Extending the accuracy limits of prediction for side-chain conformations

Categorization of hotspots into three types - weak, moderate and strong to distinguish protein-protein versus protein-peptide interactions

Protein-protein and protein-peptide interactions (PPI and PPepI) belong to a similar category of interactions, yet seemingly subtle differences exist among them. To characterize differences between protein-protein (PP) and protein-peptide (PPep) interactions, we have focussed on two important classes of residues-hotspot and anchor residues. Using implicit solvation-based free energy calculations, a very large-scale alanine scanning has been performed on benchmark datasets, consisting of over 5700 interface residues. The differences in the two categories are more pronounced, if the data were divided into three distinct types, namely - weak hotspots (having binding free energy loss upon Ala mutation, ΔΔG, ∼2-10 kcal/mol), moderate hotspots (ΔΔG, ∼10-20 kcal/mol) and strong hotspots (ΔΔG ≥ ∼20 kcal/mol). The analysis suggests that for PPI, weak hotspots are predominantly populated by polar and hydrophobic residues. The distribution shifts towards charged and polar residues for moderate hotspot and charged residues (principally Arg) are overwhelmingly present in the strong hotspot. On the other hand, in the PPepI dataset, the distribution shifts from predominantly hydrophobic and polar (in the weak type) to almost similar preference for polar, hydrophobic and charged residues (in moderate type) and finally the charged residue (Arg) and Trp are mostly occupied in the strong type. The preferred anchor residues in both categories are Arg, Tyr and Leu, possessing bulky side chain and which also strike a delicate balance between side chain flexibility and rigidity. The present knowledge should aid in effective design of biologics, by augmentation or disruption of PPIs with peptides or peptidomimetics.Communicated by Ramaswamy H. Sarma.

CAPRI-Q: The CAPRI resource evaluating the quality of predicted structures of protein complexes

Protein interactions are essential for cellular processes. In recent years there has been significant progress in computational prediction of 3D structures of individual protein chains, with the best-performing algorithms reaching sub-Ångström accuracy. These techniques are now finding their way into the prediction of protein interactions, adding to the existing modeling approaches. The community-wide Critical Assessment of Predicted Interactions (CAPRI) has been a catalyst for the development of procedures for the structural modeling of protein assemblies by organizing blind prediction experiments. The predicted structures are assessed against unpublished experimentally determined structures using a set of metrics with proven robustness that have been established in the CAPRI community. In addition, several advanced benchmarking databases provide targets against which users can test docking and assembly modeling software. These include the Protein-Protein Docking Benchmark, the CAPRI Scoreset, and the Dockground database, all developed by members of the CAPRI community. Here we present CAPRI-Q, a stand-alone model quality assessment tool, which can be freely downloaded or used via a publicly available web server. This tool applies the CAPRI metrics to assess the quality of query structures against given target structures, along with other popular quality metrics such as DockQ, TM-score and l-DDT, and classifies the models according to the CAPRI model quality criteria. The tool can handle a variety of protein complex types including those involving peptides, nucleic acids, and oligosaccharides. The source code is freely available from https://gitlab.in2p3.fr/cmsb-public/CAPRI-Q and its web interface through the Dockground resource at https://dockground.compbio.ku.edu/assessment/.

Computational design of carboxylase for the synthesis of 4-hydroxyisophthalic acid from p-hydroxybenzoic acid by fixing CO2

Carbon dioxide (CO2) emissions constitute the primary contribution to global climate change. Synthetic CO2 fixation represents an exceptionally appealing and sustainable method for carbon neutralization. Unlike the limitations of chemical catalysis, biological CO2 fixation displays high selectivity and the ability to operate under mild conditions. The superfamily of amidohydrolases has demonstrated the ability to synthesize a range of aromatic monocarboxylic acids. However, there is a scarcity of reported carboxylases capable of synthesizing aromatic dicarboxylic acids. Among these, 4-hydroxyisophthalic acid holds significant potential for applications across various fields, yet no enzyme has been reported for its synthesis. In this study, we developed for the first time that exhibits starting activity in fixing CO2 to synthesize 4-hydroxyisophthalic acid. Furthermore, we have devised a computational strategy that effectively enhances the catalytic activity of this enzyme. A focused library comprising only 13 variants was generated. Experimental validation confirmed a threefold improvement in the carboxylation activity of the optimal variant (L47M). The computational enzyme design strategy proposed in this paper demonstrates broad applicability in developing carboxylases for synthesizing other aromatic dicarboxylic acids. This lays the groundwork for leveraging biocatalysis in industrial synthesis for CO2 fixation.

PMSPcnn: Predicting protein stability changes upon single point mutations with convolutional neural network

Protein missense mutations and resulting protein stability changes are important causes for many human genetic diseases. However, the accurate prediction of stability changes due to mutations remains a challenging problem. To address this problem, we have developed an unbiased effective model: PMSPcnn that is based on a convolutional neural network. We have included an anti-symmetry property to build a balanced training dataset, which improves the prediction, in particular for stabilizing mutations. Persistent homology, which is an effective approach for characterizing protein structures, is used to obtain topological features. Additionally, a regression stratification cross-validation scheme has been proposed to improve the prediction for mutations with extreme ΔΔG. For three test datasets: Ssym, p53, and myoglobin, PMSPcnn achieves a better performance than currently existing predictors. PMSPcnn also outperforms currently available methods for membrane proteins. Overall, PMSPcnn is a promising method for the prediction of protein stability changes caused by single point mutations.

Disallowed spots in protein structures

Ramachandran (ϕ, ψ) steric map was introduced in 1963 to describe available conformation space for protein structures. Subsequently, residues were observed in high-energy disallowed regions of the map. To unequivocally identify the locations of disallowed conformations of residues, we got 36 noise-free protein structures (resolution ≤1 Å, Rwork/Rfree ≤ 0.10). These stringent criteria were applied to rule out data or model errors or any crystallographic disorders. No disallowed conformation was found in the dataset. Further, we also examined disallowed conformations in a larger dataset (resolution ≤1.5 Å, devoid of any model errors, or disorders). The observed locations of disallowed residues are referred as disallowed spots. These spots include short loops of 3-5 residues, and locations where residues participate in disulfide bonding or intramolecular interactions or inter-molecular interactions with neighboring water, metals or ligands. Conformational sampling revealed that short loops in between secondary structures hardly have any opportunity to relieve from conformational strain. Residues involved in interactions, which provide energetic compensation for high-energy conformational states, were relieved from strain once the causative interaction was removed. The present study aims to identify disallowed spots in the native state of proteins, wherein residues are forced to be trapped in high-energy disallowed conformations. Moreover, it was also observed that pre-Pro, Ser, Asp, trans-Pro, Val, Asn & Gly have higher tendency to occur in disallowed conformation, which could be attributed to factors such as conformational restrictions, residue propensity of secondary structures and compensating sidechain and mainchain interactions, stabilizing turn-mimics.

