Contingency and entrenchment in protein evolution under purifying selection

Potts Hamiltonian models of protein co-variation, free energy landscapes, and evolutionary fitness

Potts Hamiltonian models of protein sequence co-variation are statistical models constructed from the pair correlations observed in a multiple sequence alignment (MSA) of a protein family. These models are powerful because they capture higher order correlations induced by mutations evolving under constraints and help quantify the connections between protein sequence, structure, and function maintained through evolution. We review recent work with Potts models to predict protein structure and sequence-dependent conformational free energy landscapes, to survey protein fitness landscapes and to explore the effects of epistasis on fitness. We also comment on the numerical methods used to infer these models for each application.

Bridging the physical scales in evolutionary biology: from protein sequence space to fitness of organisms and populations

Bridging the gap between the molecular properties of proteins and organismal/population fitness is essential for understanding evolutionary processes. This task requires the integration of the several physical scales of biological organization, each defined by a distinct set of mechanisms and constraints, into a single unifying model. The molecular scale is dominated by the constraints imposed by the physico-chemical properties of proteins and their substrates, which give rise to trade-offs and epistatic (non-additive) effects of mutations. At the systems scale, biological networks modulate protein expression and can either buffer or enhance the fitness effects of mutations. The population scale is influenced by the mutational input, selection regimes, and stochastic changes affecting the size and structure of populations, which eventually determine the evolutionary fate of mutations. Here, we summarize the recent advances in theory, computer simulations, and experiments that advance our understanding of the links between various physical scales in biology.

Detecting Adaptation in Protein-Coding Genes Using a Bayesian Site-Heterogeneous Mutation-Selection Codon Substitution Model

Codon substitution models have traditionally attempted to uncover signatures of adaptation within protein-coding genes by contrasting the rates of synonymous and non-synonymous substitutions. Another modeling approach, known as the mutation–selection framework, attempts to explicitly account for selective patterns at the amino acid level, with some approaches allowing for heterogeneity in these patterns across codon sites. Under such a model, substitutions at a given position occur at the neutral or nearly neutral rate when they are synonymous, or when they correspond to replacements between amino acids of similar fitness; substitutions from high to low (low to high) fitness amino acids have comparatively low (high) rates. Here, we study the use of such a mutation–selection framework as a null model for the detection of adaptation. Following previous works in this direction, we include a deviation parameter that has the effect of capturing the surplus, or deficit, in non-synonymous rates, relative to what would be expected under a mutation–selection modeling framework that includes a Dirichlet process approach to account for across-codon-site variation in amino acid fitness profiles. We use simulations, along with a few real data sets, to study the behavior of the approach, and find it to have good power with a low false-positive rate. Altogether, we emphasize the potential of recent mutation–selection models in the detection of adaptation, calling for further model refinements as well as large-scale applications.

Gene Tree Discordance Can Generate Patterns of Diminishing Convergence over Time

Phenotypic convergence is an exciting outcome of adaptive evolution, occurring when different species find similar solutions to the same problem. Unraveling the molecular basis of convergence provides a way to link genotype to adaptive phenotypes, but can also shed light on the extent to which molecular evolution is repeatable and predictable. Many recent genome-wide studies have uncovered a striking pattern of diminishing convergence over time, ascribing this pattern to the presence of intramolecular epistatic interactions. Here, we consider gene tree discordance as an alternative cause of changes in convergence levels over time in a primate dataset. We demonstrate that gene tree discordance can produce patterns of diminishing convergence by itself, and that controlling for discordance as a cause of apparent convergence makes the pattern disappear. We also show that synonymous substitutions, where neither selection nor epistasis should be prevalent, have the same diminishing pattern of molecular convergence in primates. Finally, we demonstrate that even in situations where biological discordance is not possible, discordance due to errors in species tree inference can drive similar patterns. Though intramolecular epistasis could in principle create a pattern of declining convergence over time, our results suggest a possible alternative explanation for this widespread pattern. These results contribute to a growing appreciation not just of the presence of gene tree discordance, but of the unpredictable effects this discordance can have on analyses of molecular evolution.

