Patterns of epistasis in RNA viruses: a review of the evidence from vaccine design

Feature Learning of Virus Genome Evolution With the Nucleotide Skip-Gram Neural Network

Recent studies reveal that even the smallest genomes such as viruses evolve through complex and stochastic processes, and the assumption of independent alleles is not valid in most applications. Advances in sequencing technologies produce multiple time-point whole-genome data, which enable potential interactions between these alleles to be investigated empirically. To investigate these interactions, we represent alleles as distributed vectors that encode for relationships with other alleles in the course of evolution and apply artificial neural networks to time-sampled whole-genome datasets for feature learning. We build this platform using methods and algorithms derived from natural language processing (NLP), and we denote it as the nucleotide skip-gram neural network. We learn distributed vectors of alleles using the changes in allele frequency of echovirus 11 in the presence or absence of the disinfectant (ClO₂) from the experimental evolution data. Results from the training using a new open-source software TensorFlow show that the learned distributed vectors can be clustered using principal component analysis and hierarchical clustering to reveal a list of non-synonymous mutations that arise on the structural protein VP1 in connection to the candidate mutation for ClO₂ adaptation. Furthermore, this method can account for recombination rates by setting the extent of interactions as a biological hyper-parameter, and the results show that the most realistic scenario of mid-range interactions across the genome is most consistent with the previous studies.

Dynamic interactions between RNA viruses and human hosts unravelled by a decade of next generation sequencing

BACKGROUND

Next generation sequencing (NGS) methods have significantly contributed to a paradigm shift in genomic research for nearly a decade now. These methods have been useful in studying the dynamic interactions between RNA viruses and human hosts.

SCOPE OF THE REVIEW

In this review, we summarise and discuss key applications of NGS in studying the host - pathogen interactions in RNA viral infections of humans with examples.

MAJOR CONCLUSIONS

Use of NGS to study globally relevant RNA viral infections have revolutionized our understanding of the within host and between host evolution of these viruses. These methods have also been useful in clinical decision-making and in guiding biomedical research on vaccine design.

GENERAL SIGNIFICANCE

NGS has been instrumental in viral genomic studies in resolving within-host viral genomic variants and the distribution of nucleotide polymorphisms along the full-length of viral genomes in a high throughput, cost effective manner. In the future, novel advances such as long read, single molecule sequencing of viral genomes and simultaneous sequencing of host and pathogens may become the standard of practice in research and clinical settings. This will also bring on new challenges in big data analysis.

Evolution and medicine in undergraduate education: a prescription for all biology students

The interface between evolutionary biology and the biomedical sciences promises to advance understanding of the origins of genetic and infectious diseases in humans, potentially leading to improved medical diagnostics, therapies, and public health practices. The biomedical sciences also provide unparalleled examples for evolutionary biologists to explore. However, gaps persist between evolution and medicine, for historical reasons and because they are often perceived as having disparate goals. Evolutionary biologists have a role in building a bridge between the disciplines by presenting evolutionary biology in the context of human health and medical practice to undergraduates, including premedical and preprofessional students. We suggest that students will find medical examples of evolution engaging. By making the connections between evolution and medicine clear at the undergraduate level, the stage is set for future health providers and biomedical scientists to work productively in this synthetic area. Here, we frame key evolutionary concepts in terms of human health, so that biomedical examples may be more easily incorporated into evolution courses or more specialized courses on evolutionary medicine. Our goal is to aid in building the scientific foundation in evolutionary biology for all students, and to encourage evolutionary biologists to join in the integration of evolution and medicine.

