Phosphoglucose isomerases of hagfish, zebrafish, gray mullet, toad, and snake, with reference to the evolution of the genes in vertebrates

Phosphoglucose isomerase (PGI) is a protein with multiple functions. To infer its structure changes and evolution in vertebrates, we cloned cDNAs encoding PGI genes from hagfish (Paramyxine yangi), gray mullet (Mugil cephalus), zebrafish (Danio rerio), toad (Bufo melanosticus), and snake (Boiga kraepelini). Only one PGI gene was cloned in each of hagfish, toad, and snake, but two PGI genes were found in zebrafish and gray mullet, respectively. The PGI of hagfish encodes 554 amino acids, in contrast to the PGIs of bonyfishes, toad, and snake which encode 553 amino acids and the PGIs of mammals which encode 558 amino acids. Among 558 aligned amino acid sites, there are 314 sites (56.27%) totally conserved. To see if diversifying selection acts on PGI amino acids of vertebrates, we calculated the pairwise ratio of nonsynonymous versus synonymous substitution per site (Ka/Ks) and the ratio of radical amino acid changes versus conservative amino acid changes per sites (dR/dC) between PGI sequences. The average pairwise ratio between nonsynonymous substitutions per nucleotide (Ka) and synonymous substitutions per nucleotide (Ks) among vertebrate PGI sequences equals 0.047 +/- 0.019. The average pairwise ratio between radical amino acid changes and conservative amino acid changes (dR/dC) among the vertebrate PGIs equal 0.938 +/- 0.158 for charge changes, 0.558 +/- 0.085 for polarity changes, and 0.465 +/- 0.0714 when both polarity and volume are considered. There is no amino acid within the vertebrate PGIs under diversifying selection as analyzed by the method of Yang et al. (2000b). The results suggest that the present vertebrate PGIs are at evolutionary stasis and are being subjected to intense purifying selection. The purifying selection is to maintain polarity and volume of the protein but not the charge groups of amino acids. Phylogenetic analysis reveals that vertebrate PGIs can be classified into three major groups: the mammalian, amphibian-reptilian, and teleostean PGIs. The gene tree suggests that the gene duplication event of PGI in bonyfishes occurred before diversification of Acanthopterygii but after the split of bonyfishes and tetrapods. The evolution of multiple functions of PGI is discussed.     

Frequent and widespread parallel evolution of protein sequences

Understanding the patterns and causes of protein sequence evolution is a major challenge in evolutionary biology. One of the critical unresolved issues is the relative contribution of selection and genetic drift to the fixation of amino acid sequence differences between species. Molecular homoplasy, the independent evolution of the same amino acids at orthologous sites in different taxa, is one potential signature of selection; however, relatively little is known about its prevalence in eukaryotic proteomes. To quantify the extent and type of homoplasy among evolving proteins, we used phylogenetic methodology to analyze 8 genome-scale data matrices from clades of different evolutionary depths that span the eukaryotic tree of life. We found that the frequency of homoplastic amino acid substitutions in eukaryotic proteins was more than 2-fold higher than expected under neutral models of protein evolution. The overwhelming majority of homoplastic substitutions were parallelisms that involved the most frequently exchanged amino acids with similar physicochemical properties and that could be reached by a single-mutational step. We conclude that the role of homoplasy in shaping the protein record is much larger than generally assumed, and we suggest that its high frequency can be explained by both weak positive selection for certain substitutions and purifying selection that constrains substitutions to a small number of functionally equivalent amino acids.     

Evidence for Positive Selection within the PgiC1 Locus in the Grass Festuca ovina