Novelty and Innovation in the History of Life

The history of life as documented by the fossil record encompasses evolutionary diversifications at scales ranging from the Ediacaran-Cambrian explosion of animal life and the invasion of land by vascular plants, insects and vertebrates to the diversification of flowering plants over the past 100 million years and the radiation of horses. Morphological novelty and innovation has been a recurrent theme. The architects of the modern synthesis of evolutionary theory made three claims about evolutionary novelty and innovation: first, that all diversifications in the history of life represent adaptive radiations; second, that adaptive radiations are driven principally by ecological opportunity rather than by the supply of new morphological novelties, thus the primary questions about novelty and innovation focus on their ecological and evolutionary success; and third, that the rate of morphological divergence between taxa was more rapid early in the history of a clade but slowed over time as ecological opportunities declined. These claims have strongly influenced subsequent generations of evolutionary biologists, yet over the past two decades each has been challenged by data from the fossil record, by the results of comparative phylogenetic analyses and through insights from evolutionary developmental biology. Consequently a broader view of novelty and innovation is required. An outstanding issue for future work is identifying the circumstances associated with different styles of diversification and whether their frequency has changed through the history of life.

Extensively Parameterized Mutation-Selection Models Reliably Capture Site-Specific Selective Constraint

The mutation-selection model of coding sequence evolution has received renewed attention for its use in estimating site-specific amino acid propensities and selection coefficient distributions. Two computationally tractable mutation-selection inference frameworks have been introduced: One framework employs a fixed-effects, highly parameterized maximum likelihood approach, whereas the other employs a random-effects Bayesian Dirichlet Process approach. While both implementations follow the same model, they appear to make distinct predictions about the distribution of selection coefficients. The fixed-effects framework estimates a large proportion of highly deleterious substitutions, whereas the random-effects framework estimates that all substitutions are either nearly neutral or weakly deleterious. It remains unknown, however, how accurately each method infers evolutionary constraints at individual sites. Indeed, selection coefficient distributions pool all site-specific inferences, thereby obscuring a precise assessment of site-specific estimates. Therefore, in this study, we use a simulation-based strategy to determine how accurately each approach recapitulates the selective constraint at individual sites. We find that the fixed-effects approach, despite its extensive parameterization, consistently and accurately estimates site-specific evolutionary constraint. By contrast, the random-effects Bayesian approach systematically underestimates the strength of natural selection, particularly for slowly evolving sites. We also find that, despite the strong differences between their inferred selection coefficient distributions, the fixed- and random-effects approaches yield surprisingly similar inferences of site-specific selective constraint. We conclude that the fixed-effects mutation-selection framework provides the more reliable software platform for model application and future development.

Epistasis in protein evolution

The structure, function, and evolution of proteins depend on physical and genetic interactions among amino acids. Recent studies have used new strategies to explore the prevalence, biochemical mechanisms, and evolutionary implications of these interactions-called epistasis-within proteins. Here we describe an emerging picture of pervasive epistasis in which the physical and biological effects of mutations change over the course of evolution in a lineage-specific fashion. Epistasis can restrict the trajectories available to an evolving protein or open new paths to sequences and functions that would otherwise have been inaccessible. We describe two broad classes of epistatic interactions, which arise from different physical mechanisms and have different effects on evolutionary processes. Specific epistasis-in which one mutation influences the phenotypic effect of few other mutations-is caused by direct and indirect physical interactions between mutations, which nonadditively change the protein's physical properties, such as conformation, stability, or affinity for ligands. In contrast, nonspecific epistasis describes mutations that modify the effect of many others; these typically behave additively with respect to the physical properties of a protein but exhibit epistasis because of a nonlinear relationship between the physical properties and their biological effects, such as function or fitness. Both types of interaction are rampant, but specific epistasis has stronger effects on the rate and outcomes of evolution, because it imposes stricter constraints and modulates evolutionary potential more dramatically; it therefore makes evolution more contingent on low-probability historical events and leaves stronger marks on the sequences, structures, and functions of protein families.

Epistasis and the Dynamics of Reversion in Molecular Evolution

Recent studies of protein evolution contend that the longer an amino acid substitution is present at a site, the less likely it is to revert to the amino acid previously occupying that site. Here we study this phenomenon of decreasing reversion rates rigorously and in a much more general context. We show that, under weak mutation and for arbitrary fitness landscapes, reversion rates decrease with time for any site that is involved in at least one epistatic interaction. Specifically, we prove that, at stationarity, the hazard function of the distribution of waiting times until reversion is strictly decreasing for any such site. Thus, in the presence of epistasis, the longer a particular character has been absent from a site, the less likely the site will revert to its prior state. We also explore several examples of this general result, which share a common pattern whereby the probability of having reverted increases rapidly at short times to some substantial value before becoming almost flat after a few substitutions at other sites. This pattern indicates a characteristic tendency for reversion to occur either almost immediately after the initial substitution or only after a very long time.