The double-edged sword: How evolution can make or break a live-attenuated virus vaccine

Even students who reject evolution are often willing to consider cases in which evolutionary biology contributes to, or undermines, biomedical interventions. Moreover the intersection of evolutionary biology and biomedicine is fascinating in its own right. This review offers an overview of the ways in which evolution has impacted the design and deployment of live-attenuated virus vaccines, with subsections that may be useful as lecture material or as the basis for case studies in classes at a variety of levels. Live- attenuated virus vaccines have been modified in ways that restrain their replication in a host, so that infection (vaccination) produces immunity but not disease. Applied evolution, in the form of serial passage in novel host cells, is a "classical" method to generate live-attenuated viruses. However many live-attenuated vaccines exhibit reversion to virulence through back-mutation of attenuating mutations, compensatory mutations elsewhere in the genome, recombination or reassortment, or changes in quasispecies diversity. Additionally the combination of multiple live-attenuated strains may result in competition or facilitation between individual vaccine viruses, resulting in undesirable increases in virulence or decreases in immunogenicity. Genetic engineering informed by evolutionary thinking has led to a number of novel approaches to generate live-attenuated virus vaccines that contain substantial safeguards against reversion to virulence and that ameliorate interference among multiple vaccine strains. Finally, vaccines have the potential to shape the evolution of their wild type counterparts in counter-productive ways; at the extreme vaccine-driven eradication of a virus may create an empty niche that promotes the emergence of new viral pathogens.

Stickbreaking: a novel fitness landscape model that harbors epistasis and is consistent with commonly observed patterns of adaptive evolution

In relating genotypes to fitness, models of adaptation need to both be computationally tractable and qualitatively match observed data. One reason that tractability is not a trivial problem comes from a combinatoric problem whereby no matter in what order a set of mutations occurs, it must yield the same fitness. We refer to this as the bookkeeping problem. Because of their commutative property, the simple additive and multiplicative models naturally solve the bookkeeping problem. However, the fitness trajectories and epistatic patterns they predict are inconsistent with the patterns commonly observed in experimental evolution. This motivates us to propose a new and equally simple model that we call stickbreaking. Under the stickbreaking model, the intrinsic fitness effects of mutations scale by the distance of the current background to a hypothesized boundary. We use simulations and theoretical analyses to explore the basic properties of the stickbreaking model such as fitness trajectories, the distribution of fitness achieved, and epistasis. Stickbreaking is compared to the additive and multiplicative models. We conclude that the stickbreaking model is qualitatively consistent with several commonly observed patterns of adaptive evolution.

Quasispecies spatial models for RNA viruses with different replication modes and infection strategies

Empirical observations and theoretical studies suggest that viruses may use different replication strategies to amplify their genomes, which impact the dynamics of mutation accumulation in viral populations and therefore, their fitness and virulence. Similarly, during natural infections, viruses replicate and infect cells that are rarely in suspension but spatially organized. Surprisingly, most quasispecies models of virus replication have ignored these two phenomena. In order to study these two viral characteristics, we have developed stochastic cellular automata models that simulate two different modes of replication (geometric vs stamping machine) for quasispecies replicating and spreading on a two-dimensional space. Furthermore, we explored these two replication models considering epistatic fitness landscapes (antagonistic vs synergistic) and different scenarios for cell-to-cell spread, one with free superinfection and another with superinfection inhibition. We found that the master sequences for populations replicating geometrically and with antagonistic fitness effects vanished at low critical mutation rates. By contrast, the highest critical mutation rate was observed for populations replicating geometrically but with a synergistic fitness landscape. Our simulations also showed that for stamping machine replication and antagonistic epistasis, a combination that appears to be common among plant viruses, populations further increased their robustness by inhibiting superinfection. We have also shown that the mode of replication strongly influenced the linkage between viral loci, which rapidly reached linkage equilibrium at increasing mutations for geometric replication. We also found that the strategy that minimized the time required to spread over the whole space was the stamping machine with antagonistic epistasis among mutations. Finally, our simulations revealed that the multiplicity of infection fluctuated but generically increased along time.

Why do RNA viruses recombine?

Recombination occurs in many RNA viruses and can be of major evolutionary significance. However, rates of recombination vary dramatically among RNA viruses, which can range from clonal to highly recombinogenic. Here, we review the factors that might explain this variation in recombination frequency and show that there is little evidence that recombination is favoured by natural selection to create advantageous genotypes or purge deleterious mutations, as predicted if recombination functions as a form of sexual reproduction. Rather, recombination rates seemingly reflect larger-scale patterns of viral genome organization, such that recombination may be a mechanistic by-product of the evolutionary pressures acting on other aspects of virus biology.

