How inaccuracies in protein structure models affect estimates of protein-ligand interactions: computational analysis of HIV-I protease inhibitor binding

Pathogen-driven cancers from a structural perspective: Targeting host-pathogen protein-protein interactions

Host-pathogen interactions (HPIs) affect and involve multiple mechanisms in both the pathogen and the host. Pathogen interactions disrupt homeostasis in host cells, with their toxins interfering with host mechanisms, resulting in infections, diseases, and disorders, extending from AIDS and COVID-19, to cancer. Studies of the three-dimensional (3D) structures of host-pathogen complexes aim to understand how pathogens interact with their hosts. They also aim to contribute to the development of rational therapeutics, as well as preventive measures. However, structural studies are fraught with challenges toward these aims. This review describes the state-of-the-art in protein-protein interactions (PPIs) between the host and pathogens from the structural standpoint. It discusses computational aspects of predicting these PPIs, including machine learning (ML) and artificial intelligence (AI)-driven, and overviews available computational methods and their challenges. It concludes with examples of how theoretical computational approaches can result in a therapeutic agent with a potential of being used in the clinics, as well as future directions.

Protein-Nucleic Acid Interactions for RNA Polymerase II Elongation Factors by Molecular Dynamics Simulations

RNA polymerase II (Pol II) forms a complex with elongation factors to proceed to the elongation stage of the transcription process. In this work, we studied the elongation factor SPT5 and explored the protein-nucleic acid interactions for the isolated systems of KOW1 and KOW4 domains of SPT5 with DNA and RNA, respectively. We performed molecular dynamics (MD) simulations using three commonly used force fields that are CHARMM c36m, AMBER ff14sb, and ff19sb. Simulations showed strong protein-nucleic acid interactions and low electrostatic binding free energies for all force fields used. RNA was found to be highly dynamic with all force fields, while DNA had relatively more stable conformations with the AMBER force fields compared to that with CHARMM. Furthermore, we performed MD simulations of the complete elongation complex using CHARMM c36m and AMBER ff19sb force fields to compare the dynamics and interactions with the isolated systems. Similarly, strong KOW1 and DNA interactions were observed in the complete elongation complex simulations and DNA was further stabilized by a network of interactions involving SPT5-KOW1, SPT4, and rpb2 of Pol II. Overall, our study showed that the differences between CHARMM and AMBER force fields strongly affect the dynamics of the nucleic acids. CHARMM provides highly flexible DNA, while AMBER largely stabilizes the DNA structure. Although the presence of the entire interaction network stabilized the DNA and decreased the differences in the results from the two force fields, the discrepancies of the force fields for smaller systems may reflect their problems in generating accurate dynamics of nucleic acids.

Molecular dynamics and free energy studies on the Drosophila melanogaster and Leptinotarsa decemlineata ecdysone receptor complexed with agonists: Mechanism for binding and selectivity

The ecdysone receptor is a nuclear hormone receptor that plays a pivotal role in the insect metamorphosis and development. To address the molecular mechanisms of binding and selectivity, the interactions of two typical agonists Ponasterone A and 20-Hydroxyecdysone with Drosophila melanogaster (DME) and Leptinotarsa decemlineata ecdysone (LDE) receptors were investigated by homology modeling, molecular docking, molecular dynamic simulation, and thermodynamic analysis. We discover that 1) the L5-loop, L11-loop, and H12 helix for DME, L7-loop, and L11-loop for LDE are more flexible, which affect the global dynamics of the ligand-binding pocket, thus facilitating the ligand recognition of ecdysone receptor; 2) several key residues (Thr55/Thr37, Phe109/Phe91, Arg95/Arg77, Arg99/Arg81, Phe108/Leu90, and Ala110/Val92) are responsible for the binding of the proteins; 3) the binding-free energy is mainly contributed by the van der Waals forces as well as the electrostatic interactions of ligand and receptor; 4) the computed binding-free energy difference between DME-C1 and LDE-C1 is -4.65 kcal/mol, explains that C1 can form many more interactions with the DME; 5) residues Phe108/Leu90 and Ala110/Val92 have relatively position and orientation difference in the two receptors, accounting most likely for the ligand selectivity of ecdysone receptor from different orders of insects. This study underscores the expectation that different insect pests should be able to discriminate among compounds from different as yet undiscovered compounds, and the results firstly show a structural and functional relay between the agonists and receptors (DME and LDE), which can provide an avenue for the development of target-specific insecticides. Communicated by Ramaswamy H. Sarma.

