ZDOCK predictions for the CAPRI challenge

Study of recombinant antibody fragments and PAI-1 complexes combining protein-protein docking and results from site-directed mutagenesis

Elevated plasma levels of plasminogen activator inhibitor-1 (PAI-1) have been correlated with cardiovascular diseases such as myocardial infarction and venous thrombosis. PAI-1 has also been shown to play an important role in tumor development, diabetes, and obesitas. Monoclonal antibodies MA-8H9D4 and MA-56A7C10, and their single-chain variable fragments (scFv), exhibit PAI-1-neutralizing properties. In this study, a rigid-body docking approach is used to predict the binding geometry of two distinct conformations of PAI-1 (active and latent) in complex with these antibody fragments. Resulting models were initially refined by using the dead-end elimination algorithm. Different filtering criteria based on the mutagenesis studies and structural considerations were applied to select the final models. These were refined by using the slow-cooling torsion-angle dynamic annealing protocol. The docked structures reveal the respective epitopes and paratopes and their potential interactions. This study provides crucial information that is necessary for the rational development of low-molecular weight PAI-1 inhibitors.

On the computation of molecular surface correlations for protein docking using fourier techniques

The computation of surface correlations using a variety of molecular models has been applied to the unbound protein docking problem. Because of the computational complexity involved in examining all possible molecular orientations, the fast Fourier transform (FFT) (a fast numerical implementation of the discrete Fourier transform (DFT)) is generally applied to minimize the number of calculations. This approach is rooted in the convolution theorem which allows one to inverse transform the product of two DFTs in order to perform the correlation calculation. However, such a DFT calculation results in a cyclic or "circular" correlation which, in general, does not lead to the same result as the linear correlation desired for the docking problem. In this work, we provide computational bounds for constructing molecular models used in the molecular surface correlation problem. The derived bounds are then shown to be consistent with various intuitive guidelines previously reported in the protein docking literature. Finally, these bounds are applied to different molecular models in order to investigate their effect on the correlation calculation.

Integrating statistical pair potentials into protein complex prediction

The biophysical study of protein-protein interactions and docking has important implications in our understanding of most complex cellular signaling processes. Most computational approaches to protein docking involve a tradeoff between the level of detail incorporated into the model and computational power required to properly handle that level of detail. In this work, we seek to optimize that balance by showing that we can reduce the complexity of model representation and thus make the computation tractable with minimal loss of predictive performance. We also introduce a pair-wise statistical potential suitable for docking that builds on previous work and show that this potential can be incorporated into our fast fourier transform-based docking algorithm ZDOCK. We use the Protein Docking Benchmark to illustrate the improved performance of this potential compared with less detailed other scoring functions. Furthermore, we show that the new potential performs well on antibody-antigen complexes, with most predictions clustering around the Complementarity Determining Regions of antibodies without any manual intervention.

ZRANK: reranking protein docking predictions with an optimized energy function

Protein-protein docking requires fast and effective methods to quickly discriminate correct from incorrect predictions generated by initial-stage docking. We have developed and tested a scoring function that utilizes detailed electrostatics, van der Waals, and desolvation to rescore initial-stage docking predictions. Weights for the scoring terms were optimized for a set of test cases, and this optimized function was then tested on an independent set of nonredundant cases. This program, named ZRANK, is shown to significantly improve the success rate over the initial ZDOCK rankings across a large benchmark. The amount of test cases with No. 1 ranked hits increased from 2 to 11 and from 6 to 12 when predictions from two ZDOCK versions were considered. ZRANK can be applied either as a refinement protocol in itself or as a preprocessing stage to enrich the well-ranked hits prior to further refinement.

Interaction of malaria parasite-inhibitory antibodies with the merozoite surface protein MSP1(19) by computational docking

Merozoite surface protein 1 (MSP1) of the malaria parasite Plasmodium falciparum is an important vaccine candidate antigen. Antibodies specific for the C-terminal maturation product, MSP1(19), have been shown to inhibit erythrocyte invasion and parasite growth. Specific monoclonal antibodies react with conformational epitopes contained within the two EGF-like domains that constitute the antigen MSP1(19). To gain greater insight into the inhibitory process, the authors selected two strongly inhibitory antibodies (designated 12.8 and 12.10) and modeled their structures by homology. Computational docking was used to generate antigen-antibody complexes and a selection filter based on NMR data was applied to obtain plausible models. Molecular Dynamics simulations of the selected complexes were performed to evaluate the role of specific side chains in the binding. Favorable complexes were obtained that complement the NMR data in defining specific binding sites. These models can provide valuable guidelines for future experimental work that is devoted to the understanding of the action mechanism of invasion-inhibitory antibodies.

