ClusPro: performance in CAPRI rounds 6-11 and the new server

ClusPro is the first fully automated, web-based program for docking protein structures. Users may upload the coordinate files of two protein structures through ClusPro's web interface, or enter the PDB codes of the respective structures. The server performs rigid body docking, energy screening, and clustering to produce models. The program output is a short list of putative complexes ranked according to their clustering properties. ClusPro has been participating in CAPRI since January 2003, submitting predictions within 24 h after a target becomes available. In Rounds 6-11, ClusPro generated acceptable submissions for Targets 22, 25, and 27. In general, acceptable models were obtained for the relatively easy targets without substantial conformational changes upon binding. We also describe the new version of ClusPro that incorporates our recently developed docking program PIPER. PIPER is based on the fast Fourier transform correlation approach, but the method is extended to use pairwise interaction potentials, thereby increasing the number of near-native docked structures.     

ClusPro in rounds 38 to 45 of CAPRI: Toward combining template-based methods with free docking

Targets in the protein docking experiment CAPRI (Critical Assessment of Predicted Interactions) generally present new challenges and contribute to new developments in methodology. In rounds 38 to 45 of CAPRI, most targets could be effectively predicted using template-based methods. However, the server ClusPro required structures rather than sequences as input, and hence we had to generate and dock homology models. The available templates also provided distance restraints that were directly used as input to the server. We show here that such an approach has some advantages. Free docking with template-based restraints using ClusPro reproduced some interfaces suggested by weak or ambiguous templates while not reproducing others, resulting in correct server predicted models. More recently we developed the fully automated ClusPro TBM server that performs template-based modeling and thus can use sequences rather than structures of component proteins as input. The performance of the server, freely available for noncommercial use at https://tbm.cluspro.org , is demonstrated by predicting the protein-protein targets of rounds 38 to 45 of CAPRI.     

Challenges and opportunities of automated protein-protein docking: HDOCK server vs human predictions in CAPRI Rounds 38-46

Protein-protein docking plays an important role in the computational prediction of the complex structure between two proteins. For years, a variety of docking algorithms have been developed, as witnessed by the critical assessment of prediction interactions (CAPRI) experiments. However, despite their successes, many docking algorithms often require a series of manual operations like modeling structures from sequences, incorporating biological information, and selecting final models. The difficulties in these manual steps have significantly limited the applications of protein-protein docking, as most of the users in the community are nonexperts in docking. Therefore, automated docking like a web server, which can give a comparable performance to human docking protocol, is pressingly needed. As such, we have participated in the blind CAPRI experiments for Rounds 38-45 and CASP13-CAPRI challenge for Round 46 with both our HDOCK automated docking web server and human docking protocol. It was shown that our HDOCK server achieved an “acceptable” or higher CAPRI-rated model in the top 10 submitted predictions for 65.5% and 59.1% of the targets in the docking experiments of CAPRI and CASP13-CAPRI, respectively, which are comparable to 66.7% and 54.5% for human docking protocol. Similar trends can also be observed in the scoring experiments. These results validated our HDOCK server as an efficient automated docking protocol for nonexpert users. Challenges and opportunities of automated docking are also discussed.     

Evaluation of AlphaFold2 structures as docking targets

AlphaFold2 is a promising new tool for researchers to predict protein structures and generate high-quality models, with low backbone and global root-mean-square deviation (RMSD) when compared with experimental structures. However, it is unclear if the structures predicted by AlphaFold2 will be valuable targets of docking. To address this question, we redocked ligands in the PDBbind datasets against the experimental co-crystallized receptor structures and against the AlphaFold2 structures using AutoDock-GPU. We find that the quality measure provided during structure prediction is not a good predictor of docking performance, despite accurately reflecting the quality of the alpha carbon alignment with experimental structures. Removing low-confidence regions of the predicted structure and making side chains flexible improves the docking outcomes. Overall, despite high-quality prediction of backbone conformation, fine structural details limit the naive application of AlphaFold2 models as docking targets.     

