Breaking tolerance with engineered class I antigen-presenting molecules.

Successful efforts to activate T cells capable of recognizing weak cancer-associated self-antigens have employed altered peptide antigens to activate T cell responses capable of cross-reacting on native tumor-associated self. A limitation of this approach is the requirement for detailed knowledge about the altered self-peptide ligands used in these vaccines. In the current study we considered allorecognition as an approach for activating CTL capable of recognizing weak or self-antigens in the context of self-MHC. Nonself antigen-presenting molecules typically contain polymorphisms that influence interactions with the bound peptide and TCR interface. Recognition of these nonself structures results in peptide-dependent alloimmunity. Alloreactive T cells target their inducing alloantigens as well as third-party alloantigens but generally fail to target self-antigens. Certain residues located on the alpha-1/2 domains of class I antigen-presenting molecules primarily interface with TCR. These residues are more conserved within and across species than are residues that determine peptide antigen binding properties. Class I variants designed with amino acid substitutions at key positions within the conserved helical structures are shown to provide strong activating signals to alloreactive CD8 T cells while avoiding changes in naturally bound peptide ligands. Importantly, CTL activated in this manner can break self-tolerance by reacting to self-peptides presented by native MHC. The ability to activate self-tolerant T cells capable of cross-reacting on self-peptide-MHC in vivo represents an approach for inducing autoimmunity, with possible application in cancer vaccines.

Silent Tyrosinemia Type I Without Elevated Tyrosine or Succinylacetone Associated with Liver Cirrhosis and Hepatocellular Carcinoma.

Tyrosinemia type I (TYRSN1, TYR I) is caused by fumarylacetoacetate hydrolase (FAH) deficiency and affects approximately one in 100,000 individuals worldwide. Pathogenic variants in FAH cause TYRSN1, which induces cirrhosis and can progress to hepatocellular carcinoma (HCC). TYRSN1 is characterized by the production of a pathognomonic metabolite, succinylacetone (SUAC) and is included in the Recommended Uniform Screening Panel for newborns. Treatment intervention is effective if initiated within the first month of life. Here, we describe a family with three affected children who developed HCC secondary to idiopathic hepatosplenomegaly and cirrhosis during infancy. Whole exome sequencing revealed a novel homozygous missense variant in FAH (Chr15(GRCh38):g.80162305A>G; NM_000137.2:c.424A > G; NP_000128.1:p.R142G). This novel variant involves the catalytic pocket of the enzyme, but does not result in increased SUAC or tyrosine, making the diagnosis of TYRSN1 problematic. Testing this novel variant using a rapid, in vivo somatic mouse model showed that this variant could not rescue FAH deficiency. In this case of atypical TYRSN1, we show how reliance on SUAC as a primary diagnostic test can be misleading in some patients with this disease. Augmentation of current screening for TYRSN1 with targeted sequencing of FAH is warranted in cases suggestive of the disorder.

A new formulation of protein evolutionary models that account for structural constraints.

Despite the importance of a thermodynamically stable structure with a conserved fold for protein function, almost all evolutionary models neglect site-site correlations that arise from physical interactions between neighboring amino acid sites. This is mainly due to the difficulty in formulating a computationally tractable model since rate matrices can no longer be used. Here, we introduce a general framework, based on factor graphs, for constructing probabilistic models of protein evolution with site interdependence. Conveniently, efficient approximate inference algorithms, such as Belief Propagation, can be used to calculate likelihoods for these models. We fit an amino acid substitution model of this type that accounts for both solvent accessibility and site-site correlations. Comparisons of the new model with rate matrix models and alternative structure-dependent models demonstrate that it better fits the sequence data. We also examine evolution within a family of homohexameric enzymes and find that site-site correlations between most contacting subunits contribute to a higher likelihood. In addition, we show that the new substitution model has a similar mathematical form to the one introduced in Rodrigue et al. (Rodrigue N, Lartillot N, Bryant D, Philippe H. 2005. Site interdependence attributed to tertiary structure in amino acid sequence evolution. Gene 347:207-217), although with different parameter interpretations and values. We also perform a statistical analysis of the effects of amino acids at neighboring sites on substitution probabilities and find a significant perturbation of most probabilities, further supporting the significant role of site-site interactions in protein evolution and motivating the development of new evolutionary models similar to the one described here. Finally, we discuss possible extensions and applications of the new substitution model.

