Description

Author summary Proteins fold into three-dimensional structures that are essential for their function. Because these structures depend on interactions among amino acids, the fitness effect of a mutation at one site can depend on the amino acid states at other sites. Such dependencies constrain the paths that protein evolution can take, whether evolution proceeds neutrally or adtively. Although intramolecular epistasis has been demonstrated in microbial systems and in vitro, direct experimental evidence for such constraints in diploid multicellular organisms in vivo is rare. Here, we identify Drosophila proteins that exhibit clusters of closely-spaced putatively adtive amino acid substitutions and focus experimental analyses on one example, Trio, a Rho guanine nucleotide exchange factor. Using genome editing to reconstruct evolutionary intermediates in the D. melanogasterlineage, we find that all intermediate versions of Trio reduce viability or impair locomotor function. Importantly, these harmful effects are recessive, suggesting that they could be masked when paired with an ancestral version of the protein. This implies that individual amino acid changes may persist in heterozygotes within populations and later combine to contribute to adtation. If the recessivity of deleterious intermediates along adtive evolutionary paths proves to be widespread, it could have important implications for our mechanistic understanding of adtive protein evolution.

Content

?
This is an uncorrected proof.
Figures
Abstract
Intramolecular epistasis is increasingly recognized as a key factor shing patterns of evolutionary rate variation among protein sites and constraining adtive evolution. While genome-wide analyses have revealed that intramolecular epistatic interactions can drive the spatial clustering of amino acid substitutions, direct empirical evidence for such interactions and their evolutionary consequences remains limited. Using a population genetic screen for spatially-clustered and lineage-specific adtive amino acid substitutions in Drosophila proteins, we systematically identify experimentally tractable candidates for functional analysis. As proof of concept, we focus on the Trio protein, a Rho guanine nucleotide exchange factor that exhibits three spatially-clustered putatively adtive amino acid substitutions in the D. melanogaster lineage. By systematically reconstructing evolutionary intermediates in vivo using genome editing, we find that all possible intermediate states exhibit reduced viability and/or locomotor defects, providing strong evidence for epistatic constraints on evolutionary trajectories. Notably, these deleterious effects are recessive, suggesting that intermediate combinations of epistatically interacting amino acid substitutions can accumulate in heterozygotes prior to fixation, thereby circumventing parent constraints imposed by maladtive intermediate states. Together, these findings provide a rare empirical view of the fitness landsce shed by intramolecular epistasis and establish a framework for investigating the constraints on adtive protein evolution in diploid multicellular organisms.
Author summary
Proteins fold into three-dimensional structures that are essential for their function. Because these structures depend on interactions among amino acids, the fitness effect of a mutation at one site can depend on the amino acid states at other sites. Such dependencies constrain the paths that protein evolution can take, whether evolution proceeds neutrally or adtively. Although intramolecular epistasis has been demonstrated in microbial systems and in vitro, direct experimental evidence for such constraints in diploid multicellular organisms in vivo is rare. Here, we identify Drosophila proteins that exhibit clusters of closely-spaced putatively adtive amino acid substitutions and focus experimental analyses on one example, Trio, a Rho guanine nucleotide exchange factor. Using genome editing to reconstruct evolutionary intermediates in the D. melanogasterlineage, we find that all intermediate versions of Trio reduce viability or impair locomotor function. Importantly, these harmful effects are recessive, suggesting that they could be masked when paired with an ancestral version of the protein. This implies that individual amino acid changes may persist in heterozygotes within populations and later combine to contribute to adtation. If the recessivity of deleterious intermediates along adtive evolutionary paths proves to be widespread, it could have important implications for our mechanistic understanding of adtive protein evolution.
Citation: Borne F, Taverner AM, Andolfatto P (2026) Epistasis among clustered lineage-specific amino acid substitutions in the Drosophila Trio protein. PLoS Genet 22(6): e1012175. /> Editor: Colin Meiklejohn, University of Nebraska-Lincoln, UNITED STATES OF AMERICA
Received: October 10, 2025; Accepted: May 21, 2026; Published: June 3, 2026
Copyright: © 2026 Borne et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Raw data for functional assays and computational analysis pipeline and results are deposited at /> Funding: This work was funded by the National Institutes of Health R01 GM115523 to P.A. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The rate of protein evolution varies substantially among species, among proteins and among sites within proteins. This rate variation has historically been interpreted in terms of neutral, or nearly neutral models of protein evolution, which depend on the proportion of newly arising amino acid altering mutations that are neutral with respect to protein function or effectively neutral due to genetic drift [1,2]. While nearly neutral models can account for several important aspects of protein evolution [3], a debate raged for decades about the relative contribution of positive selection to protein evolution. The development and plication of the McDonald-Kreitman test [4] and similar frameworks [5–7] suggested that protein divergence is well in excess of that predicted under neutral models of protein evolution in Drosophila and other species [5,8–13]. While the McDonald-Kreitman and similar tests suffer from uncertainty of the population size over time [14,15], amino acid substitutions also exhibit genetic hitchhiking signatures at closely linked neutral sites, confirming an important role for positive selection in protein evolution [11,16–20].
Evolutionary rate variation within proteins is primarily believed to arise from differences in selective constraint, reflecting the need to maintain the integrity of specific protein domains and their functions [21–23]. However, since positive selection also contributes substantially to protein divergence, this raises the possibility that clustered substitutions may sometimes result from repeated rounds of adtive substitution in particular functional regions, or from the co-fixation of nearly neutral variants through genetic hitchhiking. Indeed, several studies have reported that positively selected amino acid substitutions tend to occur in close proximity within protein structures [24–29].
Most of the population genetic models assume that sites contribute independently to fitness. However, when considering the three-dimensional structure of proteins and their role in determining function, it becomes clear that amino acid residues act collectively to produce a coherent functional state. As sequences evolve, this coherence must be maintained or, when novel functions arise, re-established. Thus, residues are constrained to evolve non-independently, and the phenotypic and fitness effects of new amino acid–altering mutations depend on the history of substitutions at interacting sites. This dependence, termed intramolecular epistasis, is expected to play an important role in constraining protein evolutionary trajectories. This principle was first formalized in the covarion model, which proposed that only a subset of codons is free to vary at any given time and that this set shifts as substitutions occur [30–32]. Since spatially proximate amino acids are more likely to directly interact, substitutions that compensate or permit one another are expected to co-occur closely in sequence space or atomic distance in folded proteins. Consistent with this prediction, recent studies have revealed that spatially clustered amino acid substitutions exhibit characteristic signatures of epistasis and compensatory evolution [33–38]. Together, these findings suggest that intramolecular epistasis not only contributes to variation in evolutionary rate but also leaves a distinct spatial footprint in the form of clustered substitutions.
Experimental work has further confirmed the central role of intramolecular epistasis in protein evolution. A landmark study by Weinreich et al. (2006) showed that interactions among five mutations in β-lactamase dramatically constrained the number of traversable evolutionary paths to antibiotic resistance [39]. Similarly, Ortlund et al. (2007) demonstrated that the glucocorticoid receptor evolved new hormone specificity in vertebrates via stability-mediated epistasis, in which nearly-neutral permissive mutations fixed stochastically and stabilized local structures, enabling subsequent adtive substitutions to fix [40]. Later studies using ancestral protein reconstruction, experimental evolution, and large-scale mutational scans have confirmed that epistatic interactions can strongly influence evolutionary trajectories [41–48]. Furthermore, a few studies provide evidence of epistatic substitutions occurring in close spatial proximity, echoing the clustering patterns predicted from statistical analyses. For example, compensatory mutations have been observed near adtive substitutions conferring toxin resistance in the poison frog nicotinic acetylcholine receptor [49], and similar patterns pear along the evolution of vertebrate proteins like p53, hemoglobin, lysozyme and Na ⁺ ,K ⁺ -ATPase [50–54].