Assessment of CAPRI predictions in rounds 3-5 shows progress in docking procedures

The current status of docking procedures for predicting protein-protein interactions starting from their three-dimensional (3D) structure is reassessed by evaluating blind predictions, performed during 2003-2004 as part of Rounds 3-5 of the community-wide experiment on Critical Assessment of PRedicted Interactions (CAPRI). Ten newly determined structures of protein-protein complexes were used as targets for these rounds. They comprised 2 enzyme-inhibitor complexes, 2 antigen-antibody complexes, 2 complexes involved in cellular signaling, 2 homo-oligomers, and a complex between 2 components of the bacterial cellulosome. For most targets, the predictors were given the experimental structures of 1 unbound and 1 bound component, with the latter in a random orientation. For some, the structure of the free component was derived from that of a related protein, requiring the use of homology modeling. In some of the targets, significant differences in conformation were displayed between the bound and unbound components, representing a major challenge for the docking procedures. For 1 target, predictions could not go to completion. In total, 1866 predictions submitted by 30 groups were evaluated. Over one-third of these groups applied completely novel docking algorithms and scoring functions, with several of them specifically addressing the challenge of dealing with side-chain and backbone flexibility. The quality of the predicted interactions was evaluated by comparison to the experimental structures of the targets, made available for the evaluation, using the well-agreed-upon criteria used previously. Twenty-four groups, which for the first time included an automatic Web server, produced predictions ranking from acceptable to highly accurate for all targets, including those where the structures of the bound and unbound forms differed substantially. These results and a brief survey of the methods used by participants of CAPRI Rounds 3-5 suggest that genuine progress in the performance of docking methods is being achieved, with CAPRI acting as the catalyst.     

Development and testing of an automated approach to protein docking

A new version of GRAMM was applied to Targets 14, 18, and 19 in CAPRI Round 5. The predictions were generated without manual intervention. Ten top-ranked matches for each target were submitted. The docking was performed by a rigid-body procedure with a smoothed potential function to accommodate conformational changes. The first stage was a global search on a fine grid with a projection of a smoothed Lennard-Jones potential. The top predictions from the first stage were subjected to the conjugate gradient minimization with the same smoothed potential. The resulting local minima were reranked according to the weighted sum of Lennard-Jones potential, pairwise residue-residue statistical preferences, cluster occupancy, and the degree of the evolutionary conservation of the predicted interface. For Targets 14 and 18, the conformation of the complex was predicted with root-mean-square deviation (RMSD) of the ligand interface atoms 0.68 A and 1.88 A correspondingly. For Target 19, the interface areas on both proteins were correctly predicted. The performance of the procedure was also analyzed on the benchmark of bound-unbound protein complexes. The results show that, on average, conformations of only 3 side-chains need to be optimized during docking of unbound structures before the backbone changes become a limiting factor. The GRAMM-X docking server is available for public use at http://www.bioinformatics.ku.edu.     

PDBsum extras: SARS‐CoV‐2 and AlphaFold models

The PDBsum web server provides structural analyses of the entries in the Protein Data Bank (PDB). Two recent additions are described here. The first is the detailed analysis of the SARS‐CoV‐2 virus protein structures in the PDB. These include the variants of concern, which are shown both on the sequences and 3D structures of the proteins. The second addition is the inclusion of the available AlphaFold models for human proteins. The pages allow a search of the protein against existing structures in the PDB via the Sequence Annotated by Structure (SAS) server, so one can easily compare the predicted model against experimentally determined structures. The server is freely accessible to all at http://www.ebi.ac.uk/pdbsum.     

High-resolution protein-protein docking

The high-resolution prediction of protein-protein docking can now create structures with atomic-level accuracy. This progress arises from both improvements in the rapid sampling of conformations and increased accuracy of binding free energy calculations. Consequently, the quality of models submitted to the blind prediction challenge CAPRI (Critical Assessment of PRedicted Interactions) has steadily increased, including complexes predicted from homology structures of one binding partner and complexes with atomic accuracy at the interface. By exploiting experimental information, docking has created model structures for real applications, even when confronted with challenges such as moving backbones and uncertain monomer structures. Work remains to be done in docking large or flexible proteins, ranking models consistently, and producing models accurate enough to allow computational design of higher affinities or specificities.     