Computational design of highly efficient thermostable MHET hydrolases and dual enzyme system for PET recycling

Recently developed enzymes for the depolymerization of polyethylene terephthalate (PET) such as FAST-PETase and LCC-ICCG are inhibited by the intermediate PET product mono(2-hydroxyethyl) terephthalate (MHET). Consequently, the conversion of PET enzymatically into its constituent monomers terephthalic acid (TPA) and ethylene glycol (EG) is inefficient. In this study, a protein scaffold (1TQH) corresponding to a thermophilic carboxylesterase (Est30) was selected from the structural database and redesigned in silico. Among designs, a double variant KL-MHETase (I171K/G130L) with a similar protein melting temperature (67.58 °C) to that of the PET hydrolase FAST-PETase (67.80 °C) exhibited a 67-fold higher activity for MHET hydrolysis than FAST-PETase. A fused dual enzyme system comprising KL-MHETase and FAST-PETase exhibited a 2.6-fold faster PET depolymerization rate than FAST-PETase alone. Synergy increased the yield of TPA by 1.64 fold, and its purity in the released aromatic products reached 99.5%. In large reaction systems with 100 g/L substrate concentrations, the dual enzyme system KL36F achieved over 90% PET depolymerization into monomers, demonstrating its potential applicability in the industrial recycling of PET plastics. Therefore, a dual enzyme system can greatly reduce the reaction and separation cost for sustainable enzymatic PET recycling.

Persistent topological Laplacian analysis of SARS-CoV-2 variants

Topological data analysis (TDA) is an emerging field in mathematics and data science. Its central technique, persistent homology, has had tremendous success in many science and engineering disciplines. However, persistent homology has limitations, including its inability to handle heterogeneous information, such as multiple types of geometric objects; being qualitative rather than quantitative, e.g., counting a 5-member ring the same as a 6-member ring, and a failure to describe non-topological changes, such as homotopic changes in protein-protein binding. Persistent topological Laplacians (PTLs), such as persistent Laplacian and persistent sheaf Laplacian, were proposed to overcome the limitations of persistent homology. In this work, we examine the modeling and analysis power of PTLs in the study of the protein structures of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike receptor binding domain (RBD). First, we employ PTLs to study how the RBD mutation-induced structural changes of RBD-angiotensin-converting enzyme 2 (ACE2) binding complexes are captured in the changes of spectra of the PTLs among SARS-CoV-2 variants. Additionally, we use PTLs to analyze the binding of RBD and ACE2-induced structural changes of various SARS-CoV-2 variants. Finally, we explore the impacts of computationally generated RBD structures on a topological deep learning paradigm and predictions of deep mutational scanning datasets for the SARS-CoV-2 Omicron BA.2 variant. Our results indicate that PTLs have advantages over persistent homology in analyzing protein structural changes and provide a powerful new TDA tool for data science.

Using High-Throughput Molecular Dynamics Simulation to Enhance the Computational Design of Kemp Elimination Enzymes

Computational enzyme design has been successfully applied to identify new alternatives to natural enzymes for the biosynthesis of important compounds. However, the moderate catalytic activities of de novo designed enzymes indicate that the modeling accuracy of current computational enzyme design methods should be improved. Here, high-throughput molecular dynamics simulations were used to enhance computational enzyme design, thus allowing the identification of variants with higher activities in silico. Different time schemes of high-throughput molecular dynamics simulations were tested to identify the catalytic features of evolved Kemp eliminases. The 20 × 1 ns molecular dynamics simulation scheme was sufficiently accurate and computationally viable to screen the computationally designed massive variants of Kemp elimination enzymes. The developed hybrid computational strategy was used to redesign the most active Kemp eliminase, HG3.17, and five variants were generated and experimentally confirmed to afford higher catalytic efficiencies than that of HG3.17, with one double variant (D52Q/A53S) exhibiting a 55% increase. The hybrid computational enzyme design strategy is general and computationally economical, with which we anticipate the efficient creation of practical enzymes for industrial biocatalysis.

Persistent spectral theory-guided protein engineering

While protein engineering, which iteratively optimizes protein fitness by screening the gigantic mutational space, is constrained by experimental capacity, various machine learning models have substantially expedited protein engineering. Three-dimensional protein structures promise further advantages, but their intricate geometric complexity hinders their applications in deep mutational screening. Persistent homology, an established algebraic topology tool for protein structural complexity reduction, fails to capture the homotopic shape evolution during the filtration of a given data. This work introduces a Topology-offered protein Fitness (TopFit) framework to complement protein sequence and structure embeddings. Equipped with an ensemble regression strategy, TopFit integrates the persistent spectral theory, a new topological Laplacian, and two auxiliary sequence embeddings to capture mutation-induced topological invariant, shape evolution, and sequence disparity in the protein fitness landscape. The performance of TopFit is assessed by 34 benchmark datasets with 128,634 variants, involving a vast variety of protein structure acquisition modalities and training set size variations.

COVID-19 Vaccines: Computational tools and Development

The 2019 coronavirus outbreak, also known as COVID-19, poses a serious threat to global health and has already had widespread, devastating effects around the world. Scientists have been working tirelessly to develop vaccines to stop the virus from spreading as much as possible, as its cure has not yet been found. As of December 2022, 651,918,402 cases and 6,656,601 deaths had been reported. Globally, over 13 billion doses of vaccine have been administered, representing 64.45% of the world's population that has received the vaccine. To expedite the vaccine development process, computational tools have been utilized. This paper aims to analyze some computational tools that aid vaccine development by presenting positive evidence for proving the efficacy of these vaccines to suppress the spread of the virus and for the use of computational tools in the development of vaccines for emerging diseases.

