Protein co-evolution, co-adaptation and interactions

Prediction and redesign of protein-protein interactions

Understanding the molecular basis of protein function remains a central goal of biology, with the hope to elucidate the role of human genes in health and in disease, and to rationally design therapies through targeted molecular perturbations. We review here some of the computational techniques and resources available for characterizing a critical aspect of protein function - those mediated by protein-protein interactions (PPI). We describe several applications and recent successes of the Evolutionary Trace (ET) in identifying molecular events and shapes that underlie protein function and specificity in both eukaryotes and prokaryotes. ET is a part of analytical approaches based on the successes and failures of evolution that enable the rational control of PPI.

Mitochondrial-nuclear interactions: compensatory evolution or variable functional constraint among vertebrate oxidative phosphorylation genes?

Oxidative phosphorylation (OXPHOS), the major energy-producing pathway in aerobic organisms, includes protein subunits encoded by both mitochondrial (mt) and nuclear (nu) genomes. How these independent genomes have coevolved is a long-standing question in evolutionary biology. Although mt genes evolve faster than most nu genes, maintenance of OXPHOS structural stability and functional efficiency may involve correlated evolution of mt and nu OXPHOS genes. The nu OXPHOS genes might be predicted to exhibit accelerated evolutionary rates to accommodate the elevated substitution rates of mt OXPHOS subunits with which they interact. Evolutionary rates of nu OXPHOS genes should, therefore, be higher than that of nu genes that are not involved in OXPHOS (nu non-OXPHOS). We tested the compensatory evolution hypothesis by comparing the evolutionary rates (synonymous substitution rate dS and nonsynonymous substitution rate dN) among 13 mt OXPHOS genes, 60 nu OXPHOS genes, and 77 nu non-OXPHOS genes in vertebrates (7 fish and 40 mammal species). The results from a combined analysis of all OXPHOS subunits fit the predictions of the hypothesis. However, results from two OXPHOS complexes did not fit this pattern when analyzed separately. We found that the d(N) of nu OXPHOS genes for "core" subunits (those involved in the major catalytic activity) was lower than that of "noncore" subunits, whereas there was no significant difference in d(N) between genes for nu non-OXPHOS and core subunits. This latter finding suggests that compensatory changes play a minor role in the evolution of OXPHOS genes and that the observed accelerated nu substitution rates are due largely to reduced functional constraint on noncore subunits.

Coevolution of axon guidance molecule Slit and its receptor Robo

Coevolution is important for the maintenance of the interaction between a ligand and its receptor during evolution. The interaction between axon guidance molecule Slit and its receptor Robo is critical for the axon repulsion in neural tissues, which is evolutionarily conserved from planarians to humans. However, the mechanism of coevolution between Slit and Robo remains unclear. In this study, we found that coordinated amino acid changes took place at interacting sites of Slit and Robo by comparing the amino acids at these sites among different organisms. In addition, the high level correlation between evolutionary rate of Slit and Robo was identified in vertebrates. Furthermore, the sites under positive selection of slit and robo were detected in the same lineage such as mosquito and teleost. Overall, our results provide evidence for the coevolution between Slit and Robo.

Practical aspects of protein co-evolution

Co-evolution is a fundamental aspect of Evolutionary Theory. At the molecular level, co-evolutionary linkages between protein families have been used as indicators of protein interactions and functional relationships from long ago. Due to the complexity of the problem and the amount of genomic data required for these approaches to achieve good performances, it took a relatively long time from the appearance of the first ideas and concepts to the quotidian application of these approaches and their incorporation to the standard toolboxes of bioinformaticians and molecular biologists. Today, these methodologies are mature (both in terms of performance and usability/implementation), and the genomic information that feeds them large enough to allow their general application. This review tries to summarize the current landscape of co-evolution-based methodologies, with a strong emphasis on describing interesting cases where their application to important biological systems, alone or in combination with other computational and experimental approaches, allowed getting new insight into these.

The functional importance of co-evolving residues in proteins

Computational approaches for detecting co-evolution in proteins allow for the identification of protein-protein interaction networks in different organisms and the assignment of function to under-explored proteins. The detection of co-variation of amino acids within or between proteins, moreover, allows for the discovery of residue-residue contacts and highlights functional residues that can affect the binding affinity, catalytic activity, or substrate specificity of a protein. To explore the functional impact of co-evolutionary changes in proteins, a combined experimental and computational approach must be recruited. Here, we review recent studies that apply computational and experimental tools to obtain novel insight into the structure, function, and evolution of proteins. Specifically, we describe the application of co-evolutionary analysis for predicting high-resolution three-dimensional structures of proteins. In addition, we describe computational approaches followed by experimental analysis for identifying specificity-determining residues in proteins. Finally, we discuss studies addressing the importance of such residues in terms of the functional divergence of proteins, allowing proteins to evolve new functions while avoiding crosstalk with existing cellular pathways or forming reproductive barriers and hence promoting speciation.

Slow co-evolution of the MAGO and Y14 protein families is required for the maintenance of their obligate heterodimerization mode

The exon junction complex (EJC) plays important roles in RNA metabolisms and the development of eukaryotic organisms. MAGO (short form of MAGO NASHI) and Y14 (also Tsunagi or RBM8) are the EJC core components. Their biological roles have been well investigated in various species, but the evolutionary patterns of the two gene families and their protein-protein interactions are poorly known. Genome-wide survey suggested that the MAGO and Y14 two gene families originated in eukaryotic organisms with the maintenance of a low copy. We found that the two protein families evolved slowly; however, the MAGO family under stringent purifying selection evolved more slowly than the Y14 family that was under relative relaxed purifying selection. MAGO and Y14 were obliged to form heterodimer in a eukaryotic organism, and this obligate mode was plesiomorphic. Lack of binding of MAGO to Y14 as functional barrier was observed only among distantly species, suggesting that a slow co-evolution of the two protein families. Inter-protein co-evolutionary signal was further quantified in analyses of the Tol-MirroTree and co-evolution analysis using protein sequences. About 20% of the 41 significantly correlated mutation groups (involving 97 residues) predicted between the two families was clade-specific. Moreover, around half of the predicted co-evolved groups and nearly all clade-specific residues fell into the minimal interaction domains of the two protein families. The mutagenesis effects of the clade-specific residues strengthened that the co-evolution is required for obligate MAGO-Y14 heterodimerization mode. In turn, the obliged heterodimerization in an organism serves as a strong functional constraint for the co-evolution of the MAGO and Y14 families. Such a co-evolution allows maintaining the interaction between the proteins through large evolutionary time scales. Our work shed a light on functional evolution of the EJC genes in eukaryotes, and facilitates to understand the co-evolutionary processes among protein families.