BSP-SLIM: a blind low-resolution ligand-protein docking approach using predicted protein structures

We developed BSP‐SLIM, a new method for ligand–protein blind docking using low‐resolution protein structures. For a given sequence, protein structures are first predicted by I‐TASSER; putative ligand binding sites are transferred from holo‐template structures which are analogous to the I‐TASSER models; ligand–protein docking conformations are then constructed by shape and chemical match of ligand with the negative image of binding pockets. BSP‐SLIM was tested on 71 ligand–protein complexes from the Astex diverse set where the protein structures were predicted by I‐TASSER with an average RMSD 2.92 Å on the binding residues. Using I‐TASSER models, the median ligand RMSD of BSP‐SLIM docking is 3.99 Å which is 5.94 Å lower than that by AutoDock; the median binding‐site error by BSP‐SLIM is 1.77 Å which is 6.23 Å lower than that by AutoDock and 3.43 Å lower than that by LIGSITE^CSC. Compared to the models using crystal protein structures, the median ligand RMSD by BSP‐SLIM using I‐TASSER models increases by 0.87 Å, while that by AutoDock increases by 8.41 Å; the median binding‐site error by BSP‐SLIM increase by 0.69Å while that by AutoDock and LIGSITE^CSC increases by 7.31 Å and 1.41 Å, respectively. As case studies, BSP‐SLIM was used in virtual screening for six target proteins, which prioritized actives of 25% and 50% in the top 9.2% and 17% of the library on average, respectively. These results demonstrate the usefulness of the template‐based coarse‐grained algorithms in the low‐resolution ligand–protein docking and drug‐screening. An on‐line BSP‐SLIM server is freely available at http://zhanglab.ccmb.med.umich.edu/BSP‐SLIM . Proteins 2012. © 2011 Wiley Periodicals, Inc.     

Successful discrimination of protein interactions

We present results from the prediction of protein complexes associated with the first Critical Assessment of PRediction of Interactions (CAPRI) experiment. Our algorithm, SmoothDock, comprises four steps: (1) we perform rigid body docking using the program DOT, keeping the top 20,000 structures as ranked by surface complementarity; (2) we rerank these structures according to a free energy estimate that includes both desolvation and electrostatics and retain the top 2000 complexes; (3) we cluster the filtered complexes using a pairwise root-mean-square deviation (RMSD) criterion; (4) the 25 largest clusters are subject to a smooth docking discrimination algorithm where van der Waals forces are taken into account. We predicted targets 1, 6, and 7 with RMSDs of 9.5, 2.4, and 2.6 A, respectively. More importantly, from the perspective of biological applications, our approach consistently ranked the correct model first (i.e., with highest confidence). For target 5 we identified the binding region but not the correct orientation. Although we were able to find reasonable clusters for all targets, low-affinity complexes (K(d) < nM) were harder to discriminate. For four of seven targets, the top models predicted by our automated procedure were among the best communitywide predictions.     

Molecular simulation of a series of benzothiazole PI3Kα inhibitors: probing the relationship between structural features, anti-tumor potency and selectivity

The phosphatidylinositol 3-kinase α (PI3Kα) was genetically validated as a promising therapeutic target for developing novel anticancer drugs. In order to explore the structure-activity correlation of benzothiazole series as inhibitors of PI3Kα, comparative molecular field analysis (CoMFA), comparative molecular similarity indices analysis (CoMSIA) were performed on 61 promising molecules to build 3D-QSAR models based on both the ligand- and receptor-based methods. The best CoMFA and CoMSIA models had a cross-validated coefficient r(cv)(2) of 0.618 and 0.621, predicted correlation coefficient r(pred) (2) of 0.812 and 0.83, respectively, proving their high correlative and predictive abilities on both the training and test sets. In addition, docking analysis and molecular dynamics simulation (MD) were also applied to elucidate the probable binding modes of these inhibitors at the ATP binding pocket. Based on the contour maps and MD results, some key structural factors responsible for the activity of this series of compounds were revealed as follows: (1) Ring-A has a strong preference for bulky hydrophobic or aromatic groups; (2) Electron-withdrawing groups at the para position of ring-B and hydrophilic substituents in ring-B region may benefit the potency; (3) A polar substituent like -NHSO(2)- between ring-A and ring-B can enhance the activity of the drug by providing hydrogen bonding interaction with the protein target. The satisfactory results obtained from this work strongly suggest that the developed 3D-QSAR models and the obtained PI3Kα inhibitor binding structures are reasonable for the prediction of the activity of new inhibitors and be helpful in future PI3Kα inhibitor design.     