Adenine nucleotide translocase is acetylated in vivo in human muscle: Modeling predicts a decreased ADP affinity and altered control of oxidative phosphorylation.

Proteomics techniques have revealed that lysine acetylation is abundant in mitochondrial proteins. This study was undertaken (1) to determine the relationship between mitochondrial protein acetylation and insulin sensitivity in human skeletal muscle, identifying key acetylated proteins, and (2) to use molecular modeling techniques to understand the functional consequences of acetylation of adenine nucleotide translocase 1 (ANT1), which we found to be abundantly acetylated. Eight lean and eight obese nondiabetic subjects had euglycemic clamps and muscle biopsies for isolation of mitochondrial proteins and proteomics analysis. A number of acetylated mitochondrial proteins were identified in muscle biopsies. Overall, acetylation of mitochondrial proteins was correlated with insulin action (r = 0.60; P < 0.05). Of the acetylated proteins, ANT1, which catalyzes ADP-ATP exchange across the inner mitochondrial membrane, was acetylated at lysines 10, 23, and 92. The extent of acetylation of lysine 23 decreased following exercise, depending on insulin sensitivity. Molecular dynamics modeling and ensemble docking simulations predicted the ADP binding site of ANT1 to be a pocket of positively charged residues, including lysine 23. Calculated ADP-ANT1 binding affinities were physiologically relevant and predicted substantial reductions in affinity upon acetylation of lysine 23. Insertion of these derived binding affinities as parameters into a complete mathematical description of ANT1 kinetics predicted marked reductions in adenine nucleotide flux resulting from acetylation of lysine 23. Therefore, acetylation of ANT1 could have dramatic physiological effects on ADP-ATP exchange. Dysregulation of acetylation of mitochondrial proteins such as ANT1 therefore could be related to changes in mitochondrial function that are associated with insulin resistance.

Glutamine residues in Q-loops of multidrug resistance protein MRP1 contribute to ATP binding via interaction with metal cofactor.

Structural analyses of bacterial ATP-binding-cassette transporters revealed that the glutamine residue in Q-loop plays roles in interacting with: 1) a metal cofactor to participate in ATP binding; 2) a putative catalytic water molecule to participate in ATP hydrolysis; 3) other residues to transmit the conformational changes between nucleotide-binding-domains and transmembrane-domains, in ATP-dependent solute transport. We have mutated the glutamines at 713 and 1375 to asparagine, methionine or leucine to determine the functional roles of these residues in Q-loops of MRP1. All these single mutants significantly decreased Mg·ATP binding and increased the K(m) (Mg·ATP) and V(max) values in Mg·ATP-dependent leukotriene-C4 transport. However, the V(max) values of the double mutants Q713N/Q1375N, Q713M/Q1375M and Q713L/Q1375L were lower than that of wtMRP1, implying that the double mutants cannot efficiently bind Mg·ATP. Interestingly, MRP1 has higher affinity for Mn·ATP than for Mg·ATP and the Mn·ATP-dependent leukotriene-C4 transport activities of Q713N/Q1375N and Q713M/Q1375M are significantly higher than that of wtMRP1. All these results suggest that: 1) the glutamine residues in Q-loops contribute to ATP-binding via interaction with a metal cofactor; 2) it is most unlikely that these glutamine residues would play crucial roles in ATP hydrolysis and in transmitting the conformational changes between nucleotide-binding-domains and transmembrane-domains.

Assessing the accuracy of SAPT(DFT) interaction energies by comparison with experimentally derived noble gas potentials and molecular crystal lattice energies.

The density functional version of symmetry-adapted perturbation theory, SAPT(DFT), is a computationally efficient method for calculating intermolecular interaction energies. We evaluate its accuracy by comparison with experimentally determined noble gas interaction potentials and sublimation enthalpies, most of which have not been previously calculated using this method. In order to compare the results with wavefunction methods, we also calculate these quantities using MP2 and, for noble gas dimers, using CCSD(T). For the crystal lattice energy calculations, we include corrections to the dispersion, electrostatic, and induction energies that account for the finite interaction distance cutoff and higher-order induction contributions. Overall, the energy values extrapolated to the complete basis set limit show that SAPT(DFT) achieves significantly better agreement with experiment than MP2.

Predicting the effects of amino acid replacements in peptide hormones on their binding affinities for class B GPCRs and application to the design of secretin receptor antagonists.