The tangled bank of amino acids

The use of amino acid substitution matrices to model protein evolution has yielded important insights into both the evolutionary process and the properties of specific protein families. In order to make these models tractable, standard substitution matrices represent the average results of the evolutionary process rather than the underlying molecular biophysics and population genetics, treating proteins as a set of independently evolving sites rather than as an integrated biomolecular entity. With advances in computing and the increasing availability of sequence data, we now have an opportunity to move beyond current substitution matrices to more interpretable mechanistic models with greater fidelity to the evolutionary process of mutation and selection and the holistic nature of the selective constraints. As part of this endeavour, we consider how epistatic interactions induce spatial and temporal rate heterogeneity, and demonstrate how these generally ignored factors can reconcile standard substitution rate matrices and the underlying biology, allowing us to better understand the meaning of these substitution rates. Using computational simulations of protein evolution, we can demonstrate the importance of both spatial and temporal heterogeneity in modelling protein evolution.

Causes of molecular convergence and parallelism in protein evolution

To what extent is the convergent evolution of protein function attributable to convergent or parallel changes at the amino acid level? The mutations that contribute to adaptive protein evolution may represent a biased subset of all possible beneficial mutations owing to mutation bias and/or variation in the magnitude of deleterious pleiotropy. A key finding is that the fitness effects of amino acid mutations are often conditional on genetic background. This context dependence (epistasis) can reduce the probability of convergence and parallelism because it reduces the number of possible mutations that are unconditionally acceptable in divergent genetic backgrounds. Here, I review factors that influence the probability of replicated evolution at the molecular level.

Engineering and Evolution of Molecular Chaperones and Protein Disaggregases with Enhanced Activity

Cells have evolved a sophisticated proteostasis network to ensure that proteins acquire and retain their native structure and function. Critical components of this network include molecular chaperones and protein disaggregases, which function to prevent and reverse deleterious protein misfolding. Nevertheless, proteostasis networks have limits, which when exceeded can have fatal consequences as in various neurodegenerative disorders, including Parkinson's disease and amyotrophic lateral sclerosis. A promising strategy is to engineer proteostasis networks to counter challenges presented by specific diseases or specific proteins. Here, we review efforts to enhance the activity of individual molecular chaperones or protein disaggregases via engineering and directed evolution. Remarkably, enhanced global activity or altered substrate specificity of various molecular chaperones, including GroEL, Hsp70, ClpX, and Spy, can be achieved by minor changes in primary sequence and often a single missense mutation. Likewise, small changes in the primary sequence of Hsp104 yield potentiated protein disaggregases that reverse the aggregation and buffer toxicity of various neurodegenerative disease proteins, including α-synuclein, TDP-43, and FUS. Collectively, these advances have revealed key mechanistic and functional insights into chaperone and disaggregase biology. They also suggest that enhanced chaperones and disaggregases could have important applications in treating human disease as well as in the purification of valuable proteins in the pharmaceutical sector.

Connecting Active-Site Loop Conformations and Catalysis in Triosephosphate Isomerase: Insights from a Rare Variation at Residue 96 in the Plasmodial Enzyme

Despite extensive research into triosephosphate isomerases (TIMs), there exists a gap in understanding of the remarkable conjunction between catalytic loop-6 (residues 166-176) movement and the conformational flip of Glu165 (catalytic base) upon substrate binding that primes the active site for efficient catalysis. The overwhelming occurrence of serine at position 96 (98% of the 6277 unique TIM sequences), spatially proximal to E165 and the loop-6 residues, raises questions about its role in catalysis. Notably, Plasmodium falciparum TIM has an extremely rare residue--phenylalanine--at this position whereas, curiously, the mutant F96S was catalytically defective. We have obtained insights into the influence of residue 96 on the loop-6 conformational flip and E165 positioning by combining kinetic and structural studies on the PfTIM F96 mutants F96Y, F96A, F96S/S73A, and F96S/L167V with sequence conservation analysis and comparative analysis of the available apo and holo structures of the enzyme from diverse organisms.