Structure prediction of biological assemblies using GALAXY in CAPRI rounds 38-45

We participated in CARPI rounds 38-45 both as a server predictor and a human predictor. These CAPRI rounds provided excellent opportunities for testing prediction methods for three classes of protein interactions, that is, protein-protein, protein-peptide, and protein-oligosaccharide interactions. Both template-based methods (GalaxyTBM for monomer protein, GalaxyHomomer for homo-oligomer protein, GalaxyPepDock for protein-peptide complex) and ab initio docking methods (GalaxyTongDock and GalaxyPPDock for protein oligomer, GalaxyPepDock-ab-initio for protein-peptide complex, GalaxyDock2 and Galaxy7TM for protein-oligosaccharide complex) have been tested. Template-based methods depend heavily on the availability of proper templates and template-target similarity, and template-target difference is responsible for inaccuracy of template-based models. Inaccurate template-based models could be improved by our structure refinement and loop modeling methods based on physics-based energy optimization (GalaxyRefineComplex and GalaxyLoop) for several CAPRI targets. Current ab initio docking methods require accurate protein structures as input. Small conformational changes from input structure could be accounted for by our docking methods, producing one of the best models for several CAPRI targets. However, predicting large conformational changes involving protein backbone is still challenging, and full exploration of physics-based methods for such problems is still to come.     

A holistic approach to protein docking

Docking of unbound protein structures into a complex has gained significant progress in recent years, but nonetheless still poses a great challenge. We have pursued a holistic approach to docking which brings together effective methods at different stages. First, protein-protein interaction sites are predicted or obtained from experimental studies in the literature. Interface prediction/experimental data are then used to guide the generation of docked poses or to rank docked poses generated from an unbiased search. Finally, selected models are refined by lengthy molecular dynamics (MD) simulations in explicit water. For CAPRI target T27, we used information on interaction sites as input to drive docking and as a filter to rank docked poses. Lead candidates were then clustered according to RMSD among them. From the clustering, 10 models were selected and subject to refinement by MD simulations. Our Model 7 is rated number one among all submissions according to L_rmsd. Six of our other submissions are rated acceptable. As scorer, eight of our submissions are rated acceptable.     

Modeling oligomers with Cn or Dn symmetry: application to CAPRI target 10

The abundance of oligomeric proteins makes them a frequent target for structure prediction. However, homologous proteins sometimes adopt different oligomerization states, rendering the prediction of structures of whole oligomers beyond the scope of comparative modeling. This obstacle can be overcome by combining comparative modeling of the single subunit of an oligomer with docking techniques, designed for predicting subunit-subunit interfaces. We present here algorithms for predicting the structures of homo-oligomers with C(n) or D(n) (n > 2) symmetry. The prediction procedure includes a symmetry-restricted docking step followed by a C(n) or D(n) oligomer-forming step, in which the dimers from the docking step are assembled to oligomers. The procedure is applied to each of the crystallographically independent subunits in 8 C(n) and 3 D(n) oligomers, producing very accurate predictions. It is further applied to a single monomer of the tick-borne encephalitis virus coat protein E (Target 10 of the CAPRI experiment). The predicted trimer ranked 30, obtained via rigid-body geometric-hydrophobic docking followed by C(n) oligomer formation, is very similar to the experimentally observed trimer formed by domain II of this protein. Furthermore, the predicted trimer formed from the separated domain I is also close to the experimental structure.     

Homology modeling of dopamine d(2) and d(3) receptors: molecular dynamics refinement and docking evaluation

Dopamine (DA) receptors, a class of G-protein coupled receptors (GPCRs), have been targeted for drug development for the treatment of neurological, psychiatric and ocular disorders. The lack of structural information about GPCRs and their ligand complexes has prompted the development of homology models of these proteins aimed at structure-based drug design. Crystal structure of human dopamine D(3) (hD(3)) receptor has been recently solved. Based on the hD(3) receptor crystal structure we generated dopamine D(2) and D(3) receptor models and refined them with molecular dynamics (MD) protocol. Refined structures, obtained from the MD simulations in membrane environment, were subsequently used in molecular docking studies in order to investigate potential sites of interaction. The structure of hD(3) and hD(2L) receptors was differentiated by means of MD simulations and D(3) selective ligands were discriminated, in terms of binding energy, by docking calculation. Robust correlation of computed and experimental K(i) was obtained for hD(3) and hD(2L) receptor ligands. In conclusion, the present computational approach seems suitable to build and refine structure models of homologous dopamine receptors that may be of value for structure-based drug discovery of selective dopaminergic ligands.     