Predicting protein-protein interaction by the mirrortree method: possibilities and limitations

Molecular co-evolution analysis as a sequence-only based method has been used to predict protein-protein interactions. In co-evolution analysis, Pearson's correlation within the mirrortree method is a well-known way of quantifying the correlation between protein pairs. Here we studied the mirrortree method on both known interacting protein pairs and sets of presumed non-interacting protein pairs, to evaluate the utility of this correlation analysis method for predicting protein-protein interactions within eukaryotes. We varied metrics for computing evolutionary distance and evolutionary span of the species analyzed. We found the differences between co-evolutionary correlation scores of the interacting and non-interacting proteins, normalized for evolutionary span, to be significantly predictive for proteins conserved over a wide range of eukaryotic clades (from mammals to fungi). On the other hand, for narrower ranges of evolutionary span, the predictive power was much weaker.

Is the bovine lysosomal phospholipase B-like protein an amidase?

The main function of lysosomal proteins is to degrade cellular macromolecules. We purified a novel lysosomal protein to homogeneity from bovine kidneys. By gene annotation, this protein is defined as a bovine phospholipase B-like protein 1 (bPLBD1) and, to better understand its biological function, we solved its structure at 1.9 Å resolution. We showed that bPLBD1 has uniform noncomplex-type N-glycosylation and that it localized to the lysosome. The first step in lysosomal protein transport, the initiation of mannose-6-phosphorylation by a N-acetylglucosamine-1-phosphotransferase, requires recognition of at least two distinct lysines on the protein surface. We identified candidate lysines by analyzing the structural and sequentially conserved N-glycosylation sites and lysines in bPLBD1 and in the homologous mouse PLBD2. Our model suggests that N408 is the primarily phosphorylated glycan, and K358 a key residue for N-acetylglucosamine-1-phosphotransferase recognition. Two other lysines, K334 and K342, provide the required second site for N-acetylglucosamine-1-phosphotransferase recognition. bPLBD1 is an N-terminal nucleophile (Ntn) hydrolase. By comparison with other Ntn-hydrolases, we conclude that the acyl moiety of PLBD1 substrate must be small to fit the putative binding pocket, whereas the space for the rest of the substrate is a large open cleft. Finally, as all the known substrates of Ntn-hydrolases have amide bonds, we suggest that bPLBD1 may be an amidase or peptidase instead of lipase, explaining the difficulty in finding a good substrate for any members of the PLBD family.

Accounting for epistatic interactions improves the functional analysis of protein structures

Motivation: The constraints under which sequence, structure and function coevolve are not fully understood. Bringing this mutual relationship to light can reveal the molecular basis of binding, catalysis and allostery, thereby identifying function and rationally guiding protein redesign. Underlying these relationships are the epistatic interactions that occur when the consequences of a mutation to a protein are determined by the genetic background in which it occurs. Based on prior data, we hypothesize that epistatic forces operate most strongly between residues nearby in the structure, resulting in smooth evolutionary importance across the structure.

Methods and Results: We find that when residue scores of evolutionary importance are distributed smoothly between nearby residues, functional site prediction accuracy improves. Accordingly, we designed a novel measure of evolutionary importance that focuses on the interaction between pairs of structurally neighboring residues. This measure that we term pair-interaction Evolutionary Trace yields greater functional site overlap and better structure-based proteome-wide functional predictions.

Conclusions: Our data show that the structural smoothness of evolutionary importance is a fundamental feature of the coevolution of sequence, structure and function. Mutations operate on individual residues, but selective pressure depends in part on the extent to which a mutation perturbs interactions with neighboring residues. In practice, this principle led us to redefine the importance of a residue in terms of the importance of its epistatic interactions with neighbors, yielding better annotation of functional residues, motivating experimental validation of a novel functional site in LexA and refining protein function prediction.

Contact:lichtarge@bcm.edu

Supplementary information:Supplementary data are available at Bioinformatics online.

Experimental analysis of co-evolution within protein complexes: the yeast exosome as a model

Extensive bioinformatics analysis suggests that the stability and function of protein complexes are maintained throughout evolution by coordinated changes (co-evolution) of complex subunits. Yet, relatively little is known regarding the actual dynamics of such processes and the functional implications of co-evolution within protein complexes, since most of the bioinformatics predictions were not analyzed experimentally. Here, we describe a systematic experimental approach that allows a step-by-step observation of the co-evolution process in protein complexes. The exosome complex, an essential complex exhibiting a 3'→5' RNA degradation activity, served as a model system. In this study, we show that exosome subunits diverged very early during fungal evolution. Interestingly, we found that despite significant differences in conservation between Rrp41 and Mtr3 both subunits exhibit similar divergence pattern and co-evolutionary behavior through fungi evolution. Activity analysis of mutated exosomes exposes another layer of co-evolution between the core subunits and RNA substrates. Overall, our approach allows the experimental analysis of co-evolution within protein complexes and together with bioinformatics analysis can significantly deepen our understanding of the evolution of these complexes.

An improved hypergeometric probability method for identification of functionally linked proteins using phylogenetic profiles

Predicting functions of proteins and alternatively spliced isoforms encoded in a genome is one of the important applications of bioinformatics in the post-genome era. Due to the practical limitation of experimental characterization of all proteins encoded in a genome using biochemical studies, bioinformatics methods provide powerful tools for function annotation and prediction. These methods also help minimize the growing sequence-to-function gap. Phylogenetic profiling is a bioinformatics approach to identify the influence of a trait across species and can be employed to infer the evolutionary history of proteins encoded in genomes. Here we propose an improved phylogenetic profile-based method which considers the co-evolution of the reference genome to derive the basic similarity measure, the background phylogeny of target genomes for profile generation and assigning weights to target genomes. The ordering of genomes and the runs of consecutive matches between the proteins were used to define phylogenetic relationships in the approach. We used Escherichia coli K12 genome as the reference genome and its 4195 proteins were used in the current analysis. We compared our approach with two existing methods and our initial results show that the predictions have outperformed two of the existing approaches. In addition, we have validated our method using a targeted protein-protein interaction network derived from protein-protein interaction database STRING. Our preliminary results indicates that improvement in function prediction can be attained by using coevolution-based similarity measures and the runs on to the same scale instead of computing them in different scales. Our method can be applied at the whole-genome level for annotating hypothetical proteins from prokaryotic genomes.