Computational prediction of the effects of residue changes on peptide-protein binding affinities, followed by experimental testing of the top predicted binders, is an efficient strategy for the rational structure-based design of peptide inhibitors. In this study we apply this approach to the discovery of competitive antagonists for the secretin receptor, the prototypical member of class B G protein-coupled receptors (GPCRs). Proteins in this family are involved in peptide hormone-stimulated signaling and are implicated in several human diseases, making them potential therapeutic targets. We first validated our computational method by predicting changes in the binding affinities of several peptides to their cognate class B GPCRs due to alanine replacement and compared the results with previously published experimental values. Overall, the results showed a significant correlation between the predicted and experimental ΔΔG values. Next, we identified candidate inhibitors by applying this method to a homology model of the secretin receptor bound to an N-terminal truncated secretin peptide. Predictions were made for single residue replacements to each of the other nineteen naturally occurring amino acids at peptide residues within the segment binding the receptor N-terminal domain. Amino acid replacements predicted to most enhance receptor binding were then experimentally tested by competition-binding assays. We found two residue changes that improved binding affinities by almost one log unit. Furthermore, a peptide combining both of these favorable modifications resulted in an almost two log unit improvement in binding affinity, demonstrating the approximately additive effect of these changes on binding. In order to further investigate possible physical effects of these residue changes on receptor binding affinity, molecular dynamics simulations were performed on representatives of the successful peptide analogues (namely A17I, G25R, and A17I/G25R) in bound and unbound forms. These simulations suggested that a combination of the α-helical propensity of the unbound peptide and specific interactions between the peptide and the receptor extracellular domain contribute to their higher binding affinities.

Importance of each residue within secretin for receptor binding and biological activity.

Secretin is a linear 27-residue peptide hormone that stimulates pancreatic and biliary ductular bicarbonate and water secretion by acting at its family B G protein-coupled receptor. While, like other family members, the carboxyl-terminal region of secretin is most important for high affinity binding and its amino-terminal region is most important for receptor selectivity and receptor activation, determinants for these activities are distributed throughout the entire length of this peptide. In this work, we have systematically investigated changing each residue within secretin to alanine and evaluating the impact on receptor binding and biological activity. The residues most critical for receptor binding were His1, Asp3, Gly4, Phe6, Thr7, Ser8, Leu10, Asp15, Leu19, and Leu23. The residues most critical for biological activity included His1, Gly4, Thr7, Ser8, Glu9, Leu10, Leu19, Leu22, and Leu23, with Asp3, Phe6, Ser11, Leu13, Asp15, Leu26, and Val27 also contributing. While the importance of residues in positions analogous to His1, Asp3, Phe6, Thr7, and Leu23 is conserved for several closely related members of this family, Leu19 is uniquely important for secretin. We, therefore, have further studied this residue by molecular modeling and molecular dynamics simulations. Indeed, the molecular dynamics simulations showed that mutation of Leu19 to alanine was destabilizing, with this effect greater than that observed for the analogous position in the other close family members. This could reflect reduced contact with the receptor or an increase in the solvent-accessible surface area of the hydrophobic residues in the carboxyl terminus of secretin as bound to its receptor.

Lactam constraints provide insights into the receptor-bound conformation of secretin and stabilize a receptor antagonist.

The natural ligands for family B G protein-coupled receptors are moderate-length linear peptides having diffuse pharmacophores. The amino-terminal regions of these ligands are critical for biological activity, with their amino-terminal truncation leading to production of orthosteric antagonists. The carboxyl-terminal regions of these peptides are thought to occupy a ligand-binding cleft within the disulfide-bonded amino-terminal domains of these receptors, with the peptides in amphipathic helical conformations. In this work, we have characterized the binding and activity of a series of 11 truncated and lactam-constrained secretin(5-27) analogues at the prototypic member of this family, the secretin receptor. One peptide in this series with lactam connecting residues 16 and 20 [c[E(16),K(20)][Y(10)]sec(5-27)] improved the binding affinity of its unconstrained parental peptide 22-fold while retaining the absence of endogenous biological activity and competitive antagonist characteristics. Homology modeling with molecular mechanics and molecular dynamics simulations established that this constrained peptide occupies the ligand-binding cleft in an orientation similar to that of natural full-length secretin and provided insights into why this peptide was more effective than other truncated conformationally constrained peptides in the series. This lactam bridge is believed to stabilize an extended α-helical conformation of this peptide while in solution and not to interfere with critical residue-residue approximations while docked to the receptor.