ChanFAD: A Functional Annotation Database for Ion Channels

Ion channels are integral membrane protein complexes critical for regulation of membrane potential, cell volume, and other signaling events. As complex molecular assemblies with many interacting partners, ion channels have multiple structural and functional domains. While channel sequence and functional data are readily available across multiple online resources, there is an unmet need for functional annotation directly relating primary sequence information, 2D interactions, and three-dimensional protein structure. To this end, we present ChanFAD (Channel Functional Annotation Database), to provide the research community with a centralized resource for ion channel structure and functional data. ChanFAD provides functional annotation of PDB structures built on the National Center for Biotechnology Information’s iCn3D structure viewing tool while providing additional information such as primary sequence, organism, and relevant links to other databases. Here we provide a brief tour of ChanFAD functionality while showing example use cases involving drug-channel interactions and structural changes based on mutation. ChanFAD is freely available and can be accessed at https://www.chanfad.org/.

Simulation and Machine Learning Methods for Ion-Channel Structure Determination, Mechanistic Studies and Drug Design

Ion channels are expressed in almost all living cells, controlling the in-and-out communications, making them ideal drug targets, especially for central nervous system diseases. However, owing to their dynamic nature and the presence of a membrane environment, ion channels remain difficult targets for the past decades. Recent advancement in cryo-electron microscopy and computational methods has shed light on this issue. An explosion in high-resolution ion channel structures paved way for structure-based rational drug design and the state-of-the-art simulation and machine learning techniques dramatically improved the efficiency and effectiveness of computer-aided drug design. Here we present an overview of how simulation and machine learning-based methods fundamentally changed the ion channel-related drug design at different levels, as well as the emerging trends in the field.

Evolution of the Automatic Rhodopsin Modeling (ARM) Protocol

In recent years, photoactive proteins such as rhodopsins have become a common target for cutting-edge research in the field of optogenetics. Alongside wet-lab research, computational methods are also developing rapidly to provide the necessary tools to analyze and rationalize experimental results and, most of all, drive the design of novel systems. The Automatic Rhodopsin Modeling (ARM) protocol is focused on providing exactly the necessary computational tools to study rhodopsins, those being either natural or resulting from mutations. The code has evolved along the years to finally provide results that are reproducible by any user, accurate and reliable so as to replicate experimental trends. Furthermore, the code is efficient in terms of necessary computing resources and time, and scalable in terms of both number of concurrent calculations as well as features. In this review, we will show how the code underlying ARM achieved each of these properties.

Persistent spectral based ensemble learning (PerSpect-EL) for protein-protein binding affinity prediction

Protein-protein interactions (PPIs) play a significant role in nearly all cellular and biological activities. Data-driven machine learning models have demonstrated great power in PPIs. However, the design of efficient molecular featurization poses a great challenge for all learning models for PPIs. Here, we propose persistent spectral (PerSpect) based PPI representation and featurization, and PerSpect-based ensemble learning (PerSpect-EL) models for PPI binding affinity prediction, for the first time. In our model, a sequence of Hodge (or combinatorial) Laplacian (HL) matrices at various different scales are generated from a specially designed filtration process. PerSpect attributes, which are statistical and combinatorial properties of spectrum information from these HL matrices, are used as features for PPI characterization. Each PerSpect attribute is input into a 1D convolutional neural network (CNN), and these CNN networks are stacked together in our PerSpect-based ensemble learning models. We systematically test our model on the two most commonly used datasets, i.e. SKEMPI and AB-Bind. It has been found that our model can achieve state-of-the-art results and outperform all existing models to the best of our knowledge.

Database resources of the national center for biotechnology information

The National Center for Biotechnology Information (NCBI) produces a variety of online information resources for biology, including the GenBank® nucleic acid sequence database and the PubMed® database of citations and abstracts published in life science journals. NCBI provides search and retrieval operations for most of these data from 35 distinct databases. The E-utilities serve as the programming interface for the most of these databases. Resources receiving significant updates in the past year include PubMed, PMC, Bookshelf, RefSeq, SRA, Virus, dbSNP, dbVar, ClinicalTrials.gov, MMDB, iCn3D and PubChem. These resources can be accessed through the NCBI home page at https://www.ncbi.nlm.nih.gov.

Applying Bioinformatic Platforms, In Vitro, and In Vivo Functional Assays in the Characterization of Genetic Variants in the GH/IGF Pathway Affecting Growth and Development

Heritability accounts for over 80% of adult human height, indicating that genetic variability is the main determinant of stature. The rapid technological development of Next-Generation Sequencing (NGS), particularly Whole Exome Sequencing (WES), has resulted in the characterization of several genetic conditions affecting growth and development. The greatest challenge of NGS remains the high number of candidate variants identified. In silico bioinformatic tools represent the first approach for classifying these variants. However, solving the complicated problem of variant interpretation requires the use of experimental approaches such as in vitro and, when needed, in vivo functional assays. In this review, we will discuss a rational approach to apply to the gene variants identified in children with growth and developmental defects including: (i) bioinformatic tools; (ii) in silico modeling tools; (iii) in vitro functional assays; and (iv) the development of in vivo models. While bioinformatic tools are useful for a preliminary selection of potentially pathogenic variants, in vitro—and sometimes also in vivo—functional assays are further required to unequivocally determine the pathogenicity of a novel genetic variant. This long, time-consuming, and expensive process is the only scientifically proven method to determine causality between a genetic variant and a human genetic disease.

Full-length de novo protein structure determination from cryo-EM maps using deep learning

MOTIVATION

Advances in microscopy instruments and image processing algorithms have led to an increasing number of cryo-EM maps. However, building accurate models for the EM maps at 3-5 Å resolution remains a challenging and time-consuming process. With the rapid growth of deposited EM maps, there is an increasing gap between the maps and reconstructed/modeled 3-dimensional (3D) structures. Therefore, automatic reconstruction of atomic-accuracy full-atomstructures fromEMmaps is pressingly needed.

RESULTS

We present a semi-automatic de novo structure determination method using a deep learningbased framework, named as DeepMM, which builds atomic-accuracy all-atom models from cryo-EM maps at near-atomic resolution. In our method, the main-chain and Cα positions as well as their amino acid and secondary structure types are predicted in the EM map using Densely Connected Convolutional Networks. DeepMM was extensively validated on 40 simulated maps at 5 Å resolution and 30 experimental maps at 2.6-4.8 Å resolution as well as an EMDB-wide data set of 2931 experimental maps at 2.6-4.9 Å resolution, and compared with state-of-the-art algorithms including RosettaES, MAINMAST, and Phenix. Overall, our DeepMM algorithm obtained a significant improvement over existing methods in terms of both accuracy and coverage in building full-length protein structures on all test sets, demonstrating the efficacy and general applicability of DeepMM.