ZDOCK and RDOCK performance in CAPRI rounds 3, 4, and 5

We present an evaluation of the results of our ZDOCK and RDOCK algorithms in Rounds 3, 4, and 5 of the protein docking challenge CAPRI. ZDOCK is a Fast Fourier Transform (FFT)-based, initial-stage rigid-body docking algorithm, and RDOCK is an energy minimization algorithm for refining and reranking ZDOCK results. Of the 9 targets for which we submitted predictions, we attained at least acceptable accuracy for 7, at least medium accuracy for 6, and high accuracy for 3. These results are evidence that ZDOCK in combination with RDOCK is capable of making accurate predictions on a diverse set of protein complexes.

IPRO: an iterative computational protein library redesign and optimization procedure

A number of computational approaches have been developed to reengineer promising chimeric proteins one at a time through targeted point mutations. In this article, we introduce the computational procedure IPRO (iterative protein redesign and optimization procedure) for the redesign of an entire combinatorial protein library in one step using energy-based scoring functions. IPRO relies on identifying mutations in the parental sequences, which when propagated downstream in the combinatorial library, improve the average quality of the library (e.g., stability, binding affinity, specific activity, etc.). Residue and rotamer design choices are driven by a globally convergent mixed-integer linear programming formulation. Unlike many of the available computational approaches, the procedure allows for backbone movement as well as redocking of the associated ligands after a prespecified number of design iterations. IPRO can also be used, as a limiting case, for the redesign of a single or handful of individual sequences. The application of IPRO is highlighted through the redesign of a 16-member library of Escherichia coli/Bacillus subtilis dihydrofolate reductase hybrids, both individually and through upstream parental sequence redesign, for improving the average binding energy. Computational results demonstrate that it is indeed feasible to improve the overall library quality as exemplified by binding energy scores through targeted mutations in the parental sequences.

Probing Fragment Complementation by Rigid-Body Docking: in Silico Reconstitution of Calbindin D9k

Fragment complementation is gaining an increasing impact as a nonperturbing method to probe noncovalent interactions within protein supersecondary structures. In this study, the fast Fourier transform rigid-body docking algorithm ZDOCK has been employed for in silico reconstitution of the calcium binding protein calbindin D9k, from its two EF-hands subdomains, namely, EF1 (residues 1-43) and EF2 (residues 44-75). The EF1 fragment has been used both in its wild type and in nine mutant forms, in line with in vitro experiments. Consistent with in vitro data, ZDOCK reconstituted the proper fold of wild-type and mutated calbindin, locating the nativelike structures (i.e., holding a root-mean-square deviation < 1 A with respect to the X-ray structure) among the first 10 top-scored solutions out of 4000. Moreover, the three independent in silico reconstitutions of wild-type calbindin ranked a nativelike structure at the top of the output list, that is, the best scored one. The algorithm has been also successfully challenged in reconstituting the EF2 homodimer from two identical copies of the monomer. Furthermore, quantitative models consisting of linear correlations between thermodynamic data and ZDOCK scores were built, providing a tested tool for very fast in silico predictions of the free energy of association of protein-protein complexes solved at the atomic level and known to not undergo significant conformational changes upon binding.

A molecular docking model of SARS-CoV S1 protein in complex with its receptor, human ACE2

The exact residues within severe acute respiratory syndrome coronavirus (SARS-CoV) S1 protein and its receptor, human ACE2, involved in their interaction still remain largely undetermined. Identification of exact amino acid residues that are crucial for the interaction of S1 with ACE2 could provide working hypotheses for experimental studies and might be helpful for the development of antiviral inhibitor. In this paper, a molecular docking model of SARS-CoV S1 protein in complex with human ACE2 was constructed. The interacting residue pairs within this complex model and their contact types were also identified. Our model, supported by significant biochemical evidence, suggested receptor-binding residues were concentrated in two segments of S1 protein. In contrast, the interfacial residues in ACE2, though close to each other in tertiary structure, were found to be widely scattered in the primary sequence. In particular, the S1 residue ARG453 and ACE2 residue LYS341 might be the key residues in the complex formation.