Modelling three-dimensional protein structures for applications in drug design

A structural perspective of drug target and anti-target proteins, and their molecular interactions with biologically active molecules, largely advances many areas of drug discovery, including target validation, hit and lead finding and lead optimisation. In the absence of experimental 3D structures, protein structure prediction often offers a suitable alternative to facilitate structure-based studies. This review outlines recent methodical advances in homology modelling, with a focus on those techniques that necessitate consideration of ligand binding. In this context, model quality estimation deserves special attention because the accuracy and reliability of different structure prediction techniques vary considerably, and the quality of a model ultimately determines its usefulness for structure-based drug discovery. Examples of G-protein-coupled receptors (GPCRs) and ADMET-related proteins were selected to illustrate recent progress and current limitations of protein structure prediction. Basic guidelines for good modelling practice are also provided.

Improved scFv anti-HIV-1 p17 binding affinity guided from the theoretical calculation of pairwise decomposition energies and computational alanine scanning

Computational approaches have been used to evaluate and define important residues for protein-protein interactions, especially antigen-antibody complexes. In our previous study, pairwise decomposition of residue interaction energies of single chain Fv with HIV-1 p17 epitope variants has indicated the key specific residues in the complementary determining regions (CDRs) of scFv anti-p17. In this present investigation in order to determine whether a specific side chain group of residue in CDRs plays an important role in bioactivity, computational alanine scanning has been applied. Molecular dynamics simulations were done with several complexes of original scFv anti-p17 and scFv anti-p17mutants with HIV-1 p17 epitope variants with a production run up to 10 ns. With the combination of pairwise decomposition residue interaction and alanine scanning calculations, the point mutation has been initially selected at the position MET100 to improve the residue binding affinity. The calculated docking interaction energy between a single mutation from methionine to either arginine or glycine has shown the improved binding affinity, contributed from the electrostatic interaction with the negative favorably interaction energy, compared to the wild type. Theoretical calculations agreed well with the results from the peptide ELISA results.

Interaction between trehalose and MTHase from Sulfolobus solfataricus studied by theoretical computation and site-directed mutagenesis

Maltooligosyltrehalose trehalohydrolase (MTHase) catalyzes the release of trehalose by cleaving the α-1,4-glucosidic linkage next to the α-1,1-linked terminal disaccharide of maltooligosyltrehalose. Computer simulation using the hydrogen bond analysis, free energy decomposition, and computational alanine scanning were employed to investigate the interaction between maltooligosyltrehalose and the enzyme. The same residues that were chosen for theoretical investigation were also studied by site-directed mutagenesis and enzyme kinetic analysis. The importance of residues determined either experimentally or computed theoretically were in good accord with each other. It was found that residues Y155, D156, and W218 of subsites -2 and -3 of the enzyme might play an important role in interacting with the ligand. The theoretically constructed structure of the enzyme-ligand complex was further validated through an ab initio quantum chemical calculation using the Gaussian09 package. The activation energy computed from this latter study was very similar to those reported in literatures for the same type of hydrolysis reactions.

Structural modelling pipelines in next generation sequencing projects

Our capacity to reliably predict protein structure from sequence is steadily improving due to the increased numbers and better targeting of protein structures being experimentally determined by structural genomics projects, along with the development of better modeling methodologies. Template-based (homology) modeling and de novo modeling methods are being combined to fill in remaining gaps in template coverage, and powerful automated structural modeling pipelines are being applied to large data sets of protein sequences. The improved quality of 3D models of proteins has led to their routine use in assessing the functional impact of nonsynonymous single nucleotide polymorphisms (nsSNPs) in specific protein systems, with the development of approaches that may be applied in a predictive fashion to nsSNPs emerging from next-generation sequencing projects. The challenges encountered in deriving functionally meaningful deductions from structural modeling can be quite different for proteins of different protein functional classes. The specific challenges to the assessment of the structural and functional impact of nsSNPs in globular proteins such as binding and regulatory proteins, structural proteins, and enzymes are discussed, as well as membrane transport proteins and ion channels. The mapping of reliable predictions of the structural and functional impact of SNPs, generated from automated modeling pipelines, on to protein-protein interaction networks will facilitate new approaches to understanding complex polygenic disorders and predisposition to disease.