AVAILABILITY

http://huanglab.phys.hust.edu.cn/DeepMM.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A Sweep of Earth's Virome Reveals Host-Guided Viral Protein Structural Mimicry and Points to Determinants of Human Disease

Viruses deploy genetically encoded strategies to coopt host machinery and support viral replicative cycles. Here, we use protein structure similarity to scan for molecular mimicry, manifested by structural similarity between viral and endogenous host proteins, across thousands of cataloged viruses and hosts spanning broad ecological niches and taxonomic range, including bacteria, plants and fungi, invertebrates, and vertebrates. This survey identified over 6,000,000 instances of structural mimicry; more than 70% of viral mimics cannot be discerned through protein sequence alone. We demonstrate that the manner and degree to which viruses exploit molecular mimicry varies by genome size and nucleic acid type and identify 158 human proteins that are mimicked by coronaviruses, providing clues about cellular processes driving pathogenesis. Our observations point to molecular mimicry as a pervasive strategy employed by viruses and indicate that the protein structure space used by a given virus is dictated by the host proteome. A record of this paper's transparent peer review process is included in the Supplemental Information.

A twist in the road less traveled: The AMBER ff15ipq-m force field for protein mimetics

We present a new force field, AMBER ff15ipq-m, for simulations of protein mimetics in applications from therapeutics to biomaterials. This force field is an expansion of the AMBER ff15ipq force field that was developed for canonical proteins and enables the modeling of four classes of artificial backbone units that are commonly used alongside natural α residues in blended or "heterogeneous" backbones: chirality-reversed D-α-residues, the Cα-methylated α-residue Aib, homologated β-residues (β3) bearing proteinogenic side chains, and two cyclic β residues (βcyc; APC and ACPC). The ff15ipq-m force field includes 472 unique atomic charges and 148 unique torsion terms. Consistent with the AMBER IPolQ lineage of force fields, the charges were derived using the Implicitly Polarized Charge (IPolQ) scheme in the presence of explicit solvent. To our knowledge, no general force field reported to date models the combination of artificial building blocks examined here. In addition, we have derived Karplus coefficients for the calculation of backbone amide J-coupling constants for β3Ala and ACPC β residues. The AMBER ff15ipq-m force field reproduces experimentally observed J-coupling constants in simple tetrapeptides and maintains the expected conformational propensities in reported structures of proteins/peptides containing the artificial building blocks of interest-all on the μs timescale. These encouraging results demonstrate the power and robustness of the IPolQ lineage of force fields in modeling the structure and dynamics of natural proteins as well as mimetics with protein-inspired artificial backbones in atomic detail.

Computational design of new enzymes for hydrolysis and synthesis of third-generation cephalosporin antibiotics

Engineering active sites in inert scaffolds to catalyze chemical transformations with unnatural substrates is still a great challenge for enzyme catalysis. In this research, a p-nitrobenzyl esterase from Bacillus subtilis was identified from the structural database, and a double mutant E115A/E188A was designed to afford catalytic activities toward the hydrolysis of ceftizoxime. A quadruple mutant E115A/E188A/L362S/I270A with enhanced catalytic efficiency was created to catalyze the condensation reaction of ethyl-2-methoxy-amino-2-(2-aminothiazole-4-yl) acetate with 7-amino-3-nor-cephalosporanic acid to produce ceftizoxime in a fully aqueous medium. The catalytic efficiencies of the computationally designed mutants E115A/E188A/L362S/I270A and E115A/Y118 K/E188 V/I270A/L362S can be taken as starting points to further improve their properties towards the practical application in designing more ecology-friendly production of third-generation cephalosporins.

Molecular selectivity design of mitogen-inducible gene-derived phosphopeptides between oncogenic HER kinases

Mitogen-inducible gene (MIG) is a natural negative regulator of the oncogenic HER kinase signaling by binding at the activation interface of kinase domain to disrupt the kinase dimerization. In this study, we systematically examine the binding structures, dynamics and energetics of MIG region 2 to four HER kinases based on their crystal or modeled complex structures, and identify an 8-mer phosphopeptide segment pYpY from the core strand sequence of MIG region 2 as the binding hotspot of MIG protein to HER kinases. We demonstrate that the small pYpY phosphopeptide can partially restore the binding affinity of full-length MIG protein, but exhibit a moderate selectivity over different HER kinases (S = 2.3-fold). In addition, the two phosphotyrosine residues pTyr394 and pTyr395 play an essential role in MIG-HER binding; dephosphorylation of them would fully eliminate the binding capability. A machine evolution algorithm is used to optimize the wild-type pYpY phosphopeptide, aiming to simultaneously improve affinity for these kinases and to maximize the affinity gap between different kinases. Consequently, a population is computationally evolved as selective phosphopeptide candidates; the dissociation constants of four representatives with HER kinases are systematically determined using binding affinity analysis, from which their selectivity is derived. The designed pYpYp3 phosphopeptide possesses a high selectivity over different HER kinases (S = 4.8-fold) and satisfactory affinity profile to these kinase (K_D = 140-1000 μM). Structural analysis observes that the global binding modes of pYpYp3 to different kinases are roughly consistent, but its local conformation may vary considerably, thus conferring specificity to the phosphopeptide.

A combined computational and experimental strategy identifies mutations conferring resistance to drugs targeting the BCR-ABL fusion protein

Drug resistance is of increasing concern, especially during the treatments of infectious diseases and cancer. To accelerate the drug discovery process in combating issues of drug resistance, here we developed a computational and experimental strategy to predict drug resistance mutations. Using BCR-ABL as a case study, we successfully recaptured the clinically observed mutations that confer resistance imatinib, nilotinib, dasatinib, bosutinib, and ponatinib. We then experimentally tested the predicted mutants in vitro. We found that although all mutants showed weakened binding strength as expected, the binding constants alone were not a good indicator of drug resistance. Instead, the half-maximal inhibitory concentration (IC50) was shown to be a good indicator of the incidence of the predicted mutations, together with change in catalytic efficacy. Our suggested strategy for predicting drug-resistance mutations includes the computational prediction and in vitro selection of mutants with increased IC50 values beyond the drug safety window.