This experimental work has been instrumental in demonstrating the pervasiveness of epistasis and in revealing key principles of higher-order and global epistasis governing fitness landsces. However, most of these studies have explored epistatic dynamics in the context of protein activity and stability in vitro [41–44,55,56], or microbial fitness in vivo [46–48]. Consequently, a substantial g still exists in our understanding of how interacting substitutions accumulate in diploid multicellular organisms where dominance and interactions across multiple phenotypic levels may also strongly influence evolutionary outcomes.
To our knowledge, Na⁺,K⁺-ATPase (NKA) resistance to cardiotonic steroids (CTS) in insects is the only case where epistasis in protein adtation has been directly demonstrated in vivo in a multicellular organism [57–59]. In CTS-adted insects, adtation frequently targets a highly conserved domain of the NKA alpha-subunit. The most common resistance-conferring substitutions—at positions 111 and 122—almost always occur after a permissive substitution at position 119 (A119S), which reduces the negative pleiotropic effects of these resistance mutations [57,58]. Moreover, in vivo engineering of firefly CTS-resistance substitutions into Drosophila revealed similar deleterious pleiotropic effects, which were alleviated by additional background substitutions that occur within a span of 12 amino acid residues, but are not directly related to CTS resistance [59]. Further, it was shown that the deleterious effects of CTS resistance-conferring substitutions pear to be recessive, potentially explaining how they might persist in populations long enough to allow secondary compensating substitutions to arise and enable their fixation. This work highlighted the use of in vivo models in revealing pleiotropic effects on higher order phenotypes that are more closely related to organismal fitness (e.g., viability, fertility, and behavior) and would be missed in in vitro assays.
The case of CTS-resistant NKAs is a striking example of the role of epistasis in protein adtation, but it remains unclear how broadly these findings ply to other proteins. To better understand how pleiotropy, epistasis, and dominance she protein evolution during adtation, new diploid in vivo models are needed. To this end, we systematically searched the Drosophila melanogaster genome for proteins that, like NKA, are highly conserved yet contain lineage-specific clusters of positively-selected amino acid substitutions. As a case study, we focus on one such cluster in the highly-conserved protein Trio, a Rho guanine nucleotide exchange factor involved in signalling pathways that regulate several important developmental processes. Using scarless CRISPR-Cas9 genome editing, we engineered Drosophila strains carrying all possible intermediate combinations of substitutions in this cluster, enabling us to dissect the epistatic and dominance interactions involved.
Results
Lineage-specific spatially-clustered amino acid substitutions in D. melanogaster proteins
To identify proteins with lineage-specific, spatially-clustered and putatively adtive amino acid substitutions in D. melanogaster, we plied the Model-Averaged Site Selection with Poisson Random Field (MASS-PRF; [26]) to multiple sequence alignments for 9137 Drosophila proteins (See Methods). MASS-PRF estimates a site-by-site scaled selection coefficient (γ = 2Ns) based on polymorphism and divergence clustering models. While the McDonald-Kreitman test is typically plied to whole proteins, MASS-PRF is designed to quantify variation in intensities of selection across a protein sequence, highlighting potentially localized targets of adtive protein evolution. We identified 919 proteins containing at least one cluster of putatively adtive substitutions specific to the D. melanogaster lineage (Figs 1 and S1, See Methods). The identification of these spatially-clustered amino acid substitutions provides the opportunity to investigate potential epistatic interactions among individual substitutions within such clusters. We estimated that 679 proteins contain experimentally tractable clusters consisting of 2 or 3 substitutions located less than 20 amino acids away from each other.
Plotted are profiles of selection intensity (γ = 2Ns) across four D. melanogaster proteins inferred with MASS-PRF [26]. The black line corresponds to the model-averaged γ and the grey areas indicate 95% model uncertainty interval. Red lines indicate regions for which the 95% lower bound of γ > 4, which corresponds to a false positive rate of <0.1 (see Methods). Protein structure models are represented below each selection intensity plot. Functional domains are represented by colored boxes with their names in black. Lineage-specific amino acid substitutions are represented by dark gray dots and red for the cluster of interest (D. melanogaster above and D. simulans below, respectively). AlphaFold predictions of the 3D structure of the D. melanogaster proteins are inset. The prediction has been superimposed onto the structures of the homologous proteins (transparent) with ligands and in oligomerized form, when available (See enlarged structure S2 Fig). The positions of the clustered amino acid substitutions are highlighted with red arrows. (A) Trio, a Rho guanine nucleotide exchange factor. The AlphaFold prediction of the first DH-PH domain of Trio is represented in green and is superimposed onto the human Trio DH-PH structure (transparent green) in complex with Rac1 (blue) (PDB 7SJ4, [67]). (B) Fragile X messenger ribonucleoprotein 1 (Fmr1). (C) CG3544, a xylulokinase. The AlphaFold prediction is represented in green and is superimposed on the xylulokinase structure of Chromobacterium violaceum (transparent orange and green) in complex with ADP (pink) (PDB 3KZB). (D) Phosphogluconate dehydrogenase (Pgd). The AlphaFold structure prediction is represented in green and is superimposed onto the dimerized structure of the human Pgd (transparent orange and green) in complex with NADP (pink) (PDB 2JKV).
/> As proof of concept, we focus on a putatively adtive cluster of amino acid substitutions in the protein Trio (Fig 1A) as a tractable example. Trio is a Rho-guanine nucleotide exchange factor (GEF) that regulates several signaling pathways involved in functions such as neuronal migration, axonal outgrowth, axon guidance, and syntogenesis in D. melanogaster [60,61], C. elegans [62] and in humans [63,64]. Missense and nonsense mutations in trio can cause neurodevelopmental diseases such as intellectual disability and autism spectrum disorders in humans [63]. In D. melanogaster, trio mutants show specific neurodevelopmental disorders resulting in locomotion behavior, leg and wing formation defects but also more general phenotypes such as prepupal lethality and sterility [65,66].
The D. melanogaster and D. simulans Trio proteins differ by 11 amino acid substitutions, six of which occur along the D. melanogaster lineage (Figs 1A and S3). A cluster of three amino acid substitutions (V1315A; N1318S; R1335K), exhibiting lineage-specific positive selection in D. melanogaster, is located in the first GTPase binding domain DH-PH of the protein. This domain is otherwise highly conserved and these three substitutions are the only amino acid changes observed across the D. melanogaster-D. yakuba species group, which shared a common ancestor 10 Mya [67–69] (Figs 1A and S3). Further, these sites are monomorphic among 947 wild-derived D. melanogaster genome sequences from 30 populations worldwide [70], these three substitutions are monomorphic and we detected only three nonsynonymous singleton variants across the domain, indicating strong purifying selection.
While this cluster of amino acid substitutions are inferred to have been fixed by positive selection using the site-specific MASS-PRF proach, there is no additional evidence for recent or recurrent positive selection acting on the Trio protein. For example, plication of the standard McDonald-Kreitman test and a branch-site model (implemented in PAML [71]) to the entire protein do not reject neutrality and there is no obvious signature of recent selective sweep (S4 Fig). Further, the amino acid substitutions in the Trio cluster pear to be relatively conservative, preserving both polarity and charge properties (S3 Fig), though there is a trend towards incorporation of smaller functional groups. Prediction tools assessing the impact of single substitutions on protein folding-stability were plied to the D. melanogaster AlphaFold-predicted structure of the DH-PH domain [72–74]. These analyses did not identify any substitutions predicted to strongly destabilize the DH-PH domain, with all predicted ΔΔG values ranging between –1 and +1 kcal/mol (S1 Table). We note, however, that these predictions rely on a modeled AlphaFold structure and should therefore be interpreted with caution.
Strong epistasis among clustered amino acid substitutions in the Trio protein
To investigate potential phenotypic effects associated with the three amino acid substitutions in the Trio DH-PH domain (V1315A, N1318S and R1335K), we implemented scarless CRISPR-cas9/piggyBAC genome editing to the native Trio protein in D. melanogaster (see Methods). We edited D. melanogaster strains representing all possible combinations of this cluster of substitutions in the Trio protein from the ancestral state VNR to the current state ASK (Fig 2C). To avoid effects of genomic background, we homogenized the lines by crossing to well-characterized inbred strains (see Methods and S5 Fig).
