Rapid evolutionary change in trait correlations of single proteins.

Many organismal traits are genetically determined and covary in evolving populations. The resulting trait correlations can either help or hinder evolvability - the ability to bring forth new and adaptive phenotypes. The evolution of evolvability requires that trait correlations themselves must be able to evolve, but we know little about this ability. To learn more about it, we here study two evolvable systems, a yellow fluorescent protein and the antibiotic resistance protein VIM-2 metallo beta-lactamase. We consider two traits in the fluorescent protein, namely the ability to emit yellow and green light, and three traits in our enzyme, namely the resistance against ampicillin, cefotaxime, and meropenem. We show that correlations between these traits can evolve rapidly through both mutation and selection on short evolutionary time scales. In addition, we show that these correlations are driven by a protein's ability to fold, because single mutations that alter foldability can dramatically change trait correlations. Since foldability is important for most proteins and their traits, mutations affecting protein folding may alter trait correlations mediated by many other proteins. Thus, mutations that affect protein foldability may also help shape the correlations of complex traits that are affected by hundreds of proteins.

Frustration can Limit the Adaptation of Promiscuous Enzymes Through Gene Duplication and Specialisation.

Virtually all enzymes catalyse more than one reaction, a phenomenon known as enzyme promiscuity. It is unclear whether promiscuous enzymes are more often generalists that catalyse multiple reactions at similar rates or specialists that catalyse one reaction much more efficiently than other reactions. In addition, the factors that shape whether an enzyme evolves to be a generalist or a specialist are poorly understood. To address these questions, we follow a three-pronged approach. First, we examine the distribution of promiscuity in empirical enzymes reported in the BRENDA database. We find that the promiscuity distribution of empirical enzymes is bimodal. In other words, a large fraction of promiscuous enzymes are either generalists or specialists, with few intermediates. Second, we demonstrate that enzyme biophysics is not sufficient to explain this bimodal distribution. Third, we devise a constraint-based model of promiscuous enzymes undergoing duplication and facing selection pressures favouring subfunctionalization. The model posits the existence of constraints between the catalytic efficiencies of an enzyme for different reactions and is inspired by empirical case studies. The promiscuity distribution predicted by our constraint-based model is consistent with the empirical bimodal distribution. Our results suggest that subfunctionalization is possible and beneficial only in certain enzymes. Furthermore, the model predicts that conflicting constraints and selection pressures can cause promiscuous enzymes to enter a 'frustrated' state, in which competing interactions limit the specialisation of enzymes. We find that frustration can be both a driver and an inhibitor of enzyme evolution by duplication and subfunctionalization. In addition, our model predicts that frustration becomes more likely as enzymes catalyse more reactions, implying that natural selection may prefer catalytically simple enzymes. In sum, our results suggest that frustration may play an important role in enzyme evolution.

Single-cell analysis reveals the spatial-temporal expression of genes associated with esophageal malformations.

Understanding the molecular mechanisms of congenital diseases is challenging due to their occurrence within specific developmental stages. Esophageal malformations are examples of such conditions, characterized by abnormalities in the development of esophagus during embryogenesis. These developmental malformations encompass a range of anomalies, including esophageal atresia, and tracheoesophageal fistula. Here, we investigated the preferential expression of 29 genes that are implicated in such malformations and their immediate interactome (a total of 67 genes). We conducted our analyses across several single-cell atlases of embryonic development, encompassing approximately 150,000 cells from the mouse foregut, 180,000 cells from human embryos, and 500,000 cells from 24 human organs. Our study, spanning diverse mesodermal and endodermal cell populations and early developmental stages, shows that the genes associated with esophageal malformations show their highest cell-type specific expression in lateral plate mesoderm cells and at the developmental stage of E8.75-E9.0 days. In human embryos, these genes show a significant cell-type specific expression among subpopulations of epithelial cells, fibroblasts and progenitor cells including basal cells. Notably, members of the forkhead-box family of transcription factors, namely FOXF1, FOXC1, and FOXD1, as well as the SRY-box transcription factor, SOX2, demonstrate the most significant preferential expression in both mouse and human embryos. Overall, our findings provide insights into the temporal and cellular contexts contributing to esophageal malformations.