Toward the Accuracy and Speed of Protein Side-Chain Packing: A Systematic Study on Rotamer Libraries

Protein rotamers refer to the conformational isomers taken by the side-chains of amino acids to accommodate specific structural folding environments. Since accurate modeling of atomic interactions is difficult, rotamer information collected from experimentally solved protein structures is often used to guide side-chain packing in protein folding and sequence design studies. Many rotamer libraries have been built in the literature but there is little quantitative guidance on which libraries should be chosen for different structural modeling studies. Here, we performed a comparative study of six widely used rotamer libraries and systematically examined their suitability for protein folding and sequence design in four aspects: (1) side-chain match accuracy, (2) side-chain conformation prediction, (3) de novo protein sequence design, and (4) computational time cost. We demonstrated that, compared to the backbone-dependent rotamer libraries (BBDRLs), the backbone-independent rotamer libraries (BBIRLs) generated conformations that more closely matched the native conformations due to the larger number of rotamers in the local rotamer search spaces. However, more practically, using an optimized physical energy function incorporated into a simulated annealing Monte Carlo searching scheme, we showed that utilization of the BBDRLs could result in higher accuracies in side-chain prediction and higher sequence recapitulation rates in protein design experiments. Detailed data analyses showed that the major advantage of BBDRLs lies in the energy term derived from the rotamer probabilities that are associated with the individual backbone torsion angle subspaces. This term is important for distinguishing between amino acid identities as well as the rotamer conformations of an amino acid. Meanwhile, the backbone torsion angle subspace-specific rotamer search drastically speeds up the searching time, despite the significantly larger number of total rotamers in the BBDRLs. These results should provide important guidance for the development and selection of rotamer libraries for practical protein design and structure prediction studies.

Computational reconstruction of atomistic protein structures from coarse-grained models

Three-dimensional protein structures, whether determined experimentally or theoretically, are often too low resolution. In this mini-review, we outline the computational methods for protein structure reconstruction from incomplete coarse-grained to all atomistic models. Typical reconstruction schemes can be divided into four major steps. Usually, the first step is reconstruction of the protein backbone chain starting from the C-alpha trace. This is followed by side-chains rebuilding based on protein backbone geometry. Subsequently, hydrogen atoms can be reconstructed. Finally, the resulting all-atom models may require structure optimization. Many methods are available to perform each of these tasks. We discuss the available tools and their potential applications in integrative modeling pipelines that can transfer coarse-grained information from computational predictions, or experiment, to all atomistic structures.

Computational design of enhanced detoxification activity of a zearalenone lactonase from Clonostachys rosea in acidic medium

Computational design of pH-activity profiles for enzymes is of great importance in industrial applications. In this research, a computational strategy was developed to engineer the pH-activity profile of a zearalenone lactonase (ZHD101) from Clonostachys rosea to promote its activity in acidic medium. The active site pK a values of ZHD101 were computationally designed by introducing positively charged lysine mutations on the enzyme surface, and the experimental results showed that two variants, M2(D157K) and M9(E171K), increased the catalytic efficiencies of ZHD101 modestly under acidic conditions. Moreover, two variants, M8(D133K) and M9(E171K), were shown to increase the turnover numbers by 2.73 and 2.06-fold with respect to wild type, respectively, though their apparent Michaelis constants were concomitantly increased. These results imply that the active site pK a value change might affect the pH-activity profile of the enzyme. Our computational strategy for pH-activity profile engineering considers protein stability; therefore, limited experimental validation is needed to discover beneficial mutations under shifted pH conditions.

A Structure-Informed Atlas of Human-Virus Interactions

While knowledge of protein-protein interactions (PPIs) is critical for understanding virus-host relationships, limitations on the scalability of high-throughput methods have hampered their identification beyond a number of well-studied viruses. Here, we implement an in silico computational framework (pathogen host interactome prediction using structure similarity [P-HIPSTer]) that employs structural information to predict ∼282,000 pan viral-human PPIs with an experimental validation rate of ∼76%. In addition to rediscovering known biology, P-HIPSTer has yielded a series of new findings: the discovery of shared and unique machinery employed across human-infecting viruses, a likely role for ZIKV-ESR1 interactions in modulating viral replication, the identification of PPIs that discriminate between human papilloma viruses (HPVs) with high and low oncogenic potential, and a structure-enabled history of evolutionary selective pressure imposed on the human proteome. Further, P-HIPSTer enables discovery of previously unappreciated cellular circuits that act on human-infecting viruses and provides insight into experimentally intractable viruses.

OPUS-Rota2: An Improved Fast and Accurate Side-Chain Modeling Method

Side-chain modeling plays a critical role in protein structure prediction. However, in many current methods, balancing the speed and accuracy is still challenging. In this paper, on the basis of our previous work OPUS-Rota (Protein Sci. 2008, 17, 1576-1585), we introduce a new side-chain modeling method, OPUS-Rota2, which is tested on both a 65-protein test set (DB65) in the OPUS-Rota paper and a 379-protein test set (DB379) in the SCWRL4 paper. If the main chain is native, OPUS-Rota2 is more accurate than OPUS-Rota, SCWRL4, and OSCAR-star but slightly less accurate than OSCAR-o. Also, if the main chain is non-native, OPUS-Rota2 is more accurate than any other method. Moreover, OPUS-Rota2 is significantly faster than any other method, in particular, 2 orders of magnitude faster than OSCAR-o. Thus, the combination of higher accuracy and speed of OPUS-Rota2 in modeling side chains on both the native and non-native main chains makes OPUS-Rota2 a very useful tool in protein structure modeling.

RGS6 and RGS7 Discriminate between the Highly Similar Gαi and Gαo Proteins Using a Two-Tiered Specificity Strategy

RGS6 and RGS7 are regulators of G protein signaling (RGS) proteins that inactivate heterotrimeric (αβγ) G proteins and mediate diverse biological functions, such as cardiac and neuronal signaling. Uniquely, both RGS6 and RGS7 can discriminate between Gα_o and Gα_i1-two similar Gα subunits that belong to the same G_i sub-family. Here, we show that the isolated RGS domains of RGS6 and RGS7 are sufficient to achieve this specificity. We identified three specific RGS6/7 "disruptor residues" that can attenuate RGS interactions toward Gα subunits and demonstrated that their insertion into a representative high-activity RGS causes a significant, yet non-specific, reduction in activity. We further identified a unique "modulatory" residue that bypasses this negative effect, specifically toward Gα_o. Hence, the exquisite specificity of RGS6 and RGS7 toward closely related Gα subunits is achieved via a two-tier specificity system, whereby a Gα-specific modulatory motif overrides the inhibitory effect of non-specific disruptor residues. Our findings expand the understanding of the molecular toolkit used by the RGS family to achieve specific interactions with selected Gα subunits-emphasizing the functional importance of the RGS domain in determining the activity and selectivity of RGS R7 sub-family members toward particular Gα subunits.