Efficient molecular mechanics simulations of the folding, orientation, and assembly of peptides in lipid bilayers using an implicit atomic solvation model.

Membrane proteins comprise a significant fraction of the proteomes of sequenced organisms and are the targets of approximately half of marketed drugs. However, in spite of their prevalence and biomedical importance, relatively few experimental structures are available due to technical challenges. Computational simulations can potentially address this deficit by providing structural models of membrane proteins. Solvation within the spatially heterogeneous membrane/solvent environment provides a major component of the energetics driving protein folding and association within the membrane. We have developed an implicit solvation model for membranes that is both computationally efficient and accurate enough to enable molecular mechanics predictions for the folding and association of peptides within the membrane. We derived the new atomic solvation model parameters using an unbiased fitting procedure to experimental data and have applied it to diverse problems in order to test its accuracy and to gain insight into membrane protein folding. First, we predicted the positions and orientations of peptides and complexes within the lipid bilayer and compared the simulation results with solid-state NMR structures. Additionally, we performed folding simulations for a series of host-guest peptides with varying propensities to form alpha helices in a hydrophobic environment and compared the structures with experimental measurements. We were also able to successfully predict the structures of amphipathic peptides as well as the structures for dimeric complexes of short hexapeptides that have experimentally characterized propensities to form beta sheets within the membrane. Finally, we compared calculated relative transfer energies with data from experiments measuring the effects of mutations on the free energies of translocon-mediated insertion of proteins into lipid bilayers and of combined folding and membrane insertion of a beta barrel protein.

Public Protests and the Risk of Novel Coronavirus Disease Hospitalizations: A County-Level Analysis from California.

The objective of this study was to assess the relationship between public protests and county-level, novel coronavirus disease (COVID-19) hospitalization rates across California. Publicly available data were included in the analysis from 55 of 58 California state counties (29 March-14 October 2020). Mixed-effects negative binomial regression models were used to examine the relationship between daily county-level COVID-19 hospitalizations and two main exposure variables: any vs. no protests and 1 or >1 protest vs. no protests on a given county-day. COVID-19 hospitalizations were used as a proxy for viral transmission since such rates are less sensitive to temporal changes in testing access/availability. Models included covariates for daily county mobility, county-level characteristics, and time trends. Models also included a county-population offset and a two-week lag for the association between exposure and outcome. No significant associations were observed between protest exposures and COVID-19 hospitalization rates among the 55 counties. We did not find evidence to suggest that public protests were associated with COVID-19 hospitalization within California counties. These findings support the notion that protesting during a pandemic may be safe, ostensibly, so long as evidence-based precautionary measures are taken.

Orientation-dependent backbone-only residue pair scoring functions for fixed backbone protein design.

BACKGROUND: Empirical scoring functions have proven useful in protein structure modeling. Most such scoring functions depend on protein side chain conformations. However, backbone-only scoring functions do not require computationally intensive structure optimization and so are well suited to protein design, which requires fast score evaluation. Furthermore, scoring functions that account for the distinctive relative position and orientation preferences of residue pairs are expected to be more accurate than those that depend only on the separation distance. RESULTS: Residue pair scoring functions for fixed backbone protein design were derived using only backbone geometry. Unlike previous studies that used spherical harmonics to fit 2D angular distributions, Gaussian Mixture Models were used to fit the full 3D (position only) and 6D (position and orientation) distributions of residue pairs. The performance of the 1D (residue separation only), 3D, and 6D scoring functions were compared by their ability to identify correct threading solutions for a non-redundant benchmark set of protein backbone structures. The threading accuracy was found to steadily increase with increasing dimension, with the 6D scoring function achieving the highest accuracy. Furthermore, the 3D and 6D scoring functions were shown to outperform side chain-dependent empirical potentials from three other studies. Next, two computational methods that take advantage of the speed and pairwise form of these new backbone-only scoring functions were investigated. The first is a procedure that exploits available sequence data by averaging scores over threading solutions for homologs. This was evaluated by applying it to the challenging problem of identifying interacting transmembrane alpha-helices and found to further improve prediction accuracy. The second is a protein design method for determining the optimal sequence for a backbone structure by applying Belief Propagation optimization using the 6D scoring functions. The sensitivity of this method to backbone structure perturbations was compared with that of fixed-backbone all-atom modeling by determining the similarities between optimal sequences for two different backbone structures within the same protein family. The results showed that the design method using 6D scoring functions was more robust to small variations in backbone structure than the all-atom design method. CONCLUSIONS: Backbone-only residue pair scoring functions that account for all six relative degrees of freedom are the most accurate and including the scores of homologs further improves the accuracy in threading applications. The 6D scoring function outperformed several side chain-dependent potentials while avoiding time-consuming and error prone side chain structure prediction. These scoring functions are particularly useful as an initial filter in protein design problems before applying all-atom modeling.