The dimeric metabolic enzyme phosphoglucose isomerase (PGI, EC 5.3.1.9) plays an essential role in energy production. In the grass Festuca ovina, field surveys of enzyme variation suggest that genetic variation at cytosolic PGI (PGIC) may be adaptively important. In the present study, we investigated the molecular basis of the potential adaptive significance of PGIC in F. ovina by analyzing cDNA sequence variation within the PgiC1 gene. Two, complementary, types of selection test both identified PGIC1 codon (amino acid) sites 200 and 173 as candidate targets of positive selection. Both candidate sites involve charge-changing amino acid polymorphisms. On the homology-modeled F. ovina PGIC1 3-D protein structure, the two candidate sites are located on the edge of either the inter-monomer boundary or the inter-domain cleft; examination of the homology-modeled PGIC1 structure suggests that the amino acid changes at the two candidate sites are likely to influence the inter-monomer interaction or the domain-domain packing. Biochemical studies in humans have shown that mutations at several amino acid sites that are located close to the candidate sites in F. ovina, at the inter-monomer boundary or the inter-domain cleft, can significantly change the stability and/or kinetic properties of the PGI enzyme. Molecular evolutionary studies in a wide range of other organisms suggest that PGI amino acid sites with similar locations to those of the candidate sites in F. ovina may be the targets of positive/balancing selection. Candidate sites 200 and 173 are the only sites that appear to discriminate between the two most common PGIC enzyme electromorphs in F. ovina: earlier studies suggest that these electromorphs are implicated in local adaptation to different grassland microhabitats. Our results suggest that PGIC1 sites 200 and 173 are under positive selection in F. ovina.     

Evolutionary conservativeness of electric field in the Cu,Zn superoxide dismutase active site. Evidence for co-ordinated mutation of charged amino acid residues.

Equipotential lines were calculated, using the Poisson-Boltzmann equation, for six Cu,Zn superoxide dismutases with different protein electric charge and various degrees of sequence homology, namely those from ox, pig, sheep, yeast, and the isoenzymes A and B from the amphibian Xenopus laevis. The three-dimensional structures of the porcine and ovine superoxide dismutases were obtained by molecular modelling reconstruction using the structure of the highly homologous bovine enzyme as a template. The three-dimensional structure of the evolutionary distant yeast Cu,Zn superoxide dismutase was recently resolved by us, while computer-modelled structures are available for X. laevis isoenzymes. The six proteins display large differences in the net protein charge and distribution of electrically charged surface residues but the trend of the equipotential lines in the proximity of the active sites was found to be constant in all cases. These results are in line with the very similar catlytic rate constants experimentally measured for the corresponding enzyme activities. This analysis shows that electrostatic guidance for the enzyme-substrate interaction in Cu,Zn superoxide dismutases is related to a spatial distribution of charges, arranged so as to maintain, in the area surrounding the active sites, an identical electrostatic potential distribution, which is conserved in the evolution of this protein family.     

Emergence of Complex Haplotypes from Microevolutionary Variation in Sequence and Structure of <Emphasis Type="Italic">Colias</Emphasis> Phosphoglucose Isomerase

A molecular evolutionary explanation of natural genetic variation requires analysis of specific variants’ evolutionary dynamics. To pursue this for phosphoglucose isomerase (PGI) of Colias butterflies, whose polymorphism is maintained by strong natural selection, we assembled a large data set of wild haplotypes, highly variable at the amino acid and DNA levels. The most common electrophoretic, i.e., charge macrostate, allele class, 3, is conserved in its pattern of charged amino acid residues. The next most common macrostate, 4, has multiple patterns of charge, i.e., microstates, while less common (1, 2, 5, 6) macrostates are very diverse. Macrostate 4 shows significant linkage disequilibrium (LD) among its variants, especially for two groups of five haplotypes each. We find extensive intragenic recombination among all haplotypes except the two high-LD groups of macrostate 4, which display none. Phyletic relations among haplotypes are largely reticulate, again except for the high-LD groups of macrostate 4, which form clades with strong bootstrap support. Charge-changing and linked charge-neutral amino acid variants occur in diverse parts of PGI’s sequence. Homology-based modeling of PGI’s structure shows that these regions are related spatially in ways suggesting functional interaction. The high-LD groups of macrostate 4 display parallel amino acid variation in these regions. This pattern of haplotype clades with high LD among multiple varying sites, emerging from chaotically recombining variation, may be a “signature” of refinement of complex adaptive sequences by recombination and selection. It should be tested further in this study system and others as a possibly general feature of the evolution of living complexity.     