Phosphorylation sites are evolutionary checkpoints against liquid-solid transition in protein condensates.

Assemblies of multivalent RNA-binding protein fused in sarcoma (FUS) can exist in the functional liquid-like state as well as less dynamic and potentially toxic amyloid- and hydrogel-like states. How could then cells form liquid-like condensates while avoiding their transformation to amyloids? Here, we show how posttranslational phosphorylation can provide a "handle" that prevents liquid-solid transition of intracellular condensates containing FUS. Using residue-specific coarse-grained simulations, for 85 different mammalian FUS sequences, we show how the number of phosphorylation sites and their spatial arrangement affect intracluster dynamics preventing conversion to amyloids. All atom simulations further confirm that phosphorylation can effectively reduce the ß-sheet propensity in amyloid-prone fragments of FUS. A detailed evolutionary analysis shows that mammalian FUS PLDs are enriched in amyloid-prone stretches compared to control neutrally evolved sequences, suggesting that mammalian FUS proteins evolved to self-assemble. However, in stark contrast to proteins that do not phase-separate for their function, mammalian sequences have phosphosites in close proximity to these amyloid-prone regions. These results suggest that evolution uses amyloid-prone sequences in prion-like domains to enhance phase separation of condensate proteins while enriching phosphorylation sites in close proximity to safeguard against liquid-solid transitions.

Low protein expression enhances phenotypic evolvability by intensifying selection on folding stability.

Protein abundance affects the evolution of protein genotypes, but we do not know how it affects the evolution of protein phenotypes. Here we investigate the role of protein abundance in the evolvability of green fluorescent protein (GFP) towards the novel phenotype of cyan fluorescence. We evolve GFP in E. coli through multiple cycles of mutation and selection and show that low GFP expression facilitates the evolution of cyan fluorescence. A computational model whose predictions we test experimentally helps explain why: lowly expressed proteins are under stronger selection for proper folding, which facilitates their evolvability on short evolutionary time scales. The reason is that high fluorescence can be achieved by either few proteins that fold well or by many proteins that fold less well. In other words, we observe a synergy between a protein's scarcity and its stability. Because many proteins meet the essential requirements for this scarcity-stability synergy, it may be a widespread mechanism by which low expression helps proteins evolve new phenotypes and functions.

Yeast Proteins may Reversibly Aggregate like Amphiphilic Molecules.

More than a hundred proteins in yeast reversibly aggregate and phase-separate in response to various stressors, such as nutrient depletion and heat shock. We know little about the protein sequence and structural features behind this ability, which has not been characterized on a proteome-wide level. To identify the distinctive features of aggregation-prone protein regions, we apply machine learning algorithms to genome-scale limited proteolysis-mass spectrometry (LiP-MS) data from yeast proteins. LiP-MS data reveals that 96 proteins show significant structural changes upon heat shock. We find that in these proteins the propensity to phase separate cannot be solely driven by disordered regions, because their aggregation-prone regions (APRs) are not significantly disordered. Instead, the phase separation of these proteins requires contributions from both disordered and structured regions. APRs are significantly enriched in aliphatic residues and depleted in positively charged amino acids. Aggregator proteins with longer APRs show a greater propensity to aggregate, a relationship that can be explained by equilibrium statistical thermodynamics. Altogether, our observations suggest that proteome-wide reversible protein aggregation is mediated by sequence-encoded properties. We propose that aggregating proteins resemble supra-molecular amphiphiles, where APRs are the hydrophobic parts, and non-APRs are the hydrophilic parts.

The length scale of multivalent interactions is evolutionarily conserved in fungal and vertebrate phase-separating proteins.