The 3.2 Å resolution structure of a receptor: CheA:CheW signaling complex defines overlapping binding sites and key residue interactions within bacterial chemosensory arrays

Bacterial chemosensory arrays are composed of extended networks of chemoreceptors (also known as methyl-accepting chemotaxis proteins, MCPs), the histidine kinase CheA, and the adaptor protein CheW. Models of these arrays have been developed from cryoelectron microscopy, crystal structures of binary and ternary complexes, NMR spectroscopy, mutational, data and biochemical studies. A new 3.2 Å resolution crystal structure of a Thermotoga maritima MCP protein interaction region in complex with the CheA kinase-regulatory module (P4-P5) and adaptor protein CheW provides sufficient detail to define residue contacts at the interfaces formed among the three proteins. As in a previous 4.5 Å resolution structure, CheA-P5 and CheW interact through conserved hydrophobic surfaces at the ends of their β-barrels to form pseudo 6-fold symmetric rings in which the two proteins alternate around the circumference. The interface between P5 subdomain 1 and CheW subdomain 2 was anticipated from previous studies, whereas the related interface between CheW subdomain 1 and P5 subdomain 2 has only been observed in these ring assemblies. The receptor forms an unexpected structure in that the helical hairpin tip of each subunit has "unzipped" into a continuous α-helix; four such helices associate into a bundle, and the tetramers bridge adjacent P5-CheW rings in the lattice through interactions with both P5 and CheW. P5 and CheW each bind a receptor helix with a groove of conserved hydrophobic residues between subdomains 1 and 2. P5 binds the receptor helix N-terminal to the tip region (lower site), whereas CheW binds the same helix with inverted polarity near the bundle end (upper site). Sequence comparisons among different evolutionary classes of chemotaxis proteins show that the binding partners undergo correlated changes at key residue positions that involve the lower site. Such evolutionary analyses argue that both CheW and P5 bind to the receptor tip at overlapping positions. Computational genomics further reveal that two distinct CheW proteins in Thermotogae utilize the analogous recognition motifs to couple different receptor classes to the same CheA kinase. Important residues for function previously identified by mutagenesis, chemical modification and biophysical approaches also map to these same interfaces. Thus, although the native CheW-receptor interaction is not observed in the present crystal structure, the bioinformatics and previous data predict key features of this interface. The companion study of the P5-receptor interface in native arrays (accompanying paper Piasta et al. (2013) Biochemistry, DOI: 10.1021/bi400385c) shows that, despite the non-native receptor fold in the present crystal structure, the local helix-in-groove contacts of the crystallographic P5-receptor interaction are present in native arrays and are essential for receptor regulation of kinase activity.

Emerging methods in protein co-evolution

Co-evolution is a fundamental component of the theory of evolution and is essential for understanding the relationships between species in complex ecological networks. A wide range of co-evolution-inspired computational methods has been designed to predict molecular interactions, but it is only recently that important advances have been made. Breakthroughs in the handling of phylogenetic information and in disentangling indirect relationships have resulted in an improved capacity to predict interactions between proteins and contacts between different protein residues. Here, we review the main co-evolution-based computational approaches, their theoretical basis, potential applications and foreseeable developments.

Residue mutations and their impact on protein structure and function: detecting beneficial and pathogenic changes

The present review focuses on the evolution of proteins and the impact of amino acid mutations on function from a structural perspective. Proteins evolve under the law of natural selection and undergo alternating periods of conservative evolution and of relatively rapid change. The likelihood of mutations being fixed in the genome depends on various factors, such as the fitness of the phenotype or the position of the residues in the three-dimensional structure. For example, co-evolution of residues located close together in three-dimensional space can occur to preserve global stability. Whereas point mutations can fine-tune the protein function, residue insertions and deletions (‘decorations’ at the structural level) can sometimes modify functional sites and protein interactions more dramatically. We discuss recent developments and tools to identify such episodic mutations, and examine their applications in medical research. Such tools have been tested on simulated data and applied to real data such as viruses or animal sequences. Traditionally, there has been little if any cross-talk between the fields of protein biophysics, protein structure–function and molecular evolution. However, the last several years have seen some exciting developments in combining these approaches to obtain an in-depth understanding of how proteins evolve. For example, a better understanding of how structural constraints affect protein evolution will greatly help us to optimize our models of sequence evolution. The present review explores this new synthesis of perspectives.

Evidence for the robustness of protein complexes to inter-species hybridization

Despite the tremendous efforts devoted to the identification of genetic incompatibilities underlying hybrid sterility and inviability, little is known about the effect of inter-species hybridization at the protein interactome level. Here, we develop a screening platform for the comparison of protein-protein interactions (PPIs) among closely related species and their hybrids. We examine in vivo the architecture of protein complexes in two yeast species (Saccharomyces cerevisiae and Saccharomyces kudriavzevii) that diverged 5-20 million years ago and in their F1 hybrids. We focus on 24 proteins of two large complexes: the RNA polymerase II and the nuclear pore complex (NPC), which show contrasting patterns of molecular evolution. We found that, with the exception of one PPI in the NPC sub-complex, PPIs were highly conserved between species, regardless of protein divergence. Unexpectedly, we found that the architecture of the complexes in F1 hybrids could not be distinguished from that of the parental species. Our results suggest that the conservation of PPIs in hybrids likely results from the slow evolution taking place on the very few protein residues involved in the interaction or that protein complexes are inherently robust and may accommodate protein divergence up to the level that is observed among closely related species.

The origins of the evolutionary signal used to predict protein-protein interactions

Background

The correlation of genetic distances between pairs of protein sequence alignments has been used to infer protein-protein interactions. It has been suggested that these correlations are based on the signal of co-evolution between interacting proteins. However, although mutations in different proteins associated with maintaining an interaction clearly occur (particularly in binding interfaces and neighbourhoods), many other factors contribute to correlated rates of sequence evolution. Proteins in the same genome are usually linked by shared evolutionary history and so it would be expected that there would be topological similarities in their phylogenetic trees, whether they are interacting or not. For this reason the underlying species tree is often corrected for. Moreover processes such as expression level, are known to effect evolutionary rates. However, it has been argued that the correlated rates of evolution used to predict protein interaction explicitly includes shared evolutionary history; here we test this hypothesis.

Results

In order to identify the evolutionary mechanisms giving rise to the correlations between interaction proteins, we use phylogenetic methods to distinguish similarities in tree topologies from similarities in genetic distances. We use a range of datasets of interacting and non-interacting proteins from Saccharomyces cerevisiae. We find that the signal of correlated evolution between interacting proteins is predominantly a result of shared evolutionary rates, rather than similarities in tree topology, independent of evolutionary divergence.