An avidin-like domain that does not bind biotin is adopted for oligomerization by the extracellular mosaic protein fibropellin

The protein avidin found in egg white seems optimized for binding the small vitamin biotin as a stable homotetramer. Indeed, along with its streptavidin ortholog in the bacterium Streptomyces avidinii, this protein shows the strongest known noncovalent bond of a protein with a small ligand. A third known member of the avidin family, as similar to avidin as is streptavidin, is found at the C-terminal ends of the multidomain fibropellin proteins found in sea urchin. The fibropellins form a layer known as the apical lamina that surrounds the sea urchin embryo throughout development. Based upon the structure of avidin, we deduced a structural model for the avidin-like domain of the fibropellins and found that computational modeling predicts a lack of biotin binding and the preservation of tetramerization. To test this prediction we expressed and purified the fibropellin avidin-like domain and found it indeed to be a homotetramer incapable of binding biotin. Several lines of evidence suggest that the avidin-like domain causes the entire fibropellin protein to tetramerize. We suggest that the presence of the avidin-like domain serves a structural (tetrameric form) rather than functional (biotin-binding) role and may therefore be a molecular instance of exaptation-the modification of an existing function toward a new function. Finally, based upon the oligomerization of the avidin-like domain, we propose a model for the overall structure of the apical lamina.

Biologically enhanced sampling geometric docking and backbone flexibility treatment with multiconformational superposition

An efficient biologically enhanced sampling geometric docking method is presented based on the FTDock algorithm to predict the protein-protein binding modes. The active site data from different sources, such as biochemical and biophysical experiments or theoretical analyses of sequence data, can be incorporated in the rotation-translation scan. When discretizing a protein onto a 3-dimensional (3D) grid, a zero value is given to grid points outside a sphere centered on the geometric center of specified residues. In this way, docking solutions are biased toward modes where the interface region is inside the sphere. We also adopt a multiconformational superposition scheme to represent backbone flexibility in the proteins. When these procedures were applied to the targets of CAPRI, a larger number of hits and smaller ligand root-mean-square deviations (RMSDs) were obtained at the conformational search stage in all cases, and especially Target 19. With Target 18, only 1 near-native structure was retained by the biologically enhanced sampling geometric docking method, but this number increased to 53 and the least ligand RMSD decreased from 8.1 A to 2.9 A after performing multiconformational superposition. These results were obtained after the CAPRI prediction deadlines.

Phospho-Pivot Modeling Predicts Specific Interactions of Protein Phosphatase-1 with a Phospho-Inhibitor Protein CPI-17

Phospho-amino acids in proteins are directly associated with phospho-receptor proteins, including protein phosphatases. Here we produced and tested a scheme for docking together interacting phospho-proteins whose monomeric 3D structures were known. The phosphate of calyculin A, an inhibitor for protein phosphatase-1 and 2A (PP1 and PP2A), or phospho-CPI-17, a PP1-specific inhibitor protein, was docked at the active site of PP1. First, a library of 186,624 virtual complexes was generated in silico, by pivoting the phospho-ligand at the phosphorus atom by step every 5 degrees on three rotational axes. These models were then graded for probability according to atomic proximity between two molecules. The predicted structure of PP1 x calyculin A complex fitted to the crystal structure with r.m.s.d. of 0.23 A, providing a validate test of the modeling method. Modeling of PP1 x phospho-CPI-17 complex yielded one converged structure. The segment of CPI-17 around phospho-Thr38 is predicted to fit in the active site of PP1. Positive charges at Arg33/36 of CPI-17 are in close proximity to Glu274 of PP1, where the sequence is unique among Ser/Thr phosphatases. Single mutations of these residues in PP1 reduced the affinity against phospho-CPI-17. Thus, the interface of the PP1 x CPI-17 complex predicted by the phospho-pivot modeling accounts for the specificity of CPI-17 against PP1.

Computational approaches to protein-protein interaction

The interactions between proteins allow the cell's life. A number of experimental, genome-wide, high-throughput studies have been devoted to the determination of protein-protein interactions and the consequent interaction networks. Here, the bioinformatics methods dealing with protein-protein interactions and interaction network are overviewed. 1. Interaction databases developed to collect and annotate this immense amount of data; 2. Automated data mining techniques developed to extract information about interactions from the published literature; 3. Computational methods to assess the experimental results developed as a consequence of the finding that the results of high-throughput methods are rather inaccurate; 4. Exploitation of the information provided by protein interaction networks in order to predict functional features of the proteins; and 5. Prediction of protein-protein interactions.