MultiRTA: a simple yet reliable method for predicting peptide binding affinities for multiple class II MHC allotypes.

BACKGROUND: The binding of peptide fragments of antigens to class II MHC is a crucial step in initiating a helper T cell immune response. The identification of such peptide epitopes has potential applications in vaccine design and in better understanding autoimmune diseases and allergies. However, comprehensive experimental determination of peptide-MHC binding affinities is infeasible due to MHC diversity and the large number of possible peptide sequences. Computational methods trained on the limited experimental binding data can address this challenge. We present the MultiRTA method, an extension of our previous single-type RTA prediction method, which allows the prediction of peptide binding affinities for multiple MHC allotypes not used to train the model. Thus predictions can be made for many MHC allotypes for which experimental binding data is unavailable. RESULTS: We fit MultiRTA models for both HLA-DR and HLA-DP using large experimental binding data sets. The performance in predicting binding affinities for novel MHC allotypes, not in the training set, was tested in two different ways. First, we performed leave-one-allele-out cross-validation, in which predictions are made for one allotype using a model fit to binding data for the remaining MHC allotypes. Comparison of the HLA-DR results with those of two other prediction methods applied to the same data sets showed that MultiRTA achieved performance comparable to NetMHCIIpan and better than the earlier TEPITOPE method. We also directly tested model transferability by making leave-one-allele-out predictions for additional experimentally characterized sets of overlapping peptide epitopes binding to multiple MHC allotypes. In addition, we determined the applicability of prediction methods like MultiRTA to other MHC allotypes by examining the degree of MHC variation accounted for in the training set. An examination of predictions for the promiscuous binding CLIP peptide revealed variations in binding affinity among alleles as well as potentially distinct binding registers for HLA-DR and HLA-DP. Finally, we analyzed the optimal MultiRTA parameters to discover the most important peptide residues for promiscuous and allele-specific binding to HLA-DR and HLA-DP allotypes. CONCLUSIONS: The MultiRTA method yields competitive performance but with a significantly simpler and physically interpretable model compared with previous prediction methods. A MultiRTA prediction webserver is available at http://bordnerlab.org/MultiRTA.

Prediction of the binding affinities of peptides to class II MHC using a regularized thermodynamic model.

BACKGROUND: The binding of peptide fragments of extracellular peptides to class II MHC is a crucial event in the adaptive immune response. Each MHC allotype generally binds a distinct subset of peptides and the enormous number of possible peptide epitopes prevents their complete experimental characterization. Computational methods can utilize the limited experimental data to predict the binding affinities of peptides to class II MHC. RESULTS: We have developed the Regularized Thermodynamic Average, or RTA, method for predicting the affinities of peptides binding to class II MHC. RTA accounts for all possible peptide binding conformations using a thermodynamic average and includes a parameter constraint for regularization to improve accuracy on novel data. RTA was shown to achieve higher accuracy, as measured by AUC, than SMM-align on the same data for all 17 MHC allotypes examined. RTA also gave the highest accuracy on all but three allotypes when compared with results from 9 different prediction methods applied to the same data. In addition, the method correctly predicted the peptide binding register of 17 out of 18 peptide-MHC complexes. Finally, we found that suboptimal peptide binding registers, which are often ignored in other prediction methods, made significant contributions of at least 50% of the total binding energy for approximately 20% of the peptides. CONCLUSIONS: The RTA method accurately predicts peptide binding affinities to class II MHC and accounts for multiple peptide binding registers while reducing overfitting through regularization. The method has potential applications in vaccine design and in understanding autoimmune disorders. A web server implementing the RTA prediction method is available at http://bordnerlab.org/RTA/.

Elucidation of the active conformation of the amino terminus of receptor-bound secretin using intramolecular disulfide bond constraints.