Conclusions

Since interacting proteins do not have tree topologies that are more similar than the control group of non-interacting proteins, it is likely that coevolution does not contribute much to, if any, of the observed correlations.

Co-evolution of RNA polymerase with RbpA in the phylum Actinobacteria

The role of RbpA in the backdrop of M. smegmatis showed that it rescues mycobacterial RNA polymerase from rifampicin-mediated inhibition (Dey et al., 2010; Dey et al., 2011). Paget and co-workers (Paget et al., 2001; Newell et al., 2006) have revealed that RbpA homologs occur exclusively in actinobacteria. Newell et al. (2006) showed that MtbRbpA, when complemented in a ∆rbpA mutant of S. coelicolor, showed a low recovery of MIC (from 0.75 to 2 μg/ml) as compared to complementation by native RbpA of S. coelicolor (MIC increases from 0.75 to 11 μg/ml). Our studies on MsRbpA show that it is a differential marker for M. smegmatis RNA polymerase as compared to E. coli RNA polymerase at IC50 levels of rifampicin. A recent sequence-based analysis by Lane and Darst (2010) has shown that RNA polymerases from Proteobacteria and Actinobacteria have had a divergent evolution. E. coli is a representative of Proteobacteria and M. smegmatis is an Actinobacterium. RbpA has an exclusive occurrence in Actinobacteria. Since protein-protein interactions might not be conserved across different species, therefore, the probable reason for the indifference of MsRbpA toward E. coli RNA polymerase could be the lineage-specific differences between actinobacterial and proteobacterial RNA polymerases. These observations led us to ask the question as to whether the evolution of RbpA in Actinobacteria followed the same route as that of RNA polymerase subunits from actinobacterial species. We show that the exclusivity of RbpA in Actinobacteria and the unique evolution of RNA polymerase in this phylum share a co-evolutionary link. We have addressed this issue by a blending of experimental and bioinformatics based approaches. They comprise of induction of bacterial cultures coupled to rifampicin-tolerance, transcription assays and statistical comparison of phylogenetic trees for different pairs of proteins in actinobacteria.

Molecular dynamics simulations and statistical coupling analysis reveal functional coevolution network of oncogenic mutations in the CDKN2A-CDK6 complex

Coevolution between proteins is crucial for understanding protein-protein interaction. Simultaneous changes allow a protein complex to maintain its overall structural-functional integrity. In this study, we combined statistical coupling analysis (SCA) and molecular dynamics simulations on the CDK6-CDKN2A protein complex to evaluate coevolution between proteins. We reconstructed an inter-protein residue coevolution network, consisting of 37 residues and 37 interactions. It shows that most of the coevolved residue pairs are spatially proximal. When the mutations happened, the stable local structures were broken up and thus the protein interaction was decreased or inhibited, with a following increased risk of melanoma. The identification of inter-protein coevolved residues in the CDK6-CDKN2A complex can be helpful for designing protein engineering experiments.

ABC transporters of antimicrobial peptides in Firmicutes bacteria - phylogeny, function and regulation

Antimicrobial peptides (AMPs) are a group of antibiotics that mainly target the cell wall of Gram-positive bacteria. Resistance is achieved by a variety of mechanisms including target alterations, changes in the cell's surface charge, expression of immunity peptides or by dedicated ABC transporters. The latter often provide the greatest level of protection. Apart from resistance, ABC transporters are also required for the export of peptides during biosynthesis. In this review the different AMP transporters identified to date in Firmicutes bacteria were classified into five distinct groups based on their domain architecture, two groups with a role in biosynthesis, and three involved in resistance. Comparison of the available information for each group regarding function, transport mechanism and gene regulation revealed distinguishing characteristics as well as common traits. For example, a strong correlation between transporter group and mode of gene regulation was observed, with three different types of two-component systems as well as XRE family transcriptional regulators commonly associated with individual transporter groups. Furthermore, the presented summary of the state-of-the-art on AMP transport in Firmicutes bacteria, discussed in the context of transporter phylogeny, provides insights into the mechanisms of substrate translocation and how this may result in resistance against compounds that bind extracellular targets.

Protein co-evolution: how do we combine bioinformatics and experimental approaches?

Molecular co-evolution is manifested by compensatory changes in proteins designed to enable adaptation to their natural environment. In recent years, bioinformatics approaches allowed for the detection of co-evolution at the level of the whole protein or of specific residues. Such efforts enabled prediction of protein-protein interactions, functional assignments of proteins and the identification of interacting residues, thereby providing information on protein structure. Still, despite such advances, relatively little is known regarding the functional implications of sequence divergence resulting from protein co-evolution. While bioinformatics approaches usually analyze thousands of proteins to obtain a broad view of protein co-evolution, experimental evaluation of protein co-evolution serves to study only individual proteins. In this review, we describe recent advances in bioinformatics and experimental efforts aimed at examining protein co-evolution. Accordingly, we discuss possible modes of crosstalk between the bioinformatics and experimental approaches to facilitate the identification of co-evolutionary signals in proteins and to understand their implications for the structure and function of proteins.

Incorporating information on predicted solvent accessibility to the co-evolution-based study of protein interactions

A widespread family of methods for studying and predicting protein interactions using sequence information is based on co-evolution, quantified as similarity of phylogenetic trees. Part of the co-evolution observed between interacting proteins could be due to co-adaptation caused by inter-protein contacts. In this case, the co-evolution is expected to be more evident when evaluated on the surface of the proteins or the internal layers close to it. In this work we study the effect of incorporating information on predicted solvent accessibility to three methods for predicting protein interactions based on similarity of phylogenetic trees. We evaluate the performance of these methods in predicting different types of protein associations when trees based on positions with different characteristics of predicted accessibility are used as input. We found that predicted accessibility improves the results of two recent versions of the mirrortree methodology in predicting direct binary physical interactions, while it neither improves these methods, nor the original mirrortree method, in predicting other types of interactions. That improvement comes at no cost in terms of applicability since accessibility can be predicted for any sequence. We also found that predictions of protein-protein interactions are improved when multiple sequence alignments with a richer representation of sequences (including paralogs) are incorporated in the accessibility prediction.