(A) Viability of homozygous and heterozygous first instar larvae, pupae and adults. Plotted is the proportion of YFP- individuals normalized relative to the ASK hlotype (see Methods). Asterisks indicate adjusted p-values determined by Fisher’s Exact tests comparing each genotype to the ASK hlotype (*** p < 0.001; ** p < 0.01; * p 0.05). (C) Diagram representing all available genetic sequential substitution paths from the ancestral hlotype VNR to the extant D. melanogaster hlotype ASK. Colors of the circle represent the fitness of each hlotype based on viability and locomotion results.
/> To identify fitness effects associated with the three substitutions, and potential epistatic interactions among them, we first investigated homozygous effects of substitutions at these three sites on two fitness-related traits (viability and locomotion). Interestingly, most of the intermediate hlotypes showed severe defects in larval and adult viability. In particular, VNK, ASR and ANK flies are homozygous lethal and most individuals do not reach pupal stage (Fig 2A). For VSR, some homozygous individuals reached adulthood (6 males and 1 female) but failed to produce progeny when presented with fertile mates (suggesting that they may be sterile; See Methods). Only ANR, VSK and the triple reversion VNR (i.e., representing the ancestral hlotype) showed viabilities similar to the current ASK hlotype of D. melanogaster. Additionally, while the intermediate hlotypes ANR and VSK did not show viability defects, they do exhibit defects in adult locomotion (Fig 2B). Thus, it pears that all three amino acid substitutions are involved in epistatic interactions, and it is parent that all possible paths from ancestral to the current D. melanogaster state would involve transiting through a less fit intermediate state (Fig 2C).
Motivated by these findings, we next asked about the dominance of the deleterious effects associated with intermediate states. To examine dominance effects, we created heterozygotes (x/VNR) carrying one intermediate state (x) and one ancestral hlotype (VNR). We found that all intermediate states have viabilities that are not distinguishable from that of the homozygous VNR or ASK hlotypes (Fig 2A). Similarly, all the intermediate hlotypes lacked locomotion defects when heterozygous (Fig 2B). Together, these results suggest that all deleterious effects associated with intermediate states are recessive. The finding that the deleterious effects of substitutions are recessive has important implications for how we expect this cluster of substitutions to have become established in the D. melanogaster lineage. Notably, all paths involving sequential substitution of amino acid substitutions would be deleterious, however, each of these substitutions may have persisted as polymorphisms in the population for some time, assembled into a favorable hlotype and fixed concomitantly (Fig 3). Despite the locomotor defects associated with ANR and VSK (Fig 2B), the fitness effects of these defects may be small. If so, the formation of heterozygotes involving these hlotypes (i.e., x/ANR or x/VSK, where x is any other derived hlotype) could provide additional permissible evolutionary paths. Given this possibility, for a subset of cases we tested whether flies that are heterozygous for two derived hlotypes are viable (S2 Table). We find that ASR/ANR and ANK/ANR pear to be perfectly viable, as are VSK/VNK heterozygotes. While we did not test them for fertility, the viability of these genotypes has important implications for how the fitness landsce from VNR to ASK may have been traversed.
Shown is a schematic representation of the fitness landsce for the amino acid substitutions in Trio. The x-axis represents the number of mutational steps from the ancestral hlotype (VNR) to the extant D. melanogaster hlotype (ASK). The y-axis represents fitness based on viability and locomotion phenotypes of each hlotype in (A) homozygous D. melanogaster flies or (B) heterozygous flies carrying one ancestral VNR hlotype.
/> As this cluster of substitutions is inferred to be positively selected (Fig 1A), we might expect the ASK hlotype to confer a fitness advantage relative to the ancestral hlotype VNR in D. melanogaster. However, the viability and locomotion phenotypes of homozygous VNR flies are not distinguishable from ASK flies. We also find that levels of fertility of the VNR and ASK strains are not distinguishable (S6 Fig). This implies that either the ASK hlotype is not strongly beneficial (potentially even neutral) relative to the VNR hlotype, or that a fitness benefit occurs for a phenotype (or in an environment or an ancestral genomic background) that we have not investigated. While we failed to detect a difference in viability and fertility between the VNR and ASK hlotypes, it should be noted that we are underpowered to detect effects that are smaller than 30% given our sample sizes and 10% if our sample sizes were 10-fold larger (S6 Fig). In D. melanogaster populations, a 1% effect would already correspond to a very strong fitness difference in population genetic terms.
Discussion
Adtive protein evolution is widespread in plant and animal genomes and has a major impact on levels of genome-wide variability. Most inferences about this process rest on the assumptions that amino acid substitutions act independently and that the intensity of selection on them remains constant. Increasing evidence, however, suggests that interactions among substitutions—especially those in close physical proximity—may be common. These interactions raise an unresolved question: are clusters of closely linked amino acid substitutions, whether adtive or not, typically fixed sequentially or simultaneously within a species? Despite its importance, direct empirical evidence for such epistatic interactions among spatially-clustered substitutions remains scarce, particularly in multicellular eukaryotes, where most reports have been anecdotal.
To address this g, we systematically searched for proteins with experimentally tractable clusters of amino acid substitutions that are inferred to be fixed by positive selection. Among the candidates, we identified Trio, a Rho guanine nucleotide exchange factor, as a promising test case. Our analyses suggest that the cluster of three lineage-specific amino acid substitutions in Trio are unlikely to have fixed sequentially along the Drosophila melanogaster lineage. Instead, negative epistatic interactions among residues, together with the recessive deleterious effects of the derived variants, depending on epistatic interactions among them, make it more likely that all three substitutions fixed together as a single hlotype. This work provides one of the few direct demonstrations that spatially-clustered substitutions in a multicellular eukaryote can interact epistatically and have measurable effects on whole-organism phenotypes relevant to fitness.
The utility of in vivo studies in Drosophila
Most experimental studies of epistasis have been performed in vitro. These proaches are powerful because they allow systematic exploration of epistatic interactions within a protein. For example, deep mutational scanning can efficiently m interactions among sites and reconstruct a protein’s fitness landsce [44,55,56]. Such studies have been invaluable for uncovering principles of protein structure–function relationships and for providing quantitative descriptions of higher-order epistasis. However, the phenotypes measured in vitro—such as enzymatic activity or ligand binding—are tied to the specific function of the protein being studied and can be difficult to interpret in the context of their effects on organismal fitness. Moreover, many proteins have functions that are not amenable to straightforward biochemical assays.
In contrast, in vivo studies – most of which have been carried out in microbial systems [46–48] – provide the opportunity to measure phenotypes more directly relatable to fitness (for example, growth rates). They also provide a way to investigate amino acid substitutions in proteins that are otherwise difficult to assay biochemically, thereby broadening the range of proteins that can be experimentally explored. While microbial systems remain important, key differences between microbes and multicellular diploid organisms have important implications for the mechanistic basis for constraints on adtation. In particular, multicellular organisms display more complex, higher-order phenotypes—such as those related to development, survival, behavior and reproduction—that in turn increase the complexity of the potential epistatic, pleiotropic and dominance dependencies of adtive protein evolution. Our study represents one of the few to test the in vivo effects of specific adtive amino acid substitutions on phenotypic traits like viability, fertility, and locomotion in the context of a multicellular organism (see also [75]). Such studies provide a framework for more systematically exploring constraints on adtive protein evolution in the context of diploid multicellular organisms.
We note, however, that in vivo phenotypic analyses are inherently context dependent. While no additional substitutions occur within the DH-PH domain or in the interacting GTPases Rac1 and Rac2 along the D. melanogaster lineage, three other amino acid changes did occur elsewhere in the Trio protein. Additionally, we conducted our experiments in a single, contemporary D. melanogaster genomic background under laboratory conditions. The deleterious and epistatic effects observed for the intermediate hlotypes are therefore conditional on this specific genetic, genomic and environmental context. It is possible that these intermediates were tolerated in an ancestral genomic background or under alternative ecological conditions. Such caveats are an inherent limitation of all in vivo studies.