Chemical inhibition of fatty acid synthase: molecular docking analysis and biochemical validation in ocular cancer cells

Fatty acid biosynthesis is an attractive target for anti-cancer therapeutics. The ocular cancer, retinoblastoma cells were treated with fatty acid synthase (FASN) enzyme inhibitors: cerulenin, triclosan and orlistat. The IC₅₀ and dose-dependent sensitivity of cancer cells to FASN inhibitors decrease in biologic enzyme activity, and cell morphology alterations were analysed. Molecular interactions of enzyme-inhibitor complexes were studied by molecular modelling and docking simulations. The crystal structures of ketoacyl synthase (PDB ID:3HHD) (cerulenin) and thioesterase (PDB ID:2PX6) (orlistat) domains of human FASN were utilized for docking, while for the non-crystallised human FASN enoyl reductase domain (triclosan), homology model was built and used for docking. All three inhibitors showed significant binding energy indicating stable complex formation with their respective FASN subunits. The predicted Ki value of the FASN inhibitors corroborated well with their corresponding anti-cancer effects.     

A leucine-rich repeat assembly approach for homology modeling of the human TLR5-10 and mouse TLR11-13 ectodomains

So far, 13 groups of mammalian Toll-like receptors (TLRs) have been identified. Most TLRs have been shown to recognize pathogen-associated molecular patterns from a wide range of invading agents and initiate both innate and adaptive immune responses. The TLR ectodomains are composed of varying numbers and types of leucine-rich repeats (LRRs). As the crystal structures are currently missing for most TLR ligand-binding ectodomains, homology modeling enables first predictions of their three-dimensional structures on the basis of the determined crystal structures of TLR ectodomains. However, the quality of the predicted models that are generated from full-length templates can be limited due to low sequence identity between the target and templates. To obtain better templates for modeling, we have developed an LRR template assembly approach. Individual LRR templates that are locally optimal for the target sequence are assembled into multiple templates. This method was validated through the comparison of a predicted model with the crystal structure of mouse TLR3. With this method, we also constructed ectodomain models of human TLR5, TLR6, TLR7, TLR8, TLR9, and TLR10 and mouse TLR11, TLR12, and TLR13 that can be used as first passes for a computational simulation of ligand docking or to design mutation experiments. This template assembly approach can be extended to other repetitive proteins.     

Predicting P-glycoprotein-mediated drug transport based on support vector machine and three-dimensional crystal structure of P-glycoprotein

Human P-glycoprotein (P-gp) is an ATP-binding cassette multidrug transporter that confers resistance to a wide range of chemotherapeutic agents in cancer cells by active efflux of the drugs from cells. P-gp also plays a key role in limiting oral absorption and brain penetration and in facilitating biliary and renal elimination of structurally diverse drugs. Thus, identification of drugs or new molecular entities to be P-gp substrates is of vital importance for predicting the pharmacokinetics, efficacy, safety, or tissue levels of drugs or drug candidates. At present, publicly available, reliable in silico models predicting P-gp substrates are scarce. In this study, a support vector machine (SVM) method was developed to predict P-gp substrates and P-gp-substrate interactions, based on a training data set of 197 known P-gp substrates and non-substrates collected from the literature. We showed that the SVM method had a prediction accuracy of approximately 80% on an independent external validation data set of 32 compounds. A homology model of human P-gp based on the X-ray structure of mouse P-gp as a template has been constructed. We showed that molecular docking to the P-gp structures successfully predicted the geometry of P-gp-ligand complexes. Our SVM prediction and the molecular docking methods have been integrated into a free web server (http://pgp.althotas.com), which allows the users to predict whether a given compound is a P-gp substrate and how it binds to and interacts with P-gp. Utilization of such a web server may prove valuable for both rational drug design and screening.     

Interaction of c(60)-fullerene and carboxyfullerene with proteins: docking and binding site alignment

The unique properties of fullerenes have raised the interest of using them for biomedical applications. Within this framework, the interactions of fullerenes with proteins have been an exciting research target, yet little is known about how native proteins can bind fullerenes, and what is the nature of these interactions. Moreover, though some proteins have been shown to interact with fullerenes, up to date, no crystal structure of such complexes was obtained. Here we report docking studies aimed at examining the interactions of fullerene in two forms (C60 nonsubstituted fullerene and carboxyfullerene) with four proteins that are known to bind fullerene derivatives: HIV protease, fullerene-specific antibody, human serum albumin, and bovine serum albumin. Our work provides docking models with detailed binding pockets information, which closely match available experimental data. We further compare the predicted binding sites using a novel multiple binding site alignment method. A high similarity between the physicochemical properties and surface geometry was found for fullerene's binding sites of HIV protease and the human and bovine serum albumins.