Simulation of the interaction between ScyTx and small conductance calcium-activated potassium channel by docking and MM-PBSA

Computational methods are employed to simulate interaction of scorpion toxin ScyTx in complex with the small conductance calcium-activated potassium channel rsk2. All of available 25 structures of ScyTx in the Protein Data Bank determined by NMR were considered for improving performance of rigid protein docking of ZDOCK. Four main binding modes were found among a large number of predicted complexes by using clustering analysis, screening with expert knowledge, energy minimization, and molecular dynamics simulations. The quality and validity of the resulting complexes were further evaluated by molecular dynamics simulations with the generalized Born solvation model and by calculation of relative binding free energies with the molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) in the AMBER 7 suit of programs. The complex formed by the 22nd structure of the ScyTx and rsk2 channel was identified as the most favorable complex by using a combination of computational methods, which contain further introduction of flexibility without restraining residue side chain. From the resulted spatial structure of the ScyTx and rsk2 channel, ScyTx associates the mouth of the rsk2 channel with alpha-helix rather than beta-sheet. Structural analysis first revealed that Arg(13) played a novel and vital role of blocking the pore of the rsk2 channel, whose role is remarkably different from that of highly homologous scorpion toxin P05. Between the interfaces in the ScyTx-rsk2 complex, strong electrostatic interaction and hydrogen bonds exist between Arg(13) of ScyTx and Gly-Tyr-Gly-Asp sequential residues located in the four symmetrical chains of the pore region. Simultaneously, five hydrogen bonds between Arg(6) of ScyTx and Asp(341)(C), Val(366)(C), and Pro(367)(C), and electrostatic interaction between Arg(6) of ScyTx and Asp(364)(B) and Asp(341)(C) are also found by structural analysis. In addition, His(31) located at the C-terminal of ScyTx is surrounded by Val(342)(A), Asp(364)(A), Met(365)(A), Pro(367)(B), and Asn(366)(B) within a contact distance of 4.0 A. These simulation results are in good agreement with experimental data and can effectively explain the binding phenomena between ScyTx and the potassium channel at the level of molecular spatial structure. The consistency between results of molecular modeling and experimental data strongly suggests that our spatial structure model of the ScyTx-rsk2 complex is reasonable. Therefore, molecular docking combined with molecular dynamics simulations followed by molecular mechanics Poisson-Boltzmann surface area analysis is an attractive approach for modeling scorpion toxin-potassium channel complexes a priori for further biological studies.

M-ZDOCK: a grid-based approach for Cn symmetric multimer docking

Computational protein docking is a useful technique for gaining insights into protein interactions. We have developed an algorithm M-ZDOCK for predicting the structure of cyclically symmetric (Cn) multimers based on the structure of an unbound (or partially bound) monomer. Using a grid-based Fast Fourier Transform approach, a space of exclusively symmetric multimers is searched for the best structure. This leads to improvements both in accuracy and running time over the alternative, which is to run a binary docking program ZDOCK and filter the results for near-symmetry. The accuracy is improved because fewer false positives are considered in the search, thus hits are not as easily overlooked. By searching four instead of six degrees of freedom, the required amount of computation is reduced. This program has been tested on several known multimer complexes from the Protein DataBank, including four unbound multimers: three trimers and a pentamer. For all of these cases, M-ZDOCK was able to find at least one hit, whereas only two of the four testcases had hits when using ZDOCK and a symmetry filter. In addition, the running times are 30-40% faster for M-ZDOCK.

AVAILABILITY

M-ZDOCK is freely available to academic users at http://zlab.bu.edu/m-zdock/

CONTACT

zhiping@bu.edu

SUPPLEMENTARY INFORMATION

http://zlab.bu.edu/m-zdock.

Identification of the protein kinase A regulatory RIalpha-catalytic subunit interface by amide H/2H exchange and protein docking

An important goal after structural genomics is to build up the structures of higher-order protein-protein complexes from structures of the individual subunits. Often structures of higher order complexes are difficult to obtain by crystallography. We have used an alternative approach in which the structures of the individual catalytic (C) subunit and RIalpha regulatory (R) subunit of PKA were first subjected to computational docking, and the top 100,000 solutions were subsequently filtered based on amide hydrogen/deuterium (H/2H) exchange interface protection data. The resulting set of filtered solutions forms an ensemble of structures in which, besides the inhibitor peptide binding site, a flat interface between the C-terminal lobe of the C-subunit and the A- and B-helices of RIalpha is uniquely identified. This holoenzyme structure satisfies all previous experimental data on the complex and allows prediction of new contacts between the two subunits.