Elucidating the energetics of entropically driven protein-ligand association: calculations of absolute binding free energy and entropy

The binding of proteins and ligands is generally associated with the loss of translational, rotational, and conformational entropy. In many cases, however, the net entropy change due to binding is positive. To develop a deeper understanding of the energetics of entropically driven protein-ligand binding, we calculated the absolute binding free energies and binding entropies for two HIV-1 protease inhibitors Nelfinavir and Amprenavir using the double-decoupling method with molecular dynamics simulations in explicit solvent. For both ligands, the calculated absolute binding free energies are in general agreement with experiments. The statistical error in the computed ΔG(bind) due to convergence problem is estimated to be ≥2 kcal/mol. The decomposition of free energies indicates that, although the binding of Nelfinavir is driven by nonpolar interaction, Amprenavir binding benefits from both nonpolar and electrostatic interactions. The calculated absolute binding entropies show that (1) Nelfinavir binding is driven by large entropy change and (2) the entropy of Amprenavir binding is much less favorable compared with that of Nelfinavir. Both results are consistent with experiments. To obtain qualitative insights into the entropic effects, we decomposed the absolute binding entropy into different contributions based on the temperature dependence of free energies along different legs of the thermodynamic pathway. The results suggest that the favorable entropic contribution to binding is dominated by the ligand desolvation entropy. The entropy gain due to solvent release from binding site appears to be more than offset by the reduction of rotational and vibrational entropies upon binding.

The evolution of catalytic function in the HIV-1 protease

The evolution of species is a complex phenomenon based on the optimization of a multidimensional function referred to as fitness. At the level of biomolecular evolution, the fitness function can be reduced to include physiochemical properties relevant to the biological function of a particular molecule. In this work, questions involving the physical-chemical mechanisms underlying the evolution of HIV-1 protease are addressed through molecular simulation and subsequent analysis of thermodynamic properties related to the activity of the enzyme. Specifically, the impact of 40 single amino acid mutations on the binding affinity toward the matrix/capsid (MA/CA) substrate and corresponding transition state intermediate has been characterized using a molecular mechanics Poisson-Boltzmann surface area approach. We demonstrate that this approach is capable of extracting statistically significant information relevant to experimentally determined catalytic activity. Further, no correlation was observed between the effect of mutations on substrate and transition state binding, suggesting independent evolutionary pathways toward optimizing substrate specificity and catalytic activity. In addition, a detailed analysis of calculated binding affinity data suggests that ground-state destabilization (reduced binding affinity for the substrate) could be a contributing factor in the evolutionary optimization of HIV-1 protease. A numerical model is developed to demonstrate that ground-state destabilization is a valid mechanism for activity optimization given the high concentrations of substrate experienced by the functional enzyme in vivo.

Stability and dynamics of cyclic diguanylic acid in solution

Cyclic diguanylic acid (CDG) is a ubiquitous messenger involved in bacterial signaling networks. Despite its central role in motility, biofilm formation, virulence, and flagellum development, fundamental properties such as its aggregation state are still poorly understood. Here the dynamics and stability of metal-free and Mg(2+)-bound CDG are characterized. Atomistic simulations establish that the CDG dimer is slightly favored (by -5 kcal mol(-1)) over its dissociated form (2 CDG), while the Mg(2+) ion coordinated in the X-ray structure readily dissociates from (CDG)(2) in solution and prefers water coordination. As a ligand in a protein, CDG binds both as a U-shaped and a quasilinear monomer. The current results indicate that the energy difference between these two conformations is only a few kilocalories per mole, which explains the facile adaptation to different protein environments. This, together with the slight preference of (CDG)(2) over 2 CDG suggests that (CDG)(2) binding to a protein does probably not occur via sequential binding of two individual monomers.