Family B G protein-coupled receptors include several potentially important drug targets, yet our understanding of the molecular basis of ligand binding to and activation of these receptors is incomplete. While NMR and crystal structures exist for peptide ligand-associated amino-terminal domains of several family members, these only provide insights into the conformation of the carboxyl-terminal region of the peptides. The amino-terminal region of these peptides, critical for biological activity, is believed to interact with the helical bundle domain, and is, therefore, unconstrained in these structures. The aim of the current study was to provide insights into the conformation of the amino terminus of secretin as bound to its receptor. We prepared a series of conformationally constrained secretin peptides containing intramolecular disulfide bonds that were predicted by molecular modeling to approximate the conformation of the analogous region of PACAP bound to its receptor that had been determined using transfer-NOE NMR techniques. Secretin peptides with pairs of cysteine residues in positions 2-7, 3-5, 3-6, 4-7, 7-9, and 4-10 were studied as linear and disulfide-bonded forms. The analog with a disulfide bond connecting positions 7-9 had binding affinity and biological activity similar to natural secretin, supporting the relevance of this constraint to its active conformation. While this feature is shared between secretin and PACAP, absence of activity in other constrained peptides in this series also suggest that there are differences between these receptor-bound conformations. It will be critical to extend similar studies to other family members to learn what structural elements might be most conserved in this family.

Predicting protein-protein binding sites in membrane proteins.

BACKGROUND: Many integral membrane proteins, like their non-membrane counterparts, form either transient or permanent multi-subunit complexes in order to carry out their biochemical function. Computational methods that provide structural details of these interactions are needed since, despite their importance, relatively few structures of membrane protein complexes are available. RESULTS: We present a method for predicting which residues are in protein-protein binding sites within the transmembrane regions of membrane proteins. The method uses a Random Forest classifier trained on residue type distributions and evolutionary conservation for individual surface residues, followed by spatial averaging of the residue scores. The prediction accuracy achieved for membrane proteins is comparable to that for non-membrane proteins. Also, like previous results for non-membrane proteins, the accuracy is significantly higher for residues distant from the binding site boundary. Furthermore, a predictor trained on non-membrane proteins was found to yield poor accuracy on membrane proteins, as expected from the different distribution of surface residue types between the two classes of proteins. Thus, although the same procedure can be used to predict binding sites in membrane and non-membrane proteins, separate predictors trained on each class of proteins are required. Finally, the contribution of each residue property to the overall prediction accuracy is analyzed and prediction examples are discussed. CONCLUSION: Given a membrane protein structure and a multiple alignment of related sequences, the presented method gives a prioritized list of which surface residues participate in intramembrane protein-protein interactions. The method has potential applications in guiding the experimental verification of membrane protein interactions, structure-based drug discovery, and also in constraining the search space for computational methods, such as protein docking or threading, that predict membrane protein complex structures.

Predicting small ligand binding sites in proteins using backbone structure.

MOTIVATION: Specific non-covalent binding of metal ions and ligands, such as nucleotides and cofactors, is essential for the function of many proteins. Computational methods are useful for predicting the location of such binding sites when experimental information is lacking. Methods that use structural information, when available, are particularly promising since they can potentially identify non-contiguous binding motifs that cannot be found using only the amino acid sequence. Furthermore, a prediction method that can utilize low-resolution models is advantageous because high-resolution structures are available for only a relatively small fraction of proteins. RESULTS: SitePredict is a machine learning-based method for predicting binding sites in protein structures for specific metal ions or small molecules. The method uses Random Forest classifiers trained on diverse residue-based site properties including spatial clustering of residue types and evolutionary conservation. SitePredict was tested by cross-validation on a set of known binding sites for six different metal ions and five different small molecules in a non-redundant set of protein-ligand complex structures. The prediction performance was good for all ligands considered, as reflected by AUC values of at least 0.8. Furthermore, a more realistic test on unbound structures showed only a slight decrease in the accuracy. The properties that contribute the most to the prediction accuracy of each ligand were also examined. Finally, examples of predicted binding sites in homology models and uncharacterized proteins are discussed. AVAILABILITY: Binding site prediction results for all PDB protein structures and human protein homology models are available at http://sitepredict.org/.

Comprehensive inventory of protein complexes in the Protein Data Bank from consistent classification of interfaces.