Estimate of effective recombination rate and average selection coefficient for HIV in chronic infection

HIV adaptation to a host in chronic infection is simulated by means of a Monte-Carlo algorithm that includes the evolutionary factors of mutation, positive selection with varying strength among sites, random genetic drift, linkage, and recombination. By comparing two sensitive measures of linkage disequilibrium (LD) and the number of diverse sites measured in simulation to patient data from one-time samples of pol gene obtained by single-genome sequencing from representative untreated patients, we estimate the effective recombination rate and the average selection coefficient to be on the order of 1% per genome per generation (10(-5) per base per generation) and 0.5%, respectively. The adaptation rate is twofold higher and fourfold lower than predicted in the absence of recombination and in the limit of very frequent recombination, respectively. The level of LD and the number of diverse sites observed in data also range between the values predicted in simulation for these two limiting cases. These results demonstrate the critical importance of finite population size, linkage, and recombination in HIV evolution.

Simple genomes, complex interactions: epistasis in RNA virus

Owed to their reduced size and low number of proteins encoded, RNA viruses and other subviral pathogens are often considered as being genetically too simple. However, this structural simplicity also creates the necessity for viral RNA sequences to encode for more than one protein and for proteins to carry out multiple functions, all together resulting in complex patterns of genetic interactions. In this work we will first review the experimental studies revealing that the architecture of viral genomes is dominated by antagonistic interactions among loci. Second, we will also review mathematical models and provide a description of computational tools for the study of RNA virus dynamics and evolution. As an application of these tools, we will finish this review article by analyzing a stochastic bit-string model of in silico virus replication. This model analyzes the interplay between epistasis and the mode of replication on determining the population load of deleterious mutations. The model suggests that, for a given mutation rate, the deleterious mutational load is always larger when epistasis is predominantly antagonistic than when synergism is the rule. However, the magnitude of this effect is larger if replication occurs geometrically than if it proceeds linearly.

On the classification of epistatic interactions

Modern genomewide association studies are characterized by the problem of "missing heritability." Epistasis, or genetic interaction, has been suggested as a possible explanation for the relatively small contribution of single significant associations to the fraction of variance explained. Of particular concern to investigators of genetic interactions is how to best represent and define epistasis. Previous studies have found that the use of different quantitative definitions for genetic interaction can lead to different conclusions when constructing genetic interaction networks and when addressing evolutionary questions. We suggest that instead, multiple representations of epistasis, or epistatic "subtypes," may be valid within a given system. Selecting among these epistatic subtypes may provide additional insight into the biological and functional relationships among pairs of genes. In this study, we propose maximum-likelihood and model selection methods in a hypothesis-testing framework to choose epistatic subtypes that best represent functional relationships for pairs of genes on the basis of fitness data from both single and double mutants in haploid systems. We gauge the performance of our method with extensive simulations under various interaction scenarios. Our approach performs reasonably well in detecting the most likely epistatic subtype for pairs of genes, as well as in reducing bias when estimating the epistatic parameter (epsilon). We apply our approach to two available data sets from yeast (Saccharomyces cerevisiae) and demonstrate through overlap of our identified epistatic pairs with experimentally verified interactions and functional links that our results are likely of biological significance in understanding interaction mechanisms. We anticipate that our method will improve detection of epistatic interactions and will help to unravel the mysteries of complex biological systems.

Revealing evolutionary pathways by fitness landscape reconstruction

The concept of epistasis has since long been used to denote non-additive fitness effects of genetic changes and has played a central role in understanding the evolution of biological systems. Owing to an array of novel experimental methodologies, it has become possible to experimentally determine epistatic interactions as well as more elaborate genotype-fitness maps. These data have opened up the investigation of a host of long-standing questions in evolutionary biology, such as the ruggedness of fitness landscapes and the accessibility of mutational trajectories, the evolution of sex, and the origin of robustness and modularity. Here we review this recent and timely marriage between systems biology and evolutionary biology, which holds the promise to understand evolutionary dynamics in a more mechanistic and predictive manner.