Interactome-wide prediction of protein-protein binding sites reveals effects of protein sequence variation in Arabidopsis thaliana

The specificity of protein-protein interactions is encoded in those parts of the sequence that compose the binding interface. Therefore, understanding how changes in protein sequence influence interaction specificity, and possibly the phenotype, requires knowing the location of binding sites in those sequences. However, large-scale detection of protein interfaces remains a challenge. Here, we present a sequence- and interactome-based approach to mine interaction motifs from the recently published Arabidopsis thaliana interactome. The resultant proteome-wide predictions are available via www.ab.wur.nl/sliderbio and set the stage for further investigations of protein-protein binding sites. To assess our method, we first show that, by using a priori information calculated from protein sequences, such as evolutionary conservation and residue surface accessibility, we improve the performance of interface prediction compared to using only interactome data. Next, we present evidence for the functional importance of the predicted sites, which are under stronger selective pressure than the rest of protein sequence. We also observe a tendency for compensatory mutations in the binding sites of interacting proteins. Subsequently, we interrogated the interactome data to formulate testable hypotheses for the molecular mechanisms underlying effects of protein sequence mutations. Examples include proteins relevant for various developmental processes. Finally, we observed, by analysing pairs of paralogs, a correlation between functional divergence and sequence divergence in interaction sites. This analysis suggests that large-scale prediction of binding sites can cast light on evolutionary processes that shape protein-protein interaction networks.

Computational structural analysis of protein interactions and networks

Protein interactions have been at the focus of computational biology in recent years. In particular, interest has come from two different communities--structural and systems biology. Here, we will discuss key systems and structural biology methods that have been used for analysis and prediction of protein-protein interactions and the insight these approaches have provided on the nature and organization of protein-protein interactions inside cells.

Effect of reference genome selection on the performance of computational methods for genome-wide protein-protein interaction prediction

Background

Recent progress in computational methods for predicting physical and functional protein-protein interactions has provided new insights into the complexity of biological processes. Most of these methods assume that functionally interacting proteins are likely to have a shared evolutionary history. This history can be traced out for the protein pairs of a query genome by correlating different evolutionary aspects of their homologs in multiple genomes known as the reference genomes. These methods include phylogenetic profiling, gene neighborhood and co-occurrence of the orthologous protein coding genes in the same cluster or operon. These are collectively known as genomic context methods. On the other hand a method called mirrortree is based on the similarity of phylogenetic trees between two interacting proteins. Comprehensive performance analyses of these methods have been frequently reported in literature. However, very few studies provide insight into the effect of reference genome selection on detection of meaningful protein interactions.

Methods

We analyzed the performance of four methods and their variants to understand the effect of reference genome selection on prediction efficacy. We used six sets of reference genomes, sampled in accordance with phylogenetic diversity and relationship between organisms from 565 bacteria. We used Escherichia coli as a model organism and the gold standard datasets of interacting proteins reported in DIP, EcoCyc and KEGG databases to compare the performance of the prediction methods.

Conclusions

Higher performance for predicting protein-protein interactions was achievable even with 100–150 bacterial genomes out of 565 genomes. Inclusion of archaeal genomes in the reference genome set improves performance. We find that in order to obtain a good performance, it is better to sample few genomes of related genera of prokaryotes from the large number of available genomes. Moreover, such a sampling allows for selecting 50–100 genomes for comparable accuracy of predictions when computational resources are limited.

Calcineurin homologous protein: a multifunctional Ca2+-binding protein family

The calcineurin homologous protein (CHP) belongs to an evolutionarily conserved Ca(2+)-binding protein subfamily. The CHP subfamily is composed of CHP1, CHP2, and CHP3, which in vertebrates share significant homology at the protein level with each other and between other Ca(2+)-binding proteins. The CHP structure consists of two globular domains containing from one to four EF-hand structural motifs (calcium-binding regions composed of two helixes, E and F, joined by a loop), the myristoylation, and nuclear export signals. These structural features are essential for the function of the three members of the CHP subfamily. Indeed, CHP1-CHP3 have multiple and diverse essential functions, ranging from the regulation of the plasma membrane Na(+)/H(+) exchanger protein function, to carrier vesicle trafficking and gene transcription. The diverse functions attributed to the CHP subfamily rendered an understanding of its action highly complex and often controversial. This review provides a comprehensive and organized examination of the properties and physiological roles of the CHP subfamily with a view to revealing a link between CHP diverse functions.

Gene duplications contribute to the overrepresentation of interactions between proteins of a similar age

BACKGROUND

The study of biological networks and how they have evolved is fundamental to our understanding of the cell. By investigating how proteins of different ages are connected in the protein interaction network, one can infer how that network has expanded in evolution, without the need for explicit reconstruction of ancestral networks. Studies that implement this approach show that proteins are often connected to proteins of a similar age, suggesting a simultaneous emergence of interacting proteins. There are several theories explaining this phenomenon, but despite the importance of gene duplication in genome evolution, none consider protein family dynamics as a contributing factor.

RESULTS

In an S. cerevisiae protein interaction network we investigate to what extent edges that arise from duplication events contribute to the observed tendency to interact with proteins of a similar age. We find that part of this tendency is explained by interactions between paralogs. Age is usually defined on the level of protein families, rather than individual proteins, hence paralogs have the same age. The major contribution however, is from interaction partners that are shared between paralogs. These interactions have most likely been conserved after a duplication event. To investigate to what extent a nearly neutral process of network growth can explain these results, we adjust a well-studied network growth model to incorporate protein families. Our model shows that the number of edges between paralogs can be amplified by subsequent duplication events, thus explaining the overrepresentation of interparalog edges in the data. The fact that interaction partners shared by paralogs are often of the same age as the paralogs does not arise naturally from our model and needs further investigation.

CONCLUSION

We amend previous theories that explain why proteins of a similar age prefer to interact by demonstrating that this observation can be partially explained by gene duplication events. There is an ongoing debate on whether the protein interaction network is predominantly shaped by duplication and subfunctionalization or whether network rewiring is most important. Our analyses of S. cerevisiae protein interaction networks demonstrate that duplications have influenced at least one property of the protein interaction network: how proteins of different ages are connected.

Requirements for efficient correction of ΔF508 CFTR revealed by analyses of evolved sequences

Misfolding of ΔF508 cystic fibrosis (CF) transmembrane conductance regulator (CFTR) underlies pathology in most CF patients. F508 resides in the first nucleotide-binding domain (NBD1) of CFTR near a predicted interface with the fourth intracellular loop (ICL4). Efforts to identify small molecules that restore function by correcting the folding defect have revealed an apparent efficacy ceiling. To understand the mechanistic basis of this obstacle, positions statistically coupled to 508, in evolved sequences, were identified and assessed for their impact on both NBD1 and CFTR folding. The results indicate that both NBD1 folding and interaction with ICL4 are altered by the ΔF508 mutation and that correction of either individual process is only partially effective. By contrast, combination of mutations that counteract both defects restores ΔF508 maturation and function to wild-type levels. These results provide a mechanistic rationale for the limited efficacy of extant corrector compounds and suggest approaches for identifying compounds that correct both defective steps.