Using molecular dynamics simulations to evaluate active designs of cephradine hydrolase by molecular mechanics/Poisson–Boltzmann surface area and molecular mechanics/generalized Born surface area methods

The poor predictive accuracy of current computational enzyme design methods has led to low success rates of producing highly active variants that target non-natural substrates. In this report, a quantitative assessment approach based on molecular dynamics (MD) simulations was developed to eliminate false-positive enzyme designs at the computational stage. Taking cephradine hydrolase as an example, the apparent Michaelis binding constant (K_m) and catalytic efficiency (k_cat/K_m) of designed variants were correlated with binding free energies and activation energy barriers, respectively, as calculated by molecular mechanics/Poisson–Boltzmann surface area (MM/PBSA) and molecular mechanics/generalized Born surface area (MM/GBSA) methods with explicit water considered based on general MD simulation protocols. The correlation results showed that both the MM/GBSA and MM/PBSA methods with a protein dielectric constant (ε_p = 4) could rank the variants well based on the predicted binding free energies between enzyme and the substrate. Furthermore, the activation energy barriers calculated by the MM/PBSA method with an ε_p = 24 correlated well with k_cat/K_m. Thus, false-positive variants obtained by the enzyme design program PRODA were eliminated prior to experimentation. Therefore, MD simulation-based quantitative assessment of designed variants greatly enhanced the predictive accuracy of computational enzyme design tools and should facilitate the construction of artificial enzymes with high catalytic activities toward non-natural substrates.

A quantitative assessment method for computational enzyme design was developed to rank the active designs of cephradine hydrolase based on molecular dynamics simulation.

Protein side-chain packing problem: is there still room for improvement?

The protein side-chain packing problem (PSCPP) is an important subproblem of both protein structure prediction and protein design. During the past two decades, a large number of methods have been proposed to tackle this problem. These methods consist of three main components: a rotamer library, a scoring function and a search strategy. The average overall accuracy level obtained by these methods is approximately 87%. Whether a better accuracy level could be achieved remains to be answered. To address this question, we calculated the maximum accuracy level attainable using a simple rotamer library, independently of the energy function or the search method. Using 2883 different structures from the Protein Data Bank, we compared this accuracy level with the accuracy level of five state-of-the-art methods. These comparisons indicated that, for buried residues in the protein, we are already close to the best possible accuracy results. In addition, for exposed residues, we found that a significant gap exists between the possible improvement and the maximum accuracy level achievable with current methods. After determining that an improvement is possible, the next step is to understand what limitations are preventing us from obtaining such an improvement. Previous works on protein structure prediction and protein design have shown that scoring function inaccuracies may represent the main obstacle to achieving better results for these problems. To show that the same is true for the PSCPP, we evaluated the quality of two scoring functions used by some state-of-the-art algorithms. Our results indicate that neither of these scoring functions can guide the search method correctly, thereby reinforcing the idea that efforts to solve the PSCPP must also focus on developing better scoring functions.

Interplay between negative and positive design elements in Gα helical domains of G proteins determines interaction specificity toward RGS2

Regulators of G protein signaling (RGS) proteins inactivate Gα subunits, thereby controlling G protein-coupled signaling networks. Among all RGS proteins, RGS2 is unique in interacting only with the Gα_q but not with the Gα_i subfamily. Previous studies suggested that this specificity is determined by the RGS domain and, in particular, by three RGS2-specific residues that lead to a unique mode of interaction with Gα_q This interaction was further proposed to act through contacts with the Gα GTPase domain. Here, we combined energy calculations and GTPase activity measurements to determine which Gα residues dictate specificity toward RGS2. We identified putative specificity-determining residues in the Gα helical domain, which among G proteins is found only in Gα subunits. Replacing these helical domain residues in Gα_i with their Gα_q counterparts resulted in a dramatic specificity switch toward RGS2. We further show that Gα-RGS2 specificity is set by Gα_i residues that perturb interactions with RGS2, and by Gα_q residues that enhance these interactions. These results show, for the first time, that the Gα helical domain is central to dictating specificity toward RGS2, suggesting that this domain plays a general role in governing Gα-RGS specificity. Our insights provide new options for manipulating RGS-G protein interactions in vivo, for better understanding of their 'wiring' into signaling networks, and for devising novel drugs targeting such interactions.

Fundamental Insights into Proton-Coupled Electron Transfer in Soybean Lipoxygenase from Quantum Mechanical/Molecular Mechanical Free Energy Simulations

The proton-coupled electron transfer (PCET) reaction catalyzed by soybean lipoxygenase has served as a prototype for understanding hydrogen tunneling in enzymes. Herein this PCET reaction is studied with mixed quantum mechanical/molecular mechanical (QM/MM) free energy simulations. The free energy surfaces are computed as functions of the proton donor-acceptor (C-O) distance and the proton coordinate, and the potential of mean force is computed as a function of the C-O distance, inherently including anharmonicity. The simulation results are used to calculate the kinetic isotope effects for the wild-type enzyme (WT) and the L546A/L754A double mutant (DM), which have been measured experimentally to be ∼80 and ∼700, respectively. The PCET reaction is found to be exoergic for WT and slightly endoergic for the DM, and the equilibrium C-O distance for the reactant is found to be ∼0.2 Å greater for the DM than for WT. The larger equilibrium distance for the DM, which is due mainly to less optimal substrate binding in the expanded binding cavity, is primarily responsible for its higher kinetic isotope effect. The calculated potentials of mean force are anharmonic and relatively soft at shorter C-O distances, allowing efficient thermal sampling of the shorter distances required for effective hydrogen tunneling. The primarily local electrostatic field at the transferring hydrogen is ∼100 MV/cm in the direction to facilitate proton transfer and increases dramatically as the C-O distance decreases. These simulations suggest that the overall protein environment is important for conformational sampling of active substrate configurations aligned for proton transfer, but the PCET reaction is influenced primarily by local electrostatic effects that facilitate conformational sampling of shorter proton donor-acceptor distances required for effective hydrogen tunneling.