One key feature of proteins that form liquid droplets by phase separation inside a cell is multivalency-the presence of multiple sites that mediate interactions with other proteins. We know little about the variation of multivalency on evolutionary time scales. Here, we investigated the long-term evolution (∼600 million years) of multivalency in fungal mRNA decapping subunit 2 protein (Dcp2), and in the FET (FUS, EWS and TAF15) protein family. We found that multivalency varies substantially among the orthologs of these proteins. However, evolution has maintained the length scale at which sequence motifs that enable protein-protein interactions occur. That is, the total number of such motifs per hundred amino acids is higher and less variable than expected by neutral evolution. To help explain this evolutionary conservation, we developed a conformation classifier using machine-learning algorithms. This classifier demonstrates that disordered segments in Dcp2 and FET proteins tend to adopt compact conformations, which is necessary for phase separation. Thus, the evolutionary conservation we detected may help proteins preserve the ability to undergo phase separation. Altogether, our study reveals that the length scale of multivalent interactions is an evolutionarily conserved feature of two classes of phase-separating proteins in fungi and vertebrates.

Avoidance of protein unfolding constrains protein stability in long-term evolution.

Every amino acid residue can influence a protein's overall stability, making stability highly susceptible to change throughout evolution. We consider the distribution of protein stabilities evolutionarily permittable under two previously reported protein fitness functions: flux dynamics and misfolding avoidance. We develop an evolutionary dynamics theory and find that it agrees better with an extensive protein stability data set for dihydrofolate reductase orthologs under the misfolding avoidance fitness function rather than the flux dynamics fitness function. Further investigation with ribonuclease H data demonstrates that not any misfolded state is avoided; rather, it is only the unfolded state. At the end, we discuss how our work pertains to the universal protein abundance-evolutionary rate correlation seen across organisms' proteomes. We derive a closed-form expression relating protein abundance to evolutionary rate that captures Escherichia coli, Saccharomyces cerevisiae, and Homo sapiens experimental trends without fitted parameters.

Natural Selection on the Phase-Separation Properties of FUS during 160 My of Mammalian Evolution.

Protein phase separation can help explain the formation of many nonmembranous organelles. However, we know little about its ability to change in evolution. Here we studied the evolution of the mammalian RNA-binding protein Fused in Sarcoma (FUS), a protein whose prion-like domain (PLD) contributes to the formation of stress granules through liquid-liquid phase separation. Although the PLD evolves three times as rapidly as the remainder of FUS, it harbors absolutely conserved tyrosine residues that are crucial for phase separation. Ancestral reconstruction shows that the phosphorylation sites within the PLD are subject to stabilizing selection. They toggle among a small number of amino acid states. One exception to this pattern is primates, where the number of such phosphosites has increased through positive selection. In addition, we find frequent glutamine to proline changes that help maintain the unstructured state of FUS that is necessary for phase separation. Our work provides evidence that natural selection has stabilized the liquid forming potential of FUS and minimized the propensity of cytotoxic liquid-to-solid phase transitions during 160 My of mammalian evolution.

RNA nucleation by MSL2 induces selective X chromosome compartmentalization.

Confinement of the X chromosome to a territory for dosage compensation is a prime example of how subnuclear compartmentalization is used to regulate transcription at the megabase scale. In Drosophila melanogaster, two sex-specific non-coding RNAs (roX1 and roX2) are transcribed from the X chromosome. They associate with the male-specific lethal (MSL) complex1, which acetylates histone H4 lysine 16 and thereby induces an approximately twofold increase in expression of male X-linked genes2,3. Current models suggest that X-over-autosome specificity is achieved by the recognition of cis-regulatory DNA high-affinity sites (HAS) by the MSL2 subunit4,5. However, HAS motifs are also found on autosomes, indicating that additional factors must stabilize the association of the MSL complex with the X chromosome. Here we show that the low-complexity C-terminal domain (CTD) of MSL2 renders its recruitment to the X chromosome sensitive to roX non-coding RNAs. roX non-coding RNAs and the MSL2 CTD form a stably condensed state, and functional analyses in Drosophila and mammalian cells show that their interactions are crucial for dosage compensation in vivo. Replacing the CTD of mammalian MSL2 with that from Drosophila and expressing roX in cis is sufficient to nucleate ectopic dosage compensation in mammalian cells. Thus, the condensing nature of roX-MSL2CTD is the primary determinant for specific compartmentalization of the X chromosome in Drosophila.

The Combination of IFN ß and TNF Induces an Antiviral and Immunoregulatory Program via Non-Canonical Pathways Involving STAT2 and IRF9.