Mirroring co-evolving trees in the light of their topologies

MOTIVATION

Determining the interaction partners among protein/domain families poses hard computational problems, in particular in the presence of paralogous proteins. Available approaches aim to identify interaction partners among protein/domain families through maximizing the similarity between trimmed versions of their phylogenetic trees. Since maximization of any natural similarity score is computationally difficult, many approaches employ heuristics to evaluate the distance matrices corresponding to the tree topologies in question. In this article, we devise an efficient deterministic algorithm which directly maximizes the similarity between two leaf labeled trees with edge lengths, obtaining a score-optimal alignment of the two trees in question.

RESULTS

Our algorithm is significantly faster than those methods based on distance matrix comparison: 1 min on a single processor versus 730 h on a supercomputer. Furthermore, we outperform the current state-of-the-art exhaustive search approach in terms of precision, while incurring acceptable losses in recall.

AVAILABILITY

A C implementation of the method demonstrated in this article is available at http://compbio.cs.sfu.ca/mirrort.htm

Tight coevolution of proliferating cell nuclear antigen (PCNA)-partner interaction networks in fungi leads to interspecies network incompatibility

The structure and connectivity of protein-protein interaction (PPI) networks are maintained throughout evolution by coordinated changes (coevolution) of network proteins. Despite extensive research, relatively little is known regarding the molecular basis and functional implications of the coevolution of PPI networks. Here, we used proliferating cell nuclear antigen, a hub protein that mediates DNA replication and repair in eukaryotes, as a model system to study the coevolution of PPI networks in fungi. Using a combined bioinformatics and experimental approach, we discovered that PCNA-partner interactions tightly coevolved in fungal species, leading to specific modes of recognition. We found that fungal proliferating cell nuclear antigen-partner interaction networks diverged into two distinct groups as a result of such coevolution and that hybrid networks of these groups are functionally noncompatible in Saccharomyces cerevisiae. Our results indicate that the coevolution of PPI networks can form functional barriers between fungal species, and thus can promote and fix speciation.

Role of computational methods in pharmaceutical sciences

Over the past two decades computational methods have eased up the financial and experimental burden of early drug discovery process. The in silico methods have provided support in terms of databases, data mining of large genomes, network analysis, systems biology on the bioinformatics front and structure-activity relationship, similarity analysis, docking, and pharmacophore methods for lead design and optimization. This review highlights some of the applications of bioinformatics and chemoinformatics methods that have enriched the field of drug discovery. In addition, the review also provided insights into the use of free energy perturbation methods for efficiently computing binding energy. These in silico methods are complementary and can be easily integrated into the traditional in vitro and in vivo methods to test pharmacological hypothesis.

InterEvol database: exploring the structure and evolution of protein complex interfaces

Capturing how the structures of interacting partners evolved at their binding interfaces is a fundamental issue for understanding interactomes evolution. In that scope, the InterEvol database was designed for exploring 3D structures of homologous interfaces of protein complexes. For every chain forming a complex in the protein data bank (PDB), close and remote structural interologs were identified providing essential snapshots for studying interfaces evolution. The database provides tools to retrieve and visualize these structures. In addition, pre-computed multiple sequence alignments of most likely interologs retrieved from a wide range of species can be downloaded to enrich the analysis. The database can be queried either directly by pdb code or keyword but also from the sequence of one or two partners. Interologs multiple sequence alignments can also be recomputed online with tailored parameters using the InterEvolAlign facility. Last, an InterEvol PyMol plugin was developed to improve interactive exploration of structures versus sequence alignments at the interfaces of complexes. Based on a series of automatic methods to extract structural and sequence data, the database will be monthly updated. Structures coordinates and sequence alignments can be queried and downloaded from the InterEvol web interface at http://biodev.cea.fr/interevol/.

Mitochondrial-nuclear co-evolution and its effects on OXPHOS activity and regulation

Factors required for mitochondrial function are encoded both by the nuclear and mitochondrial genomes. The order of magnitude higher mutation rate of animal mitochondrial DNA (mtDNA) enforces tight co-evolution of mtDNA and nuclear DNA encoded factors. In this essay we argue that such co evolution exists at the population and inter-specific levels and affect disease susceptibility. We also argue for the existence of three modes of co-evolution in the mitochondrial genetic system, which include the interaction of mtDNA and nuclear DNA encoded proteins, nuclear protein - mtDNA-encoded RNA interaction within the mitochondrial translation machinery and nuclear DNA encoded proteins-mtDNA binging sites interaction in the frame of the mtDNA replication and transcription machineries. These modes of co evolution require co-regulation of the interacting factors encoded by the two genomes. Thus co evolution plays an important role in modulating mitochondrial activity. This article is part of a Special Issue entitled: Mitochondrial Gene Expression.

Coevolution of amino acid residues in the key photosynthetic enzyme Rubisco

BACKGROUND

One of the key forces shaping proteins is coevolution of amino acid residues. Knowing which residues coevolve in a particular protein may facilitate our understanding of protein evolution, structure and function, and help to identify substitutions that may lead to desired changes in enzyme kinetics. Rubisco, the most abundant enzyme in biosphere, plays an essential role in the process of carbon fixation through photosynthesis, thus facilitating life on Earth. This makes Rubisco an important model system for studying the dynamics of protein fitness optimization on the evolutionary landscape. In this study we investigated the selective and coevolutionary forces acting on large subunit of land plants Rubisco using Markov models of codon substitution and clustering approaches applied to amino acid substitution histories.

RESULTS

We found that both selection and coevolution shape Rubisco, and that positively selected and coevolving residues have their specifically favored amino acid composition and pairing preference. The mapping of these residues on the known Rubisco tertiary structures showed that the coevolving residues tend to be in closer proximity with each other compared to the background, while positively selected residues tend to be further away from each other. This study also reveals that the residues under positive selection or coevolutionary force are located within functionally important regions and that some residues are targets of both positive selection and coevolution at the same time.