Implications for the dynamics of adtation
Although intramolecular epistasis is recognized as an important factor of protein evolution [52,76,77], how groups of interacting amino acid substitutions arise and reach fixation in functional regions of proteins without compromising fitness remains poorly understood. In principle, epistasis can allow sets of mutations that are individually deleterious or neutral to confer a benefit when combined, yet the evolutionary paths by which such groups of substitutions become established are often opaque. A number of previous studies that have directly investigated the timing and dynamics of epistatic mutation fixation generally support a sequential fixation model [43,45,61,78,79]. In such models, mutations fix in a stepwise manner, with each substitution either being neutral or slightly deleterious until the beneficial combination is assembled. This question has largely been unexplored in the context of diploid organisms where dominance relationships of alleles may also play an important role.
For the cluster of amino acid substitutions in Trio, we show that individual substitutions, as well as pairs, are deleterious when present in the homozygous state. This implies that any evolutionary trajectory involving the stepwise, sequential substitution of these amino acids would need to traverse through strongly deleterious intermediate states (Fig 3), making such paths highly unlikely. These results therefore provide strong evidence that this cluster did not arise through a process of sequential fixation. Instead, since these deleterious effects are recessive, we propose that the cluster was more likely to have assembled at the polymorphic phase, with all three substitutions (ASK) eventually fixing concomitantly. Furthermore, ANR and VSK homozygotes are viable (Fig 2), as are ASR/ANR, ANK/ANR and VSK/VNK heterozygotes (S3 Table). If other heterozygotes of two derived hlotypes are also viable, this could potentially open additional evolutionary pathways. While waiting times for sequential mutation are expected to be long, the viability of heterozygotes implies that recombination could generate new hlotypes, reducing the time by orders of magnitude [80].
Theoretical modelling suggests that in large populations, deleterious intermediate genotypes can persist at low frequency long enough for a compensatory mutation to arise, restoring fitness and allowing the full combination to spread and reach fixation [81–83]. In diploids, recessive lethal hlotypes can persist in the population for many generations (S7 Fig), providing an opportunity for a second mutation or a recombination event to occur. This dynamic can lead to the fixation of the final hlotype at a substantially higher rate than the neutral expectation, even when the fitness advantage of the final hlotype is small [81].
Furthermore, if recessive deleterious hlotypes also confer a dominant fitness advantage (a.k.a. a dominance reversal), they can be maintained indefinitely and at a moderate frequency (S7 Fig), further enhancing the formation of favorable epistatic combinations. This scenario is comparable to a dominance shift following an environmental change described by [84]. Consistent with this view, several recent examples of adtive protein evolution have documented dominance reversals, where adtive mutations exhibit dominant beneficial effects but recessive pleiotropic costs. Such a pattern has been observed in the evolution of Na + ,K + -ATPase resistance to cardiotonic steroids [58], GABA receptor resistance to terpenoids [85], and acetylcholinesterase resistance to organophosphates [86]. The dominance of the beneficial effects associated with adtive substitutions at Trio is not known, but given the recessivity of deleterious intermediate states, it may represent a similar example of adtive protein evolution.
Notably, the scenarios described above assume that mutations arise independently. An alternative way to traverse fitness valleys caused by deleterious intermediates is that all three substitutions (or two of the three) arose simultaneously in a single mutational event. Multi-nucleotide mutational events have been documented in several systems [87–91], and such events could facilitate the coordinated emergence of interacting amino acid changes. Based on Drosophila mutation-accumulation lines and human parent-offspring trios, multi-nucleotide mutational events account for about 3% of the de novo mutations in the germline across eukaryotes [88,91]. While this mechanism could, in principle, account for the pattern observed in Trio, it seems unlikely that a single mutational event would generate multiple substitutions that are individually deleterious but jointly neutral or beneficial. In any case, multi-nucleotide mutational events may contribute to the clustered patterns observed in some of our other candidate genes. Notably, the documented excess of spatially clustered amino acid substitutions in Drosophila proteins persists between substitutions separated by introns, suggesting that multi-nucleotide mutation alone cannot fully explain the observed clustering pattern [34]. Thus experimentally examining clusters of substitutions that are separated by introns may help distinguish substitutions that arose independently from those generated by a single mutational event.
Further, although our candidate clustered substitutions are inferred to have been fixed by positive selection using a McDonald-Kreitman-base proach, this signal does not distinguish between strict adtation—where the derived hlotype confers a fitness advantage relative to the ancestral state—and compensatory evolution. Several studies have highlighted genome-wide signatures of compensatory protein evolution [33,36,77,92,93]. Compensatory evolution can also produce clustering patterns, as compensatory substitutions are often located closer to the focal mutations [33,36].
While some of our candidates for proteins with putatively adtive clusters of amino acid substitutions may be due to compensatory interactions, compensatory dynamics and positive selection are not mutually exclusive hypotheses. In the case of Trio, given the severely deleterious states of most intermediates, it seems unlikely that the fitness of the extant hlotype ASK is effectively neutral with respect to fitness. Despite this, based on the current data, neutral fixation cannot be ruled out. Importantly, even if many of the clusters we identified, including in Trio, are compensatory rather than adtive, they remain highly informative. Understanding how such interacting substitutions fix—and how they she patterns of molecular variation—is critical for interpreting population genetic signals of selection and for clarifying the mechanisms underlying protein evolution.
Implications for the interpretation of population genetic inferences
Over the past several decades, population genetics has undergone a paradigm shift in how we understand the role of positive selection in protein evolution and its impact on genomic variation in plants and animals [8,12]. In Drosophila, estimates suggest that 50–90% of amino acid divergence exceeds neutral–deleterious model predictions and is therefore attributed to positive selection [20,25,94,95]. Independent support for adtive substitutions comes from hitchhiking signatures near individual amino acid substitutions [18], but these analyses tend to yield somewhat lower estimates of the proportion of divergence that is adtive.
If, however, epistasis among amino acid substitutions (as observed for Trio) is common in protein evolution, the McDonald–Kreitman framework may overestimate the proportion of adtive divergence [96]. This is because the framework assumes independence among sites and a constant direction and strength of selection over time, both of which are violated when substitutions interact epistatically. Although this issue has been raised conceptually, the consequences of pervasive epistasis for interpreting McDonald–Kreitman results remain largely unexplored [96].
More broadly, widespread non-independence among amino acid substitutions could alter how we interpret their effects on genome-wide diversity. In standard models of genetic hitchhiking, reductions in linked neutral diversity are attributed to substitutions fixing independently [18,97,98]. However, intuitively, if several closely linked substitutions fix simultaneously rather than sequentially, the parent average strength of selection per substitution inferred from local reductions in linked diversity would be underestimated, since its impact would be distributed across multiple fixed sites. How such non-independence—particularly in light of the dominance and epistasis uncovered in the case of Trio—affects inferences of population genetic parameters is an important open question for future research.
Conclusions
By studying a cluster of putatively adtive amino acid substitutions in the D. melanogaster protein Trio, we uncovered surprisingly strong deleterious epistatic interactions among residues of this cluster. Because these deleterious effects are recessive, they are more likely to have persisted in the population as polymorphisms before combining into a beneficial hlotype that eventually became fixed in the species. This may be a common phenomenon in adtive protein evolution and has important implications for how we interpret its associated population genetic signatures. This case study lays the groundwork for more systematic in vivo studies of putatively adtive amino acid substitution, where the fitness effects of mutations can be measured at multiple levels, including whole-organism phenotypes related to fitness. Such studies will help clarify how epistasis, pleiotropy, and dominance interact to she protein evolution, deepening our understanding of the evolutionary constraints that govern evolutionary trajectories in multicellular diploid organisms.