Use of the FACTS solvation model for protein-ligand docking calculations. Application to EADock

Protein-ligand docking has made important progress during the last decade and has become a powerful tool for drug development, opening the way to virtual high throughput screening and in silico structure-based ligand design. Despite the flattering picture that has been drawn, recent publications have shown that the docking problem is far from being solved, and that more developments are still needed to achieve high successful prediction rates and accuracy. Introducing an accurate description of the solvation effect upon binding is thought to be essential to achieve this goal. In particular, EADock uses the Generalized Born Molecular Volume 2 (GBMV2) solvent model, which has been shown to reproduce accurately the desolvation energies calculated by solving the Poisson equation. Here, the implementation of the Fast Analytical Continuum Treatment of Solvation (FACTS) as an implicit solvation model in small molecules docking calculations has been assessed using the EADock docking program. Our results strongly support the use of FACTS for docking. The success rates of EADock/FACTS and EADock/GBMV2 are similar, i.e. around 75% for local docking and 65% for blind docking. However, these results come at a much lower computational cost: FACTS is 10 times faster than GBMV2 in calculating the total electrostatic energy, and allows a speed up of EADock by a factor of 4. This study also supports the EADock development strategy relying on the CHARMM package for energy calculations, which enables straightforward implementation and testing of the latest developments in the field of Molecular Modeling.

Thermodynamic resolution: how do errors in modeled protein structures affect binding affinity predictions?

The present study addresses the effect of structural distortion, caused by protein modeling errors, on calculated binding affinities toward small molecules. The binding affinities to a total of 300 distorted structures based on five different protein-ligand complexes were evaluated to establish a broadly applicable relationship between errors in protein structure and errors in calculated binding affinities. Relatively accurate protein models (less than 2 A RMSD within the binding site) demonstrate a 14.78 (+/-7.5)% deviation in binding affinity from that calculated by using the corresponding crystal structure. For structures of 2-3 A, 3-4 A, and >4 A RMSD within the binding site, the error in calculated binding affinity increases to 20.8 (+/-5.98), 22.79 (+/-11.3), and 29.43 (+/-11.47)%, respectively. The results described here may be used in combination with other tools to evaluate the utility of modeled protein structures for drug development or other ligand-binding studies.

Fourier transform convolution integrals applied to generalized Born molecular volume

Generalized Born methods are currently among the solvation models most commonly used for biological applications. We reformulate the generalized Born molecular volume method initially described by (Lee et al, 2003, J Phys Chem, 116, 10606; Lee et al, 2003, J Comp Chem, 24, 1348) using fast Fourier transform convolution integrals. Changes in the initial method are discussed and analyzed. Finally, the method is extensively checked with snapshots from common molecular modeling applications: binding free energy computations and docking. Biologically relevant test systems are chosen, including 855-36091 atoms. It is clearly demonstrated that, precision-wise, the proposed method performs as good as the original, and could better benefit from hardware accelerated boards.

QMEANclust: estimation of protein model quality by combining a composite scoring function with structural density information

BACKGROUND

The selection of the most accurate protein model from a set of alternatives is a crucial step in protein structure prediction both in template-based and ab initio approaches. Scoring functions have been developed which can either return a quality estimate for a single model or derive a score from the information contained in the ensemble of models for a given sequence. Local structural features occurring more frequently in the ensemble have a greater probability of being correct. Within the context of the CASP experiment, these so called consensus methods have been shown to perform considerably better in selecting good candidate models, but tend to fail if the best models are far from the dominant structural cluster. In this paper we show that model selection can be improved if both approaches are combined by pre-filtering the models used during the calculation of the structural consensus.

RESULTS

Our recently published QMEAN composite scoring function has been improved by including an all-atom interaction potential term. The preliminary model ranking based on the new QMEAN score is used to select a subset of reliable models against which the structural consensus score is calculated. This scoring function called QMEANclust achieves a correlation coefficient of predicted quality score and GDT_TS of 0.9 averaged over the 98 CASP7 targets and perform significantly better in selecting good models from the ensemble of server models than any other groups participating in the quality estimation category of CASP7. Both scoring functions are also benchmarked on the MOULDER test set consisting of 20 target proteins each with 300 alternatives models generated by MODELLER. QMEAN outperforms all other tested scoring functions operating on individual models, while the consensus method QMEANclust only works properly on decoy sets containing a certain fraction of near-native conformations. We also present a local version of QMEAN for the per-residue estimation of model quality (QMEANlocal) and compare it to a new local consensus-based approach.