CONCLUSION

Our results demonstrate that coevolution of residues is common in Rubisco of land plants and that there is an overlap between coevolving and positively selected residues. Knowledge of which Rubisco residues are coevolving and positively selected could be used for further work on structural modeling and identification of substitutions that may be changed in order to improve efficiency of this important enzyme in crops.

Selection of organisms for the co-evolution-based study of protein interactions

BACKGROUND

The prediction and study of protein interactions and functional relationships based on similarity of phylogenetic trees, exemplified by the mirrortree and related methodologies, is being widely used. Although dependence between the performance of these methods and the set of organisms used to build the trees was suspected, so far nobody assessed it in an exhaustive way, and, in general, previous works used as many organisms as possible. In this work we asses the effect of using different sets of organism (chosen according with various phylogenetic criteria) on the performance of this methodology in detecting protein interactions of different nature.

RESULTS

We show that the performance of three mirrortree-related methodologies depends on the set of organisms used for building the trees, and it is not always directly related to the number of organisms in a simple way. Certain subsets of organisms seem to be more suitable for the predictions of certain types of interactions. This relationship between type of interaction and optimal set of organism for detecting them makes sense in the light of the phylogenetic distribution of the organisms and the nature of the interactions.

CONCLUSIONS

In order to obtain an optimal performance when predicting protein interactions, it is recommended to use different sets of organisms depending on the available computational resources and data, as well as the type of interactions of interest.

Correlated evolution of transcription factors and their binding sites

MOTIVATION

The interaction between transcription factor (TF) and transcription factor binding site (TFBS) is essential for gene regulation. Mutation in either the TF or the TFBS may weaken their interaction and thus result in abnormalities. To maintain such vital interaction, a mutation in one of the interacting partners might be compensated by a corresponding mutation in its binding partner during the course of evolution. Confirming this co-evolutionary relationship will guide us in designing protein sequences to target a specific DNA sequence or in predicting TFBS for poorly studied proteins, or even correcting and rescuing disease mutations in clinical applications.

RESULTS

Based on six, publicly available, experimentally validated TF-TFBS binding datasets for the basic Helix-Loop-Helix (bHLH) family, Homeo family, High-Mobility Group (HMG) family and Transient Receptor Potential channels (TRP) family, we showed that the evolutions of the TFs and their TFBSs are significantly correlated across eukaryotes. We further developed a mutual information-based method to identify co-evolved protein residues and DNA bases. This research sheds light on the dynamic relationship between TF and TFBS during their evolution. The same principle and strategy can be applied to co-evolutionary studies on protein-DNA interactions in other protein families.

AVAILABILITY

All the datasets, scripts and other related files have been made freely available at: http://jjwanglab.org/co-evo.

CONTACT

junwen@uw.edu.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Protein coadaptation and the design of novel approaches to identify protein-protein interactions

Proteins rarely function in isolation but they form part of complex networks of interactions with other proteins within or among cells. The importance of a particular protein for cell viability is directly dependent upon the number of interactions where it participates and the function it performs: the larger the number of interactions of a protein the greater its functional importance is for the cell. With the advent of genome sequencing and "omics" technologies it became feasible conducting large-scale searches for protein interacting partners. Unfortunately, the accuracy of such analyses has been underwhelming owing to methodological limitations and to the inherent complexity of protein interactions. In addition to these experimental approaches, many computational methods have been developed to identify protein-protein interactions by assuming that interacting proteins coevolve resulting from the coadaptation dynamics between the amino acids of their interacting faces. We review the main technological advances made in the field of interactomics and discuss the feasibility of computational methods to identify protein-protein interactions based on the estimation of coevolution. As proof-of-concept, we present a classical case study: the interactions of cell surface proteins (receptors) and their ligands. Finally, we take this discussion one step forward to include interactions between organisms and species to understand the generation of biological complexity. Development of technologies for accurate detection of protein-protein interactions may shed light on processes that go from the fine-tuning of pathways and metabolic networks to the emergence of biological complexity.

The evolution of MDM2 family genes

MDM2 and MDM4 are proto-oncoproteins that bind to and inhibit members of the p53 protein family, p53, p73 and possibly p63. p53 is a mammalian tumor suppressor and p63 and p73 are critical for development. With the sequencing of genomes from multiple organisms there is mounting evidence for a consensus scenario of p53 gene family evolution. A single p53/p63/p73 gene is in invertebrates and required for maintenance of germline DNA. Gene duplication occurred in an ancestor in common with cartilaginous fishes, giving rise to a separate p53 gene and at least one ancestral p63/p73 gene. In bony vertebrates, all three p53 gene family paralogs, p53, p63, and p73 are distinct genes. This raises the question of how MDM2 and MDM4 genes evolved. We show evidence that MDM2 and MDM4 arose from a gene duplication event prior to the emergence of bony vertebrates more than 440 millionyears ago. Comparative genome studies indicate that invertebrate organisms have only one MDM homolog. In jawed vertebrates, the p53-binding domains of MDM2 and MDM4 proteins evolved at a high rate, approaching the evolution rate of the MDM2-binding domain of p53. However, the MDM2-binding domain of p73 exhibits markedly stronger conservation suggesting novel p53-independent functions. The most conserved domain within all MDM2 family members is the RING domain of the MDM2 ortholog which is responsible for ubiquitination of p53 and heterodimerization with MDM4. We suggest a model where oligomerization is an ancient function of MDM and ubiquitination activity was acquired later near the MDM gene duplication event coinciding with the time of the emergence of p53 as a distinct gene.

Coevolution of ABC transporters and two-component regulatory systems as resistance modules against antimicrobial peptides in Firmicutes Bacteria

In Firmicutes bacteria, ATP-binding cassette (ABC) transporters have been recognized as important resistance determinants against antimicrobial peptides. Together with neighboring two-component systems (TCSs), which regulate their expression, they form specific detoxification modules. Both the transport permease and sensor kinase components show unusual domain architecture: the permeases contain a large extracellular domain, while the sensor kinases lack an obvious input domain. One of the best-characterized examples is the bacitracin resistance module BceRS-BceAB of Bacillus subtilis. Strikingly, in this system, the ABC transporter and TCS have an absolute mutual requirement for each other in both sensing of and resistance to bacitracin, suggesting a novel mode of signal transduction in which the transporter constitutes the actual sensor. We identified over 250 such BceAB-like ABC transporters in the current databases. They occurred almost exclusively in Firmicutes bacteria, and 80% of the transporters were associated with a BceRS-like TCS. Phylogenetic analyses of the permease and sensor kinase components revealed a tight evolutionary correlation. Our findings suggest a direct regulatory interaction between the ABC transporters and TCSs, mediating communication between both components. Based on their observed coclustering and conservation of response regulator binding sites, we could identify putative corresponding two-component systems for transporters lacking a regulatory system in their immediate neighborhood. Taken together, our results show that these types of ABC transporters and TCSs have coevolved to form self-sufficient detoxification modules against antimicrobial peptides, widely distributed among Firmicutes bacteria.