Epistasis and the evolutionary dynamics of measured genotypic values during simulated serial bottlenecks

The evolutionary effects of epistasis have been primarily explored analytically and most empirical studies have utilized yeast, viral and bacterial populations. Empirical analyses in multi-cellular organisms are rare because of experimental constraints. Here, we report the results of a genome-wide scan for two-way epistasis in 16 traits related to body size and composition in F(2) mice from the LG/J by SM/J intercross. We analyze two-locus genotypic values at quantitative trait loci (QTL), which provides an especially detailed view of epistatic architectures, to evaluate their predicted evolutionary consequences via Monte Carlo simulations. Epistatic profiles vary, but all traits show complicated genetic architectures which are largely hidden in single locus QTL scans. On average, detected epistatic effects are comparable in size to marginal effects. Simulations demonstrate an expected preservation, and often inflation, of heritable variance across several generations of small effective population size for many identified epistatic pairs over a range of starting allele frequencies.

Complementation and epistasis in viral coinfection dynamics

Coinfection in RNA virus populations results in two important phenomena, complementation and recombination. Of the two, complementation has a strong effect on selection against deleterious mutations, as has been confirmed in earlier studies. As complementation delays the purging of less-fit mutations, coinfection may be detrimental to the evolution of a virus population. Here we employ both deterministic modeling and stochastic simulation to explore the mechanisms underlying the interactions between complementation and other evolutionary factors, namely, mutation, selection, and epistasis. We find that strong complementation reduces slightly the overall fitness of a virus population but substantially enhances its diversity and robustness, especially when interacting with selection and epistasis.

The frequency of fitness peak shifts is increased at expanding range margins due to mutation surfing

Dynamic species' ranges, those that are either invasive or shifting in response to environmental change, are the focus of much recent interest in ecology, evolution, and genetics. Understanding how range expansions can shape evolutionary trajectories requires the consideration of nonneutral variability and genetic architecture, yet the majority of empirical and theoretical work to date has explored patterns of neutral variability. Here we use forward computer simulations of population growth, dispersal, and mutation to explore how range-shifting dynamics can influence evolution on rugged fitness landscapes. We employ a two-locus model, incorporating sign epistasis, and find that there is an increased likelihood of fitness peak shifts during a period of range expansion. Maladapted valley genotypes can accumulate at an expanding range front through a phenomenon called mutation surfing, which increases the likelihood that a mutation leading to a higher peak will occur. Our results indicate that most peak shifts occur close to the expanding front. We also demonstrate that periods of range shifting are especially important for peak shifting in species with narrow geographic distributions. Our results imply that trajectories on rugged fitness landscapes can be modified substantially when ranges are dynamic.

Rapid adaptive amplification of preexisting variation in an RNA virus

The amount and nature of preexisting variation in a population of RNA viruses is an important determinant of the virus's ability to adapt rapidly to a changed environment. However, direct quantification of this preexisting variation may be cumbersome, because potentially beneficial alleles are typically rare, and isolation of a large number of subclones is required. Here, we propose a simpler method. We infer the initial population structure of vesicular stomatitis virus (VSV) by fitting a mathematical model of asexual evolution to an extensive set of measurements of VSV fitness dynamics under various conditions, including new and previously published data. The inferred variation of fitness in the initial population agrees very well with the results of direct experiments with subclone fitness quantification. From the same procedure, we also estimate the mean fitness effect of beneficial mutations (selection coefficient s), the percentage of sites in the genome that are under moderate positive or negative selection, and the percentage of sites where beneficial mutations may potentially occur. For VSV strain MARM U evolving in BHK-21 cells, the three parameters have values of 0.39, 9%, and 0.06%, respectively. The method can be generalized and applied easily to other rapidly evolving microbes, including both asexual microorganisms and those with recombination.

The evolution of genome compression and genomic novelty in RNA viruses

The genomes of RNA viruses are characterized by their extremely small size and extremely high mutation rates (typically 10 kb and 10(-4)/base/replication cycle, respectively), traits that are thought to be causally linked. One aspect of their small size is the genome compression caused by the use of overlapping genes (where some nucleotides code for two genes). Using a comparative analysis of all known RNA viral species, we show that viruses with larger genomes tend to have less gene overlap. We provide a numerical model to show how a high mutation rate could lead to gene overlap, and we discuss the factors that might explain the observed relationship between gene overlap and genome size. We also propose a model for the evolution of gene overlap based on the co-opting of previously unused ORFs, which gives rise to two types of overlap: (1) the creation of novel genes inside older genes, predominantly via +1 frameshifts, and (2) the incremental increase in overlap between originally contiguous genes, with no frameshift preference. Both types of overlap are viewed as the creation of genomic novelty under pressure for genome compression. Simulations based on our model generate the empirical size distributions of overlaps and explain the observed frameshift preferences. We suggest that RNA viruses are a good model system for the investigation of general evolutionary relationship between genome attributes such as mutational robustness, mutation rate, and size.

VARIABLE EPISTATIC EFFECTS BETWEEN MUTATIONS AT HOST RECOGNITION SITES IN ?X174 BACTERIOPHAGE

Epistatic interactions between mutations are widespread. Theoretical investigations have shown that variability in epistatic effects influences fundamental evolutionary processes, yet few empirical studies have identified causes or the extent of this variation. We examined variation in epistatic effects of mutations at two host recognition sites in phiX174 bacteriophage. We calculated epistatic effects from the sum of fitness effects (log scale) of two single mutants and their corresponding double mutant for five combinations of mutations in six conditions. We found that epistatic effects differed in sign, degree, and variability across conditions. The data highlight that even between single mutations at the same two sites the sign and variability of epistatic effects are affected by environment. We discuss these findings in the context of studying the role of epistasis in evolution.

Distributions of epistasis in microbes fit predictions from a fitness landscape model

How do the fitness effects of several mutations combine? Despite its simplicity, this question is central to the understanding of multilocus evolution. Epistasis (the interaction between alleles at different loci), especially epistasis for fitness traits such as reproduction and survival, influences evolutionary predictions "almost whenever multilocus genetics matters". Yet very few models have sought to predict epistasis, and none has been empirically tested. Here we show that the distribution of epistasis can be predicted from the distribution of single mutation effects, based on a simple fitness landscape model. We show that this prediction closely matches the empirical measures of epistasis that have been obtained for Escherichia coli and the RNA virus vesicular stomatitis virus. Our results suggest that a simple fitness landscape model may be sufficient to quantitatively capture the complex nature of gene interactions. This model may offer a simple and widely applicable alternative to complex metabolic network models, in particular for making evolutionary predictions.

Hemiclonal reproduction slows down the speed of Muller's ratchet in the hybridogenetic frog Rana esculenta

Rare recombination in otherwise asexually reproducing organisms is known to beneficially influence the fitness in small populations. In most of the investigated organisms, asexual and rare sexual generations with recombination follow each other sequentially. Here we present a case where clonal reproduction and rare recombination occur simultaneously in the same population. The hybridogenetic water frog Rana esculenta (E), a hybrid between R. lessonae (L) and R. ridibunda (R) produces gametes that only contain the unaltered maternal R part of their genome. New generations of R. esculenta usually arise from E x L matings. Intraspecific E x E matings produce mostly inviable offspring, but in rare cases, female R. ridibunda arise from such matings which are capable of recombination. In the absence of conspecific males, these R females have to mate with E males, which results in further R females, or with L males, which produces new E lineages. This indirect mechanism reintroduces recombination into the otherwise clonally transmitted R genomes in R. esculenta populations. In this study, we show through Monte Carlo simulations that, in most cases, it is sufficient that only between 1 % and 10 % of mixed water frog populations consist of R females to prevent or significantly reduce the fixation and accumulation of deleterious mutations.

Epistasis correlates to genomic complexity

Whether systematic genetic interactions (epistasis) occur at the genomic scale remains a challenging topic in evolutionary biology. Epistasis should make a significant contribution to variation in complex traits and influence the evolution of genetic systems as sex, diploidy, dominance, or the contamination of genomes with deleterious mutations. We have collected data from widely different organisms and quantified epistasis in a common, per-generation scale. Simpler genomes, such as those of RNA viruses, display antagonistic epistasis (mutations have smaller effects together than expected); bacterial microorganisms do not apparently deviate from independent effects, whereas in multicellular eukaryotes, a transition toward synergistic epistasis occurs (mutations have larger effects together than expected). We propose that antagonistic epistasis might be a property of compact genomes with few nonpleiotropic biological functions, whereas in complex genomes, synergism might emerge from mutational robustness.

Characterization of mycobacterial virulence genes through genetic interaction mapping

We have previously shown that approximately 5% of the genes encoded by the genome of Mycobacterium tuberculosis are specifically required for the growth or survival of this bacterium during infection. This corresponds to hundreds of genes, most of which have no identifiable function. As a unique approach to characterize these genes, we developed a method to rapidly delineate functional pathways by identifying mutations that modify each other's phenotype, i.e., "genetic interactions". Using this method, we have defined a complex set of interactions between virulence genes in this pathogen, and find that the products of unlinked genes associate to form multisubunit transporters that are required for bacterial survival in the host. These findings implicate a previously undescribed family of transport systems in the pathogenesis of tuberculosis, and identify genes that are likely to function in the metabolism of their substrates. This method can be readily applied to other organisms at either the single pathway level, as described here, or at the system level to define quantitative genetic interaction networks.

Quantifying antagonistic epistasis in a multifunctional RNA secondary structure of the Rous sarcoma virus

Recent studies have suggested that antagonistic epistasis (i.e. mutations having smaller effects in combination than alone) may be common among RNA viruses, in contrast to other biological systems. Here, by re-analysing previously published data from a random viral library, selection and epistasis coefficients were estimated in the U5-IR stem and loop of theRous sarcoma virus, a region that adopts a conserved secondary structure and is involved in various essential steps of viral infection. The estimated mutational fitness effects are extremely high and genetic interactions are antagonistic on average. This pattern might be representative of RNA virus genomes, which show high compaction and frequent secondary structures. The implications for RNA virus adaptability are explored.

Virus fitness: concept, quantification, and application to HIV population dynamics

Viral fitness has been broadly studied during the past three decades, mainly to test evolutionary models and population theories difficult to analyze and interpret with more complex organisms. More recent studies, however, are focused in the role of fitness on viral transmission, pathogenesis, and drug resistance. Here, we used human immunodeficiency virus (HIV) as one of the most relevant models to evaluate the importance of viral quasispecies and fitness in HIV evolution, population dynamics, disease progression, and potential clinical implications.

Future issues in RNA virus evolution

The study of RNA virus evolution has developed rapidly during the past 30 years. This review outlines some important recent findings, as well as a number of the remaining major challenges, particularly those that might explain why RNA viruses are the most important class of emerging diseases, yet often have difficulties adapting to sustained transmission cycles in new hosts. The author emphasizes the relevance of research on the underlying dynamics of mutation, fitness landscapes and the constraints to viral adaptation, as well as the evolution of recombination and reassortment. It is also suggested that a combination of theoretical, experimental and comparative approaches is essential for future studies of viral evolution, coupled with new genome sequence data on intrahost genetic variation.

The effect of host heterogeneity and parasite intragenomic interactions on parasite population structure

Understanding the processes that shape the genetic structure of parasite populations and the functional consequences of different parasite genotypes is critical for our ability to predict how an infection can spread through a host population and for the design of effective vaccines to combat infection and disease. Here, we examine how the genetic structure of parasite populations responds to host genetic heterogeneity. We consider the well-characterized molecular specificity of major histocompatibility complex binding of antigenic peptides to derive deterministic and stochastic models. We use these models to ask, firstly, what conditions favour the evolution of generalist parasite genotypes versus specialist parasite genotypes? Secondly, can parasite genotypes coexist in a population? We find that intragenomic interactions between parasite loci encoding antigenic peptides are pivotal in determining the outcome of evolution. Where parasite loci interact synergistically (i.e. the recognition of additional antigenic peptides has a disproportionately large effect on parasite fitness), generalist parasite genotypes are favoured. Where parasite loci act multiplicatively (have independent effects on fitness) or antagonistically (have diminishing effects on parasite fitness), specialist parasite genotypes are favoured. A key finding is that polymorphism is not stable and that, with respect to functionally important antigenic peptides, parasite populations are dominated by a single genotype.

A measure of the within-chromosome synergistic epistasis for Drosophila viability

In order to detect possible synergistic epistasis for viability in Drosophila melanogaster we assayed the relative viability of chromosomes II in: (i) panmixia, (ii) forced total homozygosity, and (iii) homozygosity for, on the average, half of their loci. As these genotypes were constructed using exactly the same set of chromosomes in the three cases, the design allows us to estimate the inbreeding depression rate at two different inbreeding levels in the absence of purging natural selection. Overall, no consistent synergistic epistasis was found. However, there was a small fraction of chromosomes whose severely deleterious effect when homozygous was almost significantly larger than expected from their viability when homozygous for half of their loci. This suggests occasional but important synergistic epistasis, which might confer evolutionary advantage to recombination in tightly linked genomes. Nevertheless, such epistasis is unlikely to be an evolutionary advantage driving the evolution of sexual anisogamous reproduction, as its contribution to overall viability is small when compared with the two-fold cost of anisogamy.

Epistasis and the adaptability of an RNA virus

We have explored the patterns of fitness recovery in the vesicular stomatitis RNA virus. We show that, in our experimental setting, reversions to the wild-type genotype were rare and fitness recovery was at least partially driven by compensatory mutations. We compared compensatory adaptation for genotypes carrying (1) mutations with varying deleterious fitness effects, (2) one or two deleterious mutations, and (3) pairs of mutations showing differences in the strength and sign of epistasis. In all cases, we found that the rate of fitness recovery and the proportion of reversions were positively affected by population size. Additionally, we observed that mutations with large fitness effect were always compensated faster than mutations with small fitness effect. Similarly, compensatory evolution was faster for genotypes carrying a single deleterious mutation than for those carrying pairs of mutations. Finally, for genotypes carrying two deleterious mutations, we found evidence of a negative correlation between the epistastic effect and the rate of compensatory evolution.

Evidence for positive epistasis in HIV-1

Reproductive strategies such as sexual reproduction and recombination that involve the shuffling of parental genomes for the production of offspring are ubiquitous in nature. However, their evolutionary benefit remains unclear. Many theories have identified potential benefits, but progress is hampered by the scarcity of relevant data. One class of theories is based on the assumption that mutations affecting fitness exhibit negative epistasis. Retroviruses recombine frequently and thus provide a unique opportunity to test these theories. Using amino acid sequence data and fitness values from 9466 human immunodeficiency virus 1 (HIV-1) isolates, we find in contrast to these theories strong statistical evidence for a predominance of positive epistasis in HIV-1.

Modular epistasis in yeast metabolism

Epistatic interactions, manifested in the effects of mutations on the phenotypes caused by other mutations, may help uncover the functional organization of complex biological networks. Here, we studied system-level epistatic interactions by computing growth phenotypes of all single and double knockouts of 890 metabolic genes in Saccharomyces cerevisiae, using the framework of flux balance analysis. A new scale for epistasis identified a distinctive trimodal distribution of these epistatic effects, allowing gene pairs to be classified as buffering, aggravating or noninteracting. We found that the ensuing epistatic interaction network could be organized hierarchically into function-enriched modules that interact with each other 'monochromatically' (i.e., with purely aggravating or purely buffering epistatic links). This property extends the concept of epistasis from single genes to functional units and provides a new definition of biological modularity, which emphasizes interactions between, rather than within, functional modules. Our approach can be used to infer functional gene modules from purely phenotypic epistasis measurements.