Interferon (IFN) ß and Tumor Necrosis Factor (TNF) are key players in immunity against viruses. Compelling evidence has shown that the antiviral and inflammatory transcriptional response induced by IFNß is reprogrammed by crosstalk with TNF. IFNß mainly induces interferon-stimulated genes by the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway involving the canonical ISGF3 transcriptional complex, composed of STAT1, STAT2, and IRF9. The signaling pathways engaged downstream of the combination of IFNß and TNF remain elusive, but previous observations suggested the existence of a response independent of STAT1. Here, using genome-wide transcriptional analysis by RNASeq, we observed a broad antiviral and immunoregulatory response initiated in the absence of STAT1 upon IFNß and TNF costimulation. Additional stratification of this transcriptional response revealed that STAT2 and IRF9 mediate the expression of a wide spectrum of genes. While a subset of genes was regulated by the concerted action of STAT2 and IRF9, other gene sets were independently regulated by STAT2 or IRF9. Collectively, our data supports a model in which STAT2 and IRF9 act through non-canonical parallel pathways to regulate distinct pool of antiviral and immunoregulatory genes in conditions with elevated levels of both IFNß and TNF.

Estimating the contribution of folding stability to nonspecific epistasis in protein evolution.

The extent of nonadditive interaction among mutations or epistasis reflects the ruggedness of the fitness landscape, the mapping of genotype to reproductive fitness. In protein evolution, there is strong support for the importance and prevalence of epistasis but the quantitative and relative contribution of various factors to epistasis are poorly known. Here, we determine the contribution of selection for folding stability to epistasis in protein evolution. By combining theoretical estimates of the rates of molecular evolution and the nonlinear mapping between protein folding thermodynamics and fitness, we show that the simple selection for folding stability imposes at least ~30% to ~40% epistasis in long-term protein evolution. Estimating the contribution of governing factors in molecular evolution such as protein folding stability to epistasis will provide a better understanding of epistasis that could improve methods in molecular evolution.

Multi-step regulation of transcription kinetics explains the non-linear relation between RNA polymerase II density and mRNA expression in dosage compensation.

In heterogametic organisms, expression of unequal number of X chromosomes in males and females is balanced by a process called dosage compensation. In Drosophila and mammals, dosage compensation involves nearly two-fold up-regulation of the X chromosome mediated by dosage compensation complex (DCC). Experimental studies on the role of DCC on RNA polymerase II (Pol II) transcription in mammals disclosed a non-linear relationship between Pol II densities at different transcription steps and mRNA expression. An ∼20-30% increase in Pol II densities corresponds to a rough 200% increase in mRNA expression and two-fold up-regulation. Here, using a simple kinetic model of Pol II transcription calibrated by in vivo measured rate constants of different transcription steps in mammalian cells, we demonstrate how this non-linearity can be explained by multi-step transcriptional regulation. Moreover, we show how multi-step enhancement of Pol II transcription can increase mRNA production while leaving Pol II densities unaffected. Our theoretical analysis not only recapitulates experimentally observed Pol II densities upon two-fold up-regulation but also points to erroneous interpretations of Pol II profiles from chromatin immunoprecipitation sequencing (ChIP-seq) or global run-on assays.

Highly expressed genes evolve under strong epistasis from a proteome-wide scan in E. coli.

Epistasis or the non-additivity of mutational effects is a major force in protein evolution, but it has not been systematically quantified at the level of a proteome. Here, we estimated the extent of epistasis for 2,382 genes in E. coli using several hundreds of orthologs for each gene within the class Gammaproteobacteria. We found that the average epistasis is ~41% across genes in the proteome and that epistasis is stronger among highly expressed genes. This trend is quantitatively explained by the prevailing model of sequence evolution based on minimizing the fitness cost of protein unfolding and aggregation. The genes with the highest epistasis are also functionally involved in the maintenance of proteostasis, translation and central metabolism. In contrast, genes evolving with low epistasis mainly encode for membrane proteins and are involved in transport activity. Our results highlight the coupling between selection and epistasis in the long-term evolution of a proteome.