Towards the prediction of protein interaction partners using physical docking

Prediction of physical protein-protein interactions represents a key challenge in computational systems biology. This study provides a proof-of-principle that high-throughput in silico protein docking results can be used to predict interaction partners.

Deciphering the whole network of protein interactions for a given proteome (‘interactome') is the goal of many experimental and computational efforts in Systems Biology. Separately the prediction of the structure of protein complexes by docking methods is a well-established scientific area. To date, docking programs have not been used to predict interaction partners. We provide a proof of principle for such an approach. Using a set of protein complexes representing known interactors in their unbound form, we show that a standard docking program can distinguish the true interactors from a background of 922 non-redundant potential interactors. We additionally show that true interactions can be distinguished from non-likely interacting proteins within the same structural family. Our approach may be put in the context of the proposed ‘funnel-energy model'; the docking algorithm may not find the native complex, but it distinguishes binding partners because of the higher probability of favourable models compared with a collection of non-binders. The potential exists to develop this proof of principle into new approaches for predicting interaction partners and reconstructing biological networks.

Correlated evolution among six gene families in Drosophila revealed by parallel change of gene numbers

Proteins involved in a pathway are likely to evolve in a correlated fashion, and coevolving gene families tend to undergo complementary gains and losses. Accordingly, gene copy numbers (i.e., repertoire size) tend to show parallel changes during the evolution of coevolving gene families. To test and verify this hypothesis, here we describe positive correlations among the repertoire sizes of six gene families, that is, trypsin-like serine protease, odorant-binding protein, odorant receptor, gustatory receptor, cytochrome P450, and glutathione S-transferase after excluding the possibility of phylogenetic constraint and random drift. The observed correlations are indicative of parallel changes in the repertoire sizes of the six gene families that are due to similar demands for the quantity of these different genes in different lineages of Drosophila. In conclusion, we propose that the correlated evolution among these six gene families in Drosophila is a signature of a parallel response to ecological adaptation.

Molecular evolution of protein conformational changes revealed by a network of evolutionarily coupled residues

An improved understanding of protein conformational changes has broad implications for elucidating the mechanisms of various biological processes and for the design of protein engineering experiments. Understanding rearrangements of residue interactions is a key component in the challenge of describing structural transitions. Evolutionary properties of protein sequences and structures are extensively studied; however, evolution of protein motions, especially with respect to interaction rearrangements, has yet to be explored. Here, we investigated the relationship between sequence evolution and protein conformational changes and discovered that structural transitions are encoded in amino acid sequences as coevolving residue pairs. Furthermore, we found that highly coevolving residues are clustered in the flexible regions of proteins and facilitate structural transitions by forming and disrupting their interactions cooperatively. Our results provide insight into the evolution of protein conformational changes and help to identify residues important for structural transitions.

Adaptive evolution of the Populus tremula photoperiod pathway

Perennial plants monitor seasonal changes through changes in environmental conditions such as the quantity and quality of light and genes in the photoperiodic pathway are known to be involved in controlling these processes. Here, we examine 25 of genes from the photoperiod pathway in Populus tremula (Salicaceae) for signatures of adaptive evolution. Overall, levels of synonymous polymorphism in the 25 genes are lower than at control loci selected randomly from the genome. This appears primarily to be caused by lower levels of synonymous polymorphism in genes associated with the circadian clock. Natural selection appears to play an important role in shaping protein evolution at several of the genes in the photoperiod pathways, which is highlighted by the fact that approximately 40% of the genes from the photoperiod pathway have estimates of selection on nonsynonymous polymorphisms that are significantly different from zero. A surprising observation we make is that circadian clock-associated genes appear to be over-represented among the genes showing elevated rates of protein evolution; seven genes are evolving under positive selection and all but one of these genes are involved in the circadian clock of Populus.

Using coevolution to predict protein-protein interactions

Bioinformatic methods to predict protein-protein interactions (PPI) via coevolutionary analysis have -positioned themselves to compete alongside established in vitro methods, despite a lack of understanding for the underlying molecular mechanisms of the coevolutionary process. Investigating the alignment of coevolutionary predictions of PPI with experimental data can focus the effective scope of prediction and lead to better accuracies. A new rate-based coevolutionary method, MMM, preferentially finds obligate interacting proteins that form complexes, conforming to results from studies based on coimmunoprecipitation coupled with mass spectrometry. Using gold-standard databases as a benchmark for accuracy, MMM surpasses methods based on abundance ratios, suggesting that correlated evolutionary rates may yet be better than coexpression at predicting interacting proteins. At the level of protein domains, -coevolution is difficult to detect, even with MMM, except when considering small-scale experimental data involving proteins with multiple domains. Overall, these findings confirm that coevolutionary -methods can be confidently used in predicting PPI, either independently or as drivers of coimmunoprecipitation experiments.

Quantitative control of organ shape by combinatorial gene activity

The development of organs with particular shapes, like wings or flowers, depends on regional activity of transcription factors and signalling molecules. However, the mechanisms that link these molecular activities to the morphogenetic events underlying shape are poorly understood. Here we describe a combination of experimental and computational approaches that address this problem, applying them to a group of genes controlling flower shape in the Snapdragon (Antirrhinum). Four transcription factors are known to play a key role in the control of floral shape and asymmetry in Snapdragon. We use quantitative shape analysis of mutants for these factors to define principal components underlying flower shape variation. We show that each transcription factor has a specific effect on the shape and size of regions within the flower, shifting the position of the flower in shape space. These shifts are further analysed by generating double mutants and lines that express some of the genes ectopically. By integrating these observations with known gene expression patterns and interactions, we arrive at a combinatorial scheme for how regional effects on shape are genetically controlled. We evaluate our scheme by incorporating the proposed interactions into a generative model, where the developing flower is treated as a material sheet that grows according to how genes modify local polarities and growth rates. The petal shapes generated by the model show a good quantitative match with those observed experimentally for each petal in numerous genotypes, thus validating the hypothesised scheme. This article therefore shows how complex shapes can be accounted for by combinatorial effects of transcription factors on regional growth properties. This finding has implications not only for how shapes develop but also for how they may have evolved through tinkering with transcription factors and their targets.