Protein evolution with dependence among codons due to tertiary structure

Markovian models of protein evolution that relax the assumption of independent change among codons are considered. With this comparatively realistic framework, an evolutionary rate at a site can depend both on the state of the site and on the states of surrounding sites. By allowing a relatively general dependence structure among sites, models of evolution can reflect attributes of tertiary structure. To quantify the impact of protein structure on protein evolution, we analyze protein-coding DNA sequence pairs with an evolutionary model that incorporates effects of solvent accessibility and pairwise interactions among amino acid residues. By explicitly considering the relationship between nonsynonymous substitution rates and protein structure, this approach can lead to refined detection and characterization of positive selection. Analyses of simulated sequence pairs indicate that parameters in this evolutionary model can be well estimated. Analyses of lysozyme c and annexin V sequence pairs yield the biologically reasonable result that amino acid replacement rates are higher when the replacements lead to energetically favorable proteins than when they destabilize the proteins. Although the focus here is evolutionary dependence among codons that is associated with protein structure, the statistical approach is quite general and could be applied to diverse cases of evolutionary dependence where surrogates for sequence fitness can be measured or modeled.     

Epistasis Creates Invariant Sites and Modulates the Rate of Molecular Evolution

Invariant sites are a common feature of amino acid sequence evolution. The presence of invariant sites is frequently attributed to the need to preserve function through site-specific conservation of amino acid residues. Amino acid substitution models without a provision for invariant sites often fit the data significantly worse than those that allow for an excess of invariant sites beyond those predicted by models that only incorporate rate variation among sites (e.g., a Gamma distribution). An alternative is epistasis between sites to preserve residue interactions that can create invariant sites. Through computer-simulated sequence evolution, we evaluated the relative effects of site-specific preferences and site-site couplings in the generation of invariant sites and the modulation of the rate of molecular evolution. In an analysis of ten major families of protein domains with diverse sequence and functional properties, we find that the negative selection imposed by epistasis creates many more invariant sites than site-specific residue preferences alone. Further, epistasis plays an increasingly larger role in creating invariant sites over longer evolutionary periods. Epistasis also dictates rates of domain evolution over time by exerting significant additional purifying selection to preserve site couplings. These patterns illuminate the mechanistic role of epistasis in the processes underlying observed site invariance and evolutionary rates.     

Evidence of positive selection at codon sites localized in the C-terminal peptide of ORC6

Origin recognition complex 6 (Orc6) plays a central role in the initiation of DNA replication in all eukaryotic systems. The exact contribution of Orc6 to replication initiation has yet to be elucidated. Here, we analyzed the evolutionary dynamics of Orc6 in 15 vertebrates. Positive selection was detected in the region of exon 6 of the Orc6 gene. Site tests revealed a proportion of codon sites that displayed evidence of positive selection (ω > 1) within the coding sequences of the vertebrate Orc6 gene. Seven positively selected amino acid sites were identified and three were located in exon6. These results suggest that amino acid residues present in the middle region of the protein are more selectively constrained, whereas amino acid residues in the C-terminal peptide of the protein evolve at a faster rate, possibly because of heightened selective pressure during the course of evolution.     

Three-dimensional window analysis for detecting positive selection at structural regions of proteins

Detection of natural selection operating at the amino acid sequencelevel is important in the study of molecular evolution. Single-siteanalysis and one-dimensional window analysis can be used todetect selection when the biological functions of amino acidsites are unknown. Single-site analysis is useful when selectionoperates more or less constantly over evolutionary time, butless so when selection operates temporarily. One-dimensionalwindow analysis is more sensitive than single-site analysiswhen the functions of amino acid sites in close proximity inthe linear sequence are similar, although this is not alwaysthe case. Here I present a three-dimensional window analysismethod for detecting selection given the three-dimensional structureof the protein of interest. In the three-dimensional structure,the window is defined as the sphere centered on the -carbonof an amino acid site. The window size is the radius of thesphere. The sites whose -carbons are included in the windoware grouped for the neutrality test. The window is moved withinthe three-dimensional structure by sequentially moving the centralsite along the primary amino acid sequence. To detect positiveselection, it may also be useful to group the surface-exposedsites in the window separately. Three-dimensional window analysisappears not only to be more sensitive than single-site analysisand one-dimensional window analysis but also to provide similarspecificity for inferring positive selection in the analysesof the hemagglutinin and neuraminidase genes of human influenzaA viruses. This method, however, may fail to detect selectionwhen it operates only on a particular site, in which case single-siteanalysis may be preferred, although a large number of sequencesis required.     