Multiple Barriers to the Evolution of Alternative Gene Orders in a Positive-Strand RNA Virus

The order in which genes are organized within a genome is generally not conserved between distantly related species. However, within virus orders and families, strong conservation of gene order is observed. The factors that constrain or promote gene-order diversity are largely unknown, although the regulation of gene expression is one important constraint for viruses. Here we investigate why gene order is conserved for a positive-strand RNA virus encoding a single polyprotein in the context of its authentic multicellular host. Initially, we identified the most plausible trajectory by which alternative gene orders could evolve. Subsequently, we studied the accessibility of key steps along this evolutionary trajectory by constructing two virus intermediates: (1) duplication of a gene followed by (2) loss of the ancestral gene. We identified five barriers to the evolution of alternative gene orders. First, the number of viable positions for reordering is limited. Second, the within-host fitness of viruses with gene duplications is low compared to the wild-type virus. Third, after duplication, the ancestral gene copy is always maintained and never the duplicated one. Fourth, viruses with an alternative gene order have even lower fitness than viruses with gene duplications. Fifth, after more than half a year of evolution in isolation, viruses with an alternative gene order are still vastly inferior to the wild-type virus. Our results show that all steps along plausible evolutionary trajectories to alternative gene orders are highly unlikely. Hence, the inaccessibility of these trajectories probably contributes to the conservation of gene order in present-day viruses.

Residue proximity information and protein model discrimination using saturation-suppressor mutagenesis

Identification of residue-residue contacts from primary sequence can be used to guide protein structure prediction. Using Escherichia coli CcdB as the test case, we describe an experimental method termed saturation-suppressor mutagenesis to acquire residue contact information. In this methodology, for each of five inactive CcdB mutants, exhaustive screens for suppressors were performed. Proximal suppressors were accurately discriminated from distal suppressors based on their phenotypes when present as single mutants. Experimentally identified putative proximal pairs formed spatial constraints to recover >98% of native-like models of CcdB from a decoy dataset. Suppressor methodology was also applied to the integral membrane protein, diacylglycerol kinase A where the structures determined by X-ray crystallography and NMR were significantly different. Suppressor as well as sequence co-variation data clearly point to the X-ray structure being the functional one adopted in vivo. The methodology is applicable to any macromolecular system for which a convenient phenotypic assay exists.

DOI:http://dx.doi.org/10.7554/eLife.09532.001

Common techniques to determine the three-dimensional structures of proteins can help researchers to understand these molecules’ activities, but are often time-consuming and do not work for all proteins. Proteins are made of chains of amino acids. When a protein chain folds, some of these amino acids interact with other amino acids and these contacts dictate the overall shape of the protein. This means that identifying the pairs of contacting amino acids could make it possible to predict the protein’s structure.

Interactions between pairs of contacting amino acids tend to remain conserved throughout evolution, and if a mutation alters one of the amino acids in a pair then a 'compensatory' change often occurs to alter the second amino acid as well. Compensatory mutations can suggest that two amino acids are close to each other in the three-dimensional shape of a protein, but the computational methods used to identify such amino acid pairs can sometimes be inaccurate.

In 2012, researchers generated mutants of a bacterial protein called CcdB with changes to single amino acids that caused the protein to fail to fold correctly. Now, Sahoo et al. – who include two of the researchers involved in the 2012 work – have developed an experimental method to identify contacting amino acids and use the CcdB protein as a test case. The approach involved searching for additional mutations that could restore the activity of five of the original mutant proteins when the proteins were produced in yeast cells. The rationale was that any secondary mutations that restored the activity must have corrected the folding defect caused by the original mutation. Sahoo et al. then predicted how close the amino acids affected by the secondary mutations were to the amino acids altered by the original mutations. This information was used to select reliable three-dimensional models of CcdB from a large set of possible structures that had been generated previously using computer models.

Next, the technique was applied to a protein called diacylglycerol kinase A. The structure of this protein had previously been inferred using techniques such as X-ray crystallography and nuclear magnetic resonance, but there was a mismatch between the two methods. Sahoo et al. found that the amino acid contacts derived from their experimental method matched those found in the crystal structure, suggesting that the functional protein structure in living cells is similar to the crystal structure. In the future, the experimental approach developed in this work could be combined with existing methods to reliably guide protein structure prediction.

DOI:http://dx.doi.org/10.7554/eLife.09532.002