Methods
Identification of adtive clusters
We generated multiple sequence alignments of orthologous proteins using D. melanogaster ISO1 (GCA_000001215.4), D. simulans w501 (GCA_016746395.2), D. yakuba NY73PB (GCA_016746365.2), D. santomea STO CAGO 1482 (GCA_016746245.2) and D. teissieri GT53w (GCA_016746235.2) One-to-one orthologs were identified using a reciprocal best-hit exonerate proach [35,99]. Predicted D. melanogaster (strain ISO1) proteins were aligned to each genome sequence to extract candidate proteins, and each extracted protein was subsequently realigned to its corresponding D. melanogaster protein. The longest predicted transcript for each protein in D. melanogaster was used to represent each protein and multispecies alignment was created using PRANK (-codon parameter, v.170427; [100]). Ancestral sequences were inferred based on this alignment and the species tree using PAML’s baseml function (v 4.9; [71]) with the following parameters: model = 7, kpa = 1.6, RateAncestor = 2. MACSE (- max_refine_iter 1 parameter, v. 2.07; [101]) was then used to align sequences from the D. melanogaster Zambia population (n = 82; [70]) with the reconstructed D. melanogaster–D. simulans ancestral sequence.
To infer selection on protein sequences, we plied Model Averaged Site Selection via Poisson Random Field (MASS-PRF), which relies on both polymorphism and divergence data [26]. In this framework, sequences from the Zambia population served as polymorphism data, while the reconstructed D. melanogaster–D. simulans ancestral sequence was used as the outgroup to quantify lineage-specific divergence. To prepare the input files required for running MASS-PRF [26], we generated consensus sequences cturing polymorphism and divergence changes for each protein, as follows. To create the polymorphism consensus sequence, monomorphic codons were replaced by *, synonymous codons by S and non-synonymous changes were replaced by R. To create the divergence consensus sequence, outgroup codons that were present in the ingroup sequences were replaced by *, outgroup codons were replaced by S or R if the ingroup codon was fixed and the outgroup codon was synonymous or non-synonymous to it respectively. If the ingroup codon was polymorphic and the outgroup codon was different from both of these, the codon was replaced by -. For proteins longer than 900 amino acids, Consensus sequences were split into fragments of 900 codons or smaller.
MASS-PRF was run on each protein fragment using parameters –ic 1 –sn 82 -o 1 -r 1 -ci_r 1 -ci_m 1 -s 1 -exact 0 -mn 30000 -t 2. Of 9137 proteins analyzed, MASS-PRF ran successfully for 5355 for at least one fragment per protein (Across all fragments, 5743: successful run; 4082: too little substitution information to run; 748: out of memory). Successful runs were screened for positive selection. A site was estimated to be under positive selection if the lower 95% confidence bound for gamma > 4, which corresponds to a false positive rate <0.1 [26]. Of the 5355 proteins for which runs were successful, 1144 had at least one site under positive selection. Among them, 919 harbors clustered substitutions, defined by at least two sites less than 20 codons art. Within this set, we found that 679 proteins contain at least one potential experimentally tractable cluster, corresponding to a total of 795 clusters. We defined a cluster as experimentally tractable if exactly 2 or 3 sites that show divergence and positive selection are less than 20 codons art from each other.
CRISPR/cas9-mediated editing: Plasmid assembly
Genomic DNA was isolated from ywISO8 flies using a squish prep protocol [106]. Homology arms were amplified with PCR from this genomic DNA using Q5 polymerase from NEB (S2 Table). The plasmid backbone and dsRed were amplified from plasmid: pScarlessHD-DsRed, a gift from Kate O’Connor-Giles (Addgene #64703) (S2 Table). The four fragments were combined at an equimolar concentration (150 fmoles each) and assembled using NEBuilder HiFi DNA assembly kit. To verify that assembly was successful, PCR was performed using primers that spanned multiple fragments. The NEBuilder HiFi DNA assembly kit result was used to transform NEB 5-alpha competent E. coli. To verify that the transformation was successful, colony PCR was performed on individual bacteria colonies using pairs of primers that spanned multiple fragments using LongAMP or regular Taq. Final plasmid assembly was verified by a Tn5 tagmentation-based Illumina sequencing protocol [107] followed by de novo assembly. Plasmids were extracted using Qiagen’s QIrep Spin Miniprep kit.
CRISPR/cas9-mediated editing: Site-directed mutagenesis
Starting from a base plasmid sequence representing the current D. melanogaster hlotype (ASK), we used site-directed mutagenesis (SDM) to create all possible intermediates leading back to the ancestral A1315V/S1318N/K1335R (VNR) hlotype. To do this, we used the Agilent QuikChange Lightning Multi Site-Directed Mutagenesis Kit in two reactions: one containing three primers: 1) A1315V, 2) S1318N, and 3) K1335R, and one containing two primers: 1) A1315V, S1318N and 2) K1335R. The standard Agilent kit protocol was followed. First, PCR was performed without the addition of QuikSolution, as it was not necessary. This was followed by DpnI digestion (provided) and transformation into the provided competent cells. Colony PCR using Taq polymerase was performed on 48 bacterial colonies from each reaction (S2 Table). Amplicons from each sample were barcoded with a unique combination of Illumina-compatible i5 and i7 adters and sequenced to assess the presence of the desired mutations. Plasmids were extracted using the Qiagen Plasmid Midi Kit for CRISPR injection. The design for mutation 3 (K1335R) is completely scarless. All the other combinations introduce a single synonymous single nucleotide polymorphism that is naturally present within D. melanogaster populations. Plasmids were obtained for the following combinations: A1315V (VSK), A1315V/S1318N/K1335R (VNR), A1315V/S1318N (VNK), S1318N (ANK), K1335R (ASR), S1318N/K1335R (ANR). One additional hlotype, A1315V/K1335R (VSR) was generated using a second SDM reaction starting from A1315V plasmid to introduce K1335R.
CRISPR/cas9-mediated editing: Injections and post-injection processing
CRISPR/cas9-mediated editing: Background homogenization
To confirm the genomic background of engineered lines, 10 female individuals were collected from each line and frozen at -20°C. Whole DNA was extracted from each fly individually using Quick-DNA 96 Kit (Zymo research). DNA from each fly from the same line was then pooled. Tn5-tagmentation libraries [107] were prepared and indexed with one specific index for each line and enriched by 12 PCR cycles (HS One Taq, NEB). Libraries were then pooled and size selected for fragments around 500nt using AMPure XP (Beckman Coulter) following manual. The pooled libraries were then sequenced with paired-end 150nt on NovaSeq X Plus 10B Illumina 2×150 (Admera Health). Illumina adters were trimmed using trim-galore (-q 0, -e 0.1) (v 0.6.7) and reads were aligned to the D. melanogaster reference genome (GCA_000001215.4) using bwa mem (v 0.7.17-r1188, [110]). Duplicates were marked with picard (v 2.18.29, Broad Institute) and reads were re-aligned with gatk3 (v 3.8-1-0, Broad Institute). Variants were called and pseudo references generated using samtools (v 1.6) and bcftools (v 1.9) and the PseudoreferencePipeline scripts of the YourePrettyGood pipeline ( Average pairwise divergence (Dxy) between engineered strains and RAL386, yw iso8 and RAL59 was calculated for 1 Mbp non-overlping windows using RandomScripts#calculatepolymorphismcpp of the YourePrettyGood pipeline ( (S5 Fig).
Drosophila rearing
All fly strains were reared in bottles or vials at 21°C, 50% in day-night cycle of 12hr-12hr on fly food containing agar (0.56%), cornmeal (6.71%), inactivated yeast (1.59%), soy flour (0.92%), corn syrup (7.0%), propionic acid (0.44%), and Tegosept (0.15%) (LabExpress).
Viability assays
Locomotion performance assay
Supporting information
S1 Table.
/> (XLSX)
S2 Table.
/> (XLSX)
S3 Table.
/> (XLSX)
S1 Fig.
/> (TIFF)
S2 Fig.
/> (TIFF)
S3 Fig.
/> (TIFF)
S4 Fig.
/> (TIFF)
S5 Fig.
/> (TIFF)
S6 Fig.
/> (TIFF)
S7 Fig.
/> (TIFF)
Acknowledgments
Thanks to N. Okami for help with fly stocks and K. O’Connor-Giles for sharing reagents. Thanks to the Andolfatto, Przeworski and Sella labs for useful discussions.
References
1. Kimura M. Evolutionary Rate at the Molecular Level. Nature. 1968;217(5129):624–6.
2. Ohta T. Slightly deleterious mutant substitutions in evolution. Nature. 1973;246(5428):96–8. pmid:4585855
3. Wright SI, Andolfatto P. The impact of natural selection on the genome: emerging patterns in Drosophila and Arabidopsis. Annu Rev Ecol Evol Syst. 2008;39:193–213.
4. McDonald JH, Kreitman M. Adtive protein evolution at the Adh locus in Drosophila. Nature. 1991;351(6328):652–4. pmid:1904993
5. Fay JC, Wyckoff GJ, Wu C-I. Testing the neutral theory of molecular evolution with genomic data from Drosophila. Nature. 2002;415(6875):1024–6. pmid:11875569
6. Smith NGC, Eyre-Walker A. Adtive protein evolution in Drosophila. Nature. 2002;415(6875):1022–4. pmid:11875568