Distinguishing structural and functional restraints in evolution in order to identify interaction sites

Structural genomics projects are producing many three-dimensional structures of proteins that have been identified only from their gene sequences. It is therefore important to develop computational methods that will predict sites involved in productive intermolecular interactions that might give clues about functions. Techniques based on evolutionary conservation of amino acids have the advantage over physiochemical methods in that they are more general. However, the majority of techniques neither use all available structural and sequence information, nor are able to distinguish between evolutionary restraints that arise from the need to maintain structure and those that arise from function. Three methods to identify evolutionary restraints on protein sequence and structure are described here. The first identifies those residues that have a higher degree of conservation than expected: this is achieved by comparing for each amino acid position the sequence conservation observed in the homologous family of proteins with the degree of conservation predicted on the basis of amino acid type and local environment. The second uses information theory to identify those positions where environment-specific substitution tables make poor predictions of the overall amino acid substitution pattern. The third method identifies those residues that have highly conserved positions when three-dimensional structures of proteins in a homologous family are superposed. The scores derived from these methods are mapped onto the protein three-dimensional structures and contoured, allowing identification clusters of residues with strong evolutionary restraints that are sites of interaction in proteins involved in a variety of functions. Our method differs from other published techniques by making use of structural information to identify restraints that arise from the structure of the protein and differentiating these restraints from others that derive from intermolecular interactions that mediate functions in the whole organism.     

Post-duplication charge evolution of phosphoglucose isomerases in teleost fishes through weak selection on many amino acid sites

Background  

The partitioning of ancestral functions among duplicated genes by neutral evolution, or subfunctionalization, has been considered the primary process for the evolution of novel proteins (neofunctionalization). Nonetheless, how a subfunctionalized protein can evolve into a more adaptive protein is poorly understood, mainly due to the limitations of current analytical methods, which can detect only strong selection for amino acid substitutions involved in adaptive molecular evolution. In this study, we employed a comparative evolutionary approach to this question, focusing on differences in the structural properties of a protein, specifically the electric charge, encoded by fish-specific duplicated phosphoglucose isomerase (Pgi) genes.     

Extreme differences in charge changes during protein evolution

Summary The maintenance of a proper distribution of charged amino acid residues might be expected to be an important factor in protein evolution. We therefore compared the inferred changes in charge during the evolution of 43 protein families with the changes expected on the basis of random base substitutions. It was found that certain proteins, like the eye lens crystallins and most histones, display an extreme avoidance of changes in charge. Other proteins, like phospholipase A2 and ferredoxin, apparently have sustained more charged replacements than expected, suggesting a positive selection for changes in charge. Depending on function and structure of a protein, charged residues apparently can be important targets for selective forces in protein evolution. It appears that actual biased codon usage tends to decrease the proportion of charged amino acid replacements. The influence of nonrandomness of mutations is more equivocal. Genes that use the mitochondrial instead of the universal code lower the probability that charge changes will occur in the encoded proteins.     

Protein Structures and Neutral Theory of Evolution

Abstract

The neutral theory of evolution is extended to the origin of protein molecules. Arguments are presented which suggest that the amino acid sequences of many globular proteins mainly represent “memorized” random sequences while biological evolution reduces to the “editing” these random sequences. Physical requirements for a functional globular protein are formulated and it is shown that many of these requirements do not involve strategical selection of amino acid sequences during biological evolution but are inherent also for typical random sequences. In particular, it is shown that random sequences of polar and unpolar amino acid residues can form α-helices and β-strands with lengths and arrangement along the chain similar to those in real globular proteins. These α- and β-regions in random sequences can form three-dimensional folding patterns also similar to those in real proteins. The arguments are presented suggesting that even the tight packing of side groups inside protein core do not require very strong biological selection of amino acid sequences either. Thus many structural features of real proteins can exist also in random sequences and the biological selection is needed mainly for the creation of active sites of proteins and for their stability under physiological conditions.     

Selective pressures at a codon-level predict deleterious mutations in human disease genes