Scoring of Side-Chain Packings: An Analysis of Weight Factors and Molecular Dynamics Structures

The protein side-chain packing problem (PSCPP) is a central task in computational protein design. The problem is usually modeled as a combinatorial optimization problem, which consists of searching for a set of rotamers, from a given rotamer library, that minimizes a scoring function (SF). The SF is a weighted sum of terms, that can be decomposed in physics-based and knowledge-based terms. Although there are many methods to obtain approximate solutions for this problem, all of them have similar performances and there has not been a significant improvement in recent years. Studies on protein structure prediction and protein design revealed the limitations of current SFs to achieve further improvements for these two problems. In the same line, a recent work reported a similar result for the PSCPP. In this work, we ask whether or not this negative result regarding further improvements in performance is due to (i) an incorrect weighting of the SFs terms or (ii) the constrained conformation resulting from the protein crystallization process. To analyze these questions, we (i) model the PSCPP as a bi-objective combinatorial optimization problem, optimizing, at the same time, the two most important terms of two SFs of state-of-the-art algorithms and (ii) performed a preprocessing relaxation of the crystal structure through molecular dynamics to simulate the protein in the solvent and evaluated the performance of these two state-of-the-art SFs under these conditions. Our results indicate that (i) no matter what combination of weight factors we use the current SFs will not lead to better performances and (ii) the evaluated SFs will not be able to improve performance on relaxed structures. Furthermore, the experiments revealed that the SFs and the methods are biased toward crystallized structures.

Reconstruction of Atomistic Structures from Coarse-Grained Models for Protein-DNA Complexes

While coarse-grained (CG) simulations have widely been used to accelerate structure sampling of large biomolecular complexes, they are unavoidably less accurate and thus the reconstruction of all-atom (AA) structures and the subsequent refinement is desirable. In this study we developed an efficient method to reconstruct AA structures from sampled CG protein-DNA complex models, which attempts to model the protein-DNA interface accurately. First we developed a method to reconstruct atomic details of DNA structures from a three-site per nucleotide CG model, which uses a DNA fragment library. Next, for the protein-DNA interface, we referred to the side chain orientations in the known structure of the target interface when available. The other parts are modeled by existing tools. We confirmed the accuracy of the protocol in various aspects including the structure deviation in the self-reproduction, the base pair reproducibility, atomic contacts at the protein-DNA interface, and feasibility of the posterior AA simulations.

Representability of algebraic topology for biomolecules in machine learning based scoring and virtual screening

This work introduces a number of algebraic topology approaches, including multi-component persistent homology, multi-level persistent homology, and electrostatic persistence for the representation, characterization, and description of small molecules and biomolecular complexes. In contrast to the conventional persistent homology, multi-component persistent homology retains critical chemical and biological information during the topological simplification of biomolecular geometric complexity. Multi-level persistent homology enables a tailored topological description of inter- and/or intra-molecular interactions of interest. Electrostatic persistence incorporates partial charge information into topological invariants. These topological methods are paired with Wasserstein distance to characterize similarities between molecules and are further integrated with a variety of machine learning algorithms, including k-nearest neighbors, ensemble of trees, and deep convolutional neural networks, to manifest their descriptive and predictive powers for protein-ligand binding analysis and virtual screening of small molecules. Extensive numerical experiments involving 4,414 protein-ligand complexes from the PDBBind database and 128,374 ligand-target and decoy-target pairs in the DUD database are performed to test respectively the scoring power and the discriminatory power of the proposed topological learning strategies. It is demonstrated that the present topological learning outperforms other existing methods in protein-ligand binding affinity prediction and ligand-decoy discrimination.

Quaternary contact in the initial interaction of CD4 with the HIV-1 envelope trimer

Binding of the gp120 envelope (Env) glycoprotein to the CD4 receptor is the first step in the HIV-1 infectious cycle. Although the CD4-binding site has been extensively characterized, the initial receptor interaction has been difficult to study because of major CD4-induced structural rearrangements. Here we used cryogenic electron microscopy (cryo-EM) to visualize the initial contact of CD4 with the HIV-1 Env trimer at 6.8-Å resolution. A single CD4 molecule is embraced by a quaternary HIV-1-Env surface formed by coalescence of the previously defined CD4-contact region with a second CD4-binding site (CD4-BS2) in the inner domain of a neighboring gp120 protomer. Disruption of CD4-BS2 destabilized CD4-trimer interaction and abrogated HIV-1 infectivity by preventing the acquisition of coreceptor-binding competence. A corresponding reduction in HIV-1 infectivity occurred after the mutation of CD4 residues that interact with CD4-BS2. Our results document the critical role of quaternary interactions in the initial HIV-Env-receptor contact, with implications for treatment and vaccine design.

13C ENDOR Spectroscopy of Lipoxygenase-Substrate Complexes Reveals the Structural Basis for C-H Activation by Tunneling

In enzymatic C-H activation by hydrogen tunneling, reduced barrier width is important for efficient hydrogen wave function overlap during catalysis. For native enzymes displaying nonadiabatic tunneling, the dominant reactive hydrogen donor-acceptor distance (DAD) is typically ca. 2.7 Å, considerably shorter than normal van der Waals distances. Without a ground state substrate-bound structure for the prototypical nonadiabatic tunneling system, soybean lipoxygenase (SLO), it has remained unclear whether the requisite close tunneling distance occurs through an unusual ground state active site arrangement or by thermally sampling conformational substates. Herein, we introduce Mn2+ as a spin-probe surrogate for the SLO Fe ion; X-ray diffraction shows Mn-SLO is structurally faithful to the native enzyme. 13C ENDOR then reveals the locations of 13C10 and reactive 13C11 of linoleic acid relative to the metal; 1H ENDOR and molecular dynamics simulations of the fully solvated SLO model using ENDOR-derived restraints give additional metrical information. The resulting three-dimensional representation of the SLO active site ground state contains a reactive (a) conformer with hydrogen DAD of ∼3.1 Å, approximately van der Waals contact, plus an inactive (b) conformer with even longer DAD, establishing that stochastic conformational sampling is required to achieve reactive tunneling geometries. Tunneling-impaired SLO variants show increased DADs and variations in substrate positioning and rigidity, confirming previous kinetic and theoretical predictions of such behavior. Overall, this investigation highlights the (i) predictive power of nonadiabatic quantum treatments of proton-coupled electron transfer in SLO and (ii) sensitivity of ENDOR probes to test, detect, and corroborate kinetically predicted trends in active site reactivity and to reveal unexpected features of active site architecture.