Superoxide dismutase 1 is positively selected to minimize protein aggregation in great apes.

Positive (adaptive) selection has recently been implied in human superoxide dismutase 1 (SOD1), a highly abundant antioxidant protein with energy signaling and antiaging functions, one of very few examples of direct selection on a human protein product (exon); the molecular drivers of this selection are unknown. We mapped 30 extant SOD1 sequences to the recently established mammalian species tree and inferred ancestors, key substitutions, and signatures of selection during the protein's evolution. We detected elevated substitution rates leading to great apes (Hominidae) at ~1 per 2 million years, significantly higher than in other primates and rodents, although these paradoxically generally evolve much faster. The high evolutionary rate was partly due to relaxation of some selection pressures and partly to distinct positive selection of SOD1 in great apes. We then show that higher stability and net charge and changes at the dimer interface were selectively introduced upon separation from old world monkeys and lesser apes (gibbons). Consequently, human, chimpanzee and gorilla SOD1s have a net charge of -6 at physiological pH, whereas the closely related gibbons and macaques have -3. These features consistently point towards selection against the malicious aggregation effects of elevated SOD1 levels in long-living great apes. The findings mirror the impact of human SOD1 mutations that reduce net charge and/or stability and cause ALS, a motor neuron disease characterized by oxidative stress and SOD1 aggregates and triggered by aging. Our study thus marks an example of direct selection for a particular chemical phenotype (high net charge and stability) in a single human protein with possible implications for the evolution of aging.

Functional interplay between MSL1 and CDK7 controls RNA polymerase II Ser5 phosphorylation.

Proper gene expression requires coordinated interplay among transcriptional coactivators, transcription factors and the general transcription machinery. We report here that MSL1, a central component of the dosage compensation complex in Drosophila melanogaster and Drosophila virilis, displays evolutionarily conserved sex-independent binding to promoters. Genetic and biochemical analyses reveal a functional interaction of MSL1 with CDK7, a subunit of the Cdk-activating kinase (CAK) complex of the general transcription factor TFIIH. Importantly, MSL1 depletion leads to decreased phosphorylation of Ser5 of RNA polymerase II. In addition, we demonstrate that MSL1 is a phosphoprotein, and transgenic flies expressing MSL1 phosphomutants show mislocalization of the histone acetyltransferase MOF and histone H4 K16 acetylation, thus ultimately causing male lethality due to a failure of dosage compensation. We propose that, by virtue of its interaction with components of the general transcription machinery, MSL1 exists in different phosphorylation states, thereby modulating transcription in flies.

Tracking evolution of myoglobin stability in cetaceans using experimentally calibrated computational methods that account for generic protein relaxation.

The evolution of cetaceans (whales, dolphins, and porpoises) from land to water is one of the most spectacular events in mammal evolution. It has been suggested that selection for higher myoglobin stability (∆G of folding) allowed whales to conquer the deep-diving niche. The stability of multi-site protein variants, including ancient proteins, is however hard to describe theoretically. From a compilation of experimental ∆∆G vs. ∆G we first find that protein substitutions are subject to large generic protein relaxation effects. Using this discovery, we develop a simple two-parameter model that predicts multi-site ∆∆G as accurately as standard methods do for single-site mutations and reproduces trends in contemporary myoglobin stabilities. We then apply this new method to the study of the evolution of Mb stability in cetaceans: With both methods the main change in stability (about 1kcal/mol) occurred very early, and stability was later relaxed in dolphins and porpoises, but was further increased in the sperm whales. This suggests that single proteins can affect whole organism evolution and indicates a role of Mb stability in the evolution of cetaceans. Transition to the deep-diving niche probably occurred already in the ancestor of contemporary baleen and toothed whales. In summary, we have discovered generic stability relaxation effects in proteins that, when incorporated into a simple model, improves the description of multi-site protein variants.

The influence of selection for protein stability on dN/dS estimations.