Deleterious mutations affecting biological function of proteins are constantly being rejected by purifying selection from the gene pool. The non-synonymous/synonymous substitution rate ratio (omega) is a measure of selective pressure on amino acid replacement mutations for protein-coding genes. Different methods have been developed in order to predict non-synonymous changes affecting gene function. However, none has considered the estimation of selective constraints acting on protein residues. Here, we have used codon-based maximum likelihood models in order to estimate the selective pressures on the individual amino acid residues of a well-known model protein: p53. We demonstrate that the number of residues under strong purifying selection in p53 is much higher than those that are strictly conserved during the evolution of the species. In agreement with theoretical expectations, residues that have been noted to be of structural relevance, or in direct association with DNA, were among those showing the highest signals of purifying selection. Conversely, those changing according to a neutral, or nearly neutral mode of evolution, were observed to be irrelevant for protein function. Finally, using more than 40 human disease genes, we demonstrate that residues evolving under strong selective pressures (omega<0.1) are significantly associated (p<0.01) with human disease. We hypothesize that non-synonymous change on amino acids showing omega<0.1 will most likely affect protein function. The application of this evolutionary prediction at a genomic scale will provide an a priori hypothesis of the phenotypic effect of non-synonymous coding single nucleotide polymorphisms (SNPs) in the human genome.     

Combining Natural Sequence Variation with High Throughput Mutational Data to Reveal Protein Interaction Sites

Many protein interactions are conserved among organisms despite changes in the amino acid sequences that comprise their contact sites, a property that has been used to infer the location of these sites from protein homology. In an inter-species complementation experiment, a sequence present in a homologue is substituted into a protein and tested for its ability to support function. Therefore, substitutions that inhibit function can identify interaction sites that changed over evolution. However, most of the sequence differences within a protein family remain unexplored because of the small-scale nature of these complementation approaches. Here we use existing high throughput mutational data on the in vivo function of the RRM2 domain of the Saccharomyces cerevisiae poly(A)-binding protein, Pab1, to analyze its sites of interaction. Of 197 single amino acid differences in 52 Pab1 homologues, 17 reduce the function of Pab1 when substituted into the yeast protein. The majority of these deleterious mutations interfere with the binding of the RRM2 domain to eIF4G1 and eIF4G2, isoforms of a translation initiation factor. A large-scale mutational analysis of the RRM2 domain in a two-hybrid assay for eIF4G1 binding supports these findings and identifies peripheral residues that make a smaller contribution to eIF4G1 binding. Three single amino acid substitutions in yeast Pab1 corresponding to residues from the human orthologue are deleterious and eliminate binding to the yeast eIF4G isoforms. We create a triple mutant that carries these substitutions and other humanizing substitutions that collectively support a switch in binding specificity of RRM2 from the yeast eIF4G1 to its human orthologue. Finally, we map other deleterious substitutions in Pab1 to inter-domain (RRM2–RRM1) or protein-RNA (RRM2–poly(A)) interaction sites. Thus, the combined approach of large-scale mutational data and evolutionary conservation can be used to characterize interaction sites at single amino acid resolution.     

The relationship between relative solvent accessibility and evolutionary rate in protein evolution

Recent work with Saccharomyces cerevisiae shows a linear relationship between the evolutionary rate of sites and the relative solvent accessibility (RSA) of the corresponding residues in the folded protein. Here, we aim to develop a mathematical model that can reproduce this linear relationship. We first demonstrate that two models that both seem reasonable choices (a simple model in which selection strength correlates with RSA and a more complex model based on RSA-dependent amino acid distributions) fail to reproduce the observed relationship. We then develop a model on the basis of observed site-specific amino acid distributions and show that this model behaves appropriately. We conclude that evolutionary rates are directly linked to the distribution of amino acids at individual sites. Because of this link, any future insight into the biophysical mechanisms that determine amino acid distributions will improve our understanding of evolutionary rates.     

Lineage-specific differences in evolutionary mode in a salamander courtship pheromone

Functionally equivalent genes may evolve heterogeneously across closely related taxa as a consequence of lineage-specific selective pressures. Such disparate evolutionary modes are especially prevalent in genes that encode postcopulatory reproductive proteins, presumably as a result of sexual selection. We might therefore expect genes that mediate reproduction prior to insemination to evolve in a similar manner. Plethodontid receptivity factor (PRF), a proteinaceous salamander pheromone produced by the male, increases female receptivity during courtship interactions. To test for lineage-specific differences in PRF's evolution, we intensively sampled PRF genes across the eastern Plethodon phylogeny (27 spp.; 34 populations) to compare gene diversification, rates of evolution, modes of selection, and types of amino acid substitution. Our analyses indicate that PRF evolutionary dynamics vary considerably from lineage to lineage. Underlying this heterogeneity, however, are two well-defined transitions in evolutionary mode. The first mode is representative of a typical protein profile, wherein neutral divergence and purifying selection are the dominant features. The second mode is characterized by incessant, cyclical evolution driven by positive selection. In this mode, the positively selected sites are bound by a limited assortment of acceptable amino acids that appear to evolve independently of other sites, resulting in a tremendous number of unique PRF alleles. Several of these selected sites are implicated in receptor binding. These sites are apparently involved in a molecular tango in which the male signal and female receptors coevolve within a confined molecular space. PRF's lineage-specific evolutionary dynamics, in combination with evidence of a molecular tango, highlight the molecular action of sexual selection on a chemical signal that is used during courtship.     

The amino acid sequence of tauropine dehydrogenase from the polychaete Arabella iricolor

The amino acid sequence of tauropine dehydrogenase (EC 1.5.1.23) from the polychaete Arabella iricolor was determined by automated sequencing of fragments obtained by cleavage with lysyl endopeptidase, endoproteinase Glu-C, and cyanogen bromide. The purified enzyme contained two isoforms that differ only in the 41st amino acid residue (Thr or Ile). Although the sequence contained eight Cys residues, intrachain disulfide bonds were not found. Two possible N-linked glycosylation sites occur in the sequences, but the enzyme does not appear to contain bound carbohydrates. Based on these data, the two isoforms of Arabella tauropine dehydrogenase are simple proteins consisted of 396 amino acid residues with calculated molecular masses of 43,085.7 Da (Thr41 isoform) and 43,097.8 Da (Ile41 isoform).     

The importance of intrinsic disorder for protein phosphorylation

Reversible protein phosphorylation provides a major regulatory mechanism in eukaryotic cells. Due to the high variability of amino acid residues flanking a relatively limited number of experimentally identified phosphorylation sites, reliable prediction of such sites still remains an important issue. Here we report the development of a new web-based tool for the prediction of protein phosphorylation sites, DISPHOS (DISorder-enhanced PHOSphorylation predictor, http://www.ist.temple. edu/DISPHOS). We observed that amino acid compositions, sequence complexity, hydrophobicity, charge and other sequence attributes of regions adjacent to phosphorylation sites are very similar to those of intrinsically disordered protein regions. Thus, DISPHOS uses position-specific amino acid frequencies and disorder information to improve the discrimination between phosphorylation and non-phosphorylation sites. Based on the estimates of phosphorylation rates in various protein categories, the outputs of DISPHOS are adjusted in order to reduce the total number of misclassified residues. When tested on an equal number of phosphorylated and non-phosphorylated residues, the accuracy of DISPHOS reaches 76% for serine, 81% for threonine and 83% for tyrosine. The significant enrichment in disorder-promoting residues surrounding phosphorylation sites together with the results obtained by applying DISPHOS to various protein functional classes and proteomes, provide strong support for the hypothesis that protein phosphorylation predominantly occurs within intrinsically disordered protein regions.     

Coevolving protein residues: maximum likelihood identification and relationship to structure

The identification of protein sites undergoing correlated evolution (coevolution) is of great interest due to the possibility that these pairs will tend to be adjacent in the three-dimensional structure. Identification of such pairs should provide useful information for understanding the evolutionary process, predicting the effects of site-directed substitution, and potentially for predicting protein structure. Here, we develop and apply a maximum likelihood method with the aim of improving detection of coevolution. Unlike previous methods which have had limited success, this method allows for correlations induced by phylogenetic relationships and for variation in rate of evolution along branches, and does not rely on accurate reconstruction of ancestral nodes. In order to reduce the complexity of coevolutionary relationships and identify the primary component of pairwise coevolution between two sites, we reduce the data to a two-state system at each site, regardless of the actual number of residues observed at that site. Simulations show that this strategy is good at identifying simple correlations and at recognizing cases in which the data are insufficient to distinguish between coevolution and spurious correlations. The new method was tested by using size and charge characteristics to group the residues at each site, and then evaluating coevolution in myoglobin sequences. Grouping based on physicochemical characteristics allows categorization of coevolving sites into positive and negative coevolution, depending on the correlation between equilibrium state frequencies. We detected a striking excess of negative coevolution (corresponding to charge) at sites brought into proximity by the periodicity of the alpha-helix, and there was also a tendency for sites with significant likelihood ratios to be close in the three-dimensional structure. Sites on the surface of the protein appear to coevolve both when they are close in the structure, and when they are distant, implying a role for folding and/or avoidance of quaternary structure in the coevolution process.