BACKGROUND: Protein-protein interactions are ubiquitous and essential for all cellular processes. High-resolution X-ray crystallographic structures of protein complexes can reveal the details of their function and provide a basis for many computational and experimental approaches. Differentiation between biological and non-biological contacts and reconstruction of the intact complex is a challenging computational problem. A successful solution can provide additional insights into the fundamental principles of biological recognition and reduce errors in many algorithms and databases utilizing interaction information extracted from the Protein Data Bank (PDB). RESULTS: We have developed a method for identifying protein complexes in the PDB X-ray structures by a four step procedure: (1) comprehensively collecting all protein-protein interfaces; (2) clustering similar protein-protein interfaces together; (3) estimating the probability that each cluster is relevant based on a diverse set of properties; and (4) combining these scores for each PDB entry in order to predict the complex structure. The resulting clusters of biologically relevant interfaces provide a reliable catalog of evolutionary conserved protein-protein interactions. These interfaces, as well as the predicted protein complexes, are available from the Protein Interface Server (PInS) website (see Availability and requirements section). CONCLUSION: Our method demonstrates an almost two-fold reduction of the annotation error rate as evaluated on a large benchmark set of complexes validated from the literature. We also estimate relative contributions of each interface property to the accurate discrimination of biologically relevant interfaces and discuss possible directions for further improving the prediction method.

Protein docking using surface matching and supervised machine learning.

Computational prediction of protein complex structures through docking offers a means to gain a mechanistic understanding of protein interactions that mediate biological processes. This is particularly important as the number of experimentally determined structures of isolated proteins exceeds the number of structures of complexes. A comprehensive docking procedure is described in which efficient sampling of conformations is achieved by matching surface normal vectors, fast filtering for shape complementarity, clustering by RMSD, and scoring the docked conformations using a supervised machine learning approach. Contacting residue pair frequencies, residue propensities, evolutionary conservation, and shape complementarity score for each docking conformation are used as input data to a Random Forest classifier. The performance of the Random Forest approach for selecting correctly docked conformations was assessed by cross-validation using a nonredundant benchmark set of X-ray structures for 93 heterodimer and 733 homodimer complexes. The single highest rank docking solution was the correct (near-native) structure for slightly more than one third of the complexes. Furthermore, the fraction of highly ranked correct structures was significantly higher than the overall fraction of correct structures, for almost all complexes. A detailed analysis of the difficult to predict complexes revealed that the majority of the homodimer cases were explained by incorrect oligomeric state annotation. Evolutionary conservation and shape complementarity score as well as both underrepresented and overrepresented residue types and residue pairs were found to make the largest contributions to the overall prediction accuracy. Finally, the method was also applied to docking unbound subunit structures from a previously published benchmark set.

Ab initio prediction of peptide-MHC binding geometry for diverse class I MHC allotypes.

Since determining the crystallographic structure of all peptide-MHC complexes is infeasible, an accurate prediction of the conformation is a critical computational problem. These models can be useful for determining binding energetics, predicting the structures of specific ternary complexes with T-cell receptors, and designing new molecules interacting with these complexes. The main difficulties are (1) adequate sampling of the large number of conformational degrees of freedom for the flexible peptide, (2) predicting subtle changes in the MHC interface geometry upon binding, and (3) building models for numerous MHC allotypes without known structures. Whereas previous studies have approached the sampling problem by dividing the conformational variables into different sets and predicting them separately, we have refined the Biased-Probability Monte Carlo docking protocol in internal coordinates to optimize a physical energy function for all peptide variables simultaneously. We also imitated the induced fit by docking into a more permissive smooth grid representation of the MHC followed by refinement and reranking using an all-atom MHC model. Our method was tested by a comparison of the results of cross-docking 14 peptides into HLA-A*0201 and 9 peptides into H-2K(b) as well as docking peptides into homology models for five different HLA allotypes with a comprehensive set of experimental structures. The surprisingly accurate prediction (0.75 A backbone RMSD) for cross-docking of a highly flexible decapeptide, dissimilar to the original bound peptide, as well as docking predictions using homology models for two allotypes with low average backbone RMSDs of less than 1.0 A illustrate the method's effectiveness. Finally, energy terms calculated using the predicted structures were combined with supervised learning on a large data set to classify peptides as either HLA-A*0201 binders or nonbinders. In contrast with sequence-based prediction methods, this model was also able to predict the binding affinity for peptides to a different MHC allotype (H-2K(b)), not used for training, with comparable prediction accuracy.