CONCLUSION

Improved model selection is obtained by using a composite scoring function operating on single models in order to enrich higher quality models which are subsequently used to calculate the structural consensus. The performance of consensus-based methods such as QMEANclust highly depends on the composition and quality of the model ensemble to be analysed. Therefore, performance estimates for consensus methods based on large meta-datasets (e.g. CASP) might overrate their applicability in more realistic modelling situations with smaller sets of models based on individual methods.

Docking and scoring with alternative side-chain conformations

We describe a scoring and modeling procedure for docking ligands into protein models that have either modeled or flexible side-chain conformations. Our methodical contribution comprises a procedure for generating new potentials of mean force for the ROTA scoring function which we have introduced previously for optimizing side-chain conformations with the tool IRECS. The ROTA potentials are specially trained to tolerate small-scale positional errors of atoms that are characteristic of (i) side-chain conformations that are modeled using a sparse rotamer library and (ii) ligand conformations that are generated using a docking program. We generated both rigid and flexible protein models with our side-chain prediction tool IRECS and docked ligands to proteins using the scoring function ROTA and the docking programs FlexX (for rigid side chains) and FlexE (for flexible side chains). We validated our approach on the forty screening targets of the DUD database. The validation shows that the ROTA potentials are especially well suited for estimating the binding affinity of ligands to proteins. The results also show that our procedure can compensate for the performance decrease in screening that occurs when using protein models with side chains modeled with a rotamer library instead of using X-ray structures. The average runtime per ligand of our method is 168 seconds on an Opteron V20z, which is fast enough to allow virtual screening of compound libraries for drug candidates.

The SWISS-MODEL Repository and associated resources

SWISS-MODEL Repository (http://swissmodel.expasy.org/repository/) is a database of 3D protein structure models generated by the SWISS-MODEL homology-modelling pipeline. The aim of the SWISS-MODEL Repository is to provide access to an up-to-date collection of annotated 3D protein models generated by automated homology modelling for all sequences in Swiss-Prot and for relevant models organisms. Regular updates ensure that target coverage is complete, that models are built using the most recent sequence and template structure databases, and that improvements in the underlying modelling pipeline are fully utilised. As of September 2008, the database contains 3.4 million entries for 2.7 million different protein sequences from the UniProt database. SWISS-MODEL Repository allows the users to assess the quality of the models in the database, search for alternative template structures, and to build models interactively via SWISS-MODEL Workspace (http://swissmodel.expasy.org/workspace/). Annotation of models with functional information and cross-linking with other databases such as the Protein Model Portal (http://www.proteinmodelportal.org) of the PSI Structural Genomics Knowledge Base facilitates the navigation between protein sequence and structure resources.

Virtual Physiological Human research initiative: the future for rational HIV treatment design?

The clinical management of HIV-positive patients receiving existing drug regimens is performed largely as a clinician’s best guess, based on available viral and patient data. The development of new antiretroviral therapies for HIV is at an impasse. How can we progress beyond this? ViroLab, an example of a new ‘style’ of research project, is attempting to answer this question by stepping into the realm of prediction. A method is in development that can model and predict the efficacy of antiretroviral treatment and may enhance clinical decision support. If extended, this method can potentially target any other pathology where a ligand–substrate interaction is concerned. This type of research falls within the remit of a new initiative: the Virtual Physiological Human (VPH). The potential impact of VPH on current and future rational treatment design is discussed.

Accurate prediction of protonation state as a prerequisite for reliable MM-PB(GB)SA binding free energy calculations of HIV-1 protease inhibitors

Binding free energies were calculated for the inhibitors lopinavir, ritonavir, saquinavir, indinavir, amprenavir, and nelfinavir bound to HIV-1 protease. An MMPB/SA-type analysis was applied to conformational samples from 3 ns explicit solvent molecular dynamics simulations of the enzyme-inhibitor complexes. Binding affinities and the sampled conformations of the inhibitor and enzyme were compared between different HIV-1 protease protonation states to find the most likely protonation state of the enzyme in the complex with each of the inhibitors. The resulting set of protonation states leads to good agreement between calculated and experimental binding affinities. Results from the MMPB/SA analysis are compared with an explicit/implicit hybrid scheme and with MMGB/SA methods. It is found that the inclusion of explicit water molecules may offer a slight advantage in reproducing absolute binding free energies while the use of the Generalized Born approximation significantly affects the accuracy of the calculated binding affinities.

Computational alanine scanning and free energy decomposition for E. coli type