Computational design to improve catalytic activity of cephalosporin C acylase from Pseudomonas strain N176

Engineering enzymes with high catalytic activities using enzyme designin silicoand a limited number of experimental evaluations is the new trend for the discovery of highly efficient biocatalysts.

New Dynamic Rotamer Libraries: Data-Driven Analysis of Side-Chain Conformational Propensities

Most rotamer libraries are generated from subsets of the PDB and do not fully represent the conformational scope of protein side chains. Previous attempts to rectify this sparse coverage of conformational space have involved application of weighting and smoothing functions. We resolve these limitations by using physics-based molecular dynamics simulations to determine more accurate frequencies of rotameric states. This work forms part of our Dynameomics initiative and uses a set of 807 proteins selected to represent 97% of known autonomous protein folds, thereby eliminating the bias toward common topologies found within the PDB. Our Dynameomics derived rotamer libraries encompass 4.8 × 10(9) rotamers, sampled from at least 51,000 occurrences of each of 93,642 residues. Here, we provide a backbone-dependent rotamer library, based on secondary structure ϕ/ψ regions, and an update to our 2011 backbone-independent library that addresses the doubling of our dataset since its original publication.

Computational design of variants for cephalosporin C acylase from Pseudomonas strain N176 with improved stability and activity

In this report, redesigning cephalosporin C acylase from the Pseudomonas strain N176 revealed that the loss of stability owing to the introduced mutations at the active site can be recovered by repacking the nearby hydrophobic core regions. Starting from a quadruple mutant M31βF/H57βS/V68βA/H70βS, whose decrease in stability is largely owing to the mutation V68βA at the active site, we employed a computational enzyme design strategy that integrated design both at hydrophobic core regions for stability enhancement and at the active site for activity improvement. Single-point mutations L154βF, Y167βF, L180βF and their combinations L154βF/L180βF and L154βF/Y167βF/L180βF were found to display improved stability and activity. The two-point mutant L154βF/L180βF increased the protein melting temperature (T _m) by 11.7 °C and the catalytic efficiency V _max/K _m by 57 % compared with the values of the starting quadruple mutant. The catalytic efficiency of the resulting sixfold mutant M31βF/H57βS/V68βA/H70βS/L154βF/L180βF is recovered to become comparable to that of the triple mutant M31βF/H57βS/H70βS, but with a higher T _m. Further experiments showed that single-point mutations L154βF, L180βF, and their combination contribute no stability enhancement to the triple mutant M31βF/H57βS/H70βS. These results verify that the lost stability because of mutation V68βA at the active site was recovered by introducing mutations L154βF and L180βF at hydrophobic core regions. Importantly, mutation V68βA in the six-residue mutant provides more space to accommodate the bulky side chain of cephalosporin C, which could help in designing cephalosporin C acylase mutants with higher activities and the practical one-step enzymatic route to prepare 7-aminocephalosporanic acid at industrial-scale levels.

Side-Chain Conformational Preferences Govern Protein-Protein Interactions

Protein secondary structures serve as geometrically constrained scaffolds for the display of key interacting residues at protein interfaces. Given the critical role of secondary structures in protein folding and the dependence of folding propensities on backbone dihedrals, secondary structure is expected to influence the identity of residues that are important for complex formation. Counter to this expectation, we find that a narrow set of residues dominates the binding energy in protein-protein complexes independent of backbone conformation. This finding suggests that the binding epitope may instead be substantially influenced by the side-chain conformations adopted. We analyzed side-chain conformational preferences in residues that contribute significantly to binding. This analysis suggests that preferred rotamers contribute directly to specificity in protein complex formation and provides guidelines for peptidomimetic inhibitor design.

Use of an Improved Matching Algorithm to Select Scaffolds for Enzyme Design Based on a Complex Active Site Model

Active site preorganization helps native enzymes electrostatically stabilize the transition state better than the ground state for their primary substrates and achieve significant rate enhancement. In this report, we hypothesize that a complex active site model for active site preorganization modeling should help to create preorganized active site design and afford higher starting activities towards target reactions. Our matching algorithm ProdaMatch was improved by invoking effective pruning strategies and the native active sites for ten scaffolds in a benchmark test set were reproduced. The root-mean squared deviations between the matched transition states and those in the crystal structures were < 1.0 Å for the ten scaffolds, and the repacking calculation results showed that 91% of the hydrogen bonds within the active sites are recovered, indicating that the active sites can be preorganized based on the predicted positions of transition states. The application of the complex active site model for de novo enzyme design was evaluated by scaffold selection using a classic catalytic triad motif for the hydrolysis of p-nitrophenyl acetate. Eighty scaffolds were identified from a scaffold library with 1,491 proteins and four scaffolds were native esterase. Furthermore, enzyme design for complicated substrates was investigated for the hydrolysis of cephalexin using scaffold selection based on two different catalytic motifs. Only three scaffolds were identified from the scaffold library by virtue of the classic catalytic triad-based motif. In contrast, 40 scaffolds were identified using a more flexible, but still preorganized catalytic motif, where one scaffold corresponded to the α-amino acid ester hydrolase that catalyzes the hydrolysis and synthesis of cephalexin. Thus, the complex active site modeling approach for de novo enzyme design with the aid of the improved ProdaMatch program is a promising approach for the creation of active sites with high catalytic efficiencies towards target reactions.

Structural Basis of Diverse Homophilic Recognition by Clustered α- and β-Protocadherins

Clustered protocadherin proteins (α-, β-, and γ-Pcdhs) provide a high level of cell-surface diversity to individual vertebrate neurons, engaging in highly specific homophilic interactions to mediate important roles in mammalian neural circuit development. How Pcdhs bind homophilically through their extracellular cadherin (EC) domains among dozens of highly similar isoforms has not been determined. Here, we report crystal structures for extracellular regions from four mouse Pcdh isoforms (α4, α7, β6, and β8), revealing a canonical head-to-tail interaction mode for homophilic trans dimers comprising primary intermolecular EC1:EC4 and EC2:EC3 interactions. A subset of trans interface residues exhibit isoform-specific conservation, suggesting roles in recognition specificity. Mutation of these residues, along with trans-interacting partner residues, altered the specificities of Pcdh interactions. Together, these data show how sequence variation among Pcdh isoforms encodes their diverse strict homophilic recognition specificities, which are required for their key roles in neural circuit assembly.