7. Sawyer SA, Kulathinal RJ, Bustamante CD, Hartl DL. Bayesian analysis suggests that most amino acid replacements in Drosophila are driven by positive selection. J Mol Evol. 2003;57 Suppl 1:S154–64. pmid:15008412
8. Sella G, Petrov DA, Przeworski M, Andolfatto P. Pervasive natural selection in the Drosophila genome?. PLOS Genetics. 2009;5(6):e1000495.
9. Slotte T, Foxe JP, Hazzouri KM, Wright SI. Genome-wide evidence for efficient positive and purifying selection in Csella grandiflora, a plant species with a large effective population size. Mol Biol Evol. 2010;27(8):1813–21. pmid:20194429
10. Charlesworth J, Eyre-Walker A. The rate of adtive evolution in enteric bacteria. Mol Biol Evol. 2006;23(7):1348–56. pmid:16621913
11. Booker T, Keightley P. Understanding the factors that she patterns of nucleotide diversity in the house mouse genome. Mol Biol Evol. 2018;35.
12. Booker TR, Jackson BC, Keightley PD. Detecting positive selection in the genome. BMC Biol. 2017;15(1):98. pmid:29084517
13. Galtier N. Adtive Protein Evolution in Animals and the Effective Population Size Hypothesis. PLoS Genet. 2016;12(1):e1005774. pmid:26752180
14. Eyre-Walker A. Changing effective population size and the McDonald-Kreitman test. Genetics. 2002;162(4):2017–24. pmid:12524367
15. Soni V, Moutinho AF, Eyre-Walker A. Changing Population Size in McDonald-Kreitman Style Analyses: Artifactual Correlations and Adtive Evolution between Humans and Chimpanzees. Genome Biol Evol. 2022;14(2):evac022. pmid:35143656
16. Andolfatto P. Hitchhiking effects of recurrent beneficial amino acid substitutions in the Drosophila melanogaster genome. Genome Res. 2007;17(12):1755–62. pmid:17989248
17. Macpherson JM, Sella G, Davis JC, Petrov DA. Genomewide spatial correspondence between nonsynonymous divergence and neutral polymorphism reveals extensive adtation in Drosophila. Genetics. 2007;177(4):2083–99. pmid:18073425
18. Elyashiv E, Sattath S, Hu TT, Strutsovsky A, McVicker G, Andolfatto P, et al. A Genomic M of the Effects of Linked Selection in Drosophila. PLoS Genet. 2016;12(8):e1006130. pmid:27536991
19. Liang Y-Y, Shi Y, Yuan S, Zhou B-F, Chen X-Y, An Q-Q, et al. Linked selection shes the landsce of genomic variation in three oak species. New Phytol. 2022;233(1):555–68. pmid:34637540
20. Williamson RJ, Josephs EB, Platts AE, Hazzouri KM, Haudry A, Blanchette M, et al. Evidence for widespread positive and negative selection in coding and conserved noncoding regions of Csella grandiflora. PLoS Genet. 2014;10(9):e1004622. pmid:25255320
21. Echave J, Spielman SJ, Wilke CO. Causes of evolutionary rate variation among protein sites. Nat Rev Genet. 2016;17(2):109–21. pmid:26781812
22. Shih C-H, Chang C-M, Lin Y-S, Lo W-C, Hwang J-K. Evolutionary information hidden in a single protein structure. Proteins. 2012;80(6):1647–57. pmid:22454236
23. Nevin Gerek Z, Kumar S, Banu Ozkan S. Structural dynamics flexibility informs function and evolution at a proteome scale. Evol pl. 2013;6(3):423–33. pmid:23745135
24. Ridout KE, Dixon CJ, Filatov DA. Positive selection differs between protein secondary structure elements in Drosophila. Genome Biol Evol. 2010;2:166–79. pmid:20624723
25. Wilson DJ, Hernandez RD, Andolfatto P, Przeworski M. A population genetics-phylogenetics proach to inferring natural selection in coding sequences. PLoS Genet. 2011;7(12):e1002395. pmid:22144911
26. Zhao ZM, Campbell MC, Li N, Lee DSW, Zhang Z, Townsend JP. Detection of Regional Variation in Selection Intensity within Protein-Coding Genes Using DNA Sequence Polymorphism and Divergence. Molecular Biology and Evolution. 2017;34(11):3006–22.
27. Adams J, Mansfield MJ, Richard DJ, Doxey AC. Lineage-specific mutational clustering in protein structures predicts evolutionary shifts in function. Bioinformatics. 2017;33(9):1338–45. pmid:28052926
28. Slodkowicz G, Goldman N. Integrated structural and evolutionary analysis reveals common mechanisms underlying adtive evolution in mammals. Proc Natl Acad Sci U S A. 2020;117(11):5977–86. pmid:32123117
29. McFerrin LG, Stone EA. The non-random clustering of non-synonymous substitutions and its relationship to evolutionary rate. BMC Genomics. 2011;12:415. pmid:21846337
30. Fitch WM. Rate of change of concomitantly variable codons. J Mol Evol. 1971;1(1):84–96. pmid:4377447
31. Fitch WM, Markowitz E. An improved method for determining codon variability in a gene and its plication to the rate of fixation of mutations in evolution. Biochem Genet. 1970;4(5):579–93. pmid:5489762
32. Miyamoto MM, Fitch WM. Testing the covarion hypothesis of molecular evolution. Mol Biol Evol. 1995;12(3):503–13. pmid:7739391
33. Davis BH, Poon AFY, Whitlock MC. Compensatory mutations are repeatable and clustered within proteins. Proc Biol Sci. 2009;276(1663):1823–7. pmid:19324785
34. Callahan B, Neher RA, Bachtrog D, Andolfatto P, Shraiman BI. Correlated evolution of nearby residues in Drosophilid proteins. PLoS Genet. 2011;7(2):e1001315. pmid:21383965
35. Taverner AM, Blaine LJ, Andolfatto P. Epistasis and physico-chemical constraints contribute to spatial clustering of amino acid substitutions in protein evolution [Internet]. bioRxiv; 2020 [cited 2024 Nov 1]. p. 2020.08.05.237594. Available from: /> 36. Chaurasia S, Dutheil JY. The Structural Determinants of Intra-Protein Compensatory Substitutions. Mol Biol Evol. 2022;39(4):msac063.
37. Stolyarova AV, Neretina TV, Zvyagina EA, Fedotova AV, Kondrashov AS, Bazykin GA. eLife [Internet]. eLife Sciences Publications Limited; 2022 [cited 2025 Jun 24]. Complex fitness landsce shes variation in a hyperpolymorphic species. Available from: /> 38. Ragsdale . Local fitness and epistatic effects lead to distinct patterns of linkage disequilibrium in protein-coding genes. Genetics. 2022;221(4):iyac097. pmid:35736370
39. Weinreich DM, Delaney NF, DePristo MA, Hartl DL. Darwinian evolution can follow only very few mutational paths to fitter proteins. Science. 2006;312(5770):111–4.
40. Ortlund EA, Bridgham JT, Redinbo MR, Thornton JW. Crystal structure of an ancient protein: evolution by conformational epistasis. Science. 2007;317(5844):1544–8. pmid:17702911
41. Biswas A, Haldane A, Arnold E, Levy RM. Epistasis and entrenchment of drug resistance in HIV-1 subtype B. eLife. 2019;8:e50524.
42. Hochberg GKA, Liu Y, Marklund EG, Metzger BPH, Laganowsky A, Thornton JW. A hydrophobic ratchet entrenches molecular complexes. Nature. 2020;588(7838):503–8. pmid:33299178
43. Gong LI, Suchard MA, Bloom JD. Stability-mediated epistasis constrains the evolution of an influenza protein. eLife. 2013.
44. Sarkisyan KS, Bolotin DA, Meer MV, Usmanova DR, Mishin AS, Sharonov GV, et al. Local fitness landsce of the green fluorescent protein. Nature. 2016;533(7603):397–401. pmid:27193686
45. Harms MJ, Thornton JW. Historical contingency and its biophysical basis in glucocorticoid receptor evolution. Nature. 2014;512(7513):203–7. pmid:24930765
46. Pkou A, Garcia-Pastor L, Escudero JA, Wagner A. A rugged yet easily navigable fitness landsce. Science. 2023;382(6673):eadh3860.
47. Good BH, McDonald MJ, Barrick JE, Lenski RE, Desai MM. The dynamics of molecular evolution over 60,000 generations. Nature. 2017;551(7678):45–50.
48. Johnson MS, Reddy G, Desai MM. Epistasis and evolution: recent advances and an outlook for prediction. BMC Biol. 2023;21(1):120. pmid:37226182
49. Tarvin RD, Borghese CM, Sachs W, Santos JC, Lu Y, O’Connell LA. Interacting amino acid replacements allow poison frogs to evolve epibatidine resistance. Science. 2017;357(6357):1261–6.
50. Mateu MG, Fersht AR. Mutually compensatory mutations during evolution of the tetramerization domain of tumor suppressor p53 lead to impaired hetero-oligomerization. Proceedings of the National Academy of Sciences. 1999;96(7).
51. Weber RE, Fago A, Malte H, Storz JF, Gorr TA. Lack of conventional oxygen-linked proton and anion binding sites does not impair allosteric regulation of oxygen binding in dwarf caiman hemoglobin. Am J Physiol Regul Integr Comp Physiol. 2013;305(3):R300-12. pmid:23720132
52. Storz JF. Compensatory mutations and epistasis for protein function. Curr Opin Struct Biol. 2018;50:18–25. pmid:29100081
53. Mohammadi S, Yang L, Harpak A, Herrera-Álvarez S, Del Pilar Rodríguez-Ordoñez M, Peng J, et al. Concerted evolution reveals co-adted amino acid substitutions in Na+K+-ATPase of frogs that prey on toxic toads. Curr Biol. 2021;31(12):2530-2538.e10. pmid:33887183
54. Mohammadi S, Herrera-Álvarez S, Yang L, Rodríguez-Ordoñez MDP, Zhang K, Storz JF. bioRxiv. 2021.
55. Starr TN, Thornton JW. Epistasis in protein evolution. Protein Sci. 2016;25(7):1204–18. pmid:26833806
56. Olson CA, Wu NC, Sun R. A comprehensive biophysical description of pairwise epistasis throughout an entire protein domain. Curr Biol. 2014;24(22):2643–51. pmid:25455030
57. Karageorgi M, Groen SC, Sumbul F, Pelaez JN, Verster KI, Aguilar JM, et al. Genome editing retraces the evolution of toxin resistance in the monarch butterfly. Nature. 2019;574(7778):409–12. pmid:31578524
58. Taverner AM, Yang L, Barile ZJ, Lin B, Peng J, Pinharanda , et al. Adtive substitutions underlying cardiac glycoside insensitivity in insects exhibit epistasis in vivo. eLife. 2019;8:e48224.
59. Yang L, Borne F, Betz A, Aardema ML, Zhen Y, Peng J, et al. Predatory fireflies and their toxic firefly prey have evolved distinct toxin resistance strategies. Curr Biol. 2023;33(23):5160-5168.e7. pmid:37989309
60. Bateman J, Shu H, Van Vactor D. The guanine nucleotide exchange factor Trio mediates axonal development in the Drosophila embryo. Neuron. 2000;26(1):93–106.
61. Iyer SC, Wang D, Iyer EPR, Trunnell SA, Meduri R, Shinwari R, et al. The RhoGEF trio functions in sculpting class specific dendrite morphogenesis in Drosophila sensory neurons. PLoS One. 2012;7(3):e33634. pmid:22442703
62. Wu Y-C, Cheng T-W, Lee M-C, Weng N-Y. Distinct rac act

Source: PLOS (Public Library of Science)

Read More

Description

A high school principal in North Carolina set a good example for educators by putting the school’s valedictorian in her place after she hijacked her commencement speech.

Content

A high school principal in North Carolina set a good example for educators by putting the school’s valedictorian in her place after she hijacked her commencement speech.
As The New York Post reported, Clayton High School senior Leen Hijaz was given the honor of delivering the institution’s commencement speech due to her academic accomplishments. Normally, such addresses are generally uneventful and at least somewhat olitical.
But Hijaz, a Muslim, soon abandoned her pre-proved remarks to unleash a vile rant against ICE and spread pro-Palestine propaganda.
Before I leave the stage, I have one last thing to say. Every single person here has a voice; we have the privilege to use it when millions around the world are struggling and suffering to be heard, Hijaz stated.
Whether it’s the millions suffering in Palestine, Sudan, Congo, Afghanistan, and so many other countries around the world, or families being torn art by ICE. These are not just an issue here; they are hpening there, they’re hpening right here as I speak, she continued.
My point is, we’re not given a voice to stay silent, Hijaz tried to add before Principal Melissa Moore proached her.
Moore promptly yanked the microphone away from Hijaz and pulled her off the stage to the teen’s dismay.
WATCH (turn to 1:02:00 to hear the relevant remarks):
Here is a shorter version posted on Hijaz’s TikTok account:
@leenhijaz_ never be afraid to speak up!! #freepalestine #ice #valedictorian #grad #graduation ♬ original sound – blade aka rui’s biggest fan
Hijaz later fumed about being shut down by Principal Moore, noting she was only able to use half of her speech.
The Muslim teen also whined about feeling oppressed.
@leenhijaz_ #claytonhighschool #johnstoncounty ♬ original sound – leeno
Johnston County Public Schools released the following statement to the New York Post regarding the incident:
During this year’s Clayton High School graduation, a student departed from her proved remarks.
School administrators intervened in order to maintain the integrity and focus of the program in real time. This action was not about limiting a student’s voice, but about ensuring that a school-sponsored event remained consistent with its intended purpose.

Source: The Gateway Pundit

Read More

Description

Content

Have you ever walked around a street, mall, or airport and noticed two or three of the same franchise restaurant within walking distance? Why might one Starbucks or McDonald’s or Wetzel’s Pretzels sometimes be built so close to another? Are they friends or competitors? And how can that possibly be profitable?
Today’s show is one such example. Our pals at Hyperfixed got a knotty question we just had to help them untangle: Why are there so many Wetzel’s Pretzels so close to one another at the Atlantic Avenue-Barclays Center Station?
To find out, Alexi Horowitz-Ghazi followed the dough all the way to the top. His journey led him to a jolly pretzel executive, a franchisee with a deep-fried American dream, and a brush with mall security.
Support:
Planet Money+
Read:
Our book: Planet Money: A Guide to the Economic Forces That She Your Life
Our weekly longform Planet Money newsletter
Our weekly Indicator round-up newsletter
Follow:
Instagram
TikTok
YouTube
Facebook
This episode was hosted by Alex Goldman and Alexi Horowitz-Ghazi. Hyperfixed is produced and edited by Emma Courtland, Amor Yates, Sari Soffer Sukenik and Tori Dominguez Peak. The music is by the mysterious Breakmaster Cylinder and Alex Goldman. It was engineered by Tony Williams. Fact checking by Naomi Barr. The Planet Money version was produced by Sam Yellowhorse Kesler and edited by Jess Jiang. It was engineered by Robert Rodriguez. Alex Goldmark is Planet Money’s executive producer.

Source: NPR

Read More

Description

Last year, a report compiled at one of California’s most prestigious public universities revealed that a significant percentage of incoming freshmen struggled with basic math. Others lacked elementary writing and language skills.

Content

Last year, a report compiled at one of California’s most prestigious public universities revealed that a significant percentage of incoming freshmen struggled with basic math. Others lacked elementary writing and language skills.
A recent revolt by University of California faculty members indicates the rot is system-wide.
In October, a joint faculty-administration committee at UC-San Diego warned that “over the past five years,” the school “has experienced a steep decline in the academic preparation of its entering first-year students. … This trend poses serious challenges both to student success and to the university’s instructional mission.”
The committee discovered that one in 12 students lacked the math skills necessary to succeed at seventh-grade levels.
Notably, most of these students “earned” outstanding grades in high school. Also notably, the UC Board of Regents, in a tip of the c to left-wing “diversity” and “equity” concerns, stopped requiring the SAT or ACT in 2020.
Last week, the controversy returned as more than 600 University of California professors, led by mathematicians at UC-Berkeley, sent a letter to system leaders expressing concerns about the readiness of incoming students, The Los Angeles Times reported.
“We now observe preparation gs so severe,” the letter observes, “that instructors must re-teach middle-school mathematics while simultaneously teaching the material students need for sciences, engineering, economics and other quantitatively demanding fields.”
One of the organizers of the letter, Zvezda Stankova, a mathematics professor at Berkeley, told The Times that she has seen the problems first-hand in her classroom.
“Something had changed drastically,” she said about a 2023 calculus class. “The bottom was taken out, and there were 25 to 30 percent of the students who were in free fall. There was nothing you could do for them. They were just not prepared.”
The idea that standardized tests are an insurmountable barrier to minority students is an example of the bigotry of low expectations. The idea that universities can maintain high academic standards and produce graduates prepared to handle the rigors of the real world by dumbing-down admission standards is similarly misguided.
Expect the professors’ plea to run into fierce resistance, but California res what it sows. It provides an unfortunate example of the dangers of progressive education fads intended to advance a political ideology rather than encourage academic achievement.

Source: Arkansas Online

Read More

Description

The National Center for Atmospheric Research is the latest example of how the Trump administration’s efforts to chainsaw the federal government can hpen too fast for the courts or Congress to counter.

Content

Courts similarly tried to slow or reverse President Donald Trump’s efforts to disband Voice of America or demolish the White House’s East Wing, only to realize that the administration’s plan was practically or physically hard to undo.
This is the Trump administration strategy in a nutshell: Break things fast before the judicial system can rule, said Samuel Bagenstos, who served as chief OMB counsel in the Biden administration.
What this administration is doing is acting in a way that breaks the government into a million pieces, so it is impossible for a court to come along and put Humpty Dumpty back together again, he said.
Even some Republicans have rebuked the assault, which is spearheaded by Budget Director Russ Vought, particularly since weather and climate research is often interlinked. Rep. Jeff Hurd (R-Colo.) pushed the administration to save NCAR, even though it’s not in his district. (The center is in Boulder, Colorado, nearby.)
I’m cautiously optimistic that this important institution is going to stay where it is, he said. It provides cabilities that are important to not only Colorado, but the nation, and I trust that the administration sees that as well.
White House officials referred questions to OMB. When asked about the cuts, a spokesperson said the administration is focused on eliminating billions of dollars in woke and wasteful programs focused on the Green New Scam and racist DEI. OMB did not immediately respond to follow up questions about the cuts.
The move to disband the National Center for Atmospheric Research is part of an administrationwide effort to neuter programs related to climate change, which Trump has repeatedly called a hoax.
Last week, Vought proposed new guidelines for federal grants that require political pointees at science agencies to prove funding in order to demonstrably advance the President’s policy priorities, a move that scientists say could be used to crush virtually all federal climate research. The unprecedented move would affect tens of billions of dollars across all agencies, shifting research priorities away from subject-matter experts to ensure that federally funded science is politicized, researchers across multiple disciplines have said.
And last month the administration announced plans to dismantle a $368 million deep-ocean observation system that monitors marine ecosystems and the powerful currents that affect the climate. Vought has also suggested that he will cut funding for the National Academies of Sciences, Engineering and Medicine, in part due to its climate work.
But the Colorado-based office has long been a specific target for Vought, who has called climate science alarmism and claimed that it must be eradicated across the federal government.
While the budget recissions and reductions in force drew the headlines, Vought has arguably been just as successful at shrinking government in quieter ways such as selling off assets and defunding offices.
NCAR, housed in a building designed by famed architect I.M. Pei, is one of the world’s most important climate and weather research labs, which the departments of Defense and Energy, the Federal Aviation Administration, NASA and NOAA all rely on for its research.
But OMB officials argued that the lab’s work should be defunded because it studies climate variability, long-term climate change, and the role of human activities in global warming, which do not align with the administration’s priorities, according to internal emails revealed during the court proceedings.
Vought’s moves against research are ignoring how much scientific innovation has rewarded the American economy, said Craig McLean, who was acting chief scientist at NOAA in the first Trump administration.
If you look at the Trump administration’s ambitions for a strong economy, he’s unplugging the engine that has contributed to this, McLean said.
Vought foreshadowed his plan to dismantle federal climate science across multiple agencies in Project 2025, the conservative policy playbook organized by the Heritage Foundation ahead of the 2024 election. In that document, Vought wrote that climate research should be eliminated because it can be used against industry in legal proceedings and regulations.
The Trump administration has been dismantling NCAR for months, even as a lawsuit filed on behalf of the center against Vought and other Trump administration officials proceeded in court. The administration sought to transfer the center’s supercomputing cacity to the University of Wyoming, court documents show. A private company has expressed interest in taking over its space weather prediction functions. Its two research aircraft may be transferred.
The lab is also having trouble hiring for key positions because plicants are rejecting job offers over fear of uncertainty about its future, according to the court records. Even plicants awarded a slot in the coveted internship program are dropping out.
As a result of OMB pressure, political leadership at the National Science Foundation, which funds the lab, has sought proposals to take over the NCAR building, its space weather observing cabilities, weather prediction functions and atmospheric research, according to the lawsuit.
The agencies’ ultimate parent goal is to destroy NCAR entirely, the suit claims.
NCAR, which has about 800 employees, has been in operation since 1960. It is operated on behalf of the National Science Foundation by the University Corporation for Atmospheric Research, a consortium of more than 100 colleges and universities.
Some longtime government employees are horrified by Vought’s determination to root out climate research and are deeply worried that it puts the nation at greater risk of physical harm from climate change and its costly impacts, said one career official, granted anonymity to avoid reprisal. They have made a mockery of climate science by calling it ‘woke,’ ‘radical.’ and ‘extremist,’ while the extreme and increasingly dangerous impacts of climate change continue to mount.
Even some administration allies are opposed to breaking up NCAR and selling it off for parts. That includes Judith Curry, a longtime climate science critic and researcher who helped write a controversial climate science report for the Department of Energy. NCAR provides invaluable research for weather research and atmospheric science that can’t be lost, said Curry, who previously served on an NCAR advisory board. She said it should be downsized but not effectively eliminated.
Should some of this be streamlined? Should some of this be rethought? I would say, sure, she said. But we need to keep the political vindictiveness out of it, and there are reasons to keep some sort of a core in Boulder.

Source: POLITICO

Read More