Understanding the relative contributions of various evolutionary processes-purifying selection, neutral drift, and adaptation-is fundamental to evolutionary biology. A common metric to distinguish these processes is the ratio of nonsynonymous to synonymous substitutions (i.e., dN/dS) interpreted from the neutral theory as a null model. However, from biophysical considerations, mutations have non-negligible effects on the biophysical properties of proteins such as folding stability. In this work, we investigated how stability affects the rate of protein evolution in phylogenetic trees by using simulations that combine explicit protein sequences with associated stability changes. We first simulated myoglobin evolution in phylogenetic trees with a biophysically realistic approach that accounts for 3D structural information and estimates of changes in stability upon mutation. We then compared evolutionary rates inferred directly from simulation to those estimated using maximum-likelihood (ML) methods. We found that the dN/dS estimated by ML methods (ωML) is highly predictive of the per gene dN/dS inferred from the simulated phylogenetic trees. This agreement is strong in the regime of high stability where protein evolution is neutral. At low folding stabilities and under mutation-selection balance, we observe deviations from neutrality (per gene dN/dS > 1 and dN/dS < 1). We showed that although per gene dN/dS is robust to these deviations, ML tests for positive selection detect statistically significant per site dN/dS > 1. Altogether, we show how protein biophysics affects the dN/dS estimations and its subsequent interpretation. These results are important for improving the current approaches for detecting positive selection.

A model of proteostatic energy cost and its use in analysis of proteome trends and sequence evolution.

A model of proteome-associated chemical energetic costs of cells is derived from protein-turnover kinetics and protein folding. Minimization of the proteostatic maintenance cost can explain a range of trends of proteomes and combines both protein function, stability, size, proteostatic cost, temperature, resource availability, and turnover rates in one simple framework. We then explore the ansatz that the chemical energy remaining after proteostatic maintenance is available for reproduction (or cell division) and thus, proportional to organism fitness. Selection for lower proteostatic costs is then shown to be significant vs. typical effective population sizes of yeast. The model explains and quantifies evolutionary conservation of highly abundant proteins as arising both from functional mutations and from changes in other properties such as stability, cost, or turnover rates. We show that typical hypomorphic mutations can be selected against due to increased cost of compensatory protein expression (both in the mutated gene and in related genes, i.e. epistasis) rather than compromised function itself, although this compensation depends on the protein's importance. Such mutations exhibit larger selective disadvantage in abundant, large, synthetically costly, and/or short-lived proteins. Selection against increased turnover costs of less stable proteins rather than misfolding toxicity per se can explain equilibrium protein stability distributions, in agreement with recent findings in E. coli. The proteostatic selection pressure is stronger at low metabolic rates (i.e. scarce environments) and in hot habitats, explaining proteome adaptations towards rough environments as a question of energy. The model may also explain several trade-offs observed in protein evolution and suggests how protein properties can coevolve to maintain low proteostatic cost.

Positively selected sites in cetacean myoglobins contribute to protein stability.

Since divergence ∼50 Ma ago from their terrestrial ancestors, cetaceans underwent a series of adaptations such as a ∼10-20 fold increase in myoglobin (Mb) concentration in skeletal muscle, critical for increasing oxygen storage capacity and prolonging dive time. Whereas the O2-binding affinity of Mbs is not significantly different among mammals (with typical oxygenation constants of ∼0.8-1.2 µM(-1)), folding stabilities of cetacean Mbs are ∼2-4 kcal/mol higher than for terrestrial Mbs. Using ancestral sequence reconstruction, maximum likelihood and bayesian tests to describe the evolution of cetacean Mbs, and experimentally calibrated computation of stability effects of mutations, we observe accelerated evolution in cetaceans and identify seven positively selected sites in Mb. Overall, these sites contribute to Mb stabilization with a conditional probability of 0.8. We observe a correlation between Mb folding stability and protein abundance, suggesting that a selection pressure for stability acts proportionally to higher expression. We also identify a major divergence event leading to the common ancestor of whales, during which major stabilization occurred. Most of the positively selected sites that occur later act against other destabilizing mutations to maintain stability across the clade, except for the shallow divers, where late stability relaxation occurs, probably due to the shorter aerobic dive limits of these species. The three main positively selected sites 66, 5, and 35 undergo changes that favor hydrophobic folding, structural integrity, and intra-helical hydrogen bonds.