . To avoid rotation of the spermatocytes in the medium , dishes were pre-treated with Cell-Tak ( BD Biosciences ) for 30 min before cell spread . Exposures of 0 . 15 sec ( for GFP ) and 0 . 025 sec ( for Hoechst 33342 ) were acquired every 30 or 7 sec using a 100× NA 1 . 40 objective on a microscope ( Olympus lV-X71 Delta Vision ) . Stacks of 7–12 optical sections with 1 µm spacing were acquired . The nuclear area and the telomere-trajectory were analyzed using ImageJ software . Immunostaining of chromosome spreads from spermatocytes was performed based on a previous study with modifications [56] . Briefly , testes were incubated in trypsin-EDTA solution at 37°C for 15 min , and washed briefly in PBS . The trypsinized testes were pipetted repeatedly and centrifuged . The cell pellets were washed several times with PBS . For native staining conditions ( Fig . 6E and Fig . 7A top ) , cells were fixed by adding the same volume of fixation buffer ( 1% PFA ) . For mild extraction ( Fig . 6F and Fig . 7A middle ) , cells were suspended in hypotonic buffer ( 30 mM Tris PH 7 . 5 , 17 mM Tris Sodium Citrate , 5 mM EDTA , 50 mM Sucrose ) for 5 min at room temperature and then fixed by adding the same volume of fixation buffer with detergent ( 1% PFA , with 0 . 1% Triton X-100 ) . For more severe extraction ( Fig . 7A bottom , Fig . 7B and S7 Figure ) , the former hypotonically-treated cells were sedimented and then suspended sequentially in alternative hypotonic buffer ( 200 mM Sucrose ) for 5 min at room temperature , and then fixed by adding the same volume of fixation buffer with detergent . The fixed cell suspensions were placed on slides and air-dried . For immunostaining , the slides were incubated with primary antibodies in PBS containing 3% BSA for 2 hr , and then with Alexa Fluor 488 , 568 , 647 ( Invitrogen ) secondary antibodies ( 1∶1 , 000 dilution ) for 1 hr at room temperature . The slides were washed with PBS , and mounted using VECTASHIELD medium with DAPI ( Vector Laboratories ) . Testes were scraped and digested with 100 µg/ml collagenase and 100 µg/ml DNase I for 15 min at 37°C , and then filtered through a 40 µm cell strainer ( FALCON ) . The testicular cells were fixed in 70% ethanol and brought to a concentration of 1–2×106 cells/ml in propidium iodide/RNase solution ( BD Biosciences 550825 ) . Cells were analyzed by a Becton , Dickinson ( Rutherford , NJ ) FACSort instrument equipped with an argon laser . Images were taken on a microscope ( Olympus IL-X71 Delta Vision; Applied Precision ) equipped with 100×NA 1 . 40 and 60×NA 1 . 42 objectives , a camera ( CoolSNAP HQ; Photometrics ) , and softWoRx 5 . 5 . 5 acquisition software ( Delta Vision ) . Acquired images were processed with Photoshop ( Adobe ) . | Meiosis is a special type of cell division for gametogenesis , errors in which cause several genetic disorders such as infertility and Down syndrome . In meiotic prophase I , chromosomes are tethered to the nuclear envelope ( NE ) through telomeres , and move rapidly along the NE to get homologs aligned and juxtaposed . Following homologous recombination and synapsis , the bivalent chromosome structure is established , which promotes genetic varieties , and also ensures accurate chromosome segregation in following anaphase I . Although there have been extensive studies addressing meiotic chromosome dynamics in yeast and worms , the same in mammalian meiosis remains largely elusive . Here , we utilized an in vivo electroporation ( EP ) technique to visualize chromosome movement in live mouse spermatocytes . We , for the first time , define the meiotic sub-stages in live cells based on telomeres and chromosome axis morphologies , and reveal chromosome movements regulated in a stage-specific manner . Putting the live-observations together with our cytological observations in fixed cells , we propose that meiotic chromosome movements in mammals are mediated by the rail-tracking movement of telomeres along the MT cables surrounding the meiotic nucleus . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biology",
"and",
"life",
"sciences",
"research",
"and",
"analysis",
"methods"
] | 2014 | The Dissection of Meiotic Chromosome Movement in Mice Using an In Vivo Electroporation Technique |
The proteomes of cells , tissues , and organisms reflect active cellular processes and change continuously in response to intracellular and extracellular cues . Deep , quantitative profiling of the proteome , especially if combined with mRNA and metabolite measurements , should provide an unprecedented view of cell state , better revealing functions and interactions of cell components . Molecular diagnostics and biomarker discovery should benefit particularly from the accurate quantification of proteomes , since complex diseases like cancer change protein abundances and modifications . Currently , shotgun mass spectrometry is the primary technology for high-throughput protein identification and quantification; while powerful , it lacks high sensitivity and coverage . We draw parallels with next-generation DNA sequencing and propose a strategy , termed fluorosequencing , for sequencing peptides in a complex protein sample at the level of single molecules . In the proposed approach , millions of individual fluorescently labeled peptides are visualized in parallel , monitoring changing patterns of fluorescence intensity as N-terminal amino acids are sequentially removed , and using the resulting fluorescence signatures ( fluorosequences ) to uniquely identify individual peptides . We introduce a theoretical foundation for fluorosequencing and , by using Monte Carlo computer simulations , we explore its feasibility , anticipate the most likely experimental errors , quantify their potential impact , and discuss the broad potential utility offered by a high-throughput peptide sequencing technology .
The basis of “next-gen” DNA sequencing is the sequencing of large numbers of short reads ( typically 35–500 nucleotides ) in parallel . Currently available next-generation sequencing platforms from Pacific Biosciences [1] and Helicos [2] monitor the sequencing of single DNA molecules using fluorescence microscopy and can allow for approx . one billion sequencing reads per run ( e . g . , for Helicos ) . Unfortunately , no method of similar scale and throughput exists to identify and quantify specific proteins in complex mixtures , representing a critical bottleneck in many biochemical , molecular diagnostic , and biomarker discovery assays . For example , consider the case of cancer biomarker discovery: nucleic acid mutations underlie nearly all cancers . However , these variants are embodied by proteins and are often expressed in bodily compartments ( saliva , blood , urine ) accessible without invasive biopsies . The use of protein biomarkers to diagnose , characterize , and monitor most , if not all , cancers [3] would be significantly advanced by an approach to sensitively identify and quantify proteins in these compartments . Indeed , the value of diagnostic biomarkers is clearly seen in the utility of detecting thyroglobulin for monitoring thyroid cancer , and in administering Herceptin specifically for breast cancers overexpressing HER2/neu [4] . Techniques applied to this problem , including mass spectrometry ( MS ) and antibody arrays , often lack sufficient sensitivity and intrinsic digital quantification to be effective [5] . What is urgently needed is a massively parallel method , akin to next-gen DNA sequencing , for identifying and quantifying individual peptides or proteins in a sample . Unlike the polymerase chain reaction ( PCR ) for nucleic acids , no methods exist for template-directed amplification of proteins . Hence , advances in proteome analysis sensitivity and throughput often focus on enhancing detectors , or even on inferring proteomes based on measuring translating mRNAs [6] . The most sensitive current methods for discriminating specific arbitrary protein sequences , such as mass spectrometry , typically exhibit attomole or femtomole sensitivities [7] , although new detection modalities for detecting specific single molecule sequences are being developed [8] . Nonetheless , current proteomics technologies tend to lag the analogous nucleic acid sequencing technologies by 3 to 5 orders of magnitude in both throughput and sensitivity . Recently , researchers have attempted to identify single protein molecules using nanopores [9 , 10] , although the challenge of deconvoluting electrical signals to sequence peptides is currently unsolved . In principle , extending classic Edman degradation protein sequencing methodology to the single-molecule level could potentially allow billions of distinct peptides to be sequenced in parallel , thereby identifying proteins composing the sample and digitally quantifying them by direct counting of peptides . Edman degradation , first described by Pehr Edman in 1949 [11] , is a standard method to determine the amino acid sequence of a purified peptide , in which the amino-terminal ( N-terminal ) amino acid residue is labeled and cleaved from the peptide without disrupting the peptide bonds between other amino acid residues [12] . In the conventional approach , the freed amino acid is identified ( e . g . , by chromatography ) , and successive cycles of this procedure reveal the sequence of amino acids in the peptide . In this paper , we have tried to anticipate how one might implement a single-molecule sequencing technology based on Edman degradation . We suggest one practical approach that would in principle be capable of generating partial peptide sequences in a highly parallel fashion , scalable to entire proteomes . Beyond biomarker discovery , such an approach would have broad applications across biology and medicine and could be as fundamental for proteins as , for example , PCR is for nucleic acid research . From a theoretical perspective , we discuss the many interesting features that data generated by such an approach would have , along with how such data might be interpreted and how sensitive the process might be to potential errors , which we model using Monte Carlo simulations .
Fig . 1 illustrates a proposed scheme for single molecule peptide sequencing . The key idea is to selectively fluorescently label amino acids on immobilized peptides , followed by successive cycles of removing peptides’ N-terminal residues ( by Edman degradation ) and imaging the corresponding decreases of fluorescence intensity for individual peptide molecules . The resulting stair-step patterns fluorescence decreases will often be sufficiently reflective of their sequences to allow unique identification of the peptides by comparison to a reference proteome . In more detail , proteins in a complex mixture are first proteolytically digested into peptides using an endo-peptidase of known cleavage specificity . Select amino acid types ( e . g . lysine , tryptophan or tyrosine ) are covalently labeled with spectrally distinguishable fluorophores , each being specific ( by reactivity ) to the given amino acid side chain . Labeled peptides are immobilized on a glass surface , as for example via the formation of a stable thioether linkage between a maleimide functionalized surface and the thiol group on cysteine residues [13] . The choice of peptidase , labeled amino acids , and anchor all convey information about the identity of a peptide and thus can be optimized for maximum effect . Using techniques such as Total Internal Reflection Fluorescence ( TIRF ) microscopy , individual peptide molecules can be imaged on such a surface , and the fluorescence intensity across all fluorophore channels can be determined for each peptide on a molecule-by-molecule basis . By monitoring decreases in fluorescence intensity following cycles of Edman degradation , we can determine the relative positions of labeled amino acids in the peptides , and thereby obtain a partial peptide sequence . This scheme might be improved by using a fluorescent Edman reagent whose coupling and decoupling can be observed , enabling the successful completion of each Edman cycle to be monitored for every single peptide , providing an additional error check . We term the pairing of an Edman degradation cycle and the subsequent observation for changes in fluorescence an experimental cycle ( see Box 1 for definitions of terminology ) . The observed sequence of luminosity drops in fluorescence across experimental cycles is a fluorosequence; the technique itself is thus fluorosequencing . For the example shown in Fig . 1 , the fluorosequence is “WKKxY” . Mapping the partial sequence back to a reference proteome of potential proteins , such as might be derived from a genome sequence , would determine if the fluorosequence uniquely identifies a peptide , and ultimately , its parent protein . Commercially available TIRF microscopes can easily monitor fluorescence changes for millions of individual peptide molecules [14] and are not dissimilar to early variants of next-generation DNA sequencers [2] . By increasing peptide density and acquiring TIRF images over a large surface area , one could in principle obtain fluorosequences for millions or billions of peptides in parallel . Critically , this approach would be intrinsically quantitative and digital , based on counting repeat peptide observations , in much the same way NextGen RNA sequencing is for identifying and quantifying RNA transcripts . Computer simulations of variations of this scheme confirm that fluorosequences can be quite information-rich; even relatively simple labeling schemes , employing only 1 to 4 amino acid-specific fluorescent labels , can yield patterns capable of uniquely identifying at least one peptide from most of the known human proteins ( Fig . 2 ) . For these simulations , we considered only labeling schemes based on known differences in side-chain reactivity and available amino acid-specific targeting chemistry [15] , such as the reactivity of diazonium groups for tyrosines [16] . Many of the above labeling schemes ( anchoring peptides via internal cysteine residues ) fail to achieve 100% coverage of the template proteome even after many experimental cycles under ideal conditions . The reason is two-fold: ( a ) Edman reactions cannot continue past the cysteine anchor or ( b ) the proteome contains paralogs and protein families differing at unlabeled amino acids that are hence indistinguishable . When simulations were repeated for the case of anchoring all cyanogen bromide cleaved peptides , not just cysteine-containing ones , by their C-termini , the coverage of the four-label scheme rose from 80% to 98% of the proteome ( Fig . 2 , top curve ) . Moreover , when simulations were performed for the case of no proteolysis and anchoring each full length protein at its C-terminus , four of the tested multiple-label schemes ( including schemes with only 2 label types ) achieved over 96% coverage of the proteome within 200 experimental cycles . The remaining proteins were unidentified due to protein families being indistinguishable by the labeling schemes employed . These simulations thus confirm that single molecule fluorosequencing is intrinsically capable of identifying a majority of proteins in a proteome even when the number of label types is small . It is also worth considering whether the linear scaling and dynamic range of photon detection by existing cameras might place a limit on the ability to discriminate luminosity drops in fluorescent intensity per peptide . For example , while it might be easy to discriminate a reduction from 5 to 4 fluorophores on a peptide , discriminating a reduction from 25 to 24 fluorophores could be difficult . However , the median count of labelable amino acids per peptide is often small . For example , when considering peptides generated by the protease GluC , this count ranges from approximately 2 ( for lysines ) to 7 ( for glutamic acid/aspartic acid residues , which we assume are indistinguishable by reactivity for labeling purposes ) ( Fig . 3 ) . This range is well within the capacity of most modern cameras , since , in practice , TIRF microscopes equipped with CCD camera variants can count up to at least 13 fluorophores; that is , up to at least 13 copies of a given fluorophore per single molecule can be quantitatively distinguished [17] . Thus , peptides from typical proteomes should not be problematic in this regard . Being a physico-chemical process , we can anticipate some of the most likely potential sources of error for an experimental implementation of the scheme . With errors , an observed fluorosequence would not reflect the true sequence of fluorescently labeled amino acids . Three of the most probable error sources are as follows: ( a ) Failure of fluorophore attachment or emission causing apparent substitutions . Steric constraints of peptides or reaction kinetics of fluorophore labeling chemistry might result in specific amino acid ( s ) not being covalently labeled . This scenario is equivalent to correctly coupled but non-emitting fluorophores , such as those observed in defective fluorophores [18] . In both circumstances , the position of a labelable amino acid would be misinterpreted as containing a non-labelable amino acid , e . g . the peptide “GK*EGK*” ( where K* represents a labeled lysine ) would mistakenly yield a fluorosequence “xxxxK” instead of “xKxxK” , for a dye failure at the first lysine . ( b ) Photobleaching of labeled fluorophores causing apparent coupled double substitutions ( “residue swaps” ) . The permanent photochemical destruction of dyes could also complicate the analysis . In this scenario , a labeled residue at one position is misinterpreted as an unlabeled residue because the label is lost by photobleaching , while another residue upstream in the peptide ( typically unlabeled ) is misinterpreted as being labeled because the photobleaching fluorophore loss coincides with that particular experimental cycle . This would shift the apparent position of the label upstream in the fluorosequence . For example , peptide GK*EGK* might be observed as xKKxx when the dye on the lysine at the fifth position photobleaches during the third imaging cycle . This situation reduces the ability to ( i ) reliably count the number of fluors lost during an experimental cycle , ( ii ) distinguish whether a change in luminosity results from fluorophore loss due to a genuine Edman degradation step or photobleaching , and ( iii ) identify which downstream fluorophore was extinguished if the loss is indeed due to photobleaching . Although fluorophore half-lives can be extended by use of oxygen scavenging systems [19] , synthesis of stable dyes [20] or even surface modification [21] , photobleaching is still a stochastic process and accounting for loss of fluorophores erroneously coincident with upstream Edman degradations would be critical to identification . Currently , there are many photo-stable dyes on the market . A recent study on the effects on dyes by oxygen radicals found that the half-life of Atto647 was roughly 3 minutes ( corresponding to 180 experimental cycles at 1 second/cycle exposure ) [22] , while Atto655 showed a mean photobleaching lifetime of 8–20 minutes [23] , corresponding to many hundreds of experimental cycles . ( c ) Inefficiency of Edman degradation chemistry causing apparent insertions . Optimization of Edman degradation over the past sixty years has resulted in efficiencies of >95% [24] . Nonetheless , failed cycles are expected at some non-zero rate and would yield an observation corresponding to no fluorescence change , even if there was a labeled amino acid in position to be removed . This corresponds to an apparent insertion of a non-labeled amino acid into the fluorosequence . Note that the use of a fluorescing Edman reagent ( e . g . , DABITC or FITC [25] ) would enable direct monitoring of every coupling and decoupling step of the chemistry , providing an internal error check for successful completion of the Edman cycle as in Fig . 1 . Nonetheless , non-fluorescent Edman reagents such as phenylisothiocyanate are much more commonly used , and we therefore investigated the effect of this parameter . To analyze how peptide sequencing efficiency is affected by the above three types of errors and to map fluorosequences to source proteins , we developed a modeling framework to simulate the process . Unlike the ideal case where fluorosequences are faithful to their source peptides , and hence mapping to the reference proteome is trivial , accounting for errors such as the three previously highlighted complicates mapping . For example , the fluorosequence “xKxxK” cannot be uniquely attributed to the “GK*EGK*” peptide , since Edman failure at the first position of peptide “K*EGK*” or a fluorophore failure on the first lysine of “K*K*EGK*” could also yield the same pattern . While errors arising from the inefficiency of Edman chemistry and fluorophore failure are tractable by analytical solutions , the non-Markovian nature of photobleaching events forces us to employ a Monte Carlo approach . We therefore developed a Monte Carlo procedure to simulate thousands of copies of each of the 20 , 252 proteins in the human proteome being subjected in silico to fluorosequencing in order to obtain a random sample of the fluorosequences produced for a specified set of error rates . Fig . 4 details the simulation steps; the Methods provide more complete descriptions of the error models and pseudo-code for the overall procedure . Each sample observation generated by the Monte Carlo simulation is a sequence of luminosity drops yielded by one individual peptide subjected to in silico Edman cycles . We conservatively assume that we cannot observe or infer the absolute number of fluorophores labeling a peptide , but that we can monitor and statistically discriminate whether , after each attempted Edman cycle , there has been a decrease in luminosity in each fluorescent channel , consistent with signals previously shown to be discernable for single molecules [17] . For the purpose of the simulation , we make the simplifying assumptions that different fluorophores have fully distinguishable signals , do not exhibit fluor-to-fluor interactions or Förster resonance energy transfer , nor exhibit channel bleed-over . The fluorosequences ( observed reads ) from the simulations are next collated into a prefix trie [26] , as illustrated for a simple example in Fig . 5 . Each fluorosequence is linked in the trie to its source protein ( s ) and associated count ( s ) of observations over the course of the simulation , thereby empirically estimating the fluorosequence’s source protein probability distribution . Fig . 6 illustrates two extreme cases of protein probability distributions for a given fluorosequence . Importantly , modeling the frequency of source proteins for fluorosequences is equivalent to obtaining ( within sample error ) the posterior probability mass functions—i . e . the set of probabilities P[pj|fi] such that given an observed fluorosequence fi , the probability that protein pj is its source ( henceforth called the attribution probability mass function ( p . m . f . ) ) . Notably , by sidestepping problems associated with developing algorithms for inverting fluoro sequences to their source peptides , and the peptides’ own derivation from source proteins , we make the strategy amenable for incorporating additional experimental parameters , including fluorophore spectral channel bleed-over or protease inefficiencies . Thus , the attribution p . m . f . ’s provide a natural framework both for modeling errors and for directly mapping actual experimentally observed fluorosequences to proteins in the proteome . Based on the properties of this distribution , a fluorosequence can be associated with the protein most likely to yield it , for example applying a confidence threshold ( see Methods ) . In future applications using attribution p . m . f . ’s to interpret fluorosequencing data from real samples , one might also wish to model realistic numbers of copies per protein processed through the simulation pipeline , since the Monte-Carlo based deconvolution of fluorosequences to source proteins will be affected by protein abundance dynamic range as well as simulation depth . For example , high simulation depth would not only reduce the sampling errors , but also accurately attribute low abundance proteins from confounding high abundance proteins that generate the same fluorosequence by a low probability event . Although we did not explore this aspect , simulating protein copies based on their prior known abundances [27] might significantly reduce Monte-Carlo simulation computational resources . The version of the simulation we performed here makes no such assumptions about protein abundance , and thus corresponds to a Bayesian flat prior expectation on protein abundance , applicable to any sample . Using the Monte Carlo scheme , we simulated sequencing the human proteome to a simulation depth of 10 , 000 copies per protein , performing a parametric sweep of 216 experimental parameter combinations ( corresponding to six values for each of the three error parameters ) . Fig . 7 illustrates the effects for three alternate labeling schemes of varying Edman efficiency and fluorophore half-life on the percentage of proteins identified after 30 Edman cycles , given fluorophore failure rates ranging between 0 and 25% . As in Fig . 2 , diversifying the labels offers the greatest improvement in proteome coverage , even with relatively poor process efficiencies . Encouragingly , the number of proteins identified is reasonably robust to changes in fluorophore failure rates . For example , a 25% increase in failure rate causes only a 0 . 8%-6 . 4% reduction ( range includes all parameter combinations ) in proteome coverage for schemes B and C ( see Fig . 7 for scheme descriptions ) . However for scheme A , a 25% increase in fluorophore failure rate causes a 19% reduction in proteome coverage under moderate estimates of photobleaching and Edman efficiency . Scheme A is less robust vis-à-vis all simulated errors because the boost in the positional information stemming from abundant aspartates and glutamates is rapidly undermined by experimental errors , as there are higher chances for fluorescently indistinguishable peptides to confound the fluorosequence . Notably , the photobleaching half-life has the greatest effect of any of the tested parameters on protein identification , causing up to 50% loss in proteome coverage ( under scheme A ) . The steepest decrease in the number of proteins identified occurs when photobleaching is considered ( comparing half-lives of infinity to 210 cycles ) and tapers with lower half-life . Although photobleaching shows the strongest impact of any of the errors considered , it is worth noting that the half-lives of commercially-available fluorophores are sufficiently longer than those simulated . Hence , we anticipate that this error source will not derail a real implementation of fluorosequencing . For example , the widely used Atto680 dye has a mean photobleaching lifetime of about 30 minutes [23] , corresponding to 1800 Edman cycles , assuming 1 second exposure per Edman cycle . Oxygen-scavenging systems are also widely used in single molecule imaging experiments to reduce the effects of photobleaching [19] . Thus , the most critical error rates appear to fall within acceptable ranges , supporting the feasibility of fluorosequencing . Fluorosequencing relies on the positional information of specific subsets of amino acids within peptide sequences . The scheme can be generalized as a framework fulfilling two conditions— ( a ) an observable event ‘e’ , which occurs by detection of a known single amino acid or a class of amino acids , and ( b ) a sequential analytical process , which increments or decrements the sequence in a known direction and by constrained number of amino acids . While we have suggested using detection of fluorescently labeled amino acids as the event , other modalities might be considered , such as detecting voltage changes or reactivity of monitored amino acids . Besides Edman degradation , other valid sequential processes could include sequential treatment with known sequence specific peptidases or directional protein translocation through a nanopore channel [9] at a defined translocation rate . The monitoring of sequenced detection events gives information-rich patterns ( such as “x-e-e-x…” where ‘x’ is one or more non-identifiable amino acids ) capable of being mapped back to a reference proteome . The nature of this information lies between the extremes of information content , wherein either every amino acid corresponds to a distinct event or there is no observable event associated with the process ( as , for example , a peptide translocating through a channel but not generating a detectable signal ) . In principle , many event-process strategies might be suitable for peptide sequencing and interpretation using a scheme similar to the one we present . We propose a strategy for the parallel identification of proteins in a complex mixture based on the positional information of amino acids in peptides . The integration of a 60-year-old , highly optimized Edman chemistry [11] with recent advances in single-molecule microscopy [28] and stable synthetic fluorophore chemistry [29] makes this strategy particularly amenable for experimental execution in the near future . Modeling of experimental errors suggests this strategy can be reasonably expected to identify a high percentage of the proteome , comparable to mass spectrometry , and potentially brings the advantages of single molecule sensitivity and—if next-generation single molecule sequencing is a reasonable proxy—throughputs of hundreds of millions or billions of molecules sequenced per run . Monte-Carlo simulations provide a framework to accommodate the inevitable experimental errors and probabilistically identify proteins from the observed fluorescent patterns . Successful experimental execution of the proposed strategy will not only lead to progress in proteomics , but enable progress in engineering and chemistry to enable the technology .
The UniProtKB/Swiss-Prot complete H . sapiens proteome ( manually reviewed ) was downloaded on 29th May 2013 and used for all simulations , comprising 20 , 252 protein sequences and ignoring alternatively spliced isoforms . Simulations were programmed in Python using Mersenne Twister [28] as the source of randomness , and implemented in parallel using the Texas Advanced Computing Center . For the purposes of simulation , the proteome can be considered dictionary pairs of protein identifiers and amino acid sequences . We began the simulations with 10 , 000 copies of each protein sequence . The first two steps in the simulation split each amino acid sequence string at residue ( s ) corresponding to the protease specificity ( e . g . E for the GluC protease ) and then discard substrings that lack the anchor residue ( e . g . substrings not containing C ) . Alternating Edman degradation steps and TIRF observations on the resulting peptides provide temporal ordering for luminosity drops , resulting in an observed fluorosequence for each peptide . In the simulation , fluorosequences were initialized from amino acid substrings’ correct fluorophore positions , and experimental errors were then introduced sequentially , modifying the fluorosequences in accordance with each type of error’s appropriate probability distribution . The three experimental sources of error sources were modeled in the Monte Carlo simulation as follows: Inefficient dye labeling—The probability of an amino acid not being labeled with its intended label or being labeled with a nonfunctional dye ( i . e . a dye that attaches but is incapable of fluorescence ) is modeled as a Bernoulli variable . For each label prepared for the experimental procedure , there is a probability u that the fluor will never be observed . Edman degradation is represented as an attempt to remove one amino acid residue per cycle . These attempts are modeled as a Bernoulli process , since every experimental cycle is independent of the preceding cycle . The probability of the N-terminus amino acid being successfully cleaved off is assigned a parameter p and the corresponding failure follows as q = 1-p . Failure of Edman chemistry delays the removal of a downstream labeled amino acid by one experimental cycle , and thus dilates the inter-label intervals in the fluorosequence . Using this model , the probability that an inter-label interval d requires d+e experimental cycles before the subsequent label is removed is d-1+ed-1pdqe . A random number is drawn from this distribution to indicate the dilation for each interval . We assume that Edman chemistry stops at the first cysteine from the N-terminus . Photobleaching is the irreversible photo-induced destruction of a fluorophore . The photobleaching process can be best described as a stochastic phenomenon and modeled by an exponential decay function [30] . Every fluorophore has a defined half-life based on solvent conditions and laser operating conditions [31] . The periodic laser excitation has an additive effect on the fluorophore’s half-life: exciting a fluorophore once for thirty seconds and , after an arbitrary delay , again for a further thirty seconds will photobleach the fluorophore with the same probability as a continuous excitation for one minute . We assume a constant period of laser exposure per experimental cycle . To model whether labeled amino acids have been cleaved , the probability of a fluorophore still on the peptide surviving k experimental cycles can be modeled as an exponential decay e-bk , where b is an experimentally-determined characteristic constant of the fluor being used , k is the number of experimental cycles performed , and e is Euler's constant . We shift labels to earlier experimental cycles based on random numbers drawn from this exponential decay . For a given simulation , all simulated fluorosequences were collated into a prefix trie whose keys were the sequences of luminosity drops and associated values represented the counts of source proteins yielding those fluorosequences . One trie was generated for each given choice of error rates , protease and labels , based upon simulating 30 Edman cycles of fluorosequencing 10 , 000 copies of each protein in the human proteome . For each fluorosequence in the resulting trie , its source proteins were counted , allowing proteome coverage to be calculated . The simulation can be summarized as pseudo-code: INITIALIZE result trie as an empty prefix trie . FOR protein IN proteome: peptides ← Proteolyse protein at the carboxyl side of a given amino acid corresponding to the protease used . FOR peptide IN peptides: Discard peptide if it does not contain at least one occurrence of the amino acid for anchoring to the surface . FOR peptide IN peptides REPEAT 10000 TIMES: Attach labels to amino acids with a given probability . Labeling probability is uniform and mutually independent for all amino acids . Adjust positions of labeled amino acids to reflect possible Edman failures . All Edman reactions for each individual peptide have a uniform probability of success specified by a given parameter , and are mutually independent . We assume the Edman reaction cannot proceed past the first amino acid anchored to the surface . Adjust positions of labeled amino acids to reflect potential photobleaching . All fluors’ survival functions are mutually independent exponential decays characterized by a given photobleaching constant . Collate final sequence of tuples ( fluorosequence ) for this peptide into the result trie . TRAVERSE THE TRIE . For each node , find the most frequent source protein yielding that fluorosequence . For the purposes of data visualization , if the most frequent protein yielded the fluorosequence at least ten times , and all other source proteins for that fluorosequence combined are responsible for less than 10% of all observations , then that fluorosequence is considered to be uniquely attributed to the protein . RETURN the set of proteins that have at least one uniquely attributed fluorosequence . More detailed pseudo-code is also provided in the supporting S1 Text . A parameter sweep was performed for the three labeling schemes as in Fig . 7 at a simulation depth of 104 copies per protein , sweeping 216 experimental parameter combinations ( testing six values for each of the three error parameters described ) spanning fluorophore failure rates of 0% , to 25% , photobleaching half-lives from 90 minutes to infinity ( i . e . , no photobleaching ) , and Edman degradation efficiencies from 90% to 100% . For more efficient use of computer memory , trie structures were calculated separately for multiple subsets of the proteome and the resulting tries merged before analysis by traversing all fluorosequences in each trie and adding each fluorosequence along with its protein counts into a master trie for that simulation . Then , the counts of each fluorosequence and affiliated peptides were analyzed to calculate a frequency distribution of the number of times peptides from a given source protein generated a given fluorosequence . For the purposes of summarizing the data , two criteria were applied to this distribution to attribute a fluorosequence uniquely to the protein: ( a ) its primary source protein yielded the fluorosequence at least 10 times out of a 104 simulation depth , and ( b ) the summation of frequency from all other source proteins were responsible for less than 10% of that fluorosequence’s occurrences . While the former criterion addresses sample error , the latter addresses confounding from other proteins . The Monte Carlo simulation Python script and C module can be accessed from github: https://github . com/marcottelab/FluorosequencingSimulation . git | The development of next-generation DNA and RNA sequencing methods has transformed biology , with current platforms generating >1 billion sequencing reads per run . Unfortunately , no method of similar scale and throughput exists to identify and quantify specific proteins in complex mixtures , representing a critical bottleneck in many biochemical and molecular diagnostic assays . What is urgently needed is a massively parallel method , akin to next-gen DNA sequencing , for identifying and quantifying peptides or proteins in a sample . In principle , single-molecule peptide sequencing could achieve this goal , allowing billions of distinct peptides to be sequenced in parallel and thereby identifying proteins composing the sample and digitally quantifying them by direct counting of peptides . Here , we discuss theoretical considerations of single molecule peptide sequencing , suggest one possible experimental strategy , and , using computer simulations , characterize the potential utility and unusual properties of this future proteomics technology . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] | [] | 2015 | A Theoretical Justification for Single Molecule Peptide Sequencing |
Our work focuses on the stability , resilience , and response to perturbation of the bacterial communities in the human gut . Informative flash flood-like disturbances that eliminate most gastrointestinal biomass can be induced using a clinically-relevant iso-osmotic agent . We designed and executed such a disturbance in human volunteers using a dense longitudinal sampling scheme extending before and after induced diarrhea . This experiment has enabled a careful multidomain analysis of a controlled perturbation of the human gut microbiota with a new level of resolution . These new longitudinal multidomain data were analyzed using recently developed statistical methods that demonstrate improvements over current practices . By imposing sparsity constraints we have enhanced the interpretability of the analyses and by employing a new adaptive generalized principal components analysis , incorporated modulated phylogenetic information and enhanced interpretation through scoring of the portions of the tree most influenced by the perturbation . Our analyses leverage the taxa-sample duality in the data to show how the gut microbiota recovers following this perturbation . Through a holistic approach that integrates phylogenetic , metagenomic and abundance information , we elucidate patterns of taxonomic and functional change that characterize the community recovery process across individuals . We provide complete code and illustrations of new sparse statistical methods for high-dimensional , longitudinal multidomain data that provide greater interpretability than existing methods .
The complex , dynamic microbial communities of the human body play essential roles in health and disease . For example , the human gut microbiota contributes to digestion , defense against pathogens , biosynthesis of essential molecules , metabolic homeostasis , and regulation of the immune system [1–3] , but has also been implicated in malnutrition , obesity , diabetes , heart disease , cancer , and autoimmune diseases [4–10] . To maintain or restore healthy states , we must better understand the nature and basis of stability in the gut microbiota , under normal and perturbed conditions . Stability , resilience , and response to perturbation are central topics in community ecology [11] . Extreme perturbations of a system , such as near-complete loss of biomass , are studied both to reveal factors that influence community structure , and as important phenomena in their own right . For example , Fisher et al . ( 1982 ) [12] examined the response of a desert creek ecosystem to flash flooding , with results that matched some but not all of Odum’s theoretical expectations about ecological succession [13] . Particular findings included a return to baseline values of community-wide measures such as diversity indices even while individual taxa were continuing to recover from the disturbance . They also found that the specific characteristics of organisms ( e . g . , the rapid post-flood emergence of motile diatoms buried in sediment , the existence of a nonaquatic adult dipteran stage that was not vulnerable to washout ) influenced community composition during recovery in ways that were not evident from the study of unperturbed intervals [12] . As part of an ongoing study of human microbiota stability and resilience , we created a flash flood-like disturbance in the human gut by inducing acute , transient non-inflammatory diarrhea using a common clinically-relevant iso-osmotic agent , thereby eliminating the vast majority of gastrointestinal biomass . Induced , iso-osmotic diarrhea ( IIOD ) differs qualitatively from the less extreme and more selective , inhibitory and stimulatory action of antibiotics [14] and of diet supplementation [15 , 16] that have more frequently been investigated as disturbances of the gut microbiota . Comparison of different types of perturbation is necessary to understand whether the traits of organisms or communities that affect resilience are specific to each type of perturbation , or act more generally . In addition , understanding the effects of IIOD on the gut microbiota has practical importance because it is a common clinical procedure ( approximately 14 million persons in the United States were subjected to this disturbance in 2013 as preparation for colonoscopy [17] ) . Furthermore , studying the effects of diarrhea per se on the gut microbiota is relevant for our understanding of infectious diarrheal disease , which remains a major cause of mortality worldwide [18] . Several previous studies have investigated the effects of induced diarrhea on the human gut microbiota using 16S rRNA gene surveys that provide a more complete representation of the community than the older culture-based techniques [19] . Most studies recruited participants who experienced both induced diarrhea and colonoscopy for screening or diagnostic purposes [20–22] , one study examined induced diarrhea without colonoscopy in healthy subjects [23] and one study induced less extreme diarrhea over several days intended to represent the physical effects of infectious diarrhea in the absence of an infectious agent [24] . Sampling strategies varied considerably between these studies , but none collected samples with sufficient frequency before or after the induced diarrhea to assess what day-to-day changes might be expected in the absence of deliberate perturbation . Furthermore , samples representing the perturbed state were separated by at least one week from any follow-up samples , so a detailed time course of gut microbiota recovery could not be investigated in these studies . We designed our sampling regime both to compare the effect of IIOD to the routine temporal variability of the gut microbiota in the same subject and to assess the timecourse of community recovery after IIOD . Some recent studies of the human gut microbiota have continued to rely on 16S rRNA gene surveys alone [25 , 26] , but it is increasingly common to combine such surveys with additional high throughput , culture independent methods , such as metagenomic ‘shotgun’ sequencing [27 , 28] , or metabolomics [29 , 30] . While all these methods provide a tremendous amount of information about microbial communities in their natural state , they present new and different challenges for data analysis and interpretation . We take the opportunity of analyzing our new human gut microbiota dataset to highlight useful recent advances in statistical methods which have yet to become widely adopted in microbiome studies . Two related challenges recognized soon after the application of next-generation sequencing to 16S rRNA gene surveys are the high dimensionality of the data ( hundreds or thousands 16S rRNA sequence variants identified per sample ) and the need to distinguish sequencing errors from genuine biological variation . A common response to both issues has been the application of ad hoc clustering methods that sweep both biological variants and error-containing sequences into bins defined by a fixed similarity threshold ( known as Operational Taxonomic Units or OTUs ) ; such an approach loses information by obscuring the existence of sequence variants that may represent ecologically distinct microbial strains [31] . In contrast , an explicit data-derived error model of Illumina amplicon sequencing allows likely ribosomal sequence variants ( RSVs ) to be distinguished both from each other and from errors , with a resolution as fine as single nucleotide differences , as demonstrated by the recent DADA2 package [32] . Once the sequence data are represented as an abundance matrix , with samples as rows and RSVs as columns , they become amenable to statistical scrutiny . However , these data present a unique set of methodological challenges; in response we present solutions based on adaptations of existing techniques or the introduction of new techniques . The first central challenge is high-dimensionality . After preliminary preprocessing , we have 419 samples and have measured 2611 RSVs and 2798 genes across these samples . Traditional methods can become unreliable and uninterpretable in this regime , where there are more measured features than samples . A second difficulty is interpretation in terms of phylogenetic units during analysis . There are few options for ordination that account for the known evolutionary relatedness between RSVs , and these methods are generally inflexible . However , incorporation of this structure leads to more informative results . Finally , standard techniques are not well-suited to simultaneous study of multiple data sources . Experiments that collect multidomain data on the same samples provide more interesting views of samples , by describing them from several angles . When such complementary data are available , it becomes interesting to characterize covariation across sources [33] . At present , there are relatively few methods designed for this purpose . To address these challenges , we repeatedly invoke a few key statistical principles . The first is that statistical methods can be improved by explicitly encoding known structure , for example , through informative priors or clever featurization . This principle motivates two methods that we introduce in this work—adaptive generalized principal components analysis ( agPCA ) [34 , 35] and tree-based sparse linear discriminant analysis ( LDA ) . By guiding statistical methods with domain knowledge—for example , about the phylogenetic relatedness of RSVs—we can typically obtain more useful results . A second principle is that ℓ1-regularization can address high-dimensionality in a way that facilitates interpretation . Indeed , regularization is foundational in modern high-dimensional statistics , and among regularization methods , ℓ1 constraints allow for the most convenient descriptions , because they induce sparsity [36] . “Sparsity” in this context means that a limited number of features , for us , either RSVs or gene ontology ( GO ) terms , are picked out as important for explaining the structure in the data . This form of regularization is used in both our tree-regularized supervised LDA and unsupervised sparse canonical correlation analysis ( sCCA ) . By implementing an intensive longitudinal sampling scheme that extended well before and after IIOD , we sought to place this perturbation to the human gut microbiota in the context of routine temporal variability . We characterized both the composition and functional potential of the gut community in eight individuals , analyzing the data with these new statistical methods and demonstrated improvements over current practice . Specifically , we pursued the following study aims: 1 ) determine whether and how quickly the gut microbiota demonstrates resilience after an IIOD perturbation , 2 ) elucidate patterns of taxonomic and functional change that characterize the community recovery process across individuals , and 3 ) innovate and apply statistical methods for high-dimensional , longitudinal multidomain data that provide greater interpretability than existing methods .
The research was approved by an Administrative Panel for the Protection of Human Subjects ( Institutional Review Board ) of Stanford University ( protocol 25268 ) . All subjects were properly informed of the risks and benefits of this study , and then signed an approved , written consent form . An unequally spaced time point design for longitudinal data with perturbations was created according to recommendations in the statistical design literature [37 , 38] . Demographic and life history factors such as gender , race and BMI , often used to stratify human populations in epidemiological studies generally have only small effects on the gut microbiota [39] . Note that a within-subject comparison of perturbed and unperturbed samples was possible because our longitudinal sampling design establishes the baseline temporal variability; insufficient sampling would increase the risk of mistaking routine temporal variability for a treatment effect . We show simulations that prove that crossover longitudinal sampling with baseline computations are more powerful than parallel designs in the supporting information ( S5 and S6 Figs ) . The response of the human gut microbiota to IIOD was evaluated by collecting fecal samples from eight healthy participants for approximately ten weeks before and ten weeks after a one-day IIOD event . IIOD is commonly used to clear the bowel prior to colonoscopy; the perturbation in this study exactly reflects a commonly-used clinical protocol for bowel preparation . On the morning of the perturbation , participants were instructed to drink ~300 mL of a solution ( GoLytely ) containing polyethylene glycol ( PEG ) and electrolytes every 10 minutes ( up to 4L total ) until their diarrhea was clear and watery . Samples were requested once per week every week , except during the week before and the week after IIOD when daily samples were requested . Five consecutive daily samples were also collected at least 6 weeks prior to IIOD . DNA was extracted from the stool samples and used for amplicon sequencing of the V4 region of the 16S rRNA gene as well as ‘shotgun’ metagenomic sequencing . The data were analyzed to reveal community composition and functional profiles , in an attempt to characterize the immediate response to IIOD , and to assess long-term effects of the perturbation . Healthy nonpregnant adults were recruited from the Stanford community , excluding individuals with chronic disease , hospitalization or antibiotic use in the previous 6 months , immunizations or international travel in the previous 4 weeks , or routine use of any prescription medication except birth control or hormone replacement therapy . Characteristics of the eight participants who completed the sampling protocol are summarized in Table 1 . Participants collected ~2g stool samples at home , which were frozen immediately without preservative in home freezers . Samples were transferred without thawing to −80°C storage in the laboratory approximately every 3 weeks . A total of 419 fecal samples were collected; the timing of samples relative to IIOD for each participant is shown in Fig 1 . Post-disturbance sampling began with the first bowel movement after IIOD in all subjects , which ranged from 1–3 days after IIOD . Some intended daily samples were not collected because participants did not produce stool that day . Samples were thawed at 4°C to a semi-solid state and ~250mg aliquots were transferred to wells of the PowerSoil -htp 96 Well Soil DNA Isolation Kit ( MoBio ) . Extraction followed the manufacturer’s centrifugation protocol , with the following modifications: stool tubes were thawed in small batches to minimize time unfrozen , the deepwell extraction plate was cooled on dry ice during sample loading , and extraction plates were returned to −80°C for at least 1 hour after loading to ensure consistent freeze-thaw cycles across all samples . Bead solution and C1 solution were added upon removal from the freezer to begin extraction with a 10 min incubation at 65°C , followed by 20 min beadbeating with the recommended MM 400 device ( Retsch ) . 6–12 extraction control blanks were included per extraction plate , as well as 52 replicate stool aliquots derived from 17 distinct samples . The V4 region of the 16S rRNA gene was amplified for sequencing using 515F and barcoded 806R primers as described by Caporaso et al . [40] . Triplicate 25 μL PCR reactions using Hot MasterMix ( 5 Prime ) with 3 μL extracted DNA as template and 10μg/μL BSA were cycled as follows: denaturation at 94°C for 3 min , 25 cycles of 94°C/45s , 52°C/60s , 72°C/120s , final extension at 72°C for 10 min . PCR amplicon libraries were purified using the UltraClean-htp 96 Well PCR Cleanup Kit ( MoBio ) . Amplicon libraries were quantified by fluorometry ( Quant-iT dsDNA High Sensitivity Kit , Invitrogen ) on a SynergyHT plate reader ( BioTek ) and combined in equimolar ratios into two pools . Pooled libraries were concentrated by ethanol precipitation and gel purified ( QIAquick Gel Extraction Kit , Qiagen ) . Each pool of V4 16S rRNA amplicons was sequenced ( 2x150 paired end ) on one lane of a HiSeq2500 sequencer ( Illumina ) at the Carver Biotechnology Center of the University of Illinois , producing an average of 237 , 800 reads per sample , with sample depths varying from 42 , 200 to a maximum of 1 , 530 , 000 , with a total of 365 , 093 , 804 reads produced for this study . The DADA2 sequence processing pipeline ( version 1 . 1 ) as described in [32] was used to infer the set of ribosomal sequence variants ( RSVs ) present and their relative abundances across the samples . Rather than clustering amplicon sequencing reads into Operational Taxonomic Units ( OTUs ) at a fixed similarity threshold , DADA2 derives an abundance distribution of distinct Ribosomal Sequence Variants ( RSVs ) , which may differ by only a single nucleotide , consistent with the observed sequence reads , based on data-derived rates of Illumina sequencing errors . Using read quality scores for the dataset , forward and reverse reads were truncated at 150bp and 130bp , respectively; other quality filtration parameters used DADA2 default values . Taxonomic assignment was performed on RSVs using the RDP classifier and reference dataset [41] following the workflow outlined in [42] . To improve power to detect subtle effects and to increase interpretability , it is often useful to include information about the phylogenetic relationships between RSVs . We use phylogeny both in the supervised context , to find groups of RSVs which distinguish samples immediately after the cleanout from the rest , as well as in the unsupervised context , to obtain a low-dimensional representation of the samples where the axes are interpretable in terms of over- or under-representation of groups of related RSVs .
Bray-Curtis dissimilarity was computed between all possible sample pairs and MDS was used to obtain a low-dimensional representation of these dissimilarities . The results are shown in Fig 2A , where the main effect is the difference across subjects . In other words , between-subject distances tend to be larger than within-subject distances . With the Bray-Curtis ordination , the pre-cleanout and post-cleanout samples do not show any systematic differences , as can be seen in S3 Fig . Community compositions in the days immediately surrounding the perturbation are displayed in supplementary S1 Fig . The analogous figure at the weekly level , is given in S2 Fig . The differences in composition across subjects is clearly evident , reinforcing the result of Fig 2 . Further , subjects AAD , AAF , AAG , and , to some extent , AAI exhibit decreases in Ruminococceae and Lachnospiraceae , though to differing degrees . These subjects also see an increase in the proportions of either Bacteroidaceae or Prevotellaceae in the days following the perturbation . The between-subject variation strongly justifies the decision to use a design in which each subject is their own control . Since we were interested in understanding the major portions of the between-sample variability which could be explained in terms of phylogenetically related groups of RSVs , we performed a phylogenetically-informed ordination of the RSV data using adaptive gPCA . The results of this ordination are shown in Figs 2B and 3A . We still see a subject effect: different subjects are localized to different regions of the principal plane and within-subject distances are generally greater than between-subject differences . However , we now also see an effect of the cleanout in that the samples immediately after the cleanout generally have more positive loadings along the first adaptive gPCA axis than other samples from the same subject . This is shown in more detail in Fig 3B , where we have plotted the scores of each sample along the first axis across time . The magnitude of the effect varies by individual , but points immediately after the cleanout tend to have the most extreme values along the first axis of any of the samples in the corresponding subject . Finally , we show the RSV loadings along the principal axes in Fig 3C . Since positive scores along the first axis seem to be associated with the samples immediately after the cleanout , we are particularly interested in RSVs which have strong loadings on this first axis . By examining Fig 3C , we see that a subset of the Bacteroidetes phylum has a strong positive loading on the first axis , and is therefore positively associated with the cleanout . This group corresponds exactly to the Bacteroides genus ( see S1 Fig ) . Since this genus seems to be associated with the cleanout , it is analyzed further below . Even without discovering the underlying mechanisms of resilience , the development of predictive diagnostics of resilience can be clinically relevant . With only 8 subjects , it is impossible to make any definitive conclusions; however , it is not unreasonable to explore methodological frameworks and propose possibly predictive factors . One approach to this problem is to define a scalar measure of resilience within each subject , and then attempt to predict this resilience measure using information known before any perturbation is performed . Any pre-perturbation features that may be predictive of resilience could become potential diagnostics . To characterize resilience , we use the relative change in Shannon diversity , computed over windows immediately preceding and following the cleanout . We use a window of length 3 days . As potential predictors of community recovery following severe perturbation , we consider taxonomic composition at the family level . Additional potential predictors include features from other measurement domains and/or derived features , a more complete supervised model would require more subjects and will be the focus of a complete followup study . While it is not unreasonable that the relative abundance of particular taxa ( e . g . nutritional generalists or specialists ) might influence community resilience , we are choosing this particular measure primarily to demonstrate the predictive methodology . Upon applying an elastic net regression to this problem , tuned by bootstrap resampling , we identify three families with nonzero coefficients , displayed in S8 Fig . There is a hint of an association between early presence of these bacteria and change in diversity after cleanout . For example , it seems that when Streptococacceae or Enterobacteriaceae are present at the onset of sampling , diversity actually increases post-cleanout , while when Prevotellaceae is more abundant at onset , diversity decreases . Of course , new data would need to be collected to validate these claims . Sequencing reads from the V4 16S survey and shotgun metagenomic sequencing are available from the NCBI Short Read Archive via BioProject PRJNA388263 . The adaptive generalized PCA programs have been combined into an R package called adaptiveGPCA on CRAN ( https://cran . r-project . org ) , the tree-aware sparse discriminant analysis code is available as the R package treeDA available on CRAN . All code Rmarkdown , R scripts , and data have been combined into the supporting information S1 Data which contains a tar file . There is also a larger docker file ( cleanout_submit . tar ) available at the Stanford digital repository permanent url: https://purl . stanford . edu/cf264md0197 for those who do not want to install R manually .
The present study constitutes an investigation of unprecedented rigor—with regard to length of sampling time period , temporal resolution of sampling , and generation of multiple data types—of the effects on the gut microbial community of a disturbance type , intestinal cleanout , relevant to clinical practice and ecological theory . As controls for comparison to the perturbed samples of each subject , we used unperturbed samples of the same subject , rather than making a comparison between distinct groups of subjects which did or did not experience IIOD . A simulation study ( the details and associated figure of which are available in the supporting materials S5 and S6 Figs ) confirms that in our context within-subject comparisons have greater power to reveal IIOD effects because inter-individual variation in the composition of the healthy adult human gut microbiota is greater than temporal variation within an individual [67 , 68] , even across experimental perturbations ( see [14 , 69 , 70] ) . As a result , this work resolves questions raised by previous studies of induced diarrhea [20–24] . These previous studies solely examined 16S rRNA taxonomic data and reached conflicting conclusions about effects of the perturbation on fecal microbiota in healthy adults . Some differences in past reported outcomes are likely attributable to variation among studies in clinical procedures and analytical methods . However , these prior studies collected only 2–5 samples per subject , with gaps of one week to one month between a sample representing the perturbed community and the earliest follow-up sample . Without fine-grained sampling beginning prior to perturbation onset and continuing until the community regained stability , these past studies could neither establish the timescale of recovery nor characterize the recovery process . In addition , by collecting samples at daily or weekly intervals for months before and after perturbation , we were able to assess the effect of this disturbance within the context of ordinary temporal variation for each subject . We have found and characterized a definitive but very transient effect of the colon cleanout . That is , we identified consistent changes in microbial community composition across all subjects in the first days following perturbation , after which the communities reverted to their pre-cleanout states . Furthermore , we found that no other phenomenon during the long sampling interval of normal temporal variation preceding IIOD compared in magnitude to the perturbation effect of IIOD on the community . In both the adaptive gPCA and the sparse CCA , the samples before and after the cleanout ( excluding those from the period immediately after the cleanout ) occupy the same region on the axes , suggesting there is no long-term compositional change resulting from cleanout . This rapid return to the pre-cleanout state is consistent with clinical observations that colon cleanout prior to colonoscopy for screening purposes in healthy individuals rarely leads to complications . The depth of sampling surrounding perturbation enabled characterization of the community recovery process using both taxonomic and metagenomic data . While community-wide metrics rapidly attain pre-perturbation states , we observe variation in recovery patterns on finer phylogenetic scales . Specifically , members of the Bacteroides genus recover quickly and dominate samples taken immediately post-perturbation while Ruminococcus genus members are slower to recover . Examining loadings of the RSVs on the agPCA axes offers insight into details of the compositional changes that accompany the cleanout . These results are partially consistent with some found in earlier studies . Gorkiewicz et al . found elevated relative abundance of OTUs within Bacteroides in fecal samples collected on the 3rd day of PEG-induced chronic diarrhea [24] . Drago et al . found reduced relative abundance of Firmicutes in fecal samples collected the day after bowel preparation with a combined stimulatory and osmotic laxative [21] . Shobar et al . found an elevated Bacteroidetes:Firmicutes ratio in dilute fecal material recovered via endoscopy from healthy subjects within a day of bowel lavage [22] . Several related biological mechanisms may explain the increased relative abundance of the Bacteroidetes phylum and Bacteroides genus , and decreased relative abundance of the Firmicutes phylum and Ruminococcus genus in the period immediately after the cleanout . We describe four potential mechanisms here: physical partitioning , substrate preference , growth rate , and differential oxygen tolerance . Physical partitioning could elevate Bacteroides abundance and decrease Ruminococcus abundance post-cleanout because paired fecal and mucosal biopsy samples from the unprepped colon of healthy humans revealed that members of Bacteroidetes are enriched in the mucosal layer , which would favor their retention during cleanout , while members of the Firmicutes are enriched in feces [71] . Furthermore , Firmicutes , and in particular members of the Ruminococcus genus , prefer attachment to undigested food particles over inhabiting the liquid phase of the gut lumen [69] . Attachment to food particles may enhance removal of Ruminococcus during cleanout . Differential use of growth substrates among the phyla may also contribute to elevated Bacteroides abundance and decreased Ruminococcus abundance post-cleanout . IIOD removes essentially all diet-derived substrates from the colon . Species capable of growth on the host-derived resources that would be available during and immediately after cleanout could begin to repopulate the colon earlier than specialist species that rely on specific diet components . Prominent gut Firmicutes tend to be nutritional specialists , whereas gut Bacteroidetes and members of the Bacteroides genus in particular are versatile foragers capable of growth on host-derived mucin [72 , 73] . A related but distinct potential explanatory mechanism for the compositional changes seen immediately post-cleanout is variation in intrinsic growth rates . Differential growth rates are related to the generalist/specialist mechanism in that generalists preferentially consume the resources that permit the fastest growth . Competition for labile substrates ensures their rapid depletion and advantages organisms capable of resource switching . On the other hand , nutritional specialists can persist in a flowing environment like the gut only if their preferred resource is reliably available , which requires that the resource not be easily degraded . The lower energy yield and/or slower rate of the catabolic reactions in degradation of a recalcitrant substrate , perhaps coupled with greater investments in requisite enzymes , imply slower maximal growth rates for specialists . In the unperturbed gut , reduced resource competition means that microorganism growth need only keep pace with the flow rate of the gut , so rapid growth is not as important for specialist fitness . The median ribosomal RNA operon copy number per genome is correlated with maximal growth rates of microbes [74 , 75] . On this basis , Bacteroides ( median 6 rrn copies/genome ) are likely to be capable of faster growth than Ruminococcus ( median 4 rrn copies/genome [76] ) Comparisons of microbial growth rates in culture are challenging to interpret because experimental conditions may not reflect the native habitat of the gut , but existing data from such experiments are consistent with the hypothesis that Bacteroides are generally capable of faster growth than Ruminococcus [77 , 78] . A final mechanism that may contribute to the over-representation of Bacteroides and underrepresentation of Ruminococcus in the post-cleanout period is differential oxygen tolerance . Under normal conditions , oxygen diffusing into the colon is rapidly depleted by facultatively anaerobic and microaerophilic microbes , allowing oxygen-sensitive anaerobes to grow in the colonic lumen [79] . The loss of most microbial biomass during cleanout and more rapid diffusion of oxygen through the less viscous intestinal contents that remain would increase oxygen concentration in the lumen of the colon . David et al . reached the same conclusion after observing a shift in the relative abundance of low-affinity vs . high-affinity cytochrome oxidases in the gut microbiome during early stages of succession following secretory diarrhea due to cholera [80] . According to the published literature summarized in Albenberg et al . , the Bacteroides genus includes both anaerobic and microaerophilic species , while Ruminococcus as well as all other genera in the Ruminococcaceae family are anaerobic [79] . The published literature may be biased by the relatively recent recognition of widespread microaerophily; a systematic investigation into the respiratory reductases encoded by 254 complete and partial genomes of human gut microbes found evidence for microaerophily in all 43 Bacteroides genomes that were examined , but only 4 of 9 genomes from Ruminococcus [81] . In addition to the 16S rRNA analyses employed by previous studies of IIOD , we collected metagenomic data and integrated analysis of the two data types using multitable methods . We applied sparse CCA to the combined 16S and metagenomic data to examine possible functional implications of the perturbation . To this end , we recovered a perturbation-related gradient across samples based on GO terms , indicating that changes in community functional capacity from baseline exist in the perturbed state . The GO terms included in this gradient may elucidate the survival advantages and disadvantages of organisms that recover quickly or more slowly , respectively , after IIOD . However , caution is necessary in the interpretation of GO terms included in the perturbation-associated gradient for several reasons . The GO terms defining the perturbation-associated gradient include a wide range of generality and specificity ( e . g . , “cellular metal ion homeostasis” and “tetrahydrobiopterin metabolic process” ) , and the method cannot establish the directionality of the causal relationship between microbial abundance and functional capabilities . That is , based on our data alone , we cannot say whether the GO terms we identified are functionally relevant to the response to perturbation or simply enriched ( or depleted ) in the genome of microbes that have a characteristic response to the perturbation due to other functional traits . In fact , we note two types of terms highlighted by the gradient: terms reflecting functions of importance for survival in the post-cleanout environment ( e . g . , catabolism of alanine and pyruvate family amino acids ) and terms carried by organisms systematically affected by the cleanout that are not themselves of direct functional importance for the carrier organisms’ survival ( e . g . , teichoic acid metabolic processes ) . We can nonetheless identify the predominant variation in functional terms , without specifying the underlying mechanism , based on Figs 3 and 5 . From the relatively small set of terms found to be more strongly associated with the perturbation , we highlight the presence of teichoic acid metabolic processes , which may indeed be a genomic marker of microbes with a characteristic response rather than a function with direct relvance to post-IIOD recovery . Teichoic acid is a cell wall component of the Gram-positive Firmicutes but not the Gram-negative Bacteroidetes , and appears with the expected positive association with both CCA axes . We have no reason to suggest that the presence or absence of teichoic acid per se influences microbial survival during the cleanout , but it is reassuring that a functional term known to be correlated with those taxa that are differentiated by other relevant functional traits is identified by this statistical technique . A functional trait that appears in several distinct clades of bacteria and is also selected by sparse CCA is more likely to reflect a function that is relevant to the perturbation , and such may be the case for the cluster of functional terms related to the catabolism of alanine and pyruvate family amino acids . Protein-coding genes from members of the Bacteroidetes , Firmicutes and Proteobacteria phyla ( among others ) are annotated with these terms and amino acid fermenting microbes belonging to 6 genera in these 3 phyla are known to associate with the human colonic mucosa [79] where they may resist elimination during the cleanout and would have access to host-derived protein . Furthermore , because proteolytic microbes are typically much less common in the gut relative to saccharolytic microbes , the selection of these functional terms by sparse CCA is less likely to be due to chance association with a broader taxonomic group . While analyzing these data , we developed new methods and applied existing methods in novel ways . Some of the key issues were: high dimensionality , the simultaneous study of multiple data sources , and our desire to have biologically interpretable results . The issues of interpretability and high dimensionality were both addressed with statistical regularization , either through the use of a sparsity constraint , through incorporation of the phylogenetic structure or using both approaches . The simultaneous study of multiple data sources was performed in an interpretable way using sCCA ( sparse CCA ) . Adaptive gPCA and tree-based discriminant analysis offer more flexible and interpretable incorporation of information regarding phylogenetic relatedness among observed RSVs than existing methods . In both cases , the aim was to obtain an explanation of the variation between the samples in terms of groups of closely-related RSVs . We expected this constraint to be useful both because groups of closely-related RSVs are more biologically interpretable than lists of unrelated RSVs and because we expect closely-related RSVs to respond in similar ways to IIOD . In adaptive gPCA , we are interested in explaining the overall variability between the samples in these terms . Currently , the most common approaches for comparing samples in microbiota studies either ignore the phylogeny entirely ( e . g . Bray-Curtis ) or incorporate the phylogeny in a fixed way ( e . g . weighted Unifrac ) . Furthermore , the ordination axes resulting from these approaches are not directly interpretable in terms of which microbial taxa are most important for positioning samples in the lower dimensional space . In contrast , adaptive gPCA allows more fine-tuned control of the extent to which phylogeny is reflected in the analysis and offers explanations of the ordination axes in terms of closely related RSVs . In tree-based discriminant analysis , we were interested in explaining the difference between the samples at baseline and the samples immediately after IIOD , but we again wanted the explanation to be in terms of groups of phylogenetically-related RSVs . By including features associated with internal tree nodes , the tree-based discriminant analysis allows identification of larger evolutionary units whose members are all associated with the response . Without this enrichment in the feature space , it is only possible to read off individual RSVs associated with the response and then attempt to assess phylogenetic relatedness in follow-up analysis . Further , in a limited sample-size setting , individual microbe effects may be undetectable , while aggregate evolutionary-unit level signals may be clear . In this situation , only a model incorporating these higher level units as features would succeed . The analysis also incorporates a sparsity constraint , meaning that in the final model most of the RSVs are considered unimportant in explaining the differences between the groups and giving us just a small number of related RSVs to focus our attention on . sCCA described here offers improved integration of analysis on multiple datatypes collected from the sample set . Most microbiota studies have employed only a single analytical technique ( most often 16S rRNA gene surveys ) , although an increasing number of studies apply additional techniques ( e . g . metagenomics , metabolomics ) to at least a subset of samples . However , the data derived from each technique has typically been studied in isolation , not exploiting the fact that various techniques have been applied to the same set of samples . By explicitly seeking aspects of the data structure that are shared across multiple data types , multitable statistical analyses can provide insight into the fundamental biological processes responsible for the patterns observed via different techniques . For example , sCCA defines ordinations based on the latent factors present across all data sources , down-weighting the influence of factors present in isolated data types . Consequently , the positions of samples in the reduced space is informed by relatedness across multiple data types . Further , the factors recovered by sCCA can illuminate sets of features across multiple data types that are correlated with one another , suggesting the presence of fundamental biological processes driving parallel changes across data types . Both sparse LDA and sCCA induce sparsity through ℓ1 regularization , reducing variance and improving interpretability in the high-dimensional regime . The high dimensionality of modern ‘omics’ data poses a problem for traditional statistics because the hundreds or thousands of identified features ( e . g . , microbial taxa , functional genes ) generally greatly exceeds the number of samples analyzed . The problem of identifying meaningful associations in high-dimensional data is often handled with a FDR approach , which seeks to provide the largest possible list of features that vary in the comparison , while keeping the rate of false positive feature identifications below a certain threshold . The resultant long lists of features can be difficult to interpret , especially when separate lists of significantly varying features are generated from different analytical techniques . An alternative approach is to apply a sparsity constraint during feature selection , which seeks to restrict the list of significant features to the small set most strongly associated with the comparison of interest . In contrast to testing , sparse models can encode specific structure—for example , phylogenetic or multidomain structure—while still providing a parsimonious description of the essential signals in a data set . Dense LDA coefficients or CCA factors can be difficult to inspect , relative to sparse versions which allow attention to be focused on the subset of coordinates with nonzero values . Further , without some form of regularization , ordinary LDA and CCA are statistically unidentifiable in the case that the number of features exceeds the number of samples , as in the IIOD experiment . Even in the case that the number of features is slightly smaller than the number of samples , unregularized models can be alarmingly unstable . Across analysis types , sparse models can encode known structure , simplify inspection of coefficients , and improve model stability . The recovery process post-IIOD observed here has potential implications for clinical practice . While colonoscopy ( and a fortiori the IIOD used to prepare the bowel for the procedure ) has a low rate of complications for healthy adults undergoing colonoscopy for colorectal cancer screening , for ulcerative colitis patients colonoscopy is associated with an exacerbation of symptoms [82] . Both the reduced abundance of Ruminococcus that are prominent producers of anti-inflammatory butyrate in the human gut [72] and the potential for increased abundance of pro-inflammatory facultative anaerobes of the Proteobacteria phylum [83] could contribute to this phenomenon . Prebiotic interventions to increase the relative abundance of butyrate-producing microbes before and after the colonoscopy [84] ( given the depletion of such organisms in IBD [85] ) , as well as irrigation of the colon with sodium butyrate solutions at the time of colonoscopy [86] may help reduce post-colonoscopy symptoms and hasten the return to a balanced microbiota in IBD patients . One way forward in illuminating the microbial and functional landscape related to perturbations and temporal variability would be to augment metagenomic data with metabolomic or transcriptomic measurements , applying the statistical techniques described here . These methods provide data that could be used to interrogate microbial function and activity with less potential for confounding due to the covariation of relevant functional traits with other genes carried on the same bacterial genomes . For example , it would be possible to directly quantify short chain fatty acids or secondary bile acids using metabolomic techniques and relate these measurements to the expression of recognizable genes from both characterized and uncharacterized microbial taxa . The provisioning of these and many other compounds are recognized as ecosystem services of the gut microbiota , with health effects both locally in the gut and systemically throughout the host [87 , 88] . The new tools and insight described in this work provide guidance and a framework for a more comprehensive assessments of stability and resilience in complex ecosystems , such as the human microbiome . The use of longitudinal study design and multidomain analysis , as we and others are now undertaking , will reveal ecosystem features that are both predictive and diagnostic of key health-associated attributes , and will guide new forms of informed intervention . | Complex dynamics of microbial communities underlie their essential roles in health and disease . To maintain or restore healthy states , we must better understand the nature and basis of stability in the gut microbiota , under normal and perturbed conditions . Stability , resilience , and response to perturbation are central topics in community ecology . Extreme perturbations such as near-complete loss of biomass from a system can reveal factors that influence community structure . Recognizing the return to baseline diversity and abundances of biomarkers in community-wide recovery after a disturbance enables us to understand the basic pillars of resilience that contribute to human health . We have designed a densely sampled longitudinal experiment in human volunteers using transient non-inflammatory diarrhea as the perturbation . In order to uncover the essential players in the recovery process , we have tailored new advances in ribosomal sequence variant detection and sparse multidomain analytics that incorporate phylogenetic structure . We show sparse meaningful multidimensional projections that exhibit the essential features in resilient recovery . This work shows how a carefully designed longitudinal study combining denoised ribosomal RNA sequence variants and metagenomic data can inform the taxa and processes involved in the recovery from loss of large proportions of intestinal biomass . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"taxonomy",
"medicine",
"and",
"health",
"sciences",
"gut",
"bacteria",
"pathology",
"and",
"laboratory",
"medicine",
"bacteroides",
"diarrhea",
"phylogenetics",
"data",
"management",
"metagenomics",
"signs",
"and",
"symptoms",
"phylogenetic",
"analysis",
"gastroenterolog... | 2017 | Multidomain analyses of a longitudinal human microbiome intestinal cleanout perturbation experiment |
The coordinated action of a variety of virulence factors allows Salmonella enterica to invade epithelial cells and penetrate the mucosal barrier . The influence of the age-dependent maturation of the mucosal barrier for microbial pathogenesis has not been investigated . Here , we analyzed Salmonella infection of neonate mice after oral administration . In contrast to the situation in adult animals , we observed spontaneous colonization , massive invasion of enteroabsorptive cells , intraepithelial proliferation and the formation of large intraepithelial microcolonies . Mucosal translocation was dependent on enterocyte invasion in neonates in the absence of microfold ( M ) cells . It further resulted in potent innate immune stimulation in the absence of pronounced neutrophil-dominated pathology . Our results identify factors of age-dependent host susceptibility and provide important insight in the early steps of Salmonella infection in vivo . We also present a new small animal model amenable to genetic manipulation of the host for the analysis of the Salmonella enterocyte interaction in vivo .
Non-typhoidal Salmonella are one of the major causal agents of bacterial gastrointestinal infections in all age groups worldwide . In addition , they also significantly contribute to systemic infection particularly in the pediatric population . Non-typhoidal Salmonella together with pneumococci and group B streptococci are among the most frequent causal agents of sepsis and meningitis in neonates and young infants in Africa [1]–[5] . Transmission of this highly endemic bacterium occurs perinatally from infected mothers or after birth from infected family members via the fecal-oral route . The underlying mechanisms of the enhanced susceptibility to systemic disease after infection during the postnatal period have not been systematically investigated . Salmonella pathogenicity is conferred by horizontally acquired chromosomal regions so called Salmonella pathogenicity islands ( SPI ) that encode sets of virulence factors required for the various steps in microbial virulence . SPI1 encodes a type III secretion systems ( T3SS ) and a set of effector molecules that are translocated into the target cell . It mediates pathogen-induced internalization in non-phagocytic cells [6]–[8] . The critical importance of cellular invasion by Salmonella and intracellular proliferation in non-phagocytic cells is supported by epidemiological data and has been extensively studied in vitro [6] , [9]–[10] . Although intracellular Salmonella have been observed in epithelial cells and cells of the lamina propria also in vivo , this host cell interaction and the functional importance have remained less defined [7]–[8] , [11]–[15] . Instead , translocation through microfold ( M ) cells has been demonstrated in adult animals . Such cells overlay Peyer's patches and forward luminal antigens and bacteria to the underlying immune cells [16]–[18] . Similarly , uptake via lamina propria resident myeloid cells that generate membrane extrusions extending into the intestinal lumen has been proposed to facilitate Salmonella barrier penetration [19] . More recently , rapid translocation via the colonic epithelium in the absence of intracellular proliferation was observed [15] . The underlying host and bacterial factors facilitating enterocyte invasion , intraepithelial proliferation and mucosal translocation have , however , remained largely undefined . Here we comparatively analyzed oral Salmonella infection of neonate and adult mice . We demonstrate marked age-dependent differences in intestinal colonization , mucosal translocation and systemic spread . We observed Salmonella enterocyte invasion in neonate mice in vivo and characterize intraepithelial microcolony formation , epithelial translocation and epithelial innate immune stimulation after oral infection . We characterize developmental peculiarities of the newborn's intestinal epithelium that allow enterocyte invasion and characterize the contribution of Salmonella virulence mechanisms . Our results identify age-dependent factors of infection susceptibility and provide a novel in vivo model that allows the use of genetically modified hosts to investigate the intimate host microbial interaction at the intestinal epithelium .
One-day-old C57BL/6 neonates were orally infected with various inocula ( 102–105 ) of mid-log grown Salmonella enterica subsp . enterica sv . Typhimurium ( S . Typhimurium ) . Colony counts and visualization of bioluminescent S . Typhimurium in the intestine using optical imaging technology demonstrated rapid colonization of the small intestine and colon ( Fig . 1A–C ) . Even low dose infection ( 102 CFU ) resulted in systemic spread with maximal bacterial counts in spleen and liver at day 4 post infection ( p . i . ) ( Fig . 1D–E ) . The absence of bacteria in lung tissue excluded accidental primary pulmonary infection ( Fig . S1A ) . Comparative analysis of 1-day-old , 6-day-old and streptomycin pretreated adult animals at day 4 after oral administration of 102 , 5×102 , and 5×108 CFU S . Typhimurium , respectively , demonstrated the efficient colonization of the neonate intestine and the marked spread to liver and spleen tissue in 1-day-old animals ( Fig . 1F–G ) . We therefore focused on 1-day-old animals and applied low dose infection and analyzed animals at 4 days p . i . for all subsequent experiments , if not stated otherwise . As SPI1 is involved in Salmonella internalization in non-phagocytic cells , we next analyzed the requirement of Salmonella SPI1 effector translocation for oral infection of neonatal mice . Spread to liver ( p<0 . 001 ) , spleen ( p<0 . 001 ) and mesenteric lymph nodes ( MLN , p<0 . 001 ) was practically abolished in the absence of a functional SPI1 system ( ΔinvC ) ( Fig . 2A–C ) . In fact , spleen tissue of all mice and liver and MLN tissue of the majority of neonates ( 8/14 and 18/23 , respectively ) remained sterile at day 2 and 4 after infection by SPI1 deficient Salmonella . Similarly , systemic dissemination of ΔinvC Salmonella was also highly significantly reduced after high dose ( 105 CFU ) infection ( Fig . S2D ) . In accordance with these results , immunostaining ( Fig . 2D ) , plating of isolated epithelial cells after gentamicin-treatment to remove extracellular Salmonella ( Fig . 2E–F ) and flow cytometry ( Fig . 2G ) detected enterocytes infected by WT but not SPI1-deficient Salmonella . Both , WT and ΔinvC Salmonella after high and low dose infection spontaneously colonized the intestinal tract ( Fig . S2A–C ) . The reduced total small intestinal organ counts at later points after low dose infection may result from the lack of intraepithelial bacteria ( Fig . 2E–F ) . In contrast to the critical role of SPI1 in neonate mice , intestinal colonization but also spread to spleen , liver and MLN in adult streptomycin-pretreated animals was largely SPI1-independent ( Fig . 2H and Fig . S2E ) . To identify differences that might account for the requirement of SPI1 in neonate but not adult animals , the gene expression profile of isolated primary neonate and adult epithelial cells was compared . Unexpectedly , genes of differentiated M cells such as Spi-B and Ccl9 were found to be markedly reduced in neonate epithelial cells ( Fig . 3A–B ) . In adult hosts , M cell-mediated bacterial translocation represents the major entry pathway of Salmonella [16] , [17] . To confirm the age-dependent differentiation of M cells , mRNA expression of the M cell specific transcription factor Spi-B was examined [18] . As expected , Spi-B expression was found to be strongly diminished in neonates and an increase was only observed starting at day 8 after birth ( Fig . 3C ) . Also , immunostaining of intestinal tissue for the established M cell markers glycoprotein 2 ( gp2 ) , Ccl9 and Ulex europaeus agglutinin ( UEA ) -1 confirmed the appearance of M cells only after the neonatal period ( Fig . 3D ) . Finally , wildtype ( WT ) and fimD mutant Salmonella disseminated to a similar degree to the spleen and liver tissue after 4 days infection of 1-day-old neonates ( Fig . S3A ) . fimD mutant Salmonella are unable to express type I pili and cannot attach to the M cell surface protein gp2 . This significantly reduces their ability to invade the adult host via M cells [20] . In contrast to a recent report on Salmonella-induced Spi-B expression in adult animals , Spi-B expression in neonate mice was reduced rather than enhanced following Salmonella infection ( Fig . S3B ) [21] . Hence , Salmonella in neonate mice spontaneously colonize the intestine , invade enterocytes , penetrate the mucosal barrier and spread to systemic organs in a SPI1-dependent but M cell-independent fashion . Salmonella infection of neonatal enterocytes in vivo was subsequently studied in more detail . Immunostaining revealed Salmonella-positive enterocytes ( Fig . 4A ) . Unexpectedly , Salmonella generated multi-bacterial intraepithelial colonies of variable size consistent with the formation of Salmonella-containing vacuoles ( SCV ) previously observed in epithelial cell lines in vitro ( Fig . 4A–B ) . Quantification of Salmonella-positive cells was achieved by flow cytometric analysis on isolated enterocytes . 0 . 24±0 . 12% and 0 . 80±0 . 61% of all murine epithelial cell adhesion molecule ( EpCAM ) + cells at 2 and 4 days p . i . , respectively were found to be infected ( uninfected controls: 0 . 07±0 . 03% , p<0 . 05 ) . Co-infection using a 1∶1 mixed inoculum of genetically green- and red-labeled bacteria revealed solely single-colored intracellular colonies ( Fig . 5A ) . This suggests that individual microcolonies originated from a single event of bacterial invasion . Electron microscopy showed that bacteria were enclosed in most instances by a detectable endosomal membrane ( Fig . 4B ) . The intra-endosomal material appeared heterogeneous with hyper- or hypodense areas ( Fig . 4Biv–v ) . Consistently , the well-established SCV transmembrane marker protein lysosomal-associated membrane protein ( Lamp ) 1 was detected in close proximity to the majority of intraepithelial microcolonies by immunostaining ( Fig . 5B–C ) . In some instances , however , direct contact of individual bacteria with the host cell cytosol could not be excluded . Clear morphological signs of intracellular bacterial proliferation were noted ( Fig . 4Bv–viii ) in accordance with a rise in the mean fluorescence intensity ( MFI ) of infected EpCAM+ enterocytes from 59 . 5±35 . 1 to 98 . 4±62 . 1 at 2 and 4 days p . i . , respectively . Despite invasion and intracellular proliferation , Salmonella-positive cells appeared morphologically intact ( Fig . 4B ) . Whereas a low number of uninfected enterocytes stained positive for active caspase 3 , no apoptotic Salmonella-positive cells were identified ( Fig . 5D–E ) . In sharp contrast , only minute numbers of ( questionably ) intraepithelial Salmonella without any indication of bacterial proliferation were detected in tissue sections of adult mice after high dose ( 1–2×108 CFU ) infection despite numerous attempts both with and without streptomycin pretreatment ( data not shown ) . Consistently , flow cytometric analysis did not detect Salmonella-infected intestinal epithelial cells ( Fig . S2F and ) . To identify molecular differences that might account for the age-dependent susceptibility of the epithelium to Salmonella invasion , the gene expression profile of neonate and adult enterocytes was compared ( Fig . 3A ) . Known SCVs constituents ( with the notable exception of iNOS ) and autophagosomal markers were equally expressed in neonate and mature adult epithelial cells . However , the neonate intestine exhibited a markedly reduced mucus layer thickness and mucin glycoprotein expression ( Fig . S3C–D ) . Also , reduced cyclin expression was noted in accordance with minimal epithelial cell turn-over and reduced crypt-villus migration in neonates [22] , [23] . Finally , a severely reduced spectrum of antimicrobial peptides was observed consistent with previous reports [24] . All of these factors might facilitate epithelial invasion , prolong the lifetime of infected epithelial cells and thereby ultimately allow intraepithelial proliferation and microcolony formation . We next investigated the neonate's host response to Salmonella infection . Highly enriched primary enterocytes ( Fig . S4 ) were subjected to quantitative RT-PCR . A significant time-dependent increase in Cxcl2 and Cxcl5 mRNA expression was measured after infection by WT or the complemented ΔinvC pinvC but not SPI1-deficient non-invasive Salmonella ( Fig . 6A–B ) . Global gene array analysis confirmed the absence of any detectable increase in epithelial gene expression upon administration of invC deficient , non-invasive Salmonella ( Fig . 6C ) . It further identified a large number of additional genes involved in metabolism , cellular responses and intercellular communication induced after Salmonella WT infection ( Fig . 6D ) . In accordance with the requirement of enterocyte invasion for innate immune stimulation ( Fig . 6C ) and the observed age-dependent susceptibility to enterocyte invasion ( Fig . 2D–G and S2F ) , Cxcl2 and Cxcl5 mRNA expression in enterocytes isolated at day 4 p . i . from infected 6-day-old or adult mice was severely reduced ( Fig . 6E–F ) . We next examined the innate immune receptors and signaling adaptors involved in Salmonella-induced enterocyte stimulation . Expression of Cxcl2 and Reg3γ mRNA by epithelial cells was severely reduced in the absence of toll-like receptor ( Tlr ) 4 , MyD88 as well as Unc93B1 and Tlr9 . Although some variability in the expression levels was noted , significantly decreased mRNA expression was also noted in epithelial cells devoid of Tlr2 , Tlr5 , or Nod2 ( Fig . 7A–B ) . Of note , intestinal colonization ( Fig . 7C ) , enterocyte invasion ( Fig . 7E–G ) and dissemination to systemic organs ( Fig . 7C–D ) was observed also in the absence of the most potently stimulated innate immune receptor , Tlr4 . Although the production of proinflammatory mediators significantly enhanced the recruitment of polymorphonuclear cells and blood macrophages to the site of infection , no gross loss of epithelial barrier integrity or histopathological signs of tissue destruction were observed at day 4 p . i . in contrast to the situation in the adult host ( Fig . 8 ) [25] . Thus , Salmonella in the neonate efficiently invades small intestinal enterocytes , activates innate immune responses mainly via Tlr stimulation and induces the formation of intraepithelial microcolonies , a hallmark of the Salmonella-enterocyte interaction .
Previous work had demonstrated enterocyte invasion in oral and intestinal loop infection models and ex vivo tissue explants of calfs , rabbits , swine and guinea pigs [26]–[32] . However , these host animals are not amenable to genetic manipulation and studies are restricted to the early course of infection . Enterocyte invasion , rapid egress at the basolateral side of the epithelium and the presence of Salmonella in lamina propria cells has also been described in the mouse large intestine [7] , [11]–[15] . The present study now provides a new small animal model that allows the use of genetically modified hosts to analyse intracellular proliferation in small intestinal epithelial cells and the formation of microcolonies by Salmonella in vivo . It might thereby facilitate a better understanding of a hallmark of Salmonella pathogenesis in vitro , the internalization by non-phagocytic epithelial cells and the formation of intraepithelial microcolonies , so called Salmonella containing vacuoles ( SCVs ) . Invasion of non-phagocytic epithelial cells is facilitated by virulence factors encoded by the pathogenicity island SPI1 [6] . The critical role of SPI1 for enteric disease is illustrated by its presence in all subspecies of Salmonella enterica [33] and its high prevalence in clinical Salmonella isolates [9] . In accordance , SPI1 effector proteins are required to cause diarrhea and mucosal inflammation in several in vivo models although also SPI2 effector molecules contribute to intraepithelial survival and proliferation [25] , [34]–[36] . The analysis of the interaction of S . Typhimurium with polarized epithelial cells within their anatomical environment represents a prerequisite to understand the functional role of individual bacterial virulence factors . Whereas previous reports using streptomycin pretreated adult animals failed to demonstrate infection of the small intestinal epithelium , we observe efficient Salmonella invasion of enterocytes in neonate mice . Known host constituents of mature SCVs and autophagosome factors are similarly expressed by neonatal and adult epithelial cells [37]–[39] . However , a number of fundamental differences exist between the neonate and adult gut epithelium in mice . For example , we demonstrate that the synthesis of various mucin glycoproteins and thus the thickness of the mucus layer are significantly enhanced in adult animals . The mucus layer was shown to significantly impair mucosal translocation of Salmonella in adult mice [40] . Also , we and others have shown that the antimicrobial peptide repertoire is severely reduced in the neonate intestine in the absence of mature Paneth cells . In adult mice , Paneth cell-derived antimicrobial peptides cells significantly influence the course of enteropathogen infection [41] , [42] and cooperate with the mucus layer to generate an antibacterial and antiinflammatory shield [43] . Finally , the well-established constant renewal of the epithelium associated with enterocyte migration and exfoliation at the villus tip is not yet established in newborn mice [22] , [23] . The reduced epithelial cell turn-over is the consequence of the lack of crypts and the reduced pool of rapidly proliferating cells during this early developmental stage . Thus , enterocytes during the neonatal period remain longer at the same anatomical position . This might allow Salmonella to proliferate intracellularly and form microcolonies . Shedding of Salmonella-infected enterocytes at the villus tip was previously observed in bovine , pig and rabbit intestinal loop models and might represent an important mechanism to remove infected enterocytes [28] , [30] , [44] . Finally , the prolonged presence of the more diverse enteric microbiota in the adult host may contribute to the enhanced resistance to Salmonella enterocyte invasion and this issue requires future investigations . In the neonate host both Salmonella enterocyte invasion and mucosal translocation were dependent on SPI1 . This suggests that penetration of the neonate's intestinal barrier occurs secondary to the observed enterocyte invasion . This is in contrast to the situation in adult mice but also other species . Here , M cells were shown to represent the major port of entry [16]–[18] , [29] . M cells as part of the follicle-associated epithelium overlaying Peyer's patches facilitate uptake of particulate antigen [17] , [18] , [45] . Thus , translocation takes place independent of bacteria-induced internalization [16] . In contrast , the absence of M cells in the neonate mouse intestine shifts the major entry pathway to enterocyte invasion . It thereby pronounces the requirement for SPI1 for mucosal translocation . This may also explain the recent finding that Yersinia enterocolitica translocation is highly reduced in neonate mice [46] . Mucosal translocation of enteropathogenic Yersinia is also mostly mediated through M cell transport [47] . However , we cannot formally exclude that additional host factors also contribute to the enhanced requirement for a functional SPI1 system in the neonate host . Strong innate immune receptor-mediated stimulation of the epithelium was noted upon infection in accordance with previous analyses of infected total mucosal tissue [48] , [49] . The present study provides a global gene expression analysis of isolated primary enterocytes after Salmonella infection in vivo . The restriction of innate immune stimulation to invasion-competent Salmonella is consistent with reports on the requirement of SPI1 for mucosal inflammation and clinical disease [25] , [34]–[35] , [49] . Epithelial stimulation was mediated by recognition of Salmonella lipopolysaccharide ( LPS ) through Tlr4 and signaling via MyD88 consistent with previous reports on the involvement of Tlr4 in the adult host defense against Salmonella [50] , [51] . Similarly , deficiency in Tlr9 and the processing molecule Unc93B1 significantly reduced the epithelial cell response to Salmonella infection . The strong effect observed in 3d UNC93B1 mutant mice indicates the possible involvement of additional UNC93B1 dependent innate immune receptors such as Tlr3 , 7 or 13 [52] . Our results indicate the synergistic action of Tlr4 and Tlr9 in the epithelial response to Salmonella in accordance with previous reports [53] . A minor but significant role was also found for Tlr2 , Tlr5 and Nod2 previously implicated in the antimicrobial response of the adult host to Salmonella [54] . The synergistic stimulation of Tlr2 , 4 and 9 was shown to be required for endosomal acidification and SPI2 effector protein translocation in bone marrow-derived macrophages [53] . A similar scenario may apply to intestinal epithelial cells . In accordance , lack of Tlr4 alone did not significantly influence enterocyte invasion or intraepithelial proliferation in vivo . Further studies are required to investigate the role of cooperative immune signaling by different receptors for the process of enterocyte infection and microcolony formation . Colonization of the neonate intestine occurred largely independent of the invasion-dependent innate immune stimulation consistent with a low degree of colonization resistance in the neonate host . Reduced organ counts in the total small intestine after low dose infection of SPI1 mutant Salmonella might result from the lack of proliferating intraepithelial Salmonella . Alternatively , invasion-induced immune stimulation might promote metabolic changes that favor pathogen colonization of the neonate intestine as recently described in adult animals [55] . In accordance , a number of metabolic genes were significantly influenced during bacterial challenge supporting the idea of a heavily altered intestinal host metabolism during neonatal infection [56] . In addition , several innate immune response genes such as the chemokines , serum amyloid proteins , iNOS , calprotectin , mucins and antimicrobial lectins were strongly upregulated in infected epithelial cells in accordance with previous reports [57] . As a consequence , large numbers of polymorphonuclear cells ( PMNs ) but also macrophages were recruited to the site of infection . The cellular immune response at this time point , however , did not lead to significant tissue alteration . The anti-apoptotic effect of NF-κB stimulation might have prevented the induction of epithelial apoptosis previously reported in both Salmonella-infected and non-infected enterocytes [44] . In conclusion , we analyzed age-dependent differences in the interaction of Salmonella with the host epithelium . We identify host and bacterial factors responsible for the infection process in the neonate host and characterize the epithelial innate immune response . Our results demonstrate enterocyte invasion and intracellular proliferation of Salmonella in differentiated and polarized epithelial cells and established a new animal model amenable to genetic manipulation to study the enterocyte-pathogen interaction in vivo . However , neonate mice exhibit an immature mucosal immune system , which might contribute to the observed phenotype and requires further characterization . Also , limitations such as the small animal size and lack of a suitable anesthesia required for intravital microscopy as well as the possible exchange of bacteria between newborn mice and the dam exist . Nevertheless , we believe that the described model opens new avenues of research to unravel the functional role of individual effector proteins in the Salmonella–enterocyte interaction and to better understand the cellular and immunological events of the Salmonella pathogenesis .
The isogenic strain MvP818 harboring a deletion of invC has been described before [58] . For complementation of the invC mutation , plasmids p3545 was generated as follows: The promoter of invF was amplified from S . Typhimurium genomic DNA using PinvF-For-EcoRI ( CCGGAATTCTCCATCCAG ATGACAATATC ) and PinvF-Rev-SmaI ( ATATCTAGATCCATCCAGATG ACAATATCTG ) . The resulting product was digested by EcoRI and SmaI and subcloned in pWSK29 to obtain p3537 . invC was amplified using invC-For–SmaI ( atacccgggtttagtcg gtcgctaatgag ) and invC-Rev-XbaI ( GTATCTAGATTAAT TCTGGTCAGCGA ATGC ) , the resulting fragment was subcloned as SmaI/XbaI fragment in p3537 to generate p3545 . Correct clones were confirmed by DNA sequencing and functional complementation of the invasion defect of MvP818 was verified in non-phagocytic intestinal epithelial m-ICcl2 cells ( Fig . S5A–C ) . S . Typhimurium ATCC 14028 and SPI-1 mutant S . Typhimurium MvP813 ΔinvC ( KanaR ) carrying a green fluorescent protein ( GFP ) expression plasmid ( pGFP , AmpR , kindly provided by Brendan Cormack , Stanford , USA ) and S . Typhimurium ATCC 14028 carrying a mCherry expression plasmid ( pFPV-mCherry , AmpR , kindly provided by Leigh Knodler , NIH , Hamilton , USA ) were used for in vivo infection experiments . Maintenance of the plasmid pGFP in S . Typhimurium under in vivo conditions was confirmed by comparative plating of spleen and liver tissue simultaneously on selective ( 100 µg/mL ampicillin ) and non-selective LB agar plates ( Fig . S6C–D ) . For oral infection with two different Salmonella strains , a 1∶1 mixture of 102 pGFP and 102 pFPV-mCherry carrying bacteria were administered orally in 2 µl PBS . Life imaging ( IVIS ) was performed employing an isogenic S . Typhimurium strain with a chromosomal insertion of the lux-operon from Photorhabdus luminescens under the control of the constitutive ß-lactamase promoter . Adult C57BL/6 wildtype and B6 . B10ScN-Tlr4lps-del/JthJ Tlr4 ( stock no . 007227 ) , B6 . 129S1-Tlr5tm1Flv/J Tlr5 ( stock no . 008377 ) , B6 . 129-Tlr2tm1Kir/J ( stock no . 004650 ) , and B6 . 129S1-Nod2tm1Flv/J Nod2 ( stock 005763 ) deficient , as well as C57BL/6J-Ticam1Lps2/J TRIF mutant ( 005037 ) mice were obtained from the Jackson Laboratory ( Bar Harbour , USA ) . 3d [59] and B6 . 129P2-Tlr9 ( tm1Aki ) Tlr9 deficient mice [60] were obtained from M . Brinkmann , Helmholtz Center for Infection Biology , Braunschweig , Germany . Mice were housed under specific pathogen-free conditions and handled in accordance with regulations defined by FELASA and the national animal welfare body GV-SOLAS ( http://www . gv-solas . de ) . Salmonella enterica subsp . enterica serovar Typhimurium ATCC14028 ( NCTC12023 ) WT and isogenic mutant strains were cultured in Luria Bertani ( LB ) broth overnight at 37°C , diluted 1∶10 and incubated at 37°C until reaching the logarithmic phase ( OD600 approximately 0 . 5 ) . Bacteria were washed and adjusted to OD600 0 . 55–0 . 60 containing approximately 1 . 5–2 . 0×108 CFU/mL and diluted to obtain the appropriate infection dose . All experiments with neonatal mice were performed using 1-day-old C57BL/6 animals . Neonates were infected orally with the indicated number of S . Typhimurium in a volume of 1 µl PBS . The administered inoculum was confirmed by serial dilution and plating . In vivo imaging was performed using the IVIS200 imaging system ( PerkinElmer ) . For image analysis and visualization the LivingImage 4 . 3 . 1 software was used . Oral infection of 6 week-old adult female mice was performed as previously described [25] . Bacterial counts were obtained after homogenization of spleen , liver and mesenteric lymph nodes by serial dilution and plating . All animal experiments were performed in compliance with the German animal protection law ( TierSchG ) and approved by the local animal welfare committee ( approval 12/0697 , 12/0693 , 13/1097 and 14/1385 of the Niedersachsische Landesamt für Verbraucherschutz und Lebensmittelsicherheit Oldenburg , Germany ) . Primary intestinal epithelial cells were isolated from small intestinal tissue after incubation in 30 mM EDTA PBS at 37°C for 10 min as described previously [61] . Cells were passed through a 100 µm nylon cell strainer ( BD Falcon ) , washed with 10% FCS/PBS , and harvested by centrifugation . The purity was verified by flow cytometry ( Fig . S4 ) . In order to determine the number of cell-associated versus intracellular bacteria , one fraction of epithelial cells was plated directly ( total number of epithelium-associated bacteria ) ; another fraction was incubated in 100 µg/mL gentamicin for 1 h prior to plating ( number of gentamicin-protected , intracellular bacteria ) . For flow cytometry analysis of intracellular Salmonella , cells were fixed with 4% PFA for 15 min . on ice . The cells were washed , stained with APC-conjugated anti-EpCAM ( anti-CD326 clone 48 . 8 , diluted 1∶500 , from eBioscience ) and finally resuspended in 5% FCS/PBS . Prior to analysis , cells were filtered through a 35 µm pore size BD Falcon Polystyrene tube ( BD , cat . no . 352235 ) . 300 , 000 events were acquired and GFP expression was detected using a FACS Calibur apparatus ( BD ) . For purity confirmation of isolated primary enterocytes , cells were fixed and incubated on ice with APC-conjugated anti-EpCAM ( clone 48 . 8 ) and PE-conjugated anti-CD45 antibodies ( clone 30-F11 , dilution: 1∶200 ) purchased from eBioscience , for 30 minutes . Cells were washed and filtered . 25 , 000 events were acquired . Postacquisition analysis of FACS data was performed using the Summit 5 . 0 software . For flow cytometric analysis of tissue invading myeloid cells , small intestines from neonatal mice were removed and opened longitudinally . Tissue was digested in Liberase ( Roche ) /DNAse I ( Roche ) /10%FCS/RPMI at 37°C and separated on a 40%/70% Percoll gradient . Cell were stained with PerCP Cy5 . 5 conjugated anti-CD45 ( Clone 104 ) , Brilliant Violet-conjugated anti-I-A/I-E ( Clone M5/114 . 15 . 2 ) , APC/Cy7-conjugated anti-CD11c ( Clone N418 , all purchased from Biolegend ) , PE-conjugated anti-CD11b ( Invitrogen ) and APC-conjugated anti-F4/80 ( Clone BM8 , from eBioscience ) . The samples were acquired on a LSRII ( BD ) and analyzed with FlowJo ( Treestar ) . Immunostaining using a chicken anti-GFP ( dilution 1∶500 ) , a mouse anti-mCherry ( diluted 1∶500 , both from Abcam ) , a mouse anti Salmonella O4-antigen ( diluted 1∶500 , from Abcam ) , a rat anti-Lamp1 ( diluted 1∶500 , from the Developmental Studies Hybridoma Bank ( DSHB ) , University of Iowa , US ) , a rabbit anti-active Caspase 3 ( diluted 1∶200 , from Cell Signaling ) , a rabbit anti-Muc2 ( diluted 1∶100 , generous gift from Gunnar Hansson , Göteborg , Sweden ) and a mouse anti-E-cadherin ( diluted 1∶100 , from BD Transduction Laboratories ) , was performed on 3 µm paraformaldehyde-fixed paraffin-embedded tissue sections in combination with the appropriate fluorophore conjugated secondary antibody ( Jackson ImmunoResearch ) . Sections were deparaffinized in xylene and rehydrated in ethanol followed by antigen retrieval in 10 mM sodium citrate and blocking with PBS 10% serum 5% BSA . Staining of constitutively GFP expressing bacteria was performed to enhance the sensitivity of detection . In addition , this approach ensured bacterial detection also under low oxygen conditions , which might be present within the intestinal lumen and have been reported to alter GFP emission . The accuracy of the anti-GFP staining to detect plasmid-bearing S . Typhimurium was verified by staining of two consecutive sections cut from the same tissue block with anti-GFP and a biotinylated 1∶100 diluted rat monoclonal anti-O4/O5 antibody generously provided by M . Kim ( Kim Laboratories , Champaign , IL ) . Both staining methods revealed a very similar staining pattern suggesting that the anti-GFP staining method used was both sensitive and specific ( Fig . S6A–B ) . Staining with the rhodamine-conjugated lectin UEA-1 ( Vector ) , a rat anti-GP2 ( diluted 1∶100 , from MBL ) or a goat anti-Ccl9 ( diluted 1∶100 , from R&D ) was performed on 8 µm sections obtained from freshly frozen OCT ( Tissue-Tek ) embedded tissue . Sections were dried and fixed using methanol at −20°C for 10 min . followed by rehydration in PBS for 15 min . Blocking in 0 . 2% BSA , 0 . 2% saponin , 10% serum in PBS was performed prior to immunostaining . Peyer's patches were visualized using a FITC-conjugated anti-CD45 antibody ( clone 30-F11 , BioLegend ) . Hematoxylin and eosin staining of paraformaldehyde-fixed tissue sections was performed according to Mayer's protocol using reagents from Roth . Slides were mounted in Vectashield ( Vector ) supplemented with DAPI and pictures were taken with an Apo-Tome microscope connected to a digital camera ( Zeiss ) . For electron microscopical analysis of infected intestinal tissue , samples were fixed in 150 mM HEPES , pH 7 . 35 , containing 4% formaldehyde and 0 . 1% glutaraldehyde at room temperature for 1 hour and then stored over night in fixative at 4°C . Samples were dehydrated in acetone and embedded in EPON . 60 nm sections were mounted onto formvar-coated copper grids , stained with 4% uranyl acetate and lead citrate as previously described by Reynolds and visualized in a Morgagni TEM ( FEI ) , operated at 80 kV [61] . Total RNA was extracted from isolated enterocytes using TRIzol ( Ambion ) and the RNA concentration was determined on a NanoDrop 1000 spectrophotometer ( Thermo Scientific ) . First-strand complementary DNA ( cDNA ) for quantitative RT-PCR was synthesized from 5 µg of RNA with Oligo-dT primers and RevertAid reverse transcriptase ( Fermentas ) . Taqman technology based RT PCR was performed using absolute QPCR ROX mix ( Thermo Scientific ) , sample cDNA and the Taqman probes hprt ( Mm00446368_m1 ) , Spi-B ( Mm03048233-m1 ) Cxcl2 ( Mm00436450_m1 ) , Cxcl5 ( Mm00436451_g1 ) and Reg3γ ( Mm01181783_g1 ) from Life Technologies . SYBR green–based real-time PCRs were performed as described in the Extended Experimental Procedures . Microarray analysis was performed using Whole Mouse Genome Oligo Microarray v2 ( 4×44k ) ( Agilent Technologies ) following the SC_AgilentV5 . 7 protocol provided by the manufacturer . Heat map analysis was performed using the MeV 4_5_1 software . Clusters of orthologous groups ( COG ) analysis was performed using the online bioinformatic tool PANTHER ( http://www . pantherdb . org/ ) . The expression array data are accessible through GEO Series accession numbers GSE51160 . Reviewer can access the raw data files using the following link: http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? token=ehufucgejpwtjwf&acc=GSE51160 . Intestinal epithelial m-ICcl2 cells were cultured for 6 days with medium change every other day to obtain a confluent cell monolayer as previously described [62] . m-ICcl2 cells were incubated with wildtype S . Typhimurium , the SPI1 mutant strain ( ΔinvC ) or the complemented SPI1 mutant strain ( ΔinvC pinvC ) at a multiplicity of infection of 1∶10 for 1 h at 37°C , washed and gentamicin ( Sigma Aldrich ) was added at a concentration of 100 µg/mL for 1 h at 37°C . Cells were washed 3 times in PBS and lysed directly or maintained in culture in the presence of 20 µg/mL gentamicin . At the indicated incubation period cells were washed with PBS and lysed in 500 µl 0 . 1% Triton X-100 for 2 min . at room temperature . Viable counts were determined by serial dilution and plating on LB agar plates . Flow cytometric analysis was performed on trypsinized m-ICcl2 cells following fixation using a FACS Calibur apparatus ( BD ) . Results for bacterial growth in organ tissues or quantitative RT-PCR show counts for individual animals plus the median . The One-way ANOVA Kruskal-Wallis test ( with Dunn's posttest ) and the Mann-Whitney test were employed for statistical analysis of bacterial growth in organ tissue at different time points , comparative analysis of quantitative RT PCR results or bacterial counts after infection with WT versus SPI1 mutant Salmonella respectively . The GraphPad Prism Software 5 . 00 was used for statistical evaluation . p values are indicated as follows: ***p<0 . 001; **p<0 . 01 , and *p<0 . 05 . | Non-typhoidal Salmonella are among of the most prevalent causative agents of infectious diarrheal disease worldwide but also very significantly contribute to infant sepsis and meningitis particularly in developing countries . The underlying mechanisms of the elevated susceptibility of the infant host to systemic Salmonella infection have not been investigated . Here we analyzed age-dependent differences in the colonization , mucosal translocation and systemic spread in a murine oral infection model . We observed efficient entry of Salmonella in intestinal epithelial cells of newborn mice . Enterocyte invasion was followed by massive bacterial proliferation and the formation of large intraepithelial bacterial colonies . Intraepithelial , but not non-invasive , extracellular Salmonella induced a potent immune stimulation . Also , enterocyte invasion was required for translocation through the mucosal barrier and spread of Salmonella to systemic organs . This requirement was due to the absence of M cells , specialized epithelial cells that forward luminal antigen to the underlying immune cells , in the neonate host . Our results identify age-dependent factors of host susceptibility and illustrate the initial phase of Salmonella infection . They further present a new small animal model amenable to genetic manipulation to investigate the interaction of this pathogen with epithelial cells and characterize the early steps in Salmonella pathogenesis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacterial",
"diseases",
"infectious",
"diseases",
"infectious",
"disease",
"immunology",
"medicine",
"and",
"health",
"sciences",
"pediatric",
"gastroenterology",
"clinical",
"immunology",
"medical",
"microbiology",
"gastroenterology",
"and",
"hepatology",
"microbial",
"pa... | 2014 | Age-Dependent Enterocyte Invasion and Microcolony Formation by Salmonella |
Helminth parasites are an assemblage of two major phyla of nematodes ( also known as roundworms ) and platyhelminths ( also called flatworms ) . These parasites are a major human health burden , and infections caused by helminths are considered under neglected tropical diseases ( NTDs ) . These infections are typified by limited clinical treatment options and threat of drug resistance . Aminoacyl-tRNA synthetases ( aaRSs ) are vital enzymes that decode genetic information and enable protein translation . The specific inhibition of pathogen aaRSs bores well for development of next generation anti-parasitics . Here , we have identified and annotated aaRSs and accessory proteins from Loa loa ( nematode ) and Schistosoma mansoni ( flatworm ) to provide a glimpse of these protein translation enzymes within these parasites . Using purified parasitic lysyl-tRNA synthetases ( KRSs ) , we developed series of assays that address KRS enzymatic activity , oligomeric states , crystal structure and inhibition profiles . We show that L . loa and S . mansoni KRSs are potently inhibited by the fungal metabolite cladosporin . Our co-crystal structure of Loa loa KRS-cladosporin complex reveals key interacting residues and provides a platform for structure-based drug development . This work hence provides a new direction for both novel target discovery and inhibitor development against eukaryotic pathogens that include L . loa and S . mansoni .
The worm parasites Loa loa ( Ll ) and Schistosoma mansoni ( Sm ) are causative agents of loiasis and schistosomiasis , respectively [1 , 2] . L . loa is a member of the nematode phyla that infects ~13 million people every year in west and central Africa causing notable morbidity , disability and socioeconomic loss [2–4] . L . loa larvae are transferred to humans after the bite of infected deerfly vector ( Chrysops spp . ) . These larvae slowly develop into mature adult parasites that traverse through various tissues and manifest angioedema , endomyocardial fibrosis , eosinophilia , encephalitis and nephropathy [2–5] . Their migration across eye conjunctiva has led to the common term of African eye worm [2–5] . Adults produce microfilariae by sexual reproduction and are re-circulated by flies during another blood meal [2–5] . These microfilariae then develop into infective larvae inside the fly [2 , 3 , 5] . Loiasis can be treated by the WHO recommended first line treatment of diethylcarbamazine or administration of alternative drugs like ivermectin and albendazole [3 , 6] . These treatments , however , are not always easily applicable and pose life threatening risks [3 , 6] . In contrast with Loiasis , schistosomiasis is a deadly neglected tropical disease that affects ~210 million people and kills >200 , 000 each year [2 , 7 , 8] . Schistosomiasis burden is mainly concentrated in the sub-Saharan Africa with highest prevalence in children and adults [2 , 7 , 8] . Human schistosomiasis is caused by three major Schistosoma species of platyhelminths phylum—S . mansoni , S . japonicum and S . haematobium [2 , 7 , 8] . These blood flukes complete their life cycle by shuttling between human and snail hosts . Adult S . mansoni reside in human vasculature and produce plentiful of eggs daily that are either excreted or deposited in the host liver . These events can trigger immune-mediated granuloma formation , hepatosplenism and periportal fibrosis leading to fatality [1 , 2 , 7 , 8] . Single dose of praziquantel ( PZQ ) is almost entirely used for treatment and control of schistosomiasis , but this mass monodrug therapy may lead to drug resistance [2 , 7 , 8] . Additionally , the drug target for praziquantel remains unknown , which can hamper attempts to rationally design and synthesize second-generation drugs based on PZQ . Hence , both Loiasis and Schistosomiasis require discovery and validation of new druggable targets as well as insights into novel chemical scaffolds that can be used for drug development . Others and we have shown that targeting of aminoacyl-tRNA synthetases ( aaRSs ) from infectious agents that cause malaria , toxoplasmosis , bacterial infections , fungal infections and leishmaniasis can be valuable [9–24] . The aaRSs control protein biosynthesis pathways by allowing pairing of cognate tRNA with amino acids [25] . Usually a cellular translational compartment contains 20 aaRSs , and depending on shared sequence motifs and topology in catalytic domains these aaRSs are divided into two classes . Class I enzymes contain the ATP binding motifs HIGH and KMSKS ( Rossmann fold ) while three conserved sequence motifs called 1 , 2 and 3 are the characteristic of class II enzymes [25 , 26] . Lysyl-tRNA synthetase ( KRS ) couples L-lysine to cognate tRNAs , and is the only aaRS that has evolved in different organisms to fall in both class I and II . While eukaryotes and most prokaryotes contain class II KRS , some bacteria and archaea contain class I [25 , 27 , 28] . The aaRSs can perform many non-canonical functions , and these have been documented for human as well as parasitic aaRSs [14 , 29] . KRSs from many organisms , including the malaria parasite P . falciparum , have also been reported to synthesize signaling molecules like diadenosine polyphosphates ( Ap4A , Ap5A ) that modulate variety of cellular functionalities such as DNA replication , gene expression and ion channel regulation to mention a few [18 , 22 , 30 , 31] . Crystal structures and functional analyses of human cytoplasmic KRS have shown that this enzyme can exist in tetrameric and dimeric forms , where the tetrameric form is bound to multi-synthetase complex and is translationally active , and the dimeric form can participate in transcription regulation and may have cytokine-like functions [32 , 33] . Thus , determining the oligomeric status of KRSs is of key importance in understanding their functionality and mechanism . Previous reports on the malaria parasite Plasmodium falciparum ( Pf ) KRS showed that this dimeric protein is inhibited by the fungal secondary metabolite cladosporin with high potency [15 , 22 , 34] . Cladosporin targets malaria in both blood and liver stages with IC50 values below 100 nM [34] . This antimalarial effect is highly selective and mammalian cells are protected as assessed by cytotoxicity assays . In an effort to understand the protein translation components responsible for supplying charged tRNAs for ribosomal protein synthesis within L . loa and S . mansoni , we first cataloged all their aaRSs and associated protein factors . We noted that L . loa and S . mansoni KRSs are present as single copy in both parasites . We discovered that cladosporin is a very potent inhibitor of L . loa and S . mansoni KRSs , and has enzyme inhibitory IC50 values in low nanomolar ranges ( ~52 nM and ~97 nM respectively ) . We provide the X-ray co-crystal structure of this drug bound L . loa KRS to demonstrate its binding mechanism and selectivity . Our proof-of-concept data on pathogen aaRSs predicts that targeting of other schistosome members and trypanosomes should also be feasible using the same chemical scaffold .
Usually one aaRS is required per amino acid and thus a complete set of 20 aaRSs is necessary for protein translation in any cellular context when alternate pathways for producing charged tRNAs are not available [25 , 26] . We thus looked for all sets of aaRSs in L . loa and S . mansoni genomes . L . loa genome encodes 35 putative aaRSs along with 5 accessory proteins–and this set likely fulfills aminoacylation requirements of its two translational chambers of cytoplasm and mitochondria ( S1 Table ) [35 , 36] . A careful analysis of the predicted cellular distribution shows that 19 aaRSs are likely to be localized in the cytoplasm with an absence of KRS ( S1 Table ) . Gene structure suggests that the single copy LlKRS absence from the cytoplasm as it contains a predicted mitochondrial N-terminal signal sequence , but its sequence alignment , domain and motif analyses suggest it to be a eukaryotic-type protein , possibly dual localized . Putative aaRSs with specificity for 16 amino acids are present in L . loa mitochondria with 4 ( cysteine , glutamine , glycine and threonine specific ) missing aaRSs ( S1 Table ) . Amongst these , the glutamine charged tRNA can be provided by indirect pathway involving putative mitochondrial glutamyl-tRNA amidotransferase [35 , 36] . In comparison to Loa loa , S . mansoni contains a set of 19 cytoplasmic aaRSs with notable absence of glycine-specific aaRS ( S2 Table ) . However , the mitochondrial GlyRS is a eukaryotic-type enzyme and its dual localization is likely , as has been reported in other organisms [13 , 37] . The S . mansoni mitochondrial aaRSs set is deficients in GlnRS , HisRS and LysRS ( S2 Table ) . Based on data from several laboratories including ours on aaRSs cellular distributions in eukaryotic pathogens , it is likely that the twin mechanisms of dual aaRs localizations and trafficking of charged tRNA across translational compartments are active in L . loa and S . mansoni as well , in order to provide all substrates required for protein translation in both compartments [13 , 17 , 37 , 38] . LlKRS and SmKRS enzymes that contained the aminoacylation and anticodon binding domains were produced recombinantly in E . coli ( Fig 1A ) . Our localization predictions and comparative sequence analysis hinted that the LlKRS contained a mitochondrial signal peptide ( 1–35 ) , while the SmKRSs was predicted to be a cytoplasmic enzyme ( Fig 1A ) . To assess the oligomer status of purified proteins , we performed gel permeation chromatography ( GPC ) experiments using a calibrated column with known standards [22] . Both wild type worm KRSs eluted at sizes corresponding to expected dimers in our GPC experiment , unlike the human counterpart that purportedly shows a tetrameric form ( Fig 1B ) [32] . Purified worm parasite proteins were used for enzyme assays using SmtRNALys and were found to be enzymatically active ( Fig 1C and 1D ) . Cladosporin is a 3 , 4-dihydro-6 , 8-dihydroxy-3- ( 6-methyl-tetrahydro-2H-pyran-2-yl ) compound that mimics the adenosine part of ATP ( Fig 2A ) . To test the activity inhibition and IC50 values of wild type SmKRS and LlKRS , enzyme assays were performed in presence of cladosporin . Human-like LlKRS V329Q/S346T mutant protein was also produced by taking cues from previous reports and structural data analysis in this work ( see next sections ) . The drug showed concentration-dependent enzymatic inhibition and revealed IC50 values of 52 nM and 97 nM for wild type LlKRS and SmKRS respectively , while a significantly higher IC50 value of 1370 nM was observed for the human-like LlKRS mutant protein ( Fig 2B ) . We also performed protein thermal shift assays to determine the binding of cladosporin to human-like LlKRS and wild type worm KRSs in presence of L-lysine and ( a ) either no ligand , or ( b ) with non-hydrolysable ATP analogue ( adenosine 5’- ( β , γ-imido ) triphosphate ( AMPPNP ) ) , or ( c ) with cladosporin in equal micromolar amounts . Data indicated that AMPPNP in a 10:1 molar ratio to KRSs was able to induce a small shift of ~0 . 3°C , ~0 . 4°C and ~0 . 4°C for LlKRS , SmKRS and for human-like LlKRS respectively indicating very weak binding ( Fig 2C ) . As expected , cladosporin when used in ten-fold higher molar concentration ( 20uM cladosporin ) relative to KRSs ( 2 μM ) was able to induce substantial shifts of 15 . 5°C and 10 . 8°C in both LlKRS and SmKRS respectively , indicating high affinity interactions with parasitic KRSs ( Fig 2C ) . On the other hand , significantly smaller shift of 5 . 3 and 7 . 2°C were observed when human-like LlKRS ( 2 μM ) was incubated with ten fold ( 20 μM ) or even hundred fold higher ( 200 μM ) concentrations of cladosporin , indicating much poorer binding of cladosporin to the mutant LlKRS ( Fig 2C ) . To determine the binding affinity of cladosporin , we performed ITC experiments and discovered Kd values of 45 . 2 ± 8 . 4 nM and 62 . 8 ± 7 . 8 nM for LlKRS and SmKRS respectively ( Fig 2D ) . Similar changes in binding enthalpy ( ΔH ) and entropy ( TΔS ) factors indicated conserved mechanism of worm KRS-cladosporin complexation ( Table 1 ) . Together , our enzyme inhibition , TSA and ITC data demonstrate strong affinity of cladosporin for these worm KRSs , and validate the potential of cladosporin to selectively bind parasitic KRSs over human counterpart where its affinity is relatively poor ( 3 . 3 μM ) [15 , 39] . To understand the atomic basis of KRS-cladosporin binding , we solved the crystal structure L . loa KRS in complex with cladosporin ( CLD ) and L-lysine ( K ) . Crystal packing analysis showed two dimers of LlKRS in the crystallographic asymmetric unit , validating our GPC results on recombinant LlKRS ( Figs 3A and 1B , Table 2 ) . LlKRS folds into a canonical eukaryotic KRS and contains N-terminal OB fold anticodon binding domain and a C-terminal catalytic domain ( Fig 3B ) . The signature motifs 1 , 2 and 3 present in the catalytic domain are also conserved ( Fig 3B ) . Cladosporin fits into the ATP binding site in LlKRS and interacts with most of the residues that accommodate adenosine moiety of ATP ( Fig 3C and 3D ) . The isocoumarin ring of cladosporin is stabilized mainly by π-π staking with Phe344 , T-stacking with His-340 and hydrogen bondings with Asn341 backbone and Glu334 ( Fig 3D ) . In addition , guanidine group from Arg563 and Arg332 also stabilize the isocoumarin moiety ( Fig 3D ) . Gly560 provides hydrophobic support to tetrahydropyran ring ( THP ) whereas the Ser346 lends suitable space for its methyl moeity . The L-lysine binds in the inner region of active site pocket and is accommodated by series of hydrogen bondings with protein atoms ( Fig 3E ) . A comprehensive sequence alignment of cladosporin-sensitive pathogen KRSs like PfKRS , SmKRS , LlKRS was used to map drug-binding residues in the active sites of these KRSs ( Fig 4A and 4B ) . HsKRS ( PDB: 4YCU ) and LlKRS share ~66% overall sequence identity and show r . m . s . d . of 1 . 88 Å in their cladosporin-bound forms for 463 Cα atoms . The earlier reported human tetramer enzyme ( PDB: 3BJU ) and our observed LlKRS dimer show differences in the amino acid sequences and topology of tetramer interface regions 1 and 2 ( Fig 4A and 4B ) [32] . Despite the comparatively conserved eukaryotic insertion 1 in LlKRS , the sequence and structural differences in tetramer interface region within it appear to have endowed only dimeric conformation to the worm KRSs ( Fig 4A and 4B ) . Interestingly , LlKRS is sequence-wise and in architectural terms ( r . m . s . d . ) closer to HsKRS than the PfKRS ( 54% sequence identity and 2 . 22 Å r . m . s . d . with LlKRS for 455 Cα atoms ) , and yet possesses cladosporin sensitivity like PfKRS . Availability of recent cladosporin-bound HsKRS structure provided an opportunity for us to compare the human and worm enzymes . The active site region and binding mechanism of cladosporin for both LlKRS and HsKRS is remarkably similar , with only difference of Ser346/Thr337 ( Ll/Hs ) and distant Val329/Gln321 ( Ll/Hs ) substitutions ( Fig 5A ) . It is clear that the residues Ser346 and Val329 provide extra space for accommodating the methyl moiety of THP ring , and they thus contribute to species selectivity [15 , 34 , 39] . PfKRS is currently the best-studied model for understanding cladosporin-binding mechanism , and in addition to two selective residues , many other structural aspects of malaria KRS that contribute to selectivity have become apparent during recent structural investigations by our group and from others [15 , 22 , 39] . To understand the LlKRS cladosporin binding and specificity , we analyzed it in backdrop of known Plasmodium and human KRS structures [15 , 22 , 39] . In PfKRS , binding of cladosporin induces a loop movement ( near motif II ) of approximately ~2 . 4 Å towards the active site , and rearrangements of His338 ( LlHis331 ) , Phe342 ( LlPhe335 ) and Arg559 ( LlArg553 ) occur to accommodate isocoumarin moiety of cladosporin ( Fig 5B and S1 Movie ) . These events coincide with formation of disulfide bond in a disordered loop region of PfKRS ( Fig 5B ) . The L-lysine induces an inward mobility in the active site roof region and also stabilizes a disordered loop ( Fig 5B ) . All four major transitions are present in the PfKRS-CLD-K complex , and the recent crystal structure of HsKRS-CLD-K also shows a structural state similar to PfKRS ( r . m . s . d . 1 . 43 Å for 490 Cα atoms ) [15 , 39] ( Fig 5C ) . The L-lysine induced changes have recently been proposed to be the main factor driving cladosporin species selectivity ( Fig 5B ) [39] . To address this , we compared cladosporin and L-lysine bound LlKRS structure to the already available HsKRS and PfKRS cladosporin-bound structures . We found that a helix in L-lysine-induced mobile body is disordered in LlKRS ( Fig 5D ) . Additionally , the stable helix ( in case of HsKRS ) or the disulfide stabilized loop ( in case of PfKRS ) are also absent in LlKRS and that this region is disordered ( Fig 5D ) . These structural observations hence support the observation that most likely the conserved pair of Val329 and Ser346 found in pathogen KRSs drive cladosporin selectivity .
Helminths represent some of the most prevalent neglected tropical disease parasites , and schistosomiasis likely ranks just below malaria as a cause of misery in context of public health [2 , 7 , 40] . The currently available monodrug treatment of schistosomiasis using praziquantel poses threat of possible drug resistance [2 , 7 , 8 , 40] . On the other hand , loiasis is prevalent in rainforest and low socioeconomic regions , and has gained prominence in recent years due to adverse effects of drug treatments during co-endemicity with other filarial pathogens [3 , 6] . Research efforts directed at understanding vital cellular processes such as protein translation machinery can hugely benefit drug discovery initiatives , especially given the promise of utility in context of other infectious diseases like malaria . This is especially of benefit to neglected tropical disease research , where efforts to develop drugs needs to be cost effective . Prompted by these concerns , we sought to dissect worm aaRSs that are responsible for protein translation in these organisms . In this report , we have provided a comprehensive overview of the aaRS distributions in genomes of L . loa and S . mansoni . In both these organisms , it is likely that aaRSs fulfill translational requirements in two cellular compartments of mitochondria and cytoplasm by evolutionarily successful mechanisms of aaRS dual localization , indirect aminoacylation pathways and trafficking of charged tRNAs [13 , 17 , 37 , 38] . Further , the presence of single copy KRSs in both pathogens presents a lucrative opportunity to target this enzyme so as to dismantle protein synthesis process in two translational compartments simultaneously . We additionally found that other members of Schistosoma genus like S . japonicum ( GeneBank: CAX83109 . 1 ) and S . haematobium ( GeneBank: KGB31491 . 1 ) also possess the conserved cladosporin-sensitive motif and hence could be targeted by cladosporin via their KRSs ( Fig 6A ) . Interestingly though , sequence analyses show that L . loa and O . volvulus could be specific targets amongst pathogenic filarial nematodes ( OVOC_0000240101-mRNA-1 ) ( Fig 6A ) . Sequence differences in L . loa , W . bancrofti ( GeneBank: EJW79634 . 1 ) and Brugia malayi ( GeneBank: CDP92701 . 1 ) KRSs suggest that the latter two might be less sensitive to cladosporin , hence providing an opportunity to selectively target L . loa and O . volvulus during co-infections ( Fig 6A ) . Poor bioavailability is the main limitation in development of cladosporin as a lead molecule [34] . Chemical synthesis protocols for cladosporin are now available which can aid in structure-guided rational synthesis of more drug-like cladosporin derivatives [41] . Apart from loiasis and schistosomiasis , suitable derivatization of cladosporin for better ADME ( absorption , distribution , metabolism , and excretion ) properties will be highly valuable for drug development against host of parasitic infections including malaria and feasibly trypanosomosis ( given the conservation in active site residues that recognize cladosporin , Fig 6B and 6C ) [15] . Hence , cladosporin-based small molecular libraries could be very good starting points for cell-based and phenotypic screening against a number of eukaryotic pathogens . The presented data here therefore provide new avenues for novel drug development against parasitic worm diseases and highlights numerous potential aaRS targets that can now be exploited .
The aaRSs and accessory proteins were identified using HMM-search tool in the HMMER web server ( http://www . ebi . ac . uk/Tools/hmmer/ ) by restricting the taxonomy against L . loa and S . mansoni and with an E-value cut-off of 0 . 01 . Pfam IDs of aaRSs and accessory proteins were used in the HMMER based search . Each hit was verified further by sequence , domain and motif analyses using SMART [42] , CD-search [43] and superfamily servers [44] . Localizations were predicted using online servers MitoProt ( mitochondrial localization- http://ihg . gsf . de/ihg/mitoprot . html ) and NucPred ( nuclear localization ) [45] . Available mitochondrial localization prediction softwares are trained on non-helminthes organisms and thus sequence alignments with respective validated mitochondrial aaRSs ( from NCBI ) were also used to identify prokaryotic/mitochondrial type aaRSs . Mitochondrial/prokaryotic type aaRSs found in our analysis , but without the predicted mitochondrial signal sequence were also assigned as putatively mitochondrial . Splice variants from single gene and any atypical aaRSs found in Uniprot [46] were verified via EnsemblMetazoa transcript database ( http://metazoa . ensembl . org/index . html ) . Protein sequences for related worm parasites were obtained from LlKRS or SmKRS protein blast while O . volvulus KRS ( OVOC_0000240101-mRNA-1 ) sequence was obtained from LlKRS protein blast in http://www . sanger . ac . uk/ . LlKRS and SmKRS protein sequences were aligned against PfKRS PDB sequence ( 4PG3 ) to identify N-terminal sequences of unknown functions or a possible signal sequence , i . e . 1 to 27 for SmKRS and 1 to 77 for the LlKRS . Gene sequences encoding protein residues 28–634 of SmKRS and 78–599 of LlKRS were designed for expression in Escherichia coli and subcloned into pETM-41 vector using NcoI and KpnI restriction sites . Human-like LlKRS V329Q / S346T mutant was created by site directed mutagenesis in two positions V329 to Q and S346 to T . Cloning , expression and purification of the human-like LlKRS protein was performed as for the wild type LlKRS . Protein expression for wild type KRSs and human like LlKRS was induced by adding 0 . 5 mM isopropyl -d-1-thiogalactopyranoside ( IPTG ) to cells grown till OD600 of 0 . 6–0 . 8 at 37°C . These cells were grown for 20 h post-induction at 18°C . Bacterial cells were harvested by centrifugation at 5000 g for 15 min and the bacterial pellets were suspended in a buffer consisting of 50 mM Tris–HCl ( pH 8 . 0 ) , 200 mM NaCl , 10 mM beta-mercaptoethanol ( βMe ) , 15% ( v/v ) glycerol , 0 . 1 mg ml-1 lysozyme and 1 mM phenylmethylsulfonyl fluoride ( PMSF ) . Cells were lysed by sonication and cleared by centrifugation at 20 , 000 g for 45 min . The cleared supernatants were applied onto amylose beads ( GE Healthcare ) . All three MBP-KRS fusion proteins were eluted with 10 mM maltose in 50 mM Tris–HCl ( pH 8 . 0 ) , 200 mM NaCl , 10 mM βMe , 1 mM DTT , 0 . 5 mM EDTA . The MBP tag was removed by incubation of eluted pure fractions with TEV protease at 293 K for 24 h . Wild type and mutant cleaved worm KRSs were concentrated using a 10 kDa cutoff Centricon centrifugal device ( Millipore ) and were purified by gel-filtration chromatography on a GE HiLoad 60/600 Superdex column . Pure fractions were checked by SDS–PAGE and pooled for crystallization . Before crystallization , the wild type LlKRS was concentrated to 10 mg/ml ( A280 , extinction coefficient- 46760 M-1 cm-1 ) and stored at -80°C . To determine the oligomer status , high molecular weight calibration standards ( GE Healthcare ) and purified proteins were run on GPC column using protein buffer mobile phase at flow rate of 0 . 5 ml/ min . Molecular weights of LlKRS and SmKRS were estimated from their elution profiles by plotting log molecular weight ( X-axis ) against partition coefficient ( Kav , Y-axis ) for known standards . SmtRNALys was synthesized using an in-vitro transcription method described earlier with minor modifications [47] . A double stranded DNA sequence ( TCAGTAGCTG AGTGGATAAT GCGA TGGCGT TTTAAGCGAA CGGTACTGGG TTCGAGTCCCAGAGTGAACCA ) encoding cytoplasmic SmtRNALys ( GeneDB: SmtRNA_01463_Lys_TTT . 1 . 1 ) containing 5’ T7 RNA polymerase promoter , CCA sequence at 3’ ( in italics ) and 2’-O methyl substitution in last two nucleotides of antisense strand was purchased ( Sigma ) . This sequence was transcribed using T7 quick high yield RNA synthesis kit ( NEB ) at 37°C for 16 h according to manufacturer’s instructions . DNA template was removed by DNase ( 10U/ml ) treatment for 3 h in ice followed by addition of EDTA ( 50 mM ) . Transcripts were extracted using phenol/CH3Cl/isoamyl alcohol ( 25:24:1 ) and ethanol precipitation method and reconstituted in 20 mM HEPES , 5 mM EDTA . Samples were further purified using DEAE column ( binding buffer 100 mM HEPES-KOH , 12 mM MgCl2 , 200 mM NaCl ( pH 7 . 5 ) and elution buffer 100 mM HEPES-KOH , 12 mM MgCl2 , 800 mM NaCl ( pH 7 . 5 ) ) . Fractions containing tRNA were ethanol precipitated , their quality was checked on SDS-urea PAGE and samples were resuspended in 5 mM HEPES-KOH , 1 mM EDTA for storage at -80°C at a concentration of 50 μM . Aminoacylation and enzyme inhibition assays for both KRSs were performed as described elsewhere [17 , 48] . SmtRNALys was refolded prior to enzyme assays by heating at 70°C for 10 minutes followed by addition of 10 mM MgCl2 and slow cooling to room temperature . Aminoacylation buffer for both worm KRSs contained 30 mM HEPES ( pH 7 . 5 ) , 150 mM NaCl , 30 mM KCl , 50 mM MgCl2 , 1mM DTT , 100 μM ATP , 500 μM L-lysine , 18 μM SmtRNALys , 2 U/ml E . coli inorganic pyrophosphatase ( NEB ) and 500 nM recombinant SmKRS or LlKRS protein at 37°C . Reaction at different time points was stopped by addition of 40 mM EDTA followed by transfer to ice . Recombinant maltose binding protein ( MBP ) was used as a control . Cladosporin inhibition assays were performed using inhibitor concentrations in log values ranging from 0 . 01 nM to 10 μM in the assay buffer . Protein melt curve assays for both worm KRSs were performed as reported earlier [49] . Both KRSs were diluted in buffer containing 20 mM Tris ( pH 8 . 0 ) , 200 mM NaCl , 5 mM MgCl2 , 1 mM L-lysine and 2 X SYPRO orange dye ( Life Technologies ) . Total of 20 μM of each ligand AMPPNP ( Sigma ) and cladosporin ( gifted by Bart Staker , SSGCID ) was added separately to 2 μM KRS proteins and incubated in ice for 10 min . Ligand bound and unbound samples of both KRSs were heated from 20 to 96°C at a rate of 1°C min-1 and fluorescence signals were monitored by StepOnePlus quantitative real-time PCR system ( Life Technologies ) . Human-like LlKRS ( 2 μM ) was tested in presence of 20 μM and 200 μM cladosporin concentrations to demonstrate weak binding even with higher inhibitor concentrations . Each curve was an average of three measurements and data were analyzed using thermal shift software ( Life technologies ) . Samples without the addition of ligands were used to determine thermal melting profile of apo proteins . Cladosporin alone and AMPPNP alone in assay buffers , along with no protein controls were used and flat lines were observed for these fluorescence readings at all temperatures . Derivative Tm ( melting temperature ) was used for analysis . ITC experiments were conducted at 30°C in the MicroCal ITC-200 apparatus ( GE Healthcare ) and results were fitted into graph using Microcal origin software . Both parasite KRSs were prepared in PBS ( phosphate-buffered saline ) pH 7 . 4 with 1 mM L-Lys and 2 mM MgCl2 , and cladosporin was solubilized in the same buffer . Cladosporin at concentrations of 240 μM and 360 μM was titrated into 21 μM and 22 . 5 μM protein concentrations of SmKRS and LlKRS respectively . For LlKRS , titrations consisted of 0 . 4 μl of first injection followed by 39 injections of 1 μl with 150 s intervals between injections . For SmKRS , same titrations were performed with 120 s intervals . Titration of cladosporin in buffer alone was performed to determine the change in enthalpy caused by ligand dilution and then subtracted as background from the actual ligand-binding experiments . Limited protein precipitation with LlKRS during our multiple ITC experiments ( in various conditions for n-value optimization trials ) was observed and possibly contributed to the lower n-value , which nonetheless is verified as value of 1 based on the crystal structure of the enzyme-drug complex we present here . Crystallization trials for both SmKRS and LlKRS were performed and crystals of LlKRS were obtained at 20°C by the hanging-drop vapor-diffusion method in 1:1 ratio of LlKRS ( 10 mg ml-1 ) and mother liquor 20% ( w/v ) PEG 3350 , 200 mM magnesium acetate and 10mM spermidine . Thin , plate-shaped crystal clusters soaked in cryoprotectant 20% glycerol were directly mounted in cooled nitrogen gas at 100 K . X-ray diffraction data were collected using a MAR CCD detector on beamline BM14 of the European Synchrotron Radiation Facility , Grenoble , France . A total of 150 images were collected with 1 min exposure and 1 oscillation per frame . The diffraction images were processed and scaled with HKL-2000 program suite [50] . The structure was solved using phaser-MR with HsKRS as template ( 66% sequence identity , PDB: 3BJU ) [51] . Initial model was built by AutoBuild in PHENIX [51] followed by multiple rounds of manual building using Coot [52] in combination with phenix . refine refinement cycles in PHENIX [51] . All structural superimpositions and preparation of figures was conducted using Chimera [53] and PyMol ( http://www . pymol . org ) . Efforts to crystallize SmKRS , though challenging , are ongoing . | The fungal metabolite cladosporin is a potent and selective inhibitor of the malaria parasite protein translation machinery enzyme lysyl-tRNA synthetase ( KRS ) . Our computational annotations of parasitic aaRSs from Loa loa and Schistosoma mansoni provide catalogs of these enzymes that drive parasitic protein translation . We have studied the drug inhibition of KRSs from two neglected tropical worm parasites L . loa and S . mansoni . Our results show that these single copy KRSs from L . loa and S . mansoni can be effectively inhibited by cladosporin with more than 60 fold better binding than for the human counterpart enzyme . Crystal structure of L . loa KRS bound to cladosporin and L-lysine shows key interacting and selectivity residues . This work hence provides a platform for structure-guided derivatization of cladosporin-based compounds for drug development against these neglected diseases . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"sequencing",
"techniques",
"schistosoma",
"invertebrates",
"schistosoma",
"mansoni",
"crystal",
"structure",
"helminths",
"condensed",
"matter",
"physics",
"animals",
"parasitic",
"protozoans",
"acylation",
"protozoans",
"sequence",
"motif",
"analysis",
"molecular",
"biolo... | 2016 | Protein Translation Enzyme lysyl-tRNA Synthetase Presents a New Target for Drug Development against Causative Agents of Loiasis and Schistosomiasis |
The host vasculature is believed to constitute the principal route of dissemination of Neisseria meningitidis ( Nm ) throughout the body , resulting in septicaemia and meningitis in susceptible humans . In vitro , the Nm outer membrane protein Opc can enhance cellular entry and exit , utilising serum factors to anchor to endothelial integrins; but the mechanisms of binding to serum factors are poorly characterised . This study demonstrates that Nm Opc expressed in acapsulate as well as capsulate bacteria can increase human brain endothelial cell line ( HBMEC ) adhesion and entry by first binding to serum vitronectin and , to a lesser extent , fibronectin . This study also demonstrates that Opc binds preferentially to the activated form of human vitronectin , but not to native vitronectin unless the latter is treated to relax its closed conformation . The direct binding of vitronectin occurs at its Connecting Region ( CR ) requiring sulphated tyrosines Y56 and Y59 . Accordingly , Opc/vitronectin interaction could be inhibited with a conformation-dependent monoclonal antibody 8E6 that targets the sulphotyrosines , and with synthetic sulphated ( but not phosphorylated or unmodified ) peptides spanning the vitronectin residues 43–68 . Most importantly , the 26-mer sulphated peptide bearing the cell-binding domain 45RGD47 was sufficient for efficient meningococcal invasion of HBMECs . To our knowledge , this is the first study describing the binding of a bacterial adhesin to sulphated tyrosines of the host receptor . Our data also show that a single region of Opc is likely to interact with the sulphated regions of both vitronectin and of heparin . As such , in the absence of heparin , Opc-expressing Nm interact directly at the CR but when precoated with heparin , they bind via heparin to the heparin-binding domain of the activated vitronectin , although with a lower affinity than at the CR . Such redundancy suggests the importance of Opc/vitronectin interaction in meningococcal pathogenesis and may enable the bacterium to harness the benefits of the physiological processes in which the host effector molecule participates .
Mucosal bacteria possess complex mechanisms of targeting their host cellular adhesion molecules and , in many cases , they adhere to specific host receptors or soluble effectors and consequently acquire enhanced colonisation and virulence potential . Understanding the molecular nature of such interactions may help identify possible targets of future strategies for prevention of infection or for altering the course of pathogenesis . In this study , we have analysed in detail , the molecular mechanisms by which Neisseria meningitidis ( Nm , meningococcus ) , a human pathogen , interacts with serum vitronectin ( Vn ) , which contains an RGD tripeptide sequence recognised by cellular integrins such as αvβ3 and αvβ5 . As these may be expressed by endothelial cells throughout the body , this interaction has the potential to enable Nm to attach to various human vascular barriers , as has been demonstrated in vitro [1] , [2] . Such Nm-endothelial interactions are particularly significant during the course of meningococcal pathogenesis; as , although Nm is a common mucosal coloniser ( up to 30% of healthy human population may carry the organism ) , it is capable of causing disseminated infections in which the host vasculature provides the primary means of distribution to distal tissues . The process of breaching the vascular endothelial barrier may involve cellular entry , and invasion of endothelial cells of distinct origins as has been observed in a number of in vitro studies [2] , [3] , [4] , [5] . It has been shown that meningococcal Opc protein , a major outer membrane adhesin , carries the property of cellular invasion , especially for endothelial cells [1] , [2] , [3] , [4] . Meningococci are capsulate bacteria and the functional efficacy of their subcapsular integral outer membrane adhesins such as Opc ( and Opa , a related but distinct class of opacity proteins ) increases when bacteria become acapsulate or in situations when their cognate target cell surface receptor expression is enhanced . This occurs following conditions in the host that increase the circulation of inflammatory cytokines . Increased receptor density serves to overcome the inhibitory effects of bacterial surface polysaccharides [3] , [6] , [7] , [8] . In addition , in vitro studies on unstimulated cells have also shown that both acapsulate and capsulate phenotypes of Nm may invade human cells in an Opc- or Opa- dependent manner [1] , [2] , [4] , [6] . It appears that in the case of capsulate phenotypes , additional adhesins such as pili ( fimbriae ) that traverse the capsule greatly increase the efficiency of initial binding to target cells; this may then lead to secondary more efficient interactions via the subcapsular bacterial adhesins [4] , [6] , [9] . Opc-mediated interactions have also been shown to result in a level of bacterial transcytosis across the endothelial barriers in vitro [3] , [10] . Although not universally present in Nm , Opc is expressed in numerous clinical isolates of Nm and retained by several meningococcal hypervirulent clonal lineages . Notably , it has been suggested that Opc expression may influence Nm pathogenic process thus altering the clinical profile of meningococcal disease . For example , Nm strains of the ET-37/ST11 clonal complex that lack the opc gene [11] , [12] have been reported to cause serious cases of septicaemia [13] , [14] but have a relatively low tendency to cause meningitis . This has led to the notion that Opc expression may enhance the bacterial ability to cause meningitis; perhaps suggesting an important role of the Opc protein in breaching the endothelial barrier to enter the brain [4] . This notion was further supported by the latter study that used Opc-deficient and Opc-expressing isolates of meningococcal clonal complexes ET-37/ST11 and ET-5/ST32 respectively [15]; the former were not able to cross human brain microvascular cell ( HBMEC ) monolayers in vitro [4] . Opc-deficient bacteria in our earlier studies also did not invade human umbilical vein endothelial cells ( HUVECs ) [2] , [3] . Opc is a transmembrane protein of the beta barrel family with five surface-exposed loops . The protein is basic in nature and has a prominent surface loop 2 . Together , the surface loops of Opc may form a positively charged crevice that may accommodate negatively-charged molecules [16] , [17] , [18] . More recent studies suggest that an induced fit mechanism , involving conformational changes in Opc , may operate for the recognition of molecular targets by the adhesin [19] . Opc has been shown to bind to heparin-like molecules and to heparan sulphate proteoglycans ( HSPG ) on human epithelial cells in vitro [20] . Its mechanisms of targeting endothelial cells are distinct and Nm requires serum factors to bind to HUVECs [1] , [2] . In these studies , the main serum factor involved was identified as vitronectin; fibronectin ( Fn ) also bound directly to Opc+ Nm , but to a lesser extent than Vn . This resulted in the localisation of the bacteria to endothelial RGD-recognising integrins particularly αvβ3 [1] . In contrast , the studies of Unkmeir et al . [4] demonstrated a prominent role of Fn but not Vn in Nm interactions with HBMECs . One notable observation made during our studies on HUVECs was that when using purified Vn , certain but not all preparations of Vn supported bacterial adhesion . As Vn may occur in a number of conformational forms [21] , [22] , [23] , we hypothesised that Opc may only bind efficiently to certain conformational forms of Vn . In this regard , Vn is believed to undergo ligand-induced conformational change to exert its role in key processes of coagulation , fibrinolysis and in the protection of bystander cells against the terminal complement membrane attack complex ( MAC ) [23]; the processes of particular importance during Gram negative bacterial sepsis [24] , [25] . Normally , Vn exists in its closed native conformation ( depicted in summary final figure ) [23] , [26] , [27] which has been shown to undergo conformational change on its activation by some of its physiological ligands which include , thrombin-antithrombin complex ( TAT ) and MAC . In one study , these were shown to be elevated during meningococcal sepsis [28] . Vitronectin conformational activation reveals a number of cryptic epitopes including the full exposure of the heparin binding site at the Vn C-terminal domain and the tyrosine sulphated residues in the connecting region ( CR ) of Vn ( Y-S , final figure ) . Notably , two tyrosines in this region have been shown unequivocally to be sulphated in secreted vitronectin [29] , [30] , [31] . The active or unfolded Vn conformation could also be more favourable for cellular targeting via the cell binding domain ( RGD ) [27] , [32] . In quantitative terms , Vn in fresh plasma is present at ∼300 µg/ml in its native conformation . However , ∼2% may be in its conformationally active form . This activated Vn concentration increases during blood coagulation to ∼7% [22] , [23] , [33] . The principal aim of our study described here was to assess the molecular mechanisms of Opc interactions with Vn and to examine the potential roles of Vn and Fn in Nm adhesion and invasion of human endothelial cells . Our studies have shown that Nm Opc can interact with Vn directly but its efficient interactions can only be shown with activated Vn . We have also demonstrated a novel mechanism of bacterial adhesion to human Vn . This involves direct binding to the Vn tyrosine sulphated region in which the sulphate moieties are necessary for bacterial adhesion . This mechanism may also be used in binding to vitronectin from other sources ( bovine , murine ) . We further present studies comparing direct and indirect binding of Opc+ Nm to activated Vn , the latter occurs via heparin . By uncovering the underlying mechanisms of Opc/Vn interactions , this study additionally provides robust evidence and a rationale for the previous apparently divergent reports regarding the roles of serum proteins , especially Vn , in supporting meningococcal interactions with human endothelial integrins and subsequent cellular invasion . The studies on Vn in relation to Nm seem pertinent also; as notably , meningococcal dissemination is associated with activation of coagulation , fibrinolysis and complement systems; the processes in which vitronectin is intimately involved as a scavenging factor for their end-products [24] , [25] , [28] . Accordingly , during meningococcal sepsis , consumption of plasma Vn has been shown to occur due to its widespread activation and subsequent removal from the circulation [28] . This would suggest that during intravascular spread , meningococci are likely to encounter a continuous supply of activated Vn in circulation . Our current studies have employed acapsulate Nm isolates of serogroup A strain C751 , which have been characterised extensively and used in a number of previous studies on meningococcal-endothelial interactions in order to unequivocally define the role of the Opc protein [1] , [3] , [10] , [34] . In addition , to evaluate the role of vitronectin during infection , we have also used capsulate isolates of a serogroup B strain MC58 representing a phenotype that may be isolated from the blood ( replete with capsule , pili , Opa and Opc ) .
Previous studies have assigned different serum proteins as bridging molecules involved in mediating Nm Opc-dependent interactions with human endothelial cells of distinct origins: Vn for human umbilical vein endothelial cells ( HUVECs ) and Fn ( but not Vn ) for human brain microvascular endothelial cells ( HBMECs ) [1] , [2] , [4] . To clarify the adhesion supporting roles and mechanisms of Opc targeting of serum components , in the current studies , the roles of the serum proteins were further examined . Initially , the ability of Opc+ and Opc− isolates to acquire serum factors from normal human serum was demonstrated by the exposure of Opc+ and Opc− variants of strain C751 to normal human serum ( NHS ) , followed by western blotting of adsorbed bacteria . Using anti-Vn and anti-Fn antibodies to overlay the blots , direct binding of Opc+ Nm to both these serum proteins could be seen; little or no binding was observed with Opc-deficient bacteria ( Figure 1A ) . In further experiments , the relative abilities of acapsulate Nm strain C751 and capsulate MC58 to attach to and invade HBMECs in the presence of NHS were compared . Several features were apparent from these studies . In serum supplemented medium , the strain C751 isolate ( acapsulate and lacking pili and Opa adhesins ) required Opc for cellular adhesion and invasion ( Figures 1B and C ) . Whereas , capsulate MC58 ( with pili and Opa expression ) , did not require Opc for a significant level of cellular adhesion . However , the Opc-deficient isolate of MC58 was not significantly invasive . In this strain however , Opc expression conferred both increased adhesion and invasion properties to the capsulate bacteria ( Figures 1B and 1C ) . Cytochalasin D controls were used to confirm cellular invasion as assessed by gentamicin protection assays [35] ( Figure 1C ) . To assess the efficacy of different serum factors in supporting cellular adhesion and invasion of the above phenotypes , purified serum factors were used to supplement the infection media . We compared the native , folded Vn ( nVn ) as well as conformationally altered activated Vn ( aVn ) in addition to cellular and plasma forms of Fn ( cFn , pFn ) . Of the purified proteins , the most efficient interaction was observed with aVn followed by pFn and then cFn . No significant cellular interactions were observed with nVn or unsupplemented media ( Figure 1D shows the full analysis for Opc-expressing C751 isolate ) . For the capsulate piliated MC58 , similar overall results were obtained . In Figure 1E , adhesion and invasion levels with the most effective serum components are shown for MC58 . As in Figure 1B , a basal level of adhesion was observed for MC58 derivative even in the absence of Opc expression and was largely unaffected by the media supplements; but notably , no significant invasion was seen . The basal level of adhesion of the Opc+ isolate in the unsupplemented medium and that of Opc− isolate under all the conditions tested is largely due to the pilus expression ( see discussion ) . From these studies it is noteworthy that in a physiologically relevant meningococcal phenotype ( MC58 phenotype used here ) , the subcapsular adhesin Opc plays an important role in endothelial cell invasion; it also enhances cellular adhesion . Both these functions appear to be largely dependent on the activated form of serum vitronectin . To identify the binding site of Opc on activated Vn , initially we investigated if the mAb 8E6 could inhibit Opc-expressing Nm binding to the activated Vn by immuno-dot blot assays . Surprisingly , the antibody almost totally inhibited Opc+ Nm binding to aVn ( Figure 4A ) and prevented aVn adsorption from NHS on to Opc+ capsulate meningococcal surface ( Figure 4B ) . Very low levels of Vn adsorbed on to some Opc− cultures could be due to Opc+ phase variants that naturally arise or due to non-specific adsorption of Vn on bacterial surface; although , the data do not rule out the possibility of Vn binding to other bacterial component/s . However , the observed difference between Opc+ and Opc− bacteria is substantial , and clearly significant . In addition , 8E6 significantly inhibited serum-mediated Opc+ Nm adhesion and invasion of HBMECs ( Figure 4C and 4D ) . Several conclusions can be drawn from these observations: that Nm can bind to activated Vn at a site overlapping or closely positioned to the mAb 8E6 binding site , that activated Vn recognised by 8E6 may represent a significant portion of the serum component involved in mediating Nm interactions with HBMECs ( Figure 4D ) and that other serum components may also participate in supporting bacterial binding to human endothelial cells but to a relatively lesser extent than activated Vn . Relatively mild acid hydrolysis of Vn using conditions that destroy the labile O-linked sulphate groups has been reported to diminish the binding of the mAb 8E6 [30] . This mAb has been shown to require sulphation of the tyrosine residues ( Y56 and Y59 ) of the connecting region of Vn for binding [30] , [31] . Accordingly , Vn expressed in E . coli does not bind to 8E6 [22] . To assess if bacterial binding is also affected by such acid treatment of Vn , aVn samples treated at 80°C with 1 M HCl for increasing time periods ( as described in methods ) were dotted on to nitrocellulose and overlaid with either 8E6 , polyclonal anti-Vn antibody or Opc+ Nm . The polyclonal antibody was used to assess if the overall vitronectin structure is affected by the acid treatment . The mAb 8E6 binding and Opc-expressing bacterial interactions ( both acapsulate C751 and capsulate MC58 ) with Vn declined in concert , and no binding of either could be detected when Vn was treated for >15 min ( Figures 5A and 5B ) . In contrast to 8E6 , polyclonal anti-Vn antibody still bound to Vn significantly at the end of acid hydrolysis ( Figure 5A ) . Overall , these data suggest that the binding sites of the mAb 8E6 and Nm Opc share similar characteristics and may be related to the sulphation of tyrosines at sites 56 and 59 . However , it could be argued that acid hydrolysis may also destroy other specific epitopes not detected by the polyclonal antibody . Therefore , to further assess if tyrosine sulphation is required for bacterial binding , we used sulphated and unsulphated peptides spanning the Vn region 48–68 ( Figure 5C ) , termed VA-21S and VA-21 respectively . These peptides otherwise carried no further modifications . As the mAb 8E6 binding requires sulphated tyrosines , only VA-21S would be expected to bind to the antibody and prevent its binding to aVn . Accordingly , in competitive assays , the binding of the mAb 8E6 to the whole immobilised aVn sample was prevented in the presence of added competing VA-21S but not with VA-21 ( Figure 5D ) . Similarly , Opc+ Nm binding to aVn was also specifically and significantly inhibited by VA-21S peptide in a concentration dependent manner but not by VA-21 ( Figure 5E ) . The data demonstrate that Nm Opc interactions with Vn involve an epitope within the CR residues 48–68 and require sulphation of tyrosines for effective interactions . The two peptides were also assessed in their ability to inhibit aVn-dependent bacterial binding and invasion of HBMECs . VA-21S , but not VA-21 , inhibited Opc+ Nm adhesion to the endothelial cells , it also dramatically inhibited invasion of HBMECs by both capsulate and acapsulate Nm ( Figure 5F ) . As the 21-mer peptides ( VA-21/S ) used above proved to be resistant to immobilisation on a variety of solid surfaces , further experiments were carried out on biotin-tagged 26-mer peptides spanning the Vn region 43-68 ( Figure 6A ) . It is noteworthy that these peptides do not contain any of the heparin binding domains of Vn ( see final figure ) . In addition to the unmodified peptide ( VA-26 ) , both sulphated ( VA-26S with Tyr56S , Tyr59S ) as well as phosphorylated ( VA-26P with Thr50P , Thr57P ) were derived . The latter peptide design was chosen as protein kinase CK2 ( casein kinase 2 ) has been reported to phosphorylate T50 and T57 of Vn extracellularly [39] , [40] , [41] . Initially , we compared the effects of the three peptides in a competitive ELISA to assess their ability to inhibit Opc+ Nm binding to immobilised aVn . In addition , we also subjected the peptides to mild acid-hydrolysis to remove the sulphate residues from VA-26S . Such a treatment ( <20 minutes in 1 M HCl at 80°C ) was not expected to affect the phosphorylation of the VA-26P peptide as prolonged hydrolysis >1 . 5 h in 6 M HCl at 110°C in vacuum has been shown to be required to extract intact phosphothreonines and phosphoserines from proteins [42] , [43] . Accordingly , mAb 8E6 binding was abrogated but the phosphate residues on VA-26P were not at all affected by acid-treatment and the anti-phosphothreonine antibody binding remained unchanged ( Figures 6B and S1 ) . Concurrently , only the untreated VA-26S inhibited Opc+ Nm binding to aVn ( Figure 6B ) . The data suggest that the VA-26S peptide behaved identically to the slightly shorter untagged peptide ( Figure 5E ) and that neither the phosphorylated nor the unmodified peptide could interfere with bacterial binding to aVn . To assess if bacteria could bind directly to peptides , ELISA plates with immobilised extravidin were loaded with increasing concentrations of the biotinylated peptides . The peptide binding was demonstrated by using 8E6 and anti-Thr-P antibodies ( Figure S2A ) . Binding of the unmodified peptide ( VA-26 ) was determined by indirect ELISA in which extravidin plates were sequentially exposed to the biotinylated peptides VA-26 followed by VA-26S . The binding of VA-26S was then assessed by the use of mAb 8E6 ( Figure S2B ) . When bacterial binding was examined to peptide-loaded plates , the level of Opc+ bacterial binding paralleled the VA-26S peptide-loading , reaching a saturation point at about 1 µg/ml peptide concentration used for immobilisation ( Figures 6C and S2A ) . In addition , Opc+ bacterial adhesion was observed only when the plates contained VA-26S but not with VA-26 or VA-26P peptides ( Figure 6D ) . The above data provide compelling evidence that sulphated tyrosines of vitronectin create a binding epitope for Opc and show that negatively charged phosphorylated threonine residues do not substitute for sulphated tyrosine residues in Opc-mediated binding at the CR of Vn . Whether phosphate groups , when present in addition to sulphate groups , further strengthen the interactions of the synthetic peptides was not investigated . The VA-26 peptides with an RGD sequence upstream of the tyrosine sulphation site , as in the native Vn , were used to examine further the role of the Vn CR in Opc+ Nm targeting of endothelial integrins . Using these peptides , specific and significant binding of Opc+ bacteria to HBMECs in the presence of VA-26S but not the other peptides could be demonstrated ( Figure 6E ) . In quantitative adhesion and invasion experiments , adhesion to and invasion of HBMECs by both capsulate and acapsulate meningococci was enhanced by VA-26S in a RGD and tyrosine-sulphation-dependent manner ( Figure 6F and G ) ; as both VA-26 peptide with an RGD sequence as well as RGDS peptides incubated with HBMECs prior to the addition of VA-26S precoated bacteria inhibited bacterial interactions . These studies highlight the sufficiency of sulphation of the tyrosines and an RGD sequence as presented in VA-26S for meningococcal invasion of human brain endothelial cells . In complementary experiments , we also investigated the effect of VA-21 and VA-21S to inhibit VA-26S-mediated HBMEC interactions . In accordance with its effect on aVn-mediated cellular interactions ( Figure 5F ) , dramatic inhibition of VA-26S-mediated adhesion and invasion occurred with VA-21S; VA-21 had no effect . With VA-21S , the inhibition of adhesion of C751 Opc+ Nm approached 80% and invasion 100% . For piliated capsulate Opc+ MC58 , adhesion was inhibited less well ( ∼40% , as pili continue to assist in cellular adhesion ) but invasion was inhibited by over 80% ( data not illustrated ) . As bovine serum is frequently used as a supplement in growth and infection media and mice are sometimes used for studies on systemic meningococcal infections , vitronectins of these animal origins relevant to Nm studies were examined for conservation of the 8E6/Nm binding sites . The relevant sequences of the human , bovine and murine vitronectins are shown in Figure 7A . Tyrosine at position 56 is conserved in all three vitronectins , and Y59 is present in human and murine but not bovine Vn . There are also significant differences in bovine and murine vitronectins especially in residues down-stream of Y56 . Notably also , there is a tyrosine residue at position 61 in both bovine and murine Vn not present in human Vn . Bovine Vn also does not contain threonine residues at positions 50 and 57 . Initially , the reactivity of the mAb 8E6 with immobilised vitronectins ( all prepared using heparin-sepharose ) from distinct sources was analysed by ELISA . Only human Vn was recognised by 8E6 ( Figure 7B ) . In a previous study , efficient recognition of Vn peptides by 8E6 was shown to be dependent on sulphated Y59 , although the maximum 8E6 binding occurred when both Y56 and Y59 were sulphated [30] . The three Vn samples were also subjected to limited heat treatment to unfold any closed conformers . This treatment did not alter the mAb 8E6 binding pattern ( not shown ) . To assess bacterial recognition of the serum proteins , immobilised Vn preparations were overlaid with Opc+ and Opc− Nm in an ELISA . Opc+ bacteria bound to human Vn to the greatest extent but significant levels of binding to bovine and murine Vn ( compared with Opc− Nm ) also occurred ( Figure 7C ) . In accordance with 8E6 recognition of human but not other vitronectins , inclusion of 8E6 with Opc+ Nm in overlay experiments resulted in inhibition of bacterial binding to human Vn observed above ( Figure 4A ) but not to other vitronectins ( data not shown ) . As with mAb 8E6 binding , prior heat treatment of the vitronectins did not alter the pattern of bacterial binding ( not shown ) . Bovine serum enhanced endothelial interactions of Nm has been reported previously [1] . Further , in accordance with Nm Opc binding to the animal vitronectins , both bovine and murine sera and purified vitronectins also supported bacterial binding to HBMECs ( Figure 7D ) . The mAb B306 against an epitope on the loop 2 of Opc [44] has been shown previously to inhibit Nm Opc-mediated , Vn-dependent interactions with human endothelial cells [2] , [3] . To assess if the direct binding to the various vitronectins can be inhibited similarly by this mAb , first an ELISA was performed using increasing concentrations of the mAb B306 and another mAb , 154 , D-11 , that binds to an epitope in loops 4/5 [44] to inhibit bacterial binding to aVn using immobilised bacteria . Concentration-dependent inhibition of binding of aVn to Opc+ Nm was observed in the presence of B306 but not with the mAb 154 , D-11 ( Figure S3 ) . Similarly , Opc+ Nm binding to immobilised bovine and murine Vn was inhibited with B306 whereas 154 , D-11 had significantly less inhibitory effect on bacterial binding ( Figure 7E ) . Taken together , the data indicate that similar mechanisms may determine the direct binding of Opc to the three vitronectins . To investigate if binding to bovine and murine Vn was dependent also on sulphated tyrosines , the Vn preparations were subjected to acid hydrolysis prior to the examination of Opc+ Nm binding by ELISA . In each case , acid hydrolysis for 20 min resulted in abrogation of bacterial binding ( Figure 7F ) consistent with the role of Y-S in bacterial binding . Further in a competition ELISA , VA-21S peptide significantly inhibited Opc+ Nm binding to immobilised bovine as well as murine Vn ( Figure 7G ) . Thus overall , the data indicate somewhat variant requirements for Nm Opc and the mAb 8E6 binding to the human and animal vitronectins , however , sulphation of tyrosine appears to be important in each case . Besides the sulphated tyrosines , Vn molecule has been shown to contain several heparin-binding domains ( HBD ) , one of which , located in the C-terminal region spanning the residues 341–380 has been shown to mediate high affinity interactions with heparin ( see schematic diagram final figure , HBD ) [45] . As Opc is also a heparin-binding protein [20] , that Opc may bind to the HBD of Vn via a heparin bridge is entirely possible; this has been implied from earlier studies [46] . To investigate the possible and relative roles of both the CR and the HBD of the activated Vn in Opc binding , we initially analysed if any heparin or heparin-like molecules become attached to bacteria grown on solid agar-containing media as bacterial acquisition of charged molecules could alter the observed effects of added heparin [47] . However , no differences were observed between C751 isolates grown on GC-agarose medium with highly purified agarose as a gelling agent ( see Methods ) , or those grown on HBHI ( Figure S4 ) . In an attempt to analyse the relative preference for bacterial binding directly to the Y-S of Vn or HBD through heparin , experiments were performed using heparin and VA-21S as inhibiting agents . Inhibition of aVn binding to bacteria was observed at increasing heparin as well as VA-21S concentrations ( Figure 8A1 ) , this may suggest a common site on Opc for binding directly to Vn and to heparin . On a closer examination of the effect of heparin when used at low concentrations in ELISA , a biphasic effect was observed: an initial decrease followed by a small rise in binding which was not sustained as heparin concentration increased and eventually there was an overall decrease in bacterial binding to aVn ( Figure 8A2 ) . This biphasic effect of heparin suggested that in the presence of low levels of heparin , two competing interactions with aVn may occur . Some bacterial binding may continue to occur at the tyrosine sulphated site; in addition , binding to HBD via the bound heparin may also come into play . An interesting observation from this study was that the presence of heparin did not produce more binding to aVn than observed in its absence at any concentration . This might suggest that while the presence of limited amount of heparin in an ELISA may alter the site of bacterial binding on aVn , this alternative binding is of a lower affinity than the binding to sulphated region of Vn CR . To finally demonstrate the dual mechanisms of Vn targeting by Opc , we raised and tested the following hypothesis , taking into account that a single site of Opc may bind to the sulphated regions of Vn and heparin . If Opc+ Nm were precoated with heparin and the excess removed , bacterial adhesion could only occur via the heparin-binding domain/s of Vn as the Y-S-binding region of Opc would be fully occupied with heparin . In the presence of excess heparin , the HBD domain of Vn should also be saturated . This should prevent all Opc interactions with Vn . However , it has to be noted that an induced-fit mechanism of ligand-targeting might be involved in binding of Opc to its ligands [19]; in this case , some differences may exist in binding of Opc to the CR of Vn and to heparin . To test this hypothesis , aVn was overlaid with uncoated or heparin-precoated bacteria in the absence or presence of excess 8E6 to block the Y-S site or excess heparin to block the HBD of Vn . The direct binding of uncoated Opc+ Nm was inhibited by the addition of 8E6 as well as in the presence of excess heparin . Pre-coating Opc+ bacteria with heparin before measuring their binding to activated Vn shows lower binding ( by ∼50% ) than uncoated bacteria ( 4th bar , Figure 8B ) suggesting a lower affinity of binding through this route . In addition , unlike uncoated bacteria , the presence of 8E6 does not affect the level of binding of heparin-precoated bacteria ( also represented by the 4th bar , Figure 8B ) . As the mAb 8E6 binds to the Y-S region , it may not be expected to interfere with Nm binding at the distally located HBD domain in the activated form of Vn . Excess heparin ( with or without 8E6 ) itself by binding to HBD , competes with heparin-coated Nm and almost completely inhibited bacterial binding to Vn in accordance with the above hypothesis . This is consistent with the binding of coated and uncoated bacteria at distinct regions of Vn . In addition , the direct binding to sulphotyrosine region of Vn has already been established above ( Figure 6 ) . A proposed sequence of events consistent with the data is illustrated in Figure 8C . Figure 9 summarises some of the Vn properties and illustrates the binding sites for Opc+ Nm on activated human vitronectin .
It is well documented that in non-piliated Nm , the capsule and sialylated LPS dampen the interactions of subcapsular adhesins with their cell-expressed receptors [2] , [3] , [20] . However , in this setting , additional expression of pili can overcome such dampening effects to bring the outer membrane adhesins and host cell receptors in close apposition [6] , [7] , [8] . In order to assess the potential of Opc in capsulate as well as acapsulate bacteria to bind to Fn or Vn conformers present in human serum , in the current study , we employed two distinct meningococcal strain isolates: those of Nm strain C751 with no surface expression of capsule and those of strain MC58 with capsule and intrinsically sialylated LPS . The latter also expressed pili , Opa and Opc , representing a phenotype of relevance in in vivo settings . The data show that the Nm Opc in such acapsulate and capsulate bacteria can bind directly to human serum factors and mediate endothelial cell adhesion and invasion . Our unpublished studies also show that , in capsulate Nm , interactions of subcapsular adhesins with the soluble serum effectors ( e . g . Vn ) are not dependent on pilus expression; whereas their further interactions with cellular receptors are facilitated by the presence of pili . Such a role of pili in initiating the initial cellular engagement and leading to facilitation/potentiation of the interactions of Nm with human cells via integral outer membrane proteins has been established in our previous studies on epithelial cells [6] , [7] , [8] . Overall , from the current studies , the levels of serum Fn bound to Opc+ Nm appear to be lower than of serum Vn and consequently serum-dependent HBMEC adhesion and invasion are significantly inhibited with 8E6 , an aVn-specific mAb . In addition , none of the purified Fn preparations supported adhesion or invasion to the extent observed either with serum or certain purified Vn preparations ( i . e . aVn ) . These data confirm that Opc can bind to human fibronectin but also suggest that Vn may be the preferred target , as observed previously [1] . Vitronectin occurs in two distinct conformations , a native conformation with a cryptic heparin binding site and an unfolded , active conformation with fully exposed binding sites for heparin/heparan sulphate and the mAb 8E6 ( see Figure 9 and associated citations ) . In our earlier studies it had been noted that some , but not all , preparations of Vn , supported efficient direct binding of Opc-expressing Nm to Vn ( unpublished observations ) . As Opc is a heparin-binding protein and as it has been implied that Opc may require sulphated bridging molecules to bind to Vn [46] , detailed investigations were undertaken to assess the direct as well as the indirect binding requirements for Opc+ Nm and the potential role of Vn in supporting adhesion and invasion of endothelial cells of human brain origin . Opc may bind to Vn and Fn via a heparin bridge [20] , [46] . However , Nm Opc clearly also interacts with Vn directly , as this occurs in the absence of added heparin . Notably , growth conditions did not result in prior acquisition of sulphated contaminants ( as occurs with Neisseria gonorrhoeae [47] ) , that could account for this observation , which could be specifically inhibited by anti-Vn mAb 8E6 and by heparin . The mAb 8E6 has been shown unequivocally to bind to the CR of Vn and require the sulphation of the tyrosine residues Y56 and Y59 in the Vn connecting region for efficient interactions with Vn [22] , [30] , [53] . However , 8E6 may inhibit bacterial binding via steric hindrance as in the case of its inhibition of the urokinase-like plasminogen activator receptor ( uPAR ) which binds to a region in the nearby somatomedin B domain [30] . Further evidence for the involvement of the 8E6 epitope as a major binding site for Nm Opc came from the use of acid-desulphated Vn which no longer supports either Opc+ bacterial binding or mAb 8E6 binding presumably due to the loss of acid-labile O-linked sulphated residues [30] . Acid desulphation under mild conditions does not destroy the protein antigenicity in general as shown in the current study ( Figure 5 ) . Studies from other groups also report that acid desulphation does not alter uPAR or PAI-1 binding to Vn [29] . It is also worth noting that any phosphorylated amino acids present ( such as threonine 50 and 57 ) may not be affected by the acid-treatment employed as O-linked phosphates are highly resistant to acid hydrolysis [42] , [43] ( as also demonstrated in Figure S1 ) . In addition , in our studies , sulphated peptide spanning the Vn residues 48–68 ( VA-21S ) or the biotinylated VA-26S ( spanning the residues 43-68 ) significantly inhibited the binding of Opc+ Nm to aVn . The equivalent non-sulphated peptides ( VA-21 or VA-26 ) were only weakly inhibitory . Further , we also analysed the effect of a synthetic peptide of Vn modified at Thr50 and Thr57 with phosphate groups ( VA-26P ) , as Vn can be modified at these residues by the action of CK2 [39] , [40] , [41] . This peptide did not inhibit bacterial binding to aVn even at high concentrations . The VA-21/S peptides could not be immobilised effectively on to solid surfaces; therefore the biotinylated VA-26/S/P peptides tethered to extravidin immobilised on ELISA plates were used to assess the direct binding of Nm . In this system also , Opc+ Nm binding to VA-26S but not to VA-26 or to VA-26P could be seen . The data show clearly that Opc+ Nm can target residues within the connecting region of vitronectin and this is facilitated by the sulphated tyrosines of the connecting region . The RGD motif in the 26-mer peptides also enabled immobilisation of the peptides on the endothelial integrins and examination of the adhesion and invasion-supporting roles of the 26-mer VA-26/S/P peptides . From these studies it is clear that sulphated tyrosines and the RGD motif as in VA-26S peptide are required for mediating Opc-dependent cellular adhesion and invasion of both capsulate and acapsulate meningococci . Tyrosine O-sulphation is a common post-translational modification of secretory and membrane-bound proteins which takes place in trans-Golgi network and is catalysed by membrane-anchored tyrosylprotein sulphotransferases . The classes of proteins that contain sulphotyrosines include G protein–coupled receptors , adhesion molecules , coagulation factors , hormones and extracellular matrix proteins [31] , [54] . In vitronectin , the residues Y56 and Y59 have been unambiguously determined as the tyrosine sulphation sites [29] , [30] , [31] . It has been proposed that the sulphated tyrosine residues of Vn have a role in its conformational stability as the highly acidic tyrosine sulphated region binds intramolecularly to the highly basic heparin-binding region at the C-terminal end of Vn thereby locking vitronectin in an inactive conformation ( as depicted in Figure 9 ) [29] , [31] . There is increasing evidence that sulphotyrosines partake in protein-protein interactions in the extracellular space and the modification has been shown in many cases to be required for efficient ligand-receptor interactions . Its importance has been reported particularly for chemokine receptor functions [55] . Interestingly , HIV enters macrophages and T cells via interacting with sulphated residues of the chemokine receptor CCR5 [56] . However , to our knowledge , such an involvement of sulphated tyrosines in bacterial targeting of host receptors and pathogenesis has not been previously reported . As sulphated residues are involved in mediating both the direct binding to the CR and in indirect binding to HBD , it is very likely that the same region of Opc is involved in these two interactions . If a common site binds to the sulphates of these diverse receptors , it may be envisaged that , at certain low heparin concentrations , bacterial binding to the HBD could occur via a heparin bridge and , a level of bacterial binding could also occur directly to the sulphated tyrosines via Opc sites unoccupied by heparin . On the other hand , in the presence of excess heparin , all Y-S binding sites on Opc may be occupied and in addition , the HBD on Vn may also be blocked by heparin . Thus , Opc-mediated interactions may be totally inhibited . This was indeed the case ( data in Figure 8 ) . In addition , in the absence of any free heparin , heparin precoated bacteria ( which can only bind to Vn via the HBD ) bind less effectively to Vn than uncoated bacteria suggesting that binding to HBD via heparin is of a lower affinity than direct binding to the Y-S site in the Vn connecting region . As iterated above , in previous studies to demonstrate the requirement for sulphated ligands in indirect Opc/Vn interactions [46] , low levels of Opc+ Nm binding to Vn occurred in the absence of added ligands . However , in the presence of dextran sulphate relatively higher binding to Vn was observed . Notably , in these studies , heparin was less efficient than dextran sulphate for the indirect binding of Opa-expressing N . gonorrhoeae to Vn and Fn . Thus , affinities of such different sulphated ligands for Opc ( and Opa ) may vary considerably and in turn affect the efficacy of vitronectin targeting . As our current studies have used heparin solely , we can only draw conclusions about the relative affinities for heparin-dependent and independent interactions of Opc with Vn . It is pertinent to note also that Opc+ Nm binding to native Vn did not change in the presence of added heparin ( data not illustrated ) , and the binding levels of Opc+ and Opc− phenotypes were similar to those shown in Figure 4 for direct Opc+ Nm interactions with native Vn . This suggests that the activated , unfolded vitronectin is the only target for effective direct or indirect binding of Nm . Vitronectin is evolutionarily highly conserved [23] . Examination of the sequences within the CR of human , bovine and murine vitronectins revealed that Y56 is conserved in all three proteins but Y59 is not present in bovine Vn . However , we observed experimentally that the mAb 8E6 binding requirements were only present in human Vn . Nonetheless , Opc+ Nm binding to both bovine and murine vitronectins was found to be significantly higher compared with Opc− Nm . As acid hydrolysis destroyed these interactions , Y56S residue and other tyrosines located at Y59/Y61 may significantly contribute to Nm binding to bovine and murine vitronectins also . Tyrosine sulphate modification of proteins is not uncommon among eukaryotic proteins and it has been observed that sulphated tyrosines are often grouped together in defined clusters and are frequently flanked by acidic residues [54] . The environments in the bovine and murine vitronectins are thus favourable for sulphation and for bacterial binding but apparently not for 8E6 binding , which clearly has a more strict epitope requirement; accordingly , the mAb has a very limited number of targets especially in human serum [53] . We also observed that , the anti-Opc mAb B306 against the Opc loop 2 inhibited Opc+ Nm binding to all three vitronectins suggesting that the adhesin binds to the three proteins using largely similar mechanisms . Finally , VA-21S ( with human Vn sequence ) also significantly inhibited Nm Opc interactions with the animal-derived vitronectins , also suggesting similar mechanisms of binding to the three vitronectins . Further site-directed mutagenesis of both Opc and of various animal/human vitronectins are required to finally confirm the observations . Vitronectin is one of the more abundant plasma proteins circulating at ∼200–400 µg/ml in humans and makes up 0 . 2–0 . 5% of total plasma proteins [23] , [26] . Only a small proportion ( ∼2% ) has been estimated to be in its activated state in plasma but after coagulation , in serum this may increase to ∼7% [33] . However , conformational changes in the protein occur in vivo as it becomes activated by the interactions of circulating ligands such as TAT and MAC complexes which increase during pathogenic conditions . Among other functions , Vn serves as scavenging receptor for the end products of the haemostasis and complement systems and complexed Vn is rapidly cleared from circulation [23] , [57] . Increased native Vn consumption due to continuous activation of Vn during disseminated Nm infections may indeed occur [28] . The availability in several such complexes for the mAb 8E6 binding site [28] , [36] suggests that Nm Opc may also be able to target any activated Vn sequestered into the complexes , which is under investigation . The frequent conversion of native Vn to activated Vn ensures a constant supply for bacterial binding which could lead to increased cellular interactions at the brain and vascular endothelial interfaces and an increased potential to traverse the barriers as observed in vitro . This study has demonstrated that capsulate meningococci can interact with HBMECs via Opc/Vn bridge; the studies of Unkmeir et al . [4] have shown this to be the case via Fn . Whether such interactions occur in vivo remains to be shown . As murine Vn may support Opc binding , suitable animal models could be used to assess this phenomenon . However , it is noteworthy that the efficacy of subcapsular adhesins in mediating cellular interactions of capsulate phenotypes may only be effectively observed when the bacteria are also piliated ( as discussed above ) . Accordingly , with respect to pilus receptor expression , humanised animal models may be required in order to assess the in vivo significance of the reported findings . Finally , as reported for other pathogens , in addition to increased cellular interactions , vitronectin binding could also enable bacteria to become more serum resistant ( N . J . Griffiths and M . Virji , manuscript in preparation; [58] ) . Thus , the targeting of activated vitronectin may impart to Opc-expressing meningococci the properties of efficient tissue penetration and resistance to host innate and adaptive immunity required for effective dissemination .
N . meningitidis phenotypes with or without the expression of Opc ( Opc+ , Opc− ) were derived from the serogroup A strain C751 and have been characterised and described previously [3] , [10] . These isolates are acapsulate and do not express the other major adhesins Opa or pili . Their outer membrane profiles by SDS-PAGE appear identical other than in the expression of Opc protein . They express pilin monomers which are incorporated in the outer membranes but these are not assembled into pili by EM and are not functionally effective [3] , [7] . Their reactivity with a number of anti-LPS mAbs is also identical , they possess lactoneotetraose in LPS which can be only extrinsically sialylated in serogroup A Nm . As blood isolates are usually capsulate and express pili and the opacity adhesins Opa and Opc , to investigate the role of Opc in adhesion and invasion of endothelial cells by such capsulate bacteria , Opc+ and Opc− isolates of serogroup B MC58 strain were used which were otherwise replete with capsule , pili and Opa expression . These isolates express L3 LPS immunotype which can be intrinsically sialylated [6] . All meningococci were normally grown on brain-heart infusion agar supplemented with heated horse blood ( HBHI ) [35] . This medium contains some sialic acids from the blood supplement and so the LPS of the serogroup A strain C751 can become extrinsically sialylated . In some experiments where indicated , a defined GC-agarose medium ( utilising ultrapure agarose as the gelling agent [59] ) was used for the growth of this strain . In this case , no serum supplement is present , thus there is no source of sialic acid for incorporation into LPS . Notably , no qualitative or significant quantitative differences in Vn targeting were observed when using C751 isolates grown on these two media ( Figure S4 ) . Human brain microvascular endothelial cell line ( HBMEC ) was kindly provided by Dr K . S . Kim , John Hopkins University , USA [60] . Cells were grown in RPMI-1640 medium supplemented with 15% ( v/v ) heat-inactivated foetal calf serum ( FCS ) ( Cambrex ) , 2 mM glutamine , 1 mM sodium pyruvate , 100 U ml−1 penicillin-streptomycin , 1% ( v/v ) MEM non-essential amino acids solution and 1% MEM vitamins solution at 37°C in 5% CO2 . Anti-Vn mAb clone 8E6 ( Millipore ) was used to detect and block the interactions of activated Vn at 1–5 µg/ml and 10 µg/ml respectively . The monoclonal anti-Vn antibody VIT-2 and polyclonal anti-Fn antibodies were obtained from Sigma and polyclonal anti-Vn antibody was from Millipore ( AB19014 ) . Anti phosphothreonine ( Sigma ) ( anti-Thr-P ) was used at a range of dilutions as indicated to detect phosphorylation of threonine residues in the synthetic peptide VA-26P . Anti-Opc mAbs B306 and 154 , D-11 were kindly provided by Prof . M . Achtman and Dr . J . Kolberg respectively [61] , [62] . Where not stated , the antibodies were used at predetermined optimum concentrations ( usually , 1–10 µg/ml ) . Native Vn was purchased from Molecular Innovations . Fibronectin and activated Vn from normal human serum ( purified using heparin-sepharose chromatography ) were obtained from Sigma . Synthetic peptides corresponding to the Vn residues 48–68 ( VA-21 ) and the corresponding sulphated peptide ( VA-21S ) were used in competition experiments . As these could not be immobilised efficiently to solid surfaces , biotinylated peptides were also used . These peptides spanned the Vn residues 43–68 ( VA-26 ) and at positions 45–47 they contained the RGD cell binding sequence as in Vn . The unmodified VA-21 , VA-26 and the phosphorylated VA-26P were obtained from GL Biochem ( Shanghai , China ) . The sulphated peptides VA-21S and VA-26S were synthesised by Cambridge Research Biochemicals ( Cambridge , UK ) . Total cell association ( adhesion ) and invasion were measured by viable count assays as described previously [35] . Briefly , for cell association , bacterial suspensions were incubated with cell monolayers at an MOI of 300∶1 in medium 199 supplemented as required with pre-determined optimum concentrations of serum or purified proteins and peptides for an infection period of 3 h . After this , target cells were lysed in a 1% saponin solution , diluted and plated . Levels of bacterial invasion were determined by gentamicin protection assays and verified by the use of cytochalasin D [10] , [35] . Optimum concentrations of supplements required to support adhesion and invasion were found to be 10% for decomplemented NHS and 10 µg/ml of the purified serum components used . To unfold Vn preparations , in some experiments , samples of purified Vn preparations ( human nVn , bovine and mouse Vn samples ) were treated by heating at 56°C for 30 min . The details of experiments to test the effects of Vn peptides or antibodies on bacterial interactions with HBMECs are described in the text . In some experiments , bacterial adhesion levels were visualised by microscopy after fluorescent labelling of bound bacteria as described previously [63] . Briefly , cell monolayers with adherent bacteria were fixed with methanol and non-specific binding sites blocked with 3% BSA in phosphate buffered saline containing 0 . 05% Tween-20 ( PBST ) . Bacteria were detected by overlaying with a rabbit polyclonal antiserum raised against Nm and secondary antibodies conjugated to rhodamine . Activated Vn , native Vn , or heat-treated nVn were dotted onto nitrocellulose at a starting concentration of 375 ng/dot followed by serial 2-fold dilutions , and non-specific binding sites blocked with 3% BSA-PBST . Nitrocellulose strips were then overlaid with anti-Vn antibodies and their binding was detected using alkaline phosphatase ( AP ) -conjugated secondary antibodies . In some experiments bacterial binding to immobilised Vn was assessed using Opc+ Nm ( 1×1010 bacteria/ml ) which were allowed to bind to immobilised Vn for 2 h . Bacterial binding was detected using polyclonal antiserum against Nm and AP-conjugated secondary antibodies . Relative levels of antibody binding were ascertained by densitometry using Scion Image software ( Scion Corporation , Maryland USA ) . Vitronectin samples and biotinylated peptides were treated with 1 M HCl to remove O-linked sulphate groups from the tyrosine residues [30] . In addition , the effect of such acid treatment on phosphorylated peptides was also monitored . Briefly , 2 . 5 µg of the appropriate Vn stock or 5 µg of VA-26S/VA-26P biotinylated peptides were incubated for up to 20 min at 80°C in 1 M HCl before neutralising with 1 M Tris . The level of sulphation of the human aVn ( or sulphated peptides ) was then determined by immuno-dot blot using the mAb 8E6 . Phosphorylation status of VA-26P after acid treatment was monitored by the use of anti-Thr-P antibody . ELISA plates ( 96 well , Dynex ) were incubated overnight with purified Vn in PBS or in carbonate buffer pH 9 . 5 . For direct binding studies , Opc+ Nm were added at 108 bacteria/well in PBS . The role of heparin was investigated by pre-coating bacteria with 50 µg/ml heparin ( Sigma ) for 30 min , followed by washing to remove excess heparin prior to the addition to wells in the presence or absence of an additional 50 µg/ml heparin , which was present throughout the initial binding period . In all cases , bacterial binding was detected by the addition of a rabbit polyclonal antiserum raised against Nm followed by AP-conjugated secondary antibody . All antibodies were incubated for 1 h at room temperature in 1% BSA-PBS . Plates were developed using SigmaFast p-Nitrophenyl phosphate substrate and absorbance was measured at 405 nm ( A405 ) . Alternative ELISA procedures for detecting the binding of Vn to immobilized bacteria were also performed and the Vn binding detected with polyclonal anti-Vn antibody and developed as above . Where heated native Vn was used for overlay , samples were heated for 30 min at 56°C prior to addition to wells . For biotinylated peptide ELISA , plates ( as above ) were incubated overnight with 10 µg/ml extravidin ( Sigma ) . Biotinylated peptides were then immobilized at concentrations described in the text , for 1 h . Unbound peptides were removed by washing . In some tests , peptide loading was assessed using the mAb 8E6 or anti-Thr-P antibody as described above or using competitive ELISA as described in Figure S2 . Peptide-containing ELISA plates were used to assess the direct bacterial binding at a variety of peptide and bacterial concentrations . Bound bacteria were detected by appropriate antibodies as above . In general , one typical of several independent experiments is shown . At least triplicate determinations were included within each experiment and mean values with standard errors have been shown . | Neisseria meningitidis is a human pathogen that can cross the natural cellular barriers to reach the blood and the brain , causing septicaemia and meningitis . One of its surface molecules , Opc , has the capacity to attach to human cells lining the blood vessels . In vitro , the bacterium can do this by coating itself with the human serum factor vitronectin; and , by mimicking as vitronectin , it can bind to human cellular vitronectin-binding proteins ( receptors ) . In this study , we have investigated the structural features of vitronectin that N . meningitidis recognises; such knowledge could help develop future strategies to control bacterial spread . We describe two different aspects of bacterial binding to vitronectin and demonstrate that one of these is a novel method , which occurs directly through the sulphated tyrosines of vitronectin available only when the molecule presents itself in an unfolded form . Such unfolded or activated vitronectin levels may be elevated during bacterial presence in the blood . Thus our observations imply that when bacteria appear in the blood , their presence may help to generate the form of vitronectin that they require for binding to and invading the cells that line human blood vessels to spread throughout body tissues including the brain . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/bacterial",
"infections",
"cell",
"biology/cell",
"adhesion",
"microbiology/cellular",
"microbiology",
"and",
"pathogenesis"
] | 2010 | Neisseria meningitidis Opc Invasin Binds to the Sulphated Tyrosines of Activated Vitronectin to Attach to and Invade Human Brain Endothelial Cells |
Proteins do not carry out their functions alone . Instead , they often act by participating in macromolecular complexes and play different functional roles depending on the other members of the complex . It is therefore interesting to identify co-complex relationships . Although protein complexes can be identified in a high-throughput manner by experimental technologies such as affinity purification coupled with mass spectrometry ( APMS ) , these large-scale datasets often suffer from high false positive and false negative rates . Here , we present a computational method that predicts co-complexed protein pair ( CCPP ) relationships using kernel methods from heterogeneous data sources . We show that a diffusion kernel based on random walks on the full network topology yields good performance in predicting CCPPs from protein interaction networks . In the setting of direct ranking , a diffusion kernel performs much better than the mutual clustering coefficient . In the setting of SVM classifiers , a diffusion kernel performs much better than a linear kernel . We also show that combination of complementary information improves the performance of our CCPP recognizer . A summation of three diffusion kernels based on two-hybrid , APMS , and genetic interaction networks and three sequence kernels achieves better performance than the sequence kernels or diffusion kernels alone . Inclusion of additional features achieves a still better ROC50 of 0 . 937 . Assuming a negative-to-positive ratio of 600∶1 , the final classifier achieves 89 . 3% coverage at an estimated false discovery rate of 10% . Finally , we applied our prediction method to two recently described APMS datasets . We find that our predicted positives are highly enriched with CCPPs that are identified by both datasets , suggesting that our method successfully identifies true CCPPs . An SVM classifier trained from heterogeneous data sources provides accurate predictions of CCPPs in yeast . This computational method thereby provides an inexpensive method for identifying protein complexes that extends and complements high-throughput experimental data .
Proteins carry out most of the work in the cell , and they frequently do so by interacting with other proteins . Therefore , understanding protein and hence cellular function often entails knowing about various types of protein-protein interactions . This paper describes a method for predicting these interactions using a supervised classification algorithm that learns from a variety of genome-wide data sets . Three classes of methods for predicting protein-protein interactions are described in the scientific literature . The first class consists of docking methods that employ detailed molecular simulations to dock two protein structures . These methods do not scale to the entire genome , both because they require protein structures and because they are computationally expensive . High-throughput computational methods fall into two classes: those that predict direct physical interactions [1]–[9] , and those that predict both direct and indirect interactions ( i . e . , co-membership in a protein complex ) [10]–[13] . The current work focuses on the latter problem: predicting co-complexed protein pairs ( CCPPs ) . We frame the problem as a supervised learning problem , and we train a support vector machine ( SVM ) classifier to discriminate between pairs of proteins that are co-complexed and pairs that are not . The SVM is a non-parametric statistical method for discriminating between two classes of data . SVMs have been applied widely in bioinformatics , in applications as diverse as protein homology detection , alternative splicing prediction , microarray analysis and mass spectrometry analysis [14] . Most relevantly , they have been used successfully to recognize physically interacting pairs of proteins [3] , [5] , [9] . The SVM operates by projecting the data into a vector space and finding a line ( or , more generally , a hyperplane ) that separates the classes in that space . SVMs are motivated by statistical learning theory , which suggests an optimal method for identifying this separating hyperplane . Furthermore , SVMs are part of a class of methods , known as kernel methods , that make use of a specific notion of pairwise similarity ( kernel functions ) to project data into a high-dimensional vector space . The benefits of the kernel approach are three-fold: the kernel function can incorporate prior knowledge of the problem domain; the kernel function can operate on non-vector data such as strings , sets or graphs , and kernel algebra allows us to combine heterogeneous types of data within a single classification framework . The SVM algorithm and its application to biological data is described in an accessible fashion in [15]; a much more detailed description of SVM applications in computational biology is available in [16] . The ability to learn from heterogeneous data is of particular value in the prediction of CCPPs , because so many types of data are relevant to this task . In this work , we define separate kernels that operate on each relevant data type . These include three kernels on protein sequences , three kernels on different types of protein networks derived from high-throughput data , and kernels on gene expression , interologs , Gene Ontology terms , co-regulation and localization data . We combine all of these kernels in a single classifier that achieves state-of-the-art predictive accuracy . In this work , we demonstrate the utility of a particular type of kernel , the diffusion kernel [17] , for predicting CCPPs . The diffusion kernel can be naturally applied to protein interaction networks . Various types of networks , representing protein physical interactions , complexes and genetic interactions , can be identified by large-scale experiments: yeast two-hybrid assays for physical interaction detection [18] , [19] , affinity purification coupled with mass spectrometry ( APMS ) for complex detection [20] , [21] , [22] , and large-scale mapping of genetic interactions [23] . The resulting protein interaction networks have been shown to exhibit several distinctive properties [11] , [23] . First , the degrees of vertices exhibit a power-law distribution , with many vertices having a small number of connections , and few vertices having a large number of connections . Second , the networks belong to the class of small world networks and contain densely connected local neighborhoods . These network properties can be exploited to improve statistical inferences about protein-protein interactions . Tong et al . [23] showed that , although there is small overlap between genetically interacting protein pairs and CCPPs , proteins sharing a large number of neighbors in the genetic interaction network tend to be members of the same complex . Goldberg and Roth [11] used a mutual clustering coefficient ( MCC ) to describe the cohesiveness in the physical interaction network . They showed that vertices with a high MCC are more likely to share an edge and that ranking by MCC improves the accuracy of edge inference . MCC considers the number of common neighbors shared by two vertices , i . e . , it only considers paths of length two . In this study , we generalize upon MCC by using the diffusion kernel , which takes into account paths of all lengths [17] . The diffusion kernel quantifies the distance between two nodes as the weighted sum of all paths connecting them , assigning larger weights to shorter paths . Our experiments show that the diffusion kernel performs much better than MCC in ranking protein pairs . In addition , we show that using a diffusion kernel in the context of an SVM classifier improves upon direct ranking by the diffusion kernel alone , and that a diffusion kernel performs much better than a linear kernel , which also only considers paths of length 2 in a network . Next , we show that integration of different data sources improves the performance of our classifier . The summation of sequence and diffusion kernels yields much better performance than either sequence or diffusion kernel alone . This classifier successfully identifies 4789 out of 10980 positives before producing the first false positive . The addition of features such as co-expression , co-regulation , interolog , co-localization and GO annotation improve the ROC50 score from 0 . 859 to 0 . 937 . We also validate our method using two recently decribed large scale APMS data sets [21] , [22] . When these data sets are not used for training , our predicted positives are significantly enriched with protein pairs that occur in both data sets . After validating our method , we trained two SVM classifiers , one using all available data , and one that excludes GO annotations , and applied both classifiers to all pairs of yeast proteins . The resulting predictions are available through the Yeast Resource Center ( http://www . yeastrc . org/pdr ) [24] .
We derive the labels for our classification task from the MIPS complex catalogue version 18052006 [25] excluding category 550: complexes by systematic analysis . The rest of the MIPS complex catalogue contains manually curated complexes derived from the scientific literature . This manually curated database is believed to be highly accurate and has been used to define gold standard CCPPs in several studies [5] , [10] , [12] , [26] . The MIPS complex catalogue organizes complexes into a hierarchy , with each lower level sub-complex contained within the corresponding upper level complex . Our CCPPs come from the lowest level and are hence the most specific complexes in the MIPS complex catalogue . The set consists of 217 complexes , containing 1190 proteins and 10 , 980 CCPPs . We select negative examples ( non-CCPPs ) at random from among all protein pairs that do not co-occur in any top level MIPS complex [9] , [12] , [26] . The resulting set of negatives may be contaminated with some positive CCPPs; however , given the ratio of co-complexed versus non-co-complexed pairs in the yeast genome , the level of contamination is likely to be low . Several studies [10] , [27] have attempted to remove these false negative CCPPs from the gold standard by requiring that non-CCPPs localize to different cellular compartments . However , Ben-hur and Noble [28] have shown that this strategy constrains the distribution of negative examples in such a way that the classification task becomes significantly easier . We use a data set with the number of negative examples the same as positive examples to compare the performance of various methods , and use a larger data set with a negative-to-positive ratio of 10 to estimate the false discovery rate . The complete collection of labeled examples , as well as all of the kernels described in the next section , are available at the on-line supplement http://noble . gs . washington . edu/proj/coco . A kernel method is an algorithm that can be written such that all occurrences of data vectors appear within a scalar product operation . When this is the case , the scalar product operation <Xi , Xj> can be replaced with a generalized similarity function K ( Xi , Xj ) , known as the kernel function . If the kernel function is positive semidefinite and symmetric , then there provably exists some vector space ( the feature space ) in which the kernel function plays the role of the scalar product . In other words , if Φ defines a mapping from the space that the data resides in ( the data space ) into the feature space , then Φ ( Xi ) ×Φ ( Xj ) = K ( Xi , Xj ) . The kernel function provides an intuitive way to encode prior knowledge about a data set . Furthermore , kernel methods provide a natural way of combining heterogeneous data sources [29] , [30] , because the sum of two kernels is itself a kernel and is equivalent to concatenating the vector representations of each data point in the two corresponding feature spaces . This capability is particularly valuable in the context of predicting CCPPs , because so many types of data are relevant . Predicting edges in a protein interaction or co-complex network presents an additional difficulty for which kernels can provide a solution . Many relevant types of data—protein sequence , gene expression , etc . —concern individual proteins , whereas the predictor evaluates protein pairs . This begs the question , how do we define a similarity between two pairs of proteins , given a similarity function that is defined on single pairs . Several groups have used SVMs to predict protein-protein interactions [5] , [9] and have used a tensor product transformation to derive a kernel on protein pairs from a kernel on individual proteins . Given a kernel K that measures the similarity between two proteins , the corresponding tensor product pair kernel ( TPPK ) Kp is defined as Kp ( ( A , B ) , ( C , D ) ) = K ( A , C ) K ( B , D ) +K ( A , D ) K ( B , C ) . It is straightforward to show that the feature space of Kp defined on protein pairs is equivalent to the tensor product of the feature vector spaces of K defined on individual proteins . In this work , we employ a variety of kernel functions . Three different amino acid sequence kernels are described in the next section . For several vector data types , we use a radial basis kernel ( RBF ) KR ( A , B ) = exp ( −γ∥A–B∥2 ) , with γ = 0 . 5 . Finally , for networks , we use the diffusion kernel , defined as follows . Given a graph G = ( V , E ) , define a generator matrix H:The generator matrix H corresponds to the adjacency matrix with the diagonal entry equal to the negative of the degree of the corresponding vertex . The diffusion kernel matrix KD is then computed as the exponential of the generator matrix: KD = eβH . KD ( i , j ) can be regarded as the sum of probabilities of reaching j from i following all paths from i to j in a random walk . The parameter β controls how rapidly a random walk diffuses away from a vertex . In this study , we use a fixed diffusion parameter of 1 . It can be shown that the exponential of any symmetric matrix is symmetric and positive semidefinite . Therefore , the matrix KD is a kernel matrix . Our CCPP predictor combines eleven different data types , listed in Table 1 . We first describe six data types that describe individual proteins , followed by five data types that describe protein pairs . For the remaining data sets , rather than define a kernel on proteins and then apply TPPK , we directly compute features of protein pairs . We then concatenate the resulting features and apply a radial basis kernel to the resulting vectors . The eleven different kernels are combined in two stages . First , the three sequence kernels and the three diffusion kernels are individually normalized by projecting onto the unit sphere , via . The six kernels are then summed in an unweighted fashion . The TPPK transformation is applied to this summed kernel , and the result is added to the RBF kernel defined on the five pairwise data types . With some abuse of notation , the final kernel can be represented as follows:where A represents the amino acid sequences , the three N's represent the three interaction networks , the five D's represent five pairwise data types , and “:” indicates vector concatenation . We use the publicly available PyML implementation of the support vector machine algorithm ( http://pyml . sourceforge . net ) . Three-fold cross-validation with C = 10 is carried out to evaluate the performance . In each split , each partition contains the same number of positive and negative data points . We measure the quality of a CCPP classifier by using receiver operating characteristic ( ROC ) curves . This curve plots number of true positives as a function of number of false positives for varying classification thresholds . Our performance metric is ROC50 , the normalized area under this curve , up to the 50th false positive . A perfect classifier receives an ROC50 score of 1 . 0; a random classifier receives a score close to 0 . The ROC curve does not take into account the negative-to-positive ratio in the data set . In the application of a classifier , we are often interested in the false discovery rate , the fraction of predicted positives that are false positives . This metric is highly dependent on the negative-to-positive ratio . We report the false discovery rate of the kernel with all features assuming a negative-to-positive ratio of 600 , which is the estimated ratio in the real scenario [26] .
We begin by demonstrating that , when directly ranking protein pairs , the diffusion kernel improves the quality of the CCPP predictor . Goldberg and Roth [11] introduced the hypergeometric mutual clustering coefficient ( MCC ) and showed that it had the best performance among four MCC formulations in ranking high confidence protein-protein interaction edges above low confidence ones . The MCC only considers paths of length two in a network . The diffusion kernel , on the other hand , considers paths of all lengths connecting two proteins . We compare ranking based on the diffusion kernel values with ranking by the hypergeometric MCC . The results in Figure 1A show that the diffusion kernel produces a better ranking than the hypergeometric MCC for three different types of networks—yeast two-hybrid , APMS and genetic interactions . This result demonstrates that taking into account paths of all lengths with the diffusion kernel improves edge inference accuracy . Next , we show that using a supervised learning algorithm improves over direct ranking . We train a support vector machine using the TPPK of the diffusion kernel , and we compare the SVM's performance with that of the simple method of ranking pairs by the diffusion kernel values between the two vertices directly . Figure 1B shows that , for all three types of networks , the SVM classifier performs better . Among the three networks , APMS yields the best performance . This is not surprising because we are predicting CCPPs , which are directly measured by APMS . Finally , we show that , in the context of SVM classification , the diffusion kernel yields better performance than the simple linear kernel . Figure 1C compares the ROC50 plots of SVMs trained using the same networks but two different kernels . To compute the linear kernel , each row of the adjacency matrix of a network is treated as a feature vector , and the inner products between two rows are computed as the corresponding kernel value . Like MCC , the linear kernels consider only paths of length two . We normalize the linear kernels and transform them using TPPK , as was done for diffusion kernels . Figure 1C shows that , in the setting of SVM classifiers , the diffusion kernels perform much better than the linear kernels for all three networks . Different types of high-throughput assays yield complementary information about CCPPs . We therefore trained a single SVM using all three networks simultaneously . Figure 2A shows the results of combining the three networks . We consider two ways to combine the networks: combining adjacency matrices and then performing the diffusion , versus performing diffusions separately on each network and then summing the kernels . Our result indicates that the second approach works better in predicting CCPPs . Combining the three networks into one network fails to preserve the different semantics associated with the three types of edges , leading to worse prediction performance . Indeed , combining the three networks in this fashion leads to even worse performance than is given by the best single network ( APMS ) . In contrast , diffusing on each network separately and subsequently summing the three diffusion kernels improves significantly over the APMS diffusion kernel alone . Finally , we combined the diffusion kernels with the sequence kernels and with the five protein pair data sets . As shown in Figure 2B , the combination of the diffusion kernels with the sequence kernels perform better than both the summation of the three diffusion kernels and the summation of the three sequence kernels . In particular , in a cross-validated test , the sequence and diffusion kernel is able to rank 4789 out of 10980 positives above all negatives . The addition of the RBF kernel based on co-expression , interolog , co-regulation , co-localization and GO annotation features further improves the ROC50 performance from 0 . 859 to 0 . 937 . The accuracy of the top-ranked predictions with the inclusion of additional RBF kernel seems to decrease compared with the TPPK of sequence and diffusion kernels alone , as indicated in the leftmost region of the ROC50 plot . In particular , six pairs are ranked high by the final classifier but are not labeled as positives according to MIPS . We investigated each of these pairs . The proteins ARC40 and ARC35 are annotated to be in complex ARp2/3 complex by the Saccharomyces Genome Database ( SGD ) , and NOP14 and UTP7 are annotated to be in complex U3snoRNP by SGD . Thus , these two pairs are likely true positives missed by the MIPS database . In another top-ranked pair , UTP9 is a component of the U3snoRNP that is involved in processing of pre-18S rRNA , and CBF5 is the pseudouridine synthase catalytic subunit of box H/ACA snoRNPs , which is also involved in rRNA processing . This pair has been identified by two APMS studies [22] , [47] . The classifier with both the RBF kernel and TPPK of sequence and diffusion kernels also predicts PDA1 and KGD1 to be in the same complex with high confidence . PDA1 is the E1 alpha subunit of the pyruvate dehydrogenase complex , and KGD1 is a component of the mitochondrial alpha-ketoglutarate dehydrogenase complex . Both proteins bind to mitochondrial DNA and are part of mitochondrial nucleoid . Finally , for two pairs , this classifier predicts one protein in the kinetochore , DAD2 or DAD4 , to be in the same complex with one protein in the spindle pole body , CNM67 or SPC98 . Although the kinetochore and the spindle pole body are both part of spindle , they are two separate components . The classifier has difficulty distinguishing these two components from each other . Thus the apparently worse performance of the classifier with the additional RBF kernel in the leftmost region of the ROC plot may partially be due to the presence of true CCPPs in the negative training set as a result of incomplete MIPS annotations . A significant concern for any method that simultaneously exploits multiple types of data arises from the increased prevalence of missing data . If each given data type is missing 10% of its entries , then in a data set consisting of four such data sources the probability that a given example will have missing data from at least one source is 1−0 . 95 = 41 . 0% . In our experiments , most of the data sources have a significant proportion of missing data , and the coverage varies across the data sets . Table 1 lists the number of proteins with information available for each type of data used in this study . Not surprisingly , the sequence kernels have the highest coverage , whereas the interolog feature and the three networks have the lowest coverage . We therefore investigated how much the SVM's performance depends on the availability of all the data sources used . We examined the 7 , 510 true positive interactions identified before the 50th false positive for the SVM trained using the summation of the three diffusion kernels . For each correctly predicted protein pair and each of the three networks , we asked whether there exists an edge between the proteins and whether there exists a path of any length between the two proteins . Figure 3 shows that most of the correctly predicted pairs are not directly linked in any of the three networks . Indeed , only a relatively small percentage of the protein pairs are linked by a path of any length in all three networks . These results demonstrate that the SVM is capable of making correction predictions from partial and indirect evidence . In the previous sections , we compared the performance of various methods using a data set with the number of negatives chosen to be the same as the number of positives . In reality , the number of negatives is much larger than that of positives , and the negative-to-positive ratio has been estimated to be around 600 [26] . We next estimate the false discovery rate of the classifier with both the RBF kernel and the TPPK kernel assuming a negative-to-positive ratio of 600 . Ideally , we would like to train on a training set with the number of negatives equal to 600 times the number of gold standard positives in our data set . However , this would involve training an SVM on a data set of 10 , 980×601 = 6 , 598 , 980 protein pairs , which is not computationally feasible . Therefore , we instead perform three-fold cross validation on a data set with the number of negatives chosen to be ten times the number of positives . In the computation of the false discovery rate , each occurrence of a false positive is then multiplied by 60 to simulate a negative-to-positive ratio of 600 . We train two classifiers , one that uses GO term annotations and one that does not . GO term annotations are sometimes derived from experimental observations of physical interactions or complex memberships . Using GO term features may therefore artificially inflate the performance of the classifier . To eliminate the possibility of circularity , we trained a classifier using the TPPK kernel and the RBF kernel without GO term features . On the other hand , biologists may be interested in the best possible predictions we can get using all available data . Especially when we apply the classifier to the prediction of protein pairs without interaction data , the issue of circularity is not a concern , and GO term annotations from sources other than physical interaction or complex membership may be useful in CCPP prediction . Therefore , we also trained a classifier that includes the GO term features . Figure 4 plots the true postive rate ( TPR ) as a function of the false discovery rate ( FDR ) for our classifiers with and without using GO term features . To make this plot , we estimate the false discovery rate separately for each fold of the cross validation with the following formula:FP and TP are the number of false and true positives for a certain threshold , respectively . For a certain number of false positives , there often exist multiple corresponding numbers of true positives . The median TP is used to compute a discrete sequence of observed FDR and TPR . Linear interpolation is then used to compute the TPR for all FDR values between any two adjacent observed FDRs . The average TPR across the three folds for a certain FDR is reported in Figure 4 . Not surprisingly , inclusion of the GO term features improves the performance . With a false discovery rate of 10% , the classifier without using GO term features achieves a true positive rate of 83 . 9% and the classifier using GO term features achieves a true positive rate of 89 . 3% . Advances in tandem affinity purification ( TAP ) followed by mass spectrometry make it possible to characterize complexes on a large scale . Recently , two groups published high throughput identifications of complexes in Saccharomyces cerevisiae [22] , [21] . Krogan et al . [22] identified 7 , 076 CCPPs , and Gavin et al . [21] identified 6 , 531 CCPPs . However , these two data sets have only 1 , 542 CCPPs in common . We used our MIPS training set to train a model based on all our features with these two data sets excluded from the APMS diffusion kernel . We then applied the trained model to the prediction of CCPPs among pairs identified by Krogan et al . or Gavin et al . Our model predicted 4536 pairs to be positive , including 1824 pairs present in the training set and 2712 new predictions . Figure 5 shows the number of pairs in the intersections between the two APMS data sets and our predicted positive data set after removal of pairs in the MIPS training set . Among the 2712 predicted positive pairs , 619 ( 22 . 8% ) pairs are present in both APMS data sets . This ratio is much higher than that in the whole data set ( 8 . 2% ) . Given that the number of pairs in either APMS data set is 10 , 226 , and the number of pairs in both APMS data sets is 839 , if we randomly pick a subset of 2712 pairs , the Fisher exact test p-value of the subset containing at least 619 pairs in both APMS data sets is 4 . 2e–198 . Because the pairs in both APMS data sets are believed to be more reliable than the rest of the pairs in the data set , it is reassuring that the positives predicted by our model are enriched in these reliable pairs . Our model predicted a larger fraction of pairs to be positive in the data set of Gavin et al . ( 48 . 5% ) than in the data set of Krogan et al . ( 38 . 0% ) Collins et al . [48] recently developed a Purification Enrichment score and used this score to combine the two APMS data sets and generate a data set of high accuracy . We compared the overlap between our predicted positives and the data set of Collins et al . Among the 10 , 226 pairs in either APMS data set but not in our training set , 2985 pairs are present in the data set of Collins et al . The 2712 pairs predicted to be positive by our classifier contain 1882 of the 2985 pairs . This large degree of overlap between our predicted positives and the data set of Collins et al . is statisitically significant according to Fisher's exact test ( p<1e–300 ) . Thus , the CCPPs predicted by our classifier are consistent with the results of Collins et al . Qi et al . [26] recently performed an extensive study comparing multiple methods on the prediction of complex co-memberships , physical interactions and co-pathway relationships . The study concludes that , among various classification algorithms , random forests performs the best , with random forest-based k-nearest neighbor and SVMs following closely . We applied our kernel methods to their gold standard data set following their learning procedure . 30 , 000 protein pairs were randomly picked as the training set with 50 from the positive data set and 29 , 950 from protein pairs not in the positive data set . Another 30 , 000 protein pairs were picked randomly from the remaining protein pairs as the test set . The test set also contained 50 pairs randomly picked from the positive data set . This training and testing procedure was repeated 5 times instead of 25 times as done by Qi et al . to save time . Our approach with both the RBF and TPPK kernels has a mean ROC50 of 0 . 69 with standard deviation of 0 . 05 . This is slightly better than the best result ( 0 . 68 ) by Qi et al . Qi et al . published their study before the availability of the two recent large scale APMS studies [21] , [22] . We removed these two data sets from the APMS network and tested on the data set of Qi et al . The mean ROC50 is 0 . 68 with a standard deviation of 0 . 05 . This is similar to what Qi et al . reported as their best performance . Qi et al . simulates a realistic scenario by using a negative-to-positive ratio of 600∶1 in the training set . However , in their setting , each classifier only learns from 50 positive pairs . Because of this relatively small number of positives in the training set , the resulting classifier will likely not generalize as well as a method that learns from all available positive pairs . This is why we instead chose to train on a data set with all available positive training pairs and a negative-to-positive ratio of 10 , and simulate the real scenario by magnifying each false positive by 60 , as described above . Having demonstrated that our method produces accurate predictions , we proceeded to apply the two classifiers described previously—trained with and without GO annotations—to all protein pairs in Saccharomyces cerevisiae , excluding 809 dubious open reading frames and 7 pseudogenes . For the SVM trained without GO term annotations , 19 , 258 out of 17 , 307 , 786 protein pairs are identified using an FDR threshold of 10% , including 3 , 946 pairs that are not already annotated in the MIPS complex catalogue . Figure 6 shows the number of predicted pairs as a function of FDR threshold for both classifiers . As expected , at a given FDR threshold , the classifier trained with GO terms predicts more protein pairs than the classifier trained without GO terms . Both sets of predictions can be downloaded from Yeast Resource Center Public Data Repository ( http://www . yeastrc . org/pdr ) , and all predictions obtained using an FDR threshold of 10% are included in the browseable interface of the repository . We analyzed the novel predictions produced by the classifier trained without GO annotations . First , we divided these novel predictions into two sets: those protein pairs in which one protein is a member of a MIPS complex , and those pairs in which neither protein is in the manually curated MIPS complex catalogue . Among the 3 , 946 novel CCPPs , 3 , 260 pairs are linked to one of the 1 , 237 members of any MIPS complex , and the remaining 686 pairs do not involve any MIPS complex . We began by investigating the extent to which the former set of predictions extend known MIPS complexes . Ideally , a newly identified member of a protein complex would be predicted to co-complex with all known members of that complex . We therefore identified all proteins that are predicted to be co-complexed with every member of a known MIPS complex containing at least five proteins . These predictions are listed in Table 2 . Not surprisingly , the majority of these predictions are not truly novel; rather , they reflect the incompleteness of the MIPS annotation that we used to train our SVMs . In fact , for all predicted new members in Table 2 , we were able to find convincing evidence in the scientific literature supporting the prediction , and the citations are given in the table . All the six predicted new members of the mRNA splicing complex—SMB1 , SNU114 , SYF2 , CLF1 , ISY1 and CUS1—have GO annotations of “nuclear mRNA splicing , via spliceosome . ” The predicted new member of the exocyst complex 160 , EXO84 has the GO annotation “exocyst . ” SOH1 ( MED31 ) , the predicted new member of the mediator complex has been annotated by SGD to be part of the mediator complex . Finally , our classifier also predicted MHR1 to be part of mitochondrial ribosomal large subunit . Although MHR1 is primarily annotated to be involved in homologous DNA recombination and genome maintenance in mitochondria , Gan et al . [49] has shown that MHR1 is present in the mitochondrial ribosomal large subunit fraction separated by sucrose density gradient centrifugation , and the stoichiometry of MHR1 in purified large subunit is roughly equal to that of MRPL1 , a member of mitochondrial ribosomal large subunit . In addition , Gavin et al . [20] , [21] found MHR1 to be associated with mitochondrial ribosomal large subunit proteins by high-throughput APMS studies . Overall , this consistent literature support suggests that our classifier makes meaningful predictions . Note that , for this analysis , we selected predictions by using very stringent criteria , requiring that each predicted new member is predicted to be in a co-complexed pair with every member in the MIPS complex . In principle , we could make a larger number of predictions with a more lenient cutoff . A table listing the predicted new members of the MIPS complexes with at least 50% overlap and two overlapping CCPPs is available in the on-line supplement at http://noble . gs . washington . edu/proj/coco . Finally , we analyze the set of novel predictions for which neither protein is a member of a MIPS complex . At an FDR threshold of 10% , this set contains 686 CCPPs among 200 proteins . These predictions can be represented as an undirected graph , with proteins as nodes and predicted co-complex relationships as edges . We identified predicted new complexes as maximal cliques in this graph , where a clique is a set of nodes with every pair of nodes in the set connected by an edge , and a maximal clique is a clique to which no node in the graph can be added to create a larger clique . In general , finding maximal cliques is an NP-hard problem , but because our network is relatively small and sparse , we were able to perform exhaustive enumeration to identify all maximal cliques . We thereby identified 199 maximal cliques with size of at least 3 , including one clique of size 11 , one clique of size 10 , seven cliques of size 9 , and so on down to 48 cliques of size 3 . Many of these cliques overlap one another . For example , the clique of size 11 and the clique of size 10 share 9 proteins in common . We therefore created a network consisting of just these predicted cliques . This network , shown in Figure 7 , consists of four connected components . We performed GO enrichment analysis on these components with GO::TermFinder [50] and summarized the results in Table 3 . All four connected components have significantly enriched GO term annotations for all three ontologies . The members in these four connected components can be found in the on-line supplement http://noble . gs . washington . edu/proj/coco .
In this paper , we developed multiple kernels from heterogeneous data sources and combined them in an SVM classifier to predict co-complexed protein pairs . We applied the diffusion kernel to the two-hybrid , APMS and genetic interaction networks , and we found that , in all three cases , a diffusion kernel performs much better than a linear kernel or the mutual clustering coefficient ( MCC ) . A diffusion kernel computes the similarity between two vertices by summing over all paths connecting the two vertices with paths of shorter lengths receiving higher weight . In contrast , a linear kernel or MCC only considers paths of length 2 . Our results indicate that taking into account the full network topology improves the prediction of CCPP edges . We also applied our prediction scheme to the protein pairs identified by two recent large scale APMS data sets [21] , [22] . Our predicted positives are enriched with protein pairs identified by both groups with high statistical significance , and are consistent with the highly accurate data set of Collins et al . [48] Our method can thus be used to select a subset of these large scale results with better accuracy and reliability . Different data sources provide complementary information , and each data source may have the best predictive power for a subset of data points . For instance , some protein pairs may have no sequence homologs , and some other protein pairs may not be included in the yeast two-hybrid screen experiments . Therefore , the combination of a variety of data sources has the potential to improve CCPP recognition . Kernel methods present a natural way to combine features by the summation of kernel matrices . Our results show that the TPPK applied to the summation of the sequence and diffusion kernels performs significantly better than either the sequence or the diffusion kernels alone . Inclusion of RBF kernels on five additional data sets improves the ROC50 performance further from 0 . 859 to 0 . 937 . We did not optimize the relative weights of the TPPK and RBF kernels . One future direction is to learn these weights by using semidefinite programming [51] , sequential minimal optimization [52] or semi-infinite programming [53] . The method described here is specifically designed to work well in the presence of heterogeneous data—primary sequence , expression , interaction networks , etc . As such , the method can be applied fairly directly to other well-studied eukaryotic genomes . The minimal requirement for applying this method , or indeed any supervised learning algorithm , to a new organism is the availability of data ( e . g . , protein sequences ) and labels ( a set of known protein-protein interactions ) . In practice , the latter is much more difficult to come by . Typically , a genome with a sufficiently large set of high-quality interaction labels will likely also have available non-sequence data such as high-throughput interaction data and expression profiles . | Many proteins perform their jobs as part of multi-protein units called complexes , and several technologies exist to identify these complexes and their components with varying precision and throughput . In this work , we describe and apply a computational framework for combining a variety of experimental data to identify pairs of yeast proteins that partipicate in a complex—so-called co-complexed protein pairs ( CCPPs ) . The method uses machine learning to generalize from well-characterized CCPPs , making predictions of novel CCPPs on the basis of sequence similarity , tandem affinity mass spectrometry data , yeast two-hybrid data , genetic interactions , microarray expression data , ChIP-chip assays , and colocalization by fluorescence microscopy . The resulting model accurately summarizes this heterogeneous body of data: in a cross-validated test , the model achieves an estimated coverage of 89% at a false discovery rate of 10% . The final collection of predicted CCPPs is available as a public resource . These predictions , as well as the general methodology described here , provide a valuable summary of diverse yeast interaction data and generate quantitative , testable hypotheses about novel CCPPs . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"computational",
"biology"
] | 2008 | Predicting Co-Complexed Protein Pairs from Heterogeneous Data |
The nuclear lamina is the structural scaffold of the nuclear envelope and is well known for its central role in nuclear organization and maintaining nuclear stability and shape . In the past , a number of severe human disorders have been identified to be associated with mutations in lamins . Extensive research on this topic has provided novel important clues about nuclear lamina function . These studies have contributed to the knowledge that the lamina constitutes a complex multifunctional platform combining both structural and regulatory functions . Here , we report that , in addition to the previously demonstrated significance for somatic cell differentiation and maintenance , the nuclear lamina is also an essential determinant for germ cell development . Both male and female mice lacking the short meiosis-specific A-type lamin C2 have a severely defective meiosis , which at least in the male results in infertility . Detailed analysis revealed that lamin C2 is required for telomere-driven dynamic repositioning of meiotic chromosomes . Loss of lamin C2 affects precise synapsis of the homologs and interferes with meiotic double-strand break repair . Taken together , our data explain how the nuclear lamina contributes to meiotic chromosome behaviour and accurate genome haploidization on a mechanistic level .
Correct segregation of the chromosomes during meiosis depends on accurate prearrangement of the homologs that culminates in their precise and unambiguous pairing . Recent studies established that the nuclear envelope ( NE ) plays an important role during these processes . It functions as a platform for telomere driven chromosome rearrangement , which is essential for chromosome pairing and synapsis [1] . This special role requires a general reorganization of the NE which also involves the nuclear lamina , a structural protein network underlying the inner nuclear membrane ( INM ) . Through its multiple interactions with a variety of proteins , the lamina functions in nuclear organization and maintenance as well as regulation of transcription [2] . Because of the many regulatory and structural roles , impaired lamina function is responsible for numerous severe human diseases , collectively termed laminopathies , which are often caused by mutations within the LMNA gene that codes for the A-type lamin proteins [3] , [4] . Mammalian meiotic cells are distinguished by the absence of three of the four lamin isoforms that are typically expressed in differentiated somatic cells [5] . Instead , they express , together with lamin B1 , a unique lamin , lamin C2 , which is a short meiosis-specific A-type lamin isoform encoded by the LMNA gene [6] , [7] . Compared to its somatic counterparts , meiotic lamin C2 is an N-terminally truncated version which lacks the complete N-terminal head including a substantial part of the rod domain . As a consequence the structure differs from that typically observed in somatic lamins [8] , [9] . This N-terminal truncation is quite remarkable , as it concerns protein domains , which have been shown to be crucial for the assembly into higher order structures [10] , [11] . Thus , lamin C2 resembles a “natural deletion mutant” that features unique properties and , consistent with this , shows altered polymerization and higher mobility compared to other A-type lamin isoforms [12] . Strikingly , the distribution of lamin C2 in meiotic cells differs significantly from the typical patterns shown for lamins in somatic cells . While somatic lamins usually distribute evenly along the NE , lamin C2 forms distinct domains within the nuclear lamina of meiocytes [13] . In early prophase I , telomeres attach to and subsequently move along the inner nuclear membrane [14] . During these movements , attached telomeres are permanently embedded within the lamin C2 enriched domains . Therefore , it was suggested that lamin C2 locally modulates the NE to allow proper telomere attachment and/or movement [13] . A first indication for such a role came from a previous study investigating spermatogenesis in Lmna−/− mice that were supposed to lack all A-type lamin isoforms [15] , [16] . In that study , A-type lamins emerged to be essential for male fertility . Furthermore , the obtained results indicated that the general integrity of A-type lamin expression is critical for correct synapsis of homologous chromosomes in male meiocytes [16] . However , Lmna−/− mice actually lack expression of both meiosis-specific lamin C2 and somatic A-type lamins A and C and , as a consequence , show a strong somatic disease phenotype . This matter sets obvious limitations to the interpretation of the results obtained in the Lmna−/− genetic background . For example , in the given genetic background one cannot exclude possible side effects that might arise from defective somatic cells of the male gonad . Therefore , it is not clear whether the observed meiotic phenotype is caused by the absence of meiotic lamin C2 or is rather a result of a more general lamin A/C dependent somatic cell dysfunction [16] . It should be noted , that we recently demonstrated that the Lmna−/− mice aberrantly express an , as yet unrecognized , short progerin-like A-type lamin , a matter which further complicates the interpretation of the former results obtained in the Lmna−/− background [17] . So far , no additional , detailed analysis of the role of A-type lamins for meiotic homologous pairing and recombination in females has been carried out . Hence , the molecular mechanism by which lamins could contribute to homologous pairing and promote normal meiotic progression has remained elusive . To address these issues , we for the present study decided to generate a lamin C2 isoform specific knockout model that , hence in a clear-cut genetic background , allows analysing the direct impact of lamin C2 on meiotic events .
We created a targeting construct to selectively eliminate the lamin C2 specific exon 1a , while all other regions of the Lmna gene were left intact to ensure regular expression of the somatic A-type lamin isoforms ( Figure 1; see also [18] ) . Successful targeting was verified through various approaches ( Figure 1 ) . Correct deletion of the lamin C2 specific exon 1a in heterozygous and homozygous animals was confirmed by Southern blot analysis ( Figure 1B ) . As expected , subsequent RT-PCR and Western blot analysis demonstrated that both lamin C2 mRNA and protein are present in the testes of wildtype males , but are clearly absent from the testes of lamin C2−/− mice . Further immunohistochemical approaches using anti-A-type lamin antibodies confirmed that lamin C2−/− male meiocytes fail to express any A-type lamin . Co-testing for somatic A-type lamin isoforms revealed that somatic lamin A/C expression is virtually not affected by the deletion of the lamin C2 specific exon 1a as in wildtype , heterozygous and lamin C2−/− mice comparable amounts of lamins A and C could be detected in liver cells as well as in somatic cells of the testes ( Figure 1C ) . These results are fully consistent with previous reports demonstrating that lamin C2 is the only A-type lamin expressed in meiocytes [7] . More importantly , they also confirm that the applied exon specific targeting strategy selectively disrupted the expression of meiosis specific lamin C2 , but not of somatic lamins A and C . Detailed analysis of the phenotype revealed that , in clear contrast to the previously described Lmna−/− mice which show severe laminopathy-associated somatic tissue dysfunction [15] , [16] , mice deficient for lamin C2 were fully viable and of normal size and weight . Repeated mating attempts of lamin C2−/− males with wildtype females never produced offspring implying that lamin C2−/− males were completely infertile . Lamin C2-deficient females , however , produced offspring when mated with wildtype males , indicating a sexual dimorphic impact of lamin C2 on fertility ( see below ) . These results are consistent with the histological appearance of the gonads . As in the wildtype , ovaries from 11 days postpartum ( dpp ) and 28 dpp lamin C2-deficient females contained growing diplotene oocytes ( data not shown ) . Lamin C2−/− males , however , had significantly smaller testes compared to wildtype controls ( Figure 1D ) . Detailed histological analyses of testes from adult lamin C2−/− mice revealed that post-meiotic stages were completely absent from the seminiferous tubules ( Figure 1D ) . Consequently , also no sperm were found within the epididymis ( Figure S1 ) . TUNEL assay on lamin C2−/− testes sections revealed a high frequency of cell death in regions of the seminiferous tubules , where prophase I stages predominated ( Figure S1 ) , indicating that mutant spermatocytes are unable to complete meiosis and are removed by apoptosis . To investigate the actual function of lamin C2 during gametogenesis in closer detail , we analysed the consequences of its absence on meiotic progression . A key feature of meiosis and an indispensable requirement for correct genome haploidization is the precise and unambiguous pairing of the homologs and their subsequent physical linkage ( synapsis ) mediated by the synaptonemal complex . Chromosome spread preparations of lamin C2−/− pachytene meiocytes revealed frequent defects in synaptic pairing of the homologs in both sexes , although sex-specific differences were observed ( Figure 2A–2C ) . In males , quantification of pairing defects ( Figure 2D ) showed that virtually none ( <1% ) of the lamin C2−/− spermatocytes showed normal meiotic progression , while in wildtype controls >98% of pachytene spermatocytes appeared normal . Univalent chromosomes were clearly the most frequent defect observed , as 96% of mutant spermatocytes showed varying numbers of unsynapsed chromosomes in cells , where fully synapsed homologs were also present ( Figure 2A″ ) . Prominently , the sex chromosomes , where synapsis is reduced to the pseudoautosomal region , remained univalent in 86% of lamin C2−/− spermatocytes ( Figure 2A′″ ) . Further frequently observed phenomena in knockout spermatocytes were heterologous associations between non-homologous chromosomes and associations between telomeres ( in 45% and 52% , respectively; Figure 2A′ , 2A″″ ) . Analysis of lamin C2-deficient females disclosed that synapsis of the homologs was also affected in oocytes , though in a less dramatic manner . While most wildtype mid-pachytene oocytes isolated from 17 . 5 days post fertilisation ( dpf ) embryos achieved full synapsis ( 91% ) , a significant portion ( 32% ) of lamin C2-deficient oocytes showed defective synapsis that , in contrast to the situation in males , most frequently manifested as pairs of homologs that initiated , but did not complete synapsis ( Figure 2B′ ) . In order to exclude a simple delay in synapsis formation at mid-pachytene stage , we also analysed synapsis defects in late-pachytene oocytes ( 19 . 5 dpf ) . In the absence of lamin C2 we found a significantly increased number of late-pachytene oocytes having overt synaptic defects ( wt: 13 . 33% , n = 75; lamin C2−/−: 27 . 38% , n = 84; Pearson's Chi2 test p-value 0 . 033 ) which is similar to the situation found in mid-pachynema . However , consistent with the pronounced differences in phenotypes of males and females , we found an increased average number of chromosomes affected by synaptic defects per cell in males compared to females ( Figure S2 ) . Overall , these results demonstrate that loss of lamin C2 significantly interferes with chromosome synapsis in mammalian meiocytes of both genders with sex-specific differences regarding the severity of meiotic complications . Telomere driven formation and release of the meiotic bouquet at the leptotene/zygotene stage is a well-conserved phenomenon that has been shown to be essential for preparing later events of meiosis [1] , [14] , [19]–[23] . Current models suggest that bouquet formation enhances homologous pairing by increasing proximity of homologous chromosomes . Furthermore , the release of the bouquet conformation may be a means of preventing incorrect associations of non-homologous chromosomes [19] . In recent years , it has been established that meiotic tethering and moving of telomeres within the nuclear envelope ( NE ) depend on SUN- and KASH-proteins [24] , [25] and that this is broadly conserved . These form LINC-complexes , thereby creating a connection between nuclear and cytoskeletal components [26] . For mammals , an involvement of SUN1 and SUN2 in NE attachment of meiotic telomeres has been reported [27] , [28] . Moreover , impairment of telomere attachment has repeatedly been shown to cause chromosome synapsis defects and thus interferes with correct progression of mammalian meiosis [28]–[30] . Nonetheless , the mechanisms by which meiotic telomeres are attached and repositioned have remained largely unclear . Particularly , direct functional evidence for an involvement of nuclear lamins in these processes is missing . Since lamin C2 is enriched at the sites of telomere attachment [13] , an obvious reason for synaptic defects and meiotic disruption as seen in lamin C2−/− males could be impairment of telomere attachment . To address this issue , we used SUN1 , an NE protein known to tether meiotic telomeres [28] , [29] , in co-localisation experiments with fluorescently labelled telomeres to quantify telomere attachment in lamin C2−/− spermatocytes [30] ( Figure 3A ) . Quantifying co-localised and non-co-localised telomere and SUN1 signals revealed no statistically significant difference between wildtype and knockout spermatocytes ( Figure 3B , p-value: 0 . 799 using Pearson's Chi2 test ) . Consistent with this , virtually all telomeres were located at the nuclear periphery in 3D reconstructed nuclei of mutant spermatocytes ( Video S1 ) . Furthermore , chromosome spreads of pachytene-like lamin C2−/− spermatocytes demonstrated that , in fact , all telomeres were connected to SUN1 as all chromosome axes had SUN1 foci on both ends ( Figure 3C , 3D ) . This clearly shows that , even though telomeres are embedded within lamin C2 enriched domains , lamin C2 is dispensable for telomere attachment to the NE . As telomere-driven meiotic chromosome rearrangement , which lead to bouquet conformation and its subsequent release , are prerequisites for intact synapsis formation [31] , we asked whether lamin C2 has a role in movement rather than in attachment of telomeres . In order to address this question , we analysed the temporal behaviour of telomere movements during bouquet formation and release over the first wave of spermatogenesis . Quantification of 3D reconstructed spermatocytes from testes tissue of sequential ages ( 10 to 14 dpp ) with regard to their state of telomere clustering ( Figure 4 , Video S2 ) revealed a distinct bouquet resolution phenotype caused by lamin C2 deficiency . In the wildtype , at 10 dpp , when most cells synchronously reach leptotene/zygotene transition [32] , 74 . 2% of spermatocytes were in the bouquet stage showing a typically clustered telomere pattern ( Figure 4A , 4C ) . Compared to the wildtype situation , at 10 dpp lamin C2−/− mice showed no significant alterations in bouquet frequency ( 77 . 6% ) . Similarly , at 11 dpp there was no difference in bouquet frequency between wildtype and knockout siblings , both showing 57 . 8% of spermatocytes with clustered telomere patterns . Hence , spermatocytes lacking lamin C2 appear to have no problem in attaining bouquet stage . With progression of spermatogenesis bouquet configuration is resolved . Accordingly , in wildtype testes we found gradually decreasing numbers of bouquet stages with 45 . 4% , 28 . 7% and 20 . 5% at 12 , 13 and 14 dpp , respectively . In corresponding lamin C2−/− littermates , however , numbers of spermatocytes in bouquet stage remained significantly elevated . In particular , knockout animals showed 51 . 8% , 45 . 4% and 43 . 9% of spermatocytes with clustered telomeres at ages of 12 , 13 and 14 dpp , respectively . Compared to their wildtype siblings at 13 and 14 dpp , knockout mice roughly showed the 1 . 5 fold and 2 . 5 fold amount of bouquet stages , resulting in statistically highly significant differences at these ages ( mean p-values for 13 and 14 dpp<0 . 01 using Pearson's Chi2 test; Figure 4C ) . Nonetheless , wildtype and knockout animals reached comparable sub-stages of prophase I at 14 dpp as judged by progression of synaptonemal complex assembly ( Figure 4D ) . Thus , our results demonstrate that while NE attachment per se is not affected , the movement of telomeres during bouquet release is significantly delayed in male mice lacking lamin C2 . Since release of the bouquet is thought to promote resolution of incorrect chromosomal associations [31] , impairment of timely bouquet resolution appears to be the basic mechanism responsible for the meiotic defects observed in lamin C2-deficient mice , i . e . defective synapsis formation . Consistent with the findings reported by other groups , our observations suggest a similar dependency of efficient homologous pairing on telomere clustering and movement in mammals as has been previously described for yeast . In fission yeast it has been shown , that telomere clustering and repositioning is required for efficient chromosome alignment and subsequent association [33] , [34] . Within this line of argument , studies in budding yeast have discussed the roles of Ndj1 and Csm4 in meiotic telomere dynamics [35] , [36] . In csm4 mutants , where meiotic telomeres are still associated with the NE , the absence of telomere-led chromosome movements , rather than altered DSB repair , serves as an explanation for the observed homolog non-disjunction . Because of the previously determined molecular properties of lamin C2 and its ability to alter NE integrity [12] , we conclude that lamin C2 locally modulates NE properties at the sites of telomere attachment to allow efficient directed telomere movement and thus promotes homologous chromosome synapsis . Lack of lamin C2 in turn could reduce local NE flexibility and , by this means , interfere with regular movement of attached telomeres as found here in the lamin C2-deficient background . Since chromosome synapsis , homologous recombination and bouquet formation and release are closely interdependent processes during mammalian meiosis [37] , [38] , we then asked whether loss of lamin C2 has a direct effect on recombination as well . To assess lamin C2 function in recombination we next examined selected markers of DSB repair . In early meiotic prophase , sites of DSBs , introduced by SPO11 , become strongly labelled by γH2AX , a phosphorylated H2A histone variant associated with unrepaired DSBs . During leptonema and zygonema it is found in large domains around the DSBs . As meiotic prophase I progresses , γH2AX labelling successively disappears from the autosomes and , in the male , becomes restricted to the sex chromosomes [39] ( Figure 5A ) . In pachytene-like lamin C2−/− spermatocytes , however , γH2AX remained associated with most of the chromosomes in a cloud-like manner , indicating that meiotic DSBs are formed , but are not efficiently repaired . Interestingly , sex chromosomes , although they often failed to synapse in knockout spermatocytes , showed strongly γH2AX labelled chromatin ( Figure 5A′ ) . Further analysis of later stages of DSB processing revealed that in lamin C2-deficient pachytene-like spermatocytes numerous RAD51 and RPA signals , which mark early and intermediate stages of DSB repair [37] , aberrantly persist along both paired and unpaired axes ( Figure 5B′ , 5C′ ) . In wildtype late pachytene spermatocytes MLH1 , a component of late recombination nodules and a marker of presumed crossing overs [40] , appeared with at least one distinct MLH1 focus per pair of synapsed homologs . These foci were consistently absent from paired and unpaired chromosome axes of lamin C2−/− spermatocytes ( Figure 5D , 5D′ ) . Overall , this indicates that recombination events are initiated in lamin C2-deficient spermatocytes , but repair is not efficient or complete and functional crossing overs do not form . Nonetheless , the formation of sex body specific chromatin does occur even if synapsis between sex chromosomes is defective . This affirms that lamin C2−/− spermatocytes initiate , but fail to complete , pachynema , a phenomenon observed in numerous knockout models of meiosis-specific proteins [41] . Inducing apoptosis during mid-pachynema of spermatocytes carrying defects , such as incomplete synapsis , points to an activation of a pachytene-checkpoint mechanism preventing defective germ cells from further maturation [42] . Aberrant persistence of RAD51 and RPA on chromosome axes and the observed lack of MLH1 in lamin C2-deficient spermatocytes might be a consequence of the activation of the male pachytene-checkpoint and the subsequent elimination by apoptosis in mid-pachynema rather than a direct effect of lamin C2-deficiency on maturation and completion of homologous recombination per se . Mammalian oogenesis apparently lacks a stringent checkpoint operating at mid-pachytene stage and , thus , allows for analysis of recombination in late stages of meiotic prophase I [43] . Hence , we performed a detailed examination of recombination in lamin C2−/− oocytes at 19 . 5 dpf . Consistent with the above described findings , lamin C2-deficient oocytes of 19 . 5 dpf embryos reached late pachynema despite the persistence of synaptic defects ( Figure 5F″ ) . Additionally , temporal progression through meiotic prophase I stages per se appeared normal as , when compared to the wildtype , the ratio of pachytene to diplotene oocytes was not significantly altered in lamin C2−/− ovaries ( n>50 each; Pearson's Chi2 test p = 0 . 5478 ) . As in the males , γH2AX staining persisted until mid-pachytene stage , thereby surrounding incompletely synapsed chromosomes in lamin C2−/− oocytes from 17 . 5 dpf embryos ( Figure 5E , 5G ) . In contrast to males however , lamin C2-deficient oocytes from 19 . 5 dpf embryos were able to recruit MLH1 onto chromosome axes at late pachynema ( Figure 5F ) . Strikingly , quantification of MLH1 , revealed a significant reduction of MLH1 foci in lamin C2−/− females , indicating a reduced rate of meiotic recombination ( Figure 5H ) . Moreover , while 80% of wildtype late pachytene oocytes had at least one obligate MLH1 focus on each bivalent , 46% of lamin C2-deficient oocytes completely failed to recruit MLH1 at least on one pair of homologs , regardless whether or not they exhibited synaptic defects ( Figure 5I ) . Absence of MLH1-marked recombination nodules indicates the lack of cross over recombination on the affected chromosomes that , in the female , may account for increased chromosome segregation defects at later stages of meiosis [40] , [44] , [45] . Notably , as shown here , lamin C2−/− females are fertile despite the fact that lamin C2−/− oocytes revealed overt defects in homologous recombination and chiasmata formation . To some extent this matter resembles the situation described for Sycp3−/− mice that show a severely disrupted synapsis and reduced chiasmata formation . Similar to the lamin C2−/− mice presented here , males deficient for SYCP3 are completely infertile . In female Sycp3−/− mice , by contrast , oocytes form chiasmata , but at a lower level than in the wildtype , which results in a reduction , but not a complete loss , of fertility [46] . Regardless of the sexual dimorphic impact on fertility , our analyses clearly demonstrated that the meiotic nuclear lamina has a central role in regulating chromosome bouquet dynamics and is therefore essential for correct progression of meiotic homologous recombination in both male and female mice . Such interdependencies between bouquet stage resolution and DSB repair and homologous recombination have been described earlier . Mice that are unable to induce meiotic DSBs due to the absence of SPO11 and those showing defects in late recombination events , as is the case in the Mlh1−/− background , also show elevated bouquet frequencies . However , in mice with altered early recombination phenotypes , caused by the lack of recombination proteins DMC1 or HOP2 , no significant increase in bouquet stages could be observed [38] . This suggests , that whilst defects in the formation of DSBs during leptonema or alterations of late recombination events during late pachynema may influence bouquet duration , the intermediate steps of DSB repair and early recombination do not . In lamin C2−/− spermatocytes DSBs are definitely induced but the stages when late recombination events normally occur are not reached due to checkpoint induced apoptosis during mid-pachynema . Therefore , neither altered early DSB repair nor defective late recombination is able to explain the observed delay in bouquet stage release in lamin C2−/− spermatocytes . Since defects in earlier DSB repair and recombination do not influence bouquet stage frequencies , inefficient DSB repair in lamin C2−/− meiocytes is likely to be a consequence of the delay in bouquet stage release , rather than the converse . Consistent with this , loss of telomere-led dynamics also affects recombination and crossing over in budding yeast meiosis [35] . Though to some respect the effects of altered meiotic chromosome movements on crossing over events differ between the budding yeast mutants and our lamin C2-deficient mouse , the significance of directed telomere-led chromosome dynamics per se for homolog recombination and disjunction seem to be as widely conserved as the bouquet formation itself .
All animal care and experiments were conducted in accordance with the guidelines provided by the German Animal Welfare Act ( German Ministry of Agriculture , Health and Economic Cooperation ) . Mouse generation , housing , breeding and experimental protocols at the CNIO , Madrid , were performed in accordance with protocols revised and approved by the Institutional Ethics Committees of the CNIO and following the European Regulation ( 2010/63/UE of September 22 , 2010 ) . Animal housing and breeding at the University of Wuerzburg was approved by the regulatory agency of the city of Würzburg ( Reference ABD/OA/Tr; according to §11/1 No . 1 of the German Animal Welfare Act ) . All aspects of the mouse work were carried out following strict guidelines to insure careful , consistent and ethical handling of mice . To generate a lamin C2 specific knockout mouse line , in which the expression of other A-type lamins is left intact , a replacement vector was constructed to selectively eliminate lamin C2 specific exon 1a and the flanking putative upstream promoter elements . Therefore , a 4 kb genomic region including exon 1a was replaced by a neomycin cassette in reverse orientation using a modified pKSloxPNT vector [47] , [48] . The vector for homologous recombination was designed as follows ( see also Figure 1A ) : a 1 . 9 kb genomic fragment located 1 . 5 kb upstream of exon 1a ( amplified from mouse genomic DNA by PCR; oligonucleotides Table S1 ) was cloned into the SalI/KpnI restriction sites downstream of the neomycin cassette and a corresponding 5 kb fragment 2 . 5 kb downstream of exon 1a ( oligonucleotides Table S1 ) was ligated into the EcoRI restriction site between the thymidine kinase and neomycin cassette . The replacement vector , linearized with KpnI , was electroporated into mouse R1 ES cells and recombinant clones were selected in the presence of G418 and gancyclovir as previously described [47] . Positively targeted ES clones were identified by PCR using external primers ( oligonucleotide sequence for genotyping: Table S1 ) and correct targeting was confirmed by Southern blot . For Southern blot analysis 15 µg NsiI ( not shown ) or SmaI digested DNA derived from ES cells ( or tail tips of lamin C2+/+ , lamin C2+/− and lamin C2−/− mice ) was separated on a 0 . 8% agarose gel , subsequently transferred to a nylon membrane and correct targeting was tested with both external ( see Figure 1A ) and neomycin probes . Blastocyst injection of one of the positive ES clones gave rise to germline transmitting chimeras that were mated to produce heterozygous founder mice . Intercrossing of lamin C2+/− founder mice produced offspring with all genotypes in mendelian ratio . To confirm the genotypes we performed RT-PCR , Southern blot and immunofluorescence analysis as described below . Gonads from wildtype , heterozygous and lamin C2−/− mice were either fixed and embedded in paraffin wax for sectioning , frozen in 2-methylbutane for swab preparations or freshly used for chromosome spreads . Gonads for paraffin embedding were fixed in either 1% PBS-buffered formaldehyde ( pH 7 . 4 ) for 3 hours or in 4% overnight . Tissues were then dehydrated in an increasing ethanol series and infiltrated with paraffin at 58°C overnight . Tissue samples for swab preparations [49] were placed in 2-methylbutane at −70°C immediately after dissection . Procedures for chromosome spreads were adapted from de Boer et al . [50] . For this , fresh tissue samples were incubated in hypotonic buffer ( 30 mM TrisHCl , 17 mM Na-citrate , 5 mM EDTA , 50 mM sucrose , 5 mM DTT; pH 8 . 2 ) . For spermatocyte spreads , testes tubules taken from the hypotonic buffer were resuspended in 20 µl of sucrose solution ( 100 mM ) and transferred to a slide covered with 1% formaldehyde solution ( 1% formaldehyde , 0 . 15% Triton X-100; adjusted with NaOH to pH 9 . 2 ) . Slides were incubated in closed moisture chambers for 2 h , followed by 30 min incubation with the lid left ajar; finally slides were dried in the opened chambers . For oocyte spreads , ovaries were transferred from hypotonic buffer to a small droplet of sucrose solution ( 50 µl ) placed on a slide . Ovaries were then decapsulated , fragmented with forceps and incubated with gentle shaking for 10 min to elute oocytes . Debris was removed from the slides and an equal amount ( 50 µl ) of 2% formaldehyde solution ( 2% formaldehyde , 0 . 15% Triton X-100; adjusted with NaOH to pH 9 . 0 ) was added to the droplet of sucrose containing the oocytes . Slides were then incubated in closed moisture chambers for 1 h , followed by a 30 min incubation with opened chambers at room temperature . Finally , slides were dried at 37°C for approximately 2 h . Standard histology was performed on 5 µm sections of paraffin-embedded tissues fixed overnight in 4% formaldehyde as described previously [16] . To visualize and identify apoptotic cells , TUNEL assays were carried out on 10 µm sections of paraffin embedded testes using the ApopTag Fluorescein In Situ Apoptosis Detection Kit ( Millipore , Schwalbach , Germany ) according to the manufacturer's protocol . To show absence of lamin C2 expression in lamin C2−/− mice and to verify that expression of somatic lamins A/C is not affected by lamin C2 isoform specific targeting , we performed RT-PCR analysis on wildtype , heterozygous and knockout mice . Total RNA was isolated from testes suspensions or liver tissues of six week old littermates . RNA isolation was performed using TriFAST™ ( Peqlab , Erlangen , Germany ) according to the manufacturer's manual . 1 µg of total RNA was used for reverse transcription using oligo ( dT ) primer and M-MLV reverse transcriptase ( Promega , Mannheim , Germany ) . Using 1 µl of RT-reaction we performed PCRs specifically amplifying either lamins A/C or lamin C2 transcripts . In case of lamins A/C , transcripts were amplified using a 5′ primer corresponding to the ATG region of the lamin A/C specific exon 1 , whereas for lamin C2 we used a 5′ primer selectively binding to the ATG region of the lamin C2 specific exon 1a . For both transcripts the same 3′ primer was chosen from a region shared by both the lamins A/C and the lamin C2 transcripts ( see Table S1 for oligonucleotide sequences ) . Primary antibodies used in this study were: rabbit anti-lamin A/C ( H-110; Santa Cruz , Heidelberg , Germany ) , rabbit anti-SYCP3 ( anti-Scp3; Novus biologicals , Littleton , CO ) , guinea pig anti-SYCP3 , rabbit anti-SYCP1 , guinea pig anti-SYCP1 [51] , rabbit anti-TRF1 ( TRF12-A; Alpha diagnostics , San Antonio , TX ) , mouse anti-γH2AX ( Millipore ) , mouse anti-RPA ( clone RPA34-20; Calbiochem , Darmstadt , Germany ) , mouse anti-MLH1 ( clone G168-15; BD Biosciences , Heidelberg , Germany ) , rabbit anti-RAD51 ( Calbiochem , Darmstadt , Germany ) and guinea pig anti-SUN1 [52] . The corresponding secondary antibodies conjugated to Cy2 , Texas red , Alexa647 ( immunofluorescence ) or peroxidase ( immunoblot ) were obtained from Dianova ( Hamburg , Germany ) . Additionally , for FISH analysis a monoclonal mouse anti-digoxigenin and an anti-digoxigenin fluorescein conjugated fab fragment ( Roche , Mannheim , Germany ) were used according to the manufacturer's protocol . To assess the efficient disruption of lamin C2 and unaffected expression of lamins A/C in lamin C2−/− mice on the protein level , western blots using cell suspensions from whole testes preparations and from liver tissue ( as a somatic control ) of six week old wildtype , heterozygous and lamin C2−/− littermates were performed . Tissues or cells were resuspended in 2× SDS sample buffer ( 120 mM Tris/HCl , 10% SDS , 20% glycerine , 20% 2-mercaptoethanol; pH 6 . 8 ) and denatured at 95°C for 15 min before applying 5×105 cells for each tissue onto a 12% SDS PAGE . After separation of the proteins with SDS-PAGE , proteins were transferred to a nitrocellulose membrane . Membranes were blocked overnight in TBST ( 10 mM Tris/HCl , 150 mM NaCl , 0 . 1%Tween 20 ) containing 10% milk powder . Anti-lamin A/C primary antibodies were diluted ( 1∶2000 ) in blocking solution and membranes were incubated for 60 min at room temperature with subsequent washing in TBST . Peroxidase-conjugated secondary antibodies were applied as specified by the manufacturer . Bound antibodies were detected using the Western Lightning Plus-ECL Enhanced Chemiluminescence Substrate ( Perkin Elmer , Rodgau , Germany ) . Immunofluorescence analysis was carried out on chromosome spreads , swab preparations from frozen tissue or paraffin sections . Immunofluorescence staining of chromosome spreads were performed according to procedures adapted from de Boer et al . [50] . Blocking of cell spreads was performed with the supernatant of centrifuged ( 16 . 000 g , 30 min ) blocking solution ( 5% milk , 5% FCS , 1 mM PMSF in DMSO; pH 7 . 4 in PBS ) . For double-label immunofluorescence , spreads were then incubated with the first primary antibody followed by washing in PBS before blocking again in blocking solution and incubating with the first secondary antibody . After another blocking step , slides were incubated with the second primary antibody followed by washing , reblocking and incubation with the second secondary antibody . For immunofluorescences on swab preparations , cells were fixed in PBS containing 1% formaldehyde for 10 min followed by permeabilisation in PBS/0 . 05% Triton X-100 for another 10 minutes . After washing and blocking in blocking solution , slides were incubated with both primary antibodies . Following this , slides were washed and blocked again and subjected to both secondary antibodies . To prepare paraffin sections for immunofluorescence , antigen retrieval and removal of paraffin was conducted as described before [53] . After washing the sections in PBS , sections were blocked with PBT ( 0 . 15% BSA , 0 . 1% Tween 20 in PBS , pH 7 . 4 ) prior to the incubation with both primary antibodies . After washing in PBS , sections were subjected to corresponding secondary antibodies . For all preparations DNA was counterstained using Hoechst 33258 ( Sigma-Aldrich , Munich , Germany ) . To directly label telomeres , we performed fluorescence in situ hybridisation using digoxigenin-labelled ( TTAGGG ) 7/ ( CCCTAA ) 7 oligomeres . After initial immunofluorescence staining ( see above ) , spreads were subsequently briefly refixed for 20 min using 4% formaldehyde in PBS and washed in PBS . After rinsing slides in 2× SSC ( 0 . 3 M NaCl , 0 . 03 Na-citrate; pH 7 . 4 ) for 5 min , spreads were incubated in RNase A ( 100 µg/ml in 2× SSC ) at 37°C for 1 hour . After rinsing in 2× SSC , cells were denatured for 20 min at 95°C in the presence of 10 pmol of each labelled probe in 40 µl of hybridisation solution ( 30% formamid , 10% dextrane sulphate , 250 µg/ml E . coli DNA in 2× SSC ) . Hybridisation was performed overnight at 37°C . After washing twice in 2× SSC at 37°C for 10 min , samples were blocked using 0 . 5% blocking-reagent ( Roche ) in TBS ( 150 mM NaCl , 10 mM Tris/HCl; pH 7 . 4 ) . Probes were incubated for 1 h with anti-digoxigenin antibodies ( Roche ) . After washing slides in TBST ( TBS , 0 . 05% Tween 20; pH 7 . 4 ) primary antibodies were detected using Cy2-conjugated anti-mouse secondary antibodies ( Dianova ) . Fluorescence images were recorded using a Leica TCS-SP2 AOBS confocal laser scanning microscope ( Leica Microsystems , Mannheim , Germany ) , equipped with a 63×/1 . 40 HCX PL APO oil-immersion objective , or an iMIC microscope with 100×/1 . 40 NA oil-immersion objective ( Till Photonics , Munich , Germany ) . Confocal images shown are calculated maximum projections of sequential single sections processed in Adobe Photoshop ( Adobe Systems ) . Images for the quantification of telomere clustering were taken using the iMIC and the Live Acquisition software package . 3D reconstruction , analysis and quantification of telomere attachment and clustering were conducted using the three-dimensional reconstruction tool of ImageJ ( version 1 . 42q; http://rsbweb . nih . gov/ij ) . All statistics shown were calculated using R ( version 2 . 10 . 1; http://www . r-project . org ) , Microsoft Office Excel 2007 or StatSoft STATISTICA 10 . P-values were generated using Pearson's Chi2 test or Mann-Whitney U-test with the significance level set <0 . 05 . | Diploid germ line cells have to undergo meiosis to produce haploid gametes . Haploidization involves pairing and recombination of homologous chromosomes as a prerequisite for their proper segregation . Pairing of homologous chromosomes requires their active repositioning within meiotic nuclei , which depends on the interaction of telomeres with the nuclear envelope . This dynamic association is vital for a faithful meiosis and thus crucial for fertility . However , very little is known about the relationship between telomeres and nuclear envelope components . Here , we have investigated the role of the nuclear lamina , a structural scaffold that is intimately associated with the inner nuclear membrane . In somatic cells , the lamina is a key player in chromatin organization and fulfils various functions such as nuclear structure maintenance and regulation of transcription . In order to understand its role in meiosis , we investigated lamin C2 , the only A-type lamin isoform expressed in mammalian meiotic cells . We demonstrate that lamin C2 is essential for timely repositioning of meiotic telomeres . In its absence , synapsis of homologous chromosomes and double-strand break repair are severely affected . These multiple meiotic defects lead to infertility in males . We conclude that the nuclear lamina contributes directly to fertility through facilitating meiotic chromosome movements . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"meiosis",
"cellular",
"structures",
"gene",
"function",
"germ",
"cells",
"animal",
"models",
"model",
"organisms",
"cell",
"nucleus",
"chromosome",
"biology",
"biology",
"mouse",
"cell",
"biology",
"genetics",
"cellular",
"types",
"molecular",
"cell",
"biology",
"g... | 2013 | The Meiotic Nuclear Lamina Regulates Chromosome Dynamics and Promotes Efficient Homologous Recombination in the Mouse |
Distributions of the backbone dihedral angles of proteins have been studied for over 40 years . While many statistical analyses have been presented , only a handful of probability densities are publicly available for use in structure validation and structure prediction methods . The available distributions differ in a number of important ways , which determine their usefulness for various purposes . These include: 1 ) input data size and criteria for structure inclusion ( resolution , R-factor , etc . ) ; 2 ) filtering of suspect conformations and outliers using B-factors or other features; 3 ) secondary structure of input data ( e . g . , whether helix and sheet are included; whether beta turns are included ) ; 4 ) the method used for determining probability densities ranging from simple histograms to modern nonparametric density estimation; and 5 ) whether they include nearest neighbor effects on the distribution of conformations in different regions of the Ramachandran map . In this work , Ramachandran probability distributions are presented for residues in protein loops from a high-resolution data set with filtering based on calculated electron densities . Distributions for all 20 amino acids ( with cis and trans proline treated separately ) have been determined , as well as 420 left-neighbor and 420 right-neighbor dependent distributions . The neighbor-independent and neighbor-dependent probability densities have been accurately estimated using Bayesian nonparametric statistical analysis based on the Dirichlet process . In particular , we used hierarchical Dirichlet process priors , which allow sharing of information between densities for a particular residue type and different neighbor residue types . The resulting distributions are tested in a loop modeling benchmark with the program Rosetta , and are shown to improve protein loop conformation prediction significantly . The distributions are available at http://dunbrack . fccc . edu/hdp .
The empirical distributions of the backbone dihedral angles φ and ψ of amino acids in proteins have been studied for over 40 years . Early efforts were based on determining those regions of the Ramachandran map that are “allowed” and those that are forbidden due to steric conflicts among the backbone atoms or between backbone and the Cβ carbon atom of side chains [1] . This steric analysis has recently been updated and refined by Ho et al . [2] , [3] . Boundaries between populated and unpopulated regions have been used as checks on the quality of newly determined experimental structures in such programs as Procheck [4] , Whatcheck [5] , and more recently MolProbity [6] . Because bond lengths and bond angles vary to only a limited extent ( although more so than is typically assumed [7] ) , protein structures are often treated with only dihedral degrees of freedom in simulations , structure prediction , and structural analysis . Ramachandran data therefore play a central role in developing empirical energy functions for structure prediction [8] and simulation [9] . We distinguish two concepts in analyzing the backbone dihedral angles of proteins . The first is a Ramachandran plot or Ramachandran map , which is simply a scatter plot of the φ , ψ values for the amino acids in a single protein structure or a set of protein structures . It may be restricted to a single amino acid type and/or a single structural feature type , such as protein loops . The second is a Ramachandran distribution , which we use here to mean a statistical representation of Ramachandran data , usually in the form of a probability density function ( N . B . by distribution , we do not mean the cumulative distribution function or CDF ) . A probability density function gives the probability of finding an amino acid conformation in a specific range of φ , ψ values . For instance , if the function is given on a 10°×10° grid from −180° to +180° in φ , ψ ( 1296 values ) , then the distribution may give the probability per 10°×10° region . It could also be expressed per degree squared or per radian squared . Such distributions may be derived for specific amino acid types and/or for specific structural features . There are several important considerations in developing Ramachandran distributions from structural data , depending on the purpose of the derived distribution . First , while glycine and proline are usually treated separately , the other 18 amino acids are often treated as a single type . However , these amino acids are quite different in their proportions of residues in the α , β , polyproline II , and left-handed helical regions . Second , quite different distributions are determined when either all residues are used or only those outside the regular secondary structures of α-helices and β-sheets [10] . The latter are often assumed to be “intrinsic” preferences of the backbone [11] , not influenced by forming specific hydrogen bonds present in regular secondary structures . Third , the quality and quantity of the data are crucial in determining distributions meant to act as quality filters for newly determined structures or for structure prediction . As more structures have become available at higher resolutions , it is now possible to use quite large datasets with resolution cutoffs of 1 . 8 Å or even better . Other filters have been used including B-factors and steric clashes to remove residues that may be modeled improperly or at least with considerable uncertainty within the electron density . For instance , by using higher resolution structures , B-factors , steric overlaps , and other checks , Lovell et al . [12] were able to determine Ramachandran distributions with smaller “allowed” and “generously allowed” regions than previous efforts . Fourth , most previous efforts have involved density estimation using simple histogram methods – the counts or proportion of counts of residues in non-overlapping square bins of the φ , ψ space . However , even when a large number of proteins are used , the distribution in φ , ψ space may be quite bumpy . It is therefore of some importance to use a proper density estimation method that results in smooth distributions and minimizes the effects of outliers . This has been accomplished in a number of ways [12] , [13] , [14] , [15] . The Richardson group used kernel density estimates to obtain Ramachandran distributions for Gly , pre-Pro , and non-Gly , non-pre-Pro residues [12] . Kernel density estimates are performed by placing a kernel function such as a Gaussian on each data point , and the density estimate is produced on a grid by summing up the values of all of the kernel functions across the data . Although not described as such , they used what in effect are adaptive kernel density estimates [16] , [17] , such that the data are smoothed to a greater extent with wider kernel functions in sparsely populated regions of the space , while in more populated regions , narrower kernel functions can be used . Because they used a narrow kernel and grouped all non-Gly , non-Pro , non-pre-Pro residues together , the resulting distributions are well-suited to structure validation . Amir et al . used non-adaptive kernel density estimates , but with removal of outliers and addition of pseudocounts in sparsely populated regions [13] . To provide smoothness and differentiability , they calculated cubic spline fits to the kernel density estimates . Pertsemlidis et al . used an exponential of a Fourier series to calculate log probability densities of φ , ψ data [18] . Hovmöller et al . produced smoothed Ramachandran distributions for all 20 amino acids , and differentiated among different secondary structures; however , the manner of smoothing was not described [19] . Dahl et al . [12] and Lennox et al . [13] used Dirichlet process mixture models to obtain Ramachandran distributions for all 20 amino acids . The Dirichlet process approach is similar to kernel density estimation in that it yields an overall density estimate that is a superposition of component density functions , but the component densities are not located at the data points , and the number of component densities is unknown and inferred from the data [20] . This latter fact places the Dirichlet process approach in the general class of so-called Bayesian nonparametric methods [21] . Here , “nonparametric” does not mean an absence of parameters , but rather means that the number of parameters is not fixed in advance and can grow as data accrue . Ramachandran distributions may also be affected by the identity or conformation of neighboring amino acids . In particular , it has long been known that residues that precede proline have quite different Ramachandran distributions [22] , with significantly less density in the α and left-handed regions of the Ramachandran map . They also exhibit additional density in the so-called ζ region [23] , near φ , ψ = ( −130° , +80° ) , due to favorable van der Waals and electrostatic interactions [2] . The effect of local sequence on backbone conformation initially was used for the purpose of secondary structure prediction [24] , [25] . A number of groups have discussed the effect of local sequence , usually plus or minus one amino acid , on backbone conformational distributions [15] , [26] , [27] , [28] , [29] , [30] , [31] . This has been examined as a violation of the Flory isolated pair hypothesis , which states that conformations of individual dihedral angle pairs in a polymer are approximately independent of the conformations and/or residue identity of their neighbors [32] . Pappu et al . demonstrated by enumerating conformations of polyalanine that this is not true for neighboring conformations in peptides [29] . Zaman et al . used molecular dynamics simulations of monomers , dimers , and trimers to determine the nearest-neighbor effects of conformation and amino acid type on backbone conformations and entropy [27] . From the same group , Jha et al . examined experimentally determined distributions in coiled regions in a set of 2020 proteins of better than 2 . 0 Å resolution and found strong neighbor dependence on the populations in the α , β , and Polyproline II ( PPII ) regions of the Ramachandran map [26] . Erman et al . also examined neighbor residue-type dependence of which regions ( α , β , etc . ) were populated as a method for predicting these regions given local sequence context from a statistical mechanical theory [30] , [31] . Betancourt and Skolnick used 7070 chains from the PDB to determine the conformational properties of triplets of amino acids , in terms of occupied basins of the Ramachandran map , as well as other distributions such as the pair [28] . They used the data to produce a low-resolution potential energy function for backbone conformations that depends on local sequence . Lennox et al . [13] found that smooth density estimates for pairs , which span a pair of residues and thus capture a limited form of neighbor dependence , yield better estimates of the Ramachandran distribution than those based on standard pairs . A limitation of efforts to capture neighbor dependence is that the data become fractionated into groups that may contain small numbers of data points . This can yield inaccurate estimates of the densities , defeating the purpose of separating the data into groups . This problem is compounded if we also wish to separate data by secondary structure , or by any of a variety of other contextual variables . Our approach to addressing this general problem is to make use of the concept of a hierarchical Bayesian model . A hierarchical model is akin to a phylogeny , where the models for individual groups of data are at the leaves , and models are related if they are nearby each other in the tree . Specifically , we make use of a recent development in Bayesian nonparametric statistics known as the hierarchical Dirichlet process ( HDP ) [33] . In the HDP approach , as we discuss in the Methods section , evidence for a region of high density in one group of data can be transferred to a related group . In particular , we can use the HDP to tie together the density estimates for a given residue with different right or left neighbor residue types . This approach allows us to exploit commonalities among these densities so as to combat the data sparsity problem while allowing the individual densities to exhibit idiosyncratic characteristics . In this paper , we determine both neighbor-independent and neighbor-dependent Ramachandran distributions for all 20 trans amino acids as well as cis proline ( 21 distributions ) and for all 420 left and 420 right-neighbor-amino acid type pairs . We use a set of 3038 proteins at resolution of 1 . 7 Å or better and use electron density calculations to remove residues that are not well-fit to the density [34] . We explore the features of different input data sets , for instance including or excluding 310-helix and turn residues from longer loop regions . We examine some clear trends in these distributions . These include not only the influence of neighboring proline residues , but also aromatics , β-branched residues , hydrogen-bonding residues , and glycine . Our primary purpose for developing these potentials is to improve protein structure prediction . We perform a number of tests including loop modeling with Rosetta [8] as well as prediction of φ , ψ values of loops residues purely from local sequence . The neighbor-dependent distributions provide better results in both cases . The distributions are available for download from http://dunbrack . fccc . edu/hdp .
The data set in this paper consisted of 3038 proteins with available electron densities from the Uppsala Electron Density Server [35] . After removing residues with electron density in the bottom 20th percentile and restricting the set to loop residues with no missing backbone atoms and at least three residues away from α helix ( H ) or β sheet ( E ) , as identified with the program Stride [36] , we obtained a set of 62 , 345 residues ( the TCBIG set , for Stride one-letter designations of Turn , Coil , β-Bridge , π-Helix , 310-Helix respectively ) . We created a second set by removing the “regular” secondary structures of 310-helices and π-helices and those residues that neighbor them . The result is a set of 44 , 112 residues ( the TCB set ) . In both sets , we kept so-called “Bridge” residues , since these sometimes occur in long loop regions as backbone-backbone hydrogen bonds . The percentages in each secondary structure for each set are given in Table 1 . The percentages in each Ramachandran region for each set are given in Table 2 . The regions are defined in Methods , and consist of A ( α helix region ) , B ( β sheet regions ) , P ( polyproline II region ) , L ( left-handed helix ) , and E ( ε or extended region , the lower right and upper right regions of the φ , ψ map , accessible primarily to glycine ) . Cis residues are counted separately . Removing the regular secondary structures , 310-helix and π-helix , from TCBIG has a large effect on both the Ramachandran distributions and contributions of turns and coil . TCBIG is 41% α while TCB is 32% . TCBIG is 50% Turn while TCB is 62% Turn . Turns contribute substantially to long loops , and also to the population in the Ramachandran α region . Finally , we also performed calculations on Turn residues alone and Coil residue alone , making the T and C sets respectively . These are relatively small sets of 27 , 532 and 13 , 945 residues . The neighbor-independent distributions are likely to be reasonable , but the neighbor-dependent ones may require larger data sets with lower resolution and/or less stringent cutoffs for electron density or mutual sequence identity . In Figures 1 and 2 , we show neighbor-independent Bayesian nonparametric density estimates of the Ramachandran distributions of all 20 amino acids for the TCBIG set with cis and trans proline plotted separately . These are smoother than many previous Ramachandran distributions and show the differences among the 20 amino acids clearly ( the jagged appearance at the top of peaks is an artifact of plotting the surface with polygons ) . In this and subsequent figures , the notation XXX . yyy or yyy . XXX means the probability density estimate for residue XXX with yyy as right or left neighbor respectively . The residue types , even outside of Gly and Pro , have quite distinct Ramachandran distributions . Because β turns are a large proportion of both the TCBIG and TCB data sets , we also calculated smooth density estimates of the Ramachandran distributions of turn and coil residues separately ( Stride designations T and C ) , and in Figure 3 we plot side-by-side the Ramachandran distributions for TCBIG , TCB , T-only , and C-only for Ala , Asn , Glu , and Ile . The TCB set loses the sharp peak near ( φ , ψ ) = ( −50° , −25° ) which is a result of the 15% of residues in the TCBIG set that are in 310-helix . As noted by others [26] , coil and turn residues ( columns 3 and 4 respectively ) have quite different distributions with the turn set having higher α content and the coil set having higher polyproline-II content . We calculated the Hellinger distance ( see Methods ) between all residue types , and the values for a subset of 12 residues are given in Table 3 for TCBIG . The table shows the calculated Hellinger distances times 100 , and we refer to these values in what follows as “the Hellinger distance , ” which will be a value between 0 for identical probability densities and 100 for completely non-overlapping densities . Very similar residue pairs such as Phe/Tyr and Val/Ile have Hellinger distances of about 8 or 9 . Very different distributions such as any residue with Gly or Pro have Hellinger distances in the range of 40 to 60 . Outside of Gly and Pro , most distances are in the range 10 to 30 . Alanine is not a typical residue; most side chains with a single γ heavy atom have smaller Hellinger distances to each other than they do to Ala . Between distributions derived from the TCBIG and TCB sets , the Hellinger distances of the same residues in each set ( e . g . , Ala in TCBIG vs . Ala in TCB ) range from 5 to 11 , with the larger values coming from hydrophobic residues , which are underrepresented in 310-helix . Comparison of turns with TCB or TCBIG for a single residue type produces Hellinger distances in the range of 9 to 14; these data sets are 50 and 62% turn respectively . On the other hand , coil distributions are quite different from the TCB and TCBIG sets with Hellinger distances in the range of 14 to 22 ( data not shown ) . In Figure 4 , we show the effect of all 20 possible right neighbors on the Ramachandran distribution of Gln . Gln behaves typically in terms of neighbor effects . Certain neighbor types have consistent effects in terms of increasing or decreasing the α , β , and/or PPII regions for most central residue types , and the residues are grouped accordingly in the figure . Pro on the right exerts the largest effect and this has been well-studied previously [2] , [22] , but with these calculations we provide smooth , statistically reasonable pre-Pro Ramachandran distributions for all 20 amino acids . The GLN . pro map shows the features typical of pre-Pro distributions – very low α ( A ) population , lower L population than non-pre-Pro distributions , and the so-called ζ conformation [2] , [23] , which is a bump just below the β ( B ) region at φ , ψ = ( −130° , +80° ) . Other groups of residues also have particular effects as neighbors . Aromatic residues as right neighbors , especially Phe and Tyr , suppress the P region and increase the A region . Val and Ile suppress A in favor of broadly distributed B and P density , while Gly strongly favors P , most likely due to an increase in Type II turns . Type II turns consist of residue 2 ( of 4 residues in the turn ) in a P conformation and residue 3 in an L conformation , most accessible to glycine . Negatively charged residues also seem to increase A . To show that these are general effects , in Figure 5 we show Ramachandran plots for ALA , LYS , TYR , and VAL with right neighbors equal to pro , gly , phe , val , and gln . Gln behaves as a relatively neutral neighbor . The other neighbor types have similar effects on these residues as they do on Gln shown in the previous figure . In Figure 6 we show the effects of left neighbors . Val and Ile tend to reduce A conformations while Pro , Ser , and Asp tend to increase A conformations . To quantify the effects , Table 4 contains the Hellinger distances for right neighbors of Gln , which shows the similar behaviors of Phe and Tyr , Val , and Ile , and the different behaviors of Gly and Pro as neighbors . Ile and Val as right neighbors of Gln are not as similar as they are for most other amino acid types , where the average Hellinger distance is about 7 . We calculated the average Hellinger distances between each pair of neighbors over all the central amino acid types , e . g . , for right neighbors Ri and Rj:and then used classical multi-dimensional scaling [37] to plot these distances approximately in two dimensions . The results are shown in Figure 7 for both left neighbors ( Figure 7A ) and right neighbors ( Figure 7B ) . For the right neighbors , we omitted Pro and Gly from the graph since they lie far from the others . Pro is at coordinate ( −25 . 5 , −1 . 3 ) and Gly is at ( 0 . 7 , 11 . 2 ) . Residues with similar properties are mostly grouped together , e . g . Val and Ile , Phe and Tyr , Lys and Arg ( as left neighbors ) , Asn and Gln ( as right neighbors ) , etc . The distances are on relatively similar scales with Val and Asp having the largest Hellinger distance at 11 . 7 for left neighbors , and Ile and Glu at 12 . 6 for the right neighbors ( excluding Gly and Pro as right neighbors ) . To test whether the neighbor-dependent Ramachandran distributions will have utility in protein structure prediction , we ran a benchmark of loop predictions developed by Soto et al . [38] , consisting of 290 loops from length 8 to length 13 . Rosetta was used to predict the structures of these loops in the context of the rest of each experimental protein structure [39] . The goal was not to judge the accuracy of Rosetta but to compare the different Ramachandran distributions for scoring and energy minimization . The results are shown in Figure 8 in the form of Q-Q plots ( quantile-quantile ) [40] . To produce a Q-Q plot , each group of data are sorted numerically , and the resulting vectors are plotted against one each other . Note that the x and y axes are RMSDs rather than 1/290th quantiles . Significant deviations from the line y = x indicate differences in the distribution of the two data sets . In Figure 8A , loop prediction Cα RMSDs using the neighbor-independent TCB Ramachandran distributions are plotted against loop predictions using the Ramachandran distributions currently in Rosetta ( “original” ) [8] . The plot shows a small benefit to the new Ramachandran distributions in the 3–6 Å range of RMSD . The TCBIG distributions show similar results ( not shown ) . In Figure 8B , the neighbor-dependent distributions are compared to the original distributions and exhibit much larger differences down to 2 . 0 Å in RMSD . To compare the neighbor-dependent and neighbor-independent distributions from the TCB set , the Q-Q plot is shown in Figure 8C . Again , the neighbor-dependent distributions have an advantage over the neighbor-independent distributions . Finally , in Figure 8D the TCB set is shown to produce better predictions than the TCBIG set , which emphasizes the need to choose the right Ramachandran distributions depending on the prediction task . To explore why , we evaluated the Stride assignments and the Ramachandran distributions for the benchmark loops . The benchmark is 59% Turn , 36% Coil , 3% Bridge , and only 1 . 5% 310 Helix . As shown in Table 1 , this distribution matches the TCB ( 62% Turn ) set much more closely than the TCBIG set ( 50% Turn ) . The benchmark is 31% α helical region and 57% β and Polyproline II region . This is much closer to the TCB set ( 32% A and 56% B+P ) than the TCBIG set ( 41% A and 48% B+P ) , as shown in Table 2 . As a test independent of loop sampling methodology and scoring functions , we took a set of 2 , 074 proteins with resolution better than 1 . 8 Å , and for loops longer than 8 residues , we predicted the φ , ψ value by selecting the grid point with the largest probability:where C is the central residues whose φ , ψ are being predicted and L and R are the identities of its left and right neighbors . We then compared these predictions with the values from the crystallographic structure . The results are shown in Table 5 , which shows the percentage of residues whose φ or ψ are predicted within 40° and also the mean absolute deviation in φ and ψ between the native and predicted values . The improvements from the neighbor-independent distributions to the neighbor-dependent distributions are evident . For instance for Leu , the neighbor-independent distributions are able to predict the ψ values for only 37 . 3% of residues within 40° , while the neighbor-dependent distributions predict 52 . 4% correct , for an improvement of 15 . 1% of all leucines . This is a 40% improvement of the neighbor-independent rate ( 15 . 1/37 . 3 ) . Dihedral φ is concentrated in a smaller region than ψ and so is harder to improve than ψ . The largest improvements are for hydrophobic residues , while residues that show different behaviors in turns vs , coils ( Gly , Pro , Asn ) show less improvement , probably because the neighbor effects are context-dependent ( Turn or Coil ) . It is of some interest to understand the origin of the neighbor-dependent backbone conformation propensities observed in this work and in similar analyses that have appeared previously . The analysis is complicated by the preponderance of β turns in the two data sets . Such turns are defined by Cα-Cα distances of less than 7 Å between residue 1 and residue 4 in a four-residue segment . β turns are categorized by the conformations of residues 2 and 3 , and follow some well-studied amino acid preferences at all four positions . [41] To determine whether the neighbor effects were dependent on whether residues were in turns or in coil regions of loops , we used the raw data to determine the percentages of residues in A , B , and P regions of the Ramachandran map depending on the neighbor residue types for coil , turns , and all residues in the TCB set . The results are presented in Table 6 . Note , these are exact values from raw counts of the data , not from the probability densities that result from them . The first row of numbers gives the percentage in A , B , or P of all Coil , Turn , or TCB non-pre-Pro residues . That is , 19 . 4% of Coil non-pre-Pro residues are in A conformation . We exclude pre-Pro residues , due to the large effect Pro has when it is a right neighbor . The numbers in the top half of the table show the effect of the right neighbors listed in the first column as changes in the percentage of all residues in each secondary structure type ( C , T , or TCB ) . That is , while 19 . 4% of non-pre-Pro Coil residues are in A , only 12 . 1% of pre-Ile residues in Coil are in A . The pre-Pro numbers in the table are relative to all residues in each group ( C , T , or TCB ) . Some effects are seen in both Turn and Coil conformations , while others are Turn or Coil specific . So for instance , Ile and Val neighbors to the right decrease A populations for both Turn and Coil , although for Coil the P population rises more than the B population compared to Turn . Aromatic residues to the right decrease P for both Turn and Coil , although in this case for Turn A goes up while for Coil B goes up . Asn , Ser , and Asp to the left decrease P and to a lesser extent B and increase A significantly both in Turn and in Coil . Ile and Val to the left decrease A and increase P for both Turn and Coil , while also increasing B for Turn , probably due to turns of type βεγL ( residues 2 and 3 ) according to the nomenclature of Wilmot and Thornton [41] . This is a form of distorted type II turns , which prefer Val and Ile at position 1 , and hence B and P at position 2 . Some effects are completely conformation specific . Gly to the right increases P and decreases A in Turn conformations ( due to Type II turns with Gly at position 3 ) , while having the opposite effect on Coil . Pro to the left increases A and decreases B and P for Turn , probably due to an increase in Type I turns with Pro at position 2 , while having exactly the opposite effects for Coil conformations . We investigated some of these preferences visually in both turns and coils in order to identify potential favorable or unfavorable interactions that may be responsible for the observed propensities listed in Table 6 . Several of these interactions are shown in Figure 9 in images of residue triples , chosen to demonstrate the interaction of the left or right neighbor side chain with the other two amino acids of the tripeptide depicted . Ile as a left neighbor to an A conformation in Coil is shown in Figure 9A . In this conformation , the β branched amino acid blocks access of hydrogen-bond acceptors to the NH of the amino acid to the right of the central residue ( i ) , whose relative position is determined by φ , ψ of residue i and ψ of residue i−1 . In this image , atom Cγ1 of Ile is 4 . 4 Å from Ni+1 . A residue with only a single Xγ atom might have Xγ in the position of Cγ2 in this residue , thereby allowing access to NHi+1 . This interaction occurs when residue i−1 is in a B conformation , which is more common for Coil than for Turn . A similar effect is shown for Val to the right in Figure 9B , with a close contact of Cγ2 with backbone atom Oi−1 . Polar interactions and side-chain/backbone hydrogen bonds are also likely responsible for some of the observed trends in Table 6 . In Figure 9C , an Asn Oδ1 at i−1 acts as a hydrogen bond acceptor to NHi+1 and thus in fact favors an A conformation at position i , both in Turn and in Coil ( as shown ) . Asp Oδ1 Oδ2 at i−1 can make the same hydrogen bond with NHi+1 , and the population of residue i is also increased in A conformations ( Table 6 ) . Ser Oγ behaves similarly ( not shown ) . It is interesting to note that these three residues , Asn , Asp , and Ser , cluster together in the two-dimensional projection of their distances as left neighbors ( Figure 7A ) but not as right neighbors ( Figure 7B ) . In Figure 9D , the well-known Pro-right interaction is shown in the very close approach of the Pro Cδ with atom N of the central residue , when this residue is in an A conformation . Figures 9E and 9F show effects of aromatic residues . A left aromatic neighbor ( Figure 9E ) may have a favorable hydrophobic interaction with the side chain of residue i+1 , when residue i is in a B conformation . A right neighbor of residue i that is aromatic can have apparent unfavorable interactions with the carbonyl oxygen of residue i−1 , as shown in Figure 9F . The position of this oxygen relative to this side chain is determined by φ , ψ of residue i , φ of residue i+1 , as well as the rotamer of the aromatic residue at i+1 . In several cases studied , when this side chain is in a χ1 rotamer of g− , which is the most common rotamer of aromatic amino acids , the π orbitals of the ring are in close proximity ( about 4 . 5 Å ) to the carbonyl oxygen , as shown in Figure 9F , thus decreasing the tendency of the A conformation when aromatic residues are at position i+1 . Also , this residue has a φ of −86° , which is consistent with both A and P conformations .
Ramachandran distributions have been produced using many different input data sets and different statistical methods . We have made several choices primarily for the purpose of protein structure prediction , in particular modeling regions of non-regular secondary structure . First , we have used a large input data set of 3 , 038 proteins with better than or equal to 1 . 8 Å resolution . Second , we discarded residues with electron density in the bottom 20th percentile [34] in order to remove conformations in potentially mobile parts of the input structures . Third , we derived data sets of all residues not in helix or sheet and a second data set also excluding 310-helix and π-helix . For comparison , we also studied residues in β turns and in coil ( i . e . , not in secondary structure or turns ) . Finally , we used hierarchical Dirichlet process methods to develop smooth and statistically reliable Ramachandran distributions for all 20 amino acid types with each of the 20 amino acids as either left or right neighbor . These so-called neighbor-dependent Ramachandran distributions are shown to be useful in loop structure prediction using the Rosetta program . Many other such statistical analyses φ , ψ data have appeared previously [11] , [12] , [14] , [18] , [27] , [29] , [42] , but few are publicly available for use in structure validation or structure prediction . Our analysis differs in a number of respects from the available distributions . For instance , the Richardson group developed smooth φ , ψ densities using adaptive kernel density estimates [16] , [17] from a filtered , high-resolution data set [12] . However , their main purpose was structure validation , and their distributions are not residue dependent; that is , all non-Gly , non-Pro , non-pre-Pro residues are treated in one density estimation . They used a quite narrow kernel in their density estimates with all 18 residue types merged into a single , very large data set . It is likely that their density estimates therefore are able to accurately represent sharp changes in the probability density between allowed and disallowed regions . This is quite important for structural validation , as demonstrated by the widespread use of the MolProbity server [6] . Quite usefully , they do provide a separate pre-Pro distribution , since this distribution exhibits the strongest neighbor-dependence , as well as Gly and Pro distributions . Amir et al . fit cubic splines to kernel density estimates to produce smooth Ramachandran distributions for each amino acid type individually for protein structure prediction [13] . They used a much smaller data set than used here , a set of 850 proteins previously produced by us for development of the backbone-dependent rotamer library [43] , [44] , and did not examine neighbor effects . Sosnick et al . calculated neighbor-dependent statistics of backbone conformations; however , these are not full Ramachandran densities but proportions in large regions of the φ , ψ space ( α , β , polyproline II , etc . ) [26] . The Flory isolated-pair hypothesis [32] states that the pair of dihedral angles in protein backbones , φ , ψ , are independent of the conformation of neighboring residues , and by extension the identity of those residues . This idea has been challenged by statistics from the PDB [26] , [28] , molecular dynamics simulations [27] , exhaustive conformational searches and energy calculations [29] , and NMR experiments [45] . The results here confirm these earlier investigations and extend them by deriving full Ramachandran probability densities for all residue-neighbor conformations and for different input data sets ( the TCBIG , TCB , T , and C sets ) . We also suggest some explanations for some of the effects observed . For left neighbors , these effects in some instances are caused by interactions of the side-chain of residue i−1 and backbone NH of residue i+1 , whose relative positions are determined by φ , ψ , of residue i , ψ of residue i−1 , as well as the χ angles of residue i−1 . For right neighbors , the effects sometimes stem from a complementary interaction – the side chain of residue i+1 and the backbone O = C of residue i−1 , whose relative positions are determined by φ , ψ , of residue i , φ of residue i+1 , and the side-chain conformation of residue i+1 . In both cases , these interactions can be electrostatic repulsions , hydrogen bonds , or steric , in some cases by blocking access of hydrogen bond donors or acceptors to the backbone . These are commonly described for some residue types in α-helices or β sheets , or capping positions of regular secondary structures , but they also operate in turn and coil conformations of long protein loops . The key idea of the statistical approach developed here is that more precise estimates of Ramachandran distributions can be found if we examine these distributions in different contexts . This differentiation by context creates a data sparsity problem , in that some contexts may yield very few data points , but , as we have shown , this problem can be addressed effectively within a hierarchical Bayesian framework . Our biochemical knowledge about relatedness can be used to reap further benefits of differentiation of different classes . The general idea of hierarchical modeling is widespread in Bayesian statistics [46] . It is most common in parametric Bayesian modeling , where parameters are often shared among multiple parametric distributions ( e . g . , the probabilities of recovery of ill patients are similar if the patients are in the same hospital , have the same doctor , etc . ) . As we have seen , however , the same basic concepts apply in nonparametric Bayesian modeling . Thus we are able to share statistical strength among multiple multi-modal distributions in which the number of modes in each distribution is unknown a priori . In particular , the hierarchical Dirichlet process allows us to separate Ramachandran distributions according to neighboring residue types and to exploit similarities in these distributions . Moreover , although we have focused on contexts that are defined by amino acid neighborhoods , the same ideas could be used to estimate Ramachandran distributions that are differentiated according to other contextual variables , such as loop structure class ( turn , coil , 310-helix , etc . ) or neighbor conformational class ( A , B , P , etc ) . We believe the neighbor-dependent distributions developed here will provide utility in a number of applications in protein structure prediction and structure determination . The appropriate distributions will be specific to each application . We hope that by making them publicly available , their properties may be further explored in various applications .
We selected a list of 3038 proteins from structures in the Protein Data Bank ( PDB ) that also had electron densities available from the Uppsala Electron Density Server ( EDS ) [35] . Using the list of PDB entries with available electron density maps , we entered this list into the PISCES server ( http://dunbrack . fccc . edu/pisces ) [47] , [48] to obtain a subset with resolution better than or equal to 1 . 7 Å , R-factor ≤0 . 25 , and mutual sequence identity less than 50% . Secondary structure was determined with the program Stride [36] , which assigns H ( helix ) , E ( sheet ) , B ( bridge ) , T ( Turn ) , G ( 310 helix ) , I ( π-helix ) , and C ( coil ) to all residues . In order to explore the distributions in longer loop regions , we excluded loops ( non-E , H segments ) of less than 6 amino acids as well as the first two non-E , H residues following each E or H and the two non-E , H residues preceding E or H . This was done to avoid secondary-structure-capping residues , which have specific distributions in order to break the secondary structure ( e . g . , N and C cap residues in helices usually have conformations outside the α region; otherwise they would likely be part of the helix ) . We used a quality measure of each residue's backbone by calculating the geometric mean of the electron density at backbone atom coordinates as described in previous work [34] . We excluded those residues in the bottom 20th percentile for each residue type from the data . This filter works similarly to a B-factor filter . We use the former because some X-ray structures do not have consistent values for B-factors versus electron density [34] . We created several sets of data for analysis with the HDP procedure: Our approach to modeling neighbor-dependent Ramachandran densities is based on a Bayesian nonparametric density model known as the hierarchical Dirichlet process ( HDP ) mixture model [33] . The HDP approach allows us to subdivide our data into groups defined by a specific central amino acid and by a particular neighboring amino acid . Given these groups , the HDP mixture model produces density estimates for each group in a manner that takes advantage of the commonalities among the groups while allowing each group to exhibit idiosyncratic features . Before providing a detailed description of the model , let us provide a non-technical overview . Our approach is based on modeling densities such as the Ramachandran distribution as mixtures ( i . e . , weighted sums ) of simple Gaussian component densities . Each such component density is a unimodal bump in the two-dimensional Ramachandran plot , and the overall density is a weighted sum of such bumps . There is a global library of component densities for a single amino acid type , and each particular density ( i . e . , the Ramachandran density corresponding to a specific right or left neighbor ) draws a number of component densities from the global library . The specific details of this model – the locations and orientations of the Gaussian bumps , the number of bumps used in each Ramachandran density , and the pattern of sharing of the bumps between the multiple Ramachandran densities – arise from a Bayesian inference procedure that combines our prior assumptions ( the HDP model described below ) with the observed data of amino acid dihedral angles in the Ramachandran maps . In essence , the inference procedure based on a Markov chain Monte Carlo procedure finds configurations of the model that are compatible with both the prior and the data . We now turn to a more technical description of the model . The HDP model can be viewed as a generalization of a simpler model , the Bayesian finite mixture model . Accordingly , we begin with a description of Bayesian finite mixtures and then develop the generalization to the HDP . A classical finite mixture considers a set of component densities:where x in this case is the two-dimensional vector of Ramachandran angles , and where is a parameter vector associated with the kth density . For example , f might be a Gaussian density , with parameter vector θ , representing the mean and covariance matrix . We assume that the number K is known and fixed; this assumption will be removed in the HDP approach . We assume that a data point xi is generated according to the following process: This process is repeated N times , yielding a data set . The probability of the th data point can be written as follows:and the overall probability of is obtained by taking a product over these probabilities . In the Bayesian approach to finite mixture models , the parameters π and are assumed to be generated from a prior probability distribution . Specifically , in the case of Gaussian mixture models , where consists of a mean and covariance matrix , a common choice for the prior iswhere Dir is the Dirichlet distribution , IW is the inverse Wishart distribution , and where α , μ0 , Λ , γ , and Γ are hyperparameters that are often fixed a priori but also can be inferred from the data . The next step in our development of the HDP mixture model is the Dirichlet process ( DP ) mixture model , which is a generalization of Bayesian finite mixture models where the number K is treated as unknown and to be inferred from the data [20] . This generalization is often described metaphorically in terms of a simple stochastic process known as the Chinese restaurant process ( CRP ) . Consider a Chinese restaurant with an infinite number of tables and consider the following seating process . The first customer to arrive in the restaurant sits at the first table with probability one . The second customer then joins that first customer with probability and starts a new table with probability , where is a parameter . The general rule is that the nth customer sits at a table with probability proportional to the number of people already sitting at that table . She may sit at a new table with probability . The CRP determines a clustering or partitioning of the customers . We can turn this clustering process into a mixture model by associating the kth table with the kth component in a mixture model . In particular , let us assume that the first customer to sit at the kth table in the restaurant selects a dish , given by parameter , for that table from some prior distribution on the parameters . Each subsequent customer who sits at that table inherits that parameter vector . By viewing the ith customer as a data point and drawing from the distribution , where is the parameter vector at the table where customer i sits , we obtain a mixture model for generating data . Note that the number of tables in the CRP , which corresponds to K in the finite mixture model , is a random variable that grows as N ( the number of data points ) grows . Indeed , the expected value of the number of occupied tables turns out to scale as log N [21] . The DP mixture model is “nonparametric” — the number of parameters grow as we obtain more data . The inferential problem associated with the DP mixture model is as follows . Given a data set X , compute a posterior probability distribution on the allocation of data points to tables and on the parameters associated with the tables . This problem can be solved with a variety of standard methods for posterior inference , including Markov chain Monte Carlo [49] , sequential Monte Carlo [50] and variational inference [51] . Although we do not provide details here , it is worth noting that the posterior inference algorithms for DP mixtures are relatively simple; for example , in the Markov chain Monte Carlo methods , we repeatedly revisit each data point and assess which table it should be moved to given the current configuration of all of the other data points . The probability of assigning a point to a table is proportional to the product of the number of data points already at that table and the likelihood of that data point given the parameter vector ( or dish ) of the table . We now turn to the hierarchical Dirichlet process ( HDP ) [33] . The HDP can also be described with a restaurant metaphor , in this case by the Chinese restaurant franchise ( CRF ) . The problem now is to model the densities associated with each of M groups of data; in particular , for a given amino acid , the M groups correspond to the set of right or left neighbors . Accordingly , in the CRF there are M = 20 restaurants . Data points are categorized as to which group they belong to . A data point ( customer ) enters the restaurant associated with its group and sits at a table with probability proportional to the number of customers currently sitting at the table . Moreover , the first customer to sit at a table selects a parameter vector for that table . In the CRF metaphor , the parameters are viewed as “dishes , ” and the dishes are obtained from a menu that is shared among all of the restaurants . When a dish is selected from the menu a check mark is placed next to that dish . When a new customer goes to the menu to select a dish for her table , she selects a dish with probability proportional to the number of check marks next to that dish and the likelihood of her data point given the parameter vector associated with the dish . Additionally , there is a probability of selecting a new dish . The global menu implements a sharing of mixture components among the restaurants . Let us consider concretely how this creates a link among multiple density estimation problems . Consider in particular the case of Gaussian mixtures , in which the parameter consists of a mean and covariance matrix . When a specific is selected by a customer in the mth restaurant , this corresponds to a Gaussian bump in the density associated with the mth group . Because this parameter vector appears on the global menu , it can then be selected by a customer in one of the other restaurants . This means that that Gaussian bump can also appear in one of the other density models . This allows us to capture commonalities among the groups . Of course , some dishes will only be selected in a single restaurant , and this allows the corresponding group of data to exhibit idiosyncratic features . Tables and dishes are selected in part based on the likelihood of the customer data points given the parameter vectors of the dishes served at each table . We now describe in greater detail the specifics of the model and estimation procedures used for Ramachandran data . We use the Gaussian mixture HDP to give density estimates of the φ , ψ , dihedral angles conditional on the either the central and left residues ( C , L ) or the central and right residues ( C , R ) . Since the central residue clearly affects the φ , ψ , angle densities far more than the neighbors , our hierarchy shares features when the central residue is the same and allows for idiosyncratic features for the different neighboring residues . The choice of conditioning on only two residues was motivated by the fact that conditioning on all three residues resulted in very small data sizes that failed to contain sufficient information to capture appropriate features . The HDP estimates were fit independently for each central residue to avoid computational issues . When proline was the central residue , the trans- and cis- configurations were treated as unique , giving a total of 21 possibilities for the central residue and 20 possible neighboring residues . A natural choice for the component distributions for the φ , ψ , angles would be the bivariate von Mises distribution , an exponential family distribution for angles [52] . Indeed , this distribution was used in earlier work using the Dirichlet process by Lennox et al . [15] However , posterior inference in the bivariate von Mises model is intractable , requiring the computation of an infinite sum of incomplete Bessel functions . Lennox et al . made use of a Gaussian approximation to the von Mises model within the framework of a Metropolis-Hastings algorithm , but this is still complex . Moreover , the major gains from the von Mises model occur near the boundaries of the Ramachandran plot , where the “wrap around” obtained from the von Mises distribution is helpful . We have pursued a simpler approach , where we rotate the data to lie in the rectangle ( 50°±180° , 90°±180° ) . This rotation ensured that the density near the boundary is low , and during model fitting , little mass is lost using component densities – in particular Gaussian densitites – which do not wrap around at the boundary . As a post-processing step , the Gaussian mixture components are treated as wrapped Gaussians to ensure that the final density estimate is smooth even at the low density regions at the boundary . This technique yields a fast algorithm that can be used on large data sets . A conjugate prior was used for the mixture component parameters; specifically , the prior distribution on μk , Σk is Normal-Inverse-Wishart with the following parameters: Mean: μ0 = ( 50° , 90° ) Diagonal covariance matrix: Λ = 252 I , where I is the unit diagonal matrix Number of pseudo-observations for the mean: κ0 = 0 . 01 Number of pseudo-observations for the variance: ν0 = 2 . 1 The numbers of pseudo-observations for the mean and variance were chosen to be small to give vague priors but large enough so that the prior is proper . An additional prior is placed on the two Dirichlet process α hyperparameters controlling the probability that a new “table” or “dish” in the CRF is sampled . The prior for each was exponential with mean 10 . Density estimates using this HDP model were obtained using Markov Chain Monte Carlo . We used the augmented Gibbs sampler described previously [33] . The mixture component parameters were integrated over , and only the sufficient statistics , the sample mean and covariance , were retained . After a burnin of 10 , 000 samples , an additional 50 , 000 samples were drawn and then thinned , with every 50th sample retained . For each sample the φ , ψ , density was evaluated on a grid of 288×288 values . These 1000 density samples were then averaged to obtain the final density estimates . Due to the large size of these grids , we fit a cubic tensor spline with 72×72 knots to reduce the size of the representation while providing an excellent approximation to the original fit . A degree d spline approximation to a function uses a piecewise polynomial of degree d to approximate the function . The location of each piece of the piecewise polynomial is determined by the knots , and the piecewise polynomial is constrained to have d−1 continuous derivatives . Thus , the compactness of the approximation is controlled by the number of knots , and the smoothness is controlled by the degree d . This spline representation has the additional advantage of being very fast to compute since evaluating a polynomial requires only a few basic multiplication and addition operations . Since the angle data are periodic , we also enforced the smoothness constraint at the boundaries at ±180° . A two dimensional tensor spline is one where each “piece” of the piecewise polynomial is of the form where and polynomials . Once the knots are defined , a 72×72 uniform grid in our case , a regression spline is easily fit by minimizing the squared error to a target , a 288×288 grid of log density estimates in our case . Compared to linear interpolation on a 72×72 grid representation , which consumes an equivalent amount of memory , the spline approximation improved the approximation error , as measured by Kullback-Leibler divergence , from 0 . 22 to 0 . 004 . We used the Hellinger distance to determine the similarities of different Ramachandran distributions . For two probability distributions , and , the Hellinger distance , H is calculated from the following equation:where the integral is taken over the domains of f and g . H satisfies the expression . We used a loop-prediction data set described by Soto et al . [38] , consisting of 290 loops ( the original set consisted of 293 targets but Rosetta was unable to complete three of them . This set consisted of loops from several previous benchmarks , with a total of 62 , 56 , 40 , 54 , 39 , and 39 loops of lengths 8 , 9 , 10 , 11 , 12 , and 13 residues respectively . Loop modeling was performed with Rosetta2 . 3 . 0 , modified to use the neighbor-dependent Ramachandran distributions . We used the standard pose-based loop modeling protocol built into Rosetta [53] , using a fixed backbone and side chains for all residues except those in the loop region to be predicted . We generated 2000 decoys for loops of length 8 and 9 , 5000 decoys for loop lengths 10–12 , and 8000 decoys for loop length of 13 . For each individual loop , a random starting conformation is constructed by arbitrarily inserting fragments in the loop region . The fragment library was generated using the standard Rosetta fragment generating tools , i . e . searching with the query sequence of each loop against representative PDB structures skipping homologous structures ( -nohoms option ) . Once the initial conformation was built , the simulation was performed in two steps . In the first ( low-resolution ) step , the side chains were represented by centroid atoms . A series of Monte Carlo perturbation steps followed by loop closure using cyclic coordinate descent ( CCD ) [54] and line energy minimization were performed . The conformation perturbation was done by inserting three-residue and one-residue fragments into the loop region . In the second ( high-resolution ) step , all atoms including hydrogen atoms were explicitly represented . The perturbation was done by introducing small random changes to one or more backbone torsions angles , followed by CCD closure , and Davidson-Fletcher-Powell minimization . Repacking of all the loop side chains was performed after every 20 cycles as well as at the end of the overall simulation . The full command line for loop modeling was: rosetta . gcc serial entry chain -pose -loops -fa_input -fold_with_dunbrack -fast -fix_natsc -ramaneighbors type -rama_file ramafile -pose_silent_out -pose_loops_file entrychain . loop -s entrychain . pdb -nstruct nStructs where variables with “$” signs were defined within a loop: The command-line options -ramaneighbors and -rama_file were added to this version of Rosetta , specifically for using the neighbor-dependent Ramachandran distributions . To obtain density estimates of φ , ψ , for the central residue conditional on all three residue types , the following model can be used to combine the HDP density estimates which are conditional only on central/left or central/right residue pairs . where S is a normalizing constant obtained by integrating the expression on the right hand side ( without the S ) . This estimate is the plug-in estimator for the full conditional probability given the assumption that the identity of the left and right residues are independent given φ , ψ , . For most residues , S was near 1 but for some residues and some neighbors , in particular proline , S was as low as 0 . 5 and as high as 1 . 5 . The normalization is therefore important . In order to characterize the effects of neighbors on populations on different regions of the Ramachandran map , we divided the φ , ψ space into non-overlapping bins as follows , for the α , β , polyproline II , left-handed , and γ conformations , respectively: A: −200°≤φ<0° , −120°<ψ≤40° , 90°≤ω≤270° B: −90°≤φ<0° , 40°<ψ≤240° , 90°≤ω<270° P: −200°≤φ<−90° , 40°<ψ≤240° , 90°≤ω<270° L: 0°≤φ<160° , −90°<ψ≤110° , 90°≤ω<270° E: 0°≤φ<160° , 110°<ψ≤270° , 90°≤ω<270° cis: −90°≤ω<90° Density plots were produced in Matlab ( the Mathworks , Inc . , Natick , MA , USA ) . The multi-dimensional scaling and QQ plots were performed in R ( the R Foundation for Statistical Computing , Vienna , Austria ) . Protein images were produced in PyMol ( DeLano Scientific , Palo Alto , CA , USA ) . All distributions are freely available to non-profit research groups at this address: http://dunbrack . fccc . edu/hdp . | The three-dimensional structure of a protein enables it to perform its specific function , which may be catalysis , DNA binding , cell signaling , maintaining cell shape and structure , or one of many other functions . Predicting the structures of proteins is an important goal of computational biology . One way of doing this is to figure out the rules that determine protein structure from protein sequences by determining how local protein sequence is associated with local protein structure . That is , many ( but not all ) of the interactions that determine protein structure occur between amino acids that are a short distance away from each other in the sequence . This is particularly true in the irregular parts of protein structure , often called loops . In this work , we have performed a statistical analysis of the structure of the protein backbone in loops as a function of the protein sequence . We have determined how an amino acid bends the local backbone due to its amino acid type and the amino acid types of its neighbors . We used a recently developed statistical method that is particularly suited to this problem . The analysis shows that backbone conformation prediction can be improved using the information in the statistical distributions we have developed . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"mathematics/statistics",
"computational",
"biology/protein",
"structure",
"prediction"
] | 2010 | Neighbor-Dependent Ramachandran Probability Distributions of Amino Acids Developed from a Hierarchical Dirichlet Process Model |
Despite its century-old use , the interpretation of local field potentials ( LFPs ) , the low-frequency part of electrical signals recorded in the brain , is still debated . In cortex the LFP appears to mainly stem from transmembrane neuronal currents following synaptic input , and obvious questions regarding the ‘locality’ of the LFP are: What is the size of the signal-generating region , i . e . , the spatial reach , around a recording contact ? How far does the LFP signal extend outside a synaptically activated neuronal population ? And how do the answers depend on the temporal frequency of the LFP signal ? Experimental inquiries have given conflicting results , and we here pursue a modeling approach based on a well-established biophysical forward-modeling scheme incorporating detailed reconstructed neuronal morphologies in precise calculations of population LFPs including thousands of neurons . The two key factors determining the frequency dependence of LFP are the spatial decay of the single-neuron LFP contribution and the conversion of synaptic input correlations into correlations between single-neuron LFP contributions . Both factors are seen to give low-pass filtering of the LFP signal power . For uncorrelated input only the first factor is relevant , and here a modest reduction ( <50% ) in the spatial reach is observed for higher frequencies ( >100 Hz ) compared to the near-DC ( ) value of about . Much larger frequency-dependent effects are seen when populations of pyramidal neurons receive correlated and spatially asymmetric inputs: the low-frequency ( ) LFP power can here be an order of magnitude or more larger than at 60 Hz . Moreover , the low-frequency LFP components have larger spatial reach and extend further outside the active population than high-frequency components . Further , the spatial LFP profiles for such populations typically span the full vertical extent of the dendrites of neurons in the population . Our numerical findings are backed up by an intuitive simplified model for the generation of population LFP .
The measurement of electrical potentials in the brain has a more than hundred year old history [1] . While the high-frequency part has been successfully used as a measure of spiking activity in a handful of surrounding neurons , the interpretation of the low-frequency part , the local field potential ( LFP ) , has proved more difficult . Current-source density ( CSD ) analysis of multisite LFP recordings across well-organized layered neural structures such as cortex and hippocampus , was introduced in the 1950's [2] . However , even if the CSD is a more local measure of neural activity than the LFP [3]–[8] , the interpretation in terms of underlying activity in neural populations is inherently ambiguous [9] , [10] . Thus in many in vivo applications , for example when investigating receptive fields in sensory systems , the LFP signal was discarded altogether . The LFP signal has seen a revival in the last decade , however . This is due to the rapid development of new silicon-based microelectrodes now allowing for simultaneous recordings of LFP at tens or hundreds of contacts [11]–[14] ( and availability of affordable computer storage ) , the realization among neuroscientists that the LFP offers a unique window into neural activity at the population level [9] , [15]–[23] , and the possibility of using the LFP signal in brain-machine interfaces [24]–[27] . To take full advantage of the opportunities offered by this new recording technology , a precise understanding of the link between the recorded LFP and the underlying neural activity is required . For example , two obvious questions regarding the ‘locality’ of the LFP that need quantitative answers are: ( 1 ) What is the size of the signal-generating region , i . e . , spatial reach , around a recording contact ? ( 2 ) How far does the LFP signal extend outside an active population due to volume conduction ? The first question has been addressed in several experimental studies , with resulting estimates for the spatial reach in cortex varying from a few hundred micrometers to several millimeters [28]–[33] . This large range in reported experimental estimates presumably reflects that the spatial reach depends strongly on the spatiotemporal properties of the underlying spiking network activity , in particular the level of correlations [34] . These critical network features will not only vary between the different brain regions and species studied , but also depend on the brain state . In cortex , thousands of neurons contribute to the LFP , making the signal inherently difficult to interpret . Fortunately , the ‘measurement physics’ , i . e . , the biophysical link between neural activity and what is measured , is well understood: According to well-established volume-conductor theory [10] , [35] , the recorded LFPs stem from appropriately weighted contributions from transmembrane currents in the vicinity of the electrode contact . Building on pioneering work by Rall in the 1960's [35] , [36] , a forward-modeling scheme incorporating detailed reconstructed neuronal morphologies in precise calculations of extracellular potentials has been established [37] and used to explore both spikes [37]–[41] and LFPs [9] , [34] , [41]–[43] generated by single neurons [37]–[40] , [42] and neural populations [9] , [34] , [41] . Unlike in experiments , this modeling scheme allows for a clear separation between volume conduction effects and effects of spatiotemporal variations in spiking network activity in determining population LFPs . In [34] it was used in a thorough investigation of the locality of LFP . It was found that the size of the LFP-generating region depends on the neuron morphology , the synapse distribution and correlations in synaptic activity . For uncorrelated activity , the LFP represents neurons in a small region ( that is , a few hundred micrometers around the electrode contact ) , while in the case of correlated input the size of the generating region is determined by the spatial range of correlated synaptic activity and could thus be much larger . Specifically , it was found that correlated synaptic inputs onto either the apical or basal dendrites of a population of pyramidal neurons could give orders of magnitude larger LFPs , and a much larger spatial reach , compared to the situations with ( 1 ) the same correlated input spread homogeneously over the neuronal dendrite or ( 2 ) similar uncorrelated synaptic inputs placed evenly or unevenly over the neurons . As shown in [34] , the relative contributions to the population LFP from neurons at different distances from the electrode will depend on three factors: First , the amplitude of the LFP generated by a single neuron decays with distance ( typically as for distances beyond a few hundred micrometers , less sharply closer to the neuron ) . Thus single neurons close to the electrode will contribute more to the LFP than if it was placed further away . Second , for a disc-like population , characteristic for a laminar population in a cortical column , it follows that with constant neuron density , the number of neurons located on a ring at a particular radial distance from the electrode will increase linearly with . Third , with correlated synaptic inputs onto a neural population , the LFP contributions from different cells will also become correlated , or synchronized , and will effectively boost the contributions to the LFP . The contributions from different rings of neurons will thus be determined by the interplay of these three factors . In [34] a simplified model for LFP generation based on these elements , ( 1 ) the decay of the single-neuron contribution with the distance from the electrode , ( 2 ) the population geometry , and ( 3 ) the correlation of LFP contributions from individual neural sources , was constructed . We found this simple model to not only give qualitative insight into the generation of population LFPs , but also quantitatively accurate predictions of the size of the signal-generating region and the decay of the signal outside an active population . Here we extend this work by examining the frequency dependence of the LFP . Strong frequency dependencies have been observed both in the tuning properties [28] , [29] and information content [18] , [22] of cortical LFPs . For example , the low-frequency LFP ( less than 12 Hz ) has been shown to carry complementary information to the gamma-range LFP ( 30–100 Hz ) in V1 of macaque monkeys during naturalistic visual stimulation [22] . To properly interpret such experiments , it is thus important to know how spatial reach of the LFP varies across frequencies and whether the biophysics of LFP signal generation boost some frequencies compared to others . The high-frequency LFP components are , for example , expected to be more local than the low-frequency components due to ‘intrinsic dendritic filtering’ [42] , i . e . due to the reduction of the ( effective ) current-dipoles with increasing frequency resulting from the capacitive properties of the dendritic membrane [10] . In [34] we used the biophysical forward-modeling scheme to investigate the total population LFP , i . e . , the total signal generated across all frequencies . Here we use the same scheme to investigate both the distribution of the power of synaptically generated LFP between different frequency bands and the frequency dependence of the locality of the LFP signal . In terms of the latter , we study the size of the signal-generating region ( spatial reach ) as well as the spatial extension of the LFP signal outside an active population — for each frequency component separately . We also use a frequency-resolved ( i . e . dealing with each frequency component separately ) version of the simplified model developed in [34] to guide our investigation of this frequency dependence . The population geometry ( factor 2 ) does obviously not change with frequency . In contrast , the single-neuron LFP contribution ( factor 1 ) decays faster with distance for higher LFP frequencies due to the intrinsic dendritic filtering effect [40] , [42] , but an equally important factor turns out to be the frequency dependence of the ‘correlation transfer’ , i . e . , how correlations in the synaptic input are transferred to correlations between the single-neuron LFP contributions ( factor 3 ) . As an example , Figure 1 illustrates how the frequency-resolved spatial reach varies with the input correlation for a pyramidal population receiving basal synaptic inputs . We show that when the frequency dependencies of factors 1 and 3 are incorporated , the simplified model can still account well for the results obtained by comprehensive numerical investigations . To allow for direct use of the simplified model in future applications , we here thus present and tabulate numerical results for the frequency dependence of these key factors for a variety of situations . Note that we here for simplicity will refer to all calculated extracellular potentials as ‘LFPs’ even if we consider frequencies as high as 500 Hz which sometimes are regarded to be outside the LFP band . Further , spikes , that is , the extracellular signatures of action potentials , may contribute to recorded extracellular potentials at frequencies as low as 100 Hz [40] , [44]–[47] . While the intrinsic dendritic filtering effect [40] and correlations [44] also are critical in determining the contribution from spikes to the LFP , our focus here is on the direct contributions from synaptic inputs . The paper is organized as follows: first we describe the biophysical model of LFP and our simulation setup , present the simplified model of the population LFP , and review its ingredients . Then we present detailed results of the simulations: we analyze the frequency content of the population LFP , the reach of different frequency components , the decay of the signal outside of the population , and the depth-dependence of the LFP . Next we discuss the implications of our results for interpretation of electrophysiological data in terms of the underlying neural activity . Finally , in Methods we give details of the simulation setup and the mathematical model .
Extracellular potentials are generated by transmembrane currents [48] . In the commonly used volume conductor theory , also employed here , the extracellular medium is modeled as a smooth three-dimensional continuum with transmembrane currents representing volume current sources . The fundamental formula relating neural activity in an infinite volume conductor to the generation of the LFP at a position is given by [10] , [37] ( 1 ) Here denotes the transmembrane current ( including the capacitive current ) in a neural compartment positioned at , and the extracellular conductivity , here assumed real ( ohmic ) , isotropic ( same in all directions ) and homogeneous ( same at all positions ) , is denoted by . A key feature of Equation 1 is that it is linear , i . e . , the contributions to the LFP from the various compartments in a neuron sum up . Likewise the contributions from all the neurons in a population add up linearly . The transmembrane currents setting up the extracellular potentials according to Equation 1 are calculated by means of standard multicompartmental modeling techniques , here by use of the simulation tool NEURON [49] . An essential part of the present work is the numerical simulation of the LFP in the center of a disc-like population of cortical cells . The simulation setup is illustrated in Figure 2 . We consider a population of cells distributed homogeneously on a planar disc with a radius of , Figure 2B . The number of cells is chosen to be the same as in [34] and translates to the planar cell density for each population . This density allows for efficient simulations and seems biologically plausible: a total planar density of , say , 50000 cortical neurons per [50] divided by the number of relevant subpopulations ( 5–10 ) , and finally multiplied by the fraction of neurons in the subpopulation receiving synaptic inputs , will give on the order of a few thousand single-neuron LFP sources per . The somas of the cells are all positioned at the same depth , and the LFP is calculated at various ‘virtual electrode’ positions inside and outside the population . In this setup we investigate how the LFP signal increases as contributions from more and more distant neurons are included , i . e . , we study how the root mean square amplitude of the population LFP ( obtained as a sum of single-cell contributions ) depends on the radius of the subpopulation of cells included in the sum ( Figure 2C ) . In the simulations we use three different morphologically-detailed cell models shown in Figure 2B: the layer-3 and layer-5 pyramidal cells , and the layer-4 stellate cells . All neuron models are passive , i . e . , without active conductances , and the extracellular signatures of action potentials ( spikes ) are thus not included . In combination with the use of current-based synapses ( see next paragraph ) this assumption makes the system linear so that each frequency ( Fourier ) component can be investigated separately . For each class of pyramidal cells we consider three different spatial patterns of synaptic input: the synapses are placed either in the apical region only , in the basal region only , or evenly over the whole cell . For the layer-4 stellate cells we consider only spatially homogeneous synaptic input , as these cells lack clearly defined dendritic regions . Each synapse is activated with a Poissonian spike train , the spike trains can be either generated independently for each cell , or chosen from a common pool to model input correlations , Figure 2A . The synaptic currents are modeled as -functions with a very short time constant ( ) to assure that no frequency filtering is imposed by the synapses themselves . In the frequency range considered in the present simulations ( up to 500 Hz ) each synaptic input current thus effectively corresponds to a -function with a white ( flat ) power spectrum . With Poissonian spike statistics , which also implies a white power spectrum , the power spectrum of the input current is flat , Figure 2D . Hence the only frequency filtering in our simulation setup will come from the intrinsic dendritic filtering effect [40] , [42] due to electrical properties of the cable and the summation of the single-neuron LFP contributions to form the population LFP , Figure 2E . If any frequency filtering was to be imposed by the synapse , such as the exponential synapse ( Figure 2D ) , the power spectra of the population LFP would be determined by the superposition of the synaptic and dendritic filters , Figure 2E , i . e . , by multiplying the transfer functions of the two filters . For further details on the simulations we refer to the Methods section . To understand how the population signal emerges from single-cell contributions we use a simplified mathematical model , which is a frequency-resolved version of the model introduced in [34] . We assume that the power spectral density ( PSD ) of the contribution to the LFP from the i-th cell at given frequency can be factorized as ( 2 ) where is the Fourier transform of the single-cell LFP , | is the PSD of the single-cell LFP , is the PSD of the synaptic input current , and is the frequency-dependent shape function of the i-th cell , which carries the information about how the root mean square amplitude of the signal at given frequency decays with distance at a given depth . Moreover , we assume that the shape function of each cell in the population can be replaced with a single , distance- and frequency-dependent function: ( 3 ) that is , we assume that the shape function only depends on the frequency and the lateral distance from the recording electrode ( Figure 2B ) , and neglect variation in the single-neuron LFP contributions due to other factors . For each particular morphology ( layer-3/layer-4/layer-5 ) and synaptic stimulation pattern ( homogeneous/apical/basal ) , the LFP contribution from each cell in the population is thus described with the function . Note that for the special case of white-noise input ( i . e . , ) , the squared shape function will be proportional to the PSD of the single-cell contribution to the LFP . The summation of single-cell LFPs to the population signal depends on the correlation between the single-cell LFP contributions . In the case of uncorrelated input this amounts to simply adding the variances of the single-cell LFPs . For a disc-like population of radius we thus obtain the following expression for the PSD of the signal at the center: ( 4 ) On the other hand , if the single-cell LFPs are fully correlated ( identical ) , the PSD of the signal is found by adding the single amplitudes , not variances , and we thus obtain ( 5 ) In our simulation setup the single-cell LFP contributions from two equidistant neurons ( i . e . , same ) are not identical even for the maximum level of input correlations : while the same spike trains are used to synaptically stimulate the cell , they will not in general activate an identical set of synapses ( see Methods ) . Moreover , as we now work in the frequency domain , the correlation between single-cell contributions to the LFP ( ) is naturally replaced by their coherence ( ) , which , in general , depends on the frequency . If we approximate the LFP coherence between each pair of cells by the population-averaged LFP coherence , then the PSD is given by ( 6 ) where is the contribution resulting from uncorrelated inputs , and represents the contribution of correlated inputs ( see Methods for the full derivation of this formula ) . Note that the root mean square amplitude of the signal ( see Figure 2 ) is related to the PSD throughwhere the integration is between and ( half the sampling frequency ) . Before embarking on the comprehensive numerical evaluation of the ingredients of the simplified model in the next Section and its use in the remainder of the Results , we illustrate in Figure 3 the key features of the model on a specific example , a population of layer-5 cells receiving basal synaptic inputs . The first ingredient that must be determined is the shape function in Equation 3 . Figures 3A and B show the numerically evaluated squared shape functions at the soma level as a function of distance from the neuron ( for three selected frequency bands ) and frequency ( for three distances ) , respectively . Figure 3C illustrates the fitting of the numerical results ( full model ) to a piecewise power-law expression ( see Equation 7 below ) for . The fitted values of the key parameter in this power-law function , the cutoff distance , are found to depend on frequency reflecting the intrinsic dendritic filtering effect ( Figure 3D ) . The second ingredient is the average coherence between single-neuron LFP contributions . The numerically evaluated , shown in Figure 3E for four values of the input correlation , is seen to depend even more strongly on frequency . Next we can plug into the integrals , Equations 4 and 5 , to obtain and , respectively . Finally , the population LFP power is evaluated by combining , and in Equation 6 . The results for the present example are displayed in Figure 3F . As observed , correlated input boosts the low-frequency population LFP up to two orders of magnitude , a key feature which is seen both in the numerical simulations ( dots ) and in the simplified model ( solid lines ) . The population LFP shown in Figure 3F is measured at the center of a population with radius . In the next sections we investigate how the LFP amplitude depends on the various factors and also investigate how local the LFP is in the various situations: First , the size of the signal-generating region is probed by studying how the LFP amplitude measured at the soma level grows when the population radius is increased . From this a measure of the spatial reach can be extracted . Next , we investigate how the measured LFP power decays when the electrode is moved outside the active population . Finally , we investigate the depth-resolved LFP profile , i . e . , the locality of the LFP changes in the vertical direction . Equation 6 implies that any frequency dependence of the population LFP ( for example , frequency dependence of the spatial reach ) in general will result from the interplay of two separate effects: ( 1 ) frequency dependence of the single-cell shape functions and ( 2 ) frequency dependence of the coherence between single-cell contributions to the population signal . These two effects are addressed next . As a first step towards exploring the spatial reach of the extracellular potential in our disc-like setup we next show how the population signal emerges from single-cell contributions and investigate frequency-related effects . In Figure 6 we present results both from the full simulation and the simplified model ( Equation 6 ) for our example situation with the population of layer-5 cells receiving basal synaptic input . In Figure 6A we show the PSD of the LFP produced by differently-sized populations of cells receiving uncorrelated synaptic input . While we observe some low-pass filtering ( especially above ) for all population sizes , the effect is not particularly strong . Figure 6D instead shows the PSD for the same uncorrelated situation as a function of the population radius . We observe that the LFP in all frequency bands saturates rather quickly with increasing population size , that is for . This implies that the contributions from uncorrelated neuronal LFP sources positioned more than a few hundred micrometers away from the electrode are negligible for all frequencies considered . The situation changes dramatically for the case of correlated synaptic input ( Figure 6B , 6C , 6E , 6F ) , both in terms of amplitude and frequency dependence . For the case with the maximum input correlations ( Figure 6C , 6F ) , we see that the low-frequency power is up to two orders of magnitude larger than for the corresponding uncorrelated case . Further , a significant low-pass filtering effect is seen . For example , the low-frequency power ( ) is an order of magnitude larger than the power at for ( Figure 6F ) . Another observation is that the low-frequency power grows much faster with increasing population radius than the high-frequency power ( Figure 6E , 6F ) . Finally , the power of the population signal no longer seems to saturate as the population radius increases [34] . The predictions from the simplified model agree qualitatively with the full simulation results; however , we observe some clear deviations: First , in Figure 6D–F we see that the simplified model overestimates the power of the low-frequency components ( ) . This is because the model here uses the approximate power-law shape functions ( Equation 7 ) which lie above the numerically evaluated shape functions for low frequencies ( Figure 3C ) . For high-frequency components ( 500 Hz ) , on the other hand , the opposite situation occurs ( results for fitted approximate power-law function not shown ) . Second , in case of correlated input the model works better for the larger populations than for smaller ones . This is as expected given the present procedure for calculating the LFP coherence used in the simplified model: here this LFP coherence was extracted from the full population ( ) simulations , and the value obtained is not surprisingly a poor approximation when applied to populations which are much smaller . With calculated for each population radius separately , the simplified model predictions significantly improve ( Figure S1 ) . We are now ready to analyze the frequency dependence of the spatial reach of extracellular potential . Following [34] we define the spatial reach as the radius of the subpopulation which yields 95% of the root mean square amplitude in the population center compared to the largest population considered ( ) . With this definition the spatial reach is easily found from the data presented in Figure 6D , 6E and 6F as the distance at which the amplitude of the LFP reaches 95% of the maximum value . The results for the spatial reach for all seven situations considered are shown in Figure 7 . The reach is seen to vary both with the frequency and the level of input correlation , but the specific effects depend sensitively on the cell morphology and synaptic stimulation pattern . For the pyramidal cells with asymmetric input ( either only basal or only apical ) the spatial reach grows significantly with increasing input correlations ( Figure 7A , 7B , 7D , 7E ) . The effect is particularly prominent for lower frequencies , i . e . , smaller levels of input correlations are needed to increase the spatial reach significantly . As a consequence , for certain correlation levels the spatial reach of the low-frequency components can differ a lot from the spatial reach of the high-frequency components . For example , in the situation with the layer-5 population receiving basal input with , the spatial reach at 100 Hz is only around , while the low-frequency reach is almost . For the case of homogeneous inputs onto pyramidal neurons ( Figure 7C , 7F ) these effects are still present , but seen to be much weaker . For the layer-4 stellate cells the spatial reach is practically independent of the frequency and the input correlation level , Figure 7G . Note that the situation with the layer-5 population receiving only apical input is again somewhat different from the other cases . Here the spatial reach for the uncorrelated input is already quite large ( ) and the levels of the input correlation required to saturate the spatial reach at a maximum value possible in our setup are significantly smaller . For the case of uncorrelated input we can obtain analytical expression for the spatial reach from the simplified model . Using Equations 4 and 7 we obtain an explicit formula for in terms of the cutoff distance and the population radius , Equation 15 . From this , we find in the limit of , that the radius of the subpopulation contributing a fraction of the asymptotic amplitude ( ) is equal to ( valid for ) . For our choice of we find the spatial reach to be . The spatial reach we have discussed above represents an ‘electrode-centric’ point of view: we ask about the distance from the recording electrode of the neurons setting up the LFP signal . However , one can also take a ‘population-centric’ approach and instead ask how rapidly the LFP signal decays with distance outside an active population [34] . In Figure 8 we show results for this situation , still with LFPs recorded at the soma level , for an example population ( ) of layer-5 cells receiving basal or apical synaptic inputs . The first observation in the case of basal synaptic input is that the low- and medium-frequency LFP components ( ) are significantly boosted , up to two orders of magnitude , by high levels of input correlations ( Figure 8A , 8B ) . This applies both inside and outside of the population . For the high-frequency signal ( 500 Hz , Figure 8C ) , however , input correlations are seen to have only a small boosting effect on the signal amplitude . In the case of apical synaptic inputs the effect of increasing input correlations is seen to be more uniform across frequency bands , with the high-frequency components ( 500 Hz ) being boosted by roughly the same factor as the low- and medium-frequency LFP components ( ) , Figure 8D–8F . The strong boosting of the LFP signal seen for correlated synaptic input for ( Figure 8A ) and 60 Hz ( Figure 8B ) has direct implications for how recorded LFP signals should be interpreted . As observed in these panels , the LFP measured a millimeter or more outside a highly-correlated populations can easily be larger than the LFP contribution from a similar , yet uncorrelated population surrounding the electrode . For the example , in Figure 8A we observe that the LFP signal recorded outside a correlated population with is still larger than the contribution recorded inside the same population receiving uncorrelated synaptic inputs ( ) . For 60 Hz ( Figure 8B ) the boosting effect is smaller , but still the signal recorded outside a correlated population may be larger than what is recorded inside an identical population receiving uncorrelated input . This dominance of LFPs from distant correlated populations over uncorrelated populations surrounding the electrode is seen to be even more pronounced for the apical-input case in the lower panels ( Figure 8D–8F ) , further highlighting that the interpretation of the recorded LFPs in terms of activity in the neurons immediately surrounding the electrode has to be done with caution . In Figure 9 we show the same PSDs as in Figure 8 , but normalized to unity at the population center . This illustrates that the decay of the LFP is more abrupt around the population edge in the uncorrelated case than in correlated cases ( this is especially prominent for the low-frequency components ) . This is consistent with an observation made in [51] ( see Figure 3 . 9 therein ) , namely that in the large-population limit the LFP signal power at the population edge will be reduced to half of power at the center for uncorrelated populations , while it will be reduced to a quarter of the center power for fully correlated populations . Here this difference between the correlated and uncorrelated cases is more pronounced for the low-frequency components , where the coherence is largest . In general , there are three key lengths determining the decay outside a population: the size of the population , the anatomical extension of the dendrites of the neurons , and the electrotonic length of the neuronal dendrites . In the examples depicted in Figures 8 and 9 we considered populations of layer-5 cells with a radius . For smaller populations the abruptness of the decay outside the population edge will be less sharp as demonstrated in [51] , but we refrain from a detailed study of the interplay of all these factors here . We next investigated the related question of detectability , i . e . , how far away from a synaptically activated population the generated LFP still can be detected above the ambient LFP ‘noise’ . This noise level will naturally vary between experimental situations , but here we assumed it to be given by the background LFP signal from neurons of the same morphology , receiving the same number and type of synaptic inputs , except that the inputs are ( 1 ) uncorrelated and ( 2 ) homogeneously spread over the neuronal membrane . ( The power of this background LFP signal is plotted as dotted lines in Figure 8 . ) The frequency-dependent signal decay and detectability outside basally-activated populations are illustrated in the 2D color plots in Figure 10 . As in Figure 8 , the population radius is fixed at , and we plot the PSD both inside and outside the population . The lines mark where the signal-to-noise ratio falls below 0 . 5 ( solid line ) and 0 . 1 ( dotted line ) , respectively . Here the signal-to-noise ratio is defined as the ratio between the root mean square amplitudes of the LFP signal ( from the basally-activated population ) and the LFP noise ( from the background population ) . A first observation is that for uncorrelated synaptic inputs ( , Figure 10A–10B ) , there is very little variation with frequency . Also the detectability of the LFP outside the active population is poor: the signal-to-noise ratio falls to 0 . 5 about outside the population , and below 0 . 1 less than outside . The situation is seen to be very different when the populations receive correlated synaptic inputs . Focusing first on the case with the largest level of input correlations ( , Figure 10G , 10H ) , we see that the lower frequencies of LFP extend further outside the population than the higher frequencies . For example , for the near-DC component ( ) the signal-to-noise ratio is seen to be almost 0 . 5 at a distance of , i . e . , outside the population edge , and 0 . 1 as far way as outside this edge . For the 125 Hz component , on the other hand , the signal-to-noise ratio is reduced to 0 . 5 as little as outside the population . The results for the intermediate cases ( ) depicted in Figures 10C–10F are seen to bridge these uncorrelated and strongly correlated cases . The results for the basally-driven pyramidal cell population in Figure 10 demonstrate a main result from this study , namely that correlations in synaptic inputs may significantly enhance the amplitude and thus also the detectability of the low-frequency LFP components relative to the high-frequency LFP components . The same effect is observed for the same population when the synaptic inputs are placed solely on the apical part of the neurons , cf . Figure 11 . However , here a sizable low-pass filtering effect in detectability is observed also for the case with uncorrelated input ( Figure 11A , 11B ) due to the intrinsic dendritic filtering effect [40] , [42] . It is also worth noting that populations of layer-5 cells stimulated apically yielded the farthest-reaching LFP signal of all cases analyzed . Note also that the low-pass filtering effect in the boosting of LFP signal with increasing correlations was seen to be largely absent in the case of a spatially homogeneous distributions of synaptic inputs onto populations made of any of our three example neuronal morphologies ( results not shown ) . Finally , inspection of Figure 8 ( and the PSD line plots in Figures 10 and 11 ) reveals that the predictions from the simplified model ( Equation 6 ) agree excellently with the full numerical simulations for the case of uncorrelated input . However , the simplified model systematically overestimates the signal power for correlated populations for positions far outside the active populations . This is because the simplified model predicts a fall-off of the LFP amplitude proportional to in the far-field limit , while in the full simulations the total LFP signal will be dominated by correlated dipoles oriented vertically . When moving horizontally from a a vertical dipole at a fixed vertical position , it follows from geometry that the dipole potential will decay as rather than [40] . As a consequence the functional form of the lateral decay of the LFP signal outside a correlated population will be close to [34] . This limitation of the simplified model can be remedied by incorporating the fact that the evaluated population-averaged coherence not only depends on the size of the population considered , but also on the electrode position along the horizontal axis from where it is evaluated , i . e . , . So far the population-averaged LFP coherence has been evaluated at the population center , i . e . , at . However , when Equation 17 is evaluated at other positions , as shown in Figure 12 , is observed to decay as for . In the formula for the simplified model in Equation 6 the power is in the correlation-dominated regime seen to be proportional to . A modified simplified theory including not only the X-dependence of [34] , [51] , but also the observed -dependence of ( i . e . , for ) , indeed predicts the correct far-field -dependence outside the active population ( see Figure S8 ) . The physical interpretation is that the dominance of the LFP signal of the correlated vertical dipoles will be incorporated in the population-averaged LFP coherence . Until now we have focused on the LFP calculated at the soma level of each population . However , in general there will be substantial transmembrane currents and thus LFP contributions across the entire dendritic structure [42] . Since the dendrites of the pyramidal cells span several cortical layers , it is natural to ask how the LFP power will depend on the depth . As for the soma-layer LFPs we observe in Figure 13 that the level of correlations is a crucial parameter also here . For example , for the case of uncorrelated ( ) , asymmetric synaptic inputs onto a layer-5 cell population the LFP is essentially located around the inputs ( superficial layers in Figure 13A , layer 5 in Figure 13E ) . However , for strongly correlated synaptic input we instead obtain a dipolar , ‘dumbbell’ pattern with two poles in each end of the dendritic structure of the neuron ( Figure 13B–D , Figure 13F–H ) . Similar behavior can be observed for the population of layer-3 pyramidal cells ( Figure S9 ) . The dipolar structure is not present in case of homogeneous synaptic input onto a layer-5 cell population ( Figure 13I–L ) and for a population of layer-4 cells ( not shown ) . Figure 13B–D , G , H also reveals the same substantial boosting of the low-frequency ( ) dumbbell-shaped LFPs for correlated synaptic inputs as previously seen in Figure 6 . For symmetric or uncorrelated inputs , on the other hand , there is no such boosting , and less relative attenuation of the signal is observed at the higher frequencies . Interestingly and encouragingly the simplified model for the population LFP in Equation 6 captures , as seen in Figure 14 , the salient features of the depth-dependence well . Now the shape curves and the population-averaged coherence depend both on depth and lateral position , as well as frequency , as depicted in Figure 15 . These functional dependencies of the elements of the simplified model also explain why the dumbbell LFP pattern arises for correlated , asymmetric synaptic inputs: As described in [34] , [51] contributions from distant neurons ( ) will dominate over neurons close by ( ) for correlated inputs , and as seen in Figure 15A–B for these distant neurons the shape functions are not too different in magnitude in the various layers . As a consequence substantial LFPs ( which more detailed analysis reveal to have a dumbbell structure ) are thus seen at most cortical depths . For uncorrelated inputs ( ) , or homogeneously distributed correlated inputs resulting in very small correlations between the individual LFP contributions ( Figure 15F ) , the neurons close by ( ) will dominate . Then for the case of basal input , for example , the somatic LFP ( layer-5 ) will be much larger than the LFP in the other layers . The dipolar LFP patterns observed for highly correlated synaptic input are consistent with the patterns observed in [9] , where strongly correlated inputs was implicitly assumed in their more simplified scheme for calculating population LFPs ( see Figure 13 therein ) .
In this computational study we have investigated the frequency dependence of the signal power and ‘locality’ of cortical local field potentials ( LFP ) . While some low-pass filtering effects of the LFP are seen also for populations of cells receiving uncorrelated synaptic inputs or homogeneously distributed correlated synaptic inputs , the large frequency-dependent effects are seen when populations of pyramidal neurons receive correlated and spatially asymmetric inputs ( i . e . , either only basal or apical ) . For example , for the case with a layer-5 population receiving correlated , Poissonian synaptic currents ( with a white-noise , i . e . , flat band , power spectra ) onto their basal dendrites , the power of the low-frequency LFP ( ) was seen to be an order of magnitude larger than the LFP power at 60 Hz . Correspondingly , the low-frequency LFP components were seen to extend much further outside the active population than high-frequency components . The correlation of synaptic input currents and their spatial placement were observed to be equally crucial for determining the vertical profile of the LFP signal . For correlated and spatially asymmetric inputs , characteristic dipolar ‘dumbbell’ LFP structures spanning the vertical extent of the dendrites of the pyramidal neurons in the populations were observed; for uncorrelated and/or spatially homogeneous inputs , the LFP was instead confined around the positions of the somas ( with the exception of uncorrelated apical input onto the layer-5 population ) . The findings from our comprehensive biophysical simulations using reconstructed neuronal morphologies were backed up by a simplified model , adapted from [34] , for generation of population LFP . This model is based on three factors: ( 1 ) the decay of the single-neuron contribution with the distance from the electrode represented by the frequency-dependent shape function , ( 2 ) the population geometry and density of neuronal LFP sources , and ( 3 ) the frequency-dependent correlation ( or , more precisely , coherence ) of the single-neuron LFP contributions from individual neural sources . Our simple model for the population LFP ( Equation 6 ) was found to give quantitatively accurate predictions , implying that it captures the salient features . While some of the observed low-pass filtering could be traced back to single-neuron properties and the intrinsic dendritic filtering effect [40] , [42] accounted for by the shape function , most of the observed low-pass filtering was due to strong low-pass filtering in the coherence between the single-neuron LFP contributions: synaptic-input correlations translated into correlated single-neuron LFP contribution to a much larger extent for lower frequencies than for higher frequencies . As a direct consequence , the low-frequency components of the extracellular potential are significantly boosted in populations with correlated synaptic input . In our model this happens purely because of dendritic filtering , as the synaptic input currents themselves have been tailored to have a flat ( white-noise ) PSD . With a colored ( frequency-dependent ) spectrum of the synaptic input , the power spectrum of the LFP would be given as the product of the PSD of this synaptic filter and the PSD from the dendritic filtering investigated here ( cf . Figure 2D and 2E ) . A key qualitative finding in our study is that the size of the signal-generating region , i . e . , the spatial reach , may in the case of correlated synaptic input vary strongly with frequency . For the example population in Figure 1 we see that for , a plausible correlation level in cortical spiking networks ( see , e . g . , Figure 6 in [34] ) , the LFP spatial reach may be reduced from close to the size of the population ( ) for to for 60 Hz . For uncorrelated input , however , the spatial reach will generally always be small ( ) for all frequencies , with the exception of the case with apical input on large pyramidal cells ( Figure 7 ) . Note that in the present simulation scheme the spatial reach is by definition less than , the size of our model population . Unlike for uncorrelated populations , the LFP power will for correlated populations keep on increasing when the population grows beyond [34] . The present definition of spatial reach ( 95% of the amplitude for ) thus underestimates the true size of the signal-generating region in this case . In a recent experimental study from macaque auditory cortex [33] it was observed that different frequency bands spread equally far from a source ( cf . Figures 5 and 6 there ) . There are , however , notable differences between this study and our present approach , making it difficult to compare the results . First , here we focus mostly on the spread of the LFP along cortical layers at the soma level , while in [33] the spread in vertical direction was studied . Second , and likely more importantly , in [33] the LFP amplitude at a given latency after stimulation was used to extract LFP decay profiles . In contrast , we here use noise input and consider the root mean square amplitude of LFP over a relatively long time period . Further , the correlation level of the synaptic input , found here to be a critical parameter in determining the frequency dependence , is not known in the situation in [33] . It is thus difficult to assess whether our results are in accordance , or not . Our results have direct consequences for the interpretation of observed cross-correlations between extracellular potentials recorded at different electrodes [52]–[57] . As demonstrated here the low-frequency LFP signal generated by a population of neurons around one electrode receiving asymmetric synaptic input , may extend a millimeter or more outside the active population ( see , e . g . , Figure 10G ) . Thus measured correlations in the low-frequency LFP components between two electrodes positioned , say , one millimeter apart , may be due to volume conduction effects . However , cross-correlation induced by such volume conduction will , as demonstrated here , have a diminishing spatial range with increasing LFP frequencies . Note also that the magnitudes of the LFP amplitude at the two adjacent electrodes will aid in the interpretation: while volume conduction may propagate the LFP a millimeter or more , the amplitude will rapidly diminish with distance ( cf . Figure 10 and 11 ) . Thus the observation of large-amplitude LFPs at both electrodes would be an indication that both electrodes are surrounded by strong LFP-generating populations . In [58] the temporal power spectra of the EEG were shown to be well fitted by power-law functions with power-law exponents varying between brain areas: in the frontal lobe was reported to be , while in the occipital lobe . Power laws have also been found in recordings of the LFP , see , e . g . , [59] , [60] , often with different exponents . In [60] was shown to vary between network states , more specifically between the slow-wave sleep and awake states . In this context it is interesting to note that the PSDs in our Figure 3F express approximate power laws with exponents highly dependent on the degree of coherence . This finding suggests that varying levels of coherence in the synaptic input may be a mechanism underlying the different experimentally observed power laws . This would also be in agreement with the experimental observations that network states with a presumably large coherence ( e . g . , slow wave sleep in [60] ) typically express a larger value of than network states for which the coherence is lower ( e . g . , awake state in [60] ) . In our modeling we have assumed the extracellular medium to have a frequency-independent conductivity , an assumption supported by a recent thorough experimental study of the electrical properties of monkey cortical tissue [61] . However , if for example low-frequency filtering of the extracellular medium should be found [62] , this filtering would superimpose directly on the filtering seen here , i . e . , the total LFP filter would be the product of the LFP filter calculated here and the filter from the extracellular medium ( ) . Here we have focused on the spatial and spectral properties of LFP signals triggered by presynaptic spikes that could originate from within the same cortical population or come from other distant brain regions . While not addressed here , it may be that the LFP signal itself influences the timing of these locally generated spikes through ephaptic coupling [63] , [64] . That would in turn influence the correlation structure of incoming spikes and thereby also the generated LFP signal . Since our simulations show that both the LFP amplitude and spatial reach is larger for low than for high frequencies , this suggests that if ephaptic effects play a role in cortical processing , they would likely be larger for low than for high frequencies . The present study has focused on LFPs generated by synaptic input currents and the associated return currents . While these synaptic contributions are thought to dominate at least low-frequency LFPs [9] , [41] , [65] , other sources will also contribute to the signal in the frequency band typically associated with the LFP ( ) . Sodium spikes , i . e . , the fast extracellular signatures of action potentials , may contribute to the LFP signal for frequencies as low as 100 Hz [40] , [44]–[47] , and slower phenomena such as calcium spikes and spike afterhyperpolarization [66] at lower frequencies still . For spikes the source of the LFP is active sodium and potassium conductances localized mainly in the soma and axon hillock , rather than synaptic currents that can be positioned all over the dendrites . Nevertheless , many of our present observations and findings also apply here , in particular , the intrinsic dendritic filtering effect that will give faster decay with distance of the single-neuron contributions for high frequencies than for low frequencies [40] and the possibility of amplification of the population signal when neuron spiking is highly correlated . Interestingly , the latter effect has recently been demonstrated in a very accomplished biophysical modeling study to be the likely mechanism behind the large LFP power observed in the 100–200 Hz frequency band in rat hippocampus [44] . In the present analysis we have modeled the dendrites as simple RC-circuits which , in combination with the use of current-based synapses , made the system linear . This greatly facilitated the present frequency-resolved analysis in that the LFPs at different frequencies were effectively decoupled , cf . the standard theory for Fourier analysis of linear systems . The present results also serve as a starting point for the exploration of non-linear effects , for example due to active membrane conductances . Close to the resting potential of the neuron , the active conductances can be linearized , and the neuron dynamics can be described by linear theory with quasi-active membrane modeled by a combination of resistors , capacitors and inductors ( see , e . g . , Ch . 10 in [67] , Ch . 9 in [68] , or [69] ) . At present it is not known to what extent such ‘generalized’ linear schemes will be able to account for the LFP generation in real neurons , but the present forward-modeling scheme , applicable for passive and active conductances alike , can be used to explore this question systematically .
The setup of the LFP simulations is almost identical to the scheme used to model cortical population LFPs in [34] . The main difference is that here we use a much smaller synaptic time constant to achieve an effectively white ( flat ) power spectrum for the synaptic currents for the frequencies of interest here ( less than 500 Hz ) . We therefore also use a smaller numerical time step . The model parameters are presented in detail ( in the format described in [70] ) in Tables S1 , S2 and S3 . For the reader's convenience we summarize the essential information below . To derive the formula in Equation 6 for the power spectral density ( PSD ) of the extracellular signal in the center of the population we start with the assumption that , the PSD of the contribution of the i-th cell at given frequency , may be factorized as ( 8 ) where is the PSD of the input current , and is the frequency-dependent shape function of the i-th cell . We also assume that the shape function depends only on frequency and distance from the center , that is: ( 9 ) Let us compute the PSD of the population signal ( dependence on frequency dropped below for convenience ) : ( 10 ) We now use Equations 8 and 9 to express in terms of shape functions and the PSD of the input current , note the trick ( multiplication by ) in the double sum: ( 11 ) We further assume that the coherence term may by replaced by its population average over pairs . This assumption , while not true in general , is a reasonable approximation because the input correlations are homogeneous across the population . We can then move the coherence term in front of the double sum: ( 12 ) As marked in Equation 12 , we denote the population-averaged coherence by . We further rewrite as ( 13 ) and finally ( 14 ) If we assume approximate , power-law shape functions parametrized by the cutoff distance ( Equation 7 ) , and change sums to integrals as in Equations 4 and 5 ( limit of large number of cells ) , then the functions and have the following closed-form representation [51]: ( 15 ) ( 16 ) which we used for calculating predictions from the simplified model . At the soma level we effectively set to zero; for modeling the LFP at any different layer we used [51] . The model can be modified to calculate the power of the signal outside the center of the population , i . e . , at positions offset from the center by the distance . In that case , the function in ( 4 ) and ( 5 ) has to be replaced by . It is no longer easy to obtain closed-form formulae for and in terms of , and we used the ( non-parametric ) shape curves obtained from the simulations , as the final integration had to be done numerically anyway . | The first recording of electrical potential from brain activity was reported already in 1875 , but still the interpretation of the signal is debated . To take full advantage of the new generation of microelectrodes with hundreds or even thousands of electrode contacts , an accurate quantitative link between what is measured and the underlying neural circuit activity is needed . Here we address the question of how the observed frequency dependence of recorded local field potentials ( LFPs ) should be interpreted . By use of a well-established biophysical modeling scheme , combined with detailed reconstructed neuronal morphologies , we find that correlations in the synaptic inputs onto a population of pyramidal cells may significantly boost the low-frequency components and affect the spatial profile of the generated LFP . We further find that these low-frequency components may be less ‘local’ than the high-frequency LFP components in the sense that ( 1 ) the size of signal-generation region of the LFP recorded at an electrode is larger and ( 2 ) the LFP generated by a synaptically activated population spreads further outside the population edge due to volume conduction . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"physics",
"cellular",
"neuroscience",
"neuronal",
"morphology",
"neurophysiology",
"central",
"nervous",
"system",
"biophysic",
"al",
"simulations",
"computational",
"neuroscience",
"electric",
"field",
"electricity",
"biology",
"computational",
"biology",
"biophysics",
"s... | 2013 | Frequency Dependence of Signal Power and Spatial Reach of the Local Field Potential |
p53-signaling is modulated by viruses to establish a host cellular environment advantageous for their propagation . The Epstein-Barr virus ( EBV ) lytic program induces phosphorylation of p53 , which prevents interaction with MDM2 . Here , we show that induction of EBV lytic program leads to degradation of p53 via an ubiquitin-proteasome pathway independent of MDM2 . The BZLF1 protein directly functions as an adaptor component of the ECS ( Elongin B/C-Cul2/5-SOCS-box protein ) ubiquitin ligase complex targeting p53 for degradation . Intringuingly , C-terminal phosphorylation of p53 resulting from activated DNA damage response by viral lytic replication enhances its binding to BZLF1 protein . Purified BZLF1 protein-associated ECS could be shown to catalyze ubiquitination of phospho-mimetic p53 more efficiently than the wild-type in vitro . The compensation of p53 at middle and late stages of the lytic infection inhibits viral DNA replication and production during lytic infection , suggesting that the degradation of p53 is required for efficient viral propagation . Taken together , these findings demonstrate a role for the BZLF1 protein-associated ECS ligase complex in regulation of p53 phosphorylated by activated DNA damage signaling during viral lytic infection .
The tumor suppressor p53 plays an important role in maintaining genomic integrity [1] , [2] . In unstressed normal cells , p53 usually exists in a hypophosphorylated form at only low levels due to rapid degradation through the ubiquitin-dependent proteasome pathway [3] . MDM2 is a key regulator of turnover by binding to p53 and promoting its ubiquitination by acting as an E3 ubiquitin ligase . In response to DNA damage , p53 is phosphorylated at S15 by ataxia-telangiectasia mutated ( ATM ) and then T18 by casein kinase 1 , preventing the interaction with MDM2 , subsequently leading to escape from proteasomal degradation [4] . The p53 protein level becomes elevated , resulting in an increase in p53-dependent transcription of its target genes , subsequently leading to cell cycle arrest or apoptosis [5] , [6] . Ubiquitination is important for the regulation of a variety of cellular processes , including signal transduction , development , apoptosis , cell cycle progression , and the immune response [7] , [8] , [9] . The ubiquitination of a substrate requires a cascade of enzymatic reactions involving an E1 activating enzyme , an E2 conjugating enzyme , and finally an E3 ligase enzyme that covalently attaches ubiquitin to a lysine residue of the target protein [10] . The latter enzyme is the most diverse , demonstrating substrate specificity and determining the rate of ubiquitin conjugation . The E3 ligase itself can be either a single protein or a multiprotein complex . Cullin-containing ligases constitute a large class of E3s [11] , primarily consisting of a substrate-specific adaptor protein , the scaffold protein Cullin , and a RING finger-containing protein that interacts with E2 ligase [12] , [13] , [14] . For evading host security responses and generating an advantageous environment for viral replication , a number of viruses have evolved sophisticated mechanisms to utilize or manipulate the host ubiquitin system . Specially , DNA viruses target p53 for inactivation through the ubiquitin-proteasome pathway . The E6 protein of the high-risk human papillomaviruses and the cellular ubiquitin-protein ligase E6AP form a complex which causes ubiquitination and degradation of p53 [15] . The adenovirus E1B 55-kDa protein binds to both p53 and E4orf6 , and recruits a Cullin-containing complex to direct the ubiquitin-mediated proteolysis of p53 [16] . However , in comparison with the effects of the smaller DNA viruses , much less is known regarding the precise mechanisms whereby the Epstein-Barr virus ( EBV ) inhibits transcriptional functions of p53 . EBV , a human gamma-herpesvirus , is associated with several B-cell and epithelial-cell malignancies and can choose between two alternative infection states; latent and lytic [17] . Infection is primarily latent , but EBV periodically reactivates and replicates in a lytic manner in a subset of B cells , which is essential for viral propagation and transmission . The switch from latent to lytic infection is triggered by the BZLF1 protein [18] , a b-Zip transcriptional factor which binds to the promoters of early lytic genes [19] , [20] . Induction of the EBV lytic program elicits a cellular DNA damage response with activation of the ATM-dependent DNA damage signal transduction pathway [21] . Although ATM-dependent DNA damage signaling is activated and consequently p53 is phosphorylated at various sites including S15 , the levels of p53-downstream targets are maintained at very low levels especially at middle to late stages of the infection [21] , [22] . This poses the fascinating puzzle of why p53-downstream signaling is blocked in EBV lytic infection even when p53 is phosphorylated through activation of the ATM-mediated DNA damage response . Recently , we found that the BZLF1 protein induces p53 degradation during the lytic infection [23] . In this study , we investigated the mechanisms by which the viral immediate-early ( IE ) BZLF1 protein targets p53 for degradation and revealed that the protein directly functions as an adaptor component of the ECS ( Elongin B/C-Cul2/5-SOCS-box protein ) ubiquitin ligase complex targeting phosphorylated p53 for degradation during the lytic replication .
We performed a bioinformatics search for structural motifs within the BZLF1 protein that interact with the established ubiquitin E3 ligase , to identify the E3 ligase responsible for the BZLF1 protein-mediated degradation of p53 during lytic infection [23] . Since the BZLF1 protein does not possess a RING finger domain , it might not have an intrinsic E3 ligase activity by itself . By this approach , we found that the BZLF1 protein possesses putative Cul2 and Cul5 binding motifs termed the Cul2-box and Cul5-box [14] , [24] , respectively ( Figure 1A ) . Cul2 and Cul5 are subunits of the well-defined ECS ubiquitin ligase complex [14] . To investigate whether the BZLF1 protein can associate with Cul2 and Cul5 , three BZLF1 protein mutants ( M1-M3 ) were constructed ( Figure 1A ) : M1 , which contains mutations at positions 44 and 45 in the Cul2 box ( LP to AA ) ; M2 , which contains mutations at positions in the Cul5 box ( LPEP to AAAA ) ; M3 , which contains mutations in both Cul2- and Cul5- boxes . As shown in Figure 1B and S1A , a series of IP assays using tagged protein showed wild-type BZLF1 protein to interact with both Cul2 and Cul5 . In contrast , the M1 mutant did not interact with Cul2 while the M2 mutant lacked any reaction with Cul5 . These observations strongly suggest that the Cul2/Cul5-boxes in the N-terminus of BZLF1 protein are important for the physical association between the BZLF1 protein and Cul2/Cul5 E3 ligases . To specifically address the question of whether BZLF1 protein interacts with Cul2 and Cul5 during lytic infection , IP assays were performed . The BZLF1 protein actually associated with both Cul2 and Cul5 and formed to ECS complex in lytic replication-induced 293/EBV cells ( Figure 1C and S1B ) . These interactions were also confirmed in lytic infection-induced B95-8 cells ( Figure S1C ) . Next , we examined if the BZLF1 M1-M3 mutants could induce reduction of p53 level . Contrary to wild-type BZLF1 and M1 , M2 mutants , the M3 mutant failed to decrease the level of p53 ( Figure 1D ) . To further investigate whether the BZLF1 protein-mediated degradation of p53 is dependent on the ECS complex , we applied RNAi experiments . Two independent sequences for each molecule were used as targets of RNAi . Since gene sequences for marmoset Cul2 and Cul5 could not be obtained from the NCBI database , we used 293/EBV cells to examine the effects of RNAi-knockdown of Cul2 and Cul5 . As shown in Figure 1E , transfection of various combinations of plasmids encoding shRNAs specific for Cul2 and Cul5 mRNAs , as well as a BZLF1 expression plasmid for induction of lytic infection , caused significant increase in the level of p53 , compared with that apparent in cells transfected with either a control plasmid for EGFP shRNA or an empty plasmid . Consistent with the findings in Figure 1D , we indeed confirmed that a knockdown of either Cul2 or Cul5 did not alter the p53 abundance in the lytic infection ( data not shown ) . Hence , these results indicate that BZLF1 protein-mediated degradation of p53 is dependent on the ECS complexes . To demonstrate that the BZLF1-ECS complex directly promotes p53 ubiquitination in vitro , we reconstituted complexes in Sf21 cells and assayed purified preparations for stimulation of p53 ubiquitination . Sf21 insect cells were coinfected with a series of recombinant baculoviruses encoding each subunit of the BZLF1-Cul2 complex ( N-terminal His-and Flag-tagged BZLF1 protein ( RHF-BZLF1 ) , myc-Rbx1 , HA-Cul2 , Elongin B and Elongin C ) , or the BZLF1-Cul5 complex ( RHF-BZLF1 , 3×myc-Rbx2 , HA-Cul5 , Elongin B and Elongin C ) , and the resultant BZLF1-ECS complexes were purified [25] . Incubation of purified p53 substrate with RHF-BZLF1-Cul2 or RHF-BZLF1-Cul5 complexes resulted in ubiquitination of p53 in the presence of E1 , E2 , GST-ubiquitin and ATP , whereas purified BZLF1 protein alone did not process activity for ligating ubiquitin ( Figure 2A ) . Omitting BZLF1 protein from the ECS complex abolished p53 ubiquitination and the RHF-BZLF1-Cul2/Cul5 complex catalyzed the ubiquitination of p53 in a dose-dependent manner ( Figure S2A ) . To further confirm p53 ubiquitination by the BZLF1 protein-associated ECS complex , we performed time-course and drop-out experiments . Ubiquitinated p53 increased in a time-dependent manner ( Figure S2B and C ) and was not detected without E1 , E2 or ATP ( Figure S2D and E ) . Substitution of the d200-227 mutant BZLF1 protein , lacking the interaction with p53 , for the wild-type in the RHF-BZLF1-Cul2/Cul5 complex eliminated the enhanced p53 ubiquitination ( Figure 2B and S2F ) . Thus , the findings suggest that the BZLF1 protein recruits p53 to the ECS ligase complex for polyubiquitination , functioning as an adaptor for substrate recognition in the complex . The mechanisms underlying regulation of the association between p53 and BZLF1 protein in cells is of considerable interest . Although the association between full-length p53 and BZLF1 protein is well-characterized [21] , [23] , [26] , the domain of p53 interacting with the BZLF1 protein remains obscure . To address this , the interaction domain of p53 with the BZLF1 protein was found to be located within the DNA binding domain from IP analyses with deletion mutants of p53 ( Figure 3A ) . Furthermore , the BZLF1 protein appeared to have affinity for C-terminus truncated mutants rather than full-length p53 . Interactions between certain classes of ubiquitin-ligating enzymes and their targets are tightly regulated by posttranslational modifications such as phosphorylation [27] . The ATM-Chk2 DNA damage signaling pathway is activated in the EBV lytic phase [21] and it was recently reported that Chk2-mediates phosphorylation of p53 at S366 and S378 in response to genotoxic stress [28] . These prompted us to assume that Chk2 might play a pivotal role in phosphorylation of p53 during lytic infection . Indeed , p53 was found to be phosphorylated at least at S15 , S20 , S366 and S378 with progression of EBV lytic infection ( Figure 3B ) . However , the presence of significant redundancy should be kept in mind since the same p53 residue can be phosphorylated by several different kinases [29] . Based on structural analysis of p53 [30] , it is speculated that the DNA binding domain is masked by its C-terminal regulatory domain rich in basic amino acids . Furthermore , phosphorylation of the C-terminal regulatory domain results in increased binding to DNA [31] . Thus , it is reasonable to assume that conformational change induced by phosphorylation of the regulatory domain enhances the association with the BZLF1 protein . To check this hypothesis , phospho-mimetic mutants of p53 ( S15E , S15E&S37E , S366E , S378E , and S366E&S378E ) His-tagged at the N-terminus were expressed in E . coli and purified . In vitro pull-down assays with His-p53 wild-type or mutants as bait were then performed using TALON His-tag affinity resin . As shown in Figure 3C , the bacterially expressed p53 S15E , S15E&S37E , S366E and S378E mutants showed negligible differences from wild-type p53 in their affinity for the BZLF1 protein . In sharp contrast , double mutations of S366 and S378 residues stimulated binding to the BZLF1 protein . To confirm the role of these p53 C-terminal phosphorylations , we generated p53 mutant in which residues 366 and 378 to nonphosphorylatable alanine . The p53 protein ( WT or S366A&S378A ) was expressed in 293T cells , phosphorylated by ionizing irradiation ( IR ) , purified by anti-FLAG antibody and pull-downed with recombinant BZLF1 protein . In contrast to wild-type , p53 S366A&S378A mutant abolished the enhanced binding to BZLF1 protein under the IR stress ( Figure 3D ) . These results imply that the phosphorylation of p53 at both S366 and S378 stimulates the association between p53 and BZLF1 protein . To further test whether phosphorylation-mediated enhancement of the association affects p53 ubiquitination , we performed in vitro ubiquitination assays . As shown in Figure 3E and S2G , BZLF1 protein-associated ECS complexes more efficiently ubiquitinated S366E&S378E mutant as compared to the wild-type p53 . Taken our results together , the C-terminal phosphorylation of p53 by Chk2 appears to stimulate ubiquitination through increase in the binding affinity for the BZLF1 protein . Induction of lytic replication by wild type BZLF1 protein results in low level of p53 , while induction by the M3 mutant did not reduce the level ( Figure 4A ) . It turned out that the inhibition of p53 degradation by the M3 mutant increased PARP cleavage , well-defined apoptotic marker ( Figure 4A ) . We further analyzed the virus yield from the lytic-induced 293/EBV cells transfected by BZLF1 wild type or M3 mutant expression vector . As shown in Figure 4B , the yield of infectious virus from the BZLF1 M3 mutant expressed cells was poor than that of virus from wild-type BZLF1 expressed cells . The simultaneous transfection of both p53 and BZLF1 protein also produced low yield of infectious virus . To further assess whether the degradation of p53 during lytic infection is linked to anticipate effects on virus production , we examined temporal linkage of the p53 effect on viral DNA replication . The compensation for p53 at the middle and late stages of the lytic infection when p53 level was decreased interfered with viral genome synthesis ( Figure 4C ) . These findings suggest that the degradation of p53 contributed to prevent apoptosis and was required for the efficient viral propagation in the lytic replication .
In this study , we revealed p53 to be degraded via an ubiquitin-proteasome pathway even under conditions of up-regulated ATM-dependent DNA damage signaling in the EBV lytic phase . Our data clearly indicate that the EBV BZLF1 IE protein plays a critical role in the degradation of p53 independent of MDM2 . The BZLF1 protein interacts with Cul2 and Cul5 through the Cul2- and Cul5-box motifs , located within its N-terminus . BZLF1-Cul2/Cul5 complexes proved capable of reconstituting a multiprotein ECS complex with ubiquitin ligase activity . The large body of evidence implicating Cul2- and Cul5-containing E3 ubiquitin ligases in regulation of diverse cellular processes [32] provides us with new insights into their significance as potential targets of viruses trying to manipulate the host cellular system . Several viral proteins have the capacity to assemble with Cullin-based ubiquitin ligase modules and act as E3 ligases . For instance , Vif encoded by HIV-1 interacts with Cul5 through a zinc-binding HCCH , a unique viral motif directing ubiquitin-mediated proteolysis of APOBEC3G , a host defense factor that causes hypermutation in newly synthesized viral DNA [33] . A major difference between the BZLF1 protein and these viral proteins is that BZLF1 protein associates not only with Cul2 but also with Cul5 , indicating redundancy in the EBV machinery to downregulate p53 . Although the molecular mechanisms are controversial , a variety of reports on the herpesvirus family have pointed to inactivation of p53 in lytic infection [34] , [35] , [36] . Consistently , overexpression of p53 in infected cells interferes with efficient expression of viral genes ( data not shown ) . Given the importance of inhibiting p53-mediated transactivation to adapt the cellular environment for viral propagation , the apparent redundancy is not so surprising . The present study indicated the existence of distinct mechanisms of p53 quantitative regulation in the latent and lytic phases of EBV infection , as schematically illustrated in Figure 5 . We observed that disruption of p53 binding to MDM2 by Nutlin-3 increased the level of p53 in latent phase but not during lytic infection [23] . Since induction of the EBV lytic program activates the ATM-Chk2 DNA damage-signaling pathway [21] , p53 is phosphorylated at least at S15 and S20 but its level is nevertheless downregulated . Under these conditions , MDM2 hardly interacts with N-terminal phosphorylated p53 [4] , [37] , implying that EBV possesses another strategy to ubiquitinate phospho-p53 to block downstream signaling during lytic infection . On the basis of our findings with the BZLF1 protein and Cul2- and Cul5-containing ubiquitin ligase complexes , we propose a model for recognition and ubiquitination of p53 by the BZLF1 protein-associated E3 ligases ( Figure 5 ) . A requirement of these complexes for effective p53 degradation was supposed to achieve the efficient viral replication . In addition , the phosphorylation at S366 and S378 by virus-induced DNA damage response enhances the association with BZLF1 protein and ubiquitination of p53 . To our knowledge , except for CARPs [38] , neither RING nor HECT type E3 ligases [7] have previously been demonstrated to recognize and ubiquitinate phosphorylated p53 for degradation , including MDM2 , COP1 , Pirh2 or E6/E6AP . Thus , this finding is one of the most interesting aspects of our study . Wen and colleagues revealed that BZLF1 is expressed as an immediate-early gene following primary EBV infection of B lymphocytes although early and late lytic gene expression is not observed [39] . They speculate that BZLF1-expressing cells are the only ones that survive and establish latency . In addition , it was reported that BZLF1 expression in early-passage lymphoblastoid cell lines may contribute to tumor formation in nudemice [40] and cellular gene expression [41] . Thus , it is noteworthy to mention the possibility that the degradation of p53 by BZLF1 protein-associated ECS ubiquitin ligases contributes to efficient establishment of latent infection at the early stages of primary EBV infection or tumor formation in vivo .
SaOS-2 cells , 293T cells , and 293/EBV cells were grown in DMEM supplemented with 10% fetal calf serum ( FCS ) . 293/EBV cells were prepared by transfection with EBV-Bac DNA [42] into 293 cells by hygromycin selection ( hygromycin B; 150 µg/ml ) . EBV-Bac was gifted by Wolfgang Hammerschmidt ( Helmholtz Zentrum München-Haematologikum , Germany ) . EBV-positive marmoset B lymphocytes B95-8 cells and Tet-BZLF1/B95-8 cells were described previously [22] . For MG132 ( Sigma ) experiments , cells were treated with MG132 ( 20 µM ) for 3–5 h , before harvesting . Anti-p53 ( FL-393 ) and anti-Cul5 ( H-300 ) rabbit polyclonal antibodies , anti-Rbx2 ( N-15 ) goat polyclonal antibodies and normal mouse IgG2a were purchased from Santa Cruz Biotechnology . Anti-Phosho-p53 ( Ser15 ) , anti-Phosho-p53 ( Ser20 ) , anti-Phopho-Chk2 ( Thr68 ) and anti-Cleaved PARP ( Asp214 ) rabbit polyclonal antibodies were obtained from Cell Signaling Technology . Mouse anti-Elongin C ( BD Transduction Laboratory ) , mouse anti-Elongin B ( BioLegend ) , mouse anti-GAPDH ( Ambion ) , monoclonal mouse anti-p53 ( Ab-6 ) ( Merck ) , mouse anti-MDM2 ( Ab-3 ) ( Merck ) and mouse anti-FLAG M2 ( Sigma ) antibodies were also used . Anti-Cul2 rabbit polyclonal antibody and horseradish-peroxidase-conjugated secondary antibodies were purchased from Zymed Laboratories . Affinity-purified anti-BZLF1 and anti-Rbx1 antibodies were prepared as described previously [43] . Anti-phosho-p53 ( Ser366 ) and ( Ser378 ) antibodies were a generous gift from Sheau-Yann Shieh ( Institute of Biomedical Sciences , Taiwan ) . These phospho-specific antibodies are used for immunoprecipitation , and are not suitable for immunoblot analysis . Mammalian expression vectors for human wild-type p53 ( pcNXRS ) and the BZLF1 protein expression vector ( pcDNA-BZLF1 ) were kindly provided by Takashi Takahashi ( Nagoya University , Japan ) and Kiyotaka Kuzushima ( Aichi Cancer Center Research Institute , Japan ) , respectively . Full-length cDNAs of BZLF1 , Cul2 , Cul5 , and ubiquitin were obtained by reverse transcriptase-PCR ( RT-PCR ) and subcloned into the pcDNA4/TO/myc-His vector ( Invitrogen ) . For expression of the epitope tagged protein , a FLAG-TEV-HA ( FTH ) or HA tag cassette was inserted into the plasmids encoding the respective cDNAs . For the expression of FLAG-tagged p53 , constructs expressing p53 full-length wild-type ( wtp53 ) and deletion mutants ( p53/1–200 , p53/100–292 , p53/169–393 , p53/309–368 , p53/1–322̂323–393 ) in p3×FLAG-CMV-14 ( Sigma ) were prepared by PCR . BZLF1 mutants ( M1 , L44A P45A; M2 , L52A P53A E54A P55A; M3 , L44A P45A L52A P53A E54A P55A in pcDNA4A ) and p53-mutant ( L323A Y327A L330A; and S366A&S378A in p3×FLAG-CMV-14 and S15E; S15&S37E; S366E; S378E and S366E&S378E in pET28b ( Novagen ) ) expression plasmids were generated by site-directed mutagenesis . A BZLF1 deletion mutant ( BZLF1 d200-227 ) , lacking part of the Zip domain , was generated by overlapping PCR . The inserted DNA sequence of each vector was confirmed by direct DNA sequencing . Cells were seeded , cultured to semi-confluence and transfected with expression plasmids using lipofection reagent ( Lipofectamine™ and Plus reagent; Invitrogen ) according to manufacturer's instructions . Knockdown of Cul2 or Cul5 was achieved by electroporation with shRNA plasmids as described previously [14] . Two days after transfection , cells were harvested and subjected to immunoblotting . Cells were lysed in lysis buffer ( 50 mM Tris-HCl pH 7 . 6 , 120 mM NaCl , 0 . 1% NP40 , 1 mM EDTA , 100 mM sodium fluoride , 2 mM sodium vanadate ) containing a protease inhibitor cocktail ( Sigma ) , and then sonicated . The debris was removed by centrifugation and the supernatants were applied for immunoprecipitation with specific antibodies . Complexes of antibody and antigen were collected by centrifugation and washed three times with NET-gel buffer ( 50 mM Tris-HCl pH 7 . 6 , 150 mM NaCl , 0 . 1% NP40 , 1 mM EDTA ) . The immunoprecipitates were then subjected to SDS-PAGE followed by immunoblot analyses . Preparation of the lysate for immunoblotting , Western blotting and detection of signals were performed as described previously [44] . Immunoreactivity was detected by Western Lightning ( Perkin-Elmer ) and images were processed with LumiVision PRO 400EX ( Aisin/Taitec Inc . ) . Signal intensity was quantified with a LumiVision Analyzer 400 . The system used in this study mounts a cooled CCD camera that has a 16 bit = 65 , 535 grayscale wide dynamic range . It enhances the accuracy of the quantitative analysis up to 100 times compared with ordinary quantitative analysis scanning of an X-ray film into the personal computer after exposing the signal to the film . Protein levels were quantified by the densitometry in triplicate experiments , and the results were expressed as ratios between the specific band under examination and appropriate internal control . For preparing reconstituted RHF-ECS complexes , lysates from Sf21 cells co-infected with baculoviruses encoding RHF-BZLF1 , HA-Cul2/5 , myc-Rbx1/2 , Elongin B and Elongin C , were applied to Ni-nitrilotriacetic acid agarose ( QIAGEN ) as described previously [16] , [25] . HA-tagged p53 was expressed in Sf21 cells infected with a recombinant baculovirus [45] , kindly provided by Carol Prives ( Columbia University ) . HA-p53 protein was purified using a monoclonal anti-HA agarose conjugate ( Sigma ) and elution was performed using an HA peptide . To prepare His-tagged p53 , His-p53 and its phospho-mimetic mutants were expressed in bacteria , and the expressed proteins were purified using Ni-nitrilotriacetic acid agarose . In vitro ubiquitination assays were performed as described previously [16] with some modifications . Reaction mixtures were incubated for 1 h at 26°C , separated by SDS-PAGE , and analyzed by IB with anti-p53 ( Ab-6 ) antibody . Hi-Five cells were infected with recombinant baculovirus AcBZLF1 , harvested 72 h post-infection , and then suspended in hypotonic buffer ( HB; 40 mM Tris-HCl pH 7 . 6 , 1 mM EDTA , 1 mM EGTA , 1 mM DTT , 1 mM PMSF , 0 . 2% Triton X-100 , 10 µg/ml leupeptin , 10 µg/ml pepstatin ) , followed by homogenization using a Dounce homogenizer . The resultant nuclei were freeze and thawed , resuspended in 0 . 1 M NaCl-HB , and precipitated again . After washing with 0 . 2 M NaCl-HB twice , extraction of the BZLF1 protein was performed with 0 . 6 M NaCl-HB . The purity of the recombinant BZLF1 protein is more than 90% . Tet-BZLF1/B95-8 cells ( 1×106 cells ) were treated with doxycycline and then transfected with p53 expression plasmid or empty plasmid as a control at the indicated times . Total DNAs were purified from the cells at 48 h post-induction . Dot-blot hybridization was performed using DIG-labeling system ( Roche ) and viral genome replication was quantified as described previously [22] . For titration of virus yields , 293/EBV cells were transfected with BZLF1 expression plasmid using a microporator ( Digital Bio ) to induce lytic replication . Cells and the culture supernatant were collected , freeze-thawed , and centrifuged . The supernatant from the centrifugation was filtered and used as a virus stock . EBV-negative Akata ( - ) cells [46] ( kindly provided by Kenzo Takada , Hokkaido University ) were infected with the virus and EGFP positive cells were counted by FACS . Data are presented as mean±S . E . . Statistical analysis has been carried out using Student's t-test Values were considered significantly different when p<0 . 05 . The Entrez Gene accession numbers for genes and gene products discussed in this study are as follows: p53 ( 7157 ) , ubiquitin ( 7314 ) , Cul2 ( 8453 ) , Cul5 ( 8065 ) , and BZLF1 protein ( 3783744 ) . | Inhibition of p53-mediated transactivation is essential for regulating the cellular environment advantageous for viral infection . Specially , DNA viruses target p53 for inactivation through the ubiquitin-proteasome pathway . The E6 protein of the high-risk human papillomaviruses and the cellular ubiquitin-protein ligase E6AP form a complex which causes ubiquitination and degradation of p53 . The adenovirus E1B 55-kDa protein binds to both p53 and E4orf6 , and recruits a Cullin-containing complex to direct the ubiquitin-mediated proteolysis of p53 . However , in comparison with the effects of the smaller DNA viruses , much less is known regarding the precise mechanisms whereby the Epstein-Barr virus ( EBV ) inhibits functions of p53 . EBV possesses two alternative life cycles , latent and lytic replication . In latent phase , p53 is regulated by MDM2 ubiquitin ligase while after induction of lytic replication p53 is phosphorylated and the level of activated p53 is regulated by a novel system independent of MDM2 . This report describes a unique functional role of the BZLF1 protein encoded by EBV in the modulation of activated p53 . In this pathway , BZLF1 protein serves as an adaptor molecule for both Cul2- and Cul5-containing E3 ubiquitin ligase complexes to stimulate the ubiquitination and degradation of p53 for inhibiting apoptosis , indicating redundancy in the EBV machinery to downregulate p53 level . Therefore , it would be possible that the complexes regulate not only p53 but also various proteins that interact with BZLF1 protein . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"biology/post-translational",
"regulation",
"of",
"gene",
"expression",
"virology",
"cell",
"biology/cell",
"signaling"
] | 2009 | Degradation of Phosphorylated p53 by Viral Protein-ECS E3 Ligase Complex |
Intracellular pathogens include all viruses , many bacteria and parasites capable of invading and surviving within host cells . Key to survival is the subversion of host cell pathways by the pathogen for the purpose of propagation and evading the immune system . The intracellular bacterium Shigella flexneri , the causative agent of bacillary dysentery , invades host cells in a vacuole that is subsequently ruptured to allow growth of the pathogen within the host cytoplasm . S . flexneri invasion has been classically described as a macropinocytosis-like process , however the underlying details and the role of macropinosomes in the intracellular bacterial lifestyle have remained elusive . We applied dynamic imaging and advanced large volume correlative light electron microscopy ( CLEM ) to study the highly transient events of S . flexneri’s early invasion into host epithelial cells and elucidate some of its fundamental features . First , we demonstrate a clear distinction between two compartments formed during the first step of invasion: the bacterial containing vacuole and surrounding macropinosomes , often considered identical . Next , we report a functional link between macropinosomes and the process of vacuolar rupture , demonstrating that rupture timing is dependent on the availability of macropinosomes as well as the activity of the small GTPase Rab11 recruited directly to macropinosomes . We go on to reveal that the bacterial containing vacuole and macropinosomes come into direct contact at the onset of vacuolar rupture . Finally , we demonstrate that S . flexneri does not subvert pre-existing host endocytic vesicles during the invasion steps leading to vacuolar rupture , and propose that macropinosomes are the major compartment involved in these events . These results provide the basis for a new model of the early steps of S . flexneri epithelial cell invasion , establishing a different view of the enigmatic process of cytoplasmic access by invasive bacterial pathogens .
The lifestyle of intracellular bacterial pathogens is generally divided into the following steps: contact and entry into the host cell , residence within a vacuole , escape into the cytosol or establishment of a membrane encased niche , and cell-to-cell spreading [1] . Some intracellular bacterial pathogens , called invasive bacteria , such as Yersinia pseudotuberculosis , Listeria monocytogenes , Salmonella enterica and Shigella flexneri , are able to induce their own entry into nonphagocytic host cells [2] . Invasive bacteria are thought to subvert various host endocytic pathways during their life cycle , allowing the establishment of their specific intracellular niche and evasion of host immunity [3] , [4] . S . flexneri is a medically important pathogen [5] that uses a type III secretion system ( T3SS ) to translocate bacterial proteins called effectors into the host cell [6] , [7] . Induction of macropinocytosis has been proposed as the invasion strategy for S . flexneri entry into non-phagocytic epithelial cells [2] . In short , effectors released upon cell contact induce major rearrangements of the host cell cytoskeleton , mainly polymerization of actin filaments to form bundles supporting membrane projections termed “ruffles” [8] , [9] , [10] , [11] . This leads to the formation of large membrane protrusions , which form a pocket enclosing the bacteria and facilitating entry [12] . Such ruffles appear similar to those described for macropinocytosis , a classical non-selective cellular uptake mechanism of molecules into large , irregular shaped vesicles , termed macropinosomes , formed by the collapse and fusion of ruffles with the plasma membrane [13] . In this classic model [2] , entry and macropinosome formation represent a single process . However , evidence exists that bacterial entry and membrane ruffling are associated with different bacterial effectors and host responses during S . flexneri invasion: for example , host vinculin is recruited specifically to entering bacteria [14] , Rho-GTPase isoforms are recruited differentially to either entering bacteria or membrane ruffles [8] , and the bacterial effector IpgD was shown to regulate ruffling morphology and is not required for bacterial entry [15] , [16] . Entry has been proposed to occur initially via effector mediated contact of S . flexneri [17] to cholesterol rich lipid raft membrane domains [18] and to be mediated by specific receptors [19] , [20] suggesting entry is akin to receptor mediated phagocytosis . In the case of Salmonella enterica , an invasive , T3SS-employing pathogen which shares many common aspects with S . flexneri entry into host cells , it was hypothesized that Salmonella containing vacuole and macropinosomes may be distinct , as they are sorted into different intracellular routes [21] . Thus , it appears that the classic model for S . flexneri invasion may be too simplistic , and a revised model could include two parallel processes: ( i ) bacterial entry and ( ii ) membrane ruffling , whose precise biological role in invasion has not been studied in detail . Upon entry , S . flexneri reside within bacteria containing vacuoles ( BCVs ) , followed by BCV rupture and escape into the cytosol within 10 minutes , a step crucial for the intracellular growth of S . flexneri [22] . Initially , it was thought that bacterial effectors , such as the translocator proteins IpaB and IpaC , induce S . flexneri vacuolar rupture [23] . For instance , IpaB was shown in vitro to oligomerize and insert into the plasma membrane of target cells , forming cation selective ion channels involved in vacuolar rupture [24] . In contrast , we recently showed that S . flexneri subverts host cell pathways for BCV rupture [25] . A siRNA screen aimed at identifying host proteins involved in vacuolar rupture yielded multiple hits related to membrane trafficking , including EEA1 , SNX1 , VAMP2 , and the small Rab GTPases Rab4 , Rab5 and Rab11 . Rab5 and Rab11 were recruited to the invasion site , with Rab11 localized to a large area surrounding invading bacteria , but not to the forming BCV . Rab11 knockdown caused a strong delay in vacuolar rupture timing , providing a functional link between membrane trafficking at the invasion site and vacuolar rupture . Additionally , the bacterial effector IpgD , a PI ( 4 , 5 ) P2 phosphatase , was shown to regulate Rab11 recruitment and to be required for efficient vacuolar rupture [25] . How this subversion of host cell pathways by S . flexneri relates to bacterial entry and vacuolar rupture has remained unclear . In this work , we applied dynamic imaging and advanced large volume correlative light electron microscopy ( CLEM ) in the form of focused ion beam/ scanning electron tomography ( C-FIB/SET ) to reveal key features of the pathogenic strategy employed by S . flexneri . We analyze the architecture of the S . flexneri invasion site in detail and reveal that it contains two distinct compartments , the BCV and macropinosomes . We demonstrate that Rab11 is recruited directly to macropinosomes and its activity is required for efficient vacuolar rupture . Finally , we reveal that the BCV and macropinosomes come into direct contact at the onset of vacuolar rupture . Our results represent a major step forward in understanding the mechanism of S . flexneri escape to the cytosol , and provide the basis for an updated model of S . flexneri invasion into epithelial cells .
Macropinosomes were previously observed at the invasion site of Salmonella enterica in epithelial cells using phase contrast microscopy and fluorescent dextran added to the infection media , acting as a fluid phase marker [21] . We set out to examine macropinosome formation at the S . flexneri invasion site by measuring internalization of fixable dextran conjugated to Alexa Fluor-647 ( Fig 1A ) . HeLa , Caco-2 and NRK cells were infected for 30 minutes with wild-type dsRed expressing bacteria , added together with the fluorescent conjugate . This was followed by washes to remove non-internalized dextran , fixation and staining for actin to visualize the site of invasion , and DNA to visualize cell nuclei . Multiple dextran containing vesicles were observed surrounding S . flexneri at the invasion site to a similar extent in all host cell types used . The size distribution of the vesicles was quantified in HeLa cells using automated software . Vesicles are heterogeneous in size , with the majority of vesicles having a diameter less than 1μm and a fraction of larger vesicles observed . As S . flexneri is generally thought to enter cells within a pocket formed by large membrane protrusions [12] , we expected the BCV to also contain detectable fluid phase marker . Surprisingly , bacteria were never observed within dextran positive vesicles , implying S . flexneri containing vacuoles exclude most extracellular fluid . The observed vesicle size heterogeneity and uptake of extracellular fluid are reminiscent of features classically associated with macropinosomes [26] , suggesting S . flexneri is surrounded by macropinosome-like vesicles ( here on referred to simply as ‘macropinosomes’ , see discussion ) at the invasion site but does not reside within them . As the bacterial effector IpgD has been previously implicated in regulation of ruffling morphology during S . flexneri invasion [15] , [27] , we examined its role in macropinosome formation ( Fig 1B and 1C ) . Confocal microscopy of HeLa cells infected by wild-type and ΔipgD strains in the presence of dextran Alexa Fluor-647 for 30 minutes revealed a reduction in the number of macropinosomes surrounding bacteria when using the mutant strain ( Fig 1B ) . Automated quantification of the number of macropinosomes formed per bacterium revealed that macropinosome formation was reduced by around 60% when using the IpgD mutant ( Fig 1C ) . Infection with ΔipgD/IpgD strain showed partial complementation of vesicle formation , in accordance with previously reported results [25] ( S1 Fig ) . We conclude that macropinosome formation at the S . flexneri invasion site is regulated by the bacterial effector IpgD . Intracellular bacterial pathogens are normally thought to subvert pre-existing host endocytic vesicles during invasion ( see introduction ) , however our study indicated that nascent macropinosomes appear to be a major endomembrane component at the invasion site of S . flexneri . We examined the relative contribution of pre-existing endocytic vesicles to the endomembrane content at the invasion site ( Fig 1D ) . To this end , we performed sequential labeling experiments , pre-loading cells with dextran Alexa Fluor-488 for two to three hours prior to infection to allow uptake by the host endocytic compartment . This was followed by extensive washes and infection in the presence of dextran Alexa Fluor-647 . Extensive labeling of punctate dextran Alexa Fluor-488 endomembrane structures was observed throughout the cells . Strikingly , we could not detect accumulation of dextran Alexa Fluor-488 containing vesicles around invading bacteria , while dextran Alexa Fluor-647 vesicles were highly enriched around bacteria . Furthermore , we did not detect any co-localization of the two differently colored dextrans around the entering bacteria at the measured time-point . Quantitative image analysis revealed 92% of total vesicle volume ( i . e . combined volume of vesicles labeled by either dextran Alexa Fluor-488 or dextran Alexa Fluor-647 ) at the invasion site is occupied by dextran Alexa Fluor-647 containing vesicles . This indicated that the local environment around invading S . flexneri is occupied by newly formed macropinosomes , without significant recruitment of pre-existing host endocytic vesicles , arguing against active subversion of pre-existing host endocytic pathways by S . flexneri during invasion . Since fine structural details of the S . flexneri invasion site are obscured by the resolution limit of standard light microscopy , we applied CLEM in the form of correlative large volume focused ion beam/scanning electron tomography ( C-FIB/SET ) to examine the S . flexneri invasion site in detail [28] ( Fig 2 ) . This emerging technique combines 3D fluorescent and large volume electron tomography into a single correlated data set , providing three features ideal for investigation of S . flexneri invasion: First , discreet and highly transient ( in the order of minutes ) stages of invasion can be targeted using fluorescent microscopy with stage-specific fluorescent markers prior to tomography acquisition . Thus , precise access to early invasion ( before BCV rupture ) and to the vacuolar rupture event itself can be gained . Secondly , the large cellular volumes ( in the order of 1000 μm3 ) acquired using this technique allow visualization of entire invasion sites , accounting for multiple bacteria , vesicles and other cellular structures present at the site . The BCV’s and other vesicles’ membranal integrity and their connectivity to other compartments can be examined in 3D from all axes , providing structural information not easily accessible by classic serial sectioning and tomography approaches [29] . Finally , the fluorescent labeling correlated in 3D to the tomography data ( with bacteria observed by light and electron microscopy acting as alignment fiducials ) provides molecular specificity to features observed within the tomography volume , albeit within the light microscopy diffraction limit ( see materials and methods ) . Overall , 15 C-FIB/SET data sets of invasion sites were acquired for this study . As fluorescence microscopy of the S . flexneri invasion site revealed that BCVs exclude detectable fluorescent dextran and are surrounded by dextran containing vesicles in all our experiments , we first set out to examine the architecture of the early invasion site in detail using C-FIB/SET . A detailed description of the correlative workflow for the data set analyzed in Fig 2 is presented ( Fig 2A , S1 Movie ) : First , a S . flexneri invasion site was imaged by confocal microscopy . As a tight actin enrichment around bacteria was previously shown to appear exclusively prior to vacuolar rupture [30] , a site containing bacteria with this feature ( as indicated by phalloidin staining ) was chosen , thereby providing a clear indication of early invasion ( Fig 2A , inset ) . After sample processing for electron microscopy , a FIB/SET data set was acquired at the exact same location . The two data sets were then combined into a single correlated volume , providing a full 3D view of the entire S . flexneri invasion site from all axes . Membrane ruffles are observed at the cell surface while BCVs and smaller vesicles are seen in the cytosol . We examined the structure of the BCV in early invasion in detail and delineated the typical BCV structural motifs ( Fig 2B , S2 Movie ) . The data in Fig 2B is representative of all datasets acquired during early invasion in our study . We found that the bacterial cytosol and membrane are surrounded by a layer of low electron density . Based on annotation from previous ultrastructural studies we associate this layer with bacterial lipopolysaccharides ( LPS ) ( e . g . [31] ) . Furthermore , this layer is also present in rupturing BCVs missing parts of their BCV membrane ( see below ) , indicating it does not represent the vacuole lumen . These bacterial structures are tightly enclosed by the vacuole membrane . An additional electron dense layer observed only around bacteria surrounded by phalloidin stain ( as identified by fluorescence microscopy ) is consequently identified as actin . Finally , we observe multiple small vesicles surrounding the BCV , but not contacting it nor each other . In all data sets observed , BCVs contained a single bacterium ( often in division ) , were structurally uniform and tightly surrounding the bacterium inside . We were not able to detect a luminal BCV space around the bacteria in any of our data sets . Overall , our high resolution correlated view of the early S . flexneri invasion site reveals the BCV is a tight , uniform compartment , structurally distinct from the surrounding heterogeneous macropinosomes . We conclude that the BCV and macropinosomes present at the early invasion site represent two distinct compartments . Our two color dextran labeling experiments ( see Fig 1D ) , indicated that the local environment around invading S . flexneri is occupied by macropinosomes formed in situ during invasion without significant recruitment of pre-existing host endocytic vesicles . We applied C-FIB/SET to correlate the total vesicle population at the invasion site with dextran Alexa Fluor-647 labeling ( added during infection ) ( S2 Fig ) . FIB/SET analysis reveals the entire vesicle population around invading S . flexneri irrespective of the fluorescent probe , while the dextran Alexa Fluor-647 label provides an indication of the volume occupied by vesicles formed during invasion . In the two datasets presented ( S2A and S2B Fig ) , 95% and 99% of vesicles found at the invasion site reside within the fluorescent dextran label ( S2C Fig ) . In data set B , five of the largest vesicles , found near the surface of the cell , lie outside of the dextran labeling , most likely representing late forming macropinosomes formed after removal of dextran in the washing phase . This result confirms that newly formed macropinosomes are the major compartment at the S . flexneri invasion site , in agreement with our two color dextran labeling experiments ( Fig 1D ) . We have recently shown that the bacterial effector IpgD is required for efficient vacuolar rupture [25] . As IpgD is also a regulator of macropinosome formation ( see Fig 1B and 1C ) , we hypothesized these two processes may be functionally related . In order to test this hypothesis , we performed two bacterial effector mutant library screens aimed at identifying the relationship between the number of macropinosomes at the invasion site and vacuolar rupture timing ( Fig 3A and 3B , selected hits are presented . For screen details see materials and methods . Full screen results are provided in S3A Fig ) . First , the library was screened for the number of macropinosomes formed at the invasion site using quantitative image analysis of fluorescent dextran containing vesicles ( Fig 3A ) . Various effectors alongside IpgD were found to affect macropinosome formation , including IpgB1 , IpgE ( a chaperone for IpgD [15] ) and IpgB2 . Secondly , the library was screened for vacuolar rupture timing using dynamic microscopy ( Fig 3B ) . We found an inverse correlation between the number of macropinosomes formed at the invasion site and the delay in vacuolar rupture timing , i . e . the fewer macropinosomes found at the invasion site , the longer rupture is delayed . Of particular interest is the effector IpgB1 , the strongest hit in both screens . IpgB1 was shown to be involved in the early stages of S . flexneri invasion , is associated with membrane ruffles , and can induce membrane ruffling via stimulation of Rac1 and Cdc42 activities . Like IpgD , it has no known function in disrupting membrane integrity or pore formation to our knowledge [32] . Thus , our mutant screens results revealed that vacuolar rupture efficiency is correlated to macropinosome availability . We also studied the effect of drugs known to inhibit macropinocytosis and actin dynamics on macropinosome formation and vacuolar rupture timing . Amiloride , a Na ( + ) /H ( + ) exchange inhibitor known to cause indirect inhibition of macropinocytosis [33] did not inhibit macropinosome formation and entry at low doses and was cytotoxic at higher doses . As we have previously identified via high-content siRNA screening that Arp2/3 subunits are among host proteins involved in S . flexneri uptake and vacuolar rupture [25] we tested CK-666 , an Arp2/3 complex inhibitor that does not stimulate dissociation of preformed actin branches in vitro [34] . CK-666 inhibited macropinosome formation without inhibiting bacterial entry , and significantly delayed vacuolar rupture timing ( S3B Fig ) . The selective inhibition of macropinosome formation by CK-666 may be due to the drug having a stronger effect on ruffle formation , a process requiring massive actin re-arrangement , as opposed to the more small scale actin re-arrangement required for bacterial uptake . Overall , our results demonstrate that vacuolar rupture timing is dependent on the availability of macropinosomes , indicating that membrane ruffling and macropinosome formation may be functionally linked to vacuolar rupture . We therefore set out to examine the cell biology underlying the relation between macropinosomes and vacuolar rupture in detail using dynamic microscopy , functional assays and structural analysis . First , we examined the spatial and temporal proximity of macropinosomes to the rupturing BCV . We performed live microscopy of S . flexneri invasion in cells transfected with a marker of phosphatidylinositol 3-phosphate ( PI3P ) , 2XFYVE-GFP , used for labeling macropinosomes , and galectin-3-mOrange used for labeling vacuolar rupture . PI3P has been previously reported to be associated with early macropinosome maturation , in particular with the cup closure step [35] . We found that 2XFYVE-GFP was partially co-localized with dextran positive vesicles at the invasion site , this partial co-localization likely due to the transient nature of its recruitment to forming macropinosomes [35] ( S3C Fig ) . Galectin-3-mOrange is commonly used as a vacuolar rupture marker as it specifically labels the BCV only after loss of vacuole integrity [36] . 3D multi-channel images were acquired every 30 seconds ( Fig 3C and 3D , S3 Movie ) and the resulting movies were quantified by counting the number of events where macropinosomes were found bordering ( within 1μm ) the rupturing BCV , as indicated by the first appearance of a galectin-3-mOrange signal ( Fig 3E left ) . In 73% of all vacuolar rupture events analyzed ( n = 30 events in three independent experiments ) , macropinosomes were found bordering the BCV , with 50% of events containing more than one bordering vesicle . This number most likely underrepresents such events as not all macropinosomes are labeled with 2XFYVE-GFP . In order to assess whether macropinosome proximity to the BCV occurs upstream of rupture or is a product of rupture , we performed high temporal resolution imaging with multi-channel confocal stacks acquired every 5 seconds ( S3D Fig , S4 Movie ) . This allowed clear detection of bordering macropinosomes at the onset of the galection-3-mOrange signal appearance and therefore at the onset of vacuolar rupture . Data quantification ( Fig 3E , right ) showed that in 92% of the subset of vacuolar rupture events containing bordering macropinosomes ( n = 36 events in three independent experiments ) , bordering occurred at least one frame before the onset of rupture , indicating this proximity is present upstream of vacuolar rupture . In all movies acquired , rupturing BCV’s ( i . e . labeled with galectin-3-mOrange ) were never co-labeled with 2XFYVE-GFP , and 2XFYVE-GFP positive vesicles were never co-labeled with galection-3-mOrange , implying macropinosome proximity to the rupturing BCV does not result in membrane fusion or ruptured macropinosomes ( see discussion ) . Rab-GTPases act as markers for macropinosome formation and maturation [13] , [37] and have been previously shown to be recruited to the S . flexneri invasion site to various degrees [25] . We examined the recruitment of Rabs to the invasion site in the context of dextran labeled macropinosomes and invading bacteria ( S4A Fig ) . Cells transfected with Rab4-GFP , Rab5-GFP , Rab7-GFP and Rab11-GFP were infected with wild-type fluorescent bacteria for 30 minutes in the presence of dextran Alexa Fluor-647 . Rab4-GFP was not recruited to the invasion site , in agreement with previous studies [25] , while other Rabs examined were recruited to macropinosomes at the invasion site to various degrees but not around invading bacteria . As Rab11’s association with macropinosomes has not been previously demonstrated in other systems to our knowledge , and its recruitment has been previously shown to be prevalent at the invasion site and is required for efficient vacuolar rupture [25] , we decided to examine its association with macropinosomes in detail . While Rab11 is classically associated with recycling endosomes [38] , [39] , our results suggested that the major compartment at the invasion site is composed of macropinosomes and that host endosomes are not recruited during invasion ( Fig 1D , S2 Fig ) . Furthermore , our fixed invasion experiments ( S4A Fig ) revealed some macropinosomes directly surrounded by Rab11-GFP . Taken together , these results suggest that during invasion Rab11 can associate directly with newly formed macropinosomes via a non-classical pathway that does not involve recycling endosomes . In order to test this hypothesis we performed dextran pulse chase experiments combined with live imaging of cells transfected with Rab11-GFP and galectin-3-mOrange ( Fig 4A–4C , S5 Movie ) . Cells were infected with wild type S . flexneri in the presence of dextran Alexa Fluor 647 for 22 minutes . At that point the cells were quickly washed to remove non-internalized dextran from the media and live imaging was initiated . Cells were imaged in 3D at 60s intervals . The time point for dextran washes ( t = 0 ) was chosen so that invasion sites already containing internalized dextran but prior to vacuolar rupture ( as indicated by galectin-3-mOrange ) could be captured . Thus the relation between dextran containing macropinosomes and Rab11-GFP recruited to the invasion site could be examined during live imaging of early invasion . A representative experiment is presented . We found that Rab11 is directly and dynamically recruited to macropinosomes prior and during vacuolar rupture , conforming the non-classical association between macropinosomes and Rab11 at the S . flexneri invasion site . As expected , live imaging also revealed that rupture events occur in close proximity to macropinosomes ( in agreement with Fig 3C–3E ) . Given the non-classical association between macropinosomes and Rab11 occurring during S . flexneri invasion , we examined the functional role of Rab11 in relation to macropinosome formation and vacuolar rupture . We employed a functionally impaired Rab11: Rab11S25N-GFP , a GDP-locked dominant negative ( Rab11 DN ) [40] . We found that it did not inhibit macropinosome formation or bacterial entry , and unlike Rab11-GFP , was not recruited to macropinosomes ( Fig 4D left , S4B and S4C Fig ) . Next we examined the functional role of Rab11 in vacuolar rupture . We transfected cells with Rab11-GFP or Rab11 DN and galectin-3-mOrange and performed live imaging of S . flexneri invasion to determine vacuolar rupture timing ( Fig 4D right , S6 Movie , S7 Movie showing representative experiments ) . We found that vacuolar rupture is significantly delayed when using the Rab11 DN . Our results indicate that while Rab11 is not required for macropinosome formation or bacterial entry , its activity is required for recruitment to macropinosomes and efficient vacuolar rupture . We hypothesize that Rab11’s role in vacuolar rupture is manifested through its well described function in vesicle trafficking regulation [39] acting uniquely on macropinosomes in the case of S . flexneri invasion ( see discussion ) . Overall our effector mutant screens , live imaging and functional studies reveal that macropinosomes at the invasion site are the target for direct recruitment of Rabs , they border the BCVs at the onset of vacuolar rupture , and their formation and trafficking are required for efficient vacuolar rupture . Together , these results provide for the first time evidence that vacuolar rupture involves two cellular compartments , the BCV and macropinosomes . We hypothesized that macropinosomes are directly implicated in vacuolar rupture via physical contacts with the BCV . Such contacts are obscured when using fluorescence microscopy due to the limited resolution of light microscopy . We therefore applied C-FIB/SET to image the very short-lived and highly dynamic rupturing event in detail , acquiring correlative data sets of invasion sites containing BCVs labeled with galectin-3-mOrange to indicate rupture ( Fig 5 ) . C-FIB/SET is particularly suited for the unambiguous identification of compartmental contact points at the invasion site , as it provides information in all three axes and within a large cellular volume likely to contain multiple contacts occurring at various orientations . Strikingly , in cells infected with the wild-type strain ( Fig 5A , S8 Movie ) rupturing BCVs were found in direct contact with multiple surrounding vesicles . A partly dissociated BCV membrane emanating into the cytosol from the contact point between a macropinosome and a BCV was observed , with most of it still attached to the bacterium ( Fig 5A , white arrows , yellow ) . Smaller intraluminal vesicles within the large macropinosome were commonly observed at the interface between the macropinosome and the BCV ( Fig 5A , black arrowhead ) . As IpgD is required for efficient rupture [25] , we examined its involvement in the formation of macropinosome-BCV contact points . Infection with a ΔipgD strain revealed a reduction in the number of vesicles in contact with the BCV , yet macropinosome-BCV contact points morphologically identical to the wild-type were still always observed ( Fig 5B , S9 Movie ) , indicating that IpgD regulates macropinosome availability but does not impact the formation of macropinosome-BCV contacts during rupture . In the rupture events presented here , most of the BCV membrane is still present ( Fig 5B , yellow ) , and rupture is observed at the opposite side of the bacterium in relation to the contact points ( Fig 5B , black arrows ) . We conclude that the BCV and surrounding macropinosomes come into direct contact during vacuolar rupture . Thus , our functional , dynamic and structural analysis ( Figs 3 , 4 and 5 ) suggest that macropinosomes are required for efficient vacuolar rupture and that macropinosome-BCV contacts are involved in this process .
We report here that macropinosomes are a central compartment in S . flexneri invasion and that they are implicated in S . flexneri vacuolar rupture . Our results lay the ground for a new model of the early steps of S . flexneri invasion into epithelial cells ( Fig 6A ) : Invasion begins with two distinct processes occurring in conjunction: bacterial entry into a tight BCV and the regulated formation of macropinosomes via membrane ruffling . Then , host Rab GTPases are recruited directly to nascent macropinosomes , followed by direct contact between macropinosomes and the BCV at the onset vacuolar rupture . Host endomembranes are not recruited to the invasion site , indicating invasion occurs without subversion of pre-existing host endocytic pathways or pre-existing endomembrane compartments . Macropinosomes are revealed to be key players in the process of vacuolar rupture , as indicated by functional experiments demonstrating that the efficiency of vacuolar rupture is dependent on macropinosome availability and Rab11 activity ( acting in direct association with macropinosomes ) . This new model provides a conceptual framework that accommodates previous studies [8] , [14] , [15] , [16] , [25] , and has predictive power in regard to the possible biological role of membrane ruffling and macropinosomes in other pathogenic systems [11] , [21] . For example , various observations made regarding the bacterial effector IpgD are put into biological context with our work: While IpgD is not required for bacterial entry [15] , it is known to promote ruffling [15] , is required for macropinosome formation ( Fig 1B and 1C ) , Rab11 recruitment to the invasion site and efficient vacuolar rupture [25] , [27] . These activities are now placed in sequence , as ruffling , macropinosome formation , Rab GTPase recruitment to macropinosomes and vacuolar rupture represent a cascade of causally connected events occurring during S . flexneri invasion ( Fig 6B ) . Vesicles induced by S . flexneri share some similarities with classically described macropinosomes formed in non-pathogenic systems [13] , [41] . Their formation in the context of membrane ruffling , size heterogeneity and uptake of fluid phase marker resemble that of macropinosomes ( and micropinosomes , accounting for smaller vesicles formed at the invasion site ) . However , their formation is induced via the action of bacterial effectors that modulate other cellular pathways as well , and thus may affect vesicle biochemistry . Indeed , our results demonstrate that S . flexneri induced macropinosomes are associated with Rab11 , a typical marker for the recycling endosomal pathway [38] , [39] , not known to label macropinosomes in other systems . Macropinosome trafficking in the absence of pathogens is cell type dependent , and while professional phagocytizing cells direct macropinosomes into the lysosomal pathway [42] , in some non-professional phagocytizing cells , macropinosomes were reported to recycle to the plasma membrane , with little or no interaction with endosomal vesicles [41] , [43] , [44] . It may be that the macropinosome-BCV contacts are a product of “misdirected” trafficking towards the BCV that we like to term “internal recycling” , instead of recycling to the plasma membrane , as both compartments present similar membranal features . Intracellular pathogens are generally thought to engage and subvert existing housekeeping endomembrane system pathways , such as the lysosomal or recycling pathway [38] . This pathogenic strategy is a hallmark of Salmonella invasion into epithelial cells , with many studies demonstrating complicated subversion patterns induced by Salmonella during invasion [4] , [45] . In stark contrast , we report here a pathogenic strategy that does not exploit a pre-existing host pathway . Instead , S . flexneri induces the formation and trafficking of macropinosomes formed in situ in a “local pathogenic pathway” that involves the recruitment of Rab GTPases to macropinosomes , and does not involve recruitment of host endomembranes . Thus at each S . flexneri invasion site a local trafficking pathway is initiated without it being normally present in the cell . This pathway is subsequently exploited by the pathogen for vacuolar escape . The strategy employed by S . flexneri represents a divergence from existing paradigms of pathogen- host interactions centered on the successive pathogen subversion of established trafficking pathways [3] . In resemblance to other pathogenic systems , such as viral infection [46] , macropinosomes are revealed to be a key constituent of the invasion process . The establishment of a new model for S . flexneri invasion presented here hinged on the utilization of the emerging technique of C-FIB/SET . First , C-FIB/SET was used to demonstrate that S . flexneri enters cells within a tight BCV ( Fig 2 ) , likely explaining the lack of detectable fluorescent dextran labeling around bacteria when studied by light microscopy ( Fig 1 ) . This result strongly supports a model where bacterial entry occurs via a phagocytosis- like mechanism leading to tightly enveloped bacteria , and not , as is classically suggested ( see introduction ) , via a macropinocytosis- like entry process—which would result in spacious , heterogeneous BCVs [2] ( see Fig 6 ) . In this “two parallel pathways” invasion model , the biological function of membrane ruffling and the resulting macropinosomes is related to the downstream step of vacuolar rupture , and not , as previously thought , to bacterial entry . Second , by specifically targeting rupturing vacuoles ( as indicated by galectin-3-mOrange fluorescent labeling ) for FIB/SET acquisition , this enigmatic and transient event became reliably and reproducibly accessible to 3D ultrastructural investigation for the first time , revealing macropinosome-BCV contacts and emanating dissociated membranes . The biological observations obtained through the application of C-FIB/SET in our study not only led to the emergence of new biological concepts regarding S . flexneri invasion , but act to demonstrate the power of this technique to investigate complex biological arenas containing multiple transient or rare events , complex 3D cellular organization and intricate structural interfaces . We reveal a functional relationship between macropinosomes and the process of vacuolar rupture , mediated by the activity of Rab11 . Macropinosome formation and trafficking ultimately result in macropinosome—BCV contact points formed at the onset of vacuolar rupture . How these contact points may drive membrane destabilization and eventual vacuolar rupture remains an open question . Our results lay the ground for future in depth investigations of the molecular mechanism and underlying biophysics driving this process , however initial insights into the mechanism of rupture can be gained from several observations: Firstly , in both fixed and dynamic light microscopy experiments , macropinosomes were never labeled with galectin-3-mOrange during vacuolar rupture . As BCV membranes become coated with galectin-3 at the loss of vacuolar integrity [36] , fused or ruptured macropinosomes ( i . e . with inner membranes exposed to the BCV or cytoplasm ) would be expected to be labeled with galectin-3 as well . Furthermore , rupturing BCVs were never labeled with fluorescent dextran ( fixed and pulse-chase experiments ) or 2XFYVE-GFP ( live microscopy; even during time-lapse studies with very high temporal resolution ) , as would be expected had dextran flowed from fusing macropinosomes into the collapsing BCV during rupture or if lipids were exchanged . These two observations argue against macropinosome -BCV fusion or macropinosome rupture as the underlying process facilitating BCV rupture , suggesting an altogether different mechanism is in place . Secondly , IpgD regulates macropinosome formation as well as rupture timing [25] . Infections with the ΔipgD strain exhibit overall fewer macropinosomes in contact with rupturing BCVs , while the typical macropinosome-BCV contact point is still present in every rupturing BCV examined ( Fig 4B ) . A decrease in the amount of contacts may limit damage to the BCV , explaining the observed delay in vacuolar rupture when using this strain and hinting at a cumulative damaging effect of macropinosome-BCV contacts . A decrease in macropinosome trafficking towards the BCV when using Rab11 KD or Rab11 DN would result in a similar effect . One possible mechanism could involve the formation of intraluminal vesicles ( ILVs ) pinched off from the BCV membrane and into maturing macropinosomes , causing increased membrane tension and eventually vacuolar rupture . Such vesicles are often observed at the interface between the BCV and connected macropinosome ( Fig 5A , black arrowhead ) . The molecular machinery driving the formation of ILVs has been described in endosomes and phagosomes , but not in macropinosomes thus far to our knowledge [41] . Finally , the involvement of multiple bacterial effectors in vacuolar rupture ( Fig 3A and 3B ) , without any single effector mutant inducing a complete arrest of rupture , implies that vacuolar rupture is not dependent on a single bacterial effector as described in other systems [47] . While other invasive bacterial pathogens employ direct lysis of their surrounding vacuole via bacterial proteins inserted directly into the membrane [48] , [49] , S . flexneri vacuolar rupture is most likely a complex process that involves multiple bacterial and host molecules acting to facilitate macropinosome formation , trafficking , contacts with the BCV and membrane destabilization . Future studies using imaging , biochemical and biophysical approaches will allow further elucidation of this critical step in the growth of S . flexneri . Macropinosomes have been identified at the invasion site of intra-cellular bacterial pathogens years ago in seminal works by S . Falkow and others [11] , [21] , [50] . However , since then , their precise role in intra-cellular bacterial invasion has remained poorly understood and their importance overlooked . We report here a central role for macropinosomes in the pathogenic strategy of S . flexneri , and predict that this compartment is exploited by other invasive bacteria as well , possessing various biological functions in the context of pathogenicity yet to be discovered .
The following S . flexneri strains were used: M90T AfaI [22] , expressing the adhesin afaI ( Figs 1B , 2 , 3C–3E and 4A–4C ) , the invasive ΔipgD AfaI [30] ( Figs 1B and 5B ) , M90T ( Figs 1C and 4D ) , ΔipgD ( 1C ) and M90T expressing dsRed ( Fig 1A and 1D ) . For the two mutant library screens ( Fig 3A and 3B ) all S . flexneri strains used were kindly provided by JR . Rohde ( Dalhousie University ) : the strains do not express afaI and are a part of a pwR100 collection [51] . The parent strain for this collection is a streptomycin-resistant strain of S . flexneri serotype 5a ( M90T-Sm ) . Growth medium was supplemented with ampicillin ( 50 μg/ml ) for all strains except for ΔipgD supplemented with tetracycline ( 5 μg/ml ) . All bacterial strains were cultured on tryptic casein soy broth ( TCSB ) agar plus 0 . 01% Congo red with 20mg/mL agar at 37°C . Human epithelial HeLa cells ( clone CCL-2 from ATCC ) and NRK fibroblasts ( Clone CRL- 1570 from ATCC ) were cultured in Dulbecco’s modified Eagle’s medium ( DMEM ) supplemented with 10% ( vol/vol ) fetal bovine serum ( FBS ) at 37°C , 5% CO2 . Caco-2 TC7 were kindly provided by P . Sansonetti ( Institut Pasteur ) and cultured as above except with 10% CO2 . ( Fig 1A , HeLa cells were used for all other experiments ) . For invasion experiments , overnight bacterial cultures were inoculated at a 1/100 dilution in TCSB with the appropriate antibiotic if required and grown to an optical density of ≈ 0 . 4 at 600 nm ( OD600 ) . Prior to infection , bacteria were washed with PBS and resuspended in EM buffer ( 120 mM NaCl , 7 mM KCl , 1 . 8 mM CaCl2 , 0 . 8 mM MgCl2 , 5 mM glucose , 25 mM HEPES , pH 7 . 3 ) and incubated with poly-L-lysine for 15 min if they did not express AfaI . For imaging of fixed samples , bacteria were added to cells at MOI 30 ( Figs 1A , 1C and 3A ) or MOI 20 for all other fixed experiments and allowed to adhere for 10 minutes at RT . Samples where then incubated at 37°C for 30 minutes , washed three times in PBS and fixed with cold PFA 4% followed by staining with DAPI and Phalloidin . Dextran Alexa Fluor-488 and Alexa Fluor-647 , 10 , 000 MW fixable ( Life Technologies ) were added to EM buffer and placed with cells for two to three hours before infection for sequential labeling assay ( Fig 1D ) or added during infection at a final concentration of 0 . 5μg/ml ( all other experiments ) . For rupture timing screen ( Fig 3B ) and live imaging ( Figs 3C–3E and 4A–4C ) bacteria were added at a MOI of 50 . HeLa cells were seeded into 96-well glass bottom plates ( Greiner ) at a density of 7000 cells per well 24 h prior to transfection . The cells were then transfected as described [25] with Rab11-GFP ( Fig 4A–4C ) , 2XFYVE-GFP ( Fig 3C–3E ) , galectin-3-mOrange ( Figs 3C–3E , 4 and 5 ) or Rab11S25N-GFP , a Rab11 GDP-locked dominant negative ( DN ) ( Fig 4D , kindly provided by B . Goud ( Institut Curie ) ) using X-tremeGENE 9 reagent ( Roche ) for 24–48 h , according to the manufacturer’s instructions . For mutant library rupture timing screen ( Fig 3B ) cells were transfected with Actin-mOrange [30] and galectin-3-GFP [36] in an identical manner . Fixed samples were imaged using a Perkin Elmer UltraView spinning disc confocal microscope , with a 60X/ 1 . 2NA water objective and a Z step size of 0 . 3 μm . For time lapse microscopy , cells were imaged with a confocal microscope ( for experiments in Figs 3C–3E and 4A–4C ) at 37°C with a 40X / 1 . 3 NA oil objective . Every 60 , 30 or 5 seconds depending on experiment , a stack of 6 z-planes ( step size of 0 . 5μm ) was acquired sequentially in two or three channels using a 488nm , 561nm and 640nm laser depending on experiment . For mutant library rupture timing screen ( Fig 3B ) and Rab11 DN experiments ( Fig 4D ) cells were imaged at 37°C using a Nikon Ti-E wide-field microscope with a 20X air objective . Imaging was performed with excitation at 465 to 500 nm and 540 to 565 nm . Images were acquired every 90 or 120 seconds ( depending on the exposure time ) for 1 to 2 hours . In each screen experiment two mutants and one control ( wild-type strain ) were screened and images of 10 positions per strain acquired . For the Rab11 DN experiments 8 positions per condition were acquired with WT and IpgD as controls . Sample preparation and acquisition were performed as previously described [25] . In short , HeLa cells were cultured on MatTek dishes with finder grid ( MatTek Corporations ) . Infections were performed as described above . Samples were fixed with 0 . 1% Glutaraldehyde and 4% paraformaldehyde for 30 minutes at room temperature . After high resolution confocal microscopy imaging ( using 60X objective ) positions of interest were marked using phase contrast and fluorescence with 10X and 20X objectives . Samples were then fixed overnight with 2 . 5% Glutaraldehyde in 0 . 1 M cacodylate buffer followed by fixation in 2 . 5% glutaraldehyde + 0 . 4% Tannic Acid pH 7 . 2 in 0 . 1 M cacodylate buffer for 30 min at RT . Samples were stained with 1% OsO4 in DDW for 30 min at 4°C . Samples were dehydrated in graded ethanol series and embedded in Epon followed by FIB/SET performed in a Helios Nanolab Dual beam ( FEI ) at the electron microscopy unit at the Weizmann Institute of Science ( Israel ) . XY pixel size range was 6 . 5–8 . 3 nm and slice thickness 10 nm . Data was aligned using ImageJ ( http://imagej . nih . gov/ij/ ) . Amira and Avizo ( FEI ) were used for 3D visualization , data correlation , manual segmentation and supporting movies . Internalized bacteria were used as correlating fiducials . Inverted contrast is presented . Overall 15 C-FIB/SEM datasets of S . flexneri invasion sites were acquired , 11 of WT strain and 4 of ΔipgD strain . Macropinosome size distribution ( Fig 1A ) was quantified using ICY ( http://icy . bioimageanalysis . org/ ) . “Spot detector” plugin was applied to segment dextran containing vesicles in 3D confocal stacks . A spherical assumption was used to extract vesicle diameter . 2300 vesicles from three separate experiments were analyzed . WT vs . ΔipgD strain vesicles/bacterium ( Fig 1C ) was quantified using CellProfiler ( http://www . cellprofiler . org/ ) . Macropinosomes were counted only when found in actin labeled ruffles that contained at least one bacterium . Quantification is based on six separate experiments including in total 1186 cells , 1877 bacteria and 4330 vesicles . Unpaired t test was used for significance . Vesicle volume distribution ( Fig 1D ) was performed using a custom ICY protocol . 34 invasion sites in two experiments were chosen for analysis . Shigella-dsRed , dextran Alexa Fluor-488 and Alexa Fluor-647 signals were segmented in 3D . The sum of volumes of all vesicles of each type within 30 voxels of bacteria in each invasion site was calculated and normalized to the total vesicle volume at the site . Overall 476 vesicles were used in analysis . The mutant library macropinosome formation screen ( Fig 3A ) was quantified using CellProfiler in an identical manner to the WT vs . ΔipgD strain vesicles formation assay described above . Results were obtained from two independent experiments , each containing at least 30 invasion sites , with 907 invasion sites analyzed in total . For mutant library rupture timing screen and Rab11 DN experiments ( Figs 3B and 4D ) , time lapse series were visually analyzed with Fiji ( http://fiji . sc ) . For rupture timing screen three independent experiments per strain were performed , at least 50 invasion events per strain were measured , with 791 events in total . For Rab11 DN experiments three independent experiments per condition were performed , with 946 rupture events analyzed in total ( WT 351 , IpgD 193 , DN 402 ) . Vacuolar rupture timing was measured as the time interval between the beginning of ruffle formation and the appearance of a Galectin-3 localized signal . Statistical analysis was performed in GraphPad Prism software v6 . The difference between WT and mutants , or DN rupture timing was evaluated using one-way ANOVA . p < 0 . 05 was considered as significant: *p<0 . 05 , **p<0 . 01 , ***p<0 . 0001 , and ****p<0 . 0001 . Time lapse microscopy ( Fig 3C–3E ) was manually quantified using Fiji , with bordering event counted when 2XFYVE-GFP and galectin-3-mOrange signal where found within 6 pixels distance of each other . Analysis of macropinosomes bordering at the onset of BCV rupture was performed by manually counting all instances of 2XFYVE-GFP positive vesicles present at the rupture site at least one frame before galectin-3 cage appearance ( positive hit ) in contrast to vesicles appearing only during or after galectin-3 cage appearance ( negative hit ) . 36 events were imaged in eight independent experiments All quantitative data in the manuscript is presented as mean , with error bars presented as s . d . Supporting figures materials and methods can be found in S1 Text . | Shigella flexneri is an intracellular bacterial pathogen and the causative agent of bacillary dysentery . It possesses the ability to invade and propagate within human cells by injecting bacterial effector proteins directly into host cells . Shortly after entry within a vacuole , S . flexneri induces vacuolar rupture and escapes into the host cytosol via an unknown mechanism . Using large volume correlative light electron microscopy ( CLEM ) and dynamic microscopy we studied discrete and highly transient steps of S . flexneri early invasion in detail . We provide the first 3D high resolution view of the S . flexneri invasion site and of vacuolar rupture itself . We find that vesicles formed at the invasion site due to injected bacterial effectors , termed macropinosomes , are functionally involved in vacuolar rupture and come into direct contact with the bacterial containing vacuole during this process . This unique and surprising pathogenic strategy stands in stark contrast to other invasive pathogens that induce direct lysis of their surrounding vacuole via the action of destabilizing bacterial proteins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"vesicles",
"pathology",
"and",
"laboratory",
"medicine",
"vacuoles",
"intracellular",
"pathogens",
"pathogens",
"microbiology",
"shigella",
"molecular",
"biology",
"techniques",
"cellular",
"structures",
"and",
"organelles",
"bact... | 2016 | Macropinosomes are Key Players in Early Shigella Invasion and Vacuolar Escape in Epithelial Cells |
Japanese encephalitis ( JE ) is a flaviviral disease of public health concern in many parts of Asia . JE often occurs in large epidemics , has a high case-fatality ratio and , among survivors , frequently causes persistent neurological sequelae and mental disabilities . In 1997 , the Vietnamese government initiated immunization campaigns targeting all children aged 1–5 years . Three doses of a locally-produced , mouse brain-derived , inactivated JE vaccine ( MBV ) were given . This study aims at evaluating the effectiveness of Viet Nam's MBV . A matched case-control study was conducted in Northern Viet Nam . Cases were identified through an ongoing hospital-based surveillance . Each case was matched to four healthy controls for age , gender , and neighborhood . The vaccination history was ascertained through JE immunization logbooks maintained at local health centers . Thirty cases and 120 controls were enrolled . The effectiveness of the JE vaccine was 92 . 9% [95% CI: 66 . 6–98 . 5] . Confounding effects of other risk variables were not observed . Our results strongly suggest that the locally-produced JE-MBV given to 1–5 years old Vietnamese children was efficacious .
Japanese encephalitis ( JE ) is a mosquito-borne flaviviral disease endemic in many regions of Asia [1] . Culex tritaeniorhynchus , the principal mosquito vector of the JE virus ( JEV ) , preferentially breeds in rice fields [2] , [3] . Swine are potent amplifiers of the virus and exhibit rapidly after virus transmission considerable viral loads . Thus , Culex mosquitoes breeding in rice fields and feeding on swine , are critical ecological factors favoring JE transmission to humans in rural areas . Prior to the availability and introduction of vaccines , JE was a significant cause of mortality in the northern provinces of Viet Nam with an annual incidence of 5–15/100 , 000 [4] . Most JE infections ( 96%–99 . 9% ) are asymptomatic or present as a mild disease only with rather non-specific flu-like symptoms . However , among symptomatic patients who exhibit symptoms of encephalitis and/or serious neurologic infection , the case-fatality ratio can be as high as 10%–30% [5] , [6] . Among survivors , 30% to 50% of individuals suffer from chronic , severe neuropsychiatric disabilities [5]–[7] . Why only a small proportion of infected individuals experience severe disease is not clear . Reasons may include host genetic factors , but also virulence factors of differing virus strains . Children younger than 15 years are at the highest risk of infection and the incidence peaks at three to ten years of age [7] . JE infections efficiently induce protective immunity [8] and seroprevalence studies indicate almost universal exposure to the infection in endemic areas by adulthood [9] . Specific antiviral treatment for JE is not available [10] and care of patients strongly depends on supportive measures . Vaccination is the primary strategy for prevention of infection [1] and has been shown to dramatically reduce the disease incidence in South Korea [11] , [12] , Japan [13] , China [14] , [15] , Thailand [15] and Taiwan [16] . In Viet Nam , children receive three pediatric doses ( 0 . 5 ml/dose ) of a locally-produced , mouse brain-derived , inactivated JE vaccine ( MBV; Nakayama strain; VaBiotech , National Institute for Hygiene and Epidemiology ( NIHE ) , Ha Noi , Viet Nam ) with the first two doses at one year of age given at an interval of two weeks followed by a third , booster dose one year later . The production of the MBV in Viet Nam was initiated in 1989 and supported by technology transfer from Japan . Bridging studies suggest that the Vietnamese MBV has an immunogenicity similar to that of the Japanese vaccine , reaching nearly 100% immunogenicity in children after the application of two doses [1] . The vaccine was integrated into Viet Nam's national Extended Program of Immunizations ( EPI ) in selected districts of Ha Tay and Hai Phong provinces in 1997 . Hospital-based surveillance of all patients presenting with an acute encephalitis syndrome ( AES ) is ongoing in the Ha Tay and Hai Phong Provincial Hospitals in the North of Viet Nam . Two provincial hospitals in the Ha Tay Province as well as the National Pediatric Hospital in Ha Noi are used as referral hospitals for JE surveillance since January 1 , 2004 . These hospitals jointly account for approximately 95% of all cases of acute encephalitis notified in the Ha Tay Province . In Hai Phong province , cases are identified through the national AES surveillance system . Cerebrospinal fluid ( CSF ) is collected at admission . The presence of immunglobulin-M ( IgM ) antibodies to JEV ( anti-JEV ) in CSF is defined as one of the criteria for a JE diagnosis [17]–[19] . Initial testing of specimens is performed at the National Pediatric Hospital Laboratory and Laboratory of Ha Tay Preventive Medicine Center , and confirmatory testing of all specimens was done by sending coded samples to the Department of Virology , Armed Forces Research Institute of Medical Sciences ( AFRIMS ) , Bangkok , Thailand . The objective of the present study was to assess in a case-control design the effectiveness of this MBV JE vaccine .
The study was conducted according to ethical principles consistent with the International Guideline for Ethical Reviews of Epidemiologic Studies [20] . The Institutional Review Board of the International Vaccine Institute , Seoul , Republic of Korea , and the local ethical committee of NIHE , Ha Noi , Viet Nam approved the study and granted ethical clearance . Vaccination status' of participants were assessed from vaccination records of the health centers . During visits of the households of children that were aimed at assessing the distances of piggeries and ricefields from the houses , oral informed consent was obtained from parents/guardians of cases and controls and documented in the questionnaire . Oral consent was considered appropriate for the study and approved by both Ethical Review Boards . For this matched case-control study , patients with a confirmed diagnosis of JE and younger than 15 years of age were identified from the database of the surveillance hospitals from January 2004 to December 2007 . Older children were not recruited , as they could not have received JE vaccines through the national immunization program . Controls were chosen from the birth registry of all births in the health center of the village of the case and matched for gender , age ( +/− six months ) , and proximity of their house . Only after four controls had been selected from birth registry and agreed upon by the study team , the vaccination record books were opened and the individual vaccination status was assessed . JE vaccination is provided during an annual mass campaign and vaccination records are kept at the health centers . The houses of cases and controls were visited and as an indicator of potential exposure to infection and the distances to piggeries and rice fields were assessed by the study team after obtaining oral informed consent from the adult/guardian present . A Student's t-test for unequal distributions was used to compare the age distribution of cases and controls . A chi-square test was employed to compare cases and controls for the proportion of males , proportion living within ≤50 meters of a piggery and percent living ≤30 meters of rice fields . A matched conditional logistic regression model was employed to evaluate the protective effect of immunization . A single dependent ( JE disease ) and independent variable ( immunization status ) was entered into the model . Odds ratios ( OR ) , 95% confidence intervals ( CI ) , and p values were calculated from model parameters . The protective effect of immunization was assessed for residents receiving three or more doses relative to two or less doses . Vaccine effectiveness ( VE ) was calculated as VE = 1−OR×100 [21] . The threshold of statistical significance was p<0 . 05 . All statistical analyses were performed using the Stata v 10 . 0 software ( Stata Corp . , Texas , USA ) .
We identified 30 laboratory-confirmed cases of JE infection and these were matched to 120 controls . After selection of controls was completed , it became evident that two controls were chosen who had no information on their immunization status in the birth registries . These two controls were excluded from the calculation of vaccine effectiveness . As cases and controls were matched for age and gender , no significant differences were observed between cases and controls for these variables . Cases and controls resided equidistantly from rice fields and piggeries ( Table 1 ) . Only individuals who had received the full course ( three doses ) of JE vaccine were considered as vaccinated; two cases and two controls received one and two cases and six controls two doses of JE vaccine , respectively , and were considered as not vaccinated ( Table 2 ) . Among the 30 laboratory-confirmed cases , the proportion of cases vaccinated was 60 . 0% , compared to 86 . 4% among the 118 control individuals . Based on the matched analysis vaccine effectiveness reached 92 . 9% [95% CI: 66 . 6–98 . 5%] ( Table 2 ) .
Three previous trials have evaluated the effectiveness of the JE MBV . A randomized double-blinded study conducted in northern Thailand , using JE MBV produced in Thailand , yielded an overall effectiveness of 91% [95% CI: 70 . 0–97 . 0] [22] . Another trial in Taiwan evaluated a Taiwanese vaccine and revealed an effectiveness of approximately 85% when two or more doses were administered [23] . A case-control study in Thailand showed an effectiveness of 94 . 6% in children ≥18 years of age [24] . The present matched case-control study suggests that the MBV produced in Viet Nam yields results that are similar to those of the Thailand and Taiwan studies . Even though the MBV has an excellent efficacy , its usage was recently restricted after considerable safety concerns were raised . Severe reactions such as hypersensitivity , including generalized urticaria and angioedema , occurred at far higher rates than observed in other routine vaccinations [25] , [26] , [27] , [28] . Consequently , Japan no longer recommends JE MBV vaccination [29] , [30] and recently introduced a Vero cell-derived JE vaccine; other countries are currently in the process of phasing out the MBV ( e . g . , Thailand , Sri Lanka ) . Nevertheless , MBV safety issues have not led to any restrictions in its use in Viet Nam . However , Viet Nam is currently developing its own Vero cell-derived vaccine , which is foreseen to be available for use in 2014/2015 . The Vietnamese government has announced that it would eliminate clinical JE by 2015 . However , in contrast to poliomyelitis virus , for which humans are the only hosts , the JE virus is enzootic and is , therefore , unlikely to be entirely eradicated from the environment . A recent JE surveillance study has shown that , even in the virtual absence of manifest human disease after vaccination , JEV is still widespread among swine [31] , showing that active transmission is perpetuated and that protective immunity of humans through persistent vaccination is a key measure to preventing disease in humans [13] . Therefore , controlling clinical JE disease through vaccination would not impact on reducing or eradicating the circulation of the virus within the vectors and animal hosts as JE is not transmitted from person to person and JE vaccination does probably not confer herd immunity [32] . Depending on the potency of enzootic transmission and the age-specific risks of natural human infection , the age for primary vaccination differs between countries . Most countries give one to two booster doses after the initial three-dose regimen [1] . In contrast to Japan and Korea , where nation-wide changes in lifestyle have provided additional contributions to the control of JE [11] , [33] , rural areas in Viet Nam have so far largely remained agrarian . Rice paddies cover extensive geographic areas , and little changes only have occurred in the natural environment and living conditions . Even though the effectiveness of the vaccination program appears to be high , an annual average incidence of 3 . 4/100 , 000 was observed among children less than 10 years of age . The current JE immunization program may be improved by immunizing younger children ( 6–23 months of age ) who are at the highest risk of patent infection , and providing a booster dose at 3–5 years after initial immunization with the three-dose regimen , reducing waning immunity in immunized children [1] . An inherent problem of inactivated vaccines is that due to their low immunogenicity , multiple vaccinations are required in order to induce and maintain sustained levels of protective immunity . Following the initial three doses given at immunization , the effectiveness of the MBV declines over years [34] . Therefore , the introduction of one or more booster doses using the adult formulation to children at school age is recommended to ensure protective immunity against JE using MBV . | Japanese encephalitis ( JE ) is a disease caused by a flavivirus transmitted by mosquitoes . Although pigs and wild birds are main reservoirs of the disease , it is occasionally transmitted to humans . The majority of infections in humans are asymptomatic . In persons developing encephalitis , JE has a high case-fatality rate and , among survivors , JE frequently causes persistent neurological sequelae and mental disabilities . Therefore , it is a public health concern in many parts of Asia and many countries vaccinate against JE . Since 1997 , children in Vietnam are vaccinated in high risk areas and receive a locally-produced vaccine . This study is aimed at evaluating the effectiveness of the Vietnamese JE vaccine through a case-control study , in which 30 cases and 120 controls were enrolled . The effectiveness of the JE vaccine was 92 . 9% [95% CI: 66 . 6–98 . 5] , which suggests that the locally-produced JE vaccine given to 1–5 year old Vietnamese children was efficacious . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"clinical",
"research",
"design",
"infectious",
"disease",
"epidemiology",
"flavivirus",
"immunology",
"vaccines",
"neglected",
"tropical",
"diseases",
"infectious",
"disease",
"control",
"japanese",
"encephalitis",
"... | 2012 | Effectiveness of the Viet Nam Produced, Mouse Brain-Derived, Inactivated Japanese Encephalitis Vaccine in Northern Viet Nam |
Transcriptome variation plays an important role in affecting the phenotype of an organism . However , an understanding of the underlying mechanisms regulating transcriptome variation in segregating populations is still largely unknown . We sought to assess and map variation in transcript abundance in maize shoot apices in the intermated B73×Mo17 recombinant inbred line population . RNA–based sequencing ( RNA–seq ) allowed for the detection and quantification of the transcript abundance derived from 28 , 603 genes . For a majority of these genes , the population mean , coefficient of variation , and segregation patterns could be predicted by the parental expression levels . Expression quantitative trait loci ( eQTL ) mapping identified 30 , 774 eQTL including 96 trans-eQTL “hotspots , ” each of which regulates the expression of a large number of genes . Interestingly , genes regulated by a trans-eQTL hotspot tend to be enriched for a specific function or act in the same genetic pathway . Also , genomic structural variation appeared to contribute to cis-regulation of gene expression . Besides genes showing Mendelian inheritance in the RIL population , we also found genes whose expression level and variation in the progeny could not be predicted based on parental difference , indicating that non-Mendelian factors also contribute to expression variation . Specifically , we found 145 genes that show patterns of expression reminiscent of paramutation such that all the progeny had expression levels similar to one of the two parents . Furthermore , we identified another 210 genes that exhibited unexpected patterns of transcript presence/absence . Many of these genes are likely to be gene fragments resulting from transposition , and the presence/absence of their transcripts could influence expression levels of their ancestral syntenic genes . Overall , our results contribute to the identification of novel expression patterns and broaden the understanding of transcriptional variation in plants .
The maize species exhibits high levels of phenotypic variation , which is likely the result of both genetic and epigenetic variation [1] . Dissection of genetic and epigenetic variation may shed light on the understanding of phenotypic variation and provide tools to accelerate maize breeding . The maize genome has a complex organization with interspersed repetitive elements and genes [2] . The genomes of different maize inbreds can vary substantially due to single nucleotide polymorphisms [3] , small insertions/deletions [4]–[5] , gene copy number variation ( CNV ) and genomic presence-absence variation ( PAV ) [2] , [6]–[7] . Transposable elements , discovered in maize by Barbara McClintock [8]–[9] , comprise a significant portion of the maize genome [2] , [10]–[13] and can contribute substantially to genomic variation among lines [14]–[17] . There are many examples illustrating the potential for transposons to capture and mobilize genes or gene fragments [14] , [18]–[24] . In addition to genetic changes , there is also evidence for epigenetic variation among maize inbred lines . The epigenetic differences vary within maize populations and show relatively stable trans-generational inheritance [25] . These diverse forms of genetic and epigenetic variation likely interact to affect relative transcript abundance , which contributes to phenotypic variation among maize individuals . Exploring transcriptome variation and elucidating the underlying mechanisms of transcriptional regulation may further our understanding of the molecular bases of phenotypic variation [26]–[27] . Several groups have used microarray profiling to compare the transcriptomes of maize inbreds [28]–[33] . A comparison of the F1 hybrids and the parents revealed that much of the parental variation resulted in additive expression with some rare examples of unexpected expression in the F1 [28] , [33] . A recent RNA-seq based analysis of transcriptomic variation in 21 maize elite inbred lines found that a substantial number of genes showed presence/absence expression patterns [34] . Genetical genomics or expression quantitative trait loci ( eQTL ) mapping is an efficient method for understanding the genetic basis of transcriptome variation [26]–[27] , [35]–[36] . eQTL mapping uses transcript abundance as a phenotypic trait and maps the genomic loci controlling the transcript abundance [35] . eQTL are generally classified as cis- or trans- depending on whether they are physically linked to the gene that is regulated or unlinked , respectively . Both cis- and trans-eQTLs have been identified in plants and while trans-eQTLs are more abundant , they generally explain less expression variation than cis-eQTLs [37]–[42] . Several eQTL mapping experiments have utilized microarrays to reveal the complexity of transcriptome variation and their underlying genetic regulators such as trans-eQTL hotspots in human , animals and plants [37] , [39] , [42]–[44] . eQTL mapping of transcriptome variation has also been employed directly to help dissect phenotypic variation [42] , [45]–[46] . The analyses of transcriptome variation in segregating populations have generally focused on exploring how a single locus contributes variation to transcript abundance in a Mendelian fashion . However , there is also the potential for non-Mendelian segregation of gene expression levels [47] . RNA-based sequencing ( RNA-Seq ) provides several key advantages for transcriptome research including robust expression detection especially for lowly expressed genes , unprecedented access to the fine structure of the transcriptome , and powerful detection of all the transcripts not depending on the reference genome annotation [48]–[49] . Here , we employed RNA-Seq on shoot apices of a well-studied maize intermated RIL population derived from B73 and Mo17 ( IBM ) [50] . We characterized the relationship of transcriptional variation between the progeny population and the parents in detail to understand how the parental variation combines to affect transcript abundance . This analysis identified a number of genes that exhibit unexpected patterns of expression variation including paramutation-like segregation patterns and presence/absence expression patterns between progeny and parents . Meanwhile , global eQTL mapping , a pair-wise epistasis scan and co-expression analysis were conducted to dissect the possible factors underlying this variation .
A population of RILs is expected to segregate 1∶1 for the parental alleles and provides an opportunity to examine variation in transcript abundance within the RILs and the relationship between the population and the parents . We first focused on the expression levels of 22 , 242 genes that were detected in both parents and at least 90% of the IBM RILs . The mean expression levels in the RILs were similar to the mid-parent values for most genes ( Figure 1B ) . Transgressive segregation , defined here as at least 10% of RILs exhibiting expression levels outside the parental range , was observed for 598 genes ( 2 . 6% ) . The other 21 , 644 ( 97 . 4% ) genes have expression levels in the RILs that are within the parental range . The level of variation for gene expression levels in the RILs was significantly correlated with the difference between the two parents ( Pearson's product-moment correlation: r = 0 . 728 , P<2 . 2E-16; Figure 1C ) . The type of distribution for expression levels within the RIL population relative to the parents was assessed using a τ score [52] . We found that 4 , 822 ( 22% ) genes fit bimodal distributions , 14 , 564 exhibited normality ( 65% ) and the remaining 2 , 856 ( 13% ) showed other unclassified distributions . Genes with little or no expression difference among the parents typically exhibited a normal distribution in the RILs ( Figure 1D ) . However , many genes with large expression differences among the parents exhibited a bimodal distribution among the RILs ( Figure 1D ) . These trends indicated that much of the variation in gene expression levels in the RILs is reflective of differences present between the parents . While the majority of genes exhibit expression patterns in the RILs that are quite predictable from the parental levels , there were a subset of genes ( 0 . 7% ) that have average expression levels in the RILs that are greater than 2-fold different than the mid-parent , indicative of other potential patterns of expression variation . It is possible that some of these genes may have expression patterns similar to those observed for genes that are subject to paramutation such that the expression levels in all RILs would be similar to the expression level of one of the parents [53] . The distribution of expression patterns in the RILs was compared to the parental expression patterns for 8 , 269 out of 28 , 603 detected genes that have at least two-fold expression level difference between B73 and Mo17 . There were 145 genes ( 86 of these genes are from the 22 , 242 genes expressed in both parents and 90% of the RILs ) with paramutation-like expression patterns for the RILs in which one parent was within the expression distribution ( two standard deviations from the population mean ) of the RILs but the other parent had an expression level at least three standard deviations from the population mean ( Figure 2A; Figure S2; Table S2 ) . It is important to note that , while these genes exhibit patterns that are similar to those expected due to paramutation these genes may either be directly regulated by paramutation or be secondary targets that are influenced by another factor that is subject to paramutation . For many ( 80/145 ) of these genes one of the two parents had an expression level that was outside the range of all RIL genotypes . The expression levels of B73 and Mo17 relative to the population mean and standard deviation helps illustrate several trends observed for these genes ( Figure 2A ) . The majority of these genes ( 124/145 ) had patterns in which the RILs were all expressed at levels similar to the lower parent as might be expected given that most examples of paramutation involve a paramutagenic allele that is expressed at lower levels than the paramutable allele ( Figure 2B ) . The expression level for these genes was assessed in the F1 hybrid relative to the two parents ( Figure S3 ) . Well characterized examples of paramutation in maize include some examples of dominant expression in the F1 as well as other examples that do not exert effects until the F2 generation [54]–[57] . The genes that had high levels of expression in all RILs were expressed at additive levels in the F1 . The genes that had expression levels similar to the lower parent included many examples of additive expression but also had a number of cases with partial to complete dominance in expression such that the F1 had levels more similar to the lower parent ( Figure S3 ) . Ten of the genes with paramutation-like patterns were selected for analysis in F2 individuals ( Figure S4 ) . Seven of the ten genes exhibited paramutation-like patterns in the F2 individuals and these include examples of both high and low expression . The basis for the regulatory variation in transcript levels was examined using a high-resolution SNP genetic map of the IBM population based upon 7 , 856 high quality SNP markers derived from the RNA-seq data to perform eQTL analysis for the 22 , 242 genes that are expressed in both parents and at least 90% of the RILs . This approach is likely to capture much of the variation for gene expression that segregates in a Mendelian fashion but is less likely to capture the basis of variation for examples of gene expression such as those described above . A total of 30 , 774 eQTLs ( α = 0 . 05 ) with a threshold logarithm of odds ( LOD ) > = 4 . 17 were identified for 19 , 304 genes , of which 5 , 303 ( 27 . 5% ) were controlled only by a single cis-eQTL , 6 , 201 ( 32 . 1% ) controlled by both cis- and trans-eQTLs and 7 , 800 ( 40 . 4% ) only by trans-eQTLs . The 30 , 774 eQTLs include 11 , 504 ( ∼37% ) cis-eQTLs and 19 , 270 ( ∼63% ) trans-eQTLs ( Figure 3A and Table S3 ) . The number of eQTLs affecting the expression level of each gene ranged from zero to six . In general , cis-eQTLs tend to have larger effects than trans-eQTLs ( Figure S5A ) . For example , 83 . 7% of cis-eQTLs account for at least 20% of the expression variation in contrast to only 12 . 7% of the trans-eQTL meeting this criterion . However , there are examples of trans-eQTLs that contribute substantially to expression variation . There were 133 trans-eQTLs that contribute at least 60% of the variation for expression of a target gene . The overall contribution of cis- and trans-eQTLs was heavily influenced by the level of expression variation in the parents ( Figure S5B ) . The contribution of cis-eQTLs increased as the parental expression level became increasingly different . In addition , the amount of variation explained by the cis-eQTL also increased as the parental expression levels become more different ( Figure S5C ) while the amounts of variation explained by trans-eQTL decreased as the parental differences increased ( Figure S5D ) . The proportion of cis- and trans-eQTL for the 598 genes exhibiting transgressive segregation was similar to the proportion of cis- and trans-eQTL for the global eQTL analysis , however , the genes with transgressive segregation were more often ( 37% ) controlled by multiple eQTLs with opposite effects than all genes ( 27% ) . The genomic distribution of trans-eQTL was assessed in an attempt to identify potential trans-eQTL hotspots that might reflect substantial regulatory differences between B73 and Mo17 . The analysis of trans-eQTL density in a 1 Mb ( which is slightly larger than the average physical distance between adjacent markers with a recombination event ) sliding window revealed 96 significant ( P<0 . 01 ) trans-eQTL hotspots ( Figure 3B and Table S4 ) , including 10 major hotspots that contain at least 200 trans-eQTLs ( Table 1 ) . These hotspots have many more trans-eQTL than other genomic regions and in the majority ( 78% ) of examples the target genes regulated at the trans-eQTL hotspots show a consistent pattern with significantly more target genes altered in expression in the same direction by the haplotype at the trans-eQTL hotspot ( haplotype bias ) . More examples in which the B73 allele ( 49 ) at the trans-eQTL hotspot promoted higher expression of the target loci than the Mo17 allele ( 26 ) were identified . The lists of target genes regulated by each of the trans-eQTL hotspots were used to search for GO enrichments; 43% of the trans-eQTL hotspots target lists exhibited enrichments for at least one GO term ( Table S5 ) . We performed further analyses for the ten trans-eQTL hotspots that had at least 200 targets ( Table 1 ) . Nine of these ten trans-eQTL hotspots showed consistent haplotype bias ( six for B73 and three for Mo17 ) and the targets for each of these hotspots had GO enrichments for at least one term . Multiple genes in the same MaizeCyc pathway [58] are observed to be co-regulated by the same trans-eQTL hotspot ( Figure 3C , Table S6 ) . These trans-eQTL hotspots may be due to functional differences in transcriptional regulators . At least in some cases it might be expected that differential expression of a regulator present at the trans-eQTL hotspot is the cause of the differences in trans-regulation . To examine the influence of structural rearrangements-gene copy number variation ( CNV ) and genomic presence/absence variation ( PAV ) on gene expression , we compared our transcriptomic data for the 28 , 603 expressed genes with previous Comparative Genomic Hybridization ( CGH ) data [59] . We focused on the full set of 28 , 603 genes as the more limited set of 22 , 242 genes assessed for eQTL analysis required expression to be present in both parents while some of the PAV are expected to abolish expression in Mo17 . There are 1 , 212 expressed genes with CNV/PAVs that affect the gene or flanking regions ( Table S7 ) . The structural rearrangements include copy number gains in B73 or Mo17 as well as PAV that are present in B73 but absent in the Mo17 genome . We might expect that copy number gains would lead to increased expression in the genotype with more copies while PAV would only be expressed in one genotype . There was evidence that this was true in many cases ( Table S8 and Figure 4 ) . eQTL mapping was conducted on these CNV/PAV-related genes and a total of 1 , 466 were identified for 1 , 009 genes , of which 704 ( 69 . 8% ) were controlled by cis-eQTLs . The cis-eQTLs proportion of genes with CNV/PAVs nearby is significantly higher ( P = 0 . 00 ) than those of all detected genes ( Figure 4A ) . Noteworthy was the observation that 89 . 2% of these genes entirely within the PAV were controlled by cis-eQTLs , while ∼10% of these genes have trans-eQTLs , indicating that other regulators underlie the expression variation in addition to PAVs . There was also evidence for an enrichment of cis-acting variation when the CNV/PAV occurred in regions surrounding the gene . Nearly half ( 120/242 ) of the genes entirely within structural variants exhibit differential expression in B73 and Mo17 . There were many examples in which the RIL genotype at the gene of interest was highly correlated with the expression difference ( Figure 4B , 4C ) . Typically , the copy number of genes entirely within CNV/PAV regions positively correlated with the genes' expression ( 99 out of the 120 differentially expressed genes between the two parents ) ( Figure 4B ) . We also noted examples ( 21/120 ) in which a copy number gain was associated with lower expression in the parents ( Figure 4C ) . We were struck that a large proportion of genes were only detected in a subset of the RILs or parents . While there were 22 , 242 genes expressed in both parents and the RILs , there were an additional 6 , 361 genes that had detectable ( False Discovery Rate-FDR>0 . 05 ) levels in at least 10% of the RILs or at least one of their parents . These 6 , 361 genes may include ( a ) some genes with very low expression levels that manage to cross the threshold of detectability in some samples but not others , ( b ) genes that are only expressed in one parent and that based on Mendelian segregation would therefore be expected to be expressed in only 50% of the RILs , and ( c ) genes with unusual regulatory mechanisms . We elected to impose a more restrictive set of filtering criteria for expression to limit the number of low-expressed genes near the detection threshold . Based on the alignment of RNA-seq reads to non-genic genomic regions , an RPKM of 1 . 03 corresponds to a FDR of 0 . 01 and 499 of the 6 , 361 genes have a RPKM value of ≧1 . 03 in at least 10% of the RILs or at least one of their parents . A substantial proportion of these genes ( 289/499 ) were expressed in only one of the parents and were observed in approximately 50% of the RILs ( with the Chi-square test at the P value<0 . 01 ) . The lack of expression in one parent and half of the RILs may reflect differences in genome content or regulatory variation . eQTL analysis of these genes revealed that 186 ( 64% ) of these genes had cis-eQTL that explained >20% of the expression variation and 54 of these genes intersect with CNV/PAVs . However , there were also 92 ( 32% ) of these genes that had evidence for at least one strong trans-eQTL with R2>20% . In total , eQTLs could explain more than 20% of the expression variation of 273 of these 289 genes ( 96 . 1% ) . The other 210 of these genes exhibited unexpected patterns of expression that could be classified into four groups ( Table S9 ) . The type I pattern included 40 genes that were expressed in both parents but were not detected ( RPKM = 0 ) in over 10% of the RILs . The type II pattern included 19 genes that were not detected ( RPKM = 0 ) in the parents but were detected in at least 10% of the RILs . The type III patterns include genes that were expressed in one parent but not the other and had expression in very few RILs ( type IIIA – 66 genes ) or the majority of the RILs ( type IIIB – 85 genes ) ( Figure 5A ) . A subset of genes ( 2 type I genes and 19 type III genes ) with unexpected expression patterns also exhibited paramutation-like expression patterns . These unexpected patterns of expression detected by RNA-seq were validated for the majority of genes tested ( 43/55 ) using RT-PCR on a subset of the RIL genotypes ( Figure S6 ) . In addition , the same type of expression patterns could be observed in an independent set of B73×Mo17 F2 individuals for all the six tested genes ( Figure 5B ) . These RT-PCR assays confirmed that the unexpected segregation patterns for presence or absence of gene expression observed in the RILs are reproducible . Further , genomic PCR was employed to assess if the expression presence/absence transcript variation might be attributable to differences in genome content . We found that genes exhibiting presence/absence transcript variation could be amplified from genomic DNA of each of the IBM RILs that were tested ( Figure S7 ) , indicating that the difference in expression was not due to segregation for genomic presence of the sequence . For each of the four patterns , the proportion of RILs expressing a gene was compared to the mean expression level in genotypes that express the gene ( Figure 5C ) . Some of these genes are quite highly expressed and there is a substantial range in the number of genotypes with expression . To further distinguish the genes with unexpected expression patterns from the genes with very low expression levels that manage to cross the threshold of detectability , we examined the maximum expression levels , population mean RPKM and the standard deviations in the population of the genes with unexpected expression patterns in comparison to all detected genes expressed in more than 90% of the RILs . Although the maximum expression levels , population mean RPKMs of genes with unexpected expression patterns are slightly lower than those of all expressed genes , the differences are not significant ( Figure S8 ) . Importantly , the standard deviation of expression levels of genes with unexpected expression patterns is similar to that of all other genes ( Figure S8 ) . The observation that there were many examples in which the proportions of RILs with detected expression was close to 25% or 75% ( Figure 5C ) may suggest that multiple genetic factors play interaction roles underlying the unexpected expression patterns for some of these genes . To test this hypothesis , a genome-wide epistasis scan with all possible pair-wise marker interactions was employed to search for evidence of two-locus interactions that control expression for genes that were detected in approximately one-quarter or three-quarters of the RILs . If two different loci are both required to achieve expression of a gene , these loci could both be present in one parent ( type III ) or could have one functional locus in either parent ( type II ) . In these examples we would expect 25% of the RILs to exhibit expression of the gene . There are 28 type IIIA and 10 type II genes with expression in only ∼25% of the RILs using Chi-square test with the p-value<0 . 01 as the cut-off . A genome-wide scan for two-locus interactions that control the variation of expression for these 38 genes found that 92% of these could be explained by a two locus interaction ( Figure S9A , S9B ) . In half of the cases in which a two-locus interaction explained a significant proportion of the expression variation we found that one of the two loci mapped in cis to the gene itself . We could also envision a scenario in which two different loci are required for loss of expression of a gene and this would be expected to result in expression in 75% of the RILs . There are 71 type IIIB and 28 type I genes that are expressed in ∼75% of the RILs and for 91% of these genes the pattern of presence/absence can be explained by a two-locus interaction , including 12 examples in which one of the two loci maps in cis to the gene itself ( Figure S9C , S9D ) . This suggests that a significant subset of the genes with unexpected patterns of presence-absence for expression can be explained by two-locus interactions . The genes that exhibit presence/absence expression patterns in progeny relative to their parents were further characterized . As a group , these genes with unexpected expression patterns were enriched for single copy genes , and for low copy number gene families relative to all maize genes ( Table 2 ) . The FGS ( Filtered Gene Set ) genes of maize represent an attempt to identify higher confidence gene models and remove gene fragments and transposon-derived sequences [2] . However , there are likely a number of gene fragments and transposon-derived sequences still present within the FGS . Comparative genomic localization can provide more confidence in syntenic genes as “real” genes [60] . Only 36/210 genes with presence/absence expression patterns are in syntenic locations relative to other grass species ( Figure 5D ) . This is a smaller proportion than expected based on the finding that 67 . 5% of all FGS genes are located in syntenic positions . It is worth noting that while the genes with unexpected patterns are enriched for non-synteny there is a subset of these genes that do have synteny and likely represent functional genes ( Table S9 ) . Annotation of the syntenic genes with unexpected presence/absence expression patterns reveal a variety of putative functions such as serine threonine protein kinase , electron transport sco1 family protein and basic leucine-zipper 44 protein , but there is no evidence for GO enrichments within this set of genes . The 174 genes with unexpected segregation patterns that are non-syntenic with other grass species may represent insertions of these genes or gene fragments in the maize genome . To test the hypothesis , the genomic regions surrounding these genes were examined for enrichment of specific classes of repetitive sequences ( Figure S10 ) . Over one-third ( 65 ) of the 174 genes had a CACTA-like element within 20 kb and these include examples of all types of unexpected expression patterns . This is significant ( P = 0 . 00 ) enrichment of CACTA-like transposable elements surrounding these genes relative to the expected genomic frequency ( Figure 6A ) . The 65 genes with CACTA-like sequences nearby ( 3 . 20 exons ) and the other 145 genes with unexpected segregation patterns ( 3 . 10 exons ) tended to have fewer exons ( P = 0 . 00 ) than the average exon number ( 4 . 88 exons ) of all maize genes ( Figure 6B ) . These features , less exons , non-syntenic genomic localization and CACTA-like element enrichment , suggest that many of these genes may be gene fragments that were captured and transposed by CACTA-like transposons . We proceeded to assess whether the non-syntenic gene fragments with presence/absence expression might affect the regulation of homologous full-length syntenic genes ( ancestral syntenic genes ) elsewhere in the maize genome . All 174 of the non-syntenic genes were homologous to at least a portion of another maize gene ( E value<1 . 0E-10 ) . The correlation between the expression level of each of these genes and the other homologous full-length sequences ( possible ancestral syntenic genes ) was assessed in the RIL population . There were 25 examples in which the presence/absence expression patterns of the non-syntenic genes were correlated with transcript abundance for ancestral syntenic genes ( Table S10 ) . For example , the presence/absence expression of a gene fragment located on chromosome 3 was highly correlated with the abundance of a transcript from its ancestral syntenic gene annotated as an Erwinia Induced Protein 1 located on chromosome 5 ( Figure 7A ) . A comparison of the expression levels for the two sequences revealed an inverse correlation such that the presence/absence of transcripts from the gene fragment correlated with low or high expression of the ancestral syntenic gene ( Figure 7B ) . However , the presence/absence of transcripts from the transposed fragment does not result from genomic differences among RILs because according to the genomic PCR amplifications this gene fragment exists in all tested RILs ( Figure 7C ) . The expression pattern of gene GRMZM2G004617 was also identified to be controlled by two-locus interaction ( Figure 7D ) . Many ( 20 ) of the other 25 examples involve similar negative correlations between presence/absence of a gene fragment and abundance of a full-length transcript ( Table S10 ) . These examples provide evidence for the ability of transposed gene fragments to influence transcript abundance of their ancestral syntenic genes .
The analysis of eQTLs allows for the dissection of the genomic regions that affect transcript abundance . Cis-regulatory eQTL reflect regulatory variation that is tightly linked to a gene and affects the allelic expression levels . In contrast , trans-eQTL reflects regulatory variation at unlinked genomic positions . The analysis of all trans-eQTL can reveal trans-eQTL hotspots , also known as trans-eQTL clusters , which are genomic regions that affect the expression of many unlinked loci [39] , [61] . These trans-eQTLs are thought to reflect differences in gene regulation that may be important for phenotypic variation [39] , [41]–[42] , [44] . Due to the limitations of mapping resolution , the identified trans-eQTL hotspots could result from the presence of a single causal regulatory factor ( pleiotropic effects ) or several tightly linked loci that affect transcript levels of different genes ( genetic coupling ) [62] . In addition , each trans-eQTL hotspot is relatively large ( ∼1 Mb ) and will likely include the targets of the hotspot itself as well as several other trans-eQTLs that only regulate a small number of genes . Most of the trans-eQTL hotspots identified in our study showed significant haplotype effect bias , which means the haplotype of one parent could increase expression levels of significantly more target genes than expected . The hotspots with haplotype effect bias are more likely to reflect “master regulators” , while some of the others may be a result of genetic linkage , even though we had already taken gene density into account . It might be expected that variation in an important regulatory locus may result in variation for transcript levels for a number of genes that share related GO annotation or are present in the same biochemical pathway . Here , the expression level of genes involved in these pathways were found to be consistently altered in the same direction by trans-eQTL hostpots , which implies that pathway variation may exhibit genetic variation underlying the phenotypic variation among different elite inbred lines . The regulatory variation provided by the trans-eQTL could be the result of differences in the expression level for a regulator located within the trans-eQTL ( a cis-eQTL ) or it could be the result of a qualitative variant for a gene located within the genomic region . If the cause of the trans-eQTL hotspot is a cis-regulatory variant then we would expect to find a cis-eQTL located within the trans-eQTL that is highly correlated with the expression level of the target genes . The analysis of these cis-eQTLs located within the trans-eQTL hotspot did not find enrichment for transcription factors . However , we did identify transcription factors or other putative regulatory genes . These candidate genes provide a potential avenue for future research to understand the basis of regulatory variation in maize ( Table S11 ) . The majority of genes behave in a manner that is predictable based on the expression levels of the parents . In general , genes with relatively little expression variation in the parental genotypes exhibit a normal distribution of expression levels centered on the parental levels in the offspring and the genes with variation between the parents exhibit a bimodal distribution in the offspring . Our results showed that for the majority of genes expression trait variation is mainly caused by additive effects , which differs from the results observed in Arabidopsis , and rice where non-additive gene action was the more common form of regulating transcript accumulation [39]–[42] . However , a portion of genes exhibit transgressive segregation in the RILs such that at least 10% of the RILs exhibit expression levels outside the parental range . The proportion of transgressive segregation for expression traits was small ( 2% ) compared with the levels reported in other species [39]–[42] . The measurement of eQTL for many genes at once provides an opportunity to assess the potential causes of transgressive segregation . One likely cause of transgressive segregation would be the presence of multiple trans-eQTL including examples in which both parental haplotypes promote expression . For example , if a single gene has two trans-eQTLs for which the B73 allele promotes higher expression and two other trans-eQTLs in which the Mo17 allele promotes higher expression then one might expect to observe a number of RILs with expression levels that are higher or lower than the parental values due to segregation of these trans-eQTLs . Indeed , we found that the 598 genes with transgressive segregation tended to have higher numbers of trans-eQTL than the other genes and that these frequently included a mixture of B73/Mo17 favorable alleles for the underlying gene expression trait . While the majority of genes behaved in predictable fashions in the RILs relative to parents and had variation that could be attributed to eQTL there were some genes with unexpected expression patterns . We focused our analysis on a couple of subsets of these genes including genes with paramutation-like pattern of expression and genes with unexpected patterns of presence/absence of the transcripts . When two parents exhibit variation in a trait it would be expected that off-spring would exhibit a similar range of variation . However , we found a number of genes for which none of the recombinant off-spring had expression levels similar to one of the parents . This is an apparent violation of Mendel's principle of segregation and might be reminiscent of paramutation . Paramutation describes instances in which there is communication between two alleles that are present in a heterozygote [53] , [63]–[65] . The paramutable allele can be altered to behave more like a paramutagenic allele . Most of the examples of paramutation have been described in maize [64] . These examples include a variety of stabilities and behaviors [64] but are often sensitive to mutations in the same genes [65]–[67] . It has been hypothesized that paramutation will affect numerous other genes but that these other examples may not have been noted due to the lack of observable phenotypes . A recent study in tomato identified several transcripts that had expression patterns in RIL genotypes that were not indicative of the parental levels and could indicate paramutation [47] . We searched for examples of genes that had expression patterns that might be expected to result from paramutation . There were 145 examples of genes for which all of the RILs had expression similar to one of the parents while the other parent had a unique expression pattern . The majority ( 55% ) of these genes represent examples in which the RILs all had expression levels similar to the lower expressing parent . The fact that these patterns were observed in RILs that have been subjected to >6 generations of inbreeding would suggest that these patterns of expression are relatively stable . While we do not have evidence to show direct interaction of the alleles in the heterozygote , we propose that the expression patterns observed for many of these 145 genes are the result of paramutation-like phenomena . Our analysis of expression in a RIL population relative to the parents suggests that paramutation-like mechanisms may contribute to regulatory variation for a number of maize loci . The analysis of F2 individuals provided further evidence for paramutation-like patterns for seven of the ten genes tested . It is possible that some of these examples may reflect spontaneous mutation or epimutation in the specific B73 and Mo17 individuals that were used for this study and these may account for the lack of validation for some examples . We also examined our dataset for genes whose expression was only detectable in a subset of the RIL population or at least one of the parents . Nearly 500 genes with various patterns of segregation for the presence/absence of transcripts were identified using a relatively stringent ( FDR = 0 . 01 ) expression threshold . If the threshold for detection was relaxed ( FDR = 0 . 05 ) , the number of genes with segregation for presence/absence of transcripts increased to 4 , 689 . These results suggest the presence of substantial qualitative as well as quantitative variation for the maize transcriptome following segregation . We further evaluated these genes to begin to understand the causes and consequences for this variation . The most likely cause for variation in presence/absence of a transcript would be examples in which one parent expresses a gene and the other parent does not . In these instances we would expect approximately 50% of the RIL progeny to exhibit expression of the gene . Over half ( 289/489 ) of the genes with segregation for the presence of transcripts exhibit this type of pattern . This pattern could be caused by a strong cis-regulatory variant or actual difference in genome content such as PAV [6] , [59] . The mapping of regulatory variation for these 289 genes revealed that many of them can be attributed to variation mapping to the location of the gene itself and likely reflect sequence differences in regulatory regions or content variation . Alternatively , the presence/absence of a transcript could reflect a strong trans-regulatory variant and a subset of the genes do exhibit trans-eQTL . This set of genes with expression in one parent and roughly 50% of RILs are expected based on previous studies of maize genome content variation and regulatory variation [68] . Many of the genes with segregation for the presence of transcripts exhibit other , unexpected , patterns of expression . These include genes that are expressed in both parents but a few RILs , genes expressed in neither parent but many of the RILs and other patterns . These segregation patterns are not expected to result from traditional single , gene segregation . We did not find evidence that there was segregation for the presence/absence of these genes within the genomic DNA of progeny . It is quite possible that many of these unexpected patterns of segregation for transcript presence reflect epigenetic or small-RNA based regulatory mechanisms . For instance , an example from tomato illustrates that a miRNA present in one of the parents can become detectably expressed in all the hybrids and their progeny [47] . In addition , there are examples of molecular dominance in siRNA levels and DNA methylation in Arabidopsis F1 plants [69]–[70] . It will be important to further understand the mechanisms that generate these unexpected patterns of segregation to understand the inheritance of traits in RIL populations . There is a growing appreciation for the qualitative variation among the genomes and transcriptomes of maize inbreds . Inbreds of maize can have substantial variation for gene content [6]–[7] , [59] , [71] . These inbreds can also have substantial variation for the presence of transcripts [29] , [34] . The F1 genotypes will contain the full set of genes found in both parents and generally tend to express this full set leading to a potential contribution to heterosis [72] . In this study , we showed that the RILs can also vary in transcriptome content relative to the parental genotypes . This leads to questions about the functional consequences of variation in transcriptome content . Many of the studies on genome content and variation in transcriptome content have found that the variable genes are under-represented for syntenic genes with functional annotations . Consistently , we found that only 36 of the 210 genes with unexpected patterns of segregation for expression were located in syntenic chromosomal positions . The variation for the presence of expression for these genes may directly impact phenotypes . The other 174 genes include a number of inserted sequences relative to gene order in other grass species . The maize genome is known to be littered with gene fragments that have been captured and mobilized by transposons [14] , [22]–[23] , [73] . In many cases , the presence of these gene fragments is variable among maize genotypes [14]–[15] , [74] and can contribute to novel transcripts [24] . Here we provide evidence that the presence/absence of transcripts from these gene fragments can act to modulate the expression level of the full-length parent gene . This suggests that some of the qualitative variation for gene fragment transcripts acts to provide a trans-acting regulator for the full-length gene and suggests a mechanism for the origin of selectable variation in expression level for single genes .
A maize IBM ( Intermated B73×Mo17 ) RIL population derived from the cross of the inbred lines B73 and Mo17 [50] was used to assess segregation of gene expression . At least 10 seedlings per genotype of 105 IBM RILs and their parents were planted in a single growth chamber . A randomized block design was employed with three replicates . The order of the flats within each block was rotated daily to minimize the effects of local environmental variation . Fourteen days after planting , at least 6 healthy seedlings were harvested and a 4 mm cubic tissue including the shoot apex were dissected and pooled for each genotype-replication combination . After separately grinding tissue from each genotype-replication pool in liquid nitrogen , RNA was extracted using the TRIzol and Qiagen RNeasy mini kit following the manufacturer's instructions . The three replicate RNA samples of each genotype were pooled with barcoding . RNA sequencing libraries were prepared and sequenced using the Illumina Hi-Seq2000 with 103–110 cycles . The resulting sequencing data were trimmed and aligned to the B73 reference genome v2 ( AGPv2 ) [2] by Data2Bio ( http://www . data2bio . com/ ) . The majority ( 69–80% ) of the trimmed reads were uniquely mapped and 94% of mapped reads were located in annotated gene regions . The uniquely-mapped reads were further analyzed for SNPs and read counts per genes in the RILs and their parents . RPKM values were determined using Cufflinks v0 . 9 . 3 ( http://cufflinks . cbcb . umd . edu/ ) based on the uniquely mapped reads of each genotype . The AGP v2 5b maize genome annotation was used as a reference , while maximum intron length = 60 , 000 bp and the quartile normalization option were employed . To establish a threshold for detectable expression , we conducted global permutation tests with 10 , 000 randomly selected non-genic fragments from B73 RNA-seq data [75] . We found the RPKMs were 0 . 055 , 1 . 03 , 2 . 02 and 5 . 41 as cutoffs for gene expression at different significant levels of FDR = 0 . 05 , 0 . 01 , 0 . 005 and 0 . 001 , respectively . For the initial analyses , a transcript presence/absence was assessed using a threshold of 0 . 055 RPKM . For the more stringent analysis of unexpected segregation patterns a threshold of 1 . 03 RPKM was employed and gene presence required values >1 . 03 and absence required a value of 0 . 0 . Intermediate values were not assigned presence or absence calls . The 22 , 242 genes expressed in more than 90% of the RILs and the parents were used to interrogate the global expression variation . The population mean and coefficient of variation of gene expression levels were summarized for the attributes of the RIL population , whereas the absolute value of log2 of the expression-level in B73 divided by the level in Mo17 was used for the expression fold change between B73 and Mo17 . The Kolmogorov-Smirnov test was applied to judge whether the expression levels of genes fit a normal distribution in the RIL population . The τ statistic , introduced by Bessarabova et al . [52] , was employed to distinguish between one-modal ( normal ) and bimodal distributions . We simulated 10 , 000 normal distribution data ( μ = 0 , σ = 1 ) , each containing 105 numbers , to obtain the global threshold of τ = 3 . 24 ( P = 0 . 01 ) . We treated the expression levels , which did not fit either normal or bimodal distribution , as unclassified distribution . The relationship between coefficient of variation and abs ( log2 ( B73/Mo17 ) ) , and the relationship between τ value and abs ( log2 ( B73/Mo17 ) ) of the variation of global gene expression were assessed by Pearson's product-moment correlation analysis in R ( http://www . R-project . org ) . Ten randomly selected genes with expression-level ( RPKM ) ranging from 0 . 05 to 2552 . 91 were selected to validate the expression profiling accuracy of RNA-seq by quantitative RT-PCR ( qRT-PCR ) using the same RNA samples as the ones used for RNA-seq . For qRT-PCR , cDNA samples ware amplified using the iQ SYBR Green Supermix on the CFX96 Real-Time PCR detection system ( Bio-Rad , Hercules , CA ) . Each PCR reaction contained 25 µl of reagent , consisting of 5 µl cDNA; 12 . 5 µl of the iQ SYBR Green Supermix; 2 . 5 µl of nuclease-free water; and 5 µl of the forward and reverse primers ( 1 µM stock ) . The qRT-PCR conditions included an initial incubation at 95°C for 3 min , followed by 40 cycles of 95°C for 10 sec , 58°C for 20 sec , and 72°C for 25 sec . To test the expression pattern of the paramutation-like genes , we examined gene expression in the shoot apex from 18 individuals from an F2 population derived from a cross between B73 and Mo17 . The F2 individuals , Mo17 and B73 were grown in a growth chamber using similar conditions as those used to obtain the RNA-seq data from the RIL population . RNA samples from the shoot apex were isolated from 2-week old seedlings and reverse-transcribed into the first strand cDNAs for the qRT-PCR quantification . Ten randomly selected paramutation-like genes were examined for the relative quantitation of expression level in the F2 individuals and their parents . qRT-PCR was performed with the SYBR Green master mix according to the manufacturer's instructions ( Applied Biosystems , Carlsbad , California ) . Three replicates were conducted to calculate the average and standard deviation of expression levels . The 2−ΔΔCT method was employed to calculate the relative quantitation of expression levels with the housekeeping gene Actin as the endogenous control and B73 as the reference genotype . To validate the unexpected expression patterns we conducted two experiments . In the first experiment , we replanted 10 IBM RIL genotypes , using the same growth conditions as used in the RNA-seq experiment , with 10 plants per genotype and sampled the shoot apices of the seedlings 14 days after planting . RNA was isolated from at least 6 healthy plants per genotype . In the second experiment , we tested the expression variation of genes with unexpected expression patterns in 18 individuals from an F2 population derived from a cross between B73 and Mo17 . A total of 55 genes with unexpected expression patterns were randomly selected for validation . RT-PCR was conducted using a Touchdown PCR program [76] . Two cycling phrases were set for the Touchdown PCR program: the TM reduced from 72°C to 62°C by 1°C every successive cycle in the first phrase with 10 cycles , while 25 other cycles were used for the amplification in the second phrase with TM = 62°C . Thus , 35 cycles were conducted . We also conducted genomic DNA PCR amplifications on the same RILs with the Touchdown PCR program on 8 randomly selected genes with unexpected expression patterns to check whether the extraordinary expression occurred only at the transcript level . The concentration of the template cDNA and DNA was 10 ng/µl for all the validations of RT-PCR and genomic PCR . All primer information can be found in Table S12 . To examine the expression patterns in hybrids of B73 and Mo17 for the paramutation-like genes , we dissected shoot apices from 10 plants from B73 , Mo17 and their reciprocal hybrids , isolated RNA and conducted RNA-seq . For this experiment , the plants were grown in the same growth chamber conditions used for the original RNA-seq experiment , using consistent protocols for sampling , library preparation , RNA-seq and analysis . For the analyses of attributes of genes with unexpected presence/absence expression patterns , we downloaded the gene family information of the whole B73 gene set from EnsemblPlants ( http://plants . ensembl . org/index . html ) . Gene family relationships were constructed through EnsemblCompara GeneTrees by using the phylogenetic approach [77] . The syntenic information of maize genes was obtained from the CoGe database ( http://genomevolution . org/CoGe/ ) . We annotated 20 Kb of flanking sequence for the genes with unexpected expression patterns ( Type I , Type II and Type IIIA and Type IIIB ) in 5 Kb windows as a fragment Bin by RepeatMasker ( http://repeatmasker . org ) . As controls , 210 genes were randomly selected and 10 , 000 permutations were conducted . Then , we annotated the adjacent fragments from 5 Kb upstream and downstream for all the FGS and summarized the number of all the different kinds of transposon-like sequences in the adjacent fragment of genes . Data2Bio ( http://www . data2bio . com/ ) identified 648 , 230 putative SNPs in 28 , 603 genes ( 72% of all maize genes ) using RNA-Seq reads from the RILs and their parents . High quality unique SNP markers with minimal missing data in the RILs were selected , grouped and integrated into chromosomes before constructing the genetic map . Maximum Likelihood Estimation with minimal threshold LOD score = 3 . 0 by JoinMap 4 . 0 [78] was employed to construct a high-resolution genetic map . The expression-levels of 22 , 242 genes were treated as expression traits ( e-traits ) for the global gene eQTL mapping . The genetic determinants controlling variation in e-traits were mapped via composite interval mapping [79]–[80] with a walking speed of 1 cM in the procedure of SRmapqtl and Zmapqtl of QTL cartographer [81] . A global permutation with 1000 randomly selected e-traits×1000 replicates were done as a representative null distribution of 1 , 000 , 000 maximum likelihood ratio test ( LRT ) statistics . A global permutation threshold as the significant cutoff of eQTL mapping was obtained at a significance level of 0 . 05 , giving a likelihood ratio test value of 19 . 23 , which corresponds to a Logarithm of Odds ( LOD ) score of 4 . 17 . The range with a 1 . 0 LOD drop on each side from the LOD peak point was selected as the confidence interval . If two adjacent peaks overlap in less than 10 cM , we treated them as one eQTL . A global permutation of randomly distribution of trans-eQTLs along the whole maize genome was performed to find the threshold of trans-eQTLs hotspots . One thousand of the maximum number of trans-eQTL scattering in 1 Mb genomic region of each permutation were obtained to compute the cutoff of hotspots . Further , we took gene density into account to rule out the gene number factor for the identification of trans-eQTL hotspots . For global trans-eQTLs hotspots , the cutoff ( #_trans-eQTLs/ ( Mb×#_genes ) ) was 1 . 25 . The GO enrichments and the pathway enrichments of the regulated genes by hotspots were conducted using BiNGO plugin [82] in Cytoscape [83] based on the annotation information from AgriGO [84] and MaizeCyc database [58] , respectively . The epistasis scan with all possible pairwise marker interactions for the genes with unexpected expression patterns was conducted with a generalized linear model . We employed an α-level of 0 . 05 ( P<2 . 1E-06 ) , which was adjusted by following the suggestion of dividing the α-level by the number of possible independent pairwise interactions among recombinant blocks [85] . We obtained genomic variation information between B73 and Mo17 from Springer et al . 2009 . The formula of CGH signal abundance of B73 and Mo17 of log2 ( Mo17/B73 ) were used to classify different CGH types [59] . The segments with a peak at log2 ( Mo17/B73 ) = 0 were simply classified as B = M , while the segments with a peak at log2 ( Mo17/B73 ) = 20 . 43 were classified as B73<Mo17_SNP . B = M_int represents segments with an intermediate value between 0 and 20 . 43 . Mo17>B73_CNV shows segments that are predicted to occur in more copies in Mo17 than in B73 . B>M_CNV indicates segments containing significantly more copies in B73 than in Mo17 , while B>M_int represents segments having intermediate more copies in B73 than in Mo17 . B>M_PAV shows segments present in B73 but absent in Mo17 . Of these genomic variants we mainly focused on CGH segments B>M_int , B>M_CNV , B>M_PAV and M>B_CNV for the relationship analyses between genomic variation and transcriptome variation in the maize IBM RIL population . First , we coordinated genes with CGH segments by coding scripts to compare the coordinates of genes ( according to the annotation of the maize reference genome AGPv2 ) with the CGH segments . Four main relationships could be obtained as genes entirely within CGH segments , genes intersecting CGH segments , genes in regions having multiple CGH segments , and other . Second , we filtered expressed genes and CGH segments . We limited the analysis to the expressed genes , which we defined as those displaying a normalized expression value ( RPKM ) of at least 1 . 03 ( corresponding to 21 reads per gene , FDR = 0 . 01 ) in more than 40% of the samples . Further , we considered the pair-wise datasets between genes and CNVs only if genes were expressed in at least 40 samples for each inferred genotype ( B73 and Mo17 ) in the RIL population . Finally , we conducted eQTL mapping of genes with CNVs nearby , for the inference of associations between structural variation and expression levels . The raw RNA-seq data on shoot apices of the IBM RIL population used in this study were submitted to NCBI's Sequence Read Archive ( SRA ) with accession number SRA055066 and will be released to public after approval of publication . The transcriptome profiling data were also deposited in MaizeGDB ( http://www . maizegdb . org/ ) . | Phenotypes are determined by the expression of genes , the environment , and the interaction of gene expression and the environment . However , a complete understanding of the inheritance of and genome-wide regulation of gene expression is lacking . One approach , called expression quantitative trait locus ( eQTL ) mapping provides the opportunity to examine the genome-wide inheritance and regulation of gene expression . In this paper , we conducted high-throughput sequencing of gene transcripts to examine gene expression in the shoot apex of a maize biparental mapping population . We quantified expression levels from 28 , 603 genes in the population and showed that the vast majority of genes exhibited the expected pattern of Mendelian inheritance . We genetically mapped the expression patterns and identified genomic regions associated with gene expression . Notably , we detected gene expression patterns that exhibited non-Mendelian inheritance . These included 145 genes that exhibited expression patterns in the progeny that were similar to only one of the parents and 210 genes with unexpected presence/absence expression patterns . The findings of non-Mendelian inheritance underscore the complexity of gene expression and provide a framework for understanding these complexities . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"plant",
"biology",
"population",
"genetics",
"crop",
"genetics",
"plant",
"science",
"cereals",
"crops",
"plant",
"genomics",
"population",
"biology",
"genetic",
"polymorphism",
"maize",
"gene",
"expression",
"plant",
"genetics",
"biology",
"agriculture",
"plant",
"b... | 2013 | Mendelian and Non-Mendelian Regulation of Gene Expression in Maize |
The evolution of bacterial pathogenicity , heavily influenced by horizontal gene transfer , provides new virulence factors and regulatory connections that alter bacterial phenotypes . Salmonella pathogenicity islands 1 and 2 ( SPI-1 and SPI-2 ) are chromosomal regions that were acquired at different evolutionary times and are essential for Salmonella virulence . In the intestine of mammalian hosts , Salmonella expresses the SPI-1 genes that mediate its invasion to the gut epithelium . Once inside the cells , Salmonella down-regulates the SPI-1 genes and induces the expression of the SPI-2 genes , which favor its intracellular replication . The mechanism by which the invasion machinery is deactivated following successful invasion of host cells is not known . Here , we show that the SPI-2 encoded transcriptional regulator SsrB , which positively controls SPI-2 , acts as a dual regulator that represses expression of SPI-1 during intracellular stages of infection . The mechanism of this SPI-1 repression by SsrB was direct and acts upon the hilD and hilA regulatory genes . The phenotypic effect of this molecular switch activity was a significant reduction in invasion ability of S . enterica serovar Typhimurium while promoting the expression of genes required for intracellular survival . During mouse infections , Salmonella mutants lacking SsrB had high levels of hilA ( SPI-1 ) transcriptional activity whereas introducing a constitutively active SsrB led to significant hilA repression . Thus , our results reveal a novel SsrB-mediated mechanism of transcriptional crosstalk between SPI-1 and SPI-2 that helps Salmonella transition to the intracellular lifestyle .
All organisms carefully regulate gene expression to ensure correct spatiotemporal deployment of gene products . For bacterial pathogens that reside in multiple niches , a mechanism to coordinate gene expression with environmental sensing is crucial for their ability to cause disease . This is achieved largely by two-component regulatory systems that sense external surroundings using a membrane sensor kinase that signals to a cytosolic response regulator that directs a transcriptional response [1] . In Salmonella , many of virulence genes required for infection are found in horizontally acquired pathogenicity islands [2] . Salmonella pathogenicity islands 1 and 2 ( SPI-1 and SPI-2 ) were acquired at different evolutionary times and have key roles in Salmonella virulence [3 , 4] . Both SPI-1 and SPI-2 encode a type III secretion system ( T3SS ) , effector proteins , chaperones , and transcriptional regulators that control the expression of the genes within each of the SPIs [3 , 5] . The SPI-1-encoded T3SS ( T3SS-1 ) and effector proteins mediate Salmonella invasion of host cells leading to gastroenteritis [3 , 4] . Following invasion , the genes within SPI-2 are required for Salmonella survival and replication within its intracellular niche , the Salmonella-containing vacuole ( SCV ) . The ability of Salmonella to replicate inside macrophages allows for dissemination , leading to systemic disease in susceptible hosts [3 , 4] . Consistent with their function , the SPI-1 genes are expressed when Salmonella is in the intestinal lumen or associated with the epithelium [6] . SPI-1 is also expressed in a subpopulation of bacteria that replicates in the cytosol of cultured epithelial cells [7] . The SPI-2 genes are mainly expressed when Salmonella is inside the SCV of epithelial cells and macrophages [7–11] . In vitro , SPI-1 genes are expressed when Salmonella is grown to early stationary phase in nutrient-rich lysogeny broth ( LB ) , whereas SPI-2 genes are expressed when Salmonella is grown to late stationary phase in LB or in acidic minimal media containing micromolar concentrations of phosphate and magnesium ions [12–14] . A transcriptional regulatory cascade comprised of HilD , HilA and InvF , positively controls the expression of the SPI-1 genes as well as several other genes outside this island that are required for Salmonella invasion of host cells [3 , 15–17] . When Salmonella is grown to late stationary phase in LB , HilD mediates transition of the gene expression program from SPI-1 to SPI-2 through activation of the SsrA-SsrB two-component system , a master regulator of SPI-2 genes [14] . In response to chemical cues detected inside host cells , the SsrA sensor kinase ( also called SpiR ) phosphorylates the SsrB response regulator leading to the activation of the genes found within SPI-2 and in other regions of the genome [3 , 18 , 19] . SsrB binds to a degenerate A+T-rich 18-bp palindrome sequence [20] , probably making few base contacts; however , the exact mechanism by which SsrB interacts with DNA may vary from gene to gene [21] . The mechanism by which the invasion machinery is repressed following invasion of host cells is not known . Here , we report that SsrB represses the expression of SPI-1 genes directly by acting on the hilD and hilA regulatory genes . Following invasion of macrophage cells SsrB represses expression of the invasion machinery encoded in the SPI-1 genes , while activating expression of the SPI-2 genes needed for intracellular survival . Consistent with this model , Salmonella mutants lacking SsrB had high levels of hilA transcriptional activity during mouse infections , whereas introducing a constitutively active SsrB led to significant hilA repression in vivo . Thus , our results reveal a regulatory switch activity for SsrB that helps Salmonella transition to the intracellular environment .
In a previous study we showed that SPI-1 and SPI-2 genes are expressed during early and late stationary phase , respectively , when S . Typhimurium is grown in LB [14] . Interestingly , the expression of SsrB during late stationary phase coincided with down-regulation of the SPI-1 regulator HilA [14] . To investigate the mechanisms controlling this regulation , we examined the chromosomal expression of InvF-FLAG by Western blot in a wild-type ( WT ) S . Typhimurium strain that constitutively expresses SsrB from the pK3-SsrB plasmid , or a strain containing the vector control pMPM-K3 . InvF is a SPI-1 regulator whose expression is dependent on HilA [3] . The chromosomal expression of SsrB-FLAG was also assessed as a control in the strain containing pMPM-K3 . As expected , in the presence of the vector pMPM-K3 the protein level of InvF-FLAG was maximal in early stationary phase and decreased during late stationary phase , whereas expression of SsrB-FLAG was induced only during late stationary phase ( Fig 1 ) . In contrast , in the presence of the pK3-SsrB plasmid InvF-FLAG was not detected at any of the time points tested ( Fig 1 ) , indicating that SsrB expression leads to InvF repression . To examine the broader impact of SsrB on SPI-1 , we determined the effect of SsrB on the effector secretion profile in WT S . Typhimurium grown in LB . Consistent with the results with InvF , in cells constitutively expressing SsrB there was reduced secretion of the SPI-1-encoded effectors SipA , SipB , SipC and SipD , as well as the flagellar protein FliC , in the culture supernatants ( Fig 2A ) . Similar results were obtained using a S . Typhimurium ΔSPI-2 mutant ( Fig 2A ) , indicating that the repressing effect of SsrB on the secretion of SipA-D and FliC proteins does not require any other SPI-2-encoded factor . Together , these results show that SsrB represses the expression of the SPI-1 and flagellar genes . Invasion of Salmonella into host cells requires the cellular functions encoded in both the SPI-1 and flagellar genes [3 , 22 , 23] . Thus , we used gentamicin protection assays to determine whether SsrB-mediated repression of the SPI-1 and flagellar genes had a phenotypic consequence on bacterial invasion . HeLa cells were infected with WT S . Typhimurium containing the pK3-SsrB plasmid or the pMPM-K3 vector and the number of intracellular bacteria was determined 1 h post-infection . S . Typhimurium ΔhilD and ΔflhDC mutants , lacking master positive regulators for the SPI-1 and flagellar genes , respectively , were used as controls . The constitutive expression of SsrB from pK3-SsrB resulted in a 500-fold reduction in invasion ( Fig 2B ) . As expected , the ΔhilD and ΔflhDC mutants also showed a very strong reduction in invasion ( Fig 2B ) . These results show that constitutive expression of SsrB negatively affects Salmonella invasion of HeLa cells , consistent with its ability to repress SPI-1 and flagellar genes . The SPI-1-encoded regulators HilD , HilA and InvF positively control the expression of the genes within this island in a cascade fashion , where HilD induces the expression of HilA and it , in turn , activates the expression of InvF [3 , 24] . To investigate how SsrB represses the SPI-1 genes , we analyzed the effect of constitutive SsrB expression on the transcription of hilD , hilA and invF , using cat transcriptional fusions . As controls for these assays the expression of sirA and csrA , which are found outside SPI-1 and encode known regulators of the SPI-1 genes , and ssaG , a SPI-2 gene whose expression is dependent on SsrB [3 , 24] , was also tested using cat transcriptional fusions . Constitutive expression of SsrB from pK3-SsrB nearly abolished the expression of the hilD-cat-364+88 , hilA-cat-410+446 and invF-cat fusions , in bacterial cultures grown in LB for 4 and 9 h , times representing early and late stationary phase of growth ( Fig 3A , 3B and 3C ) . In contrast , SsrB had a non-significant effect on the expression of the sirA-cat and csrA-cat fusions ( S1A and S1B Fig ) . SsrB induced the expression of the ssaG-cat fusion in the early stationary phase of growth , whereas in the presence of the pMPM-K3 vector its expression was only induced during late stationary phase ( Fig 3D ) . This is consistent with previous results indicating that overexpression of SsrB can activate the SPI-2 genes even in the absence of its cognate sensor kinase SsrA , while still requiring its phosphorylable Asp56 residue [18]; since small inorganic phosphate donors , such as acetyl phosphate , can also phosphorylate SsrB [25] . Together , these results demonstrate that SsrB represses the transcription of the SPI-1 regulatory genes hilD , hilA and invF . To determine whether SsrB directly or indirectly represses the expression of hilD , hilA , and invF , we analyzed the interaction of SsrB with the regulatory regions of these genes by electrophoretic mobility shift assays ( EMSAs ) . Full-length SsrB is unstable in solution , but the C-terminal DNA binding domain ( 6H-SsrBc ) is stable and can specifically bind to promoter regions of SsrB-regulated genes [18 , 25] . Therefore , purified 6H-SsrBc and the DNA fragments of each gene contained in the hilD- , hilA- and invF-cat fusions were used in these assays . 6H-SsrBc bound to the DNA fragments of hilD and hilA ( Fig 3E and 3F ) but did not bind to the DNA fragment of invF ( Fig 3G ) . As expected , 6H-SsrBc also shifted the DNA fragment of ssaG , which was used as a positive control ( Fig 3H ) but it did not shift those of the sirA or csrA negative controls ( S1C and S1D Fig ) . These results show that SsrB specifically binds to the regulatory regions of hilD and hilA . Previous work has identified a conserved yet flexible 18 bp palindrome sequence that defines the SsrB binding sequence based on a position-specific scoring matrix [20] . Scanning with this sequence ( Fig 4A ) identified two putative SsrB-binding sites in the regulatory region of hilD and nine within the hilA regulatory region . Interestingly , the two putative SsrB-binding sites near hilD are located in the promoter , whereas in hilA one putative SsrB-binding site is located upstream of the promoter , overlapping a HilD-binding site , and the others are located far upstream or downstream of the promoter ( Fig 4B ) . To determine whether SsrB represses hilD through these two putative SsrB-binding sites , three different cat transcriptional fusions were constructed , each with distinct 5’ and 3’ deletions of the hilD-cat-364+88 fusion that showed repression by SsrB ( Fig 4B ) . The fusions ( named according to the 5’ and 3’ positions of the hilD DNA fragment with respect to its transcriptional start site ) hilD-cat-108+88 , hilD-cat-48+88 and hilD-cat-37+6 were tested for CAT-specific activity in the presence of pK3-SsrB or the vector pMPM-K3 . Positive autoregulation of hilD is not essential for its expression [26] , therefore , the hilD-cat-48+88 and hilD-cat-37+6 fusions that lack the HilD-binding site upstream of hilD , were expected to be expressed . In the presence of pMPM-K3 , hilD-cat-108+88 reported expression levels similar to those from hilD-cat-364+88 ( compare Figs 5A and 3A ) , indicating that the cis-acting elements required for maximal expression of hilD are located between positions -108 to +88 . In contrast , the expression of hilD-cat-48+88 decreased by 50% relative to hilD-cat-108+88 ( Fig 5A and 5B ) , which is consistent with the reduction in hilD expression seen in the absence of autoregulation [26] . Interestingly , the hilD-cat-37+6 fusion that contains only the promoter of hilD was activated to similar levels as the hilD-cat-108+88 fusion ( Fig 5A and 5C ) , demonstrating that in the absence of negative regulatory sequences between positions +6 to +88 , the autoregulation is not required for maximal expression of hilD . Notably , the presence of pK3-SsrB significantly reduced the expression of hilD-cat-108+88 , hilD-cat-48+88 and hilD-cat-37+6 ( Fig 5A , 5B and 5C ) , indicating that SsrB negatively acts on the hilD promoter . EMSAs were performed to confirm that SsrB directly regulates the promoter of hilD . The hilD DNA fragments contained in hilD-cat-108+88 , hilD-cat-48+88 and hilD-cat-37+6 , shifted in the presence of increasing concentrations of 6H-SsrBc ( Fig 5D , 5E and 5F ) , indicating that SsrB binds to the promoter located between position -37 to +6 relative to the transcriptional start site of hilD , which is consistent with our bioinformatics analysis revealing two putative SsrB-binding sites on this region ( Fig 4B ) . These results show that SsrB binds to the promoter of hilD and thus would repress its transcription . To determine whether SsrB mediates repression of hilA at any of the SsrB-binding sites we predicted bioinformatically , four different hilA-cat transcriptional fusions were constructed that have 5’ or 3’ deletions ( or both ) with respect to the hilA-cat-410+446 fusion that showed repression by SsrB ( Fig 4B ) . The fusions ( named according to the 5’ and 3’ positions of the hilA DNA fragment with respect to its transcriptional start site ) hilA-cat-410+66 , hilA-cat-100+6 , hilA-cat-35+6 and hilA-cat-35+446 were tested for CAT-specific activity in the presence of pK3-SsrB or the pMPM-K3 vector . Previously , it was shown that sequences flanking the promoter repress hilA and in the absence of the sequence upstream or downstream of the promoter , hilA was expressed independently of HilD [27–29] . Therefore , hilA-cat-410+66 , hilA-cat-100+6 , hilA-cat-35+6 and hilA-cat-35+446 , which lack the repressing sequences , were expected to be expressed at high levels , regardless of whether they contain the HilD binding sites or not . As expected , in the presence of the pMPM-K3 vector , hilA-cat-410+66 , hilA-cat-100+6 and hilA-cat-35+6 were expressed at higher levels than hilA-cat-410+446 ( Fig 6A , 6B and 6C and Fig 3B ) . In contrast , the hilA-cat-35+446 fusion , which lacks the sequence upstream of the promoter including the HilD-binding sites had severely reduced activity ( Fig 6D ) . This suggests that the expression of hilA in the presence of the sequence downstream of the promoter , up to position +446 , requires HilD . Notably , the presence of pK3-SsrB reduced the expression of hilA-cat-410+66 and hilA-cat-100+6 , but it did not affect the activity of hilA-cat-35+6 and hilA-cat-35+446 ( Fig 6A , 6B , 6C and 6D ) , suggesting that SsrB mediates repression of hilA by acting on the region between -100 to -35 . Expression analysis of hilA-lux-740+35 and hilA-lux-36+446 transcriptional fusions further indicated that this -100 to -35 region is needed for the SsrB-mediated repression of hilA ( S2A and S2B Fig ) . To determine whether SsrB physically interacts with this region of hilA we used EMSAs with purified 6H-SsrBc . 6H-SsrBc shifted the hilA DNA fragments contained in hilA-cat-410+66 and hilA-cat-100+6 , but not those contained in hilA-cat-35+6 and hilA-cat-35+446 ( Fig 6E , 6F , 6G and 6H ) , indicating that SsrB binds between positions -100 to -35 . These results are consistent with our bioinformatics analysis that predicted a SsrB-binding site in this region , centered at position -70 , overlapping a HilD-binding site ( Fig 4B ) . To determine whether SsrB mediates direct repression of hilA at this site , we mutated this site in the hilA-cat-100+6 fusion by substituting five nucleotides within the predicted SsrB-binding site ( Fig 7 ) . The expression of the WT hilA-cat-100+6 and mutated hilA-cat-100+6 fusions was tested in WT S . Typhimurium containing pK3-SsrB or the vector control pMPM-K3 . Constitutive expression of SsrB from pK3-SsrB drastically reduced the expression of the WT hilA-cat-100+6 reporter but only slightly affected the activity of the mutated hilA-cat-100+6 fusion ( Fig 7A and 7C ) . Moreover , EMSAs showed that 6H-SsrBc binds to the hilA DNA fragment contained in WT hilA-cat-100+6 , but does not bind to the hilA-cat-100+6 fragment containing the mutated SsrB-binding site ( Fig 7E and 7F ) . Interestingly , the mutations we created within the hilA-cat-100+6 fusion also affected the regulation and binding of HilD on hilA ( S3A , S3B , S3C and S3D Fig ) . These results show that SsrB represses hilA by binding to the site centered at position -70 that overlaps a HilD-binding site , which suggested that SsrB inhibits the HilD-mediated expression of hilA . To test this , the expression of the WT hilA-cat-100+6 and mutated hilA-cat-100+6 fusions was tested in a S . Typhimurium ΔSPI-1 ΔrtsA ΔCthns triple mutant containing pK3-SsrB or the vector pMPM-K3 . This mutant lacks HilD , HilC , RtsA and the other transcriptional regulators encoded in SPI-1 , as well as the C-terminal region of H-NS . HilD , HilC and RtsA constitute a positive feed forward regulatory loop and each one can directly induce the expression of hilA [30]; on the other hand , in the absence of the C-terminal region of H-NS the expression of hilA is independent of HilD [26] . The presence of pK3-SsrB did not affect the HilD- , HilC- and RtsA-independent expression shown by the WT hilA-cat-100+6 and mutated hilA-cat-100+6 fusions in the ΔSPI-1 ΔrtsA ΔCthns mutant ( Fig 7B and 7D ) , which further indicates that SsrB inhibits the HilD-mediated expression of hilA . Taken together , these results strongly support that SsrB represses the expression of hilA by preventing HilD from binding . SsrB can also repress hilA through an indirect mechanism by negatively regulating the expression of hilD . Notably , the hilD and hilA promoter sequences contained in the hilD-cat-37+6 ( directly repressed by SsrB ) and hilA-cat-35+6 ( not repressed by SsrB ) fusions , respectively , are 65% identical ( S4 Fig ) ; thus , only 15 different positions between these sequences determine binding and thus negative regulation of SsrB on the hilD promoter , but not on the hilA promoter . Our results described above indicate that SsrB represses the expression of SPI-1 genes while activating expression of SPI-2 genes . In different in vitro SPI-2-inducing growth conditions that we have tested , invF was not de-repressed in the absence of SsrB ( S5A and S5B Fig ) , consistent with the results from a previous study [17] . Thus , detection of specific environmental cues could be required for the repression of SPI-1 by SsrB in physiological conditions , which could occur during Salmonella infection of hosts . SPI-1 and SPI-2 are known to be inversely regulated when Salmonella is within macrophages [9–11 , 31] , an environment where SsrB is active [3] . To explore whether SsrB is involved in this inverse regulation during intracellular stages of infection , we analyzed the expression of invF ( SPI-1 ) and ssaG ( SPI-2 ) in WT bacteria and in bacteria lacking SsrB following macrophage infection . For this , transcriptional fusions of invF ( SPI-1 ) and ssaG ( SPI-2 ) to the luciferase operon ( lux ) were constructed in the pCS26-Pac vector . A lux transcriptional fusion of hns , a gene constitutively expressed , was also constructed as a control . RAW264 . 7 macrophages were infected with WT S . Typhimurium or its isogenic ΔSPI-2 mutant carrying the invF-lux , ssaG-lux or hns-lux fusions . At specific time points after infection the macrophages were lysed and luminescence was measured and normalized to the number of viable intracellular bacteria . As expected , the intracellular replication of the WT strain increased over time whereas the ΔSPI-2 mutant decreased ( S6 Fig ) . The intracellular expression of invF-lux and ssaG-lux also changed as expected in the WT strain , where invF expression decreased fifteen-fold by the last time point and ssaG expression increased the same magnitude over the course of the infection ( Fig 8A and 8B ) . When comparing the expression levels of invF-lux between the WT strain and the ΔSPI-2 mutant , two distinct stages were identified . At 1 and 4 h post-infection , the invF-lux fusion showed similar expression levels in the WT strain and the ΔSPI-2 mutant , including a decrease in expression at 4 h ( Fig 8A ) . However , at later time points in the infection , invF-lux expression levels continued to decrease in the WT strain , by two to nine-fold , but not in the ΔSPI-2 mutant ( Fig 8A and 8C ) . This revealed SsrB-dependent repression of invF during intracellular infection . Furthermore , the hns-lux transcriptional fusion showed similar levels of intracellular expression in the WT and ΔSPI-2 strains at all time points of the infection ( S7 Fig ) . Thus , the differences in the intracellular expression levels shown by the invF-lux fusion in the WT strain and its derivative ΔSPI-2 mutant were not due to the different levels of intracellular bacteria at these time points . On the other hand , only background activity was detected for the ssaG-lux fusion in the ΔSPI-2 mutant ( Fig 8B ) , consistent with its expression being dependent on SsrB [3] . Interestingly , de-repression of the invF-lux intracellular expression in the ΔSPI-2 mutant coincided with the timing of induction of the ssaG-lux intracellular expression in the WT strain ( Fig 8A and 8B ) . As expected , the de-repression of the invF-lux intracellular expression was also evident in a ΔssrA and a ΔssrB mutant , whereas the expression of the hns-lux control fusion was similar in the WT strain and these two mutants ( S8 Fig ) , which indicates that both the SsrA sensor kinase and SsrB response regulator are required for intracellular repression of invF and that no other SPI-2-encoded factors are required . Together , these results show that SsrB simultaneously represses and induces the expression of invF and ssaG , respectively , inside macrophages ( Fig 8D ) . Therefore , our data support that SsrB is involved in a regulatory switch that helps to coordinate the intracellular reprogramming of Salmonella genes , by activating the genetic program required for intracellular survival while de-activating the genes involved in the now-completed invasion step of infection . To determine whether SsrB represses expression of SPI-1 during mouse infections , we tested the hilA-lux-740+350 transcriptional fusion in the WT S . Typhimurium strain , its isogenic ΔssrB mutant , and in the ΔssrB mutant complemented with a constitutive active SsrB variant in which aspartic acid 56 was replaced with glutamic acid . This SsrB D56E variant was expressed from the native ssrA promoter ( PssrA-ssrB D56E ) . C57BL/6 mice were orally gavaged with these strains and luminescence was quantified by in vivo imaging ever h for 6 h post-infection . Expression of the hilA-lux fusion was greater in the ΔssrB mutant than in the WT strain at the different times tested , which was evident by quantification of total abdominal luminescence ( Fig 9 ) . The presence of SsrB D56E reduced the expression of the hilA-lux fusion in the ΔssrB mutant ( Fig 9 ) . These results show that SsrB negatively regulates SPI-1 during the course of the intestinal infection of S . Typhimurium in a mouse model .
Salmonella has developed a complex regulatory network to express virulence genes in a highly coordinated manner within particular host niches . For example , when Salmonella is inside macrophages , it down-regulates the SPI-1 invasion machinery and flagellar-based motility genes that are required for host-cell invasion , whereas the expression of the SPI-2 genes required for intracellular survival and replication is activated [9–11 , 31] . Previously , the mechanism responsible for repressing the genes involved in invasion following successful invasion events was not known . Here , we show that this mechanism involves the SsrB response regulator , which had previously known roles in activating genes required for intracellular survival . Our data support a model in which SsrB acts as a key component of the molecular switch that helps Salmonella transition from an extracellular to an intracellular lifestyle ( Fig 10 ) . Interestingly , in a previous study it was demonstrated that SsrB , in its unphosphorylated form , drives a Salmonella lifestyle switch by relieving biofilm silencing [58] . Recent transcriptomics and proteomics data support that SsrB represses the expression of the SPI-1 and flagellar associated genes in in vitro SPI-2-inducing growth conditions [17 , 32] , and that it represses the flagellar genes when S . Typhimurium is inside macrophages [17] . In S . Typhi , a human-restricted serovar that causes systemic infections , the transcriptional regulator TviA represses the expression of the SPI-1 and flagellar genes and reduces macrophage pyroptosis [33–36]; pyroptosis and apoptosis are programmed cell death pathways stimulated by SPI-1 and flagellar gene products [37–39] . Interestingly , S . Typhimurium lacks the TviA regulator , which implied the existence of a different pathway in non-typhoidal serovars of Salmonella . The SsrB-mediated repression of the SPI-1 and flagellar genes in S . Typhimurium might be important in order to limit pyroptosis and apoptosis following infection by this serotype . Although we have not yet examined the impact of SsrB-mediated repression of invasion genes on these host cell pathways , the mechanism uncovered here may serve to limit damage to host cells as Salmonella establishes a stable intracellular niche . Our data strongly support a mechanism whereby SsrB represses the SPI-1 genes by directly acting on the hilD and hilA regulatory genes . The direct binding of SsrB to the promoter of hilD may be preventing RNA polymerase from binding to this region . In addition to reducing the levels of HilD , SsrB-binding to the sequence centered at position -70 of hilA , overlapping a HilD-binding site , inhibits the HilD-mediated expression of hilA . These findings provide further insight on the SsrB regulon , and demonstrate how SsrB can act as a negative transcriptional regulator , in addition to its well-known role as a transcriptional activator . Moreover , previous studies indicate that the regulation of the SPI-1 genes mostly involves the control of hilD at the post-transcriptional and post-translational level [3 , 24 , 40] . Our results reveal another pathway for the regulation of SPI-1 that involves repression of hilD and hilA at the transcriptional level . In Escherichia coli , the EnvZ-OmpR two-component system responds to osmotic stress signals [41] . The inverse regulation of the SPI-1 and SPI-2 genes by SsrB resembles the reciprocal control of ompC and ompF transcription by OmpR . OmpR is known to directly activate expression of ssrA-ssrB , and repress the expression of hilD [42 , 43] . In addition to OmpR , other regulators , such as SlyA and PhoP , also positively and negatively control the expression of SPI-2 and SPI-1 genes , respectively [3 , 11 , 32 , 43 , 44] . Notably , OmpR , SlyA and PhoP positively control the expression of SsrB [3] . Therefore , these regulators may provide additional input into the SsrB-dependent or independent mechanisms that inversely regulates the expression of the SPI-1 and SPI-2 genes within macrophages . In a previous study , we found that HilD mediates transcriptional crosstalk between SPI-1 and SPI-2 when S . Typhimurium is grown in LB , through growth-phase dependent activation of HilA and SsrB [14] . Here , we demonstrate that SsrB represses the expression of HilD and HilA , and thus the SPI-1 genes , revealing that the transcriptional communication between SPI-1 and SPI-2 is bi-directional . The degenerate palindromic sequence motif that SsrB recognizes on DNA [20] may make this response regulator particularly suited to dual-level control of gene expression . For example , the flexibility in the SsrB binding site may allow the bacterium to sample a wide array of new regulatory connections that can then be further optimized and selected by cis-regulatory evolution .
Animal experiments were conducted according to guidelines set by the Canadian Council on Animal Care , using protocols approved by the Animal Review Ethics Board at McMaster University under Animal Use Protocol #13-07-20 . Bacterial cultures were grown at 37ΔC in LB containing 1% tryptone , 0 . 5% yeast agar and 1% NaCl , pH 7 . 5; in N-minimal medium ( N-MM ) containing 5 mM KCl , 7 . 5 mM ( NH4 ) 2SO4 , 0 . 5 mM K2SO4 , 1mM KH2PO4 , 100 mM Tris-HCl ( pH 7 . 5 ) , 10 μM MgCl2 and 0 . 1% casamino acids; or in phosphate-carbon-nitrogen ( PCN ) minimal medium containing 80 mM MES ( pH 5 . 8 ) , 4 mM Tricine , 100 μM FeCl3 , 376 μM K2SO4 , 50 mM NaCl , 0 . 4 mM K2HPO4/KH2PO4 ( pH 5 . 8 ) , 0 . 4% glucose , 15 mM NH4Cl , 1 mM MgSO4 , 10 μM CaCl2 and micronutrients ( 10 nM Na2MoO4 , 10 nM Na2SeO3 , 4 nM H3BO3 , 300 nM CoCl2 , 100 nM CuSO4 , 800 nM MnCl2 , 1 nM ZnSO4 ) . When necessary , media were supplemented with ampicillin ( 200 μg ml-1 ) , kanamycin ( 30 μg ml-1 ) or streptomycin ( 100 μg ml-1 ) . Cultures in LB , N-MM or PCN media for chloramphenicol acetyltransferase ( CAT ) or Western blot assays were performed as described previously [12 , 14 , 45] . Briefly , overnight cultures of the Salmonella strains were sub-cultured ( 1:50 ) into 50 ml of fresh medium contained in 250 ml flaks , which were incubated at 37°C with shaking ( 200 r . p . m . ) in an Orbital shaker bath ( GYROMAX 902 , Amerex Instruments ) , during the indicated times . Bacterial strains used in this work are listed in Table 1 . Deletion of rtsA in S . Typhimurium SL1344 was performed by the λ Red recombinase system , as described previously [46] , using the primers shown in Table 2 , generating the strain DTM91 . P22 transduction was used to transfer the invF::3XFLAG-kan allele from strain DTM76 into S . Typhimurium SL1344 , generating the strain DTM85 , to transfer the ΔssrB::kan allele from the strain MJW112 into the strain DTM86 , generating the strain DTM87 , to transfer the ΔSPI-1::kan allele from the strain ΔSPI-1 into DTM92 , generating the strain DTM93 , to transfer the ΔCthns::kan allele from the strain DTM84 into the strain DTM94 , generating the strain DTM95 , to transfer the ΔssrB::kan allele from the strain 4/74 ΔssrB into S . Typhimurium SL1344 , generating the strain DTM97 , and to transfer the ΔssrA::kan allele from the strain 4/74 ΔssrA into S . Typhimurium SL1344 , generating the strain DTM98 . The kanamycin resistance cassette was excised from the strains DTM85 , ΔSPI-2::kan , JPTM30 , DTM91 , DTM93 , DTM95 , DTM97 and DTM98 , by using helper plasmid pCP20 expressing the FLP recombinase , as described previously [46] , generating the strains DTM86 , DTM89 , DTM90 , DTM92 , DTM94 , DTM96 , DTM99 and DTM100 , respectively . All mutant strains were verified by PCR amplification and sequencing . Plasmids and primers used in this work are listed in Tables 1 and 2 , respectively . To construct the plasmids containing the transcriptional fusions hilD-cat-108+88 , hilD-cat-48+88 , hilA-cat-410+66 , hilA-cat-100+6 , hilA-cat-100+6 Mut , hilA-cat-35+446 and fliC-cat , the respective segment of the regulatory region of hilD , hilA or fliC were amplified by PCR with the primer pairs hilD-108FW/hilDRHindIII ( rv ) , hilD-48FW/hilDRHindIII ( rv ) , hilA1FBam ( fw ) /hilAp+66H ( rv ) , hilA-100Ba ( fw ) /hilA+6Hind ( rv ) , hilA-100MutBamH ( fw ) /hilA+6Hind ( rv ) , hilAp+66FB ( fw ) /hilA2RHind ( rv ) or fliC-RVI-BH/fliC-FWI-Hd . The PCR products were digested with BamHI and HindIII restriction enzymes and then cloned into the BamHI and HindIII sites of the vector pKK232-8 , which carries a promotorless cat gene ( Amersham Pharmacia LKB Biotechnology ) , generating plasmids philD-cat-108+88 , philD-cat-48+88 , philA-cat-410+66 , philA-cat-100+6 , philA-cat-35+446 and pfliC-cat . To construct the plasmids containing the transcriptional fusions hilD-cat-37+6 and hilA-cat-35+6 , the complementary primers hilDPFBam ( fw ) and hilDPRHind ( rv ) or hilAPFBam ( fw ) and hilAPRHind ( rv ) , each at a final concentration of 50 μM , were annealed by heating them together at 94°C for 10 min and then slowly cooling to room temperature . The obtained double-strand products carried cohesive ends for their cloning into the BamHI and HindIII sites of the vector pKK232-8 , generating plasmids philD-cat-37+6 and philA-cat-35+6 . To construct the plasmids containing the transcriptional fusions invF-lux-306+231 , ssaG-lux-303+361 and hns-lux-967+73 , the respective segment of the regulatory region of invF , ssaG or hns were amplified by PCR with the primer pairs invF-luxR1/invF-luxF2 , ssaG-luxR1/ssaG-luxF2 or hns-luxR1/hns-luxF2 , respectively . The PCR products were digested with BamHI and XhoI enzymes and then cloned into the same restriction sites of the pCS26-Pac vector , which carries a promotorless lux operon [47] , generating the pinvF-lux , pssaG-lux and phns-lux plasmids . The hilA-lux-740+350 and hilA-lux-36+446 transcriptional fusions were constructed by replacing the em7 promoter in the pGEN-luxCDABE plasmid [48] ( Addgene plasmid # 44918 ) with the respective regulatory region of hilA . The regulatory region of hilA was amplified by PCR with the primer pairs EC30F/EC30R or EC76F/EC77R . The PCR products were digested with BamHI and SnaBI enzymes and then cloned into the same restriction sites of the pGEN-luxCDABE , generating the philA-lux-740+350 and philA-lux-36+446 plasmids . To construct the pK3-SsrB plasmid , the ssrB gene was amplified by PCR using the primer pair SRBF19-KpnI/ABR15-SacI and chromosomal DNA from the WT S . Typhimurium SL1344 as template . The PCR products were digested with KpnI and SacI restriction enzymes and then cloned into the vector pMPM-K3 [49] digested with the same restriction enzymes . The pK3-SsrB plasmid constitutively expresses SsrB from a lac promoter , since Salmonella and the vector pMPM-K3 lack the gene encoding LacI , the repressor of lac . The constitutively active SsrB variant ( PssrA-ssrB D56E ) was generated by cloning the ssrA promoter ( amplified with primers DTM17F/17 . 2R ) and ssrB coding sequence ( primers DTM17 . 1F/17R ) from S . Typhimurium SL1344 into pBluescript using SOE PCR . SDM was performed on this plasmid in pBluescript with primers DTM299F/299R to generate PssrA-ssrB ( D56E ) . This was subsequently subcloned into the low copy vector pWSK129 using the SalI and XbaI restriction sites . Protein secretion and Western blot assays were performed as we described previously [45] . Immunoblots were performed with anti-FLAG M2 ( Sigma ) or anti-DnaK ( StressGen ) monoclonal antibodies at 1:4 , 000 and 1:20 , 000 dilutions , respectively . Horseradish peroxidase-conjugated anti-mouse ( Pierce ) at a dilution of 1:10 , 000 was used as the secondary antibody . The CAT assays and protein quantification to calculate CAT specific activities were performed as previously described [50] . E . coli BL21/DE3 containing pK6-HSsrBc was grown in 200 ml of LB at 37°C with shaking . At an optical density ( OD600 ) of 0 . 6 , expression of 6H-SsrBc was induced by adding 0 . 1% L-arabinose and cultures were incubated for an additional 4 h . Bacterial cells were harvested by centrifugation at 4°C and the 6H-SsrBc protein was purified from pellet as previously described [25] . Maltose binding protein ( MBP ) -HilD was expressed in E . coli BL21/DE3 containing pMAL-HilD1 and purified by using an amylose column , as described previously [14] . Fragments of the regulatory regions of hilD , hilA , invF , ssaG , sirA and csrA were obtained by PCR amplification with the same primer pairs used to construct the respective transcriptional fusion to the cat reporter gene . PCR products were purified using the QIAquick PCR purification kit ( Qiagen ) . Each PCR product ( ≈100 ng ) was mixed with increasing concentrations of purified 6H-SsrBc in a binding buffer containing 10 mM Tris ( pH 7 . 5 ) , 50 mM KCl , 2 . 5% glycerol , 5 mM MgCl2 and 0 . 05% Nonidet P-40 , in a final volume of 20 μl . Protein-DNA binding reactions were incubated at room temperature for 20 min; then separated by electrophoresis in 6% non-denaturing acrylamide gels in 0 . 5 X Tris-borate-EDTA buffer , at room temperature . The DNA fragments were stained with ethidium bromide and visualized with an Alpha-Imager UV transilluminator ( Alpha Innotech Corp . ) . Gentamicin protection assays were performed as previously described [22] . HeLa ( human cervical adenocarcinoma epithelial ) cells ( ATCC ) were grown in high-glucose Dulbecco´s Modified Eagle Medium ( DMEM ) ( GIBCO 12100–046 ) supplemented with 10 mM sodium pyruvate solution ( SIGMA S8636 ) , 20 mM L-glutamine ( GIBCO 25030–081 ) and 10% ( v/v ) heat-inactivated fetal bovine serum ( ByProductos 13001 ) , at 37°C in a humidified atmosphere with 5% CO2 . HeLa cells were seeded 20 h prior to infection in 24-well tissue culture plates at 1 x 105 cells per well . Overnight Salmonella cultures were sub-cultured 1:33 in 20 ml of fresh LB and incubated at 37°C with shaking for 4 h . The sub-cultures were diluted ( 1:5 ) in LB to OD600 of 0 . 6 . At this point , 1 ml of each sub-culture was spun and resuspended in 1 ml of 1X PBS . Then , 10 μl of these bacterial suspensions were used to infect the HeLa cells at a multiplicity of infection ( MOI ) of 30:1 ( bacteria to eukaryotic cell ) for 10 min . Cells were then washed twice with pre-warmed 1X PBS and incubated for an additional 20 min with DMEM at 37°C . Following this incubation time , monolayers were incubated with DMEM containing 50 μg/ml gentamicin for 1 h to eliminate any extracellular bacteria . The media was then removed and the cells were lysed in 1 ml of 0 . 2% ( w/v ) sodium deoxycholate in 1 X PBS . The cell lysates and the initial starting inoculums were serial diluted and plated onto LB agar supplemented with streptomycin at 100 μg ml-1 . Overnight cultures of the Salmonella strains containing the hilA-lux transcriptional fusions were sub-cultured ( 1:50 ) into LB broth at 37°C until the cultures reached mid-exponential phase ( OD600 = 0 . 5 ) . The cultures were sub-cultured ( 1:50 ) again into LB in black 96-well polystyrene plates . Plates were incubated at 37°C with shaking , and luminescence and OD600 were measured every 30 min using the PerkinElmer Plate Reader . Luminescence was normalized to OD600 . To determine intracellular gene expression using the lux bioluminescent reporter , we performed infection assays using RAW264 . 7 murine macrophage-like cells ( ATCC ) , as described for the invasion assays with HeLa cells . The RAW264 . 7 cells were seeded at a density of 1 . 5 x 106 cells/plate in 100 mm x 20 mm culture dishes ( Corning 430167 ) and infected with the Salmonella strains carrying the lux-transcriptional fusions at an MOI of 10:1 ( bacteria to eukaryotic cell ) . Following gentamicin treatment , the cells were lysed at 1 , 4 , 8 , 12 , and 16 h post-infection in 600 μl of 0 . 2% ( w/v ) sodium deoxycholate in 1 X PBS . A 200 μl sample of the cell lysates was loaded in duplicate into a white 96 well assay plate with a clear flat bottom ( Corning 3610 ) and luminescence was measured using the GloMax-Multi Detection System ( Promega ) . Cell lysates were also plated and luminescence was normalized to bacterial CFUs . Replication was determined by enumerating the recovered CFUs at 4 , 8 , 12 , and 16 h post-infection . Fold-replication represents the CFUs recovered at 4 , 8 , 12 or 16 h relative to the CFUs at 1 h post-infection . One day prior to infection , C57BL/6 mice were orally gavaged with 20 mg of streptomycin and abdominal fur was removed using clippers and depilatory cream . The WT S . Typhimurium SL1344 strain and its isogenic ΔssrB and ΔssrB complemented with the pPssrA-ssrB ( D56E ) plasmid , each containing the hilA-lux-740+350 fusion , were grown overnight with shaking at 37°C in LB supplemented with 100 mg ml-1 ampicillin and 50 mg ml-1 kanamycin . Bacteria were washed twice in 0 . 1 M HEPES ( pH 8 ) + 0 . 9% NaCl and mice were orally gavaged with 1x108 CFUs . Following infection , mice were anaesthetized with 2% isoflourane carried in 2% oxygen and imaged dorsally in an IVIS Spectrum ( PerkinElmer ) . Grey scale and luminescent images were captured every hour for six hours . Total abdominal luminescence was quantified at each time point . Computational analyses were performed with the regulatory sequence analysis tools ( RSAT ) [51 , 52] . The position-specific scoring matrix ( PSSM ) for the DNA-binding consensus sequence of SsrB was generated using the conserved 18 bp palindrome sequence that SsrB is known to recognize [20] . The prediction of SsrB-binding sites in the regulatory regions of hilD and hilA was performed with the matrix-scan program and the PSSM we created , using a P-value of 1e-3 . Default parameters were used in these computational programs unless otherwise indicated . Data were analyzed with GraphPad Prism 5 . 0 software ( GraphPad Inc . , San Diego , CA ) using unpaired Student`s t-test . For in vivo bioluminescence analyses , data outliers were identified using the Grubbs test . One data point was identified as an outlier and was omitted from the analysis in the WT ( 1 h ) group . | Salmonella infect humans and a wide range of mammalian hosts . Successful infection requires the bacteria to sense their surroundings and regulate gene expression in a way that maximizes fitness in that particular environment . The two major lifestyles of Salmonella include extracellular stages and intracellular stages of host cell infection; however , the molecular mechanisms of how Salmonella transitions between these two lifestyles are not completely understood . Here we show that the transcriptional regulator SsrB functions in a dual capacity , activating genes required for intracellular survival while simultaneously repressing genes needed for extracellular stages of infection . Our data highlight how regulatory crosstalk is selective during infection , presumably because it helps facilitate rapid transitions in bacterial lifestyles that ultimately promote bacterial survival and replication . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"gene",
"regulation",
"pathogens",
"immunology",
"microbiology",
"plasmid",
"construction",
"dna",
"transcription",
"bacterial",
"diseases",
"regu... | 2017 | The transcriptional regulator SsrB is involved in a molecular switch controlling virulence lifestyles of Salmonella |
Autosomal dominant neovascular inflammatory vitreoretinopathy ( ADNIV ) is an autoimmune condition of the eye that sequentially mimics uveitis , retinitis pigmentosa , and proliferative diabetic retinopathy as it progresses to complete blindness . We identified two different missense mutations in the CAPN5 gene in three ADNIV kindreds . CAPN5 encodes calpain-5 , a calcium-activated cysteine protease that is expressed in retinal photoreceptor cells . Both mutations cause mislocalization from the cell membrane to the cytosol , and structural modeling reveals that both mutations lie within a calcium-sensitive domain near the active site . CAPN5 is only the second member of the large calpain gene family to cause a human Mendelian disorder , and this is the first report of a specific molecular cause for autoimmune eye disease . Further investigation of these mutations is likely to provide insight into the pathophysiologic mechanisms of common diseases ranging from autoimmune disorders to diabetic retinopathy .
Autosomal dominant neovascular inflammatory vitreoretinopathy ( ADNIV ) is a heritable autoimmune condition . It is characterized by various stages that mimic several much more common eye diseases , including: uveitis , retinitis pigmentosa , proliferative diabetic retinopathy and proliferative vitreoretinopathy [1] , [2] . Together , these diseases account for a significant fraction of visual morbidity and human blindness [3] , [4] . Identification of a gene that generates the varied pathological features of these common conditions could have a significant impact on the understanding and treatment of blindness [5] . Although there are numerous causative genes for retinitis pigmentosa , only a handful of genes have previously been associated with intraocular inflammation , neovascularization and fibrotic disease [6] . Because of its similarity to other common eye diseases , ADNIV patients are often misdiagnosed , unless the familial nature of their disease is recognized . Bennett and co-workers described the original ADNIV family , ADNIV-1 in this study [1] , in which the characteristic clinical findings were transmitted in an autosomal dominant fashion through eight-generations . The disease onset in this family varies between 10 and 30 years of age and the disease course can be divided into five stages , each lasting approximately ten years ( Figure 1 , Figure S1 ) [2] . In the first stage , ADNIV is clinically indistinguishable from an autoimmune , non-infectious uveitis [7] . Although the retina appears normal , an abnormality is detectable with electroretinography very early in the course of the disease . In the second stage , retinitis pigmentosa-like photoreceptor degeneration is apparent . In the third stage , retinal neovascularization develops , which is very similar to the pathologic angiogenesis of proliferative diabetic retinopathy [8] . In the fourth stage , intraocular fibrosis leads to retinal detachment , similar to that seen in proliferative vitreoretinopathy [9] . In the fifth stage , continued inflammation , intraocular hemorrhage , neovascular glaucoma , fibrosis and retinal detachment eventually lead to phthisis and complete blindness . There are no systemic features in this condition . This combination of overlapping clinical conditions is unusual and suggests that the disease-causing mutations may act through multiple pathways . Stone and co-workers previously mapped the genetic locus for ADNIV to chromosome 11q13 [10] . In this study , we identified two new ADNIV families , and these additional subjects provided an opportunity to refine the genetic interval and identify the causative gene . Two different mutations were identified among three ADNIV families in the gene encoding calpain-5 , an intracellular calcium-activated cysteine protease with an unknown physiological function .
The new ADNIV families displayed a phenotype very similar to the original ADNIV-1 pedigree . Specifically , affected members showed all of the previously reported clinical signs of the disease ( Figure 1 , Figure S1 ) including: non-infectious uveitis ( Figure 1A , 1B ) , early loss of the b-wave on electroretinography ( Figure 1C ) , pigmentary retinal degeneration ( Figure 1D , 1E ) , cystoid macular edema , ( Figure 1F ) , retinal and iris neovascularization , vitreous hemorrhage ( Figure 1H ) , epiretinal membrane formation , proliferative vitreoretinopathy ( Figure 1G ) , retinal detachment , cataract , neovascular glaucoma and ultimately phthisis and complete blindness ( Figure 1I ) [1] . Each pedigree was consistent with autosomal dominant inheritance with complete penetrance ( Figure 2 ) . There were sixty-one affected subjects in ADNIV-1 , seven in ADNIV-2 , and thirty-one in ADNIV-3 ( Figure 2 ) . Forty-two of these 99 affected individuals ( 42% ) were male . The clinical severity of the disease was indistinguishable between affected males and females . With the exception of psoriasis in one individual , there were no other systemic , autoimmune or inflammatory conditions present in any of the affected family members . Prior linkage analysis of the ADNIV-1 family mapped the disease-causing mutation to a 22-megabase ( chr11: 91 , 760 , 018–69 , 339 , 635 ) interval on chromosome 11q13 ( Figure 3A ) [10] . Genotyping of the ADNIV-2 and ADNIV-3 families with short tandem repeat polymorphisms was consistent with linkage to the same locus . Haplotype analysis was suggestive of an ancestral relationship between ADNIV-1 and ADNIV-2 . In addition , two affected individuals in the ADNIV-3 family were found to be recombinant within the disease interval , narrowing it to 6 . 5 megabases between D11S4139 and D11S1789 . High resolution SNP genotyping of ADNIV-1 and ADNIV-3 further reduced the interval to the 6 megabases between rs879380 and D11S1789 , a region harboring 86 known genes ( Figure 3A ) . Whole-exome sequencing was performed using DNA from two affected family members from the ADNIV-1 pedigree who were separated by seven meioses . Only one of the resultant sequence variations met the following four criteria: located within the 6 Mb ADNIV interval , shared by the two affected members of ADNIV-1 , not previously reported as a polymorphism , and nonsynonymous . This variant is a guanine to thymine nucleotide change ( c . 728G>T , p . Arg243Leu ) in exon 6 of the CAPN5 gene ( NM_004055 ) ( Figure 3B–3D ) . A combination of SSCP and Sanger sequencing of CAPN5 exon 6 verified that this mutation was present in all the affected members and none of the unaffected members of ADNIV-1 . The coding sequence of CAPN5 was then sequenced in affected members of the two other ADNIV pedigrees . All affected members of ADNIV-2 were found to harbor the same heterozygous variant ( c . 728G>T , p . Arg243Leu ) found in ADNIV-1 , supporting the suspected ancestral relationship between these two families ( Figure 3D ) . All affected members of the ADNIV-3 family , were found to harbor a heterozygous variant in the adjacent codon , a thymine to cytosine change ( c . 731T>C , p . Leu244Pro ) ( Figure 3E ) . Both of these putative disease-causing variants in exon 6 of CAPN5 were easily detectable by SSCP in ADNIV patients , but were absent from all unaffected adult members of the ADNIV families ( no asymptomatic minors were tested ) as well as 272 ethnically similar control individuals ( Figure 3F ) . None of the three variants were listed in the dbSNP or 1000 Genome databases . In addition , none of the variant alleles were found in the over 10 , 700 CAPN5 alleles sequenced in the NHLBI Exome Sequencing Project ( http://evs . gs . washington . edu/EVS/ ) Calpain-5 is an intracellular calcium-activated cysteine protease ( NP_004046 ) with evolutionarily conserved domains required for protease activity . Both ADNIV-causing mutations were found in exon 6 , which encodes a major part of the catalytic domain and contains two of the three catalytic residues that compose the active site ( Figure 4 ) . Modeling of secondary structure suggests that both the ADNIV-1/2 ( p . Arg243Leu ) and ADNIV-3 ( p . Leu244Pro ) mutations lie within a nearby alpha helical domain . The first of these mutations removes a charged residue while the second disrupts the putative helical structure ( Figure 4A ) . The amino acid sequence of CAPN5 exon 6 is highly conserved across vertebrate species ( Figure 4B ) . The catalytic residues show 100% conservation among CAPN5 orthologs . Interestingly , there is also 100% conservation of the Arg243 residue mutated in ADNIV-1/2 and 88% conservation of the Leu244 mutated in ADNIV-3 . This small evolutionary divergence at the latter codon is also quite conservative: methionine for leucine in the frog . In contrast , the disease-causing mutation at this codon introduces a new proline bend within a putative alpha helix . Previous comparisons of calpain-5 to its human paralogs demonstrated that it has diverged significantly and now belongs to its own subfamily with calpain-6 [11] . This divergence is also evident within exon 6 alone , where the calpain-5 catalytic domain shows relatively low homology to other human calpains ( Figure 4C ) . Each of the ADNIV mutant residues is conserved in four or fewer of 12 calpain paralogs , suggesting that the residues mutated in ADNIV are specifically important to calpain-5 function and may physiologically distinguish it from the other calpains . The PolyPhen2 sequence analysis program predicted both ADNIV mutations to have damaging effects on protein function ( 0 . 999 for Arg243Leu and 0 . 998 for Leu244Pro ) comparable to an active site Cys81Ser mutation ( 1 . 0 ) . The SIFT program predicted the Leu244Pro mutation to be comparably pathogenic ( 0 . 04 ) to Cys81Ser ( 0 . 03 ) but predicted the Arg243Leu mutation to be better tolerated ( 0 . 1 ) . To better examine the relationship of ADNIV mutations within the calpain-5 catalytic domain , homology modeling to calpain-2 ( m-calpain ) was used to generate a three-dimensional structure for calpain-5 ( Figure 4D ) [12] , [13] . Both mutations were outside the active site cleft and relatively far removed from the calcium-binding domains and the binding site of the endogenous inhibitor calpastatin . Interestingly , both the ADNIV-1/2 , and ADNIV-3 mutations fell into a region of low electron density , suggesting the presence of a flexible loop ( Figure 4E ) . In calpain-1 ( μ-calpain ) models , the homologous loop undergoes calcium-induced conformational changes that regulate the proximity of catalytic residues within the active site cleft [14] . This putative loop contains both ADNIV mutants and is highly conserved among all calpain-5 orthologs ( Figure 4B ) . We evaluated the CAPN5 transcript in human retinal tissue using RNA sequencing . The transcript was observed at a level of 4 . 63 fragments per kilobase of exon per million , which places it between the first quartile and the median level of expression for all transcripts observed in the retina . No significant splice variants were detected . Two antibodies against calpain-5 were used to determine whether calpain-5 protein could also be detected in human retinal tissue sections . Both antibodies showed strong calpain-5 expression in the photoreceptor cells ( Figure 5A , 5B ) . There was a punctate pattern of labeling over the nuclei and inner segments with less expression along the outer segments and outer plexiform layer . There was no significant expression in the nerve fiber layer , ganglion cell layer , inner nuclear layer , inner plexiform layer , or retinal pigment epithelium . The localization to the photoreceptor cells is consistent with both the early electrophysiologic abnormalities and the later photoreceptor degeneration seen in ADNIV patients . Intracellular compartmentalization is a key regulatory mechanism for calpains [15] , [16] , [17] . For example , mutations that disturb localized protein interactions with calpain-3 cause limb-girdle muscular dystrophy type 2A [17] , [18] . To determine the effect of the ADNIV-causing mutations on the intracellular compartmentalization of calpain-5 , HEK293T cells were transfected with normal and mutant CAPN5 constructs . A western blot with anti-myc antibody revealed a single protein species of the expected size for myc-tagged calpain-5 ( Figure 5C ) . Immunocytochemistry of HEK293T cells showed normal calpain-5 to be localized near the cell surface ( Figure 5E ) . In contrast , both ADNIV mutants were found largely within the cytoplasm ( Figure 5F and Figure S2 ) . This suggests that the ADNIV-causing mutations may alter a membrane binding property of the protein .
The calpains are an evolutionarily ancient family of calcium dependent intracellular proteases that utilize a cysteine residue in the active site to mediate limited proteolysis . The multifunctional calpains require careful regulation , since they target multiple intracellular proteins and pathways [15] , [16] , [17] , [19] , [20] . Their activity is regulated by intracellular calcium , lipid and protein interactions , subcellular localization , autocatalysis and inhibition by the endogenous peptide calpastatin [18] , [20] , [21] . There is no consensus amino acid sequence or structural motif that is targeted for cleavage by calpains , and as a result it is often difficult to identify the physiologic substrates of these enzymes , including calpain-5 [22] , [23] . Capn5 is expressed during nematode and mouse embryogenesis [24] , [25] . In adults , CAPN5 is highly expressed in the colon , kidney , liver , trachea , uterus , eye and brain [11] , [25] . Calpains have been implicated in the pathogenesis of a wide range of human diseases including cancer , multiple sclerosis , Alzheimer's disease , cataract , diabetes and muscular dystrophy [17] , [26] . Some polymorphisms in CAPN10 and CAPN5 have been shown to be risk factors for type II diabetes [27] , [28] . However , prior to this report , limb girdle muscular dystrophy ( LGMD ) type 2A was the only disease shown to be caused in a monogenic fashion by variations in a calpain's protein sequence [29] . The evidence that the two missense mutations we observed in CAPN5 are responsible for ADNIV is compelling . The gene lies within the critical region previously linked to the disease , and all living subjects in the study who are clinically affected were found to harbor a CAPN5 mutation in exon 6 . Each of these two mutations alters an amino acid in the catalytic domain that has been highly conserved throughout evolution . Neither of these mutations was found among any of the clinically unaffected members of the three kindreds we studied , or among thousands of normal individuals . There are interesting similarities and differences between calpain-associated LGMD and ADNIV . In both disorders , the affected cells ( skeletal muscle fibers and photoreceptor cells ) experience large changes in membrane potential and intracellular calcium concentration as part of their normal behavior . In both disorders , the non-mutant calpain molecules display functionally critical subcellular localization [17] , [18] , [20] , [30] . A subset of LGMD is associated with leukocyte infiltration into the tissue [31] , and all cases of ADNIV are marked by severe intraocular inflammation . The differences in these diseases are also noteworthy . Calpain-associated LGMD is inherited in a recessive fashion and appears to result from loss of calpain-3 function [29] . In contrast , ADNIV is inherited in an autosomal dominant fashion and is caused by missense mutations near the active site . Although these mutations could cause disease through haploinsufficiency , it seems more likely that they result in a gain of function of calpain-5 that causes harm to the photoreceptor cells . Capn5 knockout mice have no observable phenotype [32] , and several human neurological disorders have been associated with excess calpain activity [17] , [33] , including photoreceptor degeneration [34] . A gain of function mechanism for ADNIV is also supported by the unusual inflammation and neovascularization associated with the disease . There are dozens of monogenic disorders that cause the apoptotic death of photoreceptor cells without causing severe intraocular inflammation or neovascularization of the retina . The latter is much more typical of proliferative diabetic retinopathy than it is heritable photoreceptor disease [8] . It is easier to imagine an unregulated or mislocalized calpain promiscuously activating different signaling pathways , or being released into the extracellular space after photoreceptor death and causing inappropriate angiogenesis and leukocyte recruitment , than it is to imagine a 50% reduction of such a protein causing these dramatic complications . Whether caused by a gain or loss of function of CAPN5 , it is likely that the further elucidation of the pathogenic mechanism of ADNIV will provide important new insight into some of the most important causes of irreversible human blindness: autoimmune uveitis , retinitis pigmentosa , proliferative vitreoretinopathy and diabetic retinopathy . The latter condition alone is responsible for as much as 17% of blindness in some regions of the world [4] . The fact that an amino acid change in a single protein can lead to such a phenotype raises the possibility that a common , therapeutically accessible pathway may be shared among these conditions that could be targeted with drugs , antibodies or gene transfer approaches . It is possible that variations in the structure or expression of CAPN5 cause or modify some of these common disorders and this hypothesis will be important to test in future studies . However , given the extreme phenotypic heterogeneity of these disorders , it will be important to study a large number of subjects in such an experiment , to subdivide the patient cohorts into clinically well-characterized groups , and to screen an equal number of ethnically matched controls for each of these groups .
Phenotypic ascertainment of the pedigrees included complete ocular examination as previously described [1] . Following genotyping of all adult members of the three ADNIV pedigrees with short tandem repeat polymorphisms within the original disease interval , three members of ADNIV-1 and three members of ADNIV-3 were also genotyped with an Affymetrix GeneChip Human Mapping 50K Array . Exome sequencing was performed using NimbleGen's SeqCap EZ Human Exome v2 . 0 capture and paired-end ( 2×100 ) sequencing on an Illumina HiSeq 2000 instrument at Otogenetics ( Norcross , GA ) . The putative disease-causing mutations in CAPN5 were evaluated in the ADNIV kindreds and controls using Sanger sequencing and single-strand conformational polymorphism analysis ( SSCP ) [35] . Sequence analysis of human retinal cDNA was performed using the Illumina HiSeq 2000 instrument . Primary and secondary structure protein alignments and trees were created with Geneious Pro 5 . 4 . 6 ( http://www . geneiouspro . com ) . Yasara Structure ( version 11 . 3 . 2 ) was used to generate a homology model of human calpain-5 using calpain-2 ( m-calpain ) structures ( pdb id: 3BOW , 1DF0 , 1U5I , 1KFU ) as templates [12] , [13] . Sequence alignment with the templates was first used to build a backbone model for aligned residues followed by loop modeling and side chain optimization using a combination of steepest descent and simulated annealing minimization . The top ranking of the 20 models generated was used as the homology model of calpain-5 . The above steps were automated using Yasara's hm_build macro ( http://www . yasara . org ) . Another homology model was generated using the Phyre server ( version 0 . 2 ) ( http://www . sbg . bio . ic . ac . uk/~phyre ) , which showed good agreement with the Yasara model for the domain folds and the active site region . The putative alpha-helix region in the Yasara model also formed a helix in the best Phyre model although it was positioned further away from the calpastatin binding site region than the Yasara model . PolyPhen2 ( http://genetics . bwh . harvard . edu/pph2/ ) and SIFT ( http://sift . jcvi . org/ ) sequence analysis software were used to predict the functional effect of mutations . RNA sequence analysis was performed by extracting RNA from the retina of a human eye donor using an RNeasy kit from Qiagen ( Valencia , CA ) according to the manufacturer's instructions , preparing the sequencing library using the Illumina ( San Diego , CA ) RNA TruSeq sample preparation kit , and sequencing the latter on the Illumina HiSeq 2000 instrument at the Hudson Alpha Institute in Huntsville Alabama . The resulting sequence data were mapped using TopHat [36] and analyzed using Cufflinks [37] . The retinal expression of CAPN5 was compared to the expression of all other genes expressed in the retina . Donor eyes were received from Iowa Lions Eye Bank ( Iowa City , IA ) . Tissue was fixed in 4% paraformaldehyde solution diluted in 10 mM phosphate buffered saline ( PBS ) , pH 7 . 4 , and 7 µm sections underwent immunohistochemistry ( IHC ) using a polyclonal anti-calpain 5 primary antibody ( Santa Cruz Biotechnology , Inc . , Santa Cruz , CA ) , AlexaFluor 488 donkey anti-rabbit secondary antibody ( Invitrogen , Carlsbad , CA ) and 4′ , 6-diamidino-2-phenylindole ( DAPI; Invitrogen ) . Images were captured using an Olympus BX41 microscope equipped with fluorescent filters and the SPOT Advanced software package . HEK293T cells ( ATCC , Manassas , VA ) were transfected with normal and mutant CAPN5 pCMV6-Entry vector plasmids using Turbofectin 8 . 0 ( Origene ) transfection reagent according to the manufacturer's instructions . Cells were incubated for 48 hours post transfection . For immunocytochemistry , cells were blocked using 5% bovine serum albumin ( Amresco , Solon , OH ) diluted in PBS with 0 . 1% Triton X-100 . The polyclonal primary antibody , anti-Myc-tag , was diluted in PBS at 1∶500 and applied to the cells . Alexa Fluor 488 donkey anti-rabbit secondary , at concentration 10 µg/mL , and 0 . 0001 µg/mL of the counterstain , 4′ , 6-diamidino-2-phenylindole ( DAPI ) ( both from Molecular Probes , Eugene , OR ) , were applied to the cells , Images were captured using a Zeiss LSM 710 equipped with Zen2009 software ( Zeiss , New York , NY ) . | We care for several families with an inherited form of autoimmune inflammation inside the eye . The patients also develop bleeding , scar tissue , and eventually blindness . Using advanced gene analysis methods , we discovered the cause of this disease is gene mutations in the CAPN5 gene . This gene makes a protein , calpain-5 , which belongs to a family of calcium-activated enzymes that slice other proteins inside cells . Calpain-5 is expressed in the retina , and the disease mutations alter its location inside the cell . Future studies to understand how this protein causes inflammation and bleeding inside the eye will help us develop treatments for this condition and more common eye diseases with inflammation and bleeding . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"ophthalmology",
"genetics",
"and",
"genomics",
"clinical",
"genetics"
] | 2012 | Calpain-5 Mutations Cause Autoimmune Uveitis, Retinal Neovascularization, and Photoreceptor Degeneration |
Dynamic regulation of leukocyte population size and activation state is crucial for an effective immune response . In malaria , Plasmodium parasites elicit robust host expansion of macrophages and monocytes , but the underlying mechanisms remain unclear . Here we show that myeloid expansion during P . chabaudi infection is dependent upon both CD4+ T cells and the cytokine Macrophage Colony Stimulating Factor ( MCSF ) . Single-cell RNA-Seq analysis on antigen-experienced T cells revealed robust expression of Csf1 , the gene encoding MCSF , in a sub-population of CD4+ T cells with distinct transcriptional and surface phenotypes . Selective deletion of Csf1 in CD4+ cells during P . chabaudi infection diminished proliferation and activation of certain myeloid subsets , most notably lymph node-resident CD169+ macrophages , and resulted in increased parasite burden and impaired recovery of infected mice . Depletion of CD169+ macrophages during infection also led to increased parasitemia and significant host mortality , confirming a previously unappreciated role for these cells in control of P . chabaudi . This work establishes the CD4+ T cell as a physiologically relevant source of MCSF in vivo; probes the complexity of the CD4+ T cell response during type 1 infection; and delineates a novel mechanism by which T helper cells regulate myeloid cells to limit growth of a blood-borne intracellular pathogen .
During infection , specific immune subsets must proliferate robustly to generate a population of effector cells large enough to contain the microbial threat . A well-characterized example is the dramatic expansion of antigen-specific T lymphocytes , whose numbers may increase over a thousand-fold during the course of an immune response [1] . Myeloid cells also undergo expansion in a number of infections [2–7] . The trafficking and recruitment of monocytes into infected tissues have been defined in detail [8]; however , the mechanisms controlling myeloid proliferation during infection are much less well understood . Under homeostatic conditions the survival and renewal of the mononuclear phagocyte lineage , which includes monocytes and macrophages , is controlled primarily by the cytokine Macrophage Colony Stimulating Factor ( MCSF ) [9 , 10] . In addition , MCSF can activate myeloid cells in vitro [9] . But the extent to which MCSF also regulates macrophage and monocyte proliferation and activation under inflammatory conditions is not clearly established , in part because the grave baseline defects of mice genetically deficient in this cytokine have complicated such analysis [11] . Infection with protozoan parasites of the genus Plasmodium results in a dramatic expansion of monocytes and macrophages that has long been considered a hallmark of malaria disease in humans and other mammalian hosts [12–15] . In mouse models employing rodent-adapted parasites , myeloid expansion has been shown to involve IL-27-dependent proliferation of hematopoietic stem cells in the bone marrow [16] and interferon gamma ( IFN-γ ) -dependent mobilization of multipotent myeloid progenitor cells into the spleen [5 , 17] , where they can give rise to monocytes and , presumably , macrophages . However , the cells and cytokines that regulate differentiation and proliferation downstream of these early progenitor stages remain undefined . Recent work has demonstrated that tissue-resident macrophages can proliferate in situ during helminth infection through a process requiring the type 2 cytokine interleukin-4 ( IL-4 ) [6 , 7] . These findings raise the question of whether macrophages and monocytes undergo local expansion in response to type 1 pathogens such as Plasmodium , and if so , what factors regulate this process . In this work , we investigate the causes and consequences of myeloid proliferation and activation during infection with P . chabaudi . We find that MCSF derived from multiple sources drives proliferation of macrophages and monocytes in infected mice; moreover , the full expansion and activation of certain subsets specifically requires MCSF derived from circulating CD4+ T cells , which have not previously been demonstrated to produce this cytokine in a physiological context . We measure the inducible upregulation of Csf1 in antigen-experienced CD4+ T cells from infected mice , and show that CD4+ T cell-derived MCSF is important for control of parasitemia and recovery of host health late in infection , coinciding with the kinetics of maximal myeloid expansion . Finally , we demonstrate a previously unappreciated role for CD169+ macrophages , which are diminished in mice lacking MCSF production in CD4+ T cells , in restriction of P . chabaudi parasite burden and host survival . Thus , this study establishes a new physiological source of MCSF in vivo , and delineates a novel mechanism by which CD4+ T cells regulate the myeloid compartment to control a blood-borne intracellular infection .
In the P . chabaudi blood-stage model of malaria , parasitemia peaks approximately 7 days post-infection ( d . p . i . ) , after which it is rapidly controlled to low levels ( <5% of red blood cells infected ) ( Fig 1A , black line ) . For this study , we divided the infection conceptually into two phases: the acute phase , during which parasitemia peaks , and the resolution phase , from approximately 10–30 d . p . i . , after acute parasitemia has been controlled but before the infection has been cleared to subpatent levels . It has long been observed that myeloid cells expand in number and frequency during the blood stage of Plasmodium infection [3 , 12–14] , and previous studies demonstrate that phagocytic cells , presumed to include macrophages , are involved in control of parasitemia during the acute phase of infection [18 , 19] . However , in the P . chabaudi model , myeloid expansion does not reach its peak until the resolution phase , i . e . approximately two weeks post-infection , well after acute parasitemia has been controlled [3 , 5] ( Fig 1A , red line , and 1B ) . Additionally , macrophage surface activation markers remain elevated for days after control of acute infection [20] . Therefore , we considered the hypothesis that macrophages might also be important for limiting parasitemia during the resolution phase . To test whether this is the case , we depleted phagocytic cells in P . chabaudi-infected mice 14 d . p . i . A small recrudescence typically occurs around this time ( Fig 1A ) . We first examined the efficiency of depletion for a number of myeloid subsets in the blood and spleen , including classical monocytes ( CMs ) , defined by high expression of Ly6C; nonclassical ( Ly6Clo ) monocytes ( NCMs ) ; and red pulp macrophages ( RPMs ) , the most abundant population of splenic macrophages [21] ( gating strategy , S1 Fig ) . Flow cytometric analysis of myeloid populations in the spleen and blood following treatment revealed efficient depletion of RPMs and monocytes , whereas conventional and plasmacytoid dendritic cells were partially depleted and granulocytes actually expanded ( Fig 1C ) . We did not observe alterations in non-myeloid cell frequencies in treated mice . Strikingly , mice depleted of phagocytes late in infection experienced a rapid resurgence of parasitemia , rising as high as 60% ( Fig 1D ) , accompanied by significant mortality ( Fig 1E ) . These results demonstrate that phagocytic cells , most likely macrophages and/or monocytes , remain critical for control of parasitemia during the resolution phase of infection , coinciding with the kinetics of their maximum expansion and sustained activation . Depletion of myeloid cells could affect parasitemia directly , e . g . through loss of phagocytic and microbicidal capacity , or indirectly through effects on adaptive cells such as T cells . To search for possible effects on T cell activation , we performed intracellular cytokine staining for IFN-γ in CD4+ T cells following late phagocyte depletion in infected mice . However , we could not detect IFN-γ protein in T cells from either control or myeloid-depleted mice at this late timepoint , consistent with previous reports showing that cytokine production peaks approximately 6 d . p . i . and is virtually undetectable by two weeks [22–24] . It is possible that late depletion of myeloid cells results in defects in T cell functions other than production of IFN-γ and IL-2; however , we note that dendritic cell activation , antigen presentation , and activation of adaptive responses occur as early as 5–7 d . p . i . in the mouse model [24–26] , and robust B and T cell responses are well-established by 14 d . p . i . [27–29] . Taken together , these findings make it less likely that the observed effects on parasite control are due to impacts on antigen-presenting cells and disruption of adaptive responses . Instead , we favor the hypothesis that late phagocyte depletion crucially targets macrophages and/or monocytes , which are required during the resolution phase to phagocytose and clear infected red blood cells . The observed increase in macrophage and monocyte numbers in infected mice is due at least in part to proliferation and recruitment of progenitor cells from the bone marrow [16 , 17] , but it also might involve local proliferation of monocytes and/or macrophages in the tissues , as has been documented in helminth infection [6] . To test whether differentiated myeloid cells proliferate locally during P . chabaudi infection , we assessed levels of the nuclear proliferation marker Ki67 in myeloid subsets from infected spleens . Whereas splenic RPMs , CMs , and NCMs from naïve mice largely lacked this marker , we detected significant Ki67 expression in all three subsets 14 d . p . i . with P . chabaudi ( Fig 2A ) . Similarly , significant fractions of RPMs , CMs and NCMs from infected , but not naïve , mice incorporated the injected thymidine analog 5-ethynyl-2’-deoxyuridine ( EdU ) , indicating active DNA synthesis ( Fig 2B ) . Thus , myeloid expansion during malaria infection involves local proliferation of differentiated , tissue-resident cells . We next investigated factors that might influence this proliferation of monocytes and macrophages during P . chabaudi infection . Among other subsets , we considered a role for CD4+ T helper cells , whose importance in controlling blood-stage Plasmodium has been suggested in humans [30–32] and demonstrated in mice [33–37]; infected mice depleted of CD4+ T cells exhibit defects in parasite control reminiscent of those observed after late macrophage depletion [35 , 28 , 37 , 38] ( Fig 1D ) . Accordingly , we found that depletion of CD4+ T cells 4 d . p . i . disrupted the expansion of splenic RPMs and NCMs in infected mice , although CMs were unaffected ( Fig 2C ) . The effect of CD4+ T cell depletion was not due to increased apoptosis of myeloid cells , since the frequency of apoptotic monocytes and macrophages was not increased in splenocytes from T cell-depleted mice ( S2 Fig ) . We proceeded to examine mechanisms by which CD4+ T cells might drive this myeloid expansion . Much research on the role of CD4+ T cells in malaria has focused on their abundant production of IFN-γ[23 , 39 , 24] , and indeed , infected Ifng-/- mice have fewer splenic macrophages than wild-type mice during the acute phase of infection [39] . Nevertheless , several observations led us to consider the hypothesis that T cells might regulate myeloid cells through multiple mechanisms , in addition to production of IFN-γ . First , it was not clear to us how IFN-γ might promote proliferation of differentiated macrophages and monocytes; in fact , IFN-γ is generally considered to be anti-proliferative for these cell types [40] , although it may stimulate proliferation of hematopoietic stem cells [5 , 16 , 41] . Second , one study using the related parasite P . yoelii found that protective macrophage populations were intact in infected mice lacking IFN-γ [19] , suggesting the presence of additional mechanisms governing their activity . Thus , to identify additional T cell-dependent factors that might regulate myeloid cells during infection , we performed transcriptome analysis on activated CD4+ T cells sorted from infected mice 6 d . p . i . , when production of many cytokines peaks [22–24] . Lacking tools to detect antigen-specific cells , we used the integrins CD11a and CD49d as proxy markers for antigen-experienced cells , an approach that has been validated in this and other animal models [42 , 43] ( gating strategy , S3A Fig ) . Unexpectedly , we detected significant expression of the gene Csf1 , encoding MCSF , in antigen-experienced ( CD11a+ CD49d+ ) CD4+ T cells from infected mice , but not bulk CD4+ T cells from naive mice ( Fig 2D ) . Using quantitative RT-PCR , we confirmed that Csf1 was upregulated in antigen-experienced CD4+ cells from infected mice , but not in antigen-naive cells from infected mice or in any CD4+ T cells from naive mice , regardless of antigen exposure ( Fig 2E ) . The predominant transcript expressed was variant 3 , which encodes a soluble form of the protein that can be secreted either as a glycoprotein or a proteoglycan , depending on differential proteolysis during export [10] . Smaller amounts of variant 2 , encoding a membrane-bound form of the cytokine , were also detected ( Fig 2F ) . Intracellular cytokine staining for MCSF has not been reported to our knowledge , and our attempts to measure MCSF protein in CD4+ T cells by flow cytometry were not successful . However , by performing ELISA on supernatants from antigen-experienced CD4+ T cells sorted from infected mice 6 d . p . i . and restimulated in culture , we confirmed the expression of MCSF protein in this population ( Fig 2G ) . Simultaneously sorted TCRβ- cells cultured in parallel did not produce significant amounts of MCSF , making it unlikely that MCSF production in sorted CD4+ T cell populations actually comes from contaminating myeloid cells ( Fig 2G ) . In contrast to IFN-γ , which peaks around 6 d . p . i . and is virtually undetectable after two weeks [22–24] , MCSF levels continued to increase in the plasma of infected mice for two weeks post-infection ( Fig 2H ) , a time frame coincident with the peak of myeloid expansion ( Fig 1B ) , consistent with the hypothesis that this cytokine might support the observed proliferation of myeloid cells at later timepoints . During homeostasis , MCSF is predominantly produced by endothelial and stromal cells; under inflammatory conditions , it can also be made by activated monocytes and macrophages themselves [9 , 10] . Several early publications described Csf1 expression in cultured CD4+ T cells stimulated extensively in vitro [44–47] . In addition , one study has reported detection of lymphocyte-associated MCSF in human lymph node tumor biopsies [48] , and two others found evidence for Csf1 production in decidual T cells during pregnancy [49 , 50] . Despite these examples , T cells are not generally considered to be a biologically relevant source of MCSF . Studies of baseline expression in an MCSF reporter mouse did not detect MCSF production in T cells [51] , and to our knowledge , neither production nor a physiological role for MCSF derived from circulating T cells has ever been demonstrated in vivo . In order to characterize the nature of Csf1 expression in T cells , we performed single-cell transcriptional analysis on antigen-experienced CD4+ T cells from infected mice ( S1 Table ) . Although MCSF levels in plasma increase through day 14 ( Fig 2H ) , likely arising from multiple systemic sources , Csf1 transcript was most readily detected by qRT-PCR in blood CD4+ T cells 6 d . p . i . ; therefore , this timepoint was chosen for further RNA-Seq analysis . Moderate to abundant levels of Csf1 transcript were detected in 37% of the cells analyzed ( 13 out of 35 ) , whereas no Csf1 was detected in the remaining cells ( Fig 3A ) . Single-cell sequencing results must be interpreted with caution , as transcript levels may fluctuate significantly within a cell over time due to the stochastic nature of gene expression [52]; in addition , technical limitations of the technique introduce stochasticity in transcript detection [53] . Nevertheless , the absence of Csf1 transcript in a majority of CD4+ T cells leads us to favor the hypothesis that Csf1 expression is restricted to a subset of activated cells , in which it is strongly upregulated . T helper cells are traditionally classified into distinct lineages , each with a defining transcription factor and cytokine profile [54] . We examined whether Csf1+ CD4+ T cells expressed hallmarks of any of the canonical T helper subsets . At the time-point sampled ( 6 d . p . i . ) , the CD4+ T cell response to P . chabaudi consists primarily of Th1-polarized cells [24 , 28] . Consistent with this , we found that nearly all Csf1+ cells also expressed Tbx21 , encoding the Th1 lineage-defining transcription factor TBET , and all expressed the canonical Th1 cytokine IFN-γ . Most Csf1+ cells also expressed Il10 , which is commonly co-expressed with IFN-γ in Plasmodium-exposed subjects [55–57] ( Fig 3B ) . Expression of the Th2 transcription factor Gata3 was significantly lower in Csf1+ T cells than in Csf1- cells ( Fig 3C ) ; we did not detect the Th2 cytokines Il4 , Il5 , or Il13 , or the Th17 markers Rorc or Il17 , in any cell , regardless of Csf1 expression . Thus , Csf1+ cells express signature genes of Th1 cells , but not Th2 or Th17 cells . Next we were interested in whether Csf1+ T cells had a transcriptional signature that distinguished them from Csf1- cells . Although the vast majority of transcripts detected did not vary significantly between Csf1- and Csf1+ cells , we identified a cluster of approximately 400 differentially expressed genes , representing 2 . 7% of detected transcripts ( Fig 3D and S2 Table ) . This cluster included genes for several cell surface receptors that we subsequently examined by flow cytometry , choosing markers based on the availability of commercial antibodies as well as an expression pattern that would yield distinct populations for further transcriptional analysis . Ccr2 , which encodes a chemokine receptor , and Entpd1 , encoding the ectonuclease CD39 , were both upregulated in Csf1+ cells relative to Csf1- cells ( Fig 3E ) . Therefore we sorted CCR2hi CD39hi antigen-experienced CD4+ T cells from infected mice and used qRT-PCR to compare their Csf1 expression with CCR2- CD39- cells ( gating strategy , S2B Fig ) . Whereas Csf1 transcript was nearly undetectable in both antigen-naive cells and antigen-experienced CCR2- CD39- cells , it was readily detected in antigen-experienced CCR2hi CD39hi cells ( Fig 3F ) . Virtually no CCR2hi CD39hi cells were observed within the antigen-naïve CD4+ T cell population , excluding the possibility that Csf1 expression is linked to CCR2/CD39 expression independently of antigen experience . These results validate our RNA-Seq data at the level of protein expression and suggest that Csf1 is expressed by a subset of CD4+ T cells enriched for a distinct surface phenotype . We did not observe Csf1 expression in CD4+ T cells cultured ex vivo with either Th1- or Th2-polarizing cytokines ( Fig 3G ) , even upon restimulation ( S4 Fig ) , suggesting that additional signals provided in the context of in vivo infection are required to upregulate Csf1 . Robust induction of the canonical cytokines Ifng and Il4 in these Th1 and Th2 cultures , respectively , served as a control to confirm effective polarization ( Fig 3G ) . Altogether , our data indicate that while Csf1-producing CD4+ T cells share most features with canonical Th1 cells , they possess some distinct transcriptional and surface characteristics that may be indicative of a specialized T helper phenotype or , alternatively , a particular activation state within the Th1 subset . After detecting inducible expression of MCSF transcript and protein in CD4+ T cells , we turned to the question of whether this cytokine indeed promoted myeloid expansion in the context of Plasmodium infection by administering an MCSF-blocking antibody systemically from 3 to 13 d . p . i . We chose this time frame to immediately precede the time when myeloid numbers peak ( day 14; Fig 1B ) as well as to overlap with the observed rise of MCSF levels in the plasma ( Fig 2H ) , keeping in mind that that local production of MCSF in the tissues might well increase before elevated systemic MCSF levels could be measured [9 , 10] . After the blockade , mice treated with α-MCSF had significantly fewer splenic RPMs and NCMs on day 14 than mice treated with an irrelevant isotype control antibody ( Fig 4A ) , demonstrating a role for systemic MCSF in myeloid expansion in the spleen during P . chabaudi infection . Numbers of blood monocytes were not significantly affected by MCSF blockade ( S5 Fig ) . Since macrophages are critical for control of parasitemia during the resolution phase of infection ( Fig 1D ) and disruption of MCSF signaling reduces macrophage numbers during this window ( Fig 4A ) , we hypothesized that MCSF blockade would also result in increased parasite burden . Indeed , although it did not alter parasitemia during the acute phase , MCSF blockade significantly increased parasite recrudescence during the resolution phase of infection ( Fig 4B and 4C ) , coincident with the peak of myeloid expansion in control mice . Additionally , mice treated with α-MCSF exhibited poor recovery from infection-induced weight loss , relative to mice treated with an isotype control antibody ( Fig 4D ) . These data indicate that MCSF-driven myeloid proliferation is important for controlling parasite replication and limiting host morbidity . The fact that MCSF blockade did not affect acute parasitemia suggests that pre-existing macrophages and/or monocytes are sufficient to control the initial peak of infection , whereas expanded numbers of myeloid cells are required to suppress parasitemia as the infection persists into the resolution phase . In the same vein , systemic phagocyte depletion ( Fig 1C and 1D ) results in a more severe phenotype than MCSF blockade ( Fig 4B and 4C ) , which does not completely ablate the myeloid compartment but only reduces its numbers ( Fig 4A ) . To test directly whether MCSF derived from CD4+ T cells is important for proliferation of myeloid cells and restriction of parasite burden , we generated mice that inducibly delete Csf1 specifically in CD4+ cells ( Cd4::CreERT2; Csf1fl/fl , referred to hereafter as Csf1ΔCD4 ) . To induce Csf1 deletion , Csf1ΔCD4 mice and Csf1fl/fl littermate controls were fed tamoxifen chow beginning one month prior to infection and continuing through the duration of each experiment; disruption of Csf1 expression in CD4+ T cells was confirmed by RT-qPCR after infection ( Fig 4E ) . For comparison , we also generated and infected Ubc::CreERT2; Csf1fl/fl mice ( referred to as Csf1ΔUbc ) , which systemically delete Csf1 upon tamoxifen treatment through expression of Cre recombinase under control of the ubiquitin promoter . After one month of tamoxifen treatment , monocyte and RPM numbers were significantly diminished in the spleens of naïve Csf1ΔUbc mice , consistent with the established role of systemic MCSF in maintenance of tissue-resident myeloid cells [10] . In contrast , Csf1ΔCD4 mice were unaffected , indicating that CD4+ T cell-derived MCSF is not required to maintain myeloid cell numbers at baseline ( S6 Fig ) . Consistent with the effects of MCSF blockade ( Fig 4A–4D ) , Csf1ΔUbc mice exhibited significantly higher recrudescent parasitemia ( Fig 5A ) and lost more weight ( Fig 5B ) than control Csf1fl/fl mice . Importantly , selective deletion of Csf1 in CD4+ cells also resulted in significantly higher parasite burdens ( Fig 5C ) and delayed recovery from weight loss ( Fig 5D ) during the resolution phase of infection , directly demonstrating a role for MCSF derived from CD4+ cells in control of Plasmodium infection . In addition to ubiquitous strong expression on CD4+ T cells , CD4 is expressed on a fraction of murine dendritic cells [58] and thymic macrophages [59]; however , given the lack of reported Csf1 expression in these specific myeloid populations ( www . immgen . org ) and their relatively low representation within their respective leukocyte subsets , we consider it unlikely that they are responsible for the phenotype observed in Csf1ΔCD4 mice . Moreover , previous characterization of the Cd4::CreERT2 transgenic mouse line revealed high specificity for peripheral CD4+ T cells , with little to no recombination of floxed genes in CD11b+ myeloid cells despite reported expression of CD4 [60] . Thus we conclude that MCSF from CD4+ T cells contributes to control of parasite burden and host recovery during the resolution phase of P . chabaudi infection . We next quantified myeloid populations in the spleens of infected Csf1ΔCD4 mice 14 d . p . i . In addition to coinciding with the peak of myeloid expansion in wild-type mice ( Fig 1B ) , this timepoint falls just prior to the greatest observed differences in parasitemia between Csf1fl/fl and Csf1ΔCD4 mice ( Fig 5C ) ; by examining myeloid cells at this time , we hoped to uncover differences that would lie upstream of the divergent parasitemias observed on days 15–20 , and to avoid examining phenotypes that might arise due to differences in parasite burden , rather than directly resulting from CD4-specific deletion of Csf1 . Having observed increased parasitemia in mice lacking CD4+ T cell-specific expression of MCSF , we expected to find decreased myeloid expansion in the spleens of these mice . Surprisingly , however , numbers of splenic RPMs , CMs , and NCMs were intact in infected Csf1ΔCD4 mice ( Fig 5E ) . Combined with the baseline defects in mice that systemically delete Csf1 ( S6 Fig ) as well as the MCSF blockade results ( Fig 4A ) , these data indicate that MCSF from other sources , but not from CD4+ T cells , is required to stimulate splenic myeloid expansion during P . chabaudi infection . We therefore searched for alternative mechanisms by which CD4+ T cell-derived MCSF might promote parasite restriction . First , we hypothesized that T cell-derived MCSF might support proliferation of myeloid subsets other than those we had previously examined . Consistent with this hypothesis , we measured diminished numbers of NCMs in the blood of infected Csf1ΔCD4 mice relative to Csf1fl/fl controls 14 d . p . i . ( Fig 5F ) . The defect in NCM numbers was milder in Csf1ΔCD4 mice than in Csf1ΔUbc mice , indicating that MCSF derived both from CD4+ T cells and T cell-independent sources contributes to expansion of blood monocytes during infection ( Fig 5F ) . Additionally , we tested the hypothesis that MCSF from CD4+ T cells might promote activation of myeloid cells , as MCSF has been reported to do in vitro [9] . To accomplish this , we measured levels of activation markers on splenic myeloid subsets in infected mice . Interestingly , there was a trend toward lower expression of MHCII on RPMs and CMs from infected Csf1ΔCD4 mice compared to infected Csf1 fl/fl controls , while NCMs exhibited lower levels of CD40 ( Fig 5G ) . However , these effects were modest and not statistically significant , similar to the effects of CD4+ cell-derived MCSF on monocyte numbers ( Fig 5F ) , leading us to examine additional myeloid populations for differences that might explain the poor control of parasitemia observed in Csf1ΔCD4 mice . The CD169+ macrophages of the spleen and lymph nodes have been shown to play critical roles in the capture and clearance of blood- and lymph-borne pathogens and other antigens [61 , 62] , but they are difficult to analyze by flow cytometry due to their fragility [63] . We therefore performed immunofluorescence microscopy to examine the effects of CD4+ T cell-derived MCSF on CD169+ macrophage populations , including splenic marginal metallophillic macrophages ( MMMs ) and the subcapsular sinus macrophages ( SCSMs ) in the lymph nodes . Consistent with previous reports that show near-complete disappearance of CD169+ macrophages in the spleens of mice infected with P . chabaudi [64 , 65] , we detected few CD169+ macrophages in the spleens of either Csf1 fl/fl or Csf1ΔCD4 mice 14 d . p . i . However , in lymph nodes , there were significant differences in the distribution of CD169+ macrophages between Csf1 fl/fl and Csf1ΔCD4 mice . In the absence of a straightforward method such as flow cytometry to assess absolute numbers of CD169+ macrophages [63] , we quantified the fraction of each lymph node perimeter that was lined with CD169+ cells ( Fig 6A and 6B ) . Strikingly , in Csf1ΔCD4 mice , a significantly smaller fraction of each lymph node was lined with CD169+ macrophages , indicating a role for CD4+ T cell-derived MCSF in supporting the survival or expansion of these cells ( Fig 6B ) . To further distinguish whether CD4+ T cell-derived MCSF affects proliferation of lymph node macrophages , we also assessed the fraction of CD169+ macrophages that were Ki67+ by immunofluorescence microscopy in lymph node sections from Csf1 fl/fl and Csf1ΔCD4 mice ( S7 Fig ) . Quantification of Ki67+ cells revealed significantly fewer proliferating macrophages in lymph nodes from Csf1ΔCD4 mice ( Fig 6C ) . Thus , MCSF from CD4+ T cells is important for promoting proliferation of CD169+ macrophages in the lymph nodes of infected mice . CD169+ macrophages have not previously been examined for involvement in control of P . chabaudi infection , but a recent study demonstrated that systemic depletion of CD169+ macrophages increased tissue sequestration of parasites , morbidity , and mortality in a model of experimental cerebral malaria ( ECM ) employing the pathogen P . berghei ANKA in ECM-resistant Balb/C mice . In light of this study and the immunofluorescence data , we tested whether CD169+ macrophages are important for control of parasitemia during the resolution phase of P . chabaudi infection in B6 mice , when MCSF production by T cells is most critical for restriction . We infected transgenic CD169+/DTR mice , which express the diphtheria toxin ( DT ) receptor under control of the CD169 promoter [66] . Treatment of these mice with DT results in efficient depletion of splenic MMMs ( [66 , 67] and S8 Fig ) and lymph node SCSMs [67 , 68] . Mice depleted of CD169+ cells 12 d . p . i . developed significantly higher parasitemia ( Fig 6D ) , weight loss ( Fig 6E ) , and mortality ( Fig 6F ) relative to controls treated with a catalytically inactive point mutant ( DT*Glu ) , indicating a role for these macrophages in control of P . chabaudi . To exclude off-target effects of DT treatment , we examined frequencies of additional myeloid subsets . In addition to MMMs and SCSMs , DT treatment in CD169+/DTR mice also resulted in depletion of bone marrow macrophages , which express CD169 [69] ( Fig 5G ) ; however , using a separate transgenic mouse model , in which the DT receptor is expressed under control of the Lyz2 promoter ( Lyz2Cre/Cre; Rosa26::STOPfl/fl::DTR ) [70] , we found that depletion of bone marrow macrophages , as confirmed by flow cytometry ( Fig 5H ) , in itself had no effect on parasitemia ( Fig 5I ) . In addition , although DT treatment diminished RPM frequencies in CD169+/DTR mice ( Fig 5G ) , we have previously shown that RPMs are not required for control of P . chabaudi [71] . Therefore , these results are most consistent with a model in which CD4+ T cell-derived MCSF promotes the survival or proliferation of CD169+ SCSMs , which contribute critically to parasite control and host survival . Together with flow cytometry analyses ( Figs 4A , 5E and 5F ) , the above experiments reveal that multiple sources of MCSF drive myeloid expansion during P . chabaudi infection . T cell-independent sources are most important for proliferation of the splenic myeloid subsets examined , whereas T cell-derived MCSF is critical for maintenance and proliferation of CD169+ macrophages in the lymph nodes , as well as contributing modestly to expansion and activation of some additional subsets in the blood and spleen . In addition , by comparing the phenotype of mice depleted of CD4+ T cells ( Fig 2B ) with mice simply lacking Csf1 expression in CD4+ T cells ( Fig 5E ) , we conclude that CD4+ T cells promote myeloid expansion through multiple mechanisms , only one of which is production of MCSF ( Fig 7 ) .
Previous publications have detected production of MCSF by cultured CD4+ T cells and in the special case of by decidual CD4+ T cells during pregnancy , but production by circulating T cells in vivo has never been demonstrated to our knowledge , and no physiological role has ever been assigned to CD4+ T cell-derived MCSF . In recent years , a number of tools have become available that have proved useful for dissecting this phenomenon in vivo . Here , using conditional knockout mice and sensitive methods for cell surface and transcriptional analysis , we have identified an important role for MCSF derived from CD4+ T cells in control of blood-stage P . chabaudi . This experimental malaria model has long been used to simulate uncomplicated infection with P . falciparum , the principal etiological agent of a human disease that caused 214 million new illnesses and ~438 , 000 deaths in 2015 [72] . Thus , in addition to elucidating a novel mechanism through which T cells regulate myeloid cells to restrict infection , this work provides insight into the immune correlates of protection from a devastating pathogen for which a successful vaccine has thus far proved elusive . Because CD4+ T cells are such central mediators of myeloid cell activation , it makes sense that can they also regulate myeloid cell numbers by promoting proliferation and/or survival . Indeed , T cells likely stimulate expansion of the myeloid compartment through multiple mechanisms , depending on the context . During helminth infection , local IL-4-dependent proliferation of macrophages in the tissues has been shown to require an intact Rag locus , indicating a role for B and/or T lymphocytes [6 , 7]; it is plausible that in this model , T cells drive macrophage proliferation through production of IL-4 . Further , T cells can produce Granulocyte-Macrophage CSF , which has been shown to regulate absolute leukocyte numbers in mice infected with P . chabaudi [3] . To these mechanisms we now propose to add the inducible production of MCSF , which we have shown stimulates not only expansion , but also activation of myeloid cells . We demonstrated that selective deletion of Csf1 in CD4+ T cells during infection results in significant reductions in the abundance and proliferation of CD169+ macrophages in the lymph nodes , as well as trends towards decreased numbers of some myeloid cells ( blood NCMs ) and diminished activation of other subsets ( splenic RPMs , NCMs , and CMs ) . An alternative interpretation of our immunofluorescence data is that disruption of CD4+ T cell-derived MCSF results in diminished CD169 expression , rather than loss of CD169+ macrophages; however , in the absence of additional surface markers to positively identify SCSMs in lymph nodes , this possibility is difficult to test experimentally . It is not yet clear whether or how the more modest effects of CD4+ T cell-specific Csf1 deletion on monocyte numbers and macrophage/monocyte activation contribute to the observed increases in parasitemia and morbidity in Csf1ΔCD4 mice . However , our experiments in mice depleted of CD169+ cells clearly show an important role for this subset in control of Plasmodium parasitemia and host survival , suggesting that a primary function of CD4+ T cell-dependent MCSF during P . chabaudi infection is to support these cells . In contrast to some other macrophage populations , including the lymph node medullary macrophages and splenic RPMs , lymph node SCSMs are critically dependent on MCSF for their survival and proliferation [73]; perhaps this dependency makes them particularly vulnerable to disruption of a single source of MCSF . CD4+ T cells are certainly not the only source of MCSF during P . chabaudi infection . Indeed , depletion of CD4+ T cells did not decrease plasma concentrations of MCSF in infected mice , indicating multiple redundant producers of this cytokine . However , our data demonstrate that CD4+ T cell-derived MCSF does play a nonredundant role in sustaining SCSM abundance and limiting recrudescent parasitemia . This may reflect a requirement for localized production of MCSF , which exists in the body not only as a soluble glycoprotein , but also in membrane-bound and proteoglycan forms , the latter of which contributes to circulating MCSF levels but may also preferentially accumulate in extracellular matrix [10] . We hypothesize that antigen-experienced CD4+ T cells in the blood and tissues may deliver MCSF directly to monocytes and macrophages , perhaps in conjunction with other signals , providing a stimulus that cannot be replaced by MCSF derived from other sources . Consistent with this hypothesis , T cells have been observed to interact closely with SCSMs in the lymph nodes of both naïve mice and those infected with Toxoplasma , an apicomplexan parasite related to Plasmodium [74 , 75] . Our data do not rule out the possibility that CD169+ macrophages other than SCSMs are the relevant population for control of parasitemia; for example , it may be that despite their small numbers , the MMMs that persist in infected spleens are nonetheless critical for restriction . In addition , it remains possible that CD4+ T cell-derived MCSF contributes to parasite control through yet another alternative mechanism , such as modulation of a myeloid population that we have not examined here . Having discovered a population of activated CD4+ T cells that inducibly expresses Csf1 , we must consider how these cells fit into the established T helper paradigm . Although recent demonstrations of T cell plasticity have begun to blur the lines between different Th subsets [76] , we maintain that some key elements of a T cell lineage can be defined: individual Th lineages generally have unique master transcription factors , canonical cytokines , and chemokine receptors , and their phenotype is stable and self-reinforcing [77] . In this case , robust expression of Tbx21 and overall transcriptional similarity to Th1 cells , which constitute the majority of CD4+ T cells at this stage of Plasmodium infection , support the hypothesis that Csf1-producing cells represent a specialized subclass of Th1 cells . On the other hand , we note that in a transcriptional profiling study of Th1 , Th2 , and Tfh cells polarized in vitro , the subsets differed significantly in expression of only 300–400 genes , similar to the number of genes that were differentially expressed between Csf1+ and Csf1- cells in our analysis [78] . Further , we detected a number of transcription factors that were differentially expressed in Csf1+ compared to Csf1- T cells ( S2 Table ) ; it may be that one or more of these acts in concert with TBET to exert a distinct transcriptional program in Csf1+ cells . Indeed , one of the differentially expressed transcription factors in our dataset , Bhlhe40 , was recently shown to drive GM-CSF expression in a fraction of Th1 and Th17 cells during experimental autoimmune encephalitis [79–81] . Previous studies on the gene expression profiles of CD4+ T cells polarized in vitro detected Csf1 transcript in Th2 cells [82] , and found no defect in Csf1 expression in Bhlhe40-deficient CD4+ T cells [80]; however , given that we detected little Csf1 expression in cells polarized in vitro relative to those isolated from infected mice , we hypothesize that these previous gene expression studies lack a robust positive control for Csf1 expression , and instead are measuring relatively low levels of transcript that may not be physiologically meaningful . Monocytosis is a common feature of malaria and several other chronic infections , such as tuberculosis and leishmaniasis [83 , 84] , but its causes and significance have not been well characterized . This study elucidates one mechanism of myeloid proliferation and activation during malaria and demonstrates that expansion of macrophages and monocytes is critical for ongoing restriction of Plasmodium parasite growth . It will be of interest to determine whether a similar mechanism operates in other infectious settings in which CD4+ T cells cooperate with macrophages to limit microbial burden , and to dissect the signals required for MCSF induction . In addition , it remains to be seen whether CD4+ T cells inducibly produce MCSF in sterile disease settings , such as tumor microenvironments , in which macrophages and other myeloid cells play important roles .
All animal experiments were conducted with the approval of the UCSF Institutional Animal Care and Use Committee ( Protocol AN086391-03C ) in accordance with the “Guide for the Care and Use of Laboratory Animals , ” published by the National Research Council and endorsed by the NIH Office of Laboratory Animal Welfare . Mice were housed on a twelve hour light-dark cycle under specific pathogen free conditions . C57Bl/6 mice were from the National Cancer Institute . Cd4::CreERT2 and Ubc::CreERT2 mice ( Jackson ) were crossed to Csf1 fl/fl mice [85] ( kindly provided by S . Abboud-Werner , University of Texas Health Science Center ) to generate hemizygous CreERT2; Csf1 fl/fl mice . CD169+/DTR mice were kindly provided by J . Cyster ( UCSF ) and M . Tanaka ( RIKEN Research Center for Allergy and Immunology ) [66] . Lyz2Cre/Cre mice [70] were bred to Rosa26::STOPfl/fl::DTR mice in-house ( both Jackson ) . Female 8–12 week old mice were used for infections; littermate controls were used for all experiments with floxed and DT-treated mice . To induce deletion of floxed Csf1 , mice were fed tamoxifen chow ( Envigo ) ad libitum beginning 1 month prior to infection and through the duration of each experiment . Mice were infected with 106 Plasmodium chabaudi AS ( MRA-429; MR4 Stock Center ) parasitized RBCs , and parasitemia was monitored by thin film blood smear as described [86] . Where noted , mice were injected i . p . with the following: 300 μg α-CD4 ( GK1 . 5 ) or isotype control ( LTF2 ) ; 500 μg α-MCSF ( 5A1 ) or isotype control ( HRPN ) ( all BioXCell ) ; or 300 μL liposomes loaded with PBS or clodronate at neutral pH ( FormuMax Scientific ) . To deplete CD169+ cells , heterozygous CD169+/DTR mice were treated 12 d . p . i . with a single i . p . dose ( 80 ng/g ) of DT ( Sigma D0564 ) or a catalytically inactive point mutant ( DT*Glu; Sigma D2189 ) . To deplete Lyz2+ cells , including bone marrow macrophages , Lyz2Cre/Cre; Rosa26::STOPfl/fl::DTR mice were treated with 500 ng DT or DT*Glu on d 13 and d 15 post-infection . Blood was obtained by cardiac puncture or submandibular bleed; spleens were excised and homogenized after euthanasia according to approved protocols . Following RBC lysis in ACK buffer , samples were blocked , labeled with antibodies , and analyzed on an LSR II ( BD ) to assess myeloid cell frequencies . Intracellular Ki67 levels were measured using the Fixation and Permeabilization Buffer Set ( eBioscience ) and were compared to cells labeled with an isotype control antibody . Antibodies are listed in S3 Table . The Annexin V Apoptosis Detection Kit ( eBioscience ) was used with propidium iodide to quantify apoptotic cells . For EdU labeling , mice were injected i . p . with 750μg EdU 3 h prior to sacrifice , and the Click-It EdU Kit ( Thermo Fisher ) was used to detect EdU in splenocytes according to the manufacturer’s protocol . Myeloid cell definitions were as follows: red pulp macrophages and bone marrow macrophages ( Ly6G- Ly6C- CD11blo F4/80+ ) ; nonclassical monocytes ( Ly6G- Ly6C- CD11b+ F4/80int SSClo ) ; classical monocytes ( Ly6G- Ly6C++ CD11b+ F4/80int SSClo ) ( S1A Fig ) . Additional labeling with CD115 and CD68 was performed to confirm the identities of monocyte and macrophage populations ( S1B and S1C Fig ) . Absolute cell numbers were quantified in blood prior to RBC lysis using the Guava Viacount assay ( EMD Millipore ) and in RBC-lysed spleen samples using a hemocytometer . Antigen-experienced T cells ( S2A Fig , [42 , 43] ) were isolated from wild-type mice 6 d . p . i . by double-sorting to high purity on a FACSAria ( BD ) . RNA was isolated using the RNAqueous Micro kit ( Ambion ) and amplified ( Amino Allyl MessageAmp II kit , Life Technologies ) to generate amino allyl incorporated amplified RNA ( aaRNA ) . aaRNA was coupled to Cy3 dye ( GE Healthcare Life Sciences ) and hybridized overnight to a SurePrint G3 Mouse Gene Expression 8x60K microarray ( Agilent ) , which was washed and scanned per manufacturer’s instructions . Raw intensities were extracted using Feature Extraction software ( Agilent ) and quantile normalized using Limma [87] . Differentially expressed genes were identified using Significance Analysis for Microarrays ( SAM ) [88] . Complete microarray data can be accessed in the Gene Expression Omnibus database ( GEO; http://www . ncbi . nlm . nih . gov/geo/ ) under accession GSE81196 . Blood cells were labeled with antibodies ( S3 Table and flow cytometry methods above ) and sorted on a FACSAria ( BD ) directly into lysis buffer . RNA was isolated using the RNAqueous Micro Kit ( Ambion ) and reverse-transcribed into cDNA using Superscript III ( Life Technologies ) primed with dT ( 20 ) V . For microarray validation , cDNA was made from aaRNA with minor protocol modifications as described [89] . Quantitative PCR was performed in a Step One Plus RT PCR System ( Applied Biosystems ) using PerfeCTa 2x qPCR mix ( Quanta ) . Transcript levels were normalized to levels of actin mRNA . Primer sequences are listed in S4 Table . To examine MCSF production by T cells , CD11a+ CD49d+ CD4+ T cells were sorted on a FACSAria ( BD ) from infected mice 6 d . p . i . and cultured for 4 d in 96-well plates at 105 cells/well with PMA ( 10 ng/mL ) and ionomycin ( 1 μg/mL ) ( both Fisher ) . Cell-free supernatants were harvested for ELISA . Antigen-naive ( CD11a- CD49d- ) CD4+ T cells and non-T cells ( TCRβ- ) were sorted and cultured as controls . For plasma measurements , blood was collected from the sub-mandibular vein into K2EDTA and centrifuged to separate cells from plasma , which was snap-frozen and stored at -80°C until analysis . MCSF was measured using the Murine M-CSF ELISA Development Kit ( PeproTech ) according to the manufacturer's protocol . Mesenteric lymph nodes were obtained 14 d . p . i . and immediately frozen in Tissue-Tek OCT Compound ( Sakura Finetek ) . 10μm sections were mounted , fixed in acetone , and labeled with FITC-conjugated α-CD169 ( MOMA-1; ABD Serotec ) followed by Alexa488-conjugated α-FITC ( Jackson Immunoresearch ) . In some experiments , directly fluoroconjugated α-Ki67 ( eBioscience ) was also used . To examine MMMs , infected CD169+/DTR mice were treated 12 d . p . i . with DT or DT*Glu as described above , and spleen sections were obtained 24 h later . Sections were labeled with antibodies to B220 ( RA3 . 3A1/6 . 1 , UCSF Monoclonal Antibody Core ) , CD169 ( MOMA-1 , AbD Serotec ) , and F4/80 ( BM8 , UCSF ) to label B cells , MMMs , and RPMs , respectively . All samples were coverslipped with Vectashield Mounting Medium with DAPI ( Vector ) and visualized using an AxioCam HR camera on an AxioImagerM2 upright microscope . To quantify CD169 labeling , lymph nodes were viewed in Image J ( https://imagej . nih . gov/ij/ ) and the Segmented Lines tool was used to measure the length of the lymph node perimeter that labeled positively with α-CD169 . This was divided by the total circumference of the lymph node , measured using the Segmented Lines tool on DAPI+ cells , to obtain a numerical value for the fraction of the lymph node capsule that was lined by CD169+ cells . Four technical replicates ( i . e . , individual tissue sections , separated by at least 20μm ) were performed for each biological replicate ( i . e . , individual mouse ) . Antigen-experienced CD4+ T cells were isolated to high purity , using two consecutive rounds of FACS , from the blood of mice 6 d . p . i . Sorted cells were loaded onto a Fluidigm C1 , captured , and processed into cDNA libraries following manufacturer protocols . Capture sites with zero or more than one cell were excluded from the libraries; libraries from 40 total cells were indexed , pooled into a single library , and sequenced on a HiSeq 2500 in high output mode . Reads were aligned using RSEM 1 . 2 . 22 and STAR 2 . 4 . 2a to GRCm38; samples contained an average depth of 3 . 9 million aligned reads , with 84 . 9% of reads aligning . Samples with fewer than 0 . 5 million aligned reads were excluded from further processing . For differential gene expression analysis , edgeR 3 . 4 . 2 was used to identify significantly differentially expressed genes between cells with Csf1 expression of 0 TPM versus Csf1 expression of > 1 TPM ( FDR < 5% ) . All RNA-Seq data are available in GEO under accession GSE81197 . After RBC lysis , splenocytes from naive mice were incubated in plates coated with α-CD3 and α-CD28 ( 2 . 5 μg/mL each; UCSF Monoclonal Antibody Core ) at a concentration of 106 cells/mL in T cell media ( RPMI + 10% FBS + 1mM sodium pyruvate + 50 U/mL penicillin + 50 U/mL streptomycin + 0 . 1% β-2-mercaptoethanol ) alone or with the following polarization cocktails: for Th1 , IL-12 ( 10 ng/mL , R&D ) + α-IL-4 ( 10 μg/mL , Biolegend ) ; for Th2 , IL-4 ( 10 ng/mL , R&D ) + α-IFN-γ ( 10 μg/mL , Biolegend ) . RNA was harvested after 5 d and processed for RT-qPCR as above . Where indicated , PMA and ionomycin were added after 5 d culture and samples were incubated for 6 h before RNA harvest . Simultaneously , splenocytes from a naïve mouse were plated in PMA + ionomycin for 6 h and processed for RNA along with cultured samples . | Malaria , caused by Plasmodium parasites , places a huge disease burden on humankind . Efforts to develop an effective vaccine for this pathogen are hampered by a poor understanding of the kinds of immune responses needed for protection . When infected with Plasmodium , humans and other mammalian hosts develop greatly increased numbers of certain white blood cells called myeloid cells , but the extent to which this expansion is important for fighting infection is unknown , as are the factors that regulate it . In this work , we show that expansion of myeloid cells during malaria is critical for control of parasite replication and host recovery . Myeloid expansion and control of infection require production of the signaling protein Macrophage Colony Stimulating Factor ( MCSF ) by multiple sources , including CD4+ T cells , which had not previously been known to make this protein in a physiological setting . This study reveals a new mechanism through which immune cells cooperate to control malaria , and may guide future attempts to generate protective immune responses against this devastating pathogen . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immunology",
"parasitic",
"diseases",
"parasitology",
"animal",
"models",
"parasitemia",
"bone",
"marrow",
"cells",
"model",
"organisms",
"lymph",
"nodes",
"lymphatic",
"system",
"quantitat... | 2016 | Macrophage Colony Stimulating Factor Derived from CD4+ T Cells Contributes to Control of a Blood-Borne Infection |
Endogenous RNAi ( endoRNAi ) is a conserved mechanism for fine-tuning gene expression . In the nematode Caenorhabditis elegans , several endoRNAi pathways are required for the successful development of reproductive cells . The CSR-1 endoRNAi pathway promotes germ cell development , primarily by facilitating the expression of germline genes . In this study , we report a novel function for the CSR-1 pathway in preventing premature activation of embryonic transcription in the developing oocytes , which is accompanied by a general Pol II activation . This CSR-1 function requires its RNase activity , suggesting that , by controlling the levels of maternal mRNAs , CSR-1-dependent endoRNAi contributes to an orderly reprogramming of transcription during the oocyte-to-embryo transition .
Developmental plasticity , or pluripotency , is acquired during the oocyte-to-embryo transition , during which a highly specialized cell , the oocyte , is reprogrammed into a developmentally flexible embryo [1] . A tight regulation of this reprogramming is critical , since germ cells that precociously acquire pluripotency can give rise to teratomas , which are germ cell tumors characterized by the presence of differentiated somatic cells . Like in mammals , teratomas in C . elegans can arise in the absence of reprogramming-controlling RNA binding proteins ( RBPs ) [2] . The C . elegans RBPs GLD-1/Quaking and LIN-41/TRIM71 inhibit a precocious reprogramming of germ cells into somatic-like cells at consecutive stages of oogenic development [3 , 4] . Similar to the natural order of events during the oocyte-to-embryo transition , somatic-like differentiation of gld-1 and lin-41 germ cells is preceded by reactivation of the cell cycle and embryonic genome activation ( EGA ) [3–5] . During earlier stages of oogenic development , inhibition of the cell cycle by GLD-1 is critical to prevent precocious reprogramming [5] . The reprogramming-related function of LIN-41 in differentiating oocytes remains less clear , though it may also involve cell cycle regulation [4 , 6 , 7] . To better understand the control of reprogramming in the oocytes , we performed an unbiased genetic screen , searching for mutants displaying precocious activation of embryonic transcription in the oocyte-containing ( proximal ) part of the gonad , and identified components of endogenous small interfering RNA ( endo-siRNA ) pathways . There are two main endo-siRNA pathways operating in the C . elegans germline , which are named after the constituent Argonaute proteins; the WAGO ( worm argonaute ) and the CSR-1 ( chromosome-segregation and RNAi-deficient-1 ) pathway [8 , 9] . These Argonautes bind specific classes of small RNAs , which are 22 nucleotides in length with a guanosine at the 5’end ( 22G RNAs ) [8] . 22G RNAs are produced by RNA-dependent RNA polymerases ( RdRPs ) and act as secondary effectors of the endo-siRNA pathways . CSR-1 is the only worm Argonaute protein required for fertility and embryo survival [10] . It has been reported to function in diverse processes , including chromosome segregation [9 , 10] , chromatin organization [9 , 11] , histone mRNA processing [12] , germ granule formation [9 , 13] , alternative splicing [14] , and exogenous RNAi [10] . CSR-1 binds small RNAs that are complementary to most germline-expressed genes [9] , and has been suggested to promote the expression of these target genes [15] . Moreover , CSR-1 has been suggested to counteract gene silencing by recognizing and licensing self-sequences for expression [16–18] . Among the reported inhibitory roles of CSR-1 are the translational repression of FBF-1 target mRNAs in mitotic germ cells [19] , and the degradation of germline-expressed mRNAs dependent on the RNA-slicing activity of CSR-1 [20] . In this study , we present a new role for the CSR-1 pathway and its RNA-slicing activity in ensuring the transcriptional silencing of embryonic genes in developing oocytes . In wild-type development , the oocyte-to-embryo transition takes place in the absence of Pol II-dependent transcription . In slicer-inactive csr-1 mutants , however , the global Pol II inhibition is compromised . Our observations suggest a model , where CSR-1–dependent titration of maternal mRNAs is required to delay the onset of embryonic transcription until its physiological onset in early embryos .
During the oocyte-to-embryo transition , the RBPs GLD-1 and LIN-41 inhibit the onset of embryonic transcription in oogenic germ cells [4 , 5] . The EGA-inhibiting function of LIN-41 was uncovered through a genetic screen , in which a premature onset of embryonic transcription in the adult germline was monitored by GFP , expressed from the promoter of a gene expressed as soon as EGA takes place , vet-4 ( henceforth the EGA-GFP reporter ) [4] . Here , we continued this approach to identify additional EGA inhibitors ( Fig 1A ) . We isolated three independent mutants ( rrr2 , rrr5 and rrr9 ) expressing the EGA-GFP reporter precociously in germ cells ( Fig 1B ) . Complementation tests , using these mutants alongside gld-1 and lin-41 , suggested the identification of two new genes; one affected by the rrr2 and rrr5 mutations , and the other by the rrr9 mutation ( Fig 1C ) . The abnormal EGA observed in gld-1 or lin-41 germ cells occurs in the context of a premature exit from meiosis and the onset of mitotic proliferation , which is followed by teratoma formation [3 , 4 , 6 , 7 , 21] . In addition , EGA can be induced by precocious oocyte maturation [22] . Oocyte maturation occurs in wild-type gonads in the ( -1 ) oocyte adjacent to the spermatheca , and is characterized by nuclear envelope breakdown ( NEBD ) , cytoskeletal rearrangements and meiotic spindle assembly [23] . To test whether EGA in the new mutants reflects precocious oocyte maturation , we examined the NEBD in these mutants . To visualize nuclear envelopes , we used a strain expressing mCherry-tagged nuclear envelope protein , EMR-1 [24] . We observed that the new mutants expressed EGA-GFP in the absence of NEBD ( Fig 2A ) , indicating that EGA is not induced by precocious oocyte maturation . To follow it further , we examined additional events accompanying the progression through meiosis . In wild-type , but not in gld-1 or lin-41 mutants , centrosomes are eliminated during oocyte differentiation [7 , 25 , 26] . We monitored the centrosomes in rrr2 mutants by staining the centrosomal protein SPD-2 [27] , and observed the loss of centrosomes in mutant oocytes ( Fig 2B ) . Thus , similar to wild type , the mutants appear to eliminate centrosomes from the oocytes . Finally , wild-type oocytes , which arrest in the diakinesis stage of meiosis I , display highly condensed bivalent chromosomes . In the mutants , we observed similarly condensed chromosomes ( Fig 2B ) , suggesting a relatively normal meiotic progression and arrest . Thus , in contrast to the previously reported cases of precocious EGA , in the new mutants EGA appears to be uncoupled from reactivation of the cell cycle . To characterize further embryonic transcription in the mutant gonads , we examined expression of the endogenous vet-4 and several additional embryonic transcripts , using reverse transcription and quantitative PCR ( RT-qPCR ) . We found that the tested early embryonic transcripts ( vet-4 , vet-6 and pes-10 ) were all upregulated in the gonads dissected from rrr2 mutants , compared to wild-type gonads ( Fig 3A ) . In gld-1 or lin-41 gonads , precocious EGA is followed by embryonic-like differentiation resulting in teratomas [4 , 5] . To examine whether rrr2 germ cells attempt somatic differentiation , we examined the expression of several transcripts specific to somatic lineages . We found that these transcripts were not upregulated in the dissected rrr2 gonads ( Fig 3B ) . Thus , the rrr2 mutant germ cells appear to execute the first step in the transcriptional oocyte-to-embryo reprogramming , but in contrast to gld-1 or lin-41 mutants , do not undergo further somatic-like differentiation . To examine the extent of embryonic transcription in the rrr2 animals , we performed RNA sequencing on wild type and rrr2 mutants ( expression changes between the two biological replicates are shown in S1 Fig and S1 Dataset ) . We selected 446 early embryonic transcripts based on the transcriptome profiling of staged embryos [28] ( see Material and Methods ) . Some of these transcripts were expressed in the wild type ( Fig 4A ) . Because we used young adults prior to embryo production , we suspected that this reflects their additional expression in the adult soma . Thus , any changes in these transcripts in the germline would be likely masked by their somatic expression . To circumvent this problem , we split the embryonic transcripts between somatically expressed and not expressed ( based on the previous analysis of germline-specific expression [29] ) ( Fig 4B ) . Consistently , the embryonic transcripts , which were also expressed in the soma , were not upregulated in the rrr2 mutant compared to wild type ( p-value: 0 . 12; calculated with 2-sample Wilcoxon test , Fig 4C ) . In contrast , the “strictly” embryonic genes , which were not somatically expressed in the wild type , showed a mild , but significant upregulation in the rrr2 mutant ( p-value: 8 . 8 x 10−28; Fig 4D ) . These results are consistent with a widespread de-repression of embryonic genes in the mutant gonads . We mapped the mutants to drh-3 ( dicer-related-helicase-3 , alleles rrr2 and rrr5 ) and to ego-1 ( enhancer-of-GLP-ONE-1 , allele rrr9 ) ( Fig 5A and S2 Fig ) ; both functioning in small non-coding RNAi pathways . All mutants behaved like molecular nulls , displaying fully penetrant sterility as reported earlier for the reference drh-3 and ego-1 alleles [30 , 31] . Consistently , DRH-3 protein was not detectable in drh-3 ( rrr2 or rrr5 ) mutants by western blot ( Fig 5B ) . DRH-3 is a core component of all RdRP complexes and therefore important for the biogenesis of all classes of 22G-RNAs [8] . EGO-1 functions either redundantly with another RdRP , called RRF-1 , in the production of WAGO 22G-RNAs [8] , or alone in the production of CSR-1 22G RNAs [9] . To test which of the two pathways controls EGA , we examined expression of the EGA-GFP reporter in mutants affecting both or just one of the two pathways . In drh-3 mutants ( both newly isolated and reference ) that affect both pathways , we observed the expected gonadal EGA-GFP ( Fig 5C ) . Different primary siRNA-pathways , functioning upstream of the WAGO pathway , use the RdRP complex for the production of secondary 22G RNAs to enhance their effects . They include the maternal ERGO-1 and spermatogenesis-specific ALG-3/4 26G RNA pathways [32–34] , as well as the piRNA pathway [35–37] . To address if precocious EGA results from disrupting the 26G RNA pathways , we examined the expression of EGA-GFP in eri-1 ( mg366 ) mutants lacking the 26G RNAs [33] . To test the involvement of the piRNA pathway , we examined prg-1 ( tm892 ) mutants lacking most piRNAs [38] . We found that neither eri-1 nor prg-1 mutants expressed EGA-GFP in the gonads ( Fig 5C ) . Likewise , mut-2 or mut-7 mutants , which are deficient in the amplification of WAGO 22G RNAs [8 , 39 , 40] , displayed only occasional gonadal EGA-GFP . In contrast , 90% of the csr-1 ( tm892 ) mutants displayed the gonadal expression of EGA-GFP , which was comparable to ego-1 ( rrr9 ) or drh-3 mutants ( Fig 5C ) . To verify that the observed gonadal EGA-GFP reflects the expression of endogenous embryonic genes , we examined several embryonic transcripts by RT-qPCR . Consistent with the expression of the EGA-GFP reporter , we observed increased expression of embryonic genes in the gonads from csr-1 ( tm892 ) mutants , but not the gonads from the MAGO12 strain , which carries mutations in all WAGO genes [8] ( Fig 5D ) . Together , these results suggest that it is the CSR-1 pathway that prevents precocious EGA . Recently , CSR-1 has been reported to contribute to maternal mRNA regulation via its endonuclease activity ( also RNase or slicer activity ) [20] . To test whether the slicer activity is required to prevent precocious EGA , we examined , by RT-qPCR , the expression of several embryonic transcripts in gonads dissected from csr-1 ( tm892 ) animals , expressing either rescuing FLAG-tagged CSR-1 [18] or a FLAG-tagged CSR-1 variant , in which the DDH catalytic slicer residues were mutated to AAA ( obtained from Craig Mello’s lab ) . Similar to other csr-1 loss-of-function mutations , the slicer-inactive CSR-1 ( CSR-1SIN ) protein was unable to prevent the gonadal expression of embryonic transcripts ( Fig 6A ) . Thus , the RNase activity of CSR-1 is required to prevent precocious EGA . CSR-1 is thought to contribute to normal embryonic development by regulating the abundance of maternal transcripts loaded into the embryo [20] . Thus , to address whether CSR-1 may directly target early embryonic transcripts , we examined whether these transcripts have complementary CSR-1 22G RNAs . Early embryonic transcripts were determined as described before ( see Material and Methods ) . We then defined the number of putative CSR-1 targets based on the presence of complementary 22G RNAs associated with CSR-1 [9] . This analysis showed that a great majority of early embryonic transcripts ( 416/446 genes ) did not have any complementary CSR-1 22G RNAs ( Fig 6B ) ; among the embryonic transcripts that we routinely tested by RT-qPCR , only vet-6 had complementary 22G RNAs . Moreover , a variant of the EGA-GFP reporter , in which the vet-4 3’UTR was replaced by a non-regulated tbb-2 3’UTR , thus limiting the regulation of this reporter to the vet-4 promoter , was also de-repressed in CSR-1–depleted gonads ( S3 Fig ) . Combined , these observations suggest that CSR-1 is unlikely to regulate EGA by directly degrading embryonic transcripts . A global repression of Pol II-dependent transcription is a hallmark of the oocyte-to-embryo transition . In C . elegans , Pol II transcription is silenced from late stages of meiosis I in oocytes until the onset of EGA in early embryos [41–44] . The inhibition of Pol II is manifested by the loss of activating phosphorylations on Pol II C-terminal domain ( CTD ) ; serine 5 is phosphorylated ( PSer5 ) during transcription initiation and serine 2 ( PSer2 ) during the elongation [44] . To explore whether the premature EGA observed in the slicer-inactive csr-1 mutants might reflect defective Pol II inhibition , we stained the gonads of CSR-1SIN-expressing animals [20] with antibodies specific to PSer5 and PSer2 . We found that , in contrast to animals expressing CSR-1WT , the oocytes in CSR-1SIN-expressing animals contained Pol II phosphorylated at both Ser5 and Ser2 ( Fig 7A and 7B ) ; these phosphoepitope-specific stainings corresponded to Pol II , as they were diminished in CSR-1SIN animals depleted of the Pol II large subunit AMA-1 ( S4 Fig ) . These findings suggest that , in the absence of CSR-1-dependent slicing , the oocytes no longer undergo Pol II inhibition , potentially linking it to the premature expression of embryonic genes .
CSR-1 has been recently shown to function as an RNase that , by degrading certain maternal transcripts , ensures proper embryonic development [20] . Although it remains possible that the point mutations in CSR-1 affecting its RNase activity might compromise interactions with other factors , our findings suggest that CSR-1 uses its RNA-slicing activity to control embryonic event ( s ) already in the oocytes , where it prevents a precocious activation of embryonic transcription . The regulation by CSR-1 is expected to involve 22G RNAs , which are associated with CSR-1 to guide it to its mRNA targets . Thus , the observation that most early embryonic transcripts lack complementary 22G RNAs implies that CSR-1 is unlikely to control EGA by directly degrading EGA transcripts . In one scenario , CSR-1 could control embryonic transcription indirectly by degrading mRNA encoding a transcriptional regulator . In other species , EGA depends on specific transcription factors , which recognize sequence motifs in the embryonic promoters [45] . However , an analogous transcription factor has not been found in C . elegans to date . Thus , we tested whether the promoters of early embryonic genes misregulated in drh-3 animals are enriched for the binding motifs of specific transcription factors . Although we observed enrichment for the binding motifs of HMG-12 and LIN-29 , these motifs are very abundant in the promoters of all embryonic genes ( S5 Fig ) , potentially leading to a bias in the statistical analysis . Moreover , previous analysis argued against LIN-29 function in the activation of embryonic genes [4] . In embryos , EGA is controlled by a mechanism relying on the inhibition of general transcription factors . In one- and two-cell stage embryos , TAF-4 , a TFIID subunit , is sequestered in the cytoplasm by the RNA-binding proteins OMA-1/2 , thereby preventing the nuclear function of TAF-4 in Pol II-dependent transcription [46] . In four-cell stage embryos , the degradation of OMA-1/2 permits the nuclear translocation of TAF-4 , consequently allowing TFIID assembly , Pol II transcription , and EGA . Our observation that Pol II remains activated in slicer-inactive csr-1 oocytes suggests a possible connection between CSR-1 and Pol II repression . While the levels of most mRNAs encoding Pol II subunits appeared to remain constant in the absence of CSR-1 slicing activity , the levels of taf-11 . 2 ( encoding a general transcription factor ) and cit-1 . 2 ( a subunit of the transcription elongation factor pTEFb ) were increased [20] . Thus , CSR-1–mediated degradation of the corresponding mRNAs could potentially explain the link between CSR-1 , Pol II regulation , and EGA . However , depleting CSR-1 from taf-11 . 2 or cit-1 . 2 mutants failed to suppress germline EGA ( our observation ) , suggesting that the potential degradation of these targets by CSR-1 is not critical to inhibit EGA . In another indirect scenario , CSR-1 could control EGA by ensuring wild-type chromatin architecture , which could involve either cytoplasmic or nuclear functions of CSR-1 . For example , CSR-1 has been implicated in the biogenesis of histone mRNAs [12] . However , reducing histone levels ( by depleting CDL-1 , another histone mRNA biogenesis factor ) did not result in oocyte expression of the EGA reporter ( S6 Fig ) . Other effects on chromatin organization resulting from the depletion of CSR-1 pathway components include mislocalization of the centromere-specific histone HCP-3/CENP-A [9 , 15 , 47] and the loss of H3K9me2 deposition on unpaired chromatin [11] . Whether the CSR-1 slicer function is required for these chromatin-related functions of CSR-1 is not yet clear . Thus , we cannot exclude the possibility that the premature EGA and abnormal activation of Pol II are indirectly linked to altered chromatin . Summarizing , our studies reveal an unexpected role for the CSR-1 endoRNAi pathway in inhibiting the expression of embryonic genes during oocyte development . While the majority of existing studies suggested a positive function for CSR-1 in promoting germline expression , this and the recent study by Gerson-Gurwitz et al . suggest a negative function , involving its RNA-slicer activity . Whether CSR-1 controls EGA by degrading mRNA encoding a specific transcriptional regulator , by affecting chromatin , or through an entirely different mechanism , remains an open question . Another open issue is the relation of CSR-1 to the previously identified EGA repressors , particularly LIN-41 , which , like CSR-1 , represses EGA in the developing oocytes [4] . In lin-41 mutants , premature EGA and subsequent teratomatous differentiation are coupled to the re-activation of the cell cycle [4 , 5 , 22] . By contrast , premature EGA in CSR-1 pathway mutants is independent from the cell cycle . Because lin-41 mutants display several oocyte-to-embryo transition events , but csr-1 mutants only EGA , CSR-1 might function “downstream” from LIN-41 . Importantly , also mammalian oocytes express and utilize endo-siRNAs [48] . These endo-siRNAs are produced by an oocyte-specific Dicer isoform , DicerO , which lacks the N-terminal DExD helicase domain; the loss of DicerO from oocytes results in meiotic arrest , with spindle and chromosome segregation defects , resulting in sterility [49] . The deletion of AGO2 , the only mammalian Argonaute with slicer activity , results in a similar oocyte phenotype , and the slicer activity of AGO2 is essential for the oocyte function of endo-siRNAs [50] . Thus , the regulation by endo-siRNAs is a conserved feature of oocyte biology , and the potential role of mammalian endo-siRNAs in controlling the embryonic genome remains an exciting possibility .
Animals were maintained using standard procedures and were grown at 20°C unless stated otherwise . For alleles and transgenic lines used in this study , see the S1 Table . The screen was performed as previously described [4] . The rrr2 mutant was identified using strain #1284 , the EGA-GFP reporter strain , as a parent strain for mutagenesis . The rrr5 and rrr9 mutants were discovered using strain #1270 as parent strain , which contains in addition to the EGA-GFP reporter a transgene to visualize P granules and a glo-1 mutation to reduce autofluorescence from the gut . Mapping of the mutants was performed as described before [4] . Before WGS , each mutant was outcrossed 4–8 times to the unmutagenized parent strain . Genomic DNA was isolated using Gentra Puregene Tissue Kit ( Qiagen ) . DNA libraries were made from 50 ng of genomic DNA using the Nextera DNA kit from Illumina . Sequencing was performed using HiSeq 2000 from Illumina . The analysis was performed similarly as described in [4] . Briefly , the sequence reads were aligned to the May 2008 C . elegans assembly ( obtained from http://hgdownload . soe . ucsc . edu/goldenPath/ce6/chromosomes/ ) using ‘‘bwa” [51]; version 0 . 7 . 4 ) with default parameters , but only retaining single-hit alignments ( ‘‘bwa samse -n 1” for single reads and “bwa sampe -a 1000 -o 1000 -n 1” for paired end reads and selecting alignments with ‘‘X0:i:1” ) . The resulting alignments were converted to BAM format , sorted and indexed using ‘‘samtools”[52]; version 0 . 1 . 19 ) . In order to quantify contamination by Escherichia coli , reads were similarly aligned to a collection of E . coli genomes ( NCBI accession numbers NC_008253 , NC_008563 , NC_010468 , NC_004431 , NC_009801 , NC_009800 , NC_002655 , NC_002695 , NC_010498 , NC_007946 , NC_010473 , NC_000913 and AC_000091 ) , which typically resulted in less than 1% aligned reads . Potential PCR duplicates were identified and removed using Picard ( version 1 . 92 , http://broadinstitute . github . io/picard/ ) , reducing the number of reads to 27% to 44% in single read samples , and to 93% in the paired-end read sample . Sequence variants were identified using GATK [53] ( version 2 . 5 . 2 ) following recommended “best practice variant detection”: initial alignments were first corrected by indel realignment and base quality score recalibration , followed by SNP and INDEL discovery and genotyping using “UnifiedGenotyper” for each individual strain using standard hard filtering parameters , resulting in a total of five to six thousand sequence variations in each strain compared to the reference genome . Finally , the number of high quality ( score > = 500 ) single nucleotide substitutions of EMS-type ( G/C>A/T transitions [54] ) not found in the parent strain ( typically less than 1% of the total number of variants per strain ) were counted in sequential windows of 1 Mb to identify regions of increased variant density . We performed complementation crosses between the newly identified mutants , and mutants affecting the previously identified EGA repressors gld-1 and lin-41 ( using the gld-1 ( q485 ) and lin-41 ( rrr3 ) alleles ) . Due to sterility of the homozygous mutants , we crossed heterozygous mutants and examined the F1 progeny for the precocious EGA/sterility phenotype occurring with expected penetrance . Only in the crosses between heterozygous rrr2 and rrr5 mutants did we observe non-complementation in the F1 progeny , suggesting that the mutations affected the same gene . Based on the whole-genome sequencing ( see above ) , the only gene displaying sequence alterations in both mutants was drh-3 . We confirmed the drh-3 mutations as causal , based on non-complementation with the published drh-3 ( fj52 ) mutant , and by observing precocious EGA upon drh-3 RNAi and in other drh-3 mutants ( alleles fj52 and tm1217 ) . The rrr9 mutant contained only few variants of the EMS-type; among those only four variants were predicted to affect the corresponding proteins by non-synonymous substitutions . Among the candidates , only ego-1 mutants were previously reported to be sterile . The mutation in ego-1 was confirmed to be causal , based on non-complementation with the published ego-1 ( om71 ) mutant and a precocious EGA was observed upon ego-1 RNAi . RNA was isolated from 50 gonads of 1-day-old ( after the L4-to-adult molt ) animals using the Picopure RNA Isolation Kit ( Arcturus ) . Three independent biological replicates were collected . cDNAs were generated using the QuantiTect Reverse Transcription Kit ( Qiagen ) . Real-time PCR was performed in duplicates ( for each biological replicate ) using Absolute QPCR SYBR green ROX mix ( AbGene ) on an ABI PRISM 7700 system or a StepOnePlus system ( both Applied Biosystems ) . qPCR reactions were performed as described previously [5] . At least one primer in each pair was specific for an exon-exon junction ( see primer sequences in the S2 Table ) . Standard curves for every primer pair were generated using a serial dilution of cDNA from embryos and were used to determine the amount of each transcript in the gonads . All technical duplicate values were first averaged , and then values for a particular transcript from tested animals were normalized to the mean value for the corresponding transcript from control animals . The normalized means of the samples were used to make the bar plots . Error bars show the standard error of the mean ( SEM ) for the three biological replicates . We used act-1 , encoding an actin isoform , and tbb-2 , encoding a tubulin isoform , as reference transcripts as they are ubiquitously expressed housekeeping genes . Statistical analysis of the data was performed using two-sided Student’s t-test , assuming non-equal variance , in R . Each comparison was performed separately without correcting for multiple statistical tests . For total RNA sequencing , we collected samples from young adult drh-3 ( rrr2 ) mutants and wild type ( N2 ) animals in duplicates . The RNA sequencing samples from wild type have been uploaded previously to the GEO database ( series GSE62858 , GSM1534604 and GSM1534605 ) [29] . RNA extraction and library preparation were performed as described before [29] . The samples were sequenced on an Illumina HiSeq 2500 . The total RNA sequencing data was analyzed as previously described [55] . Differential gene expression analysis for rrr2 was performed using the edgeR package from Bioconductor [56] , and is presented in the S1 Dataset . We downloaded microarray expression data [28] for the samples ( GSM39513-GSM39519 , GSM39543 , GSM39522-GSM39526 , GSM39530 and GSM39531 ) from GEO ( www . ncbi . nlm . nih . gov/geo/ ) , representing the 4-cell ( 3x ) , 8-cell ( 4x ) , 15-cell ( 4x ) and 26-cell ( 4x ) stages . The data was normalized using the function justGCRMA from the Bioconductor package gcrma . We confirmed a high degree of consistency among the replicates and then averaged those to obtain one expression profile per stage . Embryonic transcription in C . elegans starts at the 4-cell stage [57] . Therefore , early embryonic genes were defined as genes that showed a change in expression of at least 2-fold between the 4-cell and the 8-cell stage , or between the 8-cell and the 15-cell stage and had no expression at the 4-cell stage ( expression < 3 . 75 ) . Maternally provided genes were defined as genes that showed expression at the 4-cell stage ( expression > = 3 . 75 ) . Genes not falling in any of the two above categories were defined as non-embryonic . Data from the 26-cell stage were not used for selecting genes , as there was only little change in expression when comparing to the 15-cell stage . To identify the somatically expressed early embryonic transcripts , we used RNA sequencing data ( GSE62857 ) from dissected gonads and germline-less glp-4 animals [29] . We processed and normalized the data as described previously [29] and defined somatic transcripts using a cutoff of 5 ( log2 ) . Temperature-sensitive mutants used for the EGA reporter assay in Fig 5C were grown at 15°C . Mutant animals were synchronized by bleaching and put as L1 larvae on OP50 plates to 25°C until scored for EGA-GFP expression as adults . Immunostaining against SPD-2 ( “969LA” , 1:800 ) was performed as previously described [4] . A Zeiss Axio Imager Z1 microscope equipped with an Axiocam MRm REV 2 CCD camera was used for capturing pictures . Immunostainings against PSer5 of Pol II CTD ( “3E8” , Millipore , 1:25 ) and PSer2 of Pol II CTD ( “3E10” , Millipore , 1:25 ) were performed as previously described [44] . DNA was visualized using 4-6-Diamidino-2-phenylindole ( DAPI ) . Pictures were taken using an Axio Imager M2 microscope and a Yokogawa CSU W1 Duel camera . Images were processed with Fiji and Adobe Photoshop in an identical manner and imported into Adobe Illustrator . For RT-qPCR analysis ( Fig 6A ) , we used strain 1689 for CSR-1WT [18] , and strain 1755 for CSR-1SIN ( kindly shared by the Mello group prior to publication ) . The Pol II CTD staining experiments ( Fig 7A and 7B ) were performed using published strains 1905 for CSR-1WT and 1900 for CSR-1SIN [20] . Synchronized adult worms were harvested for protein extraction as described before [5] . The drh-3 homozygous populations were obtained by worm sorting . Proteins were resolved by SDS-PAGE on 4–12% Bis-Tris Protein Gels ( NuPAGE Novex ) and transferred using the Trans-Blot Turbo Transfer System ( BIO RAD ) . Membranes were blocked with 4% milk in PBST and incubated with the primary antibodies α-DRH-3 [58] or α-ACT-1 ( MAB1501 , Millipore ) in blocking buffer at 4°C overnight . Membranes were washed three times in PBST and incubated with the secondary ( HRP-conjugated ) antibody ( GE Heathcare ) in blocking buffer at RT for 1 hour , and washed again three times with PBST . Chemiluminescence was performed using Pierce ECL Western Blotting Substrate ( Thermo Scientific ) . | During the oocyte-to-embryo transition , the control of development is transferred from the mother to the embryo . A key event during this transition is the transcriptional activation of the embryonic genome , which is tightly controlled . Here , by using the nematode C . elegans , we uncover a role for endogenous RNA interference in this process . We demonstrate that a specific endoRNAi pathway , employing the Argonaute protein CSR-1 , functions as a break on gene-specific , and potentially global , activation of embryonic transcription in the developing oocytes . Our findings reveal a new layer of control over the transcriptional reprogramming during the oocyte-to-embryo transition , raising questions about its potential conservation in mammalian development . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"reproductive",
"system",
"rna",
"interference",
"gonads",
"caenorhabditis",
"messenger",
"rna",
"animals",
"dna",
"transcription",
"germ",
"cells",
"animal",
"models",
"developmental",
"biology",
"oocytes",
"cae... | 2018 | The CSR-1 endogenous RNAi pathway ensures accurate transcriptional reprogramming during the oocyte-to-embryo transition in Caenorhabditis elegans |
Translation of hundreds of small ORFs ( smORFs ) of less than 100 amino acids has recently been revealed in vertebrates and Drosophila . Some of these peptides have essential and conserved cellular functions . In Drosophila , we have predicted a particular smORF class encoding ~80 aa hydrophobic peptides , which may function in membranes and cell organelles . Here , we characterise hemotin , a gene encoding an 88aa transmembrane smORF peptide localised to early endosomes in Drosophila macrophages . hemotin regulates endosomal maturation during phagocytosis by repressing the cooperation of 14-3-3ζ with specific phosphatidylinositol ( PI ) enzymes . hemotin mutants accumulate undigested phagocytic material inside enlarged endo-lysosomes and as a result , hemotin mutants have reduced ability to fight bacteria , and hence , have severely reduced life span and resistance to infections . We identify Stannin , a peptide involved in organometallic toxicity , as the Hemotin functional homologue in vertebrates , showing that this novel regulator of phagocytic processing is widely conserved , emphasizing the significance of smORF peptides in cell biology and disease .
Multicellular organisms contain specialised cells in appropriate parts of the body , performing tasks that allow the formation and maintenance of a fully functional organism . This specialisation relies upon the modification of basic cellular processes , for example enhanced cytoskeletal mechanics in muscle cells , or enhanced endocytic activity in phagocytic cells [1] . At the molecular level , such modifications rely on tissue-specific gene products that regulate specific cell biology and physiology pathways . These regulators offer great promise as specific therapeutic targets , yet for many tissues we still ignore their identity and mechanism of action . Some unidentified cell regulators might be proteins whose functions have not been investigated yet; alternatively , some might be encoded by noncanonical gene products , such as peptides encoded by small Open Reading Frames ( smORFs ) of less than 100 amino acids . smORFs have been largely disregarded by genome annotations and considered nonfunctional , but recently a number of ribosomal profiling and peptidomics studies have highlighted the apparent translation of hundreds of smORFs in the genomes of animals [2–4] . However , the functionality of these smORFs remains an open question , although a few smORFs have been studied and characterised functionally [5–8]; reviewed in [9 , 10] . Recently , we described a class of smORFs of about 80 codons long with a propensity to encode hydrophobic peptides with predicted alpha helix domains that localise to membranes and cell organelles [3] . The few examples of these smORFs with annotated function are widely expressed and involved in housekeeping processes , such as oxidative phosphorylation in mitochondria [3] , but in principle these hydrophobic smORFs have the capacity to act as regulators in other membrane-based cellular processes , as the sarcolamban/sarcolipin smORF family of calcium signalling regulators illustrates [7 , 11] . Here we characterize hemotin ( hemo ) , a tissue-specific smORF gene , which belongs to this class of smORFs , encoding an 88aa peptide with alpha-helical domains . hemo is expressed in hemocytes ( Drosophila macrophages ) , where it regulates endosomal maturation during phagocytosis , the specific function of this histotype . Hemocytes are the main component of the cellular branch of the insect immune system , and like vertebrate macrophages , they are professional phagocytes tasked with removing dying cells and microorganisms invading the body [12–15] . Although phagocytosis is a basic and ancestral cellular function that predates multicellularity , this function is greatly enhanced in these “professional” phagocytes . The molecular mechanisms underlying this cellular specialisation are actively studied , and have shown a surprising degree of conservation between insects and humans [16 , 17] . Central to phagocytosis seems to be the formation of the phagosome , a specialised endocytic vesicle containing the phagocytosed material [16] , and its subsequent maturation and degradation through the endolysosomal pathway [18] . This processing requires basic endocytic components but also specific proteins and regulators , whose identity and functions are not yet fully clarified [1 , 16] . Furthermore , pathogenic microorganisms are often able to override this cellular defence of phagocytes by interfering with the processing and maturation of the endo-phagolysosome [19] . Here , we show that regulation of endocytic maturation by Hemotin is essential for hemocytes to digest phagocytosed bacteria effectively . Removal of hemo compromises the ability of the animal to clear bacteria from the body and severely reduces lifespan . Molecularly , endosomal maturation requires “molecular labels” for the sorting of membrane vesicles into their appropriate endocytic compartments and their processing by fusion with appropriate organelles such as lysosomes . Some of these labels are provided by distinct phosphorylated states of Phosphatidylinositol ( PI ) . In early endosomes , PI is phosphorylated to form PI ( 3 ) P ( phosphatidylinositol-3-phosphate ) , which is required for endosomes to progress through the maturation process [18 , 20] . This phosphorylation step is mediated by PI ( 3 ) kinases such as class II PI3K68D ( Phosphatidylinositol 3 kinase 68D ) and class III Vps34 ( Phosphatidylinositol 3 kinase Vps34 ) , whereas the reverse dephosphorylation of PI ( 3 ) P is mediated by phosphatases such as Myotubularin ( Mtm ) [21] . Here , we show that Hemotin peptides bind and repress the adaptor protein 14-3-3ζ , and that in turn , 14-3-3ζ binds and promotes the function of PI3K68D . Thus , the hemo gene indirectly represses the PI3K68D-mediated labelling of early endosomes and this regulates subsequent steps of phagocytic processing . We also observe that this regulatory mechanism is conserved across evolution . We have identified Stannin ( Snn ) , encoded by a smORF of 88 codons , as the vertebrate homologue of Hemotin . The snn gene had been previously characterised as a mediator of cytotoxicity by organometallic compounds such as tri-methyl-tin ( TMT ) [22] . Our results show that Stannin is a functional homologue of Hemotin in fly hemocytes and mouse macrophages . Thus , we postulate that the previously unknown , nontoxic , and endogenous role of Stannin is also to modulate endosomal maturation by inhibiting the 14-3-3ζ-mediated stimulation of PI kinase function . Our results add to the body of evidence revealing the homology between the innate immune system of vertebrates and invertebrates [16 , 17] , by identifying a new conserved member of this system with an essential pathogen-fighting activity . Hemotin/Stannin is , thus , a new example of a smORF conserved across vast evolutionary distances and fulfilling an important cellular regulatory function , suggesting that the structure-function conservation of the sarcolamban-sarcolipin smORF family from invertebrates to humans [7] is likely not to be an exception , but an example of a wider trend .
We originally identified hemo in a bioinformatics search for putative functional smORFs in the Drosophila melanogaster genome [23] . hemo encodes a putative transmembrane peptide of 88 amino acids that matches a putative peptide fragment from a proteomics study of Drosophila S2 cell membrane extracts ( S1A and S1B Fig ) [24] . hemo originally mapped to a region of the genome devoid of any annotated gene , but subsequently , this locus has been annotated by the Drosophila Genome Project , and its structure is shown in Fig 1A . A potential polycistronic RNA is expressed by this locus and contains the 88aa hemo-ORF and a second short ORF ( Open Reading Frame ) of 59 codons ( ORF2 ) , currently annotated in Flybase as CG43194 and CG43210 , respectively . Poly-Ribo-Seq data from Drosophila S2 cells shows that hemo transcripts are actively translated ( Fig 1B ) [3] . We assayed expression of hemo throughout development by reverse transcription polymerase chain reaction ( RT-PCR ) and found that transcription of hemo is temporally regulated , showing higher expression from prepupal to adult stages , in agreement with modEncode RNAseq data ( S1 File;Flybase ) . However , in situ hybridisation also showed that hemo is specifically expressed in Drosophila embryos in a pattern similar to crq , a hemocyte-specific marker ( Fig 1C–1F ) [26–28] . Accordingly , hemo was detected by RT-PCR from prepupal hemocyte RNA extractions ( S1C Fig ) . Hemocytes are fly macrophages involved in the engulfment of cellular corpses throughout development , and they are the cellular branch of the fly innate immune system [12] . To elucidate the function of hemo , we generated a deletion of the locus by FRT-mediated recombination using the P ( RS3 ) fray CB-0706-3 and PBac ( WH ) fru f02684 P-element insertions [29] . This deletion ( hemoA4 ) removes the entire hemo locus , CG7691 , and also one 5’ noncoding exon of each of the frayed ( fray ) and fruitless ( fru ) genes ( Fig 1A and S1C Fig ) . Homozygous hemoA4 mutants reached adulthood and did not show overt morphological defects; however , their hemocytes exhibited a more vacuolated morphology in comparison to wild-type controls ( Fig 1G and 1H and S1D and S1E Fig ) . The vacuolated phenotype in hemoA4 mutant hemocytes was reflected in the microtubule cytoskeleton , and the volume of the vacuoles was significantly larger than those present in wild-type hemocytes ( Fig 1I , 1J and 1O ) . Further genetic analysis corroborates that this hemocyte phenotype in hemoA4 is specific to loss of the Hemotin peptide . The fru and fray genes have functions in dimorphic sexual behaviour and nerve fasciculation , respectively [25 , 30] . We discarded them as candidates for providing the requirement for hemocyte vacuolation since fru was not expressed in hemocytes ( S1 File ) , and the function of fray is provided by a shorter transcript containing the coding exons ( Fig 1A ) , whose expression was not affected in the hemoA4 deletion ( S1 File ) . In agreement with this , the hemoA4 and frayPZ07551 mutations ( Fig 1A ) [25] complemented each other giving rise to hemocytes with normal vacuoles ( Fig 1O ) and larvae with normally fasciculated nerves ( S2 File ) . For the CG7691 gene , we generated a 12 Kb genomic construct containing its coding and upstream regulatory regions . Hemocytes carrying this CG7691 genomic fragment ( GF ) in a hemoA4 mutant background expressed CG7691 at wild-type levels ( S1 File ) but still possessed large vacuoles ( Fig 1O ) , indicating that deletion of CG7691 in hemoA4 is not the cause of hemocyte vacuolation . In contrast , loss of function and rescue experiments pinpoint the hemo transcript as responsible for these vacuolation phenotypes . First , reduction of hemo RNA expression with a ds-RNA construct ( UAS-hemo-RNAi ) in hemocytes produced enlarged vacuoles mimicking the hemoA4 phenotype ( Fig 1K and 1O ) . Second , expression of the hemo full-length transcript rescued hemoA4 phenotypes ( Fig 1L and 1O ) , whereas expression of the same construct carrying a frame-shift in each ORF , which produces scrambled peptides , did not ( Fig 1O ) . Next , we assessed the contribution of each ORF to hemo function . Expression of UAS-minigenes containing hemo-ORF alone ( Fig 1M and 1O ) or hemo-ORF-GFP ( green fluorescent protein-tagged peptides ( hemo-GFP ) ( Fig 1O ) rescued the vacuolated hemoA4 phenotype , whereas UAS-ORF2 did not ( Fig 1B , 1N and 1O ) , indicating that the peptide encoded by hemo-ORF represents the functional unit of the gene in this context . Poly-Ribo-Seq data from Drosophila S2 cells [3] indicated high translation of hemo-ORF but a lower , or non-productive , rate of ORF2 translation ( Fig 1B ) . Finally , carboxyl terminal-tagged GFP fusions of each peptide in a full-length transcript only showed expression of Hemo-GFP , which localized to intracellular membrane structures in S2 cells ( S2 File ) and Kc167 cells ( S1F–S1F” Fig ) . We could only detect ORF2-GFP peptides from an ORF2-GFP fusion in a short transcript minigene that excludes hemo-ORF ( Fig 1B; S2 File ) . Thus , although ORF2 does show potential for translation , altogether our results support that the 88aa transmembrane peptide encoded by hemo-ORF alone provides the requirement for normal vacuolation in Drosophila hemocytes . We next investigated the origin of the mutant vacuoles . hemoA4 hemocytes showed an increased accumulation of acidic compartments as revealed by pH-sensitive Lysotracker ( S1G , S1H and S1K Fig ) , and vacuoles disrupting the microtubule cytoskeleton contained the lysosomal marker LAMP1 ( Figs 2A–2A” and S2A–S2A” ) . Expression of hemo full-length or hemo-ORF constructs in hemoA4 mutants rescued the enlarged acidic compartment phenotype ( S1I–S1K Fig ) . These pieces of evidence suggest that the enlarged vacuoles are some type of abnormal degradation compartment and that hemo is necessary for the processing of phagocytic or recycled materials . We have used further endocytic markers to ascertain the nature and integrity of the acidic compartments in hemoA4 null hemocytes . The hemoA4 enlarged acidic intracellular organelles show an extensive overlap of early and late endocytic markers: FYVE ( a PI ( 3 ) P-binding zinc finger domain , early endosomal marker , named after being found in Fab1 , YOTP , Vac1 , EEA1 [31] ) ( Fig 2B and 2B’ and 2E and S2E and S2E’ Fig ) , Rab7 ( late endosomal marker ) ( S2C–S2C’ , S2E–S2E’ Fig ) , and Lysotracker ( acidic organelle marker ) ( Fig 2B , 2B” and 2E and S2C and S2C” Fig ) in comparison with wild-type hemocytes ( Fig 2C–2C” and 2E and S2B–S2B” and S2D–S2D” Fig ) . This combination of early and late markers identifies the enlarged compartments as aberrant endolysosomes [21] . We quantified the occupied FYVE area index ( see Materials and Methods; Fig 2D ) and the average diameter of FYVE vesicles ( see Materials and Methods; S6 Fig ) . hemoA4 mutant hemocytes showed larger FYVE compartments containing larger vesicles than wild-type . Expression of UAS-hemo full-length transcript , UAS-hemo-ORF , and UAS-hemo-GFP constructs rescued these phenotypes , whereas expression of a UAS-hemo frame-shift , UAS-ORF2 and a CG7691 GF constructs failed to do so ( Fig 2D ) . The simplest interpretation of our results is that Hemotin peptides are required for completing the maturation of at least some early-to-late endosomes , and hence for their subsequent endolysosomal degradation . Other disruptions of this traffic could also perturb endosomal maturation , since endolysosomal homeostasis depends on the balance between trafficking of membrane from the endocytic pathway ( influx ) and exit of membrane ( efflux ) into early recycling compartments [18] . We explored whether the membrane efflux to early recycling compartments was disrupted in hemoA4 mutants by visualizing Rab11 , a marker of recycling endosomes . In hemoA4 mutant hemocytes , the size of Rab11-positive compartments seemed normal in comparison with wild-type ( S2F–S2F’ and S2G–S2G’ Fig ) , suggesting that membrane efflux to early recycling compartments is not grossly affected . Thus , the role of Hemotin peptides seems to focus at promoting early to late endosomal maturation . To corroborate this hypothesis , we ascertained the precise subcellular localisation of Hemotin peptides in hemocytes . The Hemo-GFP peptides , which are functional and able to rescue hemoA4 mutants ( Fig 1B and 1O and Fig 2D ) , localised to early endosomes , as shown by their colocalization with the FYVE marker ( Fig 2F and 2G–2G” ) and HRS proteins [32] ( S2H–S2H” Fig ) but only limited colocalization with the late endocytic marker Lysotracker ( Fig 2F and 2H–2H” ) . Thus , the in vivo Hemotin peptide localization correlates with their inferred role in promoting endosomal maturation . Hemotin’s requirement in endolysosomal maturation could be related to the digestion of phagocytic materials and hence have an effect on immune defence and individual survival . We analysed ex vivo the ability of hemocytes to phagocytise and digest bacteria by monitoring uptake of pHrodo-E . coli bacterial particles , which fluoresce in acidic compartments such as endolysosomes , while simultaneously revealing early endosomes with the FYVE marker ( Fig 3A–3E; S1 and S2 Videos ) . hemoA4 mutant hemocytes internalised pHrodo particles at similar rate to wild-type hemocytes ( S3A Fig ) , indicating that the initial phagocytic uptake is not affected . However , in the hemo mutants , the internalised particles acidified at a slower rate and to a lesser degree , and remained for longer in FYVE-positive vesicles ( Fig 3A–3E ) , suggesting a reduced ability of hemoA4 mutant hemocytes to digest phagocytosed material . Such impairment could compromise the mutant fly’s ability to deal with invading and commensal microorganisms , especially for those usually cleared through phagocytosis by hemocytes [33 , 34] . Indeed , hemoA4 mutant hemocytes also showed in vivo a similar normal uptake but slower processing of intact E . coli bacteria expressing mCherry ( Fig 3F–3J ) , and homozygous hemoA4 mutants carried a higher bacterial load than wild-type flies raised simultaneously in the same vial ( Fig 3K and 3L ) . This increased bacterial load seems to affect lifespan , since hemo mutants had a median life span that is only 47% of the wild-type ( Fig 3M ) in normal cultures , but this increases to 77% when raised in germ-reduced media supplemented with antibiotics ( Fig 3M ) . Furthermore , hemoA4 mutant flies show reduced resistance when infected with normally nonpathogenic E . coli bacteria , comparable to other mutants that have a reduced capacity for phagocytosis [33 , 34] ( S3B and S3D Fig ) . These results suggest that hemo mutants have a reduced cellular immunity . However , their humoral immunity ( driven by antibacterial peptides [13] ) does not seem affected . hemoA4 median life span after infection with pathogenic bacteria such as Micrococcus luteus and Enterobacter cloacae , which overcome hemocytes but fully engage the humoral production of antibacterial peptides [13 , 33 , 34] , was not significantly affected ( S3B , S3C and S3D Fig ) , and accordingly the production of antibacterial peptides in hemoA4 mutants was not impaired ( S1 File ) . Thus , our results suggest that Hemotin peptides in the early endosomes of fly macrophages are required for normal phagocytic processing and that absence of these peptides results in abnormal and slower maturation and degradation of phagocytic materials . In turn , this abnormal phagocytic processing reduces the ability of the organism to fight off bacteria and has a direct impact on fly immunity and viability . To elucidate the molecular function of the Hemotin peptide , we searched for similarity to known structural domains using the Phyre2 engine [35] . The human Stannin peptide is encoded by an 88aa smORF-like Hemotin and appeared as a possible match ( Fig 4A ) . Structural analyses of the Stannin peptide suggest a transmembrane peptide containing two alpha helices , an N-terminal one that spans the lipid bilayer and a C-terminal helix at the cytosolic side ( Fig 4A ) [22 , 36] , and our independent analysis of the Hemotin sequence using a transmembrane topology prediction program [37] also revealed a very similar potential transmembrane alpha-helical domain ( Fig 4A , S1B Fig ) . Threading the Hemotin sequence onto the predicted human Stannin tertiary structure ( see Materials and Methods ) confirms a structural compatibility similar to vertebrate members of the Stannin family ( Fig 4C ) . We further searched for homologues of Hemotin and of Stannin following a bioinformatics pipeline used in Magny et al . , 2013 [7] . We identified homologues of Hemotin peptides in other dipterans ( mosquitoes ) and in other insects such as hymenopterans ( bees , ants , and wasps ) , plus new homologues of Stannin in ancestral vertebrates , such as hagfish , lamprey , and cartilaginous fishes ( Fig 4A and 4B and S4A and S4B Fig ) , and these sequences also show structural compatibility with Stannin ( Fig 4C and S4C Fig ) . Despite considerable amino acid sequence variation , comparison of Hemotin and Stannin sequences reveals conservation in the alpha-helices and in the intervening sequence , which have been implicated in Stannin function [22] ( Fig 4A ) . The resulting tree of Stannin and Hemotin sequences shows a good correlation with the animal phylogeny ( Fig 4B ) and correctly locates the 88aa mitochondrial ribosomal protein S21 as an outgroup ( S4B Fig ) . Altogether , the amino acid sequence analyses suggest that that Stannin and Hemotin are members of the same peptide family , displaying sequence and structural homology . Stannin is a peptide involved in organometallic toxicity , but its endogenous physiological and cellular functions have remained elusive [22] . In rodent models , snn expression has been detected in hematopoietic organs and immune cells including macrophages [39–41] . Similarly as with Hemotin ( see above; unp . obs . ) , Flag-tagged Snn peptides have been detected in membrane fractions of intracellular compartments , such as endoplasmic reticulum , peroxisomes , mitochondria , and endosomes in mouse NIH-3T3 cell lines [42] . To explore the endogenous cellular function of hemo’s vertebrate relative , snn , in a vertebrate innate immune cell context , we used mouse RAW264 . 7 ( macrophage-like ) cells [43] . Firstly , we confirmed expression of snn in this cell line ( S5A and S5B Fig ) . Secondly , we knocked down the expression of snn in RAW264 . 7 cells using siRNAs ( small interfering RNAs ) ( S5A and S5B Fig ) and monitored the formation of acidic compartments with Lysotracker . snn si-RNA-treated RAW 264 . 7 cells showed abnormally large acidic aggregates ( Fig 5A–5C ) mimicking the hemoA4 null phenotype observed in hemocytes ( Fig 2A” and 2L” and S1G–S1K Fig ) . Expressing a GFP-tagged human Stannin ( Snn-GFP ) peptide in Drosophila Kc167 cells revealed its localisation to cellular membranes , comprising intracellular punctate compartments ( S5D–S5D” Fig ) . In addition , coexpression of tagged Stannin and Hemotin peptides in these cells show their colocalisation in membrane intracellular organelles ( Fig 5D–5D” ) . These results suggest that Stannin and Hemotin peptides localize to similar cellular compartments and are involved in the formation of acidic compartments , suggesting a functional homology . To test this possibility , we measured the ability of Stannin to rescue hemoA4 mutant phenotypes in hemocytes: expression of the human Stannin peptide rescued the vacuolated phenotype ( Fig 5G ) and the size of the FYVE-positive vesicles ( Fig 5H , 5I , 5J and 5L; S6 Fig ) of hemoA4 mutants almost as effectively as endogenous Hemotin peptides ( Fig 5G , 5H , 5I , 5J and 5K; S6 Fig ) , supporting that hemo and snn are functional homologues . Stannin peptides bind to the 14-3-3ζ adaptor protein , although the molecular and cellular implications of this binding have not been fully clarified [44] . 14-3-3ζ proteins form dimers that bind to phosphorylated amino acid residues of target proteins and modulate their functions [45] . We investigated whether a similar interaction exists between Hemotin and 14-3-3ζ during endosomal maturation in hemocytes . We observed that Hemo-GFP peptides ( which localize to early endosomes , Fig 2G–2G” and S2H–S2H” Fig ) colocalised with 14-3-3ζ in hemocytes ( Fig 5E–5E”‘ ) . In addition , Hemo-GFP peptides coimmunoprecipitated with Nt-tagged 14-3-3ζ proteins expressed in Drosophila Kc167 culture cells ( Fig 5F and S5C Fig ) and hemocytes ( S1 File ) . The coimmunoprecipitation appears moderate but reproducible and specific , perhaps reflecting a moderate affinity , or a limited amount of the multifunctional 14-3-3ζ protein [45] being engaged by Hemotin peptides . Our genetic analysis reveals a negative functional relationship between 14-3-3ζ and hemo/snn during endosomal maturation in hemocytes . Thus , excess of 14-3-3ζ function ( in He-Gal4;UAS-14-3-3ζ hemocytes ) resembled loss of hemo function ( as in hemoA4 ) by producing large vacuoles ( Fig 5G ) and FYVE vesicles ( Fig 5H , 5I , 5J and 5M ) , thus suggesting that 14-3-3ζ and hemo work in opposite directions . An antagonistic , yet closely related , function is further suggested by gene dosage interactions . First , the 14-3-3ζ gain of function phenotypes were corrected by simultaneous gain of either hemo or snn: both UAS-hemo and UAS-snn corrected the extra vacuolation of UAS-14-3-3ζ ( Fig 5G ) , while UAS-hemo corrected the large FYVE vesicles produced by UAS-14-3-3ζ . Second , reducing 14-3-3ζ function rescued the hemoA4 phenotype to near wild-type , as indicated by vacuolation ( after removing a copy of 14-3-3ζ in a hemoA4 homozygous background , Fig 5G ) and the size of FYVE vesicles ( by expressing 14-3-3ζ RNAi in a hemoA4 homozygous background , Fig 5J and 5N ) . These negative dosage interactions suggest that 14-3-3ζ works in a common pathway , yet antagonistically , with hemo and snn [46 , 47] . Because 14-3-3ζ dosage is able to modify hemoA4 null phenotypes , the formal interpretation is that 14-3-3ζ acts downstream of hemo , or in other words , that Hemotin regulates 14-3-3ζ . Altogether , these genetic results suggest that Hemotin function represses or down-regulates 14-3-3ζ activity . Thus , our results suggest that Hemotin and Stannin are functional homologues that are required at the cellular level for endosomal maturation; and at the molecular level , to bind and antagonize 14-3-3ζ . Specific steps in the endosomal maturation process depend on the phosphorylation states of PI [18 , 20] . In hemocytes , regulation of PI ( 3 ) P endocytic pools is key for early endosomal trafficking [21 , 48] . It has been shown that PI ( 3 ) P homeostasis depends on the class II PI ( 3 ) Kinase , PI3K68D , which phosphorylates PI at the carbon 3 position to form PI ( 3 ) P and the Mtm phosphatase that dephosphorylates this residue to revert to PI [21] . Either mutations in mtm or overexpression of PI3K68D increase PI ( 3 ) P in early endosomes and induce the formation of abnormal enlarged endolysosomal compartments retaining larger amounts of the PI ( 3 ) P sensor FYVE-GFP ( Fig 6E , 6F and 6J ) [21] , similar to that observed in hemoA4 mutants ( Fig 2B–2B” , S2C–S2C” and S2E–S2E” Fig ) . The similarities of the cellular phenotypes between hemoA4 mutants and overproduction of PI ( 3 ) P by PI ( 3 ) P enzymes prompted us to carry out a genetic analysis of their functional interactions during endosomal maturation in hemocytes . The PI3K68D gain of function phenotype of enlarged FYVE organelles and vesicles was suppressed by coexpression of either a hemo full-length transcript ( Fig 6J and S6 Fig ) or a 14-3-3ζ-RNAi construct ( Fig 6H and 6J and S6 Fig ) . Similarly , reduction of mtm function by expressing an mtm-RNAi construct produced aberrant enlarged FYVE compartments with larger-sized vesicles ( Fig 6F and 6J; S6 Fig ) [21] . Codepletion of mtm and 14-3-3ζ with RNAi constructs corrected the size of these abnormal FYVE compartments and vesicles ( Fig 6I and 6J and S6 Fig ) . Formally , these results indicate that 14-3-3ζ cooperates with PI3K68D , whereas Hemotin and Mtm antagonize their action . Three further results suggest that mtm and PI3K68D act downstream of Hemotin and 14-3-3ζ . Either overexpression of mtm or removing a copy of the PI3K68D corrected the abnormally large FYVE vesicles of hemoA4 null hemocytes ( Fig 6A–6C and 6J and S6 Fig ) . Similarly , reducing PI3K68D function with a RNAi construct produces little or no phenotype on its own ( Fig 6J ) [21] but is also able to rescue the enlarged size of FYVE vesicles produced by overexpression of 14-3-3ζ ( Fig 6G and S6 Fig ) . Thus , in three different genetic conditions , a reduction of PI ( 3 ) P synthesis was able to suppress the enlarged and abnormal early-endosome-like vesicles produced by total loss of hemo or gain of function of 14-3-3ζ . These epistatic results strongly suggest that mtm and PI3K68D act downstream of Hemotin and 14-3-3ζ during endosomal maturation; in other words , that Hemotin and 14-3-3ζ fulfil their roles through regulation of the PI ( 3 ) P enzymes . Two independent lines of evidence corroborate this hypothesis . First , Hemo-GFP peptides colocalize with PI3K68D kinase at early endosomes ( Fig 6K–6K”‘ ) . Second , PI3K68D is able to pull down 14-3-3ζ from hemocyte protein extracts ( Fig 6L ) , suggesting that 14-3-3ζ directly binds PI3K68D . Altogether , the genetic , cellular , and biochemical results support a model where Hemotin peptides indirectly affect the PI ( 3 ) P labelling of early endosomes by binding 14-3-3ζ , and hence repressing the positive effect of 14-3-3ζ on PI3K68D kinase ( Fig 7 ) .
Hemocytes are part of the innate immune surveillance system in Drosophila and are involved in the uptake of cell corpses during development and in overriding bacterial infections throughout the life cycle of the fly [12 , 49] . Their ease of detection and observation , coupled with the arsenal of Drosophila genetic techniques , makes hemocytes an excellent model system in which to characterize at molecular level basic cellular processes shared by other metazoan cells . For example , insect hemocytes display functions similar to vertebrate macrophages , such as directed migration and phagocytosis [12–15] . Similarities in the molecular mechanisms controlling these processes have been noted before [16 , 43 , 50] , but the question of whether hemocytes are truly homologous to vertebrate white blood cells , representing a kind of ancestral macrophage-like cell , is not settled . Homology is supported by the clear similarities in the recognition of exogenous microorganisms through pattern recognition receptors of the Toll and Imd signalling pathways [17 , 51 , 52] . Our work further supports such homology through the identification of a modulator of endosomal maturation that is specifically expressed in hemocytes and vertebrate white blood cells and is essential for their phagocytic activity . This modulator is the Hemotin-Stannin peptide . During endosomal maturation , turnover of PI ( 3 ) P pools at endocytic compartments is essential for PI ( 3 ) P-mediated recruiting of effectors involved in sorting specific cargo proteins and endosomal vesicles to specific trafficking routes ( Fig 7A ) [18 , 20] . In hemocytes , PI ( 3 ) P synthesis at early endosomes depends on a scaffold complex comprising both PI ( 3 ) P-turnover enzymes , PI3K68D kinase , and Mtm phosphatase bridged by Sbf ( an Mtm-pseudophosphatase ) , indicating that PI ( 3 ) P synthesis must be tightly regulated and spatially linked to the traffic of membrane vesicles through the endosomal pathway towards degradation [21 , 50] . It makes sense that the function of these complexes must be tightly controlled in cell types involved in intense phagocytosis , such as hemocytes , in order to regulate PI ( 3 ) P-turnover and maintain endosomal homeostasis under the stress of enhanced membrane influx . Our results show that Hemotin peptides are specifically expressed in hemocytes and locate to early endosomes , where they bind and repress 14-3-3ζ activity . In turn , the genetic and biochemical results suggest that Hemotin-Stannin peptides prevent 14-3-3ζ from assisting the class II PI3K68D kinase at these endocytic vesicles . While the genetic suppression of the hemo phenotype by 14-3-3ζ and PI3K68D indicate the functional link with these proteins , this could be based on direct or indirect effects . Our ability to co-IP the two proteins is consistent with a direct effect . However , further studies are needed to confirm this and to determine how Hemotin and 14-3-3ζ can together regulate PI3Kinase activity . Thus , our results indicate that 14-3-3ζ binds PI3K68D , but do not reveal the molecular function of this binding , nor the nature of its positive impact . 14-3-3ζ could either directly stimulate the enzymatic activity of PI3K68D or promote the recruitment or retention of PI3K68D at the Sbf complex . In any case , removal of Hemotin or Mtm , or excess of either 14-3-3ζ or PI3K68D result in large PI ( 3 ) P-labelled vesicles ( as revealed by the PI ( 3 ) P sensor FYVE-GFP , Figs 2 and 5 ) , as observed in other systems when PI ( 3 ) P kinase function is deregulated [18 , 21] . These enlarged PI ( 3 ) P vesicles display mixed endosomal-lysosomal characteristics , and thus the normal transitory overlap of endosomal and lysosomal markers in the cell appears increased ( Fig 2 , S2 Fig , Fig 7B ) . These enlarged intermediate vesicles could occur when the increase in early endosomal vesicle formation overpowers the ability of the subsequent endosomal machinery to complete their maturation into proper endolysosomes . These mutant endosomal vesicles seem to have an impaired ability to process phagocytic material , which displays slower and less acute acidification and longer trafficking times ( Fig 3 ) . The PI ( 3 ) P-rich nature of the mutant vesicles might preclude the formation of proper endo-lysosomes , but the process of endolysosomal acidification and maturation is still not completely understood [18 , 53] . We propose that the cellular role of Hemotin is to reduce the endosomal PI ( 3 ) P labelling produced by the 14-3-3ζ-PI3K68D-Sbf-Mtm complex , and hence to reduce production of vesicles with early endosome characteristics ( Fig 7 ) . This modulation by Hemotin would allow for the subsequent maturation and degradation part of the endophagocytic cycle to keep up with the intake of phagocytic material , which is unaffected in hemo mutants . This “endocytic modulation” appears essential in cells with high phagocytic activity such as hemocytes and macrophages to ensure proper digestion of phagocytised bacteria . In the absence of hemo , commensal and invading bacteria are less efficiently cleared from the body , and this ultimately compromises the viability of the organism ( Fig 3 , S3 Fig ) . Interestingly , much of this effect can be attributed to bacteria that are commensal or else not normally pathogenic , which are normally kept in check by the hemocytes , but that when this cellular defence is compromised , they overgrow in the body and lead to death ( Fig 3 , S3 Fig ) [33] . This has interesting parallels with the severe effects of infection by normally mildly pathogenic bacteria in human patients with immunodeficiencies [54] . Regulation of endosomal maturation by Hemotin-Stannin appears to be a conserved mechanism . We have identified Hemotin homologues in other insects , and our phylogenetic analyses indicates that Hemotin is a member of a conserved peptide family , including the vertebrate Stannin , encoding peptides with similar tertiary structure ( Fig 4 , S4 Fig ) . In addition , we have demonstrated that Hemotin and Stannin are functional homologues in hemocytes during endosomal maturation , a role that appears to be conserved in vertebrates , as hemo mutant-like endosomal abnormalities are produced by reduction of snn function in mouse macrophage-like cells ( Fig 5 ) . Finally , we have shown that both hemo and snn display antagonistic genetic relationships with 14-3-3ζ during endosomal maturation . The binding of Hemotin to 14-3-3ζ offers a direct molecular basis for these interactions ( Fig 5 ) . This seems a conserved regulatory mechanism as Stannin peptides have been shown to bind 14-3-3ζ proteins in vertebrates [44] and physical interactions between 14-3-3ζ and the vertebrate PI ( 3 ) P enzyme homologues , Mtmr1 phosphatase and class II PI ( 3 ) P kinase have been reported by proteomics of 14-3-3ζ pull-down protein extracts from human HeLa cells and mouse neural cells [55 , 56] . Given 14-3-3ζ multiple functions and near-ubiquitous expression , we surmise that cooperation of 14-3-3ζ with the PI3K68D kinase might be a general cellular mechanism that is modulated by Hemotin/Stannin in specific cell types , although further experiments must test this point . Interestingly , results relating to the role of Stannin as a promoting factor for organometallic-mediated cytotoxicity may be relevant to our model . Organotins such as TMT are used in industry as plastic stabilisers , but they are acutely toxic , producing cytotoxicity in specific tissues and eventually death of the affected individual [22] . snn was originally identified as a cDNA specifically expressed in TMT-sensitive tissues such as hematopoietic organs and immune system lineages [57] . Subsequent studies showed binding of TMT to Stannin [58–60] and of Stannin to 14-3-3ζ [44] . Although the molecular and cellular consequences of such binding and the roles of Stannin in TMT toxicity and under normal physiology have not been fully elucidated , several results might suggest a relationship with PI metabolism . The first cellular symptom of exposure to organometallic compounds is an increase of endolysosomal-like vacuoles [61 , 62] , which resembles both the hemo mutant phenotype and the snn siRNA phenocopies in mouse macrophage-like cells ( S1G and S1H Fig , Fig 5A and 5B ) . Lithium treatment , which inhibits PI synthesis , has a protective effect on TMT cytotoxicity [62] , but addition of exogenous PI ( 3 ) P ( myophosphatidylinositol ) to such TMT+Lithium-treated cells reverts this Lithium protective effect , altogether suggesting that PI promotes TMT-mediated cytotoxicity [62] . Thus , TMT could alter Stannin and cause a toxic excess of 14-3-3ζ-related PI ( 3 ) Kinase function , leading to an increase of endocytic PI ( 3 ) P labelling , and hence to enlarged endolysosomal compartments and eventual apoptosis in snn-expressing tissues . We have not been able to identify Hemotin-Stannin homologues in diblastic animals or unicellular eukaryotes , so a compatible hypothesis is that Hemotin-Stannin appeared as an adaptation to enhanced levels of endocytic activity in specific phagocytic cells during the evolution of complex body plans . Interestingly , this is also the point when the Sarcolamban family appears in the animal tree [7] . The sarcolamban-sarcolipin smORF family also encodes peptides regulating a basic cellular process ( Calcium homeostasis ) fundamental to a specific cell type ( muscle cells ) [11 , 63] . smORFs are widespread , having been identified in bacteria [64] , yeast [65] , and plants [66] . In eukaryotes , translated smORFs have been recently identified in putative long noncoding RNAs [2 , 3] , and interestingly , expression of long noncoding RNAs can be highly tissue-specific ( [67] , unp . obs . ) . We have shown that hydrophobic smORF peptides have a propensity to encode transmembrane alpha-helices and to localise to cell membranes and organelles [3] . Finally , it has been proposed that smORFs are a source of evolutionary new peptides [68] . Altogether , it is tempting to speculate that hydrophobic smORFs could provide a source of emerging tissue-specific modulators of organelle-based cellular processes . Regardless of this speculation , the accumulating evidence points to the potential of smORF peptides to fill gaps in our understanding of cell biology and physiology , and its associated diseased states .
Fly stocks and crosses were cultured at 25°C . The Oregon-Red line was used as our wild-type strain . The following lines were obtained from the Bloomington Stock Centre: frayPZ07551 , Df ( 3 ) BSC626 ( PI3K68D ) , fru1 , 14-3-3ζBL12 , UAS-Rab7-YFP , UAS-2XFYVE-GFP , UAS-LAMP1-GFP , UAS-PI3K68DIR ( GL00159 ) . For ectopic and rescue experiments , we used Hemese-Gal4 , and crq-Gal4 lines as described in Sampson et al . , 2013 [15] . Fly strains used in this study: UAS-HA-14-3-3ζ and UAS-14-3-3ζIR [69] , UAS-14-3-3ζIR ( VDRC#48725 ) . UAS-mtmIR , UAS-mtm-Cherry , UAS 2XFYVE-Cherry , UAS-PI3K68D-Cherry , and UAS-PI3K68D-GFP [21] . frayR1 [25] . The P{RS3}fray ( CB-0706-3 ) and the P-Bac{WH}fru ( f02684 ) from DGRC and Exelixis , respectively , were used to generate a 34-Kb deficiency by FRT-mediated recombination [29] depleting the first 5’exon of fru and fray genes and the whole gene locus of CG7691 and CG43210 ( hemo ) genes , which was confirmed by Taq-polymerase PCR ( Qiagen , Venio , Netherlands ) from genomic DNA extractions . In addition , detection of mRNA levels of these genes were conducted by RT-PCR from mRNA extraction using Trizol ( Ambion/Life Technologies , Carlsbad , CA ) of hemoA4 larval haemolymph . Primary Drosophila hemocytes were isolated from postembryonic life stages as indicated in Sampson et al . , 2013 [15] . Extraction of free-flowing hemocytes was achieved by bleeding individual specimens using a 25-gauge dissecting needle in a culture medium ( 80% Schneider’s Drosophila medium [Pan-Biotech , Dorset , UK] and 20% fetal bovine serum [Invitrogen/Life Technologies] with no antibiotics ) . For measuring vacuolation primary hemocytes from white prepupae ( >100 h AEL ) were cultured into a glass-bottom dish containing collagen extracellular matrix ( MatTek Corp . , MA , US ) and stained with anti-beta tubulin and phalloidin ( Invitrogen/Life Technologies ) following the procedures described in Sampson et al . , 2013 . Similar procedures were followed to identify lysosomes using lysotracker ( Invitrogen/Life Technologies ) or by detecting the LAMP1-GFP marker . Detection of GFP and Cherry–2XFYVE , Hrs , Cherry-PI3K68D , Cherry-Mtm , HA-14-3-3ζ , and hemo-ORF1-GFP was achieved by culturing primary hemocytes from wandering late third instar larvae and white prepupa stages in glass multisport microscope slides ( Hendley-Essex , Essex , UK ) and immunostaining was performed as described in Sampson et al . , 2013 . Antibodies used in this study are: mouse anti-GFP ( Roche ) 1:500; rabbit anti-cherry ( Invitrogen ) 1:5000; mouse anti-HA ( Roche ) 1:500; anti-Hrs [70] , mouse anti-FLAG M2 ( SIGMA ) 1:1000 . Actin cytoskeleton was labelled with fluorophore-conjugated phalloidin ( Molecular-Probes . Invitrogen ) 1:50 . For measuring lysosomal aggregation , live primary hemocytes were stained with 50 nM Red-lysotracker ( Invitrogen ) in culture medium for 15 min and then washed several times and finally replaced with normal culture medium for live imaging for 20–30 min . For embryonic hemocyte in vivo imaging , adult flies were allowed to lay eggs on apple juice agar plate . Embryos collected at stage 15 , dechorionated and mounted on double-sided tape stuck to a standard glass slide . Voltalef oil was applied to the embryos to prevent dehydration , and a coverslip fixed with nail polish applied on top . For pupal hemocyte in vivo imaging , pupae at stage P6 ( 50 h APF ) were mounted ventral side down on double-sided tape applied to a glass slide and dissected to remove a window in the pupal case over the thorax as described in Sampson et al . , 2013 . After application of 10S Voltalef Oil on the ROI , a ring of petroleum jelly was made around the samples , a coverslip was rested on this ring , and then it was pressed down on the sample . Confocal laser microscopy was used to image fluorescence in live and fixed cells . An inverted Zeiss Axiovert 200M series microscope with a LSM-510 confocal laser attachment was used . Images were captured using a Hamamatsu ORCA-ER C4742-95 camera . Most cells were observed at 63x using the Zeiss Apochromat 63x oil objective NA = 1 . 4 , but in some cases 100x magnification was required , using a Zeiss Epiplan—NeoFluar 100x oil objective with NA = 1 . 3 . Fluorescent imaging utilised the automated prior stage for Z-stacking throughout the cells imaged . Live cells were imaged at 0 . 5–1 μm path length between each imaging slice . Fixed cells were also imaged within the same range of path length for nondetailed images , whilst detailed images , particularly for 3-D reconstruction , were imaged at a path length of 0 . 2–03 μm between each imaging slice . LSM image browser and imageJ v . 1 . 46 were used to analyse acquired images and Photoshop ( Adobe ) was used for editing . Phagocytosis assays were performed on wild-type and hemoA4 mutant live primary hemocytes expressing FYVE-GFP endocytic marker supported in ex vivo cell culture without ECM coating and using HEPES+HBSS pH 7 . 4 . The phagocytic stimulation used was in the form of pHrodo pH-sensitive conjugated heat-killed bacterial particles . E . coli ( Gram - ) bio-particles were used ( Life Technologies ) . A 1:1 dilution of pHrodo-conjugated bacterial particles ( 0 . 5 mg/mL ) with HEPES+HBSS pH 7 . 4 buffer was sonicated for 5 min to break apart large conglomerates . Further dilution was carried out by using 16 μl ( 1:1 dilution ) into a final volume of 200 μl HEPES+HBSS pH 7 . 4 which was added to the ex vivo hemocyte culture . Time-lapse Imaging was conducted using a SP8 Leica confocal microscope with a short Z-stack throughout the cell at 0 . 5 μm path length between each image slice . Bright Field Phase , 488 and 543 nm laser illuminations were used to observe the cell outline and the fluorescent FYVE and pHrodo particles . Live cells were imaged every 90 to 120 seconds with a Z-stack for a period lasting for 90 min . The integrated pHrodo intensity over time was measured for each particle from the moment the particles docked into the cell membrane and for each time point , using an identical ROI of 0 . 9 μm2 for every series on single confocal z-stack plane across on the middle of the pHrodo particle . To quantify FYVE prevalence , the number of frames for which individual pHrodo particles were surrounded by FYVE , using a single z-stack plane across on the middle of the pHrodo particle , were counted and converted into time according to the frame rate of each time lapse . For viability assays , virgin male flies were collected from each genotype and kept at 18°C until sufficient numbers were reached ( maximum 2 d , at least 50 flies per genotype or condition ) . Flies ( 2-d-old ) were then kept at a density of 10–25 flies/tube , reared at 25°C and scored daily for number of surviving flies . To assess the viability of flies in antibiotic media , 100 μL of Penicillin-Streptomycin solution ( 10 , 000 units penicillin and 10 mg streptomycin/mL , Sigma-Aldrich ) was added to the surface of standard corn-meal food media vials ( containing approximately 10ml of media ) , and left over night at 25°C , until fully absorbed . To assess the effect of bacterial infection on viability , nonpathogenic Escherichia coli ( DH5α ) , or pathogenic Micrococcus luteus ( gram+ ) and Enterobacter cloacae ( gram− ) bacterial strains were used . Flies were infected by wounding at the top of the abdominal segment , under the haltere , using a tungsten dissection needle dipped in bacterial culture grown to a cell density of ( OD600 = 0 . 8 ) . To quantify the number of mCherry E . coli bacteria ( K12 E . coli expressing PDSpRSETD-cherry plasmid from Dr . Stephan Mesnage ) in hemocytes in vivo , He-Gal4; UAS-FYVE-GFP or he-Gal4; UAS-FYVE-GFP , hemoA4 flies were injected with 0 . 2 μL of bacterial culture grown to OD600 = 0 . 05 , using a glass capillary microinjection needle . Flies were then dissected as described in Magny et . al 2013 [7] to expose the dorsal abdominal cuticle in order to image the hemocytes associated to the dorsal vessel , as described in Horn et al . 2014 [71] . Flies were dissected after 20 or 120 min postinjection , in PBS , and the cuticles fixed in 4% PFA in PBS for 20 min . The preparations were then washed in , PBTX , incubated 30 min in PBS with phalloidin-Cy5 ( Sigma-Aldrich ) ( 1:10 ) , washed in PBS , and mounted in Vectashield ( vector ) . Hemocytes were imaged with a Zeiss laser scanning microscope LSM 5 . 10 on a Zeiss Axioskop 2 stage , with a 40X Achroplan objective , and bacterial cells within hemocytes , as determined with the phalloidin and FYVE counter-stains , were counted in Z-stack reconstructions . To quantify endogenous bacterial contents , we followed the method described in Khalil et al . 2015 [72] . Briefly , hemoA4 and wild-type Oregon Red L3 larvae were collected , washed in PBTX and PBS , and reared together in the same vial at a density of 20 flies per vial , in order to minimize the effect of external bacteria present in the media . Wild type and mutant flies were collected at either 10 d or 30 d after eclosion , and homogenised individually in 100 μL of autoclaved and sterile filtered PBS . The homogenates were diluted serially , and plated in antibiotic-free LB-agar plates . The plates were incubated for 48 h at 29°C , and the bacterial colonies from each homogenate spot counted . The full-length hemo cDNA was obtained by RT-PCR ( RACE ) ( PCR from overlapping EST ( Flybase ) using specific primers and cloned in TOPO vector ( Invitrogen ) . pUAST-hemo full length vector was constructed by inserting flanking restriction sites to the 5’ and 3‘ end of the hemo cDNA respectively by PCR amplification . hemo cDNA was inserted in the pUASt ( AttB ) vector using these restriction sites . A similar strategy was used to clone ORF1 and ORF2 into the pUASt ( AttB ) vector . Generation of the pUASt-hemo-ORF carboxyl-tagged with GFP was carried out by generating a single PstI restriction site at the end of the ORF removing the stop codon in the hemo cDNA in TOPO by PCR and subsequently inserting a PstI flanking GFP cds in frame lacking its kozak and methionine sequences . Then the hemo-GFP was cloned in the pUASt-vector . The pUASt ( AttB ) -hemo-GFP was constructed for site-directed transgenesis by removing the hemo-GFP from pUASt and cloned in the pUASp ( ATTB ) vector ( DGRC ) . The pUASt-ORF2 carboxyl terminal GFP tagged construct was made by amplification of ORF2 sequence and a short 5’UTR according to the short hemo cDNA described in flybase by PCR . The PCR product was cloned in pENTR vector ( Invitrogen ) . LR recombination ( Invitrogen ) was used to clone the ORF2 into the pUASt-ctGFP vector ( Murphy collection , Carnegie Institute ) . The hemo frameshift construct was engineered by introducing a double nucleotide insertion in each ORF ( in ORF1 a CG insertion three codons after the ATG and in ORF2 a GC insertion just after the ATG ) by site directed mutagenesis ( Stratagene , CA , USA ) using the hemo full length cDNA in TOPO as a template . Subsequently , the hemo frameshift cDNA was cloned in pUASt ( ATTB ) vector . A CG7691 Genomic Fragment spanning 11 . 8 Kb was constructed by amplifying independently four sequential genomic fragments by Long Expand Range PCR ( Roche , Basel , Switzerland ) each containing the EcoRI-Kpn1 , KpnI-NotI , NotI-XbaI and XbaI-AscI restriction sites respectively and then cloned in TOPO vector ( Invitrogen ) . Each fragment was sequentially cloned in the Casper 5 vector using the restriction sites . Proof of the presence of the wild-type the CG7691 gene was obtained by sequencing . CG7691 mRNA expression from the construct was detected in adult transgenic flies over an hemoA4 deletion background by RT-PCR as described in [7] . To generate an amino terminal tagged version of 14-3-3ζ the ORF and 3’UTR from RH61958 cDNA ( DGRC ) were amplified by PCR and subsequently the PCR product was cloned in the pENTR vector ( Invitrogen ) . The pUASt-Ntmyc-14-3-3ζ construct was generated by LR recombination into the acceptor vector ( Murphy’s collection; Carnegie Institute ) . A snn ORF clone in an entry vector ( GeneCopoeia , MD , US ) was inserted in the pUASp vector ( Murphy’s collection , Carnegie Institute ) by LR recombination ( Invitrogen ) . Similarly , a pUASt- amino-terminal GFP- or FLAG-tagged SNN constructs were generated using the former snn ORF entry vector and destination vectors ( Murphy’s collection , Carnegie Institute ) . To generate the hemo-RNAi construct used the method described by Kondo et al . ( 2006 ) [73] . Briefly , we used the following primers ( hemo RNAi Fw GTTCCACAGAGATATCGTCT hemo RNAi Rv ACCACGAAGCTAACGCACAGC ) to amplify a 354 bp DNA fragment from genomic DNA , corresponding to a region of the hemo locus with no homology to other genomic regions . This fragment was then cloned into the pRISE vector and used to generate transgenic flies by Bestgene . S2 cell culture , transfections , and immunocytochemistry were performed as described in [7] . Kc167 cells ( 250X105 ) were cultured in a 6-well plate dish in M3 insect medium ( Sigma , MO , USA ) containing 10% FBS and 1% pellicillin/streptomycin ( Sigma ) . After 24 h , pUASt- and Act5-Gal4-constructs were transfected using FuGene ( Roche ) . 3 d later , cells were pelleted for immunoprecipitation . Mouse RAW264 . 7 cells were cultured in RPMi-1640 with L-glutamine ( Sigma ) supplemented with 10% FBS and 1% penicillin-streptomycin ( Sigma ) at 37°C and 5% CO2 conditions . 300 x 105 cells/well were grown in a 6-well plate ( for lysotracker they were placed on acid-treated coverslips ) . siRNA treatments were performed as described by Ulvila et al . , 2011 [43] . A mixture of two 5’ FITC labelled siRNAs were generated for snn , one targeting the ORF ( sense sequence-GGCCAUGUGUGGAAAGAAAUU ) and the other the 3’UTR ( sense-sequence-GGGAGGAGCUGUAGGGAAGUU ) , and a control sample of nontargeting siRNAs pool manufactured by Dharmacon ( GE Healthcare , NJ , US ) were transfected using RNAmax reagent ( Invitrogen ) into cells at 24h and 48h after seeding the cells . After 24 h , cells were pelleted for Trizol ( Ambion ) mRNA extraction . Retrotranscription was conducted using random and polydT primers using RNA isolation kit ( Promega , CA , US ) . snn and elongation translation factor 4 primers were used for RT-PCR . For quantification of lysosomes with Lysotracker , the media was removed and cells were washed with PBS1x . Afterwards , 50nM Lysotracker in culture media was added and left for 15 min . Cells were washed with culture media and mounted for acquisition of live cell imaging ( 30min ) using confocal microscopy with a Zeiss LSM 510 NLO AXIOSkop microscope . Three independent siRNA treatment repeats were performed in our analysis . Late third instar larvae ( 80 ) and pelleted Kc167 cells were homogenized in lysis buffer ( 50 mM Tris pH 7 . 5;150 mM NaCl;1 mM EDTA;1 mM EGTA; 2 . 5 mM pyrophosphate; 1 mM Na3VO4; 1 mM glycerol phosphate ) for 1 h ( 4°C ) . Cellular debris was spun at 3 , 000 rpm for 15 min at 4°C . Supernatant was again spun at 9 , 500 rpm for 45 min at 4°C . Supernatant was added to equilibrated GFP-beads ( Chromo Tek , NY , US ) and left rotating overnight at 4°C . Beads were washed several times and then boiled in Laemmi Loading Buffer ( Biorad , CA , US ) . Beads were loaded onto 8% or 12% polyacrylamide gel and proteins were separated by SDS-PAGE ( Biorad ) . Specific proteins were detected by Western Blot using a semidry blotting or tetra cell ( Biorad ) . Antibodies used were: mouse anti-GFP 1:3000 ( Roche ) ; anti-HA 1:5000 ( Roche ) ; anti-Myc 1:1000 ( Upstate ) . To measure the size of the vacuoles of in vivo embryonic and pupal thoracic hemocytes , we calculated the OAI , which is the accumulative area of intracellular vacuoles divided by the cellular area ( >30 cells for most of the genotypes randomly selected as described above ) in hemocytes in the wild-type and hemoA4 mutant . Z-stack images were processed in the ImageJ software , and the outlines of the vacuole and cell were highlighted using the draw tool and the areas were measured . Student t test was used for statistical analysis . For measuring vacuolation in ex vivo hemocytes , we determined the OVI , which corresponds to the addition of the volume of intracellular vacuoles ( voids that disrupt the tubulin cytoskeleton ) /total cytoplasmic volume of the cell . The Z-stack images were loaded onto the ImageJ software . To calculate the total cytoplasmic volume , we used the oblate ellipsoid volume formula: 4/3πab2 ( a = radius , b = height ) . The radius of the cytoplasm was measured at widest Z-slice of the cell and the height at the highest point at the orthogonal projection . The average wild-type cytoplasmic volume of ex vivo hemocytes ( >30 cells for most of samples were selected at random across >3 separate imaging foci and 3 experimental repeats ) was 311 . 06 μm3 , which diverged only slightly from A4 mutants ( 375 . 57 μm3 ) , therefore the wild-type cytoplasmic volume was used for our analysis . To calculate the volume of intracellular vacuoles , we used the sphere volume formula: = 4/3πr3 , the radius was calculated by measuring the diameter of the widest point of the vacuole , only the vacuoles with a radius equal to or larger than 1 . 5 μm were considered for OVI analysis , with cells lacking vacuoles of at least 1 . 5 μm giving an OVI of 0 . To measure early endosome cellular compartments ( FYVE-positive ) in ex vivo hemocytes , we calculated the OAI , meaning the addition of FYVE particle area/total cytoplasmic cell area ( volume inside of cortical actin ring ) . ImageJ software was used to calculate the OAI . A maximum projection of the Z-stacks was made and both FYVE and Phalloidin channels were separated . FYVE image was transformed into a binary image and particle analyses ( size pixel μ2: 20-infinity and circularity: 0–1 ) were used to calculate the area of particles . Addition of FYVE particle areas was done in Excel . Measurement of the cytoplasmic area was performed manually by using the free hand selection tool . OAI of endosomes of 20–60 cells selected at random as described above were calculated per genotype . In addition , we measured FYVE diameter ratio consisting of the average of three FYVE vesicle diameters per cell ( randomly selected 20–60 cells from >3 imaging foci and 3 experimental repeats per genotype ) divided by the averaged FYVE diameter of the He-Gal4 UAS-hemo full-length transcript in an A4 mutant background to diminish the genetic background effects possibly caused by the Gal4 , hemoA4 and UAS-docking chromosomes . Lysosome aggregates in ex vivo hemocytes were measured by calculating the area of lysosome particles as described above ( visualized with lysotracker ) /number of lysosomal particles . For each genotype >30 cells were selected randomly as described above and were measured . We calculated the OAI as described above as a measure of the lysosome occupied area in the siRNAs experiments in mouse RAW264 . 7 cells . 80–110 cells were measured per sample in each of the three repeats . Statistical analysis was performed with one-way ANOVA to assess whether the means of the groups were significantly different ( p < 0 . 0001 ) , and a post hoc Bonferroni’s test was carried out to compare multiple groups with p < 0 . 05 considered as significant using the Prism suite ( Graph Pad ) . As a measure of the colocalisation of endocytic markers and hemotin-GFP peptides , we used the Pearson’s correlation coefficient ( ImageJ WCIF colocalisation plugins ) that evaluates the amount of signal intensity from one channel that occurs in the same location in the other colour channel . Pearson’s coefficients range between −1 to 1 , with values closer to 1 indicating reliable colocalisation . Pearson’s coefficients of hemocyte cells ( n = 17–32 per genotype ) were analysed with two-tailed Mann-Whitney test ( p < 0 . 0001 ) . The Phyre2 online search engine was used to find structural homologues using the Hemo-ORF peptide sequence as an input . To find Hemotin sequence homologues , we used the pipeline described in Magny et al . , 2013 [7] that is based on the identification of closest homologues , followed by expanded searches using consensus sequences weighted by the phylogeny from the alignment of such close homologues . In addition , we introduced an extra corroboration step of reciprocal best Blast hit in doubtful cases . Alignments and trees of peptide sequences were generated using MAFFT and Clustal programs , with MAFFT parameters set at 5 iterations and global pair alignment . The alignments were visualized with Jalview . To evaluate the ability of hemo peptides to adopt the tertiary peptide structure of human Stannin , we threaded the sequence of insect Hemotin peptides ( D . melanogaster and Microplitis demolitor ) and the basal vertebrate Stannin peptide ( hagfish ) onto the human Stannin peptide structure ( 1zzA; RCSB protein data bank ) using the Phyre2 server [35] on one-to-one threading mode . These preliminary structures were then refined , and Global distances test values were obtained with respect to the original human 1zzA Snn structure using the KobaMIN server [38] . | In our genomes there are millions of short open reading frames that could produce small peptides of less than 100 amino-acids if translated . These sequences have been so far disregarded , but increasing evidence supports the notion that a subset of these–termed smORFs–are translated; however the function of most of the resulting peptides remains unclear . Here we characterise hemotin , a smORF gene that encodes a transmembrane peptide of 88 amino-acids expressed in Drosophila macrophages , and we show that Hemotin localizes to early endosomes–vesicles involved in the traffic of material between the cell membrane and the cytoplasm . Macrophages found in hemotin mutants have enlarged and abnormal endosomes that delay digestion of phagocytosed bacteria . Accordingly , these mutant flies fight bacterial infections poorly , and die early . In humans we identify Stannin–a peptide previously involved in heavy metal toxicity–as the Hemotin homologue , and show that it shares Hemotin function in macrophages . These results identify the Hemotin-Stannin smORF peptides as regulators of phagocytosis , and suggest that they could have contributed to the ancestral origin of macrophage-like cells . This provides a strong example of a smORF conserved for hundreds of millions of years . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"bacteriology",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"lysosomes",
"immune",
"cells",
"vesicles",
"immunology",
"microbiology",
"vertebrates",
"animals",
"animal",
"models",
"drosophila",
"melanogaster",
"model",
"organisms",
"bacteri... | 2016 | Hemotin, a Regulator of Phagocytosis Encoded by a Small ORF and Conserved across Metazoans |
The human brain has the impressive capacity to adapt how it processes information to high-level goals . While it is known that these cognitive control skills are malleable and can be improved through training , the underlying plasticity mechanisms are not well understood . Here , we develop and evaluate a model of how people learn when to exert cognitive control , which controlled process to use , and how much effort to exert . We derive this model from a general theory according to which the function of cognitive control is to select and configure neural pathways so as to make optimal use of finite time and limited computational resources . The central idea of our Learned Value of Control model is that people use reinforcement learning to predict the value of candidate control signals of different types and intensities based on stimulus features . This model correctly predicts the learning and transfer effects underlying the adaptive control-demanding behavior observed in an experiment on visual attention and four experiments on interference control in Stroop and Flanker paradigms . Moreover , our model explained these findings significantly better than an associative learning model and a Win-Stay Lose-Shift model . Our findings elucidate how learning and experience might shape people’s ability and propensity to adaptively control their minds and behavior . We conclude by predicting under which circumstances these learning mechanisms might lead to self-control failure .
The human brain has the impressive ability to adapt how it processes information and responds to stimuli in the service of high level goals , such as writing an article [1] . The mechanisms underlying this behavioral flexibility range from seemingly simple processes , such as inhibiting the impulse to browse your Facebook feed , to very complex processes such as orchestrating your thoughts to reach a solid conclusion . Our capacity for cognitive control enables us to override automatic processes when they are inappropriate for the current situation or misaligned with our current goals . One of the paradigms used to study cognitive control is the Stroop task , where participants are instructed to name the hue of a color word ( e . g . , respond “green” when seeing the stimulus RED ) while inhibiting their automatic tendency to read the word ( “red” ) [2] . Similarly , in the Eriksen flanker task , participants are asked to report the identity of a target stimulus surrounded by multiple distractors while overcoming their automatic tendency to respond instead to the distractors . Individual differences in the capacity for cognitive control are highly predictive of academic achievement , interpersonal success , and many other important life outcomes [3 , 4] . While exerting cognitive control improves people’s performance in these tasks , it is also effortful and appears to be intrinsically costly [5 , 6] . The Expected Value of Control ( EVC ) theory maintains that the brain therefore specifies how much control to exert according to a rational cost-benefit analysis , weighing these effort costs against attendant rewards for achieving one’s goals [7] . In broad accord with the predictions of the EVC theory , previous research has found that control specification is context-sensitive [8 , 9] and modulated by reward across multiple domains [10 , 11] , such as attention , response inhibition , interference control , and task switching . While previous theories account for that fact that people’s performance in these task is sensitive to reward [7 , 12–14] , it remains unclear how these dependencies arise from people’s experience . Recently , it has been proposed that the underlying mechanism is associative learning [15 , 16] . Indeed , a number of studies have demonstrated that cognitive control specification is plastic: whether people exert cognitive control in a given situation , which controlled processes they employ , and how much control they allocate to them is learned from experience . For instance , it has been demonstrated that participants in visual search tasks gradually learn to allocate their attention to locations whose features predict the appearance of a target [17] , and a recent study found that learning continuously adjusts how much cognitive control people exert in a Stroop task with changing difficulty [18] . Furthermore , it has been shown that people learn to exert more cognitive control after their performance on a control-demanding task was rewarded [10] and learn to exert more control in response to potentially control-demanding stimuli that are associated with reward than to those that are not [11] . These studies provide evidence that people can use information from their environment ( e . g . , stimulus features ) to learn when to exert cognitive control and how to exert control , and it has recently been suggested that this can be thought of in terms of associative learning [15 , 16] . Other studies suggested that cognitive control can be improved through training [19–21] . However , achieving transfer remains challenging [22–25] , the underlying learning mechanisms are poorly understood , and there is currently no theory that could be used to determine which training regimens will be most effective and which real-life situations the training will transfer to . Developing precise computational models of the plasticity of cognitive control may be a promising way to address these problems and to enable more effective training programs for remediating executive dysfunctions and enabling people to pursue their goals more effectively . In this article , we extend the EVC theory to develop a theoretical framework for modeling the function and plasticity of cognitive control specification . This extension incorporates recent theoretical advances inspired by the rational metareasoning framework developed in the artificial intelligence literature [26 , 27] . We leverage the resulting framework to derive the Learned Value of Control ( LVOC ) model which can learn to efficiently select control signals based on features of the task environment . The LVOC model can be used to simulate cognitive control ( e . g . , responding to a goal-relevant target that competes with distractors ) and , more importantly , how it is shaped by learning . According to the LVOC model , people learn the value of different cognitive control signals ( e . g . , how much to attend one stimulus or another ) . A key strength of this model is that it is very general and can be applied to phenomena ranging from simple learning effects in the Stroop task to the acquisition of complex strategies for reasoning and problem-solving . In order to demonstrate the validity and generality of this model , we show that it can capture the empirical findings of five cognitive control experiments on the plasticity of visual attention [17] , the interacting effects of reward and task difficulty on the plasticity of interference control [10 , 11] , and the transfer of such learning to novel stimuli [8 , 9] . Moreover , the LVOC model outperforms alternate models of such learning processes that rely only on associative learning or a basic win-lose-stay-shift strategy . Our findings shed light on how learning and experience might shape people’s ability and propensity to adaptively control their minds and behavior , and the LVOC model predicts under which circumstances these mechanisms might lead to self-control failure .
At an abstract level , all cognitive control processes serve the same function: to adapt neural information processing to achieve a goal [28] . At this abstract level , neural information processing can be characterized by the computations being performed , and the extent to which the brain achieves its goals can be quantified by the expected utility of the resulting actions . From this perspective , an important function of cognitive control is to select computations so as to maximize the agent’s reward rate ( i . e . , reward per unit time ) . This problem is formally equivalent to the rational metareasoning [26 , 29] problem studied in computer science: selecting computations so as to make optimal use of the controlled system’s limited computational resources ( i . e . , to achieve the highest possible sum of rewards with a limited amount of computation ) . Thus , rational metareasoning suggests that the specification of cognitive control is a metacognitive decision problem . In reinforcement learning [30] , decision problems are typically defined by a set of possible actions , the set of possible states , an initial state , the conditional probabilities of transitioning from one state to another depending on the action taken by the agent , and a reward function . Together these five components define a Markov decision process ( MDP [30] ) . In a typical application of this framework the agent is an animal , robot , or computer program , actions are behaviors ( e . g . , pressing a lever ) , the state characterizes the external environment ℰ ( e . g . , the rat’s location in the maze ) , and the rewards are obtained from the environment ( e . g . , pressing a lever dispenses cheese ) . In general , the agent cannot observe the state of the environment directly; for instance , the rat running through a maze does not have direct access to its location but has to infer this from sensory observations . The decision problems posed by an environment that is only partially observable can be modelled as a partially observable MDP ( POMDP [31] ) . For each POMDP there is an equivalent MDP whose state encodes what the agent knows about the environment and is thus fully observable; this is known as the belief-MDP [31] . Critically , the belief-MDP formalism can also be applied to the choice of internal computations [27]–such as allocating attention [32] or gating information into working memory [33 , 34]–rather than only physical actions . In the rational metareasoning framework , the agent is the cognitive control system whose actions are control signals that specify which computations the controlled systems should perform . The internal state of the controlled systems is only partially observable . We can formally define the problem of optimal cognitive control specification as maximizing reward the in the meta-level MDP M= ( S , s0 , C , T , r ) , ( 1 ) where S is the set of possible information states , comprising beliefs about the external environment ( e . g . , the choices afforded by the current situation ) and beliefs about the agent’s internal state ( e . g . , the decision system’s estimates of the choices’ utilities ) , s0 denotes the initial information state , C is the set of possible control signals that may be discrete ( e . g . , “Simulate action 1 . ” ) or continuous ( e . g . , “Increase the decision threshold by 0 . 175 . ” or “Suppress the activity of the word-reading pathway by 75% . ” ) , T is a transition model , and r is the reward function that cognitive control seeks to maximize . The transition model specifies the conditional probability of transitioning from belief state s to belief state s′ if the control signal is c by T ( s , c , s′ ) . The meta-level reward function r combines the utility of outcome X ( of actions resulting from control signal c in belief state s ) with the computational cost associated with exerting cognitive control: r ( s , c ) =u ( X ) -cost ( s , c ) , ( 2 ) where X is the outcome of the resulting action , u is utility function of the brain’s reward system , and cost ( s , c ) is the cost of implementing the controlled process . Within this framework , we can define a cognitive control strategy π:S→C as a mapping from belief states s∈S to control signals c∈C . The optimal cognitive control strategy π⋆ is the one that always chooses the computation with the highest expected value of computation ( EVOC ) : π⋆:s↦argmaxcEVOC ( c , s ) . ( 3 ) The EVOC is the expected sum of computational costs and benefits of performing the computation specified by the control signal c and continuing optimally from there on: EVOC ( c , s ) =Qπ⋆ ( s , c ) =E[r ( s , c ) +Vπ⋆ ( St+1 ) |St=s , Ck=c , T] , ( 4 ) where Qπ⋆ is known as the Q-function of the optimal control strategy π⋆ , and Vπ⋆ ( St+1 ) is the expected sum of meta-level rewards of starting π⋆ in state St+1 . In summary , cognitive control specification selects the sequence of cognitive control signals that maximizes the expected sum of rewards of the resulting actions minus the cost of the controlled process . The optimal solution to this problem is given by the optimal control policy π⋆ . So far , we have assumed that the cognitive control system chooses one control signal at a time , but c could also be a vector comprising multiple control signals ( e . g . , one that increases the rate at which evidence is accumulated towards the correct decision via an attentional mechanism and a second one that adjusts the decision threshold ) . Furthermore , overriding a habit by a well-reasoned decision also requires executing a coordinated sequence of cognitive operations for planning and reasoning . Instead of specifying each of these operations by a separate control signal , the cognitive control system might sometimes use a single control signal to instruct the decision system to execute an entire planning strategy . The rational metareasoning framework allows us to model cognitive strategies as options [35–38] . An option is a policy combined with an initiation set and a termination condition [38] . Options can be treated as if they were elementary computations and elementary computations can be interpreted as options that terminate after the first step . With this extension , the optimal solution to the cognitive control specification problem becomes π⋆ ( s ) =argmaxo∈OQ⋆ ( s , o ) , ( 5 ) where the set of options O may include control strategies and elementary control signals . Critically , this rational metareasoning perspective on cognitive control covers not only simple phenomena , such as inhibiting a pre-potent automatic response in the Stroop task , but also more complex ones , such as sequencing one’s thoughts so as to follow a good decision strategy , and very complex phenomena such as reasoning about how to best solve a complex problem . The computations required to determine the expected value of control may themselves be costly and time consuming . Yet , in some situations cognitive control has to be engaged very rapidly , because maladaptive reflexes , impulses , and habitual responses have to be inhibited before the triggered response has been executed . In such situations , there is simply not enough time to compute the expected value of control on the fly . Fortunately , this may not be necessary because an approximation to the EVOC can be learned from experience . We therefore hypothesize that the cognitive control system learns to predict the context-dependent value of alternative control signals . By understanding how this learning occurs , we might be able to explain the experience-dependent changes in how people use their capacity for cognitive control , which we will refer to as the plasticity of cognitive control specification . In addition to these systematic , experience-driven changes cognitive control is also intrinsically variable . To model the plasticity and the variability of cognitive control , this section develops a model that combines a novel feature-based learning mechanism with a new control specification mechanism that explores promising control signals probabilistically to accelerate learning which of them is most effective . The previous section characterized the problem of cognitive control specification as a sequential meta-decision problem . This makes reinforcement learning algorithms [39] a natural starting point for exploring how the cognitive control systems learns the EVOC from experience . Approximate Q-learning appears particularly suitable because the optimal control strategy can be expressed in terms of the optimal Q-function ( Eqs 3–5 ) . From this perspective , the plasticity mechanisms of cognitive control specification serve to learn an approximation to the value Qt ( s , c ) of selecting control signal c in state s based on one’s experience with selecting control signals c = ( c1 , ⋯ , ct ) in states s = ( s1 , ⋯ , st ) and receiving the meta-level rewards r = ( r1 , ⋯ , rt ) . Learning an approximate Q-function Qt from this information could enable the cognitive control system to efficiently select a control strategy by comparing learned values rather than reasoning about their effects . Learning the optimal meta-level state-value function Q⋆ can be challenging because the value of each control signal may depend on the outcomes of the control signals selected afterwards . Furthermore , the state space of the meta-level MDP has a very high dimensionality as it comprises all possible states that the controlled system could be in . To overcome these challenges , a neural system like the brain might learn a linear approximation to the meta-level state value function instead of estimating each of its entries separately . Concretely , the cognitive control system might learn to predict the value of selecting a control strategy ( e . g . , focusing on the presenting speaker instead of attending to an incoming phone call ) by a weighted sum of features of the internal state and the current context ( e . g . being in a conference room ) . For instance , the value Q⋆ ( s , c ) of choosing control signal c in the internal state s can be predicted from the features fk ( s ) , the implied control signal intensities c , their interactions with the features , that is fk ( s ) ci , and their costs . Concretely , the EVOC of selecting control signal c in state s is approximated by the Learned Value of Control ( LVOC ) , LVOC ( s , c;w ) =w0+ ( ∑k=1Kwk ( f ) ⋅fk ( s ) ) + ( ∑l=1Lwl ( c ) ⋅cl ) + ( ∑k=1K∑l=1Lwk , l ( f×c ) ⋅fk ( s ) ⋅cl ) −cost ( c ) −w ( T ) ⋅T , ( 6 ) where the weight vector w includes the offset w0 , the weights wk ( f ) of the states’ features , the weights w ( c ) of the control signal intensities , the weights wk , l ( f×c ) of their interaction terms , the weight w ( T ) of the response time T , and cost ( c ) is the intrinsic cost of control which scales with the amount of cognitive control applied to the task . The optimal way to update the weights based on experience in a stationary environment is given by Bayes rule . Our model therefore maintains and continues to update an approximation to the posterior distribution P ( w|e1 , ⋯ , t ) ∝P ( w|e1 , ⋯ , t-1 ) ⋅P ( et|w ) , ( 7 ) on the weight vector w given its experience e1 , ⋯ , t up until the present time t , where each experience ei = ( si , ci , ri , Ti , si+1 ) comprises the state , the selected control signal , the reward , the response time , and the next state . In simple settings where a single control signal determines a single reward our model’s learning mechanism is equivalent to Bayesian linear regression [40 , 41] . In more complex settings involving a series of control signals or delayed rewards the learning rule approximates the Bayesian update by substituting the delayed costs and benefits of control by the model’s predictions . For more details , see S1 Text . If the value of control is initially unknown , the optimal way to select control signals is to balance exploiting previous experience to maximize the expected immediate performance with exploring alternative control allocations that might prove even more effective . Our model solves this dilemma by an exploration strategy similar to Thompson sampling: It draws k samples from the posterior distribution on the weights and averages them , that is w~1 , ⋯ , w~k∼P ( w|e1 , ⋯ , t ) , w~=1k⋅∑i=1kw~i . ( 8 ) According to the LVOC model the brain then selects a control signal by maximizing the EVOC predicted by the average weight w~ , that is ct≈argmaxcLVOC ( st , c;w~ ) . ( 9 ) Together , Eqs 6–9 define the LVOC model of the plasticity of cognitive control . The LVOC model extends the EVC theory [7] which defines optimal control signals in terms of the EVOC ( Eq 3 ) , by proposing two mechanisms through which the brain might be able to approximate this normative ideal: learning a feature-based , probabilistic model of the EVOC ( Eqs 6 and 7 ) and selecting control signals by sampling from this model ( Eqs 8 and 9 ) . This model is very general and can be applied to model cognitive control of many different processes ( e . g . , which location to saccade to vs . how strongly to inhibit the word-reading pathway ) and different components of the same process ( e . g . , rate of evidence accumulation towards the correct decision vs . the decision threshold ) . The LVOC model’s core assumptions are that the brain learns to predict the EVOC of alternative control specifications from features of the situation and the control signals , and that the brain then probabilistically selects the control specification with the highest predicted value of control . Both of these components could be implemented by many different mechanisms . For instance , instead of implementing the proposed approximation to Bayesian regression , the brain might learn to predict the EVOC through the reward-modulated associative plasticity mechanism outlined in the SI . We are therefore not committed to the specific instantiation we used ( Eqs 7–9 ) for the purpose of the simulations reported below . The LVOC model instantiates the very general theory that the brain learns how to process information via metacognitive reinforcement learning . This includes not only the plasticity of cognitive control but also how people might discover cognitive strategies for reasoning and decision-making and how they learn to regulate their mental activities during problem solving . As a proof of concept , the following sections validate the LVOC model against five experiments on the plasticity of attention and interference control . In principle , the control-demanding behavior considered in this paper could result from simpler mechanisms than the ones proposed here . In this section , we consider two simple models that we use as alternatives to compare against the more complex LVOC model . The first model relies on the assumption that the plasticity of cognitive control can be understood in terms of associative learning [15 , 16] . We therefore evaluate our model against an associative learning model based on the Rescorla-Wagner learning rule [42] . This model forms stimulus-control associations based on the resulting reward . The association As , c between a stimulus s and a control signal c is strengthened when it is accompanied by ( intrinsic or extrinsic ) reward and weakened otherwise . Concretely , the association strengths involving the chosen response were updated according to the Rescorla-Wagner rule , that is As , c=As , c+α⋅ ( R-∑sIs⋅As , c ) , ( 10 ) where α is the learning rate , R is the reward and the indicator variable Is is 1 when the stimulus s was present and 0 else . Given the learned associations , the control signal is chosen probabilistically according to the exponentiated Luce’s choice rule , that is each control signal c is selected with probability p ( c ) =exp ( As , c ) ∑cexp ( As , c ) . ( 11 ) The second alternative model is based on previous research suggesting that people sequentially adjust their strategy through a simple Win-Stay Lose-Shift mechanism [43] . On the first trial , this mechanism chooses a strategy at random , and on each subsequent trial it either repeats the previous strategy when it was successful or switches to a different strategy when the current strategy failed . Here , we apply this idea to model how the brain learns which control signal to select . Concretely , our WSLS model repeats the previous control signal ( e . g . , “Attend to green . ” ) when it leads to a positive outcome ( Win-Stay ) and randomly selects a different control signal ( e . g . , “Attend to red . ” ) otherwise ( Lose-Shift ) . In contrast to the LVOC mode , the two alternative models assume that control signals are discrete rather than continuous . In the context of visual attention , they choose their control signal c from the set {1 , 2 , 3 , ⋯ , 12} of possible locations to attend , and in the context of inhibitory control they decide to either inhibit the process completely or not at all ( c ∈ {0 , 1} ) . To evaluate the proposed models , we used them to simulate the plasticity of attentional control in a visual search task [17] as well as learning and transfer effects in Stroop and Flanker paradigms [8–11] . Table 1 summarizes the simulated phenomena and how the LVOC model explains each at a conceptual level .
Lin et al . [17] had participants perform a visual search task for which the target of attention could either be predicted ( training and predictable test trials ) or not ( unpredictable test trials ) ( Fig 1a ) . For this task , given its core reinforcement learning assumption ( Table 1 ) , the LVOC model predicts that 1 ) people should learn to attend to the circle with the predictive color and thus become faster at finding the target over the course of training , 2 ) continue to use the learned attentional control strategy in the test block and hence be significantly slower when the target appears in a circle of a different color during the test block , and 3 ) gradually unlearn their attentional bias during the test block ( Fig 1c ) . As shown Fig 1b , all three predictions were confirmed by Lin and colleagues [17] . We compared the performance of LVOC to two plausible alternative models of these control adjustments: a Win-Stay Lose-Shift model and a simple associative learning model based on the Rescorla-Wagner learning rule . We found that the Win-Stay Lose-Shift model failed to capture that people’s performance improved gradually during training , and it also failed to capture the difference between people’s response times to predicted versus unpredicted target locations in the test block ( see Fig 1d ) . As Fig 1e shows , the fit of the associative learning model ( estimated learning rate: 0 . 0927 ) captures that after learning to exploit the predictive regularity in the training block participants were significantly slower in the test block . However , this simple model predicted significantly less learning induced improvement and significantly slower reaction times than was evident from the data by [17] . A quantitative model comparisons using the Bayesian Information Criterion [60 , 61] provided very strong evidence that the LVOC model explains the data by [17] better than the Rescorla-Wagner model or the Win-Stay Lose-Shift model ( BICLVOC = 1817 . 8 , BICRW = 9763 . 2 , BICWSLS = 3449 . 9 ) . This reflects that our model was able to accurately predict the data from [17] without any free parameters being fitted to those data . In conclusion , findings suggest that the LVOC model correctly predicted essential learning effects observed by [17] and explains these data significantly better than a simple associative learning model and a Win-Stay Lose-Shift model . To more accurately capture both the slow improvement in the training block and the rapid unlearning in the test block simultaneously , the LVOC model could be extended by including a mechanism that discounts what has been learned or increases the learning rate when a change is detected [62 , 63] . Next , we evaluate the LVOC model against empirical data on the plasticity of inhibitory control . We found that our model can capture reward-driven learning effects in Stroop and Flanker tasks , as well as how people learn to adjust their control allocation based on features that predict incongruence and the transfer of these learning effects to novel stimuli . In each case , the LVOC model captured the empirical phenomenon more accurately than either a simple Win-Stay Lose-Shift model or a simple associative learning model . The following two sections present these results in turn . The expected value of computation depends not only on the rewards for correct performance but also on the difficulty of the task . In easy situations , such as the congruent trials of the Stroop task , the automatic response can be as accurate , faster , and less costly than the controlled response . In cases like this , the expected value of exerting control is less than the EVOC of exerting no control . By contrast , in more challenging situations , such as incongruent Stroop trials , the controlled process is more accurate and therefore has a positive EVOC as long as accurate performance is sufficiently important . Therefore , on incongruent trials the expected value of control is larger than the EVOC of exerting no control . Our model thus learns to exert control on incongruent trials but not on congruent trials . Our model achieves this by learning to predict the EVOC from features of the stimuli . This predicts that people should learn to exert more control when they encounter a stimulus feature ( such as a color or word ) that is predictive of incongruence than when they encounter a feature that is predictive of congruence ( see Table 1 ) . Consistent with our model’s predictions , Bugg and colleagues [8] found that people learn to exert more control in response to stimulus features that predict incongruence than stimulus features that predict congruence . Their participants performed a color-word Stroop task with four colors and their names printed either in cursive or regular font . Our model captured the effects of congruency-predictive features on control allocation with a plausible set of parameters ( see Table 2 ) . As shown in Fig 4a and 4b , the LVOC model predicted that responses should be faster ( 655 ± 9 ms vs . 722 ± 11 ms; t ( 49 ) = 5 . 39 , p < 0 . 0001 ) and more accurate ( 2 . 85 ± 0 . 2% errors vs . 4 . 3 ± 0 . 3% errors; t ( 49 ) = 5 . 01 , p < 0 . 0001 ) on incongruent trials if the word was predictive of incongruence than when it was not . To their surprise , Bugg and colleagues observed that adding an additional feature ( font ) that conveyed the same information about congruence as the color , did not enhance learning . This is exactly what our model predicted because the presence of a second predictive feature reduces the evidence for the predictive power of the first one and vice versa–this is directly analogous to a phenomenon from the Pavlovian literature known as blocking , whereby an animal fails to learn an association between a stimulus and an outcome that is already perfectly predicted by a second stimulus [64] . Since our model learns about the predictive relationship between features and the EVOC , it predicts that all learning effects should transfer to novel stimuli that share the features that were predictive of the expected value of control in the training trials ( see Table 1 ) . A separate study by Bugg and colleagues [9] confirmed this prediction . They trained participants in a picture-word Stroop task to associate particular images of certain categories ( e . g . , cats and dogs ) with incongruence and associated particular images of other categories ( e . g . , fish and birds ) with congruence . As expected , participants learned to exert more control when viewing the stimuli associated with incongruence . More importantly , these participants also exerted more control when tested on novel instances of the category associated with incongruence ( e . g . , cats ) than on novel instances of the category associated with congruence ( e . g . , fish ) . This finding provides strong evidence for the feature-based learning mechanism that is at the core of our model of the plasticity of cognitive control and is entirely accounted for by our model . As shown in Fig 4e and 4f , our model correctly predicted the positive and the negative transfer effects reported by [9] with reasonable parameters ( see Table 2 ) : The model’s responses were faster ( 709 ± 3 ms vs . 685 ± 2 ms; t ( 99 ) = −8 . 13 , p < 0 . 0001 ) and more accurate ( 4 . 8 ± 0 . 3% errors vs . 3 . 2 ± 0 . 1% errors; t ( 99 ) = −5 . 06 , p < 0 . 0001 ) on incongruent trials if the word was predictive of incongruence than when it was not ( positive transfer ) . Conversely , on congruent trials , the predicted responses were slightly slower when the features wrongly predicted incongruence ( 527 ± 0 . 2ms vs . 530 ± 0 . 1ms , t ( 99 ) = 9 . 28 , p < 0 . 0001; negative transfer ) .
The model developed in this article builds on two previous theories: the EVC theory , which offered a normative account of control specification [7] , and the rational metareasoning theory of strategy selection [53] , which suggested that people acquire the capacity to select heuristics adaptively by learning a predictive model of the execution time and accuracy of those heuristics . The LVOC model synergistically integrates these two theories: it augments the EVC theory with the metacognitive learning and prediction mechanisms identified by [53] , and it augments rational metareasoning models of strategy selection with the capacity to specify continuous control signals that gradually adjust parameters of the controlled process ( see S2 Text ) . All else being equal , the proposed learning rules ( see Eq 7 , S1 Text Equations 1–7 , and S3 Text Equations 13–14 ) predict that people’s propensity to exert cognitive control should increase when the controlled process was less costly ( e . g . , faster ) or generated more reward than expected [19] . The experience that less controlled ( more automatic ) processing was more costly or less rewarding than expected should also increase our propensity to exert cognitive control [71–73] . Conversely , if a controlled process performed worse than expected or if an automatic process performed better than expected , people’s propensity to exert cognitive control should decrease [74] . At a more detailed level , our theory predicts that the influence of environmental features on control allocation generalizes across contexts , to the extent that their features are similar . Thus , adding or removing features to the internal predictive model of the EVOC should have a profound effect on the degree to which observed performance of the controlled process in Context A changes people’s propensity to select it in Context B , and vice versa . This mechanism can account for empirical evidence that suggests a role for feature-binding in mechanisms of task switching [75–78] . These studies suggest that participants associate the task that they perform on a stimulus with the features of that stimulus . Once they are asked to engage in a new task on that stimulus , the old ( associated ) task interferes , leading to switch costs . Furthermore , our theory predicts that increasing the rewards and punishments for the outcomes of the controlled or automatic processes should increase the speed with which people’s control allocation adapts to new task requirements , because the resulting weight updates will be larger; this becomes especially apparent when the updates are rewritten in terms of prediction errors ( see S3 Text , Eqs 1 and 2 ) . Finally , when the assumptions of the internal model are met and its features distinguish between the situations in which each controlled process performs best , then control signal selection should become increasingly more adaptive over time [79 , 80] . But in situations where the internal model’s assumptions are violated , for instance because the value of control is not additive and linear in the features , then the control system’s plasticity mechanisms may become maladaptive . This prediction has been confirmed in a recent experiment with a novel color-word Stroop paradigm comprising two association phases and a test phase [81] . In the first association phase , participants learned that color naming was rewarded for certain colors whereas word reading was rewarded for the other colors . In the second association phase , participants learned that color-naming was rewarded for certain words whereas word-reading was rewarded for other words . Critically , in the test phase , naming the color was rewarded if either the word or the color had been associated with color naming ( SINGLE trials ) ; but when both the color and the word were associated with color naming then participants had to instead read the word ( BOTH trials ) . This non-linear relationship between stimulus-features and control demands caused mal-transfer from SINGLE trials to BOTH trials that significantly interfered with participants’ performance ( resulting in participants incorrectly engaging in color-naming , the more control-demanding task which in that context was also less rewarding ) . The LVOC model may thus be able to explain the puzzling phenomenon that people sometimes overexert cognitive control even when it hurts their performance . For instance , if your past experience has taught you to choose your words very carefully on a certain topic then receiving an email on that topic might compel you to mentally compose a perfect response even when you would be better off thinking about how to open the talk you have to deliver in 5 minutes . According to the LVOC model , control allocation is a process of continuing gradual adjustment ( Eq 13 ) . This means that the control intensity for a new situation starts out with the control intensity from the previous situation and is then gradually adjusted towards its optimal value—just like in anchoring-and-adjustment [48 , 82] . This might provide a mechanism for commonly observed phenomena associated with task set inertia and switch costs [47] . Since control adjustment takes time , this mechanism predicts that increased time pressure could potentially lead to decreased control adjustment , thereby biasing people’s control allocation to its value on the previous trial and thus decreasing their cognitive flexibility . Finally , thinking about the neural implementation of the LVOC model leads to additional neural predictions as detailed in the S3 Text . We view rational metareasoning as a general theoretical framework for modeling the allocation and plasticity of cognitive control . As such , it could be used to develop unifying models of different manifestations of cognitive control , such as attention , response inhibition , and cognitive flexibility . Furthermore , rational metareasoning can also be used to connect existing models of cognitive control [6 , 7 , 32–34 , 46 , 65 , 66 , 79 , 80 , 83 , 84] . Interpreting previously proposed mechanisms of control allocation as approximations to rational metareasoning and considering how else rational metareasoning could be approximated might facilitate the systematic evaluation of alternative representations and computational mechanisms and inspire new models . While our computational explorations have focused on which control signal the cognitive control system should select , future work might also shed light on how the cognitive control system monitors the state of the controlled system by viewing the problem solved by the cognitive control system as a partially observable MDP . Concretely , the function of cognitive monitoring could be formulated as a meta-level MDP whose computational actions include sensing operations that update the cognitive control system’s beliefs about the state of the monitored system . Future work should further evaluate the proposed computational mechanism and its neural implementation by performing quantitative model comparisons against simpler models across a wider range of cognitive control phenomena . This line of work should also evaluate the performance of the proposed metacognitive learning mechanism and evaluate it against alternative mechanisms ( e . g . , temporal difference learning mechanisms with eligibility traces [30] ) . Another interesting direction will be to use the learning models to investigate the plasticity of people’s cognitive control skills . We are optimistic that this line of work will lead to better quantitative models of control plasticity that can be used to develop interventions to improve people’s executive functions via a combination of cognitive training and augmenting environments where people’s automatic responses are maladaptive with cues that prime them to employ an appropriate control signal . In addition , future work may also explore model-based metacognitive reinforcement learning [85] as a model of the plasticity of cognitive control specification . Model-based hierarchical reinforcement learning approaches [37] , such as option models [38] , could be used to integrate the learning mechanisms for the value of individual control signals with the strategy selection model to provide an account of how the brain discovers control strategies . This might explain how people learn to adaptively coordinate their thoughts and actions to pursue increasingly more challenging goals over increasingly longer periods of time . Finally , the rational metareasoning framework can also be used to model how people reason about the costs and benefits of exerting mental effort and to delineate self-control failure from rational resource-preservation through a normative account of effort avoidance [13 , 86] . Our simulation results suggested that the LVOC model provides a promising step towards a mathematical theory of cognitive plasticity that can serve as a scientific foundation for designing cognitive training programs to improve people’s executive functions . This illustrates the utility of formalizing the function of cognitive control in terms of rational metareasoning . Rational metareasoning provides a unifying framework for modeling executive functions , and thus opens up exciting avenues for future research . We are optimistic that the connection between executive functions and metareasoning will channel a flow of useful models and productive ideas from artificial intelligence and machine learning into the neuroscience and psychology of cognitive control . | The human brain has the impressive ability to adapt how it processes information to high level goals . While it is known that these cognitive control skills are malleable and can be improved through training , the underlying plasticity mechanisms are not well understood . Here , we derive a computational model of how people learn when to exert cognitive control , which controlled process to use , and how much effort to exert from a formal theory of the function of cognitive control . Across five experiments , we find that our model correctly predicts that people learn to adaptively regulate their attention and decision-making and how these learning effects transfer to novel situations . Our findings elucidate how learning and experience might shape people’s ability and propensity to adaptively control their minds and behavior . We conclude by predicting under which circumstances these learning mechanisms might lead to self-control failure . | [
"Abstract",
"Introduction",
"Models",
"Results",
"Discussion"
] | [
"learning",
"control",
"theory",
"engineering",
"and",
"technology",
"social",
"sciences",
"neuroscience",
"learning",
"and",
"memory",
"control",
"engineering",
"cognitive",
"psychology",
"systems",
"science",
"mathematics",
"cognition",
"vision",
"computer",
"and",
"i... | 2018 | Rational metareasoning and the plasticity of cognitive control |
Electrophysiological studies of the human heart face the fundamental challenge that experimental data can be acquired only from patients with underlying heart disease . Regarding human atria , there exist sizable gaps in the understanding of the functional role of cellular Ca2+ dynamics , which differ crucially from that of ventricular cells , in the modulation of excitation-contraction coupling . Accordingly , the objective of this study was to develop a mathematical model of the human atrial myocyte that , in addition to the sarcolemmal ( SL ) ion currents , accounts for the heterogeneity of intracellular Ca2+ dynamics emerging from a structurally detailed sarcoplasmic reticulum ( SR ) . Based on the simulation results , our model convincingly reproduces the principal characteristics of Ca2+ dynamics: 1 ) the biphasic increment during the upstroke of the Ca2+ transient resulting from the delay between the peripheral and central SR Ca2+ release , and 2 ) the relative contribution of SL Ca2+ current and SR Ca2+ release to the Ca2+ transient . In line with experimental findings , the model also replicates the strong impact of intracellular Ca2+ dynamics on the shape of the action potential . The simulation results suggest that the peripheral SR Ca2+ release sites define the interface between Ca2+ and AP , whereas the central release sites are important for the fire-diffuse-fire propagation of Ca2+ diffusion . Furthermore , our analysis predicts that the modulation of the action potential duration due to increasing heart rate is largely mediated by changes in the intracellular Na+ concentration . Finally , the results indicate that the SR Ca2+ release is a strong modulator of AP duration and , consequently , myocyte refractoriness/excitability . We conclude that the developed model is robust and reproduces many fundamental aspects of the tight coupling between SL ion currents and intracellular Ca2+ signaling . Thus , the model provides a useful framework for future studies of excitation-contraction coupling in human atrial myocytes .
In cardiac myocytes , the process triggered by the action potential ( AP ) and resulting in the contraction of the myocyte is commonly referred to as excitation-contraction coupling ( ECC ) [1] . The transient elevation of intracellular Ca2+ concentration ( [Ca2+]i ) that underlies the contraction is initiated by the Ca2+ influx from the extracellular space through the L-type calcium channels ( LTCCs ) , which causes the release of more Ca2+ from the sarcoplasmic reticulum ( SR ) via the SR calcium release channels ( ryanodine receptors; RyRs ) , This mechanism is known as Ca2+ induced Ca2+ release ( CICR ) [2] . During one contraction cycle , the Ca2+ influx has to be balanced with an efflux to the same compartments , in order that Ca2+does not start to accumulate and impede contraction . The majority of Ca2+ is re-circulated back to the SR by the SR Ca2+-ATPase ( SERCA ) , leaving a smaller fraction of Ca2+ to be extruded from the cell by the Na+/Ca2+ exchanger ( NCX ) and plasmalemmal Ca2+-ATPase ( PMCA ) . Whilst the same CICR mechanism initiates the transient elevation of [Ca2+]i in both ventricular and atrial myocytes , there are substantial spatiotemporal differences in the properties of the atrial and ventricular Ca2+ transients [3] , [4] due to the divergent intracellular ultrastructures . Mammalian atrial myocytes lack a prominent transverse tubular system [5] , which in ventricular myocytes establishes the tight coupling of the SR to sarcolemma , enabling a Ca2+ release that is virtually uniform throughout the cell [6] . In atrial myocytes , however , the Ca2+ wave arises in the periphery ( junctional-SR ) and then propagates to the center of the cell , activating secondary release from the corbular ( non-junctional ) SR compartments [3] , [7] , [8] . Facing the complexity of a highly integrated and interdependent system , mathematical modeling has become an established complement to the experimental approach in elucidation of the mechanisms that underlie cardiac electrophysiology [9] . In human studies , the role of mathematical modeling is perhaps even more important because there are substantial limitations in the quantity and quality of the human cardiac tissue that is available for in vitro experiments . In this study , we present a model of the adult human atrial myocyte that has a spatially detailed and physiologically based formulation of the Ca2+ release from and uptake to the SR . Based on a realistic description of the interrelations between [Ca2+]i , sarcolemmal ( SL ) ion currents , and the SR Ca2+ release , the aim of this study was to elucidate to what extent intracellular Ca2+ regulates the AP waveform , cellular excitability , and rate-dependent electrophysiological mechanisms . The results indicate that the peripheral ( junctional ) SR Ca2+ release sites define the interface between Ca2+ and AP , whereas the central ( non-junctional ) release sites are important for the fire-diffuse-fire propagation of Ca2+ diffusion . Moreover , our analysis suggests that the rising intracellular Na+ concentration is an important modulator of the action potential duration at increasing heart rates . Finally , the results predict that SR Ca2+ release is a strong modulator of AP duration and thus affects the human atrial myocyte refractoriness .
The dynamic components of the model cell are the membrane current-voltage system and the intracellular Ca2+ , Na+ , and K+ and SR Ca2+ concentrations ( Figure 1 ) . The intracellular and SR Ca2+ also have a spatial dimension . The cell was modeled as a cylinder , with a length of 122 . 051 µm and a radius of 6 . 02 µm ( Figure 1B ) . These yield a 50 pF capacitance for the cell membrane when the top and bottom of the cylinder are not included in the area [10] . The intracellular space is divided into the junctional cytosol , which is a 0 . 02 µm deep region below the cell membrane [11] , [12] , and the bulk cytosol , which represents the rest of the cytosol below the junctional cytosol . The bulk cytosol and the SR are further divided into four 1 . 625 µm deep compartments ( Figure 1B ) . The RyR and SERCA in the first SR compartment interact with the junctional cytosol compartment , whereas the RyRs and SERCAs in the other three SR compartments interact with the corresponding bulk cytosolic compartments ( Figure 1B ) . This structure in our model is similar to the structure seen in immunolabeleled images of RyRs and SERCAs in atrial myocytes , in which the non-junctional release sites form a regular structure with ∼2 µm distances in the transverse direction and the junctional release site is apart from this structure [12] , [13] . The volume of the SR was set to 2 . 25% of the volume of the bulk cytosol in each of these compartments , and thus also in the whole cell [14] . The complete set of geometrical and physical parameters is shown Table 1 . The Ca2+ diffusion in the bulk cytosol and in the SR was modeled with Fick's second law of diffusion . The components of cytosolic Ca2+ buffering are shown in Table 2 . The effect of the mobility of Ca2+ buffers in the bulk cytosol was implemented as described previously [15] . The amount of SR Ca2+ buffer ( calsequestrin ) was fitted based on experimental SR Ca2+ content [16]; see Results for details . The diffusion between the junctional and bulk cytosol was modeled as an analytical diffusion equation [17] . The accessible volume for the Ca2+ diffusion in the cytosol and in the SR was set to 50% of the total volume of the compartments [11] . Also the accessible area for Ca2+ diffusion between junctional and bulk cytosol was set to 50% of the total area between these compartments . The diffusion coefficients are shown in Table 2 . Previously , it has been shown that the effective diffusion coefficient for Ca2+ in the SR is 8–9 µm2/s [18] . The effective diffusion coefficient is smaller than the free diffusion coefficient due to the Ca2+ buffering in the SR [15] . In our model , the free diffusion coefficient for Ca2+ in the SR is 44 µm2/s , which yields an effective coefficient of 8–12 µm2/s in the relevant range of [Ca2+]SR ( 0 . 3–0 . 6 mM ) . The components of cytosolic Ca2+ buffering have not been characterized from atrial myocytes in the same detail as in ventricular myocytes [1] . Previously , ventricular data has been used to model the atrial Ca2+ buffering [11] . However , implementation of a ventricular buffering system in our model resulted in almost non-existent Ca2+ transients with the experimental SR Ca2+ content [16] ( data not shown ) . In our model , the ‘known’ cytosolic Ca2+ buffers are the sarcolemma , which is assumed to be as in ventricular myocytes based on [11] , and the SERCA [19] , which is fitted based on Ca2+ transient kinetics and SR Ca2+ content . The rest of the buffering , consisting most likely of troponin , calmodulin and myosin , is modeled as an single arbitrary mobile buffer , which has the mobility and Kd of calmodulin [11] . The Ca2+ buffering to and uptake by the SERCA and the passive Ca2+ leak from the SR to the cytosol were modeled as previously [19] . See Table S1 in the Supporting Information for parameter values . To describe the Ca2+ release flux through the RyR channel , we developed a novel phenomenological model ( Figure 2 ) that employs a Hodgkin-Huxley type of formalism . The model has three gating variables ( Figure 2A ) : adaptation gate , open gate and closed gate . The complete set of equations ( Text S1 ) and the parameter values ( Table S2 ) are listed in the Supporting Information . Time constants of RyR gating were adjusted so that the calcium release from a release unit situated in non-junctional compartments represents a signal that is spread wider both spatially and temporally compared to the junctional space ( Table S2 ) . The simple structure makes the RyR model computationally efficient , especially for modeling the Ca2+ wave propagation , and limits the amount of unnecessary free variables . However , as shown in Figure 2B , the model is still capable of reproducing complex features of RyR Ca2+ release , i . e . the dependence of the release on both the intracellular and SR [Ca2+] and the adaptation of the RyR open probability's dependence on intracellular Ca2+ [20] , [21] . The SL ion currents were mostly formulated as in the previously published model of human atrial AP [10] . All the major modifications and novel features of the ion current submodels are described in the Text S1 . Minor adjustments of parameter values are listed in Table S1 .
As in other excitable cells , APs in atrial myocytes reflect the coordinated activation of several voltage-gated inward ( depolarizing ) and outward ( repolarizing ) ion channel currents ( Figure S3 in the Supporting Information ) . The major depolarizing currents in the initial phase of the AP are the INa and ICaL , and INCX during the later phase of the AP ( Figure S3 B&C&G ) . The Ito and Isus generate large repolarizing currents in the beginning of the AP , which quickly repolarize the membrane voltage back to −30 mV after the spike ( Figure S3 A&D ) . Following this , the repolarization is carried mostly by IK1 ( Figure S3E ) with very little contribution by IKs and IKr ( Figure S3F ) . Although there is a significant amount of If present in human atrial myocytes [22] , it does not contribute substantially to the action potential ( Figure S3F ) , since it is activated at voltages below −80 mV ( see Figure S2 ) . The PMCA creates only a very small current ( Figure S3G ) . As a principal validation , we compared the characteristics of the emergent AP waveform of our model to published data during 1 Hz pacing ( 3; see also Discussion ) . The model reproduces the experimental values for the resting membrane potential ( −77 mV ) , AP upstroke velocity ( 170 mV/ms ) , AP amplitude ( 119 mV ) , and AP duration ( APD; APD30 = 11 ms APD90 = 239 ms , ) at different stages of repolarization ( see Table S3 in the Supporting Information for detailed comparison ) . In our model , the average cytosolic Ca2+ signal has a resting concentration of 0 . 15 µM and an amplitude of 0 . 58 µM ( Figure 3A ) at 1 Hz pacing . The reported single exponential decay constants of the Ca2+ transient range from 92 ms to 160 ms [4] , [23] , [24] and in comparative studies between atrial and ventricular myocytes , the atrial myocytes have more rapid decays [3] , [4] , [25] . In our model , the decay constant of the Ca2+ transient is 131 ms , which is in line with data from atrial myocytes and more rapid than that reported in human ventricular myocytes [26] ( Table 3 ) . As a high-level validation of the relative contribution of Ca2+ transport mechanisms that underlie Ca2+ dynamics of the model , we simulated the effect of elevation of extracellular Ca2+ from 0 . 9 to 3 . 2 mM ( data not shown ) . In line with experimental findings [27] , the diastolic and systolic [Ca2+]i increased by 52% and 88% , respectively . Although the local Ca2+ release homogenizes the cytosolic Ca2+ signal , the Ca2+ transients at different parts of the cytosol differ from the average signal ( Figure 3A ) . In atrial myocytes , the [Ca2+]i can peak in the periphery of the cell to ∼1 µM during slow pacing , whereas in the center there is only a small increase in Ca2+ concentration [28] . This spatial heterogeneity is reproduced by our model ( Figure 3 A&B ) . Also the temporal characteristics of the simulated Ca2+ dynamics agree well with in vitro findings . The time-to-peak for Ca2+ in peripheral cytosol in atrial cells has been estimated to be ∼20 ms [28] and 57 . 1±4 . 0 ms [3] , and at the center of cell 123 . 7±7 . 8 ms [3] . In the model , the peripheral peak is at 34 . 8 ms and the central peak at 123 . 6 ms , in line with the experimental data . As shown in Figure 3 C&D , the delay between the peripheral and central Ca2+ release in atrial myocytes yields a biphasic increment during the upstroke of the whole cell Ca2+ transient [7] . This phenomenon is reproduced in our model with similar upstroke dynamics ( Figure 3E ) as recorded in human atrial myocytes [7] . We simulated a voltage clamp experiment with a corresponding protocol and obtained the release rates with a linear fit to normalized Ca2+ transients . In human atrial myocytes , the diastolic SR Ca2+ content , measured as the integral of NCX current during a caffeine-induced Ca2+ transient , has been reported to be 8 . 3±1 . 2 amol/pF [16] . Based on the calculations of Hove-Madsen et al . [16] , this corresponds to 50 . 9±7 . 4 µM of accessible cytosolic volume with the dimensions of our model . The SR Ca2+ content in our model is 76 . 2 µM . However , if we use the model to reproduce the caffeine-pulse experiment [16] and calculate the integral of the generated NCX current , we get a comparable value of 7 . 5 amol/pF for the SR Ca2+ content in the model ( data not shown ) . A possible source for the difference between the integrated NCX value and the actual SR Ca2+ content is the fraction of Ca2+ that is extruded from the cell by the PMCA , which was not considered in the experimental analysis [16] . The Ca2+ release from the SR generates 79±6% of the Ca2+ transient amplitude in human atrial myocytes [7] and 77% in our model ( Figure 4D ) . Most of the Ca2+ release is generated in the junctional compartment ( Figure 4 A&B ) . During the uptake of Ca2+ from the cytosol to the SR , the SERCA buffers the Ca2+ and generates a delay in the fluxes between cytosol to SERCA ( Figure 4C ) and SERCA to SR ( Figure 4D ) . At the end of the diastolic phase there is some diffusion of Ca2+ in the SR , which balances the concentration differences in different parts of the SR ( Figure 4D ) . Having established that the AP characteristics and Ca2+ dynamics of our model are in line with in vitro findings , we wanted to exploit the potential of the model to elucidate the roles of junctional and non-junctional SR Ca2+ release sites . Accordingly , we conducted two in silico experiments , in which either the non-junctional ( Figure 5A ) or junctional ( Figure 5C ) SR Ca2+ release was blocked . Results indicate that inhibition of the release of non-junctional sites in the bulk cytosol has only a small impact on the ECC ( Figure 5A ) . The amplitude of the global Ca2+ transient is decreased by 31% and the APD at 90% repolarization ( APD90 ) is decreased by 10% ( Figure 5B ) . The AP appears to be shortened because the lower [Ca2+]i in the junctional compartment does not activate the depolarizing INCX ( Figure 5 B ) to the normal extent . The most significant effect of inhibiting the release sites in the bulk cytosol is that the Ca2+ signal becomes relatively inhomogeneous and it is not carried at all to the center of the cell ( Figure 5A , lower panel ) . Thus , the release sites in the bulk cytosol are not a significant source of Ca2+ but act more as amplifiers of the Ca2+ signal during fire-diffuse-fire propagation . Compared to the previous scenario , inhibition of the junctional SR Ca2+ release site yields partly opposite results ( Figure 5C ) . The amplitude of the global Ca2+ transient is decreased by 58% and the APD90 is decreased by 34% ( Figure 5D ) . The AP is shortened , because the [Ca2+]i in the junctional compartment is elevated only slightly and thus the activation of the depolarizing INCX is reduced dramatically ( Figure 5 D ) . However , similar to the situation where the release sites in the bulk cytosol were inhibited ( Figure 5A ) , inhibition of the junctional release results in a failure in the propagation of the Ca2+ signal to the center of the cell . Without the junctional SR Ca2+ release , the Ca2+ signal coming from the LTCCs is too weak to trigger the CICR at the first release site in the bulk cytosol ( Figure 5C , lower panel ) . Above we have shown that our myocyte model can reproduce the experimentally observed impact of SR Ca2+ release on the inactivation of ICaL ( Figure S1C ) and amplitude of the Ca2+ transient ( Figure 4B ) . In addition , our results indicate that the SR Ca2+ release , especially junctional , affects the membrane voltage also via the NCX ( Figure 5 B&D ) . To further study the effect of intracellular Ca2+ dynamics on AP morphology , we simulated the acute effect of total block and 3-fold increase of Ca2+ release ( Figure 6 ) . The results indicate that the amplitude of the Ca2+ transient ( Figure 6A ) has a substantial effect on the APD ( Figure 6B ) . Blocking the SR Ca2+ release slows down the early phase of repolarization ( Figure 6B , inset ) , but speeds up the late repolarization ( Figure 6B ) , whereas the 3-fold increase of Ca2+ release has an opposite effect . Slowed AP repolarization causes the sodium channels to remain inactivated for a longer time . Consequently , the duration of the refractory period is increased and a prematurely applied second stimulus is unable to trigger the next AP ( Figure 6D ) . To further dissect the role of [Ca2+]i in AP morphology , we simulated separately the effect of decay and amplitude modulation of intracellular Ca2+ transients on the APD ( Figure 7 ) . That is , the junctional [Ca2+] was “clamped” to three different modes ( Figure 7 A&E ) in both cases . This modification was implemented by replacing differential variable of junctional [Ca2+] with an analytical equation that was fitted manually to the simulated control [Ca2+] trace . Then either the decay ( Figure 7A ) or the amplitude ( Figure 7E ) of the junctional Ca2+ transient was modified . As the results show , the accelerated decay shortens the AP substantially , whereas deceleration of decay has an opposite effect ( Figure 7B ) . The decay modulation has little or no effect on the early repolarization of the membrane voltage ( Figure 7B , inset ) . The underlying mechanism appears to be the changed INCX ( Figure 7C ) , whereas ICaL is not affected ( Figure 7D ) . Similar to the effect of decay modulation , the amplitude of the [Ca2+] transient affects the APD substantially ( Figure 7F ) . However , compared to the modulation by the [Ca2+]i transient decay , increasing the amplitude also affects the early phases of repolarization ( Figure 7F , inset ) by enhancing the outward peak current via the NCX and accelerating the inactivation of ICaL ( Figure 7H ) . This results in a more complex modulation scheme , in which the increased [Ca2+] transient amplitude accelerates the early repolarization and decelerates the late repolarization ( Figure 7F ) . Increasing the pacing rate causes an immediate ( within a few APs ) and then a gradual ( reaching steady state over several minutes ) decrease in the APD of atrial myocytes; this has been shown in numerous studies . Experimental findings indicate that this adaptation coincides with functional changes in the LTCC following calcium overload [29] , [30] , and it is thus seen as one of the main mechanisms that underlie the changes in the ADP [31] , [32] . However , in ventricular myocytes , one of the important factors in rate dependence has been shown to be the accumulation of cytosolic Na+ during fast pacing [33] , [34] , [35] , [36] . To further study this phenomenon in atrial myocytes , we simulated pacing experiments within a physiologically relevant range of frequencies or basic cycle lengths ( BCLs ) . To account for the other rate-dependent mechanisms that affect the APD , we implemented two additional variants of the myocyte that are described in detail in the Supporting Information ( Text S1 ) . Briefly , we added a subsarcolemmal Na+ compartment ( model variant: “vCaNass” ) to the developed model ( vCa ) , and we also included the recently updated description of K+ currents according to [37] ( vCaNassIk ) . Simulation results shown in Figure 8 were obtained as in a previously used experimental protocol [38] . Figure 8A shows the overall changes in [Ca2+]i dynamics: diastolic and systolic [Ca2+]i increase and decrease with faster pacing , respectively , and accumulation of intracellular Na+ . The simulated values of APD30 that were calculated for each BCL fall within the range of reported experimental values ( Figure 8B ) . The steep dependence of APD90 on the BCL is reproduced faithfully by the model ( Figure 8C ) . Although the variance of absolute APD90 values reported in the literature is large , the shape of the curve is similar in experiments and simulations . The relative change of APD90 in the range of BCL = [1600 , 400] in our model ( −32% , −30% and −27% for the model variants vCa , vCaNass and vCaNassIk , respectively ) fits well to the in vitro values of Boutjdir et al . [39] ( −44% ) and Dawodu et al . [38] ( −38% ) . A multitude of measurement protocols are used to study the rate dependence of AP morphology in cardiac myocytes . Most importantly , the length of the period , after which the APD is determined , ranges from tens of seconds [31] , [40] to five minutes [41] . As the continued pacing potentially increases the [Na+]i , we wanted to study the rate dependence of AP with a second pacing protocol , in which simulation is always started from the quiescent steady-state and continued for five minutes for each BCL , separately ( representative data is shown in Figure 8D ) . To account for all rate-dependent mechanisms described above , we performed these simulations with the vCaNassIk model variant . Comparison of APD90 after 30 seconds and 5 minutes of pacing highlights the dynamic nature of rate dependence ( Figure 8E ) . That is , the shortening of the AP in response to faster pacing becomes more pronounced as pacing is continued . This is affected substantially by the increasing [Na+]ss ( Figure 8D ) . If the intracellular [Na+] is “clamped” to the quiescent steady-state value ( 7 . 8 mM ) , the effect of continued fast pacing on the APD is dramatically reduced ( Figure 8E , black line ) . Furthermore , the simulation results suggest that Na+ accumulation is actually a mechanism that is required for rate-dependent adaptation of APD , because the myocyte model failed to produce a normal AP at BCL 300 ms when [Na+] was “clamped” the quiescent steady-state value ( Figure 8E , open triangle in the black trace ) . This failure was present already during the first few seconds of the simulation at BCL 300 ms ( open diamond , in the grey trace ) . Results shown in Figure 8E suggest that roughly half of the rate-dependent adaptation of APD comes from the short-term ionic mechanisms that operate in the timescale of seconds , whereas the long-term adaptation responsible for the other half of APD decrease takes minutes to develop . Furthermore , the longer pacing protocol results in a much steeper dependence of APD in the BCL range of [1600 , 400] ms ( −35% ) , compared to the shortening of APD ( −27% ) shown in Figure 8C . Simulation results suggest that the mechanism that links AP shortening ( Figure 8F ) to Na+ accumulation is the Na+/K+-ATPase ( NKA ) . The enhanced pumping function of NKA lead to a dramatic increase in the current ( INKA ) ( Figure 8G ) . Accordingly , this mechanism has been reported previously in human atrial fibers [42] and guinea pig ventricular myocytes [43] . To confirm that this feature of the myocyte model does not depend on the Na+ parameters of model variant , we simulated the same protocol with model variant vCa ( no junctional Na+ compartment ) and found that fast pacing resulted in a similar increase in INKA ( data not shown ) . To evaluate the effect that Na+ accumulation due to fast pacing has on the Ca2+ dynamics via modulation of the NCX function , we simulated a previously used pacing protocol [44] , in which the pacing frequency was increased in a stepwise manner from 1 to 2 Hz and further to 3 Hz . The results provide a general level validation for the rate dependence of Ca2+ dynamics of the model ( Figure 9A ) that correspond qualitatively to the measured force ( Figure 9B ) . To further study the role of NCX in the rate dependence of Ca2+ dynamics , we evaluated the role of the NCX as a secondary trigger of the CICR process . Figure 9E shows that whereas the Ca2+ influx via LTCC is decreased slightly ( integral decreases by ∼14% ) during fast pacing , the flux via reverse mode of NCX is increased considerably ( integral increases over two-fold ) . However , since relative contribution of NCX is much smaller , we performed an additional simulation analysis to quantitate its role in rate-dependent modulation of Ca2+ dynamics . In line with a previous study [45] , blocking the reverse mode of the NCX delays the peak of [Ca2+]i by 1 . 8 ms ( 109 . 1 ms vs . 110 . 9 ms for control and block , respectively ) when the pacing frequency is 1 Hz . This effect is more pronounced during fast pacing: the delay increases to 2 . 4 and 3 . 5 ms with 2 and 3 Hz , respectively . Thus , the results suggest that the rate-dependent modulation of NCX could mediate a shortening of the delay between the electrical excitation and the peak of Ca2+ transient .
It is well known that in the human atria the cellular AP shape varies depending , for example , on gender , age and personal ongoing medication plan . In addition to the physiological variation , studies using human tissue samples are always affected by the parallel pathophysiology of the patients who are undergoing surgery due to some other cardiac malfunction . Experimentally , the morphology of the human atrial AP ranges from a triangular AP shape with no sustained plateau to a long AP with a spike-and-dome shape [46] , [47] , [48] . The AP heterogeneity has been shown to correlate tightly with the relative expression levels of ion channels in different regions of the atria in an experimental canine model [49] and in human patients [50] . In the model presented here , the AP characteristics ( resting membrane potential , upstroke velocity , amplitude and duration ) emerging from an accurate description of individual ion currents are well within the range of experimental data reported in the literature . During the last decade , mathematical modeling has become an established complement to experimental work in attempts to elucidate the ionic mechanisms that underlie the electrophysiology of cardiac myocytes [9] . In the case of human atrial myocyte models , the platform was established by the individual works of Nygren et al . [10] and Courtemanche et al . [31] . The usability of these comprehensive frameworks has been established , e . g . in consecutive in silico studies of AP morphology [37] , [51] and atrial fibrillation [52] , [53] , [54] . While these models provide a detailed description of the transmembrane ion currents , very little emphasis has been placed on the accurate description of the spatiotemporal properties of intracellular Ca2+ . Experimental findings indicate that there are substantial spatiotemporal differences in the properties of the atrial and ventricular Ca2+ transients [3] , [4] , thus we have developed a novel model that considers the atrial-specific properties of Ca2+ signaling . In contrast to the virtually uniform Ca2+ release in ventricular myocytes [6] , it is characteristic for atrial myocytes that the Ca2+ wave , initiated by the CICR mechanism , arises first in the periphery and then propagates to the cell center [3] , [7] , [8] . Consequently , the delay between the peripheral and central Ca2+ release in atrial myocytes yields a biphasic increment during the upstroke of the Ca2+ transient [7]; a phenomenon that is reproduced in our model with accurate spatiotemporal parameters . This delay can be decreased by inotropic interventions that promote increased SR Ca2+ content , and consequently enhanced SR Ca2+ release; thus , it establishes a mechanism through which the interval between the electrical excitation and the peak of Ca2+ transient can be modulated . Our results highlight the crucial role of the junctional SR in mediating the CICR , while the inhibition of non-junctional Ca2+ release sites causes only an attenuation of the Ca2+ signal during fire-diffuse-fire propagation . Furthermore , junctional SR Ca2+ release sites appear to define the interface between Ca2+ and AP , i . e . , decreasing the junctional Ca2+ release shortens the AP substantially . These findings are in line with the previously suggested role of the non-junctional SR as an inotropic release reserve that can be recruited when greater contractility is required [13] . Cumulative evidence suggests that changes in the [Ca2+]i homeostasis may initiate electrical remodeling during atrial fibrillation , which is characterized by a marked shortening of the action potential plateau phase [55] , [56] , [57] . In future studies , the presented model , which is based on an atrial-specific description of Ca2+ signaling , has thus great potential in elucidating the function of the remodeled cells with altered Ca2+ homeostasis and SL ion currents . The tight coupling of SL ion currents , which underlie the AP shape , and the [Ca2+]i has been well established in experimental setups of both ventricular [33] , [58] , [59] and atrial myocytes [48] , as well as in computational studies [36] , [60] . Our results suggest that both amplitude and decay modulation of the Ca2+ transient produce significant changes to the APD compared to the control situation . Changes in [Ca2+]i in the vicinity of the SL affects both INCX and ICaL . However , while both increased amplitude and decelerated decay of the Ca2+ transient enhance the inward INCX , the inactivation of ICaL is affected only by the former . These findings indicate that in human atrial myocytes the NCX is more important than the LTCC in linking the amount of Ca2+ released from the SR to changes in the APD . This scheme also concurs with a previously reported role of NCX as the main mediator of the inotropic effect of AP prolongation in canine atrial myocytes [61] . Understanding the role of this mechanism is of great importance because it not only links inotropic interventions ( increased amplitude of the Ca2+ transient ) but also situations such as hypothyroidism ( slowed decay of the Ca2+ transient ) to APD modulations . When the effects of amplitude and decay modulation of the Ca2+ transient are combined to a strongly modulated physiological frameset of blocked or increased SR Ca2+ release , the resulting changes in AP morphology are even more pronounced . That is , decreasing the release decelerates the early repolarization and accelerates the late repolarization of the membrane voltage , whereas increased SR Ca2+ release promotes a faster early repolarization and a slower late repolarization of the AP . Most interestingly , the increased SR Ca2+ release enhances the refractoriness of the AP . That is , the decelerated repolarization of the AP ( due to increased INCX ) slows down the recovery from inactivation of the sodium current and thus a premature electrical stimulus is unable to trigger a second AP . The primary physiological context , in which the duration of the AP is modulated , relates to changes in heart rate [62] . Numerous parameters of the in situ heart are affected simultaneously in that dynamic process , but in single cell preparations the situation is significantly simpler . In ventricular myocytes , one of the main mechanisms that has been shown to underlie the rate dependence of the AP is the accumulation of cytosolic Na+ during fast pacing [33] , [34] , [35] , [36] . That is , as the pacing rate increases , the Na+ influx per unit time is increased . Since this change is not fully compensated with increased efflux , [Na+]i increases with continued fast pacing . As previously demonstrated in ventricular myocytes [35] , this mechanism emerges from the interplay of NKA and NCX . If only NKA or NCX senses the accumulated [Na+]i in response to continued fast pacing , the APD is decreased less compared to the normal situation , in which both of them sense an increased [Na+]i . A similar phenomenon is apparent in our atrial model: the mere increase in pacing frequency ( or reduction in BCL ) does not produce a maximal shortening of the AP . Rather , the increase of [Na+]i and the consequent change in the NKA function , caused by the continued high frequency pacing , promote a substantial further decrease of APD . Furthermore , the results indicate that the other ionic determinants of APD , emerging from the rate dependence of Isus [37] and Ca2+ dynamics , are effective already within a shorter time-scale ( ∼10 s ) of adaptation compared to that of Na+ ( ∼minutes ) . Thus , our findings underline the importance of considering the effect of Na+ accumulation in both in vitro and in silico studies of AP rate dependence when results obtained with different protocols are interpreted or compared . In myocytes with long AP , the increase of Ca2+ transient shortens the AP ( via inactivation ICaL ) . Interestingly , our simulation results indicate that , in this respect , the human atrial myocytes behave similar to animal with short AP . That is , increased Ca2+ transient amplitude ( due to e . g . enhanced SR Ca2+ release ) promotes an inward INCX , thus lengthening the AP . The role of the NCX as a trigger of SR Ca2+ release has been studied extensively in ventricular myocytes during recent years; the findings , however , are controversial [60] . While some results indicate that Ca2+ entry via the reverse mode of NCX is significant in physiological conditions [63] , [64] , other studies suggest it to be important only in pathological situations [1] , [65] . Our simulation results suggest that in atrial myocytes the contribution of NCX to the Ca2+ influx is increasingly significant at higher pacing rates , even though the bulk of the flux goes via the LTCC . However , when the temporal role of these two mechanisms is compared , it is apparent that during the first few milliseconds of the CICR process the NCX contributes rather equally to the Ca2+ influx . Accordingly , the acute inhibition of the reverse mode of NCX delays the peak of [Ca2+]i by a few milliseconds , depending on the pacing rate . These findings are in line with the reported delay of the Ca2+ transient if the Ca2+ entry via the NCX is inhibited in ventricular myocytes [45] . Our results thus indicate that the NCX could have an important role in accelerating the rise of [Ca2+]i with increasing heart rate , even though LTCC is the primary trigger of CICR , as suggested previously [66] . Hence , the modulation of the NCX could promote a rate-dependent reduction of the electro-mechanical interval in the development of contractile force . The in vivo significance of the delay modulation by the NCX presents an interesting question for future simulation studies in tissue and/or whole-atria models . Variability of experimental results imposes a fundamental challenge for modeling studies that utilize data measured from isolated human atrial tissues or cells . It would thus be an irrelevant and futile effort to try to fit the electrophysiological characteristics of the model perfectly to one single set of in vitro data . Instead , it is more essential that the simulation results agree qualitatively and semiquantitatively with the majority of the measured results , which has been established for the AP morphology in our model . There is a further challenge for validation of the Ca2+ dynamics due to incomplete experimental data . To compensate for this , we have also compared the model behavior to ventricular data and measurements obtained with atrial animal models . The developed RyR module of the myocyte model is an approximate description of macroscopic SR Ca2+ release , sharing the general limitations ‘common pool’ models . Thus , the justification for most of the chosen parameter values cannot be derived directly from the biophysical properties of RyR channels [67] , [68] . Instead , the ad hoc parameter values of the RyR modules are based on indirect fitting: the dynamics of intracellular Ca2+ were adjusted to be in line with macroscopic experimental observations [3] , [7] , [8] . The simple structure makes the RyR module computationally efficient , while still being capable of reproducing essential features of RyR Ca2+ release characterize the emergent properties of atrial Ca2+ dynamics: biphasity of the increasing Ca2+ concentration during the Ca2+ transient , and the relative contribution of SL and SR Ca2+ fluxes to the Ca2+ transient . A more complex RyR module might be needed in future studies if the simulations are run beyond the conditions that were investigated in this study . Although our model can reproduce rather accurately the rate-dependent changes in ion dynamics , it should be noted that the description of the underlying mechanisms is by no means exhaustive . For example , one potential regulatory pathway is the Ca2+/calmodulin-dependent protein kinase ( CaMK ) II that has been studied extensively in ventricular myocytes [69] . Recent findings indicate that CaMKII could , for example , have an important role in the regulation of RyR [70] in atrial myocytes . As more experimental data become available , the effects of CaMKII on its phosphorylation targets should be considered in future modeling studies . We have chosen to use Hodgkin-Huxley formalism for the ion currents . We acknowledge that Markovian models would allow for a more detailed description of the complex kinetics of processes ( activation , deactivation , inactivation , and recovery from inactivation ) that the channels exhibit . However , it can be very difficult to meet the information requirements for defining the transitional rate constants [71] . Furthermore , Markovian models can be computationally expensive compared to Hodgkin-Huxley formalism [71] . We conclude that the novel myocyte model provides significant insight into the excitation and ion dynamics of human atria . Our results underline the tight coupling of AP morphology to SR Ca2+ release and intracellular Na+ that are subject to strong modulation under physiological conditions . Furthermore , with a physiologically accurate description of intracellular Ca2+ dynamics the model is a potential tool , for example , in the elucidation of mechanisms that link changes in Ca2+ homeostasis to pathophysiological conditions . Thus , it offers attractive possibilities to study the electrical remodeling during atrial fibrillation , and to find potential targets that affect the refractoriness of the AP . Finally , as the computational cost of the model is low , it is also a feasible component for multi-scale models of tissue and/or heart . | In the human heart , the contraction of atrial and ventricular muscle cells is based largely on common mechanisms . There is , however , a fundamental difference in the cellular calcium dynamics that underlie the contractile function . Here , we have developed a computational model of the human atrial cell that convincingly reproduces the experimentally observed characteristics of the electrical activity and the cyclic fluctuations of the intracellular calcium concentration . With the model , we evaluate the relative roles of the most important cellular calcium transport mechanisms and their impact on the electrical behavior of the cell . Our simulations predict that the amount of calcium released from the cellular stores during each electrical cycle crucially regulates the excitability of the human atrial cell . Furthermore , the results indicate that the cellular sodium accumulation related to faster heart rates is one of the main mechanisms driving the adaptation of cardiac electrical activity . Finally , we conclude that the presented model also provides a useful framework for future studies of human atrial cells . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"cell",
"biology/cell",
"signaling",
"cardiovascular",
"disorders/arrhythmias,",
"electrophysiology,",
"and",
"pacing",
"physiology/cardiovascular",
"physiology",
"and",
"circulation",
"physiology/muscle",
"and",
"connective",
"tissue",
"computational",
"biology/signaling",
"netw... | 2011 | Impact of Sarcoplasmic Reticulum Calcium Release on Calcium Dynamics and Action Potential Morphology in Human Atrial Myocytes: A Computational Study |
According to World Health Organization ( WHO ) prevalence estimates , 1 . 1 million people in Mexico are infected with Trypanosoma cruzi , the etiologic agent of Chagas disease ( CD ) . However , limited information is available about access to antitrypanosomal treatment . This study assesses the extent of access in Mexico , analyzes the barriers to access , and suggests strategies to overcome them . Semi-structured in-depth interviews were conducted with 18 key informants and policymakers at the national level in Mexico . Data on CD cases , relevant policy documents and interview data were analyzed using the Flagship Framework for Pharmaceutical Policy Reform policy interventions: regulation , financing , payment , organization , and persuasion . Data showed that 3 , 013 cases were registered nationally from 2007–2011 , representing 0 . 41% of total expected cases based on Mexico's national prevalence estimate . In four of five years , new registered cases were below national targets by 11–36% . Of 1 , 329 cases registered nationally in 2010–2011 , 834 received treatment , 120 were pending treatment as of January 2012 , and the treatment status of 375 was unknown . The analysis revealed that the national program mainly coordinated donation of nifurtimox and that important obstacles to access include the exclusion of antitrypanosomal medicines from the national formulary ( regulation ) , historical exclusion of CD from the social insurance package ( organization ) , absence of national clinical guidelines ( organization ) , and limited provider awareness ( persuasion ) . Efforts to treat CD in Mexico indicate an increased commitment to addressing this disease . Access to treatment could be advanced by improving the importation process for antitrypanosomal medicines and adding them to the national formulary , increasing education for healthcare providers , and strengthening clinical guidelines . These recommendations have important implications for other countries in the region with similar problems in access to treatment for CD .
Chagas disease is clinically manifested in two stages – an acute stage and a chronic stage . The acute stage lasts for approximately 4–8 weeks and is characterized by flu-like symptoms or a characteristic local swelling at the site of parasite entry [8] , [9] , following which an infected person enters the indeterminate form of the chronic phase of infection . Among those with the indeterminate chronic form , about 20–30% of patients progress to the chronic cardiac or digestive forms of Chagas disease [10] . The most common course of Chagasic cardiomyopathy includes conduction system abnormalities early in the disease , resulting in heart failure . In all phases , serological tests such as the enzyme-linked immunosorbent assay ( ELISA ) test , the indirect haemagglutination assay ( IHA ) , and the indirect immunofluorescent antibody test ( IIF ) are used for diagnosis [4] , [9] , [11] . Because these tests can be difficult to interpret , the WHO recommends the use of two concomitantly positive tests to make a confirmed diagnosis [11] , [12] . Currently , benznidazole and nifurtimox are the only antitrypanosomal medicines available to treat T . cruzi infection . Antitrypanosomal therapy is strongly recommended by WHO for acute , congenital or reactivated infections , and for chronic infection in children under the age of 18 [13] , [14] , [15] . Recent scientific evidence about the clinical effectiveness of these medications has led to the expansion of treatment indications to include adults in the chronic phase of the disease without advanced cardiomyopathy [1] , [11] , [16] , [17] , [18] , [19] . Though no randomized controlled trial has directly compared the two medications [11] , WHO guidance and the clinical literature place greater emphasis on the use of benznidazole [4] as a first-line therapy because there is more clinical evidence for its efficacy , and it has a more favorable side-effect profile and is better tolerated by adult patients [9] , [15] , [16] , [17] , [18] , [20] . A randomized clinical trial of benznidazole is underway to determine its efficacy in slowing progression of disease among patients with early to moderate stage Chagasic cardiomyopathy [21] , [22] . Both benznidazole and nifurtimox have undergone changes to their global supply chains over the past decade . Benznidazole was manufactured by Roche until 2003 , at which time the rights and manufacturing technology were transferred to the Pernambuco state pharmaceutical laboratory in Brazil , Laboratorio Farmaceutico do Estado Pernambuco ( LaFepe ) [23] , [24] . Between 2004 and 2006 , LaFepe produced several batches of benznidazole using active pharmaceutical ingredient that was donated by Roche [24] . Then , after a period of no production , LaFepe resumed production of benznidazole in late 2011 and the medicine is now distributed by several entities including LaFepe , WHO , and Masters Pharmaceuticals . Nifurtimox is manufactured by Bayer HealthCare in El Salvador . In 2007 Bayer reached an agreement with WHO for Bayer to donate nifurtimox to WHO and for WHO to distribute the medicine through the WHO-Bayer Nifurtimox Donation Program [25] . Access to treatment for Chagas disease in Mexico must be considered in the context of the Mexican health system and its recent reforms . Mexico has three major national insurance schemes , the Instituto Mexicano del Seguro Social ( IMSS ) , Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado ( ISSSTE ) , and Seguro Popular ( SP ) [26] . IMSS and ISSSTE together offered coverage to approximately 42 . 6 million private sector ( IMSS ) and public sector ( ISSSTE ) employees in 2010 [26] . As of 2011 , SP , a social health insurance program started in 2003 , offers a package of 284 essential services to approximately 51 . 8 million Mexicans , according to the Mexican government [10] , [27] , [28] , [29] . Affiliation with SP requires a fixed family contribution that is based on a progressive scale by income , though individuals and families who fall in the lowest two income deciles are exempt from payment of a premium [26] , [27] . The national Program on Onchocerciasis , Leishmaniasis and Chagas Disease within the Mexican Secretary of Health's National Center for the Prevention and Control of Diseases ( CENAPRECE ) is the unit responsible for establishing guidelines and coordinating national activities for Chagas disease control . The State Secretaries of Health report patients who are diagnosed by ISSSTE , IMSS and SP systems to the national Program , which then turn provides medicines to treat confirmed cases . Figure 1 shows the process of case registration for a patient with Chagas disease .
IRB exemption was obtained from Harvard School of Public Health ( Protocol# 21514-101 ) and the National Institute for Public Health ( INSP ) located in Cuernavaca , Mexico . Oral informed consent was obtained from all interviewees .
Table 2 provides a list of national level obstacles to treatment access for Chagas disease , based on our analysis of data collected in this study . The list includes all obstacles that were mentioned during interviews and could be triangulated using a second data source .
This study provides evidence regarding the extent of treatment access for Chagas disease in Mexico and the barriers that influence the level of access . In particular , the study demonstrates that the number of Chagas disease cases registered at the national level in Mexico since 2007 is approximately 0 . 41% of expected cases and that 120 registered , eligible cases were awaiting treatment at the time of the study . These findings also indicate that Mexico has made an effort to register new cases and provide treatment at both the state and national level and thus show an increased commitment to addressing this disease in Mexico . Our findings also demonstrate that epidemiologic surveillance for Chagas disease remains a challenge in Mexico and that the complexity of the case registration system may delay or limit registration . Evidence from national data shows that problems in the supply chain of medicines make it difficult to ensure timely access to treatment as cases are registered and further , that the medicine provided by the national program since 2009 has exclusively been nifurtimox , a medicine that has been identified in the clinical literature and international guidelines as second-line therapy [20] . The lack of awareness and understanding of the disease and its treatment among both physicians and populations at risk was another important challenge related to the persuasion policy intervention area [46] . Patient and provider awareness of the disease has implications for efforts to strengthen epidemiologic surveillance and the willingness of physicians to treat infected patients when medicines are available . Additionally , access to treatment for Chagas disease has until 2012 been further weakened by its exclusion from the package of health interventions that are covered under SP [27] , [28] . While its addition to the CAUSES in 2012 represents an important step ( organization ) toward increasing access to treatment , clinical information about the disease is still lacking in this document and neither benznidazole nor nifurtimox is listed as a treatment for the diseases in this category . In addition to these barriers , it is important to acknowledge the role of international actors and policies as barriers to access to treatment for Chagas disease in Mexico and potentially in other countries as well . The global shortage of benznidazole in 2011 and the challenges in obtaining nifurtimox through WHO exist outside the Mexican context but directly affect efforts by the Mexican national and state control programs to increase access to treatment [23] , [24] . These findings provide new information on the state of treatment for Chagas disease in Mexico and the barriers that prevent more widespread access . Previous work on this subject has suggested that efforts to control and treat Chagas disease in Mexico are insufficient [36] , [53] but no study has previously measured the gap in access to treatment or analyzed related obstacles . In addition , a recent study estimated the economic burden associated with Chagas disease to exceed seven billion dollars globally and several studies have described the need for increased treatment globally [3] , [5] , [33] , [53] , [54] . This study is one of the first to examine the multiple complex factors within the health system that prevent more widespread treatment access in a particular country setting . It is important to note , however , that the state of Morelos did successfully procure benznidazole and offers an important case for showing how a state can take significant initiative in improving access to treatment for Chagas disease . Some of the findings from the Mexican experience may be relevant to treatment access for Chagas disease in other countries in the region . For instance , reliance on nifurtimox as a first-line therapy in both the 2010 Mexican guidelines for vector-borne diseases and in procurement of medicines at the national level raises questions about the reasons for this choice and whether other countries may also choose to procure nifurtimox through the donation program now or in the future instead of purchasing benznidazole through the private market . In the case of Mexico , the regulatory status of the drugs , especially the lack of commercial permits for them , and the exclusion of antitrypanosomal therapies for Chagas disease from the Mexican national formulary have severely limited sources of financing to buy benznidazole , causing the national program to instead rely on the free nifurtimox . However , little information exists about whether other countries also rely on nifurtimox as a first-line therapy and if so , why . Though clinical guidelines overwhelmingly suggest that benznidazole is better tolerated and that the clinical evidence of its efficacy is more robust , clear international consensus guidelines for the treatment of Chagas disease have not been published and relatively limited data are available about the use and clinical outcomes for the two drugs by different countries around the world . There are several limitations to this study . First , data on the prevalence of Chagas disease are limited both in Mexico and globally . This constitutes an important challenge to efforts to address this disease in Mexico . In this analysis , we use the official 2010 prevalence estimate from the Mexican Secretary of Health because it is more conservative than the most recent WHO estimate and because the WHO estimate does not have a clear evidence base . This choice may result in our analysis showing greater access to treatment ( as a proportion of total infected cases ) than may actually exist in Mexico . Some actors within the Mexican Secretary of Health have argued that the epidemiology of Chagas disease in Mexico is focal and that states with a high burden of disease should undertake activities to address this disease at a state level , while others have maintained that the prevalence of Chagas disease is substantial across much of the country and that the disease should be a national priority , especially given the migration of populations from endemic areas both within Mexico and from neighboring countries to Mexico [36] , [50] . To provide a more reliable estimate of national prevalence , a nationally representative epidemiologic survey could be conducted , both nationally and by state . This would advance efforts by both the state and national programs to make more informed decisions about the priority and resources that are warranted for Chagas disease treatment . A second limitation is that we consider benznidazole as the first line antitrypanosomal medicine , despite the lack of definitive international consensus on this issue . We made this decision because benznidazole is being used exclusively as the reference treatment regimen in clinical trials of new drugs , is named as the first line therapy in the treatment guidelines of several non-governmental organizations [20] , and is cited as such in the vast majority of the clinical literature [9] , [17] , [18] . It is worth noting , however , that there is some diversity on treatment regimens within Mexico . Although the national program has used nifurtimox from the WHO donation program , the state of Morelos in Mexico has purchased benznidazole for its treatment program . Morelos registered 263 cases between 2007 and 2011 , and treated 148 cases with benznidazole and 4 with nifurtimox . This study was also limited by lack of data availability at the national and global levels . At the national level in Mexico , this included a lack of national treatment guidelines or data prior to 2010 , a dearth of information about treatment eligibility or patient refusal of treatment , and a lack of data on treatment dose , completion or clinical outcomes . In particular , it was difficult to determine what proportion of patients would be treatment eligible according to the guidelines given that no data were available on co-morbidities or patient clinical history that would allow a more thorough analysis of patients in whom treatment may be contraindicated . Furthermore , there is limited evidence about access to treatment in other countries to provide a comparison for assessing Mexico's achievement in this area . Of note , however , a recent study estimated that less than 1% of those infected with T . cruzi receive treatment globally , suggesting that the extent of access in Mexico is likely to be similar in other countries [5] . Based on these findings , there are three important strategies that could be undertaken to increase access to treatment for Chagas disease in Mexico . First , under regulation , an effort could be made to ease the importation process for these drugs . Ideally , this could be accomplished by securing COFEPRIS approval for both medicines and adding them to the national formulary , which could require actions by the relevant producers of benznidazole and nifurtimox . However , as noted above , benznidazole and nifurtimox are not approved by the United States Federal Drug Administration or the European Medicines Agency , in part because full clinical trials have not been completed for either drug . This lack of approval from two leading regulatory bodies may affect the willingness of other national regulatory bodies to approve the medicines . That said , both medications are included on the WHO Essential Medicine List [55] . In addition , clinical evidence continues to accumulate in favor of these drugs and efforts by institutions such as the Drugs for Neglected Diseases Initiative are being made to register the drugs in countries such as Colombia , Paraguay and Bolivia . In other contexts , alternative regulatory approaches such as investigational protocols are being utilized to make the drugs available [18] . Also with respect to regulation , countries with a high burden of Chagas disease may consider instituting laws that mandate rigorous epidemiologic surveillance and health education as well as prevention , diagnosis and treatment of the disease . For instance , Argentina offers a model for such legislation in National Law No . 26281 . This law requires , among other things mandatory diagnostic testing and reporting for Chagas disease in all pregnant women and in newborns in the first year of life born to infected mothers . Second , under persuasion , efforts could be expanded to provide disease-specific health education programs on Chagas disease for physicians , healthcare providers and populations at risk . Increased awareness of the disease and a better understanding of appropriate treatment methods is a critical aspect of strengthening case registration and access to treatment . In addition , health education activities have been emphasized in other national control programs such as those in Guatemala [56] and the Southern Cone initiative and have been used alongside vector control to increase awareness of the disease in high risk communities and among physicians and health workers . Increased awareness of the disease and of treatment methods is a critical aspect of strengthening case registration and access to treatment . Given the importance of this programming , the WHO and PAHO also play a potentially important role in terms of encouraging these programs and providing guidance on their design and implementation . Third , under organization , it is important to strengthen existing guidelines in Mexico for the diagnosis and treatment of Chagas disease and information availability about the supply chains for these two medicines . This includes the addition of a clinical description of Chagas disease and the two medicines to its entry in the CAUSES and the creation of a clinical guide for diagnosis and treatment as this information is critically important to strengthen awareness of treatment for Chagas disease and information for practitioners about how to diagnose and treat the disease . In addition , better public reporting of medicines released and used at the state , national and global levels is needed . In conclusion , this study found that access to treatment for Chagas disease in one high burden country ( Mexico ) is limited in important ways and identified three critical obstacles to treatment access: regulatory barriers to importation , a lack of understanding of the disease and its treatment , and a dearth of clinical guidelines [5] . Several of these barriers are likely to affect access in other countries as well , especially the lack of regulatory approval and registration of benznidazole and nifurtimox and the lack of publically available information on their supply chains . Finally , the study proposed a series of actions that could be taken in Mexico , based on a general analytical framework , to improve access to treatment for Chagas disease . These recommendations have important implications for other countries in the region with similar problems in access to treatment for Chagas disease . | Chagas disease is a vector-borne disease caused by the parasite Trypanosoma cruzi . The disease is most frequently transmitted by triatomine insects but can also be passed through blood donation or from mother to child at birth . Experts estimate that 8 million people are infected with Chagas disease globally and that 1 . 1 million of these infections are found in Mexico . Most public health programs for Chagas disease focus on preventing new infections through vector control and screening the blood supply . However , in recent years there has been a greater focus on treating the disease with one of two available medications , benznidazole or nifurtimox . This study explores access to these two drugs in Mexico . The study shows that less than 0 . 5% of those who are infected with the disease received treatment in Mexico in years . The study also identified important factors that limit access in Mexico , including the exclusion of both drugs from the national health insurance program and problems importing these medications . Finally , the paper suggests ways that these problems can be overcome in Mexico , while providing helpful insight for other countries that struggle with similar problems in treating this disease . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2013 | Barriers to Treatment Access for Chagas Disease in Mexico |
Endothelial cells ( EC ) are the main target for Orientia tsutsugamushi infection and EC dysfunction is a hallmark of severe scrub typhus in patients . However , the molecular basis of EC dysfunction and its impact on infection outcome are poorly understood . We found that C57BL/6 mice that received a lethal dose of O . tsutsugamushi Karp strain had a significant increase in the expression of IL-33 and its receptor ST2L in the kidneys and liver , but a rapid reduction of IL-33 in the lungs . We also found exacerbated EC stress and activation in the kidneys of infected mice , as evidenced by elevated angiopoietin ( Ang ) 2/Ang1 ratio , increased endothelin 1 ( ET-1 ) and endothelial nitric oxide synthase ( eNOS ) expression . Such responses were significantly attenuated in the IL-33-/- mice . Importantly , IL-33-/- mice also had markedly attenuated disease due to reduced EC stress and cellular apoptosis . To confirm the biological role of IL-33 , we challenged wild-type ( WT ) mice with a sub-lethal dose of O . tsutsugamushi and gave mice recombinant IL-33 ( rIL-33 ) every 2 days for 10 days . Exogenous IL-33 significantly increased disease severity and lethality , which correlated with increased EC stress and activation , increased CXCL1 and CXCL2 chemokines , but decreased anti-apoptotic gene BCL-2 in the kidneys . To further examine the role of EC stress , we infected human umbilical vein endothelial cells ( HUVEC ) in vitro . We found an infection dose-dependent increase in the expression of IL-33 , ST2L soluble ST2 ( sST2 ) , and the Ang2/Ang1 ratio at 24 and 48 hours post-infection . This study indicates a pathogenic role of alarmin IL-33 in a murine model of scrub typhus and highlights infection-triggered EC damage and IL-33-mediated pathological changes during the course of Orientia infection .
Orientia tsutsugamushi is an obligately intracellular bacterium and the etiological agent of scrub typhus with a geographical distribution that encompasses much of the Asia-Pacific region [1] . Scrub typhus is a neglected but important tropical disease , which puts one-third of the world’s population at risk . The disease is transmitted by the bite of an infected larval Leptotrombidium mite or chigger . After 7–14 days of incubation , patients exhibit signs of infection such as an inoculation site eschar followed by fever and rash accompanied by non-specific flu-like symptoms . Although the endothelial tropism of Orientia can lead to disseminated endothelial infection that affects all organs; macrophages , dendritic cells and cardiac myocytes are also the targets of infection [2 , 3] . Primary characteristics of fatal scrub typhus pathology include diffuse interstitial pneumonia , hepatic lesions , glomerulonephritis , meningoencephalitis , and coagulation disorders [3–6] . Scrub typhus often presents as an acute febrile illness [1 , 7] . Without appropriate treatment , scrub typhus can cause severe multi-organ failure with a relatively high mortality rate [8] . Several antibiotics ( doxycycline , azithromycin , rifampicin , chloramphenicol , etc . ) have been used to treat Orientia infection . Although these antibiotics are effective if given early [9–12] , misdiagnosis , inappropriate antibiotic treatment , and antibiotic failures have occurred , emphasizing the need for a vaccine and alternative therapeutics [1] . Understanding the molecular mechanism of the infection will be beneficial for vaccine design and future therapeutic strategies . Infection-induced renal dysfunction , as well as acute kidney injury , has often been described in moderate-to-severe scrub typhus [13–18] . The severity of disease is often correlated with the extent of renal dysfunction [13 , 17 , 19 , 20]; however , the molecular mechanism that accounts for such renal dysfunction is poorly understood . The endothelium provides a crucial interface between tissues and circulating inflammatory cells . During tissue damage , endothelial cells ( ECs ) become activated , expressing adhesion molecules that alert circulating leukocytes to possible insults and further allow the leukocyte to transmigrate across the endothelial layer . In the case of scrub typhus , ECs are the primary target cells once the bacteria has disseminated [3 , 21] . During infection the ECs become activated , attracting inflammatory cells , resulting in the observed pathology . Endothelial activation and dysregulation can lead to tissue damage and organ dysfunction . Understanding the molecules that are released during infection is crucial to understanding the role of ECs in the host response during scrub typhus . Damage-associated molecular pattern molecules ( DAMPs ) are molecules that can initiate and perpetuate an immune response within the noninfectious and infectious inflammatory responses . Among them , IL-33 , a member of interleukin-1 family , locates in the nucleus as a chromatin-associated nuclear factor . IL-33 can modulate inflammatory responses when released [22 , 23] . In damaged tissues , necrotic cells can directly release endogenous IL-33 , which can signal through its receptor IL-33R/ST2L on target cells [24 , 25] . IL-33 has pro- or anti-inflammatory roles , depending on the disease models and tissues involved [26] . For example , recombinant IL-33 ( rIL-33 ) treatment can exacerbate cisplatin-induced acute kidney injury by increasing CD4+ T cell infiltration , CXCL1 production , and acute tubular necrosis [27] . IL-33 also mediates inflammatory responses in human lung tissue cells involved in the chronic allergic inflammation of the asthmatic airway [28] . However , IL-33 can be hepatoprotective in viral infection and ischemia/reperfusion-induced acute liver injury [29 , 30] . At present , the role of IL-33 in severe scrub typhus is unclear . In this study , we found that mice infected with O . tsutsugamushi Karp strain had significantly increased expression levels of IL-33 and its receptor in the kidneys and liver , but not in the lungs . IL-33 deficiency resulted in decreased kidney cellular infiltration and apoptotic cells , as well as delayed bodyweight loss . Compared to WT mice , the endothelium stress and activation in the kidneys of IL-33-/- mice were significantly attenuated , as evidenced by increased angiopoietin ( Ang ) 1 and endothelial nitric oxide synthase ( eNOS ) , and decreased endothelin-1 ( ET-1 ) . To further confirm the role of IL-33 in scrub typhus , we injected rIL-33 to sub-lethally infected mice , and observed the exacerbated illness and increased mortality . Moreover , rIL-33 treatment resulted in increased vascular dysregulation in the kidneys of infected mice . In vitro , Orientia infection significantly stimulated IL-33 and ST2 expression in human EC and increased EC activation . These data suggest that IL-33 plays a significant role in modulating host immune and endothelial responses during scrub typhus infection .
We have recently reported a strong type 1 immune response , but a repressed type 2 response , accompanied with severe tissue damage in multiple organs during lethal infection with O . tsutsugamushi Karp strain in B6 mice [31 , 32] . Since Orientia lacks the classical ligands for TLR2/4 stimulation , we speculated that the host DAMP molecule , IL-33 plays a role in modulating inflammation responses in this infection . B6 mice were challenged with a lethal dose of Orientia and serially sampled on 2 , 6 and 10 days post-infection ( dpi ) . IL-33 expression began to rise at 6 dpi , and the increase was significant at 10 dpi in the kidneys ( Fig 1A ) . Along with this increase , its receptor ST2L also significantly increased at 6 and 10 dpi in the kidneys ( Fig 1A ) . As IL-33 increased as early as 2 dpi in the livers and remained at similar levels throughout the course of infection ( Fig 1B ) , ST2L expression in the liver was also significantly increased at 6 and 10 dpi ( Fig 1B ) . Unlike the kidneys and liver , the lungs had significantly decreased IL-33 expression at 6 and 10 dpi and no statistically significant changes in ST2L expression ( Fig 1C ) . These findings suggest tissue-specific expression of IL-33/ST2L during the infection . To assess the role of IL-33 in scrub typhus progression , we challenged IL-33-/- and WT mice with a lethal dose of Orientia and monitored them daily for disease manifestations . IL-33-/- and WT mice both lost body weight from 4 to 8 dpi; however , IL-33-/- mice had significantly less weight loss than did WT mice ( Fig 2A ) and were considerably more active throughout the course of infection as compared to WT mice . Nevertheless , both groups of mice were moribund at 9 dpi , with similar bacterial loads in the kidneys , livers and lungs ( Fig 2B–2D ) . We also examined a panel of cytokines and chemokines at 9 dpi at the RNA and protein levels ( Fig 3 ) . IL-33 deficiency led to significantly higher gene expressions of IL-6 , IL-10 , and IFN-γ in the kidneys , but no major changes for the expression of Th2 cytokines ( IL-4 and IL-13 ) or chemokines ( CXCL9 and CXCL10 ) ( Fig 3A ) . Similar trends by these cytokines were confirmed in protein levels from the kidneys ( Fig 3B ) . The gene expression levels for IFN-γ , TNF-α , and IL-4 in lung and livers were comparable , while IL-13 expression was undetectable ( S1 and S2 Figs ) . IL-33 appears to play a role in pathogenesis of the kidneys [27] . To further gauge inflammatory responses and renal pathology in IL-33-/- mice , we examined and found fewer cellular infiltrates in the kidneys of IL-33-/- mice . Intertubular infiltration was evident , as well as cellular infiltrations in and around the glomeruli in WT mice , whereas IL-33-/- mice had very few instances of infiltration in the kidneys ( Fig 4A ) . A considerable number of dense/fragmented nuclei resembling apoptotic cells were observed in endothelial locations in WT mice ( Fig 4A , WT-box ) . To confirm that these fragmented nuclei were indeed apoptotic cells , TUNEL immunohistochemistry was performed allowing quantification and comparison of apoptotic cell numbers . The kidneys of WT animals had more intense positive staining compared to that in IL-33-/- mice ( Fig 4B ) , especially in the endothelium ( Fig 4C ) . While apoptotic ECs constituted approximately 50% of the total number of apoptotic cells in kidneys from both groups , the WT kidneys had 5-fold more apoptotic cells than did their IL-33-/- counterparts ( Fig 4D ) . To verify the TUNEL findings , we compared the expression of the anti-apoptotic gene BCL-2 and found a significantly higher level of BCL-2 transcripts in IL-33-/- kidneys than seen in WT controls ( Fig 4E ) . There were no major differences of pathology in lungs and livers between infected WT and IL-33-/- mice ( S1 and S2 Figs ) . Orientia infection can result in EC stress and activation in lung and liver tissue , as judged by the changes in the Ang2/Ang1 ratio [32] . In the kidneys , elevation in the Ang2/Ang1 ratios were evident as early as 2 dpi and peaked at 10 dpi ( Fig 5A and 5B ) . When EC activation between WT and IL-33-/- mice in the kidneys was compared at 9 dpi ( Fig 5C–5E ) , we found that IL-33-/- mice had significantly attenuated EC stress and activation compared to that in WT controls , as judged by higher levels of Ang1 ( an EC-stabilizing factor ) and eNOS ( a synthase of EC-relaxing factor nitric oxide ( NO ) [33] ) , as well as a lower Ang2/Ang1 ratio and lower levels of ET-1 ( an important factor in the development of vascular dysfunction by inhibiting eNOS and NO production [34] ) . The kidney eNOS/ET-1 ratio was significantly higher in the infected IL-33-/- mice ( Fig 5F ) , implying that deficiency in DAMP molecule IL-33 alleviated the endothelial dysfunction in the kidneys of Orientia-infected mice . In the liver and lungs , the Ang2/Ang1 ratios were comparable in infected WT and IL-33-/- mice , indicating similar levels of EC activation in these organs ( S1 and S2 Figs ) . The above data indicated that the absence of IL-33 during Orientia infection resulted in an attenuated weight loss and cellular apoptosis in the kidneys during lethal challenge , but it was not sufficient to increase mouse survival . To validate the function of IL-33 , we infected WT mice with a sub-lethal dose of Orientia and then i . p . delivered rIL-33 or PBS every other day for 10 days . As shown in Fig 6A , the rIL-33 group lost weight more rapidly starting on 8 dpi than did their PBS-injected counterparts . While the PBS-injected mice recovered part of their body weight after 10 dpi , rIL-33-injected mice exhibited severe signs of disease , with a 64 . 7% mortality rate ( Fig 6B ) . This increased mortality in the IL-33-injected mice seemed not due to an increase in bacterial loads in the livers or kidneys ( S3 Fig ) . To examine the underlying mechanisms , we examined endothelial markers in the kidneys . We found evidence for increased EC stress and endothelial dysfunction in the kidneys of Orientia-infected , IL-33-injected mice , including a significantly reduced Ang1 expression and a near 2-fold increase in Ang2/Ang1 ratio ( Fig 7A ) , which was accompanied with a significantly reduced eNOS/ET-1 ratio ( Fig 7B ) . Exogenous IL-33 also increased the liver inflammation and EC stress , as evidenced by increased liver Ang2/Ang1 ratios ( S4 Fig ) . The down-regulated BCL-2 expression , plus increased CXCL1 expression , in the kidneys of rIL-33-treated mice ( Fig 7C ) suggested an increased cellular apoptosis and increased IL-33-mediated pro-inflammatory reaction , as previously reported [27] . To further examine the role of EC stress , we infected human umbilical vein endothelial cells ( HUVEC ) in vitro at 3 and 10 multiples of infection ( MOI ) , respectively . At 24 hours post-infection ( hpi ) , we found an infectious dose-dependent increase in the expression of IL-33 , soluble ST2 ( sST2 ) , membrane-bound ST2L , and the Ang2/Ang1 ratio ( Fig 8A ) . At 48 hpi , IL-33 levels were similar to those in controls , but elevation in sST2 , ST2L , and Ang2/Ang1 ratio remained significant , especially for high-dose infection groups ( Fig 8B ) . We also examined the secretion of IL-33 proteins in culture supernatants of infected HUVECs ( MOI 3 and MOI 10 ) at 0 , 3 , 24 and 48 hpi by using an ELISA assay . Appreciable IL-33 was detected among high-dose infection groups , rather than in samples infected with 3 MOI and control samples ( S5 Fig ) . Our in vitro data were consistent with those from mouse studies in vivo , implying an important role for IL-33/ST2-mediated responses during Orientia infection .
Invasion of O . tsutsugamushi can cause acute tubular necrosis , leading to renal failure in patients [35 , 36]; however , the underlying mechanism is unclear . The DAMP molecule IL-33 is known to be a potent endothelial activator , promoting angiogenesis and vascular permeability [37] , and can also selectively target the non-quiescent ECs , driving pro-inflammatory cell activation [38] . IL-33 also contributes to the pathogenesis of cisplatin-induced acute kidney injury [27] . However , it is unclear whether IL-33 modulates tissue injury and progression of scrub typhus . Here , we demonstrate that there was a significant increase in IL-33 and its ST2L receptor expression in the kidneys and liver during O . tsutsugamushi infection in mice and that IL-33 contributed to renal pathology and endothelial damage during experimental scrub typhus infection . The absence of IL-33 signaling during lethal infection attenuated cellular and tissue damage and delayed the onset of disease ( i . e . weight loss ) , although such changes were not sufficient to rescue the mice from death . Conversely , addition of exogenous IL-33 during a sub-lethal infection exacerbated the activation of renal endothelium and lethality . Moreover , Orientia infection alone was capable of inducing gene expression of IL-33 and its receptors , as well as endothelial activation , in human endothelial cells . We have proposed a pathogenic role of IL-33 in endothelial dysregulation during the infection ( Fig 9 ) . This is the first study to address the role of IL-33 in a mouse model of scrub typhus . IL-33 has crucial and diverse roles in infectious diseases , depending on the type of infectious agents , acute or chronic infection stages , tissues involved , and host immune microenvironments [39] . The protective roles of the IL-33/ST2 axis have been reported during chronic viral infection in the liver , via promoting CD8+ T-cell responses [40] , repressing inflammatory cytokine TNF-α , inducing type 2 innate lymphoid cells ( ILC2 ) , and protecting the liver in acute adenovirus infection [30] . IL-33-induced ILC2 also promotes lung tissue homeostasis in influenza virus infection [41] . However , IL-33 also plays deleterious roles during Cryptococcus neoformans-induced lung mycosis and allergic inflammation in the lungs [42 , 43] . Research concerning the role of IL-33 in kidney infection is relatively limited , and a few reports are focused on cisplatin or Candida albicans-induced renal injury [13 , 44] . Our data demonstrate that infection with O . tsutsugamushi Karp strain can increase gene expression of IL-33 and ST2L in the kidneys and liver ( Fig 1 ) . The cellular sources of IL-33 in Orientia-infected tissues was not examined in this study , due to technical issues with respect to cell isolation in the ABSL3 facility; but the possible candidates may include ECs in the kidneys [22] and hepatocytes in the liver [45] . The marked reduction of IL-33 expression in the lung tissues at 6 and 10 dpi may not be surprising , given the massive cellular necrosis and tissue damage [32] . Yet , the tendency of reduced IL-33 expression at 2 dpi in the lung was interesting . Regardless of the underlying mechanisms , our data suggest tissue-specific roles of endogenous IL-33 and highlight its contributions in renal injury , cellular apoptosis , and endothelial activation in mouse model of severe scrub typhus . To further investigate how IL-33 regulates immune responses and why IL-33-/- mice have attenuated weight loss and kidney injury ( Figs 2–4 ) , we examined a panel of immune cytokines in the kidney after infection . We have demonstrated previously that Orientia infection induces strong type 1 , but impaired , type 2 immune responses , in several tissues [32] . In the present study , we found that IL-33 deficiency or exogenous rIL-33 did not drastically change type 1 or type 2 cytokine expression during Orientia infection , as reported in a C . albicans-induced renal injury model [44] . While IL-33 did not reprogram type 1 vs . type 2 responses during Orientia infection , the IL-33/ST2 signaling significantly amplified the magnitude of pro-inflammatory responses ( Fig 3 ) , cellular apoptosis , EC stress and activation , and host death ( Figs 4–7 ) . On one hand , we observed the upregulated CXCL1 , but decreased anti-apoptotic gene BCL-2 , in the kidneys by rIL-33 treatment . This finding is similar to that in the cisplatin-induced renal failure model [27] , suggesting the unique mechanism that IL-33/CXCL1 axis may play a critical role in renal injury not only in the toxic reagent-induced model but also in infectious diseases . On the other hand , we found that IL-33-/- mice have a higher expression of anti-inflammatory cytokine IL-10 at both the gene and protein levels in Orientia infection . This increased IL-10 may play a role in renal protection in infected IL-33-/- mice [46] . Some acute inflammatory mediators ( e . g . IL-6 , IL-12 and IFN-γ ) may contribute to bacterial control [47 , 48] . Orientia infection in vitro can activate ECs , leading to cell apoptosis [49 , 50] . ECs are known to be the source of nucleus IL-33 [22]; however , few studies have focused on the interaction and DAMP molecule expression in ECs infected with Orientia . We have provided evidence that Orientia infection in vitro increased IL-33 and ST2L expression in and the activation of human ECs by 24 h . Prolonged stimulation ( 48 h ) , however , did not alter the IL-33 gene expression levels , but dramatically increased both the soluble and membrane-bound ST2 forms of receptors ( Fig 8 ) ; this may partially explain our difficulty in detecting IL-33 proteins in the culture supernatants . Since sST2 can bind IL-33 and block intracellular IL-33/ST2L signaling [51] , this increased sST2 level may counterbalance the excessive IL-33 signal and keep the homeostasis . In addition to being the source of IL-33 , ECs are the target of IL-33 [38] . It was previously shown that angiogenesis in ECs was induced by stimulating endothelial NO production via the ST2/TRAF6-Akt-eNOS signaling pathway [37] . It will be interesting to further examine the intracellular signaling events that regulate IL-33/ST2 expression during O . tsutsugamushi infection . Based on our in vitro studies in human ECs and in vivo studies in WT and IL-33-/- mice , we have proposed a pathogenic role of IL-33 in endothelial dysregulation during the infection ( Fig 8 ) . High-dose O . tsutsugamushi infection in ECs and other cell types can trigger EC stress , dysfunction , and apoptosis . In the WT mice , the increased IL-33 and ST2 expression on EC and IL-33 production may further exacerbate EC stress and damage . These IL-33/ST2-mediated effects are diminished or markedly reduced in O . tsutsugamushi-infected IL-33-/- mice , leading to attenuated renal endothelium activation but higher levels of Ang1 in the kidneys . We have provided evidence that endogenous IL-33 promotes EC inflammation during Orientia infection , via multiple mechanisms , which includes reduced Ang1 and eNOS expression , but increased Ang2 and ET-1 expression in the infected kidneys , as in reports for other models [37 , 52] . The interplays among Ang1 , eNOS and ET-1 in Orientia-infected ECs warrant further investigation [34] . As expected , our results reveal that IL-33 regulates the balance of Ang1/Ang2 as well as that of eNOS/ET-1 , modulating the EC inflammation and tissue dysregulation in the kidneys during severe scrub typhus . Our findings are important in the context of a recent report , showing that IL-33 concentrations in human serum strongly correlated with the severity of Hantaan infection , another endotheliotropic pathogen [53] . Therefore , while IL-33 plays a protective role in other models such as viral hepatitis , it has a pathogenic role in endotheliotropic diseases . Overall , this study indicates a significant role of IL-33 alarmin in endothelial activation and renal damage , highlighting infection-triggered EC damage and IL-33-mediated pathological changes during the course of O . tsutsugamushi infection . This study provides a better understanding of the pathogenesis and a potential biomarker for monitoring disease progression of scrub typhus cases .
Female WT B6 mice were purchased from Jackson Laboratory . IL-33-/- mice on the B6 background were kindly provided by Dr . Rene de Waal Malefyt ( Merck , Palo Alto , CA ) . Mice were maintained under specific pathogen-free conditions and used at 8- to 12 weeks of age following protocols approved by the Institutional Animal Care and Use Committee ( protocol # 1302003 ) at the University of Texas Medical Branch ( UTMB ) in Galveston , TX . All mouse infection studies were performed in the ABSL3 facility in the Galveston National Laboratory located at UTMB; all tissue processing and analysis procedures were performed in the BSL2 or BSL3 facilities . All procedures were approved by the Institutional Biosafety Committee , in accordance with Guidelines for Biosafety in Microbiological and Biomedical Laboratories . UTMB operates to comply with the USDA Animal Welfare Act ( Public Law 89–544 ) , the Health Research Extension Act of 1985 ( Public Law 99–158 ) , the Public Health Service Policy on Humane Care and Use of Laboratory Animals , and the NAS Guide for the Care and Use of Laboratory Animals ( ISBN-13 ) . UTMB is a registered Research Facility under the Animal Welfare Act and has a current assurance on file with the Office of Laboratory Animal Welfare , in compliance with NIH Policy . O . tsutsugamushi Karp strain was used herein , and all infection studies were performed with the same bacterial stock prepared from liver extracts pooled from several infected mice . Infectious organisms were then quantified via a focus forming assay as described previously [31 , 32] . WT and IL-33-/- mice were inoculated intravenously ( i . v . ) with a lethal dose of O . tsutsugamushi ( 4 . 5 x 106 FFU in 200 μl ) . Control mice were similarly injected with PBS . At 9 dpi , serum and tissue samples were collected and inactivated for subsequent analyses . To study the effect of excess IL-33 mice received a mostly sub-lethal injection of 8 . 5 x 105 organisms . After infection mice were injected intraperitoneally with either PBS or 1 μg of rIL-33 at 2 , 4 , 6 , 8 , and 10 dpi , respectively . Animals were monitored for signs of disease progression daily until the end of the experiment ( 13 dpi ) . HUVECs were cultured as described previously [54] . Briefly , HUVECs ( Cell Application , San Diego , CA ) were cultivated in Prigrow I medium supplemented with 10% ( vol/vol ) heat-inactivated FBS in 5% ( vol/vol ) CO2 at 37°C . All experiments were performed between passages 5 and 7 , and cells were maintained in Prigrow I medium with 3% ( vol/vol ) FBS . When HUVECs were confluent , they were collected and seeded onto 24-well plates ( Corning Inc . , Corning , NY ) . Once all wells were confluent , the HUVEC monolayers were infected with either 3 MOI , 10 MOI , or media only . Total RNA was extracted from each plate at 3 , 24 , and 48 hours post-infection ( hpi ) by using an RNeasy mini kit ( Qiagen , Valencia , CA ) and digested with RNase-free DNase ( Qiagen ) . Gene expression was determined as described below . Cell-free culture supernatants were collected and stored in -80°C until protein analysis . IL-33 concentrations in supernatants of control and infected HUVECs were determined by using human IL-33 Quantikine ELISA kits ( R&D Systems , Minneapolis , MN ) following the manufacturer’s protocol . Briefly , 100 μl of supernatant was added to each well of the anti-hIL-33-coated , 96-well ELISA plate . The plate was analyzed using a Versamax Turntable Microplate Reader ( Molecular Devices , Sunnyvale , CA ) and Softmax Pro V . 4 . 0 . All procedures were performed in the BSL3 facility . Mouse tissues were collected in an RNALater solution ( Ambion , Austin , TX ) at 4°C overnight to inactivate infectious bacteria and stored at -80°C for subsequent analyses . Total RNA was extracted from tissue by using an RNeasy mini kit ( Qiagen , Valencia , CA ) and digested with RNase-free DNase ( Qiagen ) . cDNA was synthesized with the iScript cDNA synthesis kit ( Bio-Rad Laboratories , Hercules , CA ) . The abundance of target genes was measured by qRT-PCR by using a Bio-Rad CFX96 real-time PCR apparatus , and a SYBR Green Master mix ( Bio-Rad ) was used for all PCR reactions . PCR reactions were started at 95°C for 3 min , followed by 39 cycles of 95°C for 10 sec , and 60°C for 10 sec , and ended with an elongation step at 72°C for 10 sec . Dissociation melting curves were obtained after each reaction to confirm the purity of PCR products . Relative abundance of mRNA expression was calculated by using the 2-ΔΔCT method . Glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) and β-actin were used as the housekeeping genes . Primer sequences are listed in S1 Table . Bacterial loads were assessed by quantitative real-time PCR as described previously [31 , 32] . DNA was extracted by using a DNeasy Kit ( Qiagen , Gaithersburg , MD ) from the tissue samples , and the bacterial load at each time point and for each organ sampled was determined by quantitative real-time PCR . The gene for a 47-kDa protein was amplified by using specific primers ( OtsuF630 and OtsuR747 ( IDT , Coralville , IA ) . PCR products were detected with a specific probe OtsuPr665 ( Applied Biosystems , Foster City , CA ) . Bacterial loads were normalized to total nanogram ( ng ) of DNA per μL for the same sample , and data are expressed as the gene copy number of 47-kDa protein per picogram ( pg ) of DNA . 47-kDa gene copy number was determined by known concentrations of control plasmid containing single-copy inserts of the gene . The plasmid concentration was determined and serially diluted 10-fold for the standards . All tissues were fixed in 10% neutral-buffered formalin and embedded in paraffin , and sections ( 5-μm thickness ) were stained with hematoxylin and eosin . Apoptosis was detected by using a Millipore ApopTag Peroxidase In Situ Apoptosis Detection kit . Kidneys were assessed for positive staining; five , 40x images were taken on an Olympus BX53 microscope . Images were used so that multiple observers could assess the same fields for apoptosis . DAB-positive cells were counted as apoptotic and divided into endothelial cells , based on cellular and nuclear morphology , and other cells . Cells that were rounded or otherwise not recognizable as ECs were counted as other cells . The number of apoptotic cells was counted per 40x field-of-view . The observers’ counts were pooled and averaged and then numbers compared as total number of apoptotic cells per view and apoptotic ECs only . Cytokine profiles in the tissues were characterized by using Procarta Plex Mouse Cytokine Panel ( eBioscience , San Diego , CA ) . Briefly , kidney protein was extracted by using RIPA ( Cell Signaling Technology , Danvers , MA ) plus Protease Inhibitor Cocktails ( Sigma , St . Louis , MO ) . The concentration of protein was determined by a Pierce BCA Protein Assay kit ( Thermo Scientific , Waltham , MA ) . Colored magnetic beads coated with different antigens were mixed together with kidney protein samples , and then allowed to incubate for overnight at 2–8°C . After three wash cycles , detection antibody was added and allowed to incubate for 1 h at room RT , followed by incubation with Streptavidin-Phycoerythrin for 30 min at RT . After removal of excess conjugate , 150 μl of sheath fluid was added to each well . The beads were read on a Bio-Rad Bio-Plex 200 System . Raw data were measured as the relative fluorescence intensity and then converted to the concentration according to the standard curve . Data were presented as mean ± standard errors of the mean ( SEM ) . Differences between individual treatment and control groups were determined by using Student’s t test . One-way ANOVA was used for multiple group comparisons . Statistically significant values are referred to as * , p < 0 . 05; ** , p < 0 . 01 , *** , p < 0 . 001 , **** , p < 0 . 0001; NS , no significance . | Scrub typhus is a life-threatening disease , caused by infection with O . tsutsugamushi , a Gram-negative bacterium that preferentially infects and replicates in the endothelium . Every year , approximately one million people are infected globally , especially in the Asia-Pacific region . However , the molecular mechanism ( s ) of tissue pathogenesis and immune responses in scrub typhus remain poorly understood . IL-33 is a damage-associated molecular pattern factor , which can modulate host inflammatory responses in several infectious diseases . In this study , we compared the severity of disease between wild-type ( WT ) and IL-33-/- mice infected with O . tsutsugamushi and used exogenous IL-33 to further examine the function of IL-33 during the infection . Our studies in mouse models , as well as in vitro studies in human endothelial cells , have revealed a pathogenic role of IL-33 in promoting endothelial cell stress , cellular apoptosis , tissue damage , and host death . This study will help us understand the pathogenesis of severe scrub typhus . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"death",
"typhus",
"medicine",
"and",
"health",
"sciences",
"innate",
"immune",
"system",
"immune",
"physiology",
"cytokines",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"cell",
"processes",
"rna",
"extraction",
"microbiology",
"immunology",
"a... | 2016 | IL-33-Dependent Endothelial Activation Contributes to Apoptosis and Renal Injury in Orientia tsutsugamushi-Infected Mice |
Trypanosoma cruzi is a protist parasite that causes Chagas disease . Several proteins that are essential for parasite virulence and involved in host immune responses are anchored to the membrane through glycosylphosphatidylinositol ( GPI ) molecules . In addition , T . cruzi GPI anchors have immunostimulatory activities , including the ability to stimulate the synthesis of cytokines by innate immune cells . Therefore , T . cruzi genes related to GPI anchor biosynthesis constitute potential new targets for the development of better therapies against Chagas disease . In silico analysis of the T . cruzi genome resulted in the identification of 18 genes encoding proteins of the GPI biosynthetic pathway as well as the inositolphosphorylceramide ( IPC ) synthase gene . Expression of GFP fusions of some of these proteins in T . cruzi epimastigotes showed that they localize in the endoplasmic reticulum ( ER ) . Expression analyses of two genes indicated that they are constitutively expressed in all stages of the parasite life cycle . T . cruzi genes TcDPM1 , TcGPI10 and TcGPI12 complement conditional yeast mutants in GPI biosynthesis . Attempts to generate T . cruzi knockouts for three genes were unsuccessful , suggesting that GPI may be an essential component of the parasite . Regarding TcGPI8 , which encodes the catalytic subunit of the transamidase complex , although we were able to generate single allele knockout mutants , attempts to disrupt both alleles failed , resulting instead in parasites that have undergone genomic recombination and maintained at least one active copy of the gene . Analyses of T . cruzi sequences encoding components of the GPI biosynthetic pathway indicated that they are essential genes involved in key aspects of host-parasite interactions . Complementation assays of yeast mutants with these T . cruzi genes resulted in yeast cell lines that can now be employed in high throughput screenings of drugs against this parasite .
Glycosylphosphatidylinositol ( GPI ) is an abundant component of the plasma membrane of protist parasites . In most eukaryotic cells , GPIs are found as free molecules or as lipid anchor for proteins that are bound to the cell surface [1] . They are complex molecules that are synthesized in the ER by sequential addition of sugar residues and other substituents , e . g . ethanolamine-phosphate , to the phosphatidylinositol ( PI ) precursor and transported to the cell surface , as a free GPI also known as GIPL ( glycoinositolphospholipid ) or linked to the C-terminus of a protein that contains a GPI signal sequence [2] . Numerous studies with different parasites clearly show that GIPLs and GPI-anchored proteins play important roles in different processes related to host-parasite interaction . Also , it has been suggested that , because of the existence of differences in the structure of GPI from several parasite species as well as between GPIs of the parasite and their host cells [2] , [3] , [4] , these molecules constitute promising targets for studies towards the development of new anti-microbial drugs [5] . Trypanosoma cruzi is a parasitic protist that causes Chagas disease , an illness not only prevalent in Latin America , where an estimated 8 million people are infected , but a worldwide health issue for which there is an urgent need for the development of new chemotherapeutic agents and more effective prophylactic methods ( www . who . int/topics/chagas_disease/en/ ) . The surface of T . cruzi is covered by a large amount of GPI-anchored proteins whose structure and chemical composition have been extensively studied [6] and are expressed in all developmental stages of the parasite life cycle [3] , [7] . Analysis of the T . cruzi genome indicated that 12% of the parasite genes encode proteins anchored by GPI , a percentage that is much higher when compared with other organisms [8] . Many of these proteins play important roles in the invasion process and , since they show varying sequences , they could also participate in the processes responsible for evasion of the host immune response [9] , [10] . Two main components of the T . cruzi surface , the trans-sialidases and mucins , which act , respectively , as enzymes responsible for the transfer and acceptors for sialic acid molecules , are GPI-anchored glycoproteins [11] . It has also been demonstrated that T . cruzi GPI-anchored mucins as well as free GPI anchors act as potent pro-inflammatory agents that are recognized by Toll like receptors [12] and , because of their role in activating the innate immune response , they have been used as adjuvants in immunization protocols [13] . In Saccharomyces cerevisiae , biosynthesis of GPI is essential for cell growth and occurs in eleven steps beginning with the transfer of a molecule of N-acetyl-glucosamine ( GlcNAc ) from UDP-GlcNAc to PI [14] , [15] . After the addition of mannose molecules using dolichol-P-mannose as a donor , followed by the transfer of ethanolamine-phosphate ( EtNP ) to the third mannose residue , GPI is transferred to proteins that have a predicted GPI addition signal at their C-terminal end , in a reaction catalyzed by the GPI-transamidase complex [16] . Genes encoding enzymes involved in GPI pathway from various organisms , including protist parasites such as Trypanosoma brucei , Leishmania mexicana and Plasmodium falciparum have been cloned and their products characterized by functional complementation in mammalian cells and in yeast mutants [17] , [18] , [19] , [20] . Although the main structure of GPI is conserved in all organisms , several studies have shown differences in the biosynthetic pathway and additional modifications to GPI structures present in mammalian and parasite cells [2] , [3] , [4] . Substrate analogues of enzymes of the GPI biosynthetic pathway showing trypanocidal activity have been described [21] . Since enzymes involved in the basic steps common to the biosynthesis of GPI in the different organisms have different sensitivities to various inhibitors [22] , [23] , [24] , [25] , [26] , [27] , we sought to characterize the genes involved in biosynthesis of GPI anchors in T . cruzi . Orthologous sequences of all genes involved in biosynthesis of T . cruzi GPI anchors were identified and , for three of them , we were able to show that they complement yeast conditional mutants of genes of this pathway . Unsuccessful attempts to generate T . cruzi knockouts for three of these genes suggest that GPI is an essential component of the parasite . Since specific inhibition of GPI biosynthesis may affect the expression of a large number of T . cruzi proteins that are essential for host-parasite interactions , targeting this pathway can be considered a promising strategy for the development of new chemotherapy against Chagas disease . The availability of yeast mutants expressing T . cruzi enzymes constitutes the first step in that direction .
Epimastigotes of the CL Brener clone of T . cruzi were maintained in logarithmic growth phase at 28°C in liver infusion tryptose ( LIT ) medium supplemented with 10% fetal bovine serum as described by Camargo [28] . Metacyclic trypomastigotes were obtained after metacyclogenesis in LIT medium , observed after 15–20 days of culture [28] and were used to infect Vero cells . Intracellular amastigotes and tissue culture derived trypomastigotes were obtained from Vero cells grown in Dulbecco's Modified Eagle Medium ( DMEM ) supplemented with 5% fetal bovine serum , at 37°C and 5% CO2 as previously described [29] . Sequence analyses were conducted using the T . cruzi genome database ( www . tritrypdb . org ) to identify all orthologous genes involved in the parasite GPI biosynthesis . Sequences from different organisms , such as T . brucei , P . falciparum and S . cerevisiae [16] , [20] , were used as queries in Blastp analyses ( www . ncbi . nlm . nih . gov/blast/Blast . cgi ) and ClustalW ( www . clustal . org/ ) for multiple alignments between the predicted T . cruzi protein sequences and homologous sequences present in other organisms . Total DNA was purified from 109 T . cruzi epimastigotes that were harvested from exponentially growing cultures , according to previously described protocols [29] . Total RNA was isolated from epimastigotes , tissue culture derived trypomastigotes and amastigotes using the RNeasy kit ( Qiagen ) . For northern blot analyses , 10 µg of total RNA/lane was separated in 1 . 2% agarose/MOPS/formaldehyde gel . The RNA was transferred to Hybond-N membrane ( GE-Healthcare ) and hybridized with GPI8 , GPI10 and 24Sα rRNA probes previously labeled with [α-32P]-dCTP using the Amersham Ready-to-Go DNA Labeling Beads ( GE-Healthcare ) , according to the suppliers protocol . The hybridization was carried out as previously described [30] in 50% formamide buffer at 42°C . After washing twice with 2X SSC/0 . 2% SDS at 60°C for 20 min , the membranes were exposed to a phosphor screen of the STORM 820 phosphor image ( GE-Healthcare ) . Reverse-transcription amplifications ( RT-PCR ) were carried out with total RNA isolated from transfected yeast mutants and T . cruzi epimastigotes according to published protocols [30] . After first strand cDNA synthesis using oligo ( dT ) 18 or gene-specific primers ( see primer sequences in supplementary material , Table S1 ) and the SuperScript II Reverse Transcriptase ( Life Technologies ) , the cDNAs were amplified using Taq Polymerase ( Promega ) and primers specific for each gene and analyzed in 1% agarose gels stained with ethidium bromide . The S . cerevisiae strain used in this work were: YPH499 ( Mat a , ura3-52 , lys2-801amber , ade2-101ochre , trp1-63 , his3-200 , leu2-1 ) ( Stratagene ) , used as a control , and conditional lethal yeast mutants for GPI biosynthesis ( YPH499-HIS-GAL-DPM1 , YPH499-HIS-GAL-GPI3 , YPH499-HIS-GAL-GPI8 , YPH499-HIS-GAL-GPI10 , YPH499-HIS-GAL-GPI12 , YPH499-HIS-GAL-GPI14 , YPH499-HIS-GAL-GAA1 , and YPH499-HIS-GAL-AUR1 ) , which were generated by replacement of the endogenous yeast promoter by a galactose regulated promoter , as described [31] . S . cerevisiae strains were grown in YPGR medium ( 1% w/v yeast extract , 2% w/v bacto-peptone , 2% w/v galactose , 1% w/v raffinose ) , or in SD medium ( 0 . 17% yeast nitrogen base , 0 . 5% ammonium sulfate , 2% glucose , containing the nutritional supplements necessary to complement the auxotrophic samples or to allow selection of transformants ) . Before complementation , yeast clones were cultivated in SGR medium ( 4% galactose , 2% raffinose , 0 . 17% yeast nitrogen base , 0 . 5% ammonium sulfate ) in which glucose is replaced by galactose/raffinose as a carbon source . Sequences encompassing the full-length coding regions of TcDPM1 , TcGPI3 , TcGPI8 , TcGPI10 , TcGPI12 , TcGPI14 , TcGAA-1 , and TcIPCS were PCR amplified from total DNA of T . cruzi epimastigotes prepared as described above , using primers specific for each gene ( Table S1 ) . The amplicons were inserted into the S . cerevisiae expression vector pRS426Met [32] . Full-length coding sequences corresponding to orthologous S . cerevisiae genes were also PCR amplified with specific primers ( Table S1 ) and cloned into the same vector . Transformation of yeast mutants were carried out using the standard lithium acetate procedure [33] . Conditional lethal mutants were transformed with pRS426Met plasmids carrying either the S . cerevisiae ( Sc ) or the T . cruzi ( Tc ) genes and transformed cells were plated on minimal medium lacking histidine and uracil containing either galactose ( SGR ) or glucose ( SD ) and incubated at 30°C . Control YPH499 cells , mutant yeasts ( YPH499-HIS-GAL ) and mutant yeasts carrying pRS426Met containing yeast or T . cruzi genes were grown in SGR to saturation and used to inoculate SD ( 2% glucose ) , in which they were grown for about 16 h . Cells ( 1×108 ) were washed twice in SD without inositol medium ( 2% glucose ) , resuspended in 1 ml of SD without inositol ( 2% glucose ) and depleted of inositol for 20 min before the addition of 30 µCi of [2-3H]myo-inositol ( American Radiolabeled Chemicals , St . Louis , USA ) . Cells were labeled for 1 hour . Protein extraction was done according to Damasceno et al . [34] with the following modifications: radiolabeled cells were harvested , washed twice in phosphate-buffered saline ( PBS 1X ) at pH 7 . 4 , and resuspended in 100 µl of Yeast Breaking Buffer [50 mM sodium phosphate , pH 7 . 4; 1 mM phenylmethylsulfonyl fluoride ( PMSF ) ; 1X protease inhibitor cocktail ( Amresco , Solon , USA ) ; 1 mM EDTA , and 5% ( v/v ) glycerol] . Yeast cells were lysed by the addition of acid-washed glass beads ( 425–600 µm ) vortexing for 1 min with 1 min intervals on ice , repeated twenty times . The lysate was centrifuged at 2 , 000×g for 5 min at 4°C and the supernatant was collected . The remaining pellet containing cell debris and glass beads was resuspended in 75 µl of Yeast Breaking Buffer containing 2% ( w/v ) sodium dodecyl sulfate ( SDS ) by vortexing for 1 min with 1 min intervals on ice , repeated five times . After removing cellular debris by centrifugation , the lysates were combined and the proteins were then separated by 10% SDS-polyacrylamide gel electrophoresis . Protein bands containing labeled inositol were detected by fluorography . Wild type and yeast mutant cell lysates were prepared as previously described [35] . Briefly , exponential-phase yeast cultures corresponding to 1 . 5×107 cells/ml of cells grown in glucose-containing medium ( nonpermissive ) or in galactose-containing medium ( permissive medium ) were lysed after incubation in 1 . 0 ml of 1 M sorbitol/1 mM EDTA containing Zymolyase at 37°C and glass beads for 30 min , harvested by centrifugation ( 1800×g , 10 min , 4°C ) and resuspended in 200 µl of TM buffer ( 50 mM Tris/HCl , pH 7 . 5 , containing 5 mM MgCl2 and 0 . 2% 2-mercaptoethanol ) . Ninety µl for lysates ( corresponding to 3×108 cells for each assay ) were assayed directly for Dol-P-Man synthase activity as described [36] . Briefly , incubation mixtures contained 5 µl of GDP-[3H]Man ( 1 µCi/ml ) , 1 µl of Dol-P ( 5 mg/ml dispersed in 1 . 0% Triton X-100 by sonication ) and water to give a final volume of 10 µl . Amphomycin and tunicamycin ( final concentrations 1 mg/ml ) were added to some samples . After the addition of 90 µl of cell lysates and incubation at 30°C for 30 min , the reactions were terminated by the addition of 1 . 5 ml of ice-cold chloroform/methanol ( 2∶1 , v/v ) . The reactions were centrifuged ( 1500×g , 5 min , 4°C ) and the pellet extracted twice with 500 µl of chloroform/methanol . Equivalent amounts of radiolabeled , chloroform/methanol extractable reaction products were analyzed by TLC on Silica 60 plates ( Merck ) with chloroform/methanol/acetic acid/water ( 25∶15∶4∶2 , by vol . ) as solvent and Dol-P-Man as a reference . Plates were screened for radioactivity with a Berthold LB 2842 Automatic TLC-Linear Analyzer . Full-length TcDPM1 , TcGPI3 , and TcGPI12 coding sequences were PCR amplified from genomic DNA purified from cultures of the T . cruzi epimastigotes , using forward and reverse primers carrying XbaI and EcoRI restriction sites , respectively ( Table S1 ) . The amplicons were inserted into the XbaI-EcoRI sites of the T . cruzi expression vector pTREXnGFP [37] , generating pTREX-TcDPM1-GFP , pTREX-TcGPI3-GFP , and pTREX-TcGPI12-GFP that contain TcDPM1 , TcGPI3 and TcGPI12 genes fused to the N-terminus of the green fluorescent protein ( GFP ) . A total of 100 µg of each plasmid construction was used to transfect T . cruzi epimastigotes as previously described [37] . Twenty four hours post-transfection , parasites were fixed with 4% paraformaldehyde for 30 min at 4°C , permeabilized with 0 . 1% Triton X-100 for 5 min at room temperature and blocked with 5% fetal bovine serum in PBS ( blocking solution ) for 20 min at 4°C . Staining of the parasite ER was done with rabbit anti-T . brucei BiP antibody ( [38]; kindly provided by Renato Mortara , Universidade Federal de São Paulo ) , at a 1∶1000 dilution , and secondary goat anti-rabbit IgG antibody conjugated to Alexa Fluor 555 ( 1∶1000 dilution ) ( Molecular Probes/Life Technologies ) . After nuclei staining with 1 µg/ml of 4′ , 6-diamidino-2-phenylindole ( DAPI , Molecular Probes/Life Technologies ) , cover slides were mounted with 90% glycerol , 10% 1 M Tris HCl pH 9 . 0 , and 2 . 3% DABCO ( Sigma ) . Images were obtained with a fluorescence microscope ( Nikon Eclipse Ti ) or with the 5 LIVE confocal microscope ( Zeiss ) , both at the Center of Electron Microscopy ( CEMEL ) , at the Instituto de Ciências Biológicas , UFMG . Transfections of HT1080 human fibrosarcoma cells were done with 1 µg of pcDNA3 . 1/NT-GFP-TOPO ( Life Technologies ) containing the different T . cruzi genes inserted in fusion with GFP ( for primer sequences , see Table S1 ) and the FuGENE transfection reagent ( Roche ) , following the manufacturer's instructions . All plasmids were co-transfected with pGAG-DsRed-ER , a mammalian expression vector that encodes the Discosoma sp . red fluorescent protein ( DsRed ) in fusion with ER targeting sequences and the ER retention sequence , KDEL ( Clontech ) . DNA constructs designed to delete both TcGPI8 alleles in the T . cruzi CL Brener genome by homologous recombination were prepared after PCR amplification of the 5′ and 3′ regions of the TcGPI8 gene ( for primer sequences , see Table S1 ) . The generated PCR products ( with 487 bp and 647 bp , respectively ) were cloned sequentially into the SacI/SpeI and XhoI/XbaI sites of pCR2 . 1 TOPO vector ( Invitrogen ) , flanking the neomycin phosphotransferase ( NeoR ) or hygromycin phosphotransferase ( HygR ) resistance markers that were cloned into this vector . To improve mRNA expression in the parasite , the 3′ UTR plus downstream intergenic sequences of the T . cruzi gliceraldehyde-3-phosphate dehydrogenase ( gapdh ) gene was inserted downstream from the HygR marker . Similar constructs using 5′ and 3′ flanking sequences derived from TcGPI3 and TcGPI10 genes were generated . Epimastigote transfections were performed by electroporation with 50 µg DNA as described previously [37] . Twenty-four hours after transfection , 200 µg/ml of hygromycin B or G418 was added to the cultures and selected populations were obtained approximately 30 days after transfection . Cloned cell lines were obtained by plating on semisolid blood agar plates , after another 30 days of incubation at 28°C . Epimastigotes were fixed in 5% glutaraldehyde in 0 . 1 M cacodylate buffer pH 7 . 2 and processed following standard protocols , including post-fixation in osmium tetroxide followed by block counterstained with uranyl acetate and embedding in Epon resin . Ultrathin sections were counterstaining with lead citrate and analyzed in the Transmission Electron Microscope Tecnai G2-12 - SpiritBiotwin FEI - 120 kV located at the Center of Microscopy at the Universidade Federal de Minas Gerais , Belo Horizonte , Brazil . Approximately 109 epimastigotes were lysed in 20 mM Hepes , 10 mM KCl , 1 . 5 mM MgCl2 , 250 mM sucrose , 1 mM DTT , 0 . 1 mM PMSF , with five cycles of freezing in liquid nitrogen and thawing at 37°C . Total cell lysate was centrifuged at a low speed ( 2 , 000×g ) for 10 min and the supernatant was subjected to ultracentrifugation ( 100 , 000×g ) for one hour . The resulting supernatant was analyzed as soluble , cytoplasmic fraction ( C ) whereas the pellet , corresponding to the membrane fraction ( M ) was resuspended in lysis buffer . Volumes corresponding to 20 µg of proteins from total parasite cell lysate ( T ) , cytoplasmic ( C ) and membrane ( M ) fractions were loaded onto a 12 . 5% SDS-PAGE gel , transferred to nitrocellulose membranes , blocked with 5 . 0% non-fat dry milk and incubated with the anti-mucin antibody 2B10 ( gently provided by Nobuko Yoshida , Universidade Federal de São Paulo ) , at 1∶200 dilution followed by incubation with peroxidase conjugated anti-mouse IgG and the ECL Plus reagent ( GE-Healthcare ) . For flow cytometric analysis , epimastigotes were stained with anti-mucin 2B10 ( dilution 1∶450 ) and Alexa Fluor 488 conjugated secondary antibodies . Data were acquired on a FACScan flow cytometer ( Becton Dickinson ) .
Eighteen T . cruzi genes involved in 8 steps of the GPI biosynthetic pathway were identified based on their similarities to the yeast , mammals , Trypanosoma brucei and Plasmodium falciparum sequences [15] , [16] , [17] , [20] , ( Table 1 ) . For the majority of these genes , annotated as putative T . cruzi orthologs in the TriTrypDB ( www . tritrypdb . org ) , both alleles , belonging to the two CL Brener haplotypes , were identified . Since CL Brener is a hybrid strain , as described by El-Sayed et al . [39] , the two haplotypes corresponding to the two ancestral genomes that originated the CL Brener genome , named Esmeraldo-like and non-Esmeraldo-like , were separated during the T . cruzi genome assembly . In Table 1 , the genes corresponding to the non-Esmeraldo haplotype were indicated by their identification numbers in the TriTrypDB database . For all listed genes , the amino acid identities between the two alleles were greater than 94% . Based on these sequences and the known structure of the GPI anchor in this parasite ( Figure 1A ) [3] , we proposed that the T . cruzi GPI biosynthetic pathway occurs in the ER according to the diagram shown in Figure 1B . Dolichol-phosphate mannose synthase ( DPM1 ) , also named dolichol-phosphate-β-D-mannosyltransferase , catalyses the transfer of a mannose residue from GDP-mannose to dolichol-phosphate ( Dol-P ) generating Dol-P-mannose , used as a donor for all mannosylation reactions that are part of the GPI biosynthetic pathway [40] , [41] . Comparisons among DPM1 of various organisms [42] , [43] , [44] showed that , together with S . cerevisiae , T . brucei , and Leishmania mexicana [45] and in contrast to P . falciparum DPM1 , T . cruzi DPM1 belongs to a group that includes monomeric enzymes that have a C-terminal hydrophobic tail . The glycosyltransferase complex that is responsible for transferring N-acetylglucosamine ( GlcNAc ) from UDP-GlcNAc to phosphatidylinositol ( PI ) to generate N-acetylglucosaminyl-PI ( GlcNAc-PI ) has six and seven proteins , respectively , in yeast and mammalian cells [16] . TcGPI3 was identified as the gene encoding the catalytic subunit of the T . cruzi glycosyltransferase complex since it shares 41% and 49% of sequence identity with the yeast GPI3 and mammalian PIG-A , respectively . Among other components of the glycosyltransferase complex present in yeast , we identified the T . cruzi orthologs of GPI1 , GPI2 , GPI15 , and GPI19 . In mammalian cells , DPM2 , a non-catalytic subunit of dolichol-P-mannose synthase , is physically associated with PIG-A , PIG-C and PIG-Q and enhances GlcNAc-PI transferase activity [46] . A T . cruzi gene encoding a protein with 17% identity to human DPM2 and containing a DPM2 domain , which probably acts as a regulatory component of the N-acetyl-glucosamine transferase complex , was also identified . Only one component of this complex , named ERI1 in yeast [47] , and PIG-Y in mammals [48] , was not identified either in T . cruzi , P . falciparum or T . brucei . The T . cruzi ortholog of yeast GPI12 ( named PIG-L in mammals ) [49] , encoding the enzyme responsible for the de-N-acetylation of GlcNAc-PI , which has been well characterized in T . brucei [50] , [51] , was also identified . Since differences in substrate recognition among the mammal and T . brucei enzyme have been described [52] , this enzyme has been considered as a suitable target for drug development . As depicted in Figure 1B , the first two reactions of the GPI biosynthetic pathway occur on the cytoplasmic face of the ER , whereas mannosylation reactions occur in the ER lumen . After deacetylation , the GPI precursor is transported across the ER membrane to the ER lumen , a step that requires distinct flippases [53] . In yeast and mammalian cells , the addition of mannose residues to GlcN-PI after flipping this precursor into the ER lumen requires acylation of the inositol ring and , after mannosylation and the attachment of GPIs to proteins , this group is removed [54] . In contrast , in T . brucei , inositol acylation occurs after the addition of the first mannose residue [55] since both acylated and non-acylated GPI intermediates exist during transfer of the Man2 and Man3 to GPI intermediates [56] . Although analyses of GPI precursors synthesized in T . cruzi cell-free systems indicated that this organism also has the ability to acylate the inositol ring [57] , sequences encoding an enzyme responsible for acylation of the inositol ring , named PIG-W in mammals and GWT1 in yeast [54] , [58] were not identified either in T . cruzi or in T . brucei [2] . In spite of that , the two alleles encoding the ortholog of the enzyme responsible for inositol deacylation , named GPIdeAc2 in T . brucei [56] , were found in the T . cruzi genome ( Tc00 . 1047053508153 . 1040 and Tc00 . 1047053506691 . 22 ) . All three genes encoding mannosyltransferases , responsible for the addition of the first , second and third mannose residues to GlcN-PI , named TcGPI14 ( α-1 , 4-mannosyltransferase ) , TcGPI18 ( α-1 , 6-mannosyltransferase ) and TcGPI10 ( α-1 , 2-mannosyltransferase ) , were identified in the T . cruzi genome . Since the predicted T . cruzi proteins exhibit sequence identities with yeast and human proteins ranging from 17% to 30% , for some of these genes , functional assays are necessary to confirm these predictions . It is noteworthy that no T . cruzi ortholog encoding the enzyme responsible for the addition of the fourth residue of mannose ( step 6 ) , named SMP3 in yeast and PIG-Z in human , was identified . Similarly , no ortholog of the SMP3 gene was found in P . falciparum , even though the presence of a fourth mannose residue has been shown by structural studies of the GPI anchor from both organisms [3] , [20] , [59] . Furthermore , genes encoding an essential component of the mannosyltransferase I complex named PBN1 in yeast and PIG-X in mammals , have not been identified either in T . cruzi or in T . brucei [60] , [61] . In mammals and yeasts there are three enzymes that add ethanolamine-phosphate ( EtNP ) to different mannose residues: PIG-N/MCD4 ( EtNP addition to Man1 ) , PIG-G/GPI7 ( Man2 ) , and PIG-O/GPI13 ( Man3 ) [2] , resulting in the structure to which the protein will be linked . In T . cruzi , T . brucei and P . falciparum , EtNP addition occurs only at the third mannose [2] , [20] and , as expected , only a T . cruzi GPI13 ortholog was identified . However , it has also been shown in different T . cruzi strains , that GPI-linked proteins as well as free GIPLs have 2-aminoethylphosphonate ( AEP ) replacing EtNP at the third mannose residue and that an additional AEP is linked to GlcN in T . cruzi GPI anchors ( for recent reviews , see [62] , [63] ) . After being assembled , the transfer of the GPI anchor to the C-terminal end of a protein is mediated by a transamidase complex that cleaves the GPI-attachment signal peptide of the nascent protein . In human and yeast , this complex consists of five ER membrane proteins , PIG-K/GPI8 , PIG-T/GPI16 , PIG-S/GPI17 , PIG-U/GAB1 and GAA1 [64] in which GPI8 is considered the catalytic subunit [16] , [65] . As shown in Table 1 , we identified T . cruzi GPI8 , GAA1 and GPI16 orthologs . Although orthologs of GPI17 and GAB1 were not identified in other trypanosomatids , genes encoding two other components of the transamidase complex , known as trypanosomatid transamidase 1 ( TTA1 ) and TTA2 , were also found in T . cruzi [66] . Besides differences in the glycan core , in T . cruzi GPI anchors , the phosphatidylinositol ( PI ) is replaced by inositolphosphorylceramide ( IPC ) , a molecule also present in plants , fungi but not present in mammals [4] . This change in the lipid portion of the anchor occurs during the differentiation of epimastigotes into metacyclic trypomastigotes [67] and is observed in members of the large family of trans-sialidases [68] . Although it may not be considered part of the GPI biosynthetic pathway , the T . cruzi IPC synthase ( TcIPCS ) is thought to be a highly attractive drug target [69] . Based on that , Denny and collaborators [70] identified the ortholog of AUR1 , that encodes the yeast IPC synthase [71] , in Leishmania major and two closely related T . cruzi sequences encoding proteins sharing 52–53% identity with the Leishmania IPC synthase [70] . Our analysis confirmed that the two sequences described by Denny and collaborators [70] correspond to the two alleles of the T . cruzi IPC synthase ( TcIPCS ) gene present in the CL Brener genome , which are synthenic with the L . major and T . brucei orthologs . To verify whether the genes identified through the in silico analyses described above are expressed in T . cruzi , sequences encoding two enzymes of the GPI biosynthetic pathway were used as probes in northern blot hybridizations performed with total RNA purified from epimastigote , trypomastigote and amastigote forms of the parasite . As shown in Figure 2 , transcripts with 1 , 300 nt and 2 , 100 nt , approximately , corresponding to TcGPI8 and TcGPI10 mRNAs were detected in all three stages of the parasite life cycle . As expected , increased levels of both transcripts were found in the two proliferative stages , epimastigotes and amastigotes , compared to the infective , nonproliferative trypomastigote stage . To provide further evidence for the role of the proteins encoded by the T . cruzi genes identified through in silico analyses as components of the GPI biosynthetic pathway , we determined the subcellular localization of three of these proteins expressed as GFP fusion in T . cruzi epimastigotes . The coding regions of TcDPM1 , TcGPI3 and TcGPI12 genes were cloned in the T . cruzi expression vector pTREXnGFP and , after transfection into epimastigotes , the cells were examined by fluorescence microscopy . Figure 3 shows that all three fusion proteins in transfected parasites that were stained with anti-BiP antibodies [38] co-localize with BiP , a known ER marker . Similar results were obtained with confocal microscopy analyses ( not shown ) , thus confirming that these enzymes are part of the GPI biosynthetic pathway . In addition , transfection of T . cruzi genes TcDPM1 , TcGPI3 , TcGPI8 and TcGPI12 in fusion with GFP in the HT1080 human fibrosarcoma cells also resulted in the expression of the GFP fusion T . cruzi proteins with a cellular localization compatible with the ER ( Figure S1 ) . One of the main goals of this work is to develop a strategy for high-throughput screening of drugs against T . cruzi enzymes involved in the GPI biosynthetic pathway . S . cerevisiae has been largely used as surrogate system to express heterologous proteins from diverse parasites including Leishmania spp and T . brucei . Therefore , not only to assay for the functions of the T . cruzi genes but also to create yeast cells expressing T . cruzi target enzymes for future drug studies , conditional lethal yeast mutants were transformed with an expression vector containing the coding sequences for the T . cruzi genes TcDPM1 , TcGPI3 , TcGPI12 , TcGPI14 , TcGPI10 , TcGAA1 , TcGPI8 as well as with the TcIPCS . These mutants were constructed by replacing the endogenous promoter of each one of the GPI genes by the GAL 1 promoter , resulting in yeast cell lines that could only grow in the presence of galactose [31] . By inhibiting the expression of the endogenous GPI genes in medium containing glucose , the complementation of yeast cells with the T . cruzi genes can be easily accessed by comparing the growth of transformed colonies in glucose and galactose-containing medium . As shown in Figure 4A and Table 2 , we tested eight T . cruzi genes for which yeast mutants were available . Three of them , TcDPM1 , TcGPI10 and TcGPI12 , once transformed into yeast , allowed the yeast mutants to grow on plates containing glucose as well as galactose . For all tested yeast mutants , we verified that transformation with plasmids containing the orthologous yeast gene allows them to grow on glucose-containing medium . Figure 4A also shows that , when the mutants were plated on glucose-containing medium supplemented with uracil , none of them were able to grow . As expected , wild type yeast , which has histidine deficiency , does not grow in minimum media lacking histidine . As an additional control , we verified , by RT-PCR analyses , the expression of two T . cruzi genes transformed into yeast mutants , for which we did not observed the complementation , i . e . , that did not grow in nonpermissive media . Transcripts derived from the T . cruzi TcGPI8 or TcIPCS genes , as well as from the orthologous yeast genes , were detected in the corresponding yeast mutants growing in galactose-containing media ( Figure S2 ) , indicating that the inability of these mutants to grow in the presence of glucose is not due to the lack of expression of the T . cruzi genes in the transfected yeasts . To evaluate whether the expression of T . cruzi enzymes in yeast results in the correct synthesis of GPI anchor precursors by the complemented mutants , SDS-PAGE and fluorography analyses of yeast proteins containing [2-3H]myo-inositol were performed . As shown in Figure 4B , after 1 hour growing in medium containing glucose and [2-3H]myo-inositol , a complex pattern of proteins is visualized by fluorography in wild type cells as well as in yeast mutants expressing the T . cruzi genes . The protein patterns in yeast mutants expressing TcDPM1 and TcGPI12 genes growing in glucose-containing medium were indeed indistinguishable from the pattern observed with molecules synthesized by wild type yeasts or by mutants transformed with the orthologous yeast genes . On the other hand , a much weaker signal was detected in non-transformed yeast mutants , indicating that the expression of T . cruzi orthologs encoding enzymes of the GPI biosynthetic pathway restores the mutants' ability to synthesize GPI molecules . Corroborating the functional complementation of yeast mutants with the TcDPM1 gene , thin layer chromatography ( TLC ) of yeast mutants expressing the T . cruzi gene or the yeast ScDPM1gene , as a positive control , showed the presence of dolichol-P-mannose . Yeast cell extracts were preincubated with dolichol-phosphate and labeled in vitro with GDP-[2-3H]mannose . Labeled dolichol-P-mannose was detected in wild type yeast cells as well as in DPM1 mutants that were transfected with the TcDPM1 or with the yeast ScDPM1 gene , confirming that the expression of the T . cruzi enzyme rescues the mutant ability to synthesize dolichol-P-mannose ( Figure S3 ) . Knockout parasites of GPI8 , GPI16 and GPI10 were generated in T . brucei whereas a GPI8 knockout was described in L . mexicana [18] , [19] , [72] , [73] . To further investigate the role of GPI anchors in T . cruzi , we tried to generate parasite cell lines in which both alleles of TcGPI3 , TcGPI8 and TcGPI10 genes were deleted by homologous recombination . Although we were able to generate heterozygote epimastigotes carrying a drug resistance marker inserted in each one of the TcGPI8 alleles ( Figure 5A–B ) , several attempts to generate double-resistant , null mutant epimastigotes with both TcGPI8 alleles deleted were unsuccessful . Also unexpectedly , transfection with plasmid constructs containing TcGPI3 and TcGPI10 sequences flanking the neomycin resistance gene did not result in G418 resistant parasites , indicating that disruption of even one allele of a gene involved in the initial steps of the GPI biosynthesis pathway results in non-viable parasites ( not shown ) . Thus , our results suggest that , in contrast to T . brucei and L . mexicana , the GPI biosynthesis may be an essential pathway in epimastigotes of T . cruzi . In agreement with PCR analyses that showed the disruption of single alleles of TcGPI8 ( Figure 5B ) , northern blot assays ( Figure 5C ) showed that both heterozygous TcGPI8 mutants have the expression of TcGPI8 mRNA reduced by about 40% . Although a few double-resistant epimastigote clones were generated and PCR analyses indicated that the neomycin and hygromycin resistance genes were inserted into both TcGPI8 alleles , PCR amplifications also indicated that additional sequences corresponding to the TcGPI8 gene were present in a different genomic location in the double resistant parasites ( Figure 6A–B ) . It should be noted that it was possible to generate the double resistant parasites only after we prepared different plasmid constructs in which the resistance genes were linked to trans-splicing and polyadenylation signals from the glyceraldehyde-3-phosphate dehydrogenase ( gapdh ) and the ribosomal protein TcP2β ( HX1 ) genes and performing drug selection by gradually increasing drug concentrations . Northern blot analyses ( Figure 6C ) indicate that the recombination events that resulted in viable , double resistant parasites allowed the expression of an aberrant TcGPI8 mRNA population . Among this TcGPI8 mRNA population transcribed in the double resistant mutants , mature , trans-spliced mRNAs were detected by RT-PCR using primers specific for TcGPI8 sequences and the T . cruzi spliced leader ( Figure 6D ) , thus indicating that this gene is still active in these mutants . Although no significant changes in either growing or overall morphology of the TcGPI8 mutants were observed , transmission electron microscopy showed striking alterations in the dense glycocalyx that covers the parasite surface . As shown in Figure 7 , cell membranes of epimastigotes from TcGPI8 heterozygous mutants ( +/−N ) present a thinner layer of the surface glycocalyx compared to wild type ( WT ) epimastigotes . In contrast , cell membranes from both clones of double resistant parasites ( N/H ) , which may have suffered recombination events involving TcGPI8 sequences , present an increased thickness of their glycocalyx compared to the heterozygous mutants ( Figure 7 ) . Although no significant differences in the levels of mucins were detected in the heterozygous mutants , western blot analyses of membrane proteins of WT and double resistant TcGPI8 mutants using the anti-mucin monoclonal antibody 2B10 [74] showed increased amounts of the 35–50 kDa glycoproteins ( also known as Gp35/50 mucins ) expressed on the surface of epimastigotes of the double resistant clones ( Figure 8 ) . Flow cytometry of epimastigotes stained with 2B10 antibodies also showed increased amounts of surface mucins in the double resistant parasites ( Figure S4 ) .
Several T . cruzi surface proteins known to be involved in parasite infectivity or escape from the host immune response are anchored to the parasite membrane by covalent linkage to glycosylphosphatidylinositol ( GPI ) . T . cruzi GPI anchors are also strong proinflammatory molecules , being critical in the modulation of the host immune response against this parasite . T . cruzi belongs to a group of early branching eukaryotic protists that are responsible for several neglected human tropical diseases , for which there is a strong need for new drug treatments . The elucidation of the structures of molecules that play a direct role in host-parasite interactions and the understanding of the biosynthetic pathways that generate these specific parasite molecules , such as the T . cruzi GPI biosynthetic pathway , represent a significant contribution towards this goal . Using a combination of sequence similarity analyses based on yeast , mammal , T . brucei and P . falciparum previously characterized genes , cellular localization and functional expression in yeast mutants we identified 18 orthologous genes encoding components of the GPI biosynthetic pathway present in the published T . cruzi CL Brener genome database . In addition , the gene encoding the IPC synthase , an enzyme responsible for a key modification that occurs in the lipid anchor during in the infective , trypomastigote stage of the parasite , was also identified . Although most sequences were correctly annotated in the TriTrypDB genome database , TcGPI15 , TcGPI19 , TcDPM2 and TcGPI16 were not correctly identified . It should be noted however that , because the assembly of the CL Brener genome is not complete , our in silico analyses might still contain a few missing genes whose presence or absence can only be unambiguously determined by a total sequencing read based analysis or through degenerate PCR experiments . Efforts towards identifying additional genes , such as the ones encoding a component of the mannosyltransferase I complex , α-1 , 2-mannosyltransferase IV or the acyltransferase , responsible for acylation of the inositol ring , are underway . By expressing these genes in yeast mutants , we generated yeast cell lines that can now be used in high throughput screening assays for drugs that are specifically targeted to T . cruzi enzymes . Using yeast as a tool for drug screening against parasites is a strategy that has been successfully employed [5] , [21] , [75] . This system allows the identification of drugs acting specifically on the parasite enzyme since their effect on transfected yeast mutants growing in permissive and nonpermissive media can be compared ( for a recent review , see [76] ) . Alternatively , specific inhibitors can be discovered using cell-free system assays , as it was shown for T . brucei and L . major enzymes involved with GlcNAc-PI de-N-acetylation , mannosylation and inositol acylation [23] , [24] , [25] . It is noteworthy that , among all tested genes , we observed functional complementation in yeast only for those whose products are not part of a protein complex . Among the T . cruzi genes that we were able to show complementation is the DPM1 gene . Since all four mannose residues are likely to be transferred from dolichol-P-mannose , DPM1 , a gene encoding the dolichol-P-mannose synthase is considered an excellent candidate gene to be targeted for drug test studies . In contrast to DPM1 , for which the T . cruzi homologous protein has high levels of amino acid identity with the yeast enzyme , TcGPI10 was also able to complement the yeast mutation even though it has only 21% identity with the yeast enzyme . On the other hand , the T . cruzi IPC synthase , which presents 10% identity with the yeast enzyme and is also a promising target for chemotherapy against trypanosomiases , is not functional in yeast . This is an unexpected result , since it has been shown that the Leishmania major IPC synthase gene ( also known as AUR1 gene ) restored the growth of yeast AUR1 mutants in nonpermissive , glucose-containing media [70] . We further confirmed the role of these genes by analyzing the cellular localization and mRNA expression of their gene products . Sequences corresponding to TcDPM1 , TcGPI3 and TcGPI12 genes , expressed as GFP-fusion proteins in epimastigotes , showed a cellular localization compatible with ER . Interestingly , the T . cruzi sequences containing ER localization signals can be recognized by the mammalian protein trafficking machinery , since we were also able to show similar localization of GFP fusions of TcDPM1 , TcGPI3 , TcGPI8 and TcGPI12 in the ER of transfected HT1080 human fibrosarcoma cells . As expected , analyses of mRNA levels of TcGPI8 and TcGPI10 indicated that the components of the GPI biosynthetic pathway are more actively produced in the two proliferative stages of the parasite life cycle , epimastigotes and amastigotes . To gain further insights into the role of GPI molecules as well as GPI-anchored proteins , we tried to generate T . cruzi null mutants for some of these genes . Because a large number of T . cruzi proteins involved in host-parasite interactions such as members of the large trans-sialidase , mucin and MASP families are GPI anchored , the availability of T . cruzi cell lines with disrupted genes of the GPI biosynthetic pathway would allow us to perform a number of studies regarding the effect of the absence of these proteins on the parasite surface during infection . Given that it encodes the catalytic subunit of the GPI:protein transamidase complex , responsible for transferring GPI anchor to the proteins , we sought to disrupt the TcGPI8 gene , which would have resulted in parasites containing only surface GIPLs , but no GPI-anchored proteins . Not surprisingly , the deletion of a single TcGPI8 allele could be easily achieved by homologous recombination between sequences from each allele flanking the neomycin or hygromycin resistance genes . Accordingly , mRNA expression analyses showed that both TcGPI8 heterozygous mutants have decreased mRNA levels . On the other hand , several attempts to delete the second TcGPI8 allele did not result in viable parasites . When the plasmid constructs were modified and drug selection protocol was conducted in such a way that drug concentrations were increased gradually , rare double resistant cell lines were obtained . However , these parasites seem to have undergone large gene rearrangement involving GPI8 sequences . Although frequently described in Leishmania spp , where gene amplification and overexpression of sequences have been observed after disruption of essential genes [45] , [77] , this phenomenon has been rarely reported for T . cruzi [78] . Together with the results of northern blot and RT-PCR analyses , preliminary data on pulse field gel electrophoresis and southern blot hybridizations ( not shown ) suggested that the amplification of TcGPI8 sequences involved the production of episomal DNA molecules . Thus , the anomalous expression of TcGPI8 mRNA sequences from distinct genomic locations , indicated by a large smear of high molecular weight RNA bands in northern blots and the amplification of spliced leader containing TcGPI8 mRNA allowed the growth of mutants in which both TcGPI8 alleles were disrupted by drug resistance markers . Surprisingly , although no major morphological alterations were evident , electron microscopy analyses of cell membrane structures of epimastigotes showed that TcGPI8 mutants have changes in their glycocalyx layer . Even though the small reduction in the glycocalyx layer observed in the heterozygous mutants could not be correlated with changes in the levels of mucins , western blot with membrane fractions , confirmed by flow cytometry using anti-mucin antibodies indicated that double-resistant parasites present a small increase in the amount of surface glycoproteins , most likely due to an increased expression of the translocated copies of TcGPI8 gene . Mucins play a critical role during infection , since they are the acceptors of sialic acid that allows trypomastigotes to build a negatively charged coat that protects them from killing by host anti-α-galactopyranosyl antibodies [79] . Whether the genomic rearrangements that resulted in the expression of TcGPI8 from different genomic locations have affected the expression of other T . cruzi genes , it remains to be determined . It will be also important to determine which are the mechanisms employed by the parasite that resulted in the genomic rearrangement observed with the double resistant clones . Interestingly , despite being viable in culture , T . brucei mutants lacking TbGPI8 resulted in the absence of GPI-anchored surface proteins , accumulation of non-protein-linked GPI and incapacity of procyclic forms to establish infections in the tsetse midgut [80] . In contrast , GPI8 RNAi knock-down in bloodstream forms resulted in accumulation of unanchored variant surface glycoprotein ( VSG ) and cell death with a phenotype indicative of blocking cytokinesis [72] . On the other hand , L . mexicana GPI8 knockouts , although deficient of GPI-anchored proteins , display normal growth in culture , are capable of differentiating into amastigotes , and are able to infect mice [19] . In addition to GPI8 , procyclic T . brucei lacking the TbGPI12 and TbGPI10 were also obtained . Although unable to synthesize GPI structures beyond GlcNAc-PI , TbGPI12−/− parasites are viable in culture , but are not able to colonize the tsetse midgut [51] . Deletion of TbGPI10 also interferes with the ability of procyclic mutants to infect tsetse flies [18] . These reports are in contrast with our results indicating that disruption of only one allele of a gene involved in the initial steps of the GPI pathway such as TcGPI3 or TcGPI10 resulted in non-viable T . cruzi epimastigotes . On the other hand , similarly to the genomic alterations we observed in the T . cruzi double resistant TcGPI8 mutants , an attempt to create a L . mexicana knockout by targeted deletion of the gene encoding the dolichol-phosphate-mannose synthase resulted in amplification of this chromosomal locus [45] . Thus , our contrasting results attempting to generate T . cruzi null mutants of genes involved with GPI biosynthesis , compared to similar studies described in T . brucei and L . mexicana , suggest that , although considered closely related organisms , the different members of the trypanosomatid family have significant peculiarities that deserve detailed analyses of major biochemical pathways in each parasite species . | Chagas disease , considered one of the most neglected tropical diseases , is caused by the blood-borne parasite Trypanosoma cruzi and currently affects about 8 million people in Latin America . T . cruzi can be transmitted by insect vectors , blood transfusion , organ transplantation and mother-to-baby as well as through ingestion of contaminated food . Although T . cruzi causes life-long infections that can result in serious damage to the heart , the two drugs currently available to treat Chagas disease , benznidazole and nifurtimox , which have been used for more than 40 years , have proven efficacy only during the acute phase of the disease . Thus , there is an urgent need to develop new drugs that are more targeted , less toxic , and more effective against this parasite . Here we described the characterization of T . cruzi genes involved in the biosynthesis of GPI anchors , a molecule responsible for holding different types of glycoproteins on the parasite membrane . Since GPI anchored proteins are essential molecules T . cruzi uses during infection , besides helping understand how this parasite interacts with its host , this work may contribute to the development of better therapies against Chagas disease . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"molecular",
"cell",
"biology",
"microbiology",
"biology",
"genomics"
] | 2013 | Identification and Functional Analysis of Trypanosoma cruzi Genes That Encode Proteins of the Glycosylphosphatidylinositol Biosynthetic Pathway |
Amebiasis , a global intestinal parasitic disease , is due to Entamoeba histolytica . This parasite , which feeds on bacteria in the large intestine of its human host , can trigger a strong inflammatory response upon invasion of the colonic mucosa . Whereas information about the mechanisms which are used by the parasite to cope with oxidative and nitrosative stresses during infection is available , knowledge about the contribution of bacteria to these mechanisms is lacking . In a recent study , we demonstrated that enteropathogenic Escherichia coli O55 protects E . histolytica against oxidative stress . Resin-assisted capture ( RAC ) of oxidized ( OX ) proteins coupled to mass spectrometry ( OX-RAC ) was used to investigate the oxidation status of cysteine residues in proteins present in E . histolytica trophozoites incubated with live or heat-killed E . coli O55 and then exposed to H2O2-mediated oxidative stress . We found that the redox proteome of E . histolytica exposed to heat-killed E . coli O55 is enriched with proteins involved in redox homeostasis , lipid metabolism , small molecule metabolism , carbohydrate derivative metabolism , and organonitrogen compound biosynthesis . In contrast , we found that proteins associated with redox homeostasis were the only OX-proteins that were enriched in E . histolytica trophozoites which were incubated with live E . coli O55 . These data indicate that E . coli has a profound impact on the redox proteome of E . histolytica . Unexpectedly , some E . coli proteins were also co-identified with E . histolytica proteins by OX-RAC . We demonstrated that one of these proteins , E . coli malate dehydrogenase ( EcMDH ) and its product , oxaloacetate , are key elements of E . coli-mediated resistance of E . histolytica to oxidative stress and that oxaloacetate helps the parasite survive in the large intestine . We also provide evidence that the protective effect of oxaloacetate against oxidative stress extends to Caenorhabditis elegans .
Entamoeba histolytica is a protozoan parasite , which inhabits the gastrointestinal tract , and an E . histolytica infection is a substantial health risk in almost all countries where the barrier between food and water and human feces is inadequate . The major clinical manifestations of an E . histolytica infection are amebic colitis , amebic liver abscess , and extraintestinal amebiasis . It is estimated that amebiasis accounted for 55500 deaths and 2 . 237 million disability-adjusted life years ( the sum of years of life lost and years lived with disability ) in 2010[1] . This mortality rate makes an E . histolytica infection the second leading cause of death due to a parasitic infection . E . histolytica is a dimorphic organism whose life cycle has two stages: a trophozoite , a cell-invasive form which can be found in the human intestine , and a cyst , an infective form which is found in the external environment . The conversion between the two stages is usually reversible[2] . Infection of the host occurs upon ingestion of water or food contaminated with cysts . After ingestion , the cysts pass through the stomach , excyst in the small intestine where they produce ameboid trophozoites , which then colonize the large intestine . In the colon , the trophozoites can either asymptomatically colonize the gut , re-encyst , and be expelled in the feces or cause invasive disease[3] . Although the exact conditions , which trigger the onset of invasive disease , are still unknown , the interaction between the parasite’s virulence factors and the host’s response contribute to the development of disease[4] . The human gastrointestinal tract is a nutrient-rich environment which harbors a complex and dynamic population of 38 trillion microbes [5] . About 500–1000 bacterial species colonize the adult intestine , with 30–40 species comprising up to 97% of the total population[6] , [7] . The majority reside in the colon where densities approach 1011 - 1012 cells/ml[8] . Following colonization of the gut , the parasite is constantly interacting with the gut microbiota whose contribution to the manifestation of disease is poorly understood . The trophozoites are quite selective in their interactions with different bacterial species and only those bacteria which have the appropriate recognition molecules get attached to the trophozoites and are ingested[9] . The relationship between E . histolytica and the gut microbiota was the subject of many studies which concluded that the gut microbiota affects greatly several aspects of E . histolytica’s physiology[10 , 11] , [12 , 13] , [14] . In areas where amebiasis is endemic , mixed intestinal infections of E . histolytica and enteropathogenic Escherichia coli ( EPEC ) are common[15] . Bacteria in the intestinal flora including EPEC have been proposed as inducers of amebic virulence , but the causes or mechanisms which are responsible for the induction are still undetermined [16 , 17] . The presence of enteropathogenic bacteria[15] or the presence of Prevotella copri[18] , a normal component of the gut microbiota , has been correlated to a symptomatic outcome of E . histolytica infection in young children . In contrast , mice which were inoculated with commensal Clostridia spp . and segmented filamentous bacteria are protected against an E . histolytica infection[19] . Amebiasis is marked by acute inflammation with the release of cytokines ( tumor necrosis factor alpha ( TNFα ) , interleukin 8 ( IL-8 ) , IL-1β , interferon gamma ( IFN-γ ) , reactive oxygen species ( ROS ) , and nitric oxide ( NO ) from activated cells of the immune system . Depending on their concentration , ROS and NO have been reported to ( a ) trigger stress responses , ( b ) control the activity of E . histolytica virulence factors , and ( c ) be cytotoxic [20] , [21–24] . Recent evidence suggests that the gut microbiome can control oxidative stress and inflammation in the gut [25–29] . Recently , we demonstrated that enteropathogenic Escherichia coli O55 protects E . histolytica against oxidative stress and that this bacterium exerts a strong influence on the transcriptome of oxidatively stressed parasites [30] . In this report , we inform on the mechanism of E . coli-mediated resistance of E . histolytica to oxidative stress by examining the redox proteome of the parasite exposed to E . coli and oxidative stress . We found that live E . coli trigger significant changes in the redox proteome profile of oxidatively stressed E . histolytica . We also found that E . coli malate dehydrogenase ( MDH ) and its product , oxaloacetate , are essential for protecting the parasite against oxidative stress .
The results of our recent investigation indicate that preincubation of E . histolytica trophozoites with live E . coli O55 , but not with heat-killed E . coli O55 , confers resistance to H2O2-induced oxidative stress to the parasite [30] . In order to obtain insights into the mechanisms of survival of oxidatively stressed E . histolytica trophozoites , we did an OX-RAC analysis of the proteins in E . histolytica trophozoites which were exposed to live or heat-killed E . coli O55 for 30 minutes and then exposed to 2 . 5 mM H2O2 for 60 minutes at 37°C ( Fig 1A ) [22] . The purification of OX-proteins by OX-RAC analysis , which has been previously described in detail [22] , has three steps: ( i ) blocking by N-ethylmaleimide ( NEM ) of non-oxidized cysteine residues present in E . histolytica proteins; ( ii ) reduction of oxidized cysteine residues with dithiothreitol ( DTT ) ; and ( iii ) binding of the cysteine residues reduced by DTT to a thiopropyl sepharose resin . The OX-proteins are then eluted from the thiopropyl sepharose resin and identified by mass spectrometry . We identified 329 OX-proteins in those trophozoites which were exposed to heat-killed E . coli O55 and H2O2 ( THK ) and 300 OX-proteins in those trophozoites which were exposed to live E . coli O55 and H2O2 ( TL ) ( S1 & S2 Tables ) . Using the PANTHER sequence classification tool[31] , we found that a third of the OX-proteins are shared by THK and TL ( ( Fig 1B and S3 & S4 Tables ) . We also found that OX-proteins in the following biological processes were significantly enriched in THK ( Fig 1C and S5 Table ) : redox homeostasis ( GO:0045454 ) ( exemplified by thioredoxin EHI_004490 or EHI_006700 ) , lipid metabolism ( GO:0044255 ) ( exemplified by 3-ketoacyl-CoA synthase EHI_111000 or geranylgeranyl pyrophosphate synthase EHI_105060 ) , small molecule metabolism ( GO:0044281 ) ( exemplified by glyceraldehyde-3-phosphate dehydrogenase EHI_008200 and threonine dehydratase EhTD1 EHI_092390 ) , carbohydrate derivative metabolism ( GO:1901135 ) ( exemplified by mannosyltransferase EHI_103330 and glucosidase 2 subunit beta EHI_135420 ) , and organonitrogen compound biosynthesis ( GO:1901566 ) ( exemplified by alpha-1 , 3-mannosyltransferase ALG2 , EHI_162230 and dolichyl-diphosphooligosaccharide—protein glycosyltransferase subunit 1 EHI_029540 ) . In contrast , we found that OX-proteins in the following biological process were significantly enriched in TL: redox homeostasis ( GO:0045454 ) ( exemplified by thioredoxin EHI_004490 or EHI_006700 ) ( Fig 1C and S5 Table ) . The small number of shared OX-proteins and biological processes in THK and TL suggests that exposure of the parasite to live E . coli influences the parasite’s redox proteome . We also found 70 E . coli proteins that were co-purified with E . histolytica OX-proteins ( S6 Table ) . It has been previously reported that a catalase-deficient strain of Salmonella dysenteriae , which cannot decompose H2O2 to water and oxygen , cannot boost the virulence of oxidatively stressed trophozoites whereas a wild-type S . dysenteriae does [32] . Accordingly , we posited that those E . coli proteins which are involved in the bacterium’s resistance to oxidative stress also participate in the parasite’s protective mechanism against oxidative stress . Five of the 70 E . coli proteins were selected because of their participation to the resistance of E . coli to oxidative stress , namely , 60 kDa chaperonin ( it protects proteins which are denatured by oxidative stress from misfolding and aggregating [33 , 34] ) ; superoxide dismutase ( it catalyzes the decomposition of the superoxide free radical [35] ) ; MDH ( it catalyzes the formation of oxaloacetate from malate and it protects E . coli against oxidative stress [36] ) ; aspartate ammonia-lyase ( it catalyzes the formation of ammonia and promotes the formation of antioxidant polyamines in E . coli [37] [38] ) , and alkyl hydroperoxide reductase subunit C ( it catalyzes the NADH-dependent reduction of H2O2 to H2O [39] ) . We then investigated the ability of E . coli K12 and the following E . coli mutants: 60 kDa chaperonin ( JW 4103- Δ gro ) , superoxide dismutase ( JW1648- Δsod ) , MDH ( JW 3205- Δ mdh ) , aspartate ammonia-lyase ( JW4099- ΔaspA ) , and alkyl hydroperoxide reductase subunit C ( JW0598- Δ ahpc ) to protect E . histolytica against H2O2-induced oxidative stress . We found that the levels of protection that are conferred by E . coli O55 [30] and E . coli K12 on oxidatively stressed E . histolytica trophozoites are similar ( Table 1 ) . Among the E . coli K12 mutants , we found that the MDH mutant E . coli JW 3205 was the only one that did not protect E . histolytica against H2O2-induced oxidative stress ( Table 1 ) . In order to demonstrate whether MDH is essential for protecting E . histolytica against H2O2-induced oxidative stress , E . coli JW 3205 was complemented with a plasmid which harbored wild-type E . coli mdh . MDH activity was determined in the whole lysate of E . coli K12 , E . coli JW 3205 , and E . coli JW 3205 complemented with mdh ( Table 2 ) . We found MDH activity in E . coli K12 and in E . coli JW 3205 complemented with mdh but not in E . coli JW 3205 ( Table 2 ) . We also found that E . coli JW 3205 complemented with mdh , but not E . coli JW 3205 , protects E . histolytica against H2O2-induced oxidative stress ( Table 1 ) . MDH catalyzes the reversible transformation of malate into oxaloacetate and it has been reported that oxaloacetate in millimolar concentrations protects E . coli against H2O2-induced oxidative stress[36] . In order to test the hypothesis that oxaloacetate is essential for protecting E . histolytica against H2O2-induced oxidative stress , we determined the viability of E . histolytica trophozoites first exposed to different concentrations of oxaloacetate ( 0–2 mM for 15 minutes ) and then to H2O2 ( 2 . 5 mM for one hour ) . We found that oxaloacetate at concentrations higher than 0 . 25 mM protects the parasite against H2O2-induced oxidative stress ( Table 1 and S1 Fig ) . We also found that malate ( 0 . 5 mM ) has no effect on the resistance of the parasite to H2O2-induced oxidative stress ( Table 1 ) . Paraquat has been previously used to trigger oxidative stress in E . histolytica[40] . In order to demonstrate whether oxaloacetate protects the parasite against paraquat , trophozoites were incubated with oxaloacetate ( 2 mM ) and then exposed to paraquat ( 2 . 5 mM ) . We found that oxaloacetate does not protect the parasite against paraquat ( S2 Fig ) . Next , we tested whether E . coli O55 also protects E . histolytica against nitrosative stress . For this purpose , we exposed E . histolytica trophozoites to the nitric oxide ( NO ) donor , S-nitrosoglutathione ( 350 μM for 120 minutes ) [41] . We found that E . histolytica trophozoites exposed to live or heat-killed E . coli O55 are not protected against S-nitrosoglutathione-induced nitrosative stress ( S3 Fig ) . We also found that oxaloacetate ( 2 mM ) does not protect the parasite against S-nitrosoglutathione-induced nitrosative stress ( S3 Fig ) . It has been reported that E . coli can boost the virulence of E . histolytica and that this boosting is contact dependent[42] . However , it is not known whether contact between E . coli and the parasite is necessary for protecting the parasite against H2O2-induced oxidative stress . This question was addressed by physically separating the trophozoites from E . coli O55 or E . coli K12 with a polycarbonate insert ( 0 . 4 μm ) prior to the exposure of the parasite to H2O2 . We found that E . coli O55 and E . coli K12 inside the polycarbonate insert protect the parasite against H2O2-induced oxidative stress ( S4 Fig ) . Since E . coli MDH ( EcMDH ) , which is secreted by E . coli[43] , is essential for protecting E . histolytica against H2O2-induced oxidative stress ( Table 2 ) and this protection is contact independent ( S4 Fig ) , we posited that EcMDH contributes to the resistance of E . histolytica to H2O2-induced oxidative stress . This hypothesis was tested by first incubating E . histolytica with commercial His-tagged EcMDH and then exposing the parasite to H2O2-induced oxidative stress . As a prerequisite to this experiment , we checked the activity of the commercial His-tagged EcMDH and found that the recombinant protein is catalytically active ( Table 2 ) . We found that the presence of His-tagged EcMDH did not protect the parasite against H2O2-induced oxidative stress ( Fig 2 ) . However , we found that the parasite is protected against H2O2-induced oxidative stress when the parasite is incubated with both His-tagged EcMDH ( 1 . 5 μg ) and L-malate ( 50 mM ) prior to its exposure to H2O2-induced oxidative stress ( Fig 2 ) . No MDH activity was detected in a whole lysate and the secretory products of control E . histolytica trophozoites [21] ( Table 2 ) despite the presence of six putative MDH genes in the genome of the parasite [44] ( EHI_152670 , EHI_067860 , EHI_165350 , EHI_030810 , EHI_092450 and EHI_014410 ) and one MDH in the secretome of E . histolytica [45] ( EHI_092450 ) . When the E . histolytica MDH ( EhMDH , EHI_067860 ) is overexpressed in E . coli , its activity is less than 40% of its E . coli homolog ( Table 2 ) . These results raise a question on whether the parasite can express an active MDH . To answer this question , E . histolytica trophozoites were complemented with HA-tagged EcMDH . The expression of HA-tagged EcMDH in the trophozoites was confirmed by western blotting ( Fig 3A ) and detecting MDH activity in their whole lysates ( Table 2 ) . When we compared the resistance of HA-tagged EcMDH trophozoites to H2O2-induced oxidative stress ( Fig 3B ) and trophozoites transformed with the pcontrol plasmid[41] , we found that the sensitivity of the HA-tagged EcMDH trophozoites and the pcontrol trophozoites to H2O2-induced oxidative stress was identical . Incubation of the HA-tagged EcMDH trophozoites in presence of malate ( 2 mM ) prior to their exposure to H2O2-induced oxidative stress did not increase their resistance to H2O2-induced oxidative stress ( Fig 3B ) . The intracellular concentration of oxaloacetate was determined in pcontrol trophozoites , HA-tagged EcMDH trophozoites , E . coli K12 , and E . coli JW 3205 . We were able to detect 12 . 7 nmol oxaloacetate/mg proteins in E . coli K12 , but no oxaloacetate was detected in the pcontrol trophozoites , the HA-tagged EcMDH trophozoites , and E . coli JW 3205 . No oxaloacetate was also detected in the supernatant of the growth medium of E . coli or E . histolytica . Ketoacids act as non-enzymatic antioxidants due to their ability to scavenge H2O2 [46 , 47] . We determined the amount of H2O2 in presence or absence of oxaloacetate and confirmed that the amount of H2O2 dropped in presence of oxaloacetate ( S5 Fig ) . The antioxidant property of oxaloacetate was also evaluated by determining the formation of OX-proteins in trophozoites exposed to H2O2 and trophozoites exposed to H2O2 and oxaloacetate using OX-RAC ( Fig 4; left panel ) . The amount of OX-proteins in those trophozoites exposed to H2O2 was four times bigger than that in those trophozoites exposed to H2O2 and oxaloacetate . Gal/GalNac lectin is a surface protein which is essential for the binding of the parasite to target cells [48] and its binding activity is inhibited by the oxidation of the cysteine residues present in the carbohydrate recognition domain of the heavy subunit Gal/GalNAc lectin ( Hgl ) ( EHI_012270 ) [22] . We decided to determine the redox status of Hgl in trophozoites exposed to H2O2 and trophozoites exposed to H2O2 and oxaloacetate by western blotting of the OX-proteins ( Fig 4; right panel ) . We detected a strong Hgl signal in those trophozoites exposed to H2O2 whereas Hgl is barely detectable in those trophozoites exposed to H2O2 and oxaloacetate . Collectively , these results indicate that oxaloacetate reduces the formation of OX-proteins in trophozoites exposed to H2O2 . It has been previously reported that H2O2-induced oxidative stress impairs the cytopathic activity of E . histolytica [16] . We found that the cytopathic activity of E . histolytica trophozoites incubated with E . coli K12 or E . coli JW 3205 complemented with mdh was not impaired when they were exposed to H2O2-induced oxidative stress and was impaired when they were exposed to E . coli JW 3205 ( Fig 5A ) . In order to test whether oxaloacetate can protect the cytopathic activity of oxidatively stressed E . histolytica trophozoites , E . histolytica trophozoites were exposed or not exposed to oxaloacetate prior to their exposure to H2O2 ( Fig 5B ) . We found that the cytopathic activity of the oxidatively stressed E . histolytica trophozoites is substantially impaired , but this impairment does not occur when the parasite was preincubated with oxaloacetate before being exposed to H2O2 ( Fig 5B ) . The amebicidal activity of murine macrophages depends on the formation of reactive oxygen and nitrogen species and the addition of catalase to the culture medium of activated murine macrophages reduces their amebicidal activity [49] . Based on these data , we determined the effect of oxaloacetate on the amebicidal activity of activated murine macrophages ( Fig 6 ) . We found that the amebicidal activity of activated murine macrophages is substantially impaired when these macrophages were incubated with oxaloacetate ( Fig 6 ) . The effect of oxaloacetate on the survival of E . histolytica in the large intestine was tested in a mouse strain resistant to intestinal amebiasis C57BL/6 ( B6 ) [50] and in a mouse strain susceptible to intestinal amebiasis CBA/J [51] . The survival of the parasite was determined by counting the number of trophozoites in the stool after cultivation and by amplification of E . histolytica 18S rRNA in DNA extracted from the stool by polymerase chain reaction ( PCR ) . We found that intracecal injection of the parasite with oxaloacetate ( 2 mM ) helps the parasite to survive in the intestine . ( Fig 7A & 7B ) . It has been reported that the glycolytic activities of oxidatively stressed E . histolytica trophozoites are impaired and this impairment results in a redirection of the metabolic flux toward glycerol production[40] . Therefore , the amount of glycerol in the parasite could be used as an indicator of the level of oxidative stress sensed by the parasite . We found that the amount of glycerol in those parasites exposed to H2O2 was two-fold greater than that in the unexposed parasites ( Fig 8 ) . We also did not detect any differences in the amount of glycerol in those parasites exposed to oxaloacetate and in those trophozoites exposed to oxaloacetate and H2O2 . Husain et al . [40] reported that the amount of isocitrate in control ( unstressed ) and oxidatively stressed trophozoites is similar . This finding suggests that the intracellular amount of isocitrate could be used as an internal standard for our metabolomic data in the control and oxidatively stressed trophozoites . We found no difference in the amounts of isocitrate in the oxidatively stressed and control parasites which confirmed the finding of Husain et al . [40] and concluded that the quality of our metabolomic data is good . ( Fig 8 ) . The environment in which C . elegans lives contains bacteria and C . elegans uses these bacteria as its food source . Based on our findings in E . histolytica , we tested the hypothesis that bacteria can also influence the ability of C . elegans to resist H2O2-induced oxidative stress . It has been previously reported that oxaloacetate increases the life span of C . elegans [52] . Additionally , our knowledge on the effect of oxaloacetate on the resistance of the nematode to H2O2-induced oxidative stress is lacking . In order to fill these knowledge gaps , C . elegans at the L1 developmental stage were incubated with different concentrations of oxaloacetate and then exposed to H2O2 . We found that the survival rate of those worms which were treated with oxaloacetate ( 2 . 5 mM or higher ) is higher than those worms which were treated with oxaloacetate ( 1 mM or less ) prior to their exposure to H2O2 ( Fig 9A ) . C . elegans were also incubated in M9 medium without bacteria or with E . coli K12 or E . coli JW 3205 prior to their exposure to H2O2 . We found that both E . coli K12 and E . coli JW 3205 protect C . elegans against H2O2 ( Fig 9B ) whereas only E . coli K12 protected E . histolytica against H2O2 ( Table 1 ) . These findings suggest that the mechanism of protection of C . elegans against H2O2 is more complex than that in E . histolytica and that it does not solely depend on the activity of EcMDH .
In a previous report , we informed on the results of our redox proteomics analysis of oxidatively stressed E . histolytica trophozoites [22] . In order to understand the mechanisms of survival of E . histolytica trophozoites that were incubated with E . coli prior to their exposure to H2O2 , we did a redox proteomics analysis of the parasite which was exposed first to heat-killed or live E . coli and then exposed to H2O2 . The proteins involved in redox homeostasis , mainly , thioredoxin ( TRX ) and protein disulfide isomerase ( PDI ) were identified as OX-proteins in oxidatively stressed E . histolytica trophozoites which were exposed to heat-killed or live E . coli . This result indicates that the presence of heat-killed or live E . coli has no detectable effect on the oxidation status of these proteins . In previous investigations , we reported that TRX and PDI are oxidized and nitrosylated when E . histolytica trophozoites are exposed to oxidative and nitrosative stress [22 , 53] . PDIs are oxidoreductases and isomerases which are involved in the unfolded protein response[54] . In the oxidation-reduction reaction to reduce peroxyredoxin or decompose H2O2 into H2O , TRX is the first substrate to be transformed by TRX reductase ( TrxR ) [55] . Since TRX is susceptible to oxidation[56] and is reduced as part of its antioxidant activity[57 , 58] , these properties may be reasons why TRX is oxidized in the oxidatively stressed parasite after its exposure to heat-killed or live E . coli . We detected OX-proteins in E . histolytica trophozoites which were incubated with heat-killed E . coli and exposed to H2O2 . Since these OX-proteins were not detected in E . histolytica trophozoites which were incubated with live E . coli and exposed to H2O2 , their absence suggests that their oxidation in E . histolytica trophozoites depends on the presence of live E . coli . One of these OX-proteins is the 60S acidic ribosomal protein L9 which belongs to the “organonitrogen compound biosynthesis” class of proteins . In our previous investigation , we also identified the 60S acidic ribosomal protein L9 ( EHI_193080 ) as an OX-protein[22] . We have previously discussed that the inhibition of protein synthesis in oxidatively stressed E . histolytica trophozoites results from the oxidation of different components of the parasite’s translational machinery which includes the 60S acidic ribosomal proteins[22] . The absence of oxidized ribosomal proteins in trophozoites which were incubated with live E . coli prior to their exposure to H2O2 suggests that the presence of live E . coli protects these proteins from oxidation . We can also deduce that this presence facilitates the survival of the oxidatively stressed parasite . The results of previous studies have shown that a short preincubation of E . histolytica trophozoites strain HM-1:IMSS with E . coli O55 can boost the parasite’s ability to destroy monolayers of cultured cells[10] . Our findings indicate that the impaired cytopathic activity of oxidatively stressed E . histolytica trophozoites can be regained by pre-incubating the parasite with E . coli or oxaloacetate . These results and those of others[20] suggest that the parasite’s virulence and ability to resist oxidative stress are linked . This boosting of E . histolytica’s virulence by E . coli O55 is contact dependent[10] and relies on the presence of galactose lectin on the parasite’s surface . We found that the protective effect of E . coli on oxidatively stressed E . histolytica trophozoites does not rely on the binding of the bacteria to the parasite but on the formation of oxaloacetate by E . coli . Pyruvate , oxalo-ketoglutarate , and other ketoacids function as non-enzymatic antioxidants due to their ability to scavenge H2O2[46 , 47] . It has also been reported that ketoacids can protect E . coli[36] , several eukaryotic cell types[59] , [60] , [61–63] and even whole organs , such as the heart and kidney[64–66] against oxidative stress . It is also interesting to note that the increase in the life span of C . elegans after exposure to oxaloacetate[52] may be due to the ability of oxaloacetate to scavenge H2O2 . The ability of oxaloacetate to scavenge H2O2 is supported by our findings of a reduction in ( i ) the amount of OX-proteins in trophozoites exposed to H2O2 and oxaloacetate; ( ii ) the intracellular amount of glycerol in trophozoites exposed to H2O2 and oxaloacetate and ( iii ) the concentration of H2O2 is reduced in presence of oxaloacetate . The mechanism of H2O2 scavenging by oxaloacetate is not compatible with the detoxification of s-nitrosothiol groups which are formed during nitrosative stress[67] . Since it is also not compatible with the intracellular formation of superoxide by paraquat[68] , this finding may explain why oxaloacetate cannot protect E . histolytica against nitrosative stress and oxidative stress induced by S-nitrosoglutathione and paraquat , respectively . We found that oxaloacetate produced by E . coli protects E . histolytica against H2O2induced oxidative stress . It is possible that H2O2 is detoxified by oxaloacetate inside the bacteria . The need of an active EcMDH for protecting E . histolytica against H2O2 is supported by the results of our experiment in which we found that heat-killed E . coli are unable to protect the parasite against H2O2 . It is also possible that H2O2 is detoxified by oxaloacetate which is secreted into the culture medium . To our knowledge , such secretion of oxaloacetate by E . coli and other microorganisms is not supported by the literature . Our attempts to detect oxaloacetate in the extracellular medium using metabolomics or an enzymatic-based kit failed because it is a very unstable metabolic intermediate[69] . However , it is possible that oxaloacetate is formed extracellularly by secreted EcMDH because ( i ) EcMDH is part of E . coli’s secretome [43] and ( ii ) the addition of recombinant EcMDH to the medium in presence of malate confers resistance to the parasite against H2O2-induced oxidative stress ( this work ) . Despite the presence of six putative MDH genes in the genome of the parasite [44] , we and others [21] were unable to detect any MDH activity in a total lysate of the parasite . Several different explanations may account for this experimental observation: ( i ) EhMDHs are very sensitive to environmental conditions and are denatured once the parasite is lysed , ( ii ) the parasite has lost an active MDH during the course of evolution because it can relies on malate and oxaloacetate produced by the gut microbiota and ( iii ) EhMDHs may have a different enzymatic activity unrelated to their metabolic function as reported for a number of E . histolytica moonlighting enzymes [41 , 70] . This third hypothesis is supported by the unexpected presence of two MDHs ( EHI_030810 and EHI_165350 ) on the parasite’s surface [71] . Since E . histolytica is able to express a functional EcMDH ( based on the activity measured in a total lysate of the EcMDH trophozoites ) , its sensitivity to H2O2-induced oxidative stress was comparable to that of pcontrol trophozoites . Since oxaloacetate was not being detected in the EcMDH-overexpressing trophozoites even when malate was present in the medium , we concluded that ( i ) the concentration of intracellular malate in the parasite is limiting , ( ii ) the parasite cannot transport external malate , and ( iii ) EcMDH is not functional inside the parasite . Another possibility is that the newly formed oxaloacetate is quickly converted to a non-protective metabolite in the parasite . Pyruvate:ferredoxin oxidoreductase ( PFOR ) is an enzyme which uses oxaloacetate as an alternative substrate to pyruvate may be responsible for this conversion[72] . Alternatively , it is possible that EcMDH is active inside the parasite , but the production of oxaloacetate is insufficient to neutralize the cytotoxic effect of H2O2 . Accordingly , any neutralization or antioxidant mechanism must be operative extracellularly in order to be effective . We found that the survival of oxaloacetate-treated parasites is better than that of parasites which were not exposed to oxaloacetate in a mouse model of amebic colitis . It has been previously shown that a strong inflammatory response occurs after injecting the parasite into the large intestine of mice [73] . This inflammatory response is essential for killing the parasite because neutrophil-depleted or dexamethasone-treated C3H or CBA mice are more susceptible than untreated mice [73] . It is possible that oxaloacetate neutralizes H2O2 which is produced by neutrophils inside the colon of the trophozoite-infected mice and consequently promotes the parasite’s survival . This hypothesis is supported by our in vitro data which demonstrated that the amebicidal activity of activated murine macrophages which depends on the formation of ROS and NO [74] is reduced in presence of oxaloacetate . The susceptibility of mice to an E . histolytica infection depends in part on the content of sialic acids in intestinal mucins and the binding of the parasite to these mucins is mediated by Hgl [75] . The better colonization of oxaloacetate-treated parasites than untreated parasites in the large intestine of mice may be explained by the presence of a functional Hgl in the oxaloacetate-treated parasite . Specifically , Hgl in the untreated parasite is inactivated or becomes non-functional when it is oxidized [22] . Although we don’t know the exact concentration of oxaloacetate in the human large intestine due to the instability of oxaloacetate , the concentration of malate , its precursor , is in the millimolar range [76] . Based on this information , it is tempting to speculate that enough oxaloacetate is produced in the human large intestine to protect the parasite against H2O2-induced oxidative stress . To conclude , we have done the first redoxomics of E . histolytica incubated with E . coli and exposed to H2O2-induced oxidative stress . Although it is difficult to deduce from our data whether changes in the redox status of E . histolytica proteins actually occur when the parasite resides in its host , this investigation highlights that the interaction between the parasite and the gut flora is more complex than the predator-prey relationship . A complex interaction has also been recently described in C . elegans . This bacteria-feeding nematode can avoid pathogenic bacteria , such as Pseudomonas aeruginosas , by sensing some of their secondary metabolites [77] . Other parasitic protozoa and helminths which are also in a tight relationship with the host’s intestinal microbiota may benefit from the antioxidant properties of oxaloacetate which is produced by the gut microbiota [78] . This proposition is supported by our data about the protection of C . elegans by oxaloacetate against H2O2-induced oxidative stress . Strategies that counteract the protective effect of oxaloacetate against oxidative stress may be valuable in the treatment of amebiasis .
E . histolytica trophozoites HM-1:IMSS strain were grown under axenic condition at 37°C in Trypticase Yeast Extract Iron Serum ( TYI-S-33 ) medium prepared according to a previously reported protocol[79] . The trophozoites were harvested during the logarithmic phase of growth by chilling the culture tubes at 4°C and pelleted by centrifugation at 500 g for five minutes . The pellet was washed twice with ice-cold phosphate-buffered saline . The bacterial strains used in this study are listed in S7 Table . E . coli was grown at 37°C in Luria-Bertani ( LB ) medium[80] . C . elegans strain Bristol N2 were grown and maintained at 16°C on nematode growth media ( NGM ) agar using a previously reported protocol[81] . E . coli OP50 was used as their food source . For synchronizing the worms , gravid adults were treated with a freshly prepared 20% sodium hypochlorite solution to isolate embryos . Embryos were then incubated overnight in M9 solution in a nutator at 20°C without food to allow hatching to the L1 developmental stage and to prevent further development . The number of worms was estimated using a hemocytometer . HeLa cells ( a kind gift from Dr . Kleinberger , Faculty of Medicine , Technion ) were maintained in continuous culture using a previously described protocol [41] . The transfection of E . histolytica trophozoites was done using a previously described protocol[82] . E . histolytica trophozoites ( 1×106 ) in TYI-S-33 medium ( without serum ) were exposed to different concentrations of H2O2 ( 0-5mM ) for one hour at 37°C . The viability of the trophozoites was determined by the eosin dye exclusion method[41] . E . histolytica trophozoites were first cultivated in standard TYI-S-33 medium in 7 ml culture tubes for 12 hours at 37°C . The culture medium was then replaced with fresh and warm culture medium and paraquat ( 2 . 5 mM final concentration ) was added and the culture was continued for 12 hours . The viability of the trophozoites was determined by the eosin dye exclusion method[41] . C . elegans ( 100 L1 larvae ) after synchronization in M9 medium were placed into each well of a 24-well plate . The worms were first exposed to different concentrations of oxaloacetate ( 0–5 mM ) for five minutes , exposed to 2 . 5 mM H2O2 , and then incubated in an orbital shaker at 20°C for 2 , 4 , 6 , and 24 hours . To determine their viability after each exposure time , the worms were seeded on one side of a NGM agar plate and E . coli OP50 was seeded on the opposite side . After a 1-hour incubation at room temperature , the number of viable worms was assessed by measuring their mobility[83] . At least three biological replicates were performed for each experiment . The viability of C . elegans was also measured after preincubation of the worms with E . coli K12 or E . coli JW3205 for five minutes . Subsequently , the worms were exposed to 2 . 5 mM H2O2 for 2 , 4 , 6 , and 24 hours . The viability of the worms was assessed by measuring their mobility using the previously described method . E . histolytica trophozoites ( 1×106 ) in TYI-S-33 medium ( without serum ) were exposed to 350μM S-nitrosoglutathione ( Sigma-Aldrich , St . Louis , MO , USA ) for two hours at 37°C . The viability of the trophozoites was determined by the eosin dye exclusion method[41] . E . histolytica trophozoites ( 1×106 ) in 500 μl TYI-S-33 medium ( without serum ) were seeded into each well of a 24-well plate ( Nunclon delta surface , Thermo Scientific , Israel ) . A polycarbonate SPL insert ( 0 . 4 μm ) ( SPL Biosciences , Israel ) was introduced into each well and E . coli O55 or E . coli K12 ( 1×109 bacteria in 500 μl TYI-S-33 medium ( without serum ) ) were introduced into the SPL insert . No bacteria were introduced in the SPL insert of the control trophozoite culture . Oxidative stress was generated in some wells by adding 2 . 5 mM H2O2 directly into the SPL insert . The viability of the trophozoites was determined by the eosin dye exclusion method[41] . The cytopathic activity of E . histolytica trophozoites was determined using a previously described protocol [84] . The amebicidal activity of activated murine macrophages was determined using previously described protocol [49] . Briefly , RAW 267 . 7 macrophages ( a kind gift from Dr . Moran Benhar , Faculty of Medicine , Technion ) were activated by a 22-hour incubation with lipopolysaccharides ( LPS ) ( 1 μg/μl ) and interferon γ ( INF-γ ) ( 100 U/ml ) in absence of presence of oxaloacetate ( 2mM ) . Activated macrophages ( 2 × 106/ml ) and E . histolytica trophozoites ( 2 × 104/ml ) were co-incubated at 37°C for six hours . The viability of trophozoites was determined by the eosin dye exclusion method [41] . C57BL/6 and CBA/J mice were purchased from the Jackson Laboratory ( Japan ) . The mice were maintained under specific pathogen-free conditions . Trophozoites for intracecal injections were originally derived from laboratory strain HM1:IMSS ( American Type Culture Collection ) that were sequentially passaged in vivo through the mouse cecum [50] . For all intracecal injections , axenic trophozoites were grown to the log phase and counted with a hemacytometer , and 1 × 106 trophozoites in 200 μl TYI-S-33 medium were injected in the presence or absence of oxaloacetate ( 2 mM ) into the proximal , middle , and apical regions of the cecum [85] of mice anesthetized with Domitor ( medetomidine hydrochloride , 0 . 1 mg/kg ) and Dormicum ( midazolam , 0 . 1 mg/kg ) . At the end of the experiment , mice were sacrificed by barbiturate overdose . We anesthetized mice with Domitor ( medetomidine hydrochloride , 0 . 1 mg/kg ) and Dormicum ( midazolam , 0 . 1 mg/kg ) . Mice were sacrified by Barbiturate overdose . All experiments that involved mice were reviewed and approved by the Committee for Ethics on Animal Experiments in the Graduate School of Gunma University , and were conducted under the control of the Guidelines for Animal Experiments in the Graduate School of Medicine , Gunma University , and the Law ( No . 105 ) and Notification ( No . 6 ) of the Japanese Government . The protocol number 16–041 has been assigned by the Committee for Ethics on Animal Experiments in the Graduate School of Gunma University after approval of the animal experiments described in this study . QIAamp DNA stool kits ( Qiagen , Valencia , CA ) was used for DNA extraction from the mice feces according to the manufacturer’s instructions . To quantify the presence of E . histolytica trophozoites in stool , real-time quantitative PCR was performed by using using SYBR Green Supermix ( Life Technologies , TA , CA , USA ) in the Quant Studio 7 Flex Real-Time PCR System ( Applied Biosystems® , Life Technologies , CA , USA ) . Primer sets specific to E . histolytica 18S rRNA were EntaF and EhR ( S8 Table ) . To make a standard curve , DNA extracted from E . histolytica trophozoites was serially diluted from 105 to 100 . Based on the standard curve and the stool weight , the number of trophozoite/mg stool was calculated . The presence of live trophozoites in the stool was confirmed by cultivation of stool in TYI-S-33 medium in presence of 103 unit/ml penicillin G , 1 mg/ml streptomycin , and 2 . 5 μg/ml amphotericin B . Overnight cultures of E . coli were diluted 100-fold in 10 ml of fresh LB medium and incubated at 37°C until the OD600 reached 0 . 5 . The cells were then harvested by centrifugation , washed once with phosphate-buffered saline , and finally re-suspended in 500 μl buffer which contained 0 . 1M Tris ( pH 7 . 4 ) , 2 mM EDTA , 0 . 2M DTT , and 0 . 5mM PMSF . The cells were lysed by ultrasonic disintegration using an ultrasonic disintegrator ( Topas GmbH ) which was operated five times for ten seconds at 50% output power at 4°C . The resultant homogenates were centrifuged at 15 , 000×g for 15 minutes at 4°C , and the supernatants were used for measuring the MDH activity . Proteins secreted by E . histolytica trophozoites were isolated using a previously described protocol [45] . Enzyme assays were performed in a 1-cm cuvette which contained 890 μl MDH assay buffer ( 50 mM Tris buffer and 2 mM NAD+ ) and 10 μl of test sample . The reaction was initiated by the addition of 100 μl L-malate ( 500mM ) and the rate of formation of the reduced form of nicotinamide adenine dinucleotide ( NADH ) was monitored at 340 nm using spectrophotometer ( Pharmacia Biotech Ultrospec 2000 ) . One unit of MDH activity is defined as 1 μmol of NAD+ converted to its reduced form/min/mg protein . The protein concentration was determined by the Bradford method [86] . The commercial E . coli His-tagged MDH was purchased from Abcam ( Recombinant E . coli mdh protein ab124594 ) . The recombinant E . coli MDH protein was diluted with MDH assay buffer to a concentration of 60 μg/ml . Viability of E . histolytica trophozoites preincubated with His-tagged EcMDH ( 1 . 5 μg ) with or without malate ( 50 mM ) was performed as described above . For the cloning of E . coli MDH , the E . coli MDH gene was amplified from E . coli genomic DNA using the E . coli MDH 5’ and E . coli MDH 3’ primers ( S8 Table ) . The PCR product was sub-cloned using the pGEM-T easy vector system ( Promega , Madison , Wisconsin , USA ) . For construction of the pJST4-E . coli MDH expression vector that was used to express HA-tagged EcMDH in the parasite , a synthetic E . coli MDH gene was ordered ( Synthezza , Israel ) . The synthetic gene was digested with the restriction enzymes KpnI and BamHI . The released MDH gene was cloned into the pJST4 vector that has been previously linearized with KpnI and BamHI restriction enzymes . The construction of the pcontrol plasmid , which was used in this investigation , has been previously described [41] . For complementation of the mutated E . coli strain JW3205 with E . coli MDH , E . coli MDH gene was amplified from E . coli genomic DNA using primers 5’ BamHI MDH and 3’ EcoRI MDH ( S8 Table ) and the PCR product was cloned in the pGEM-T easy vector . E . coli MDH was digested with BamHI and EcoRI and then cloned in the vector pGFP ( Genbank accession No: U17997 ) which had been previously linearized with BamHI and EcoRI . For complementation of the mutated E . coli strain JW3205 with E . histolytica MDH , the E . histolytica MDH gene was amplified from a PGEX-EhMDH vector using the primers EhMDH BamHI 5’and EhMDH EcoRI 3’ ( S8 Table ) . The PCR product was cloned in the pGEM-T easy vector . Eh MDH was digested with BamHI and EcoRI and then cloned in the pGFP vector ( Genbank accession number: U17997 ) which had been previously linearized with BamHI and EcoRI . Primers used in this study are displayed in S8 Table . The detection of OXs by OX-RAC was performed using a previously described protocol[22] . Captured proteins were eluted with 30 μl elution buffer which contained 10 mM HEPES , 0 . 1 mM EDTA , 0 . 01 mM neocuproine , 0 . 1% SDS and 100 mM 2-mercaptoethanol for 20 minutes at room temperature . Proteins in a 10-μl aliquot of each eluent were resolved on a 12 . 5% SDS-PAGE gel . Each gel was then stained with silver ( Pierce Silver Stain ) and each gel slice was independently analyzed by mass spectrometry ( MS ) . A protein was considered to be oxidized when its relative amount in the dithiothreitol ( DTT ) -treated lysates was significantly less than that in the DTT-untreated lysates ( p <0 . 05 according to the results of a unpaired t-test ) . In gel proteolysis by trypsin and analysis by LC-MS/MS on Q Exactive plus ( Thermo ) and data analysis with MaxQuant 1 . 5 . 2 . 8 [87] and the Uniprot database as the reference were done using a previously described protocol[22] . The data was quantified by LF analysis using the same software . The identifications are filtered for proteins identified with a false discovery rate of <0 . 01 and at least two identified peptides in the project . The intensities are presented as raw intensities without normalization and as LFQ with normalization , both presented as log2 intensities . The OX-proteins were classified according to their protein class using PANTHER software ( Protein ANalysis THrough Evolutionary Relationships ) Classification System ( http://www . pantherdb . org/ ) [31] . Following the OX-RAC procedure , proteins in a 10 μl aliquot of each eluent were resolved on an 8% SDS-PAGE gel and stained with silver or transferred onto a nitrocellulose membrane ( Whatman , Protran BA83 ) . The blots were first blocked using 3% skim milk , and then probed with 1:500 rabbit polyclonal Gal/GalNAc lectin antibody ( a kind gift from Dr . N . Guillen , Pasteur Institute , Paris , France ) for 16 hours at 4°C . After incubation with the Gal/GalNac lectin antibody , the blots were incubated with 1:5000 secondary rabbit antibody for one hour at room temperature ( Jackson ImmunoResearch ) , and then developed by enhanced chemiluminescence . The detection of oxaloacetate in E . coli and in E . histolytica was done using a commercial enzymatic-based kit ( Abcam , Zotal , Israel ) . Oxaloacetate ( 0 . 25 mM ) and H2O2 ( 3 . 5 mM ) in 10 mM phosphate buffer , pH 7 . 4 , were mixed together at 25°C and the concentration of H2O2 was determined by UV spectrophotometric analysis according to a previously described protocol [47] on a UV spectrophotometer NanoDrop 2000c ( Thermo Fisher Scientific , USA ) . E . histolytica trophozoites were suspended in 2 ml ice cold methanol and transferred into a 4-ml tube which contained glass beads ( 0 . 10–0 . 11-mm diameter ) ( Sartorius AG ) . Cell disruption was performed by using a FastPrep-24 instrument ( MP Biomedicals , LLC ) twice for 40 seconds each at 6 . 0 m/s . Methanolic cell extract was transferred to 15-ml tubes after centrifugation for five minutes at 4°C . Cell debris and glass beads were washed twice with 1 ml ultrapure water as a second extraction step . The aqueous and methanolic cell extracts were combined . An aliquot ( 0 . 4 ml ) of chloroform was added to the cell extracts and the suspension was vortexed and shaken five times for ten seconds . For separation of the aqueous and organic layers , the samples were stored at -20°C for ten minutes . After centrifugation for five minutes at 4°C and 10 , 015g , the upper layer was transferred to a 50-ml tube , diluted with water , and stored at -80°C for lyophilization . Derivatization of lyophilized samples was done using a previously described protocol[88] . Analysis was performed by using an Agilent 7890B GC system with an autosampler ( model G4513A ) , and a coupled mass selective detector ( model 5977B MSD ) ( Agilent ) . The 2-μl injection volume of the sample was split 1:10 at 250°C with an inlet split flow of 10 ml/min . Helium was used as the carrier gas at a pressure of 8 . 8 lb/in2 . Chromatographic separation on a 30-m HP 5-ms column ( Agilent Technologies ) with a 0 . 25-mm inner diameter and a 2 . 5-μm film thickness was performed at a constant gas flow of 1 ml/min . The oven program started with an initial temperature which was held at 70°C for one minute , continued at a heating rate of 1 . 5°C/minute up to 76°C , followed by heating at 5°C/minute up to 220°C , and 20°C/minute up to 325°C , with a hold time of eight minutes . The analytes were transferred to the mass selective detector via the transfer line at 325°C and ionized by electron impact ionization at 230°C . After a solvent delay of six minutes , mass spectra were acquired using a quadrupole temperature of 150°C and SIM acquisition mode . , The selected quantifier ion for glycerol was m/z 205 . 1 and for isocitrate was m/z 245 . 1 . Data analysis was done by using MassHunter Workstation software Quantitative Analysis 8 . 0 . The area of the quantifier ion of each metabolite was integrated and normalized to the area of the quantifier ion of one internal standard ( glycerol and isocitrate were normalized to p-chlorophenylalanine ) . This ratio represents the relative metabolite amount and was normalized to the protein content of the sample . | Entamoeba histolytica is a unicellular parasite which infects millions of humans worldwide via contaminated food and water . It resides in the colon and most infected individuals are asymptomatic . In some people , the parasite can spread into both intestinal and extraintestinal tissues , and results in amebiasis . E . histolytica feeds on bacteria in the colonic microflora . Since changes in the composition of the colonic microflora coincide with the onset of symptomatic amebiasis in affected individuals , a potential role of this microflora in disease manifestation has been suggested . Some modulating effects of the intestinal bacteria and the nature of the native intestinal flora on amebic virulence have been reported , but the exact mechanisms have not been described . Here , we report that E . coli confers increased resistance against oxidative stress to the parasite via the production of oxaloacetate . This antioxidant metabolite is the result of the oxidation of malate by malate dehydrogenase . Our results indicate that E . histolytica can use oxaloacetate of bacterial origin to increase its resistance against oxidative stress and that this oxaloacetate promotes its survival in the large intestine of mice with experimentally-induced amebiasis . We also present evidence that oxaloacetate can protect Caenorhabditis elegans , another bacteria-feeding organism , against oxidative stress . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"trophozoites",
"parasite",
"groups",
"medicine",
"and",
"health",
"sciences",
"oxidative",
"stress",
"caenorhabditis",
"tropical",
"diseases",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"animals",
"parasitology",
"animal",
"models",
"apicomplexa",... | 2018 | Escherichia coli mediated resistance of Entamoeba histolytica to oxidative stress is triggered by oxaloacetate |
Perception is fundamentally underconstrained because different combinations of object properties can generate the same sensory information . To disambiguate sensory information into estimates of scene properties , our brains incorporate prior knowledge and additional “auxiliary” ( i . e . , not directly relevant to desired scene property ) sensory information to constrain perceptual interpretations . For example , knowing the distance to an object helps in perceiving its size . The literature contains few demonstrations of the use of prior knowledge and auxiliary information in combined visual and haptic disambiguation and almost no examination of haptic disambiguation of vision beyond “bistable” stimuli . Previous studies have reported humans integrate multiple unambiguous sensations to perceive single , continuous object properties , like size or position . Here we test whether humans use visual and haptic information , individually and jointly , to disambiguate size from distance . We presented participants with a ball moving in depth with a changing diameter . Because no unambiguous distance information is available under monocular viewing , participants rely on prior assumptions about the ball's distance to disambiguate their -size percept . Presenting auxiliary binocular and/or haptic distance information augments participants' prior distance assumptions and improves their size judgment accuracy—though binocular cues were trusted more than haptic . Our results suggest both visual and haptic distance information disambiguate size perception , and we interpret these results in the context of probabilistic perceptual reasoning .
For well over a century [1] , [2] psychologists have considered the question of how the brain uses visual angle sensations to make judgments of an object's size , overcoming the confounding effect of its distance - but the topic remains unsettled . Holway and Boring [3] found that when strong sensations of an object's distance were made available , human size matching performance at different distances was high , but when distance sensations were removed human perception of an object's size was erroneously dominated its visual angle . Epstein et al . [4] surveyed literature regarding the “size-distance invariance hypothesis” [5] , which holds that retinal visual angle constrains perception of an object's size and distance such that their ratio holds a constant value ( e . g . doubling an object's physical distance while hold its retinal image size constant causes its perceived size to double ) , and concluded the size-distance invariance hypothesis was subject to a variety of failures . Several studies attributed participants' mistaken size perceptions [4] , [6]–[12] to misjudgments of physical distance , while others point out that specific experimental design choices and task demands contribute to reported failures of size constancy [13]–[16] . Recently Combe and Wexler [17] reported that size constancy is stronger when the relative distance between observer and object varies due to observer motion , than when due to object motion . Such findings highlight the unsettled state of current empirical knowledge about human size and distance perception , which is exacerbated by the absence of a unified theoretical account for normative size/distance perception . We hypothesize that the brain makes size inferences by incorporating multiple sensations based on knowledge of their generative relationship with physical environment properties , and that failures like inaccuracy and systematic biases are due to poverty , unreliability , and/or mistrust , of observed sensations . Our experiments tackle the issue of how the brain incorporates distance information , in particular binocular and haptic ( touch ) , to jointly perceive of how an object's size is changing . Size-change perception , which surprisingly has not been studied in the size/distance perception literature , bears close similarity to static size perception because size-change judgments based on retinal image size are ambiguous if information about the object's motion-in-depth is unknown . However when auxiliary sensations indicating motion-in-depth are available , an observer may rule out size-change/motion combinations that are inconsistent with the auxiliary sensations , and unambiguously infer whether the object is inflating or deflating . We predicted that despite the inherent novelty of the stimuli ( i . e . objects do not typically change in size while moving in depth ) , participants' abilities to discriminate whether an object inflated or deflated would depend on the availability and quality of information about its motion-in-depth . Because binocular and haptic sensations provide information about depth , we predicted that they would each be incorporated for improving size-change judgments . Thus our study answers two key questions: 1 ) Does the brain use distance-change information for size-change perception ? 2 ) What are the roles of binocular and haptic distance-change information ? Our size-change discrimination task ( Figure 1 ) presented participants with an object that either inflated or deflated while simultaneously either approaching or receding , and asked them to discriminate whether it inflated or deflated ( Figure 2 ) . Most static size perception tasks use matching paradigms , and our task was advantageous because it allowed us to present a single stimulus per trial , and avoid issues regarding relative comparison of pairs of stimuli . We provided participants with different types of auxiliary motion-in-depth information , binocular [3] , [16] , [17] and haptic [18] , [19] , both in isolation and simultaneously , and examined their inflation/deflation judgments to evaluate how auxiliary distance information influenced perceived size-change . Evidence for the use of binocular and haptic distance information in size-change perception has not been reported , and previous studies of cue integration [20] suggest the brain combines haptic and binocular information in proportion to its reliability to jointly improve spatial perception . We found that when distance-change information was absent , participants' size-change judgments closely matched object's image size-change . However , when we provided participants with auxiliary distance-change sensations , participants incorporated this additional information to form more accurate size percepts that were consistent with both monocular and auxiliary sensations . Moreover when both binocular and haptic information was presented , most participants showed greater disambiguation of size than when either was presented in isolation . These results suggest size-change perception uses knowledge of how multi-modal size and distance sensations are related to interpret the scene . We interpret these findings in the framework of probabilistic perceptual inference , in which available sensations are combined according to their relationship to scene properties and their respective reliabilities [21] , [22] .
Experiment 1 contained four distance-cue conditions ( H−/B− , H+/B− , H−/B+ , H+/B+ ) that provided the four possible combinations of the presence ( + ) or absence ( − ) of haptic ( H ) and binocular ( B ) cues to the ball's distance-change . Haptic cues include proprioceptive and pressure information generated by the ball's movement in depth , and binocular cues include vergence and relative retinal disparity information that gives direct information about the ball's trajectory ( see Methods and Text S1 ) . Figures 2A–B show grids on which we plot the ball's size- and distance-change rates for all stimuli ( black dots ) . The diagonal dashed line divides the stimuli into those in which the ball's image size increases ( lower-right ) versus decreases ( upper-left ) in size , and the vertical dotted line divides the stimuli into those in which ball's physical size inflates ( right ) versus deflates ( left ) . Our specific analysis and results are as follows . We separated balls' distance-change rates into three distance-direction groups: receding , intermediate , and approaching ( colored lines , Figures 2A–B ) . For each group we fit individual psychometric functions ( cumulative Gaussian ) , where the height of the function at a particular size-change rate indicates the percentage of trials the participant judged “inflating” . Figures 2C–D depict the results for one participant corresponding to the distance-direction group in Figures 2A–B . Figures 2E–F illustrates the relationship between the psychometric function fits and the shaded regions in Figures 2A–B . Within each distance-cue condition , we found each psychometric function's 50% point , and fit a line between these points . We termed these best-fit lines participants' discrimination boundaries between “inflating” and “deflating” responses , and interpreted them as measures of participants' confusion . Specifically , we computed the best-fit slope with respect to distance-change rate ( y-axis ) , and normalized it into a confusion ratio . A confusion ratio of 1 meant the participant discriminated inflation from deflation depending exclusively on the sign of the image size-change rate , which corresponded to the locus of physical distance- and size-change rates that produced an image-change rate of 0 ( diagonal line , Figure 2A ) . A confusion ratio of 0 meant the participant discriminated inflation from deflation depending on the sign of the physical size-change rate ( vertical line , Figure 2B ) . Simply put , when a participant's discrimination judgments were independent of the nuisance distance property they did not confuse distance-change for size change ( zero confusion ) , and when their discrimination judgments were dependent on the nuisance distance property they confused distance-change with size change ( confusion of 1 ) . Our “confusion ratio” is related to the Brunswick and Thouless ratios , which apply to static size matching tasks [23] . Notably , those ratios scale inversely to ours: they take values of 1 when participants comparison size judgments match the standard stimulus size ( confusion of 0 ) , and 0 when the comparison size judgment matches the image size ( confusion of 1 ) . In the trials that contained no distance cues ( H−/B− ) , we predicted participants would rely on prior assumptions that the ball tends to stay still ( or move slowly ) . This is a sort of motion analog to the “specific distance tendency” [10] . Slow movement priors have previously been reported for 2D motion perception [24]–[26] and others [27] find similar priors in 3D [28] . Assuming slow , or no , movement would bias participants to attribute increasing image size largely to inflation and in turn lead them to judge stimuli with increasing image sizes as “inflating” ( shaded grey in Figures 2A–B ) . All participants display precisely this pattern; Figure 3 ( top-left box ) shows the specific pattern for a typical participant ( 5 ) in the H−/B− condition , and Figure 4 summarizes all participants ( white bars ) . The evidence suggests that participants used prior assumptions that objects tend to stay at rest to disambiguate the scene . But because the ball was often approaching or receding , these often-incorrect prior assumptions led to erroneous perceptual size judgments . However , if we had allowed participants to decide whether the ball was changing size or changing distance , they may have preferred changing distance in some cases - it may be that the role of the prior is guided by the task's demands . In those conditions that contained auxiliary distance-change cues ( H+/B− , H−/B+ , H+/B+ ) , we predicted participants would perceive trials with increasing image size as “inflating” ( shaded regions in Figure 2B ) only when the ball's movement in depth could not account for the changing image size; in other words , the participant will not perceive a rapidly approaching ball as inflating if the image size is only increasing a small amount . Likewise , when the ball's image size was decreasing , we predicted participants would perceive the ball as deflating only when the recession rate was not great enough to account for the image size change . All participants exhibited this pattern when the auxiliary binocular cue was present ( H−/B+ and H+/B+ ) , and 7 of 10 also showed size disambiguation when the haptic cue alone was present ( H+/B− ) ; again , Figure 3 ( bottom-left , and right boxes ) shows the specific pattern for a typical participant ( 5 ) in the H+/B− , H−/B+ , and H+/B+ conditions , and Figure 4 summarizes all participants ( grey bars ) . These results indicate that participants disambiguate the scene using both haptic and binocular distance-change cues , by augmenting their prior assumptions to make more accurate inflation discriminations . Figure 4 presents confusion for all participants in all distance-change conditions . A two-way , repeated-measures ANOVA found a significant reduction of confusion across participants for both haptic ( F ( 1 , 9 ) = 17 . 42 , p<0 . 005 ) and binocular ( F ( 1 , 9 ) = 212 . 5 , p<0 . 0001 ) distance-change cues , and no significant interaction ( F≈0 , p>0 . 05 ) ( though the fact that the binocular cue almost fully disambiguated the inflation/deflation rate for most participants means any interaction effect would be masked by the ceiling ) . Our results indicate participants use binocular distance-change cues significantly more than haptic cues for disambiguating the scene and improving physical size judgments ( H+/B− vs . H−/B+ conditions compared in a paired sign test , p<0 . 002 ) . Previous cue combination studies [20] , [29]–[38] have demonstrated integration of cues in proportion to their relative reliabilities . If each auxiliary cue , binocular and haptic , was trusted by the observer to provide information about the ball's distance-change , we hypothesized that their disparate explaining-away effects were due to the binocular cues' greater reliability over the haptic cues' . We examined whether Experiment 1's binocular/haptic discrepancy was due to differences in haptic and binocular cue reliabilities in Experiment 2 . We measured the haptic and binocular cues' noise ( see [39] ) to determine whether differences in their respective reliabilities could explain their discrepant effects on disambiguating the balls' inflation/deflation rates in Experiment 1 . Participants observed two moving balls sequentially , and judged which ball moved faster , in a two-interval forced choice ( 2IFC ) discrimination task . Experiment 2 used binocular and haptic cues in different conditions , so we could measure their respective reliabilities in isolation . The ball's movements were always restricted to the depth axis ( with slight fronto-parallel oscillation described in the Methods ) as in Experiment 1 , and also spanned the same speed range as Experiment 1 . In the haptic condition , the ball was not visible during the stimulus interval; in the binocular condition the ball was visible and its image size changed under accurate perspective projection ( see Methods for details ) . Our results show that with the exception of one participant , the haptic and binocular cue reliabilities do not explain their differential uses in Experiment 1 . Figure 5 shows the haptic and binocular distance-change noise magnitudes for each participant , where each pair of bars represents the haptic and binocular noise magnitudes ( standard deviation ) for a participant . Qualitatively it is clear that the binocular and haptic noises have comparable magnitudes . By comparing the set of bootstrap-resampled binocular and haptic noise magnitudes , we can perform a hypothesis test of the prediction that the binocular noise is less than the haptic noise . All participants fail this test ( p>0 . 05 ) , except participant 9 ( p<0 . 05 ) . Thus , differences in cue reliabilities cannot explain Experiment 1's discrepant use of binocular and haptic cues to reduce confusion . This effect is consistent with the observer trusting the binocular cue more greatly than the haptic , thus integrating less of the haptic cue information .
Our study finds that humans use within- ( binocular ) and cross-modal ( haptic ) distance-change sensations to disambiguate otherwise ambiguous monocular image size sensations , resulting in more accurate judgments of object size . Binocular distance-change cues influenced participants' size judgments more strongly than haptic cues . When both modalities' distance-change cues were presented simultaneously , nine of ten participants' physical size judgments were virtually confusion-free . In order to use the distance-change to improve size-change judgments , the brain must use generative knowledge of how an object's physical size and distance cause monocular image size- and distance-change cues to alleviate the confounding effects of physical distance-change . Such knowledge may be abstractly represented ( the laws of physics ) or encoded in a more applied manner ( a look-up table relating size , distance , and image cues ) . This is consistent with a core feature of Bayesian reasoning termed explaining-away [40] . Knowledge about the relationships between world properties and sensations provides perceptual inference processes with a common representation for integrating prior knowledge with sensory evidence , and probabilistically “solving for” scene properties based on sensations . Bayesian reasoning as a framework for interpreting perceptual behavior has attracted considerable attention because it provides a principled theoretical framework for describing the brain's recovery of scene properties from sensations [22] , [41] and has allowed quantitative confirmation that humans exhibit near-optimal perceptual performance across many tasks [20] , [22] , [29] , [33] , [35]–[38] . Various studies have found that when humans judge single scene properties that produce multiple pieces of sensory information , or cues ( Figure 6A ) , they average the cues in proportion to their reliability [25] , which is the Bayes'-prescribed perceptual strategy . Others report [42]–[44] perceptual “discounting” , in which prior knowledge is used to disambiguate otherwise ambiguous sensory cues , which requires knowledge of the generative relationship between a cue and the scene properties that cause it . Our study examines a more complex situation ( Figure 6B ) where , unlike discounting [42]–[44] ( Figure 6A ) , correct inference of the desired scene property ( physical size-change ) requires an inference strategy that exploits generative knowledge of the relationships between multiple scene properties ( physical size-change and physical distance-change ) and multiple sensations ( retinal image size-change , binocular and haptic distance-change cues ) . No single sensation alone , retinal image size-change or distance-change cue , constrains the physical size-change inference uniquely due to the confounding influence of nuisance scene properties – properties that affect sensations but do not contribute to the judgment - in this case , physical distance-change ( Figure 6B ) . Because the nuisance property ( physical distance ) confounds the direct cue ( retinal image size-change ) to the desired property ( physical size-change ) , incorporating auxiliary cues ( distance-change sensations ) can explain-away the influence of the nuisance physical distance-change and allow unambiguous judgments of physical size-change . Explaining-away can characterize other perceptual tasks in which multiple scene properties influence multiple cues in the manner depicted by Figure 6B; for example , estimating surface reflectance from sensed lightness despite the confounding influence of illumination [45] , estimating object shape from image contours despite the confounding influence of pose , and the general class of “perceptual constancy” effects . Also explaining-away is a general Bayesian perspective on a specialized concept [46] termed “cue promotion” - in which a relative cue ( like stereoscopic disparity ) is able to be incorporated into perceptual judgments ( promoted ) only because a second , auxiliary cue ( like depth from vergence ) provides information to make it an absolute cue . Many unimodal perceptual phenomena are characteristic of explaining-away [3] , [47]–[48] . Multimodal perceptual explaining-away is less documented , but explaining-away in bistable percepts has been reported [49]–[52] as well as in continuous percepts [53]–[54] . Our results extend previous reports of explaining-away to include continuous , multimodal scene property judgments [47] , [50]–[51] , [53] , [55]–[56] . Explaining-away is only appropriate when the auxiliary cues are dependent on scene properties that influence cues to the desired scene property . This typically occurs when the nuisance variable causes the auxiliary cue . There is evidence suggesting that non-visual sensory cues are integrated less efficiently than their reliabilities afford [33] or in a less committed , reversible manner [39] , [57]–[58] , and some have attributed lack of cue integration to weak conditional dependency between cues and world properties [31] , [58]–[62] . Reliability reflects the quality of a cue; if the sensory signal is corrupted by noise the reliability decreases . Trust reflects the degree to which the observer believes the cue is related to the desired scene property; there may be other scene properties that influence the auxiliary cue which diminishes the cue's diagnosticity for the desired scene property . In cases in which all auxiliary cues are trusted equally , they should be integrated in proportion to their relative reliabilities only . However , if trust in the auxiliary cues is unequally distributed they should integrated in proportion to the relative reliabilities and their trust . Previous studies that tested multisensory disambiguation of bistable stimuli reported mixed results [50]–[51] . It is possible that these different findings are due to non-visual cues being trusted less due to their frequent independence from visual cues . Alternatively the mixed results may be due to variable cue reliabilities [63] , for instance when visual cues to a bistable stimulus's structure vary in relative reliability compared with tactile cues , tactile cues may influence perceived structure in proportion to their reliability . Our experiment was sensitive to partial disambiguation , because participants discriminated percepts that lied on a continuous axis ( rate of distance-change ) , which may reconcile previous mixed results by demonstrating the graded roles of auxiliary cue information . We found different effects of individual haptic and visual cues , and strongest influence when both were present , which argues for the reliability-weighted integration of that information . One potential reason that binocular distance-change cues were more useful than haptic cues for disambiguating size perception in our experiment may be that the haptic cue is more weakly coupled with the image cue than the binocular cue , perhaps reflecting the causal structure of the world . In decoupled situations , in which different world properties influence different cues independently , it is inappropriate to combine cues . For instance , in natural settings binocular depth and monocular image size cues are transmitted to the eyes by the same light patterns , thus are usually highly dependent . Because , sensory channels for visual and haptic information differ , and there are many situations in which the felt position of an object differs from its visual position , like manipulating a tool , playing with a yo-yo , or touching an object that is occluded by a nearer object . In our experiment the haptic cue was somewhat atypical , because we forced the fingertip to always be positioned at the center of the ball , not the edge , so the size-change would not be directly measurable by radial pressure toward or away from the ball's center . It is plausible that this atypicality degraded participants' belief that haptic and visual cues were caused by the same object . Recent reports of visual-auditory cue integration have found causality-modulated cue integration [59]–[62] , and it may explain why the haptic cue is trusted less for disambiguation compared with the disparity cue in our experiment . Another possibility derives from the brain's algorithm used to compute the size-change rate . Per Rushton and Wann ( [64] , Figure 1 caption ) , the B+ conditions allow the possibility of estimating the size-change rate without explicitly estimating the distance-change rate ( by computing the ratio between image-size-change and binocular vergence angle-change rates , which causes the explicit distance-change rate terms to cancel ) . This means that a potential source of noise in the B+ conditions , incurred during estimation of the distance-change rate , would be removed , allowing higher fidelity disambiguation of the size-change rate in those conditions . If this were the case , Experiment 2 may have overestimated the effect of noise in Experiment 1's B+ conditions depending on how noise enters the system: if noise only corrupts the brain's estimates of binocular vergence angle-change rates , then Experiment 2's binocular noise estimates are valid . However , if noise additionally corrupts the ability to make binocular distance-change judgments , then Experiment 2's binocular noise estimates would be overestimates of the true noise afflicting Experiment 1's B+ conditions . This logic may be moot if the distance-change is used to drive oculomotor vergence dynamics ( i . e . tracking in depth ) because in that case the noisy distance-change rate would influence the binocular vergence-change rate . Either way , in order to apply the ratio algorithm [64] for computing size-change still requires the brain to understand the generative relationships among size , distance , and the image and binocular sensory cues , which does not diminish our findings . One future challenge is directly assessing what prior assumptions the perceptual system has about the world , and how reliability and trust in various cues are learned [63] . With quantitative estimates of prior assumptions , one can predict how reliable auxiliary cues must be and how much they should be trusted , to override conflicting priors . Other studies [5] refer to a “specific distance tendency” in which participants assume objects appear at a canonical distance . In the 2D motion perception domain and [24] , [26] each reported that humans exhibit strong prior preferences for “slow and smooth” movement , and our study suggests participants assume objects move slowly in 3D , but a stronger direct test of 3D motion priors requires quantitative predictions . Measuring prior knowledge directly is difficult , but developing indirect methods is an important topic of recent and continuing research [26] . Our results indicate that the brain uses multisensory distance-change cues to improve perceptual size-change disambiguation . Haptic and binocular distance-change cues are both effective , binocular more than haptic , which is not explained by their relative reliabilities , but is consistent with causal cue integration models [61]–[62] . Our findings support the view that perceptual processing employs knowledge of the sensory generative process to infer scene properties and disambiguate competing interpretations .
Experiments were undertaken with the understanding and written consent of each subject , with the approval of the Ethik-Kommission der Medizinischen Fakultät und am Universitätsklinikum Tübingen , and in compliance with national legislation and the Code of Ethical Principles for Medical Research Involving Human Subjects of the World Medical Association ( Declaration of Helsinki ) . 11 right-handed participants ( ages 18 to 35 ) with normal/corrected-to-normal vision ( Snellen-equivalent of 20/25 or better ) and normal stereopsis ( 60 s of arc or better - Stereotest circles; Stereo Optical , Chicago ) were recruited from MPI Tuebingen's Subject Database and compensated 8 €/h . All participants completed both Experiments 1 and 2 , with the exception of one who was excluded from reported results because her responses indicated she did not follow the experimenters' instructions . Participants sat in a virtual workbench that presented both graphical and haptic stimuli ( Figure 1; see [20] for details ) . Participants' heads were stabilized with a chin-and-forehead rest 45 deg forward . Visual stimuli were presented on a monitor ( 21″ GDM-F500R SONY , 38 . 2×29 . 8 cm , resolution of 1280×1024 pixels , refresh rate 100 Hz ) whose center was 50 cm from the eyes reflected on a first-surface mirror , and whose top was tilted 22 deg backwards from the fronto-parallel plane . Binocular stimuli were presented through CrystalEyes TM ( StereoGraphics ) liquid-crystal shutter glasses which allowed different images to be presented to each eye . Haptic stimuli were presented using a Premium PHANToM force-feedback device ( SensAble Technologies ) , to which the index finger was attached by a thimble and elastic band , allowing six degrees of freedom movements . The 3D fingertip position was monitored continuously , and the computer applied simulated normal forces when the tip reached the positions of the virtual haptic objects . The apparatus was calibrated to spatially align the visual and haptic stimuli , simulating a single scene . There were two experiments , 1 . Distance cue disambiguation for size perception and 2 . Distance cue reliability , that each contained haptic and binocular distance cues . At the start of each trial , a 35 mm diameter red ball was placed between 443 mm and 455 mm from the observer ( 4 . 4–4 . 5 deg visual angle ) . In trials containing a binocular distance cue , the ball was presented binocularly to the observer's two eyes , rendered to simulate an interocular distance of 58 mm . The participant signaled he or she was ready to begin the trial by reaching and contacting the ball with the index finger ( attached to the PHANToM device ) . Once contact was made , the PHANToM device applied forces to the fingertip to guide it to the center of the ball . At this point the experimental phase of the trial began: the ball began moving in depth with respect to the participant , while simultaneously changing in size , for a duration of 1000 ms . If the trial contained a haptic distance cue , as the ball moved appropriate forces were applied to the fingertip to maintain its position at the center of the ball; otherwise no forces were applied to the fingertip once the ball began to move and participants typically held their fingertips at a roughly constant position . The ball also slightly oscillated in the observer's fronto-parallel plane following a sinusoidal displacement ( with amplitude between 5 . 0 and 15 . 0 mm ) in a random direction and at a random frequency ( between 0 . 35 and 0 . 5 Hz ) . This was intended to both decrease the similarity of the visual and haptic trajectories across trials , increase their perceptual fusion , as well as obviate local edge motion information as a direct indicator of image size-change . Although fixation was not precisely controlled or monitored , our experience and observations of participants suggested they fixated the ball in monocular and binocular conditions . Also , our stimuli were constructed to eliminate two potential sources of size-change information from binocular cues . One source is “Da Vinci” stereopsis , which refers to depth information that results from points on the object that are visible in only one eye due to object self-occlusion . This cue requires identifying object points without correspondences between the eyes . Because the ball has no horizontal luminance/color contrast , Da Vinci stereopsis was eliminated as a cue to size-change . A second potential source of binocular size-change information was disparities due to the ball's oscillation . For a ball in the mid-sagittal plane there are no binocular disparity cues to size change . We determined that the slight oscillatory movements the balls made out of the mid-sagittal plane created sub-threshold ( undetectable ) relative disparity cues to ball size . See Text S1l for an in-depth examination and schematic of the binocular cue . Lastly , accommodation was a potential cue , uncontrolled except that the screen depth was fixed . After 1000 ms , the ball disappeared . In Experiment 1 , only a single stimulus interval was presented . In the Experiment 2 , two stimulus intervals were presented; following the first interval a new ball appeared and the second interval proceeded just as the first . Once the stimulus interval ( s ) were finished , two buttons appeared on the left side of the scene and participants were instructed to press the button that corresponded to his or her judgment of the scene . The trial ended once the button was pressed , and the subsequent trial began immediately . In Experiment 1 the buttons were labeled “inflating” and “deflating” , and the participant pressed the button corresponding to his or her perception of the ball's physical size change . We interpreted participants' choices as their discriminations of the ball's absolute size-change rate . In Experiment 2 , each trial was designed as two-interval forced-choice ( 2IFC ) . In every trial , both balls moved in the same direction with respect to the participant ( approaching/receding ) , but their speeds were different relative to each other . Also , the balls never changed in size ( equivalent to 0 mm/s size-change rate in the main experiment ) . In haptic trials , the ball disappeared from view as soon as it began to move . Following the two intervals participants were instructed to press one button among two choices , labeled “1st” and “2nd” , indicating which interval contained the faster ball . All confidence intervals were estimated by nonparametric bootstrapping [65] , comparable to those used by [66]–[67] . Error bars on some figures were computed using the “median absolute deviations with finite sample correction factors” ( MADC ) from the LIBRA Robust Statistics toolbox for Matlab [68] . MADC approximates standard deviation estimates of the mean of the sample for normally-distributed data , but it is more robust for skewed and kurtotic distributions . | To perceive your surroundings your brain must distinguish between different possible scenes , each of which is more or less likely . In order to disambiguate interpretations that are equally likely given sensory input , the brain aggregates multiple sensations to form an interpretation of the world consistent with each . For instance , when you judge the size of an object you are viewing , its distance influences its image size that projects to your eyes . To estimate its true size , your brain must use extra information to disambiguate whether it is a small , near object , or large , far object . If you touch the object your brain could use the felt distance to scale the apparent size of the object . Cognitive scientists do not fully understand the computations that make perceptual disambiguation possible . Here we investigate how people disambiguate an object's size from its distance by measuring participants' size judgments when we provide different types of distance sensations . We find that distance sensations provided by viewing objects with both eyes open , and by touching the object , are both effective for disambiguating its size . We provide a general probabilistic framework to explain these results , which provides a unifying account of sensory fusion in the presence of ambiguity . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"neuroscience/sensory",
"systems",
"computational",
"biology/computational",
"neuroscience",
"neuroscience/psychology",
"neuroscience/experimental",
"psychology",
"neuroscience/natural",
"and",
"synthetic",
"vision"
] | 2010 | Within- and Cross-Modal Distance Information Disambiguate Visual Size-Change Perception |
Human T-lymphotropic virus type 1 ( HTLV-1 ) has worldwide distribution and is considered endemic in many world regions , including southwestern Japan and Brazil . Japanese immigrants and their descendants have a high risk of acquiring this infection due to intense population exchange between Brazil and Japan . This cross-sectional study aimed to estimate the prevalence of HTLV , analyze the main risk factors associated with this infection , identify the main circulating types and subtypes of HTLV in Japanese immigrants and descendants living in Campo Grande-MS ( Middle-West Brazil ) , as well as analyze the phylogenetic relationship among isolates of HTLV . A total of 219 individuals were interviewed and submitted to blood collection . All collected blood samples were submitted for detection of anti-HTLV-1/2 using the immunoassay ELISA and confirmed by immunoblot method . The proviral DNA of the 14 samples HTLV- 1 positive were genotyped by nucleotide sequencing . The overall prevalence of HTLV-1 was 6 . 8% ( IC 95%: 3 , 5-10 , 2 ) . Descriptive analysis of behavioral risk factors showed statistical association between HTLV-1 and age greater than or equal to 45 years . The proviral DNA of HTLV-1 was detected in all HTLV-1 positive samples . Of these , 14 were sequenced and classified as Cosmopolitan subtype , and 50% ( 7/14 ) belonged to subgroup A ( transcontinental ) and 50% ( 7/14 ) to the subgroup B ( Japanese ) . The high prevalence of HTLV-1 found evidence of the importance of early diagnosis and counseling of individuals infected with HTLV-1 for the control and prevention of the spread of this infection among Japanese immigrants and their descendants in Central Brazil .
The retrovirus human T-lymphotropic virus type 1 ( HTLV-1 ) is associated with many severe diseases , including adult T-cell leukemia/lymphoma ( ATL ) and HTLV-1-associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) , but most infected people remain asymptomatic [1] . Seven genetic subtypes have been defined ( a-g ) based on analyses of the HTLV-1 long terminal repeat ( LTR ) region . The Cosmopolitan subtype ( 1a ) is the most widespread [2] . Approximately 10 million people are estimated to be infected with HTLV-1 throughout the world [2] . A high prevalence of HTLV-1 infection can be found in the endemic regions of equatorial Africa , the Caribbean islands , Japan , Colombia , northeast Australia , Papua New Guinea and Brazil , that has heterogeneous geographic distribution . However , the highest infection rate has been observed to occur in the islands of Kyushu and Okinawa , in southwestern Japan , and Hokkaido , in north of Japan , with approximately 1 . 1 million of infected individuals [3 , 4] . In Brazil , HTLV-1 was first described in 1986 among Japanese immigrants from Okinawa , Southern Japan , residing in the city of Campo Grande , state of Mato Grosso do Sul . Prevalence rates of 13% in the immigrants and 8% in their descendents were observed [5] . Since confirmatory tests for the diagnosis of HTLV infection were not available in the 80s , the prevalence found by Kitagawa and cols ( 1986 ) could be overestimated due to the presence of false positives . Therefore , considering the lack of regional studies on HTLV infection , the Japanese immigration wave to Brazil , particularly to Mato Grosso do Sul state and the risk of intrafamilial transmission of HTLV-1 , the main goal of this study was to revisit the situation of HTLV-1 epidemiology , especially its estimated prevalence and molecular characterization around 30 years after the first published epidemiological study in Japanese immigrants and their descendants living in Campo Grande , MS .
This cross-sectional study was conducted between April 2012 and October 2013 in the city of Campo Grande , capital of the state of Mato Grosso do Sul in midwestern Brazil , which has about 786 , 797 inhabitants with an estimated contigent of 20 , 000 Japanese-Brazilian descendants from 4 , 000 families , showing high migration of this population to Campo Grande [6 , 7] . In order to get a statistical power of 80% ( β = 20% ) , a significance level of 95% ( α <0 . 05 ) and an accuracy of 4% , the calculated sample size to estimate the prevalence of HTLV-1 infection , should include at least 216 individuals , based on the prevalence of 10 . 0% for anti-HTLV-1/2 found by Kitagawa et al . ( 1986 ) . Two hundred and twenty two potential subjects met the sample criteria . Of them , 219 ( 88 . 6% ) agreed to participate . Therefore , the study population included 219 participants , who were from , approximately , 70 families in a population of about 18 , 000 Japanese immigrants and Japanese descendants living in Campo Grande-MS . Individuals were eligible to participate if they were older than 18 years of age , Japanese immigrant ( born in Japan ) , Japanese descendant ( had any genetic relationship to a Japanese immigrant ) and non-descendant ( had relationship to a Japanese immigrant only by marriage ) and able to provide written informed consent . Pregnant women were excluded from the study . The presence of other illness or other infections was not used as selection criteria . The participants were previously advised about the campaign through posters and leaflets placed in the Okinawa Association and Brazilian Nipo Association . In addition to anti-HTLV test , free testing for diabetes and blood pressure were offered by the health campaign . The participants did not know their serological profile for anti-HTLV-1/2 . They were informed of the objectives and methods of the study and written informed consent was obtained from all study participants prior to enrollment . Following enrollment , a questionnaire was administered to obtain sociodemographic characteristics and identify risk factors associated with HTLV transmission , including age , gender , immigrant generation , history of residence in Japan , history of blood transfusion , surgery , personal item/sharps sharing , injection drug use , breastfeeding , multiple partners , unprotected sex and sexually transmitted diseases . Blood samples were collected from study participants and serum samples were evaluated by an enzyme-linked immunosorbent assay ( ELISA ) test for the presence of anti-HTLV 1 and 2 antibodies ( MP Diagnostic HTLV-1/2 ELISA 4 . 0—MP Biomedicals , Singapore ) and confirmed by Western Blot ( WB ) assay with a commercial test ( MP Diagnostics HTLV BLOT 2 . 4–Singapore ) . The samples reactive by screening and positive by WB were considered positive for HTLV-1 or 2 infections . DNA was extracted from whole blood samples using the DNA Genomic Purification Kit ( Wizard® Genomic , Promega , Madison , WI , USA ) , according to manufacturer’s instructions . Nested PCR amplified the targeted portion of the 5’LTR of HTLV-I . Reactions were performed as previously described [8] . Internal primers were HFL9 ( AAGGCTCTGACGTCTCCCCCC ) and HFL10 ( TCCCGGACGAGCCCCCAA ) , corresponding to nucleotide positions 124–144 and 779–796 from the ATK-1 sequence . Products of the second PCR reaction with molecular-weight size marker were subjected to electrophoresis on 1 . 8% agarose gel in TAE buffer ( Tris-acetate EDTA ) at 70V for 30 minutes . After being stained with ethidium bromide , gels were visualized on the Kodak Gel Logic 112 photo documentation system . The LTR 672 bp amplicons were purified using illustraTM PCR DNA and Gel Band Purification Kit ( GE Healthcare ) according to the manufacturer’s instructions . The fragments were sequenced using BigDye Terminator Cycle Sequencing Ready reaction Kit and ABI 1373 ( Applied Biosystems , Foster City , CA , USA ) . Nucleotide sequences were aligned and compared with published HTLV-1 sequences from various geographic regions using ClustalX and Mega 5 . 05 . Philogenetic trees were constructed by use of the neighbor-joining ( NJ ) method . The NJ tree was evaluated by bootstrap analyses of 1000 replicates . Data were entered into Epi Info 7 . 1 . 2 . 0 ( Centers for Disease Control and Prevention [CDC] , Atlanta , GA ) , a statistical software package . Statistical analyses were performed using SPSS Statistics Data Editor ( Statistical Package for Social Science , Chicago , Illinois , USA ) . HTLV prevalence was estimated using 95% confidence intervals ( 95% CI ) . Potential risk factors for HTLV infection were assessed in univariate analysis . Chi-square and Fisher’s exact tests were performed for comparison of categorical parameters . Chi-square and Fisher tests were two-tailed . Statistical significance was considered at p≤0 . 05 . The present study was approved by the Ethics Committee on Research Involving Human Subjects of the Federal University of Mato Grosso do Sul ( CEP/UFMS ) , under protocol number 2249 CAAE 0329 . 0 . 0 . 049 . 000–11 . All adult subjects provided written informed consent , and a parent or guardian of any child participant provided written informed consent on their behalf .
A total of 219 individuals , Japanese immigrants or descendants and non-descendants of Japanese immigrants living in Campo Grande-MS were enrolled in this study . Most of those were female ( 61 . 2% ) . Mean age was 54 . 4 years ( range 11 to 101 years ) and the majority ( 69 . 9% ) had 45 years of age or older . The majority ( 88 . 1% ) was born in Brazil and 10 . 9% were born in different regions of Japan ( Okinawa , Tokyo , Yamaguchi , Hokkaido , Aichi and Kumamoto ) and 1 . 0% was from Japanese descendants from Peru . The Brazilian group ( n = 193 ) was composed of 187 ( 96 . 8% ) Japanese descendants and 6 ( 3 . 1% ) non-Japanese descendants . The majority ( 54 . 8% ) of participants reported having a steady sexual partner ( married ) or cohabitating with a sex partner . Regarding education , 74% had completed secondary education or college . The prevalence of HTLV-1 infection was 6 . 8% ( 95% IC: 3 . 5 to 10 . 2 ) . None of the 219 individuals presented anti-HTLV-2 antibodies . None of the positive subjects showed clinical signs compatible with HAM/TSP or ATL , and therefore , were classified as asymptomatic . The frequency of anti-HTLV seropositivity was higher among Okinawan descendants ( 8 . 1% vs . 2 . 1% ) than among descendants from other regions of Japan and non-Japanese descendants; however , this difference was not statistically significant . Among studied population , a frequency of 6 . 1% of HTLV-1 infection was found among Japanese-Brazilian descendants participants , while 12 . 5% was observed in those who declared themselves as foreign , which comprise 22 Japanese and 2 Peruvian nationals . Most infected subjects ( n = 14 ) claimed to be Okinawan descendant , and only one who was not a descendant was positive for anti-HTLV . Risk factors for HTLV infection among the studied population are shown in Table 1 . In the univariate analysis , HTLV-1 infection was associated only with age ≥ 45 years ( P = 0 . 04 ) . The mean age of subjects positive for HTLV-1 was 70 years , which was greater than that of 53 years for non-reactive subjects . Fig 1 shows the variation of age in individuals negative and positive for HTLV-1 . Age of infected participants was concentrated in the seventh decade of life , with only two outliers representing younger infected individuals ( between 40 and 60 years ) . However , the age of uninfected people ranged from 11 to 101 years , mainly in the range of 40 to 70 years . All infected subjects were breastfed as children; however , this risk factor was not statistically significant . Some risk factors were absent or infrequently reported in this population , such as history of tattooing , injection drug use and multiple partners . The proviral DNA of HTLV-1 was detected by amplification of the 5 'LTR region by nested PCR in all samples ( n = 15 ) HTLV-1 positive . Of them , 14 were sequenced and classified as Cosmopolitan subtype and 50% ( 7/14 ) belonged to subgroup A ( Transcontinental ) and 50% ( 7 / 14 ) to the subgroup B ( Japanese ) . The new nucleotide sequence samples described in this study have been deposited in GenBank with accession number as follows: OKW21 ( KM023750 ) , OKW24 ( KM023751 ) , OKW46 ( KM023752 ) , OKW63 ( KM023753 ) , OKW72 ( KM023764 ) , OKW84 ( KM023754 ) , OKW107 ( KM023761 ) , OKW112 ( KM023755 ) , OKW131 ( KM023756 ) , OKW151 ( KM023757 ) , OKW152 ( KM023758 ) , OKW165 ( KM023759 ) , OKW209 ( KM023760 ) , OKW235 ( KM023767 ) were compared with nucleotide sequences of 17 isolates of HTLV-1 available from GenBank ( Fig 2 ) . The GenBank accession numbers for the sequences of the 5 'LTR region of HTLV-1 included in the phylogenetic analysis are: ATK1 ( J02029 ) , H5 ( M37299 ) , pyg19 ( L76310 ) , ITIS ( Z32527 ) , Me3 ( Y16480 ) , CA423 ( EU108724 ) , Mel5 ( Lo2534 ) , K344 ( GQ443755 ) , BRLO14-02 ( JF271836 ) , BRRP438 ( DQ323811 ) , Qu3 ( Y16477 ) , 1066/05 ( HQ606137 ) , 526MZ ( GU194504 ) , Ni2 ( Y16487 ) , efe1 ( Y17014 ) . The HTLV-1aA strains of this study clustered closely with other isolates from Latin America , mainly from Brazil . Further , the HTLV-1aB strains of this study clustered closely with other isolates from Japan and from a Japanese descendant from Peru .
Currently , there are approximately an estimated 1 . 5 million Japanese descendants living in Brazil , whose vast majority reside in the state of São Paulo ( Southwest region ) , followed by the state of Paraná ( South region ) and Mato Grosso do Sul ( Central-West region ) . In Brazil , 10% of the descendants of Japanese immigrants originate from the province of Okinawa , and the second largest Okinawan community in Brazil is in Mato Grosso do Sul State [9] . The population of Okinawan immigrants is considered at risk of acquiring HTLV-1 due to originating from the endemic islands of Okinawa , Japan , where the prevalence of HTLV-1 varies between 14% and 31% [4] . In the present study , 6 . 8% HTLV-1 infection prevalence was observed . This percentage is considered high , indicating there should be greater attention given to HTLV-related diseases in this population . This prevalence is 40 and 52 times higher than those found in prime blood donors ( 0 . 17% ) and pregnant women ( 0 . 13% ) from Campo Grande ( Central-West Brazil ) , respectively , and 13 . 6 times higher than that found in African descendants residing in Central Brazil ( 0 . 5% ) [10 , 11 , 12] . In another study conducted with Japanese immigrants in the city of Tome-Açu in northern Brazil , the prevalence ( 2 . 8% ) was smaller than that found in this study [13] . When comparing the prevalence of HTLV-1 among members of the Japanese community of Campo Grande in the year of 1986 with the current prevalence found in this study , we observed the maintenance of high prevalence over the last 27 years ( 10% vs 6 . 8% ) . This difference was not significant ( P = 0 . 28 ) [5] . Maybe , the prevalence of 6 . 8% was attributed to the improvement in serological and confirmatory tests over the years [14] . In addition , the prevalence of 10% found in the 1986 study may also be due to the higher number of Japanese immigrants ( n = 46 ) investigated when compared to the current study , in which the majority of subjects were descendants of Japanese immigrants and only 24 were Japanese immigrants . Moreover , all participants in the previous study were descendants of Okinawans , and in the present study , 78 . 5% of participants had Okinawan ancestry [5] . Increasing age was significantly associated with HTLV-1 infection . This finding was also reported in previous studies and may be due to the cumulative risk of HTLV-1 infection over the lifetime of individuals surveyed [15 , 16] . The increased prevalence in older individuals may also be a reflection of close kinship to Japanese immigrants from endemic areas , who were likely infected in Japan and brought the virus in the recent past to Brazil . There may have been an age cohort effect resulting from declining HTLV-1 seroprevalence over the past decades . In two studies conducted in Japan , one reported a high seropositivity rate ( above 40% ) in people over 40 years of age and another , more specifically in Okinawa , found a 21% rate of HTLV-1 in the general population over 40 years [17 , 18] . Also , a change in the age distribution of patients with HTLV-1 infection in several studies was observed . Given in the 80s most of the patients were between 50–59 years of age , then in 2007 , the range would be from 60 to 80 years , as encountered in the present study [19 , 20] . Although the association between sex and HTLV-1 infection was not statistically significant in our study , higher prevalence of HTLV-1 infection found among females may be due to male-to-female transmission being more efficient during sexual intercourse [21] . These data is in agreement with the prevalence rates in Japan , which show the same pattern of older age and sex-specific prevalence [22] . Although breastfeeding is known to play an important role in the transmission of HTLV-1 , it was not statistically significant in our study [23 , 24] . In 2003 , the State Coordination of Epidemiological Surveillance of Mato Grosso do Sul determined the notification of every case of a pregnant woman infected with HTLV as well as provided advice to suspend breastfeeding and started to furnish formula milk . These preventative measures were considered important strategies for controlling the spread of the virus by vertical transmission [11] . All of HTLV-1 isolates were classified as Cosmopolitan subtype ( HTLV-1a ) and half of them ( 50% ) belonged to subgroup A ( Transcontinental ) and 50% to subgroup B ( Japanese ) . In Brazil , the Cosmopolitan is the most prevalent subtype of HTLV-1 and the Transcontinental subgroup ( HTLV-1aA ) is the most prevalent among them , followed by the Japanese subgroup ( HTLV-1aB ) , which is detected among Japanese-descendants and immigrants , such as in the present study [25] . In Japan , the existence of subgroups named Transcontinental and Japanese was reported , so that , the geographical distribution of subgroups presents a difference defined according to ethnic group and origin of HTLV-infected individual . In Okinawa region , most infected individuals with HTLV-1 belongs to the Japanese subgroup of HTLV-1a ( 65 . 2% ) [26 , 27] . A study of the population of Japanese immigrants living in Tome-Acu , Para , also found infection with HTLV-1 subgroup Japanese and Transcontinental , the largest part of the Japanese subgroup ( HTLV-1aB ) [13] . The presence of HTLV-1aB was identified in three voluntary blood donation in São Paulo , asymptomatic carriers , two of whom were descendants of Japanese [28] . All individuals who are positive for HTLV-1 infection in this study were asymptomatic , and therefore , mostly likely unaware of their serological profile . Thus , the individual may remain asymptomatic and continue to infect sexual partners and family [29] . Considering the results here , we emphasize the importance of implementing preventive and diagnostic public health policies to decrease the risk of HTLV transmission among family members of Japanese communities throughout Brazil .
The HTLV-1 prevalence of 6 . 8% ( 95% CI: 3 . 5 to 10 . 2 ) found in the present study indicates the maintenance of high prevalence of this infection in the Japanese community of Campo Grande-MS over the past 27 years . These findings also showed that the prevalence of HTLV-1 infection is distinct among regions of Brazil , and although high , this rate did not contribute to an increase in HTLV-1 infection prevalence among the general Brazilian population . Further HTLV-1 nucleotide sequencing may provide more information on the molecular epidemiology of this infection in Japanese descendants population living in Brazil , which may be helpful in understanding the HTLV-1 transmission in these isolated communities . | The population of Okinawan immigrants is considered vulnerable to human T-lymphotropic virus type 1 ( HTLV-1 ) infection because the Okinawa region in Japan is an endemic area . The second Brazilian largest Okinawan community is set in Campo Grande , Middle-West Brazil . This study aimed to estimate the prevalence and risk factors associated with HTLV infection among Japanese immigrants and their descendants living in Campo Grande . The prevalence of 6 . 8% of HTLV-1 infection that was found is considered high . The HTLV-1 infection was associated with age ranged from 45 years old or older . Most infected individuals are Okinawan descendants . The HTLV-1 rate found in the present study indicates that the prevalence of this infection remains high among this Japanese community . This study emphasizes the importance of implementing preventive and diagnostic public health policies to decrease the risk of HTLV-1 transmission among Japanese communities throughout Brazil . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion",
"Conclusion"
] | [] | 2015 | High Prevalence of HTLV-1 Infection among Japanese Immigrants in Non-endemic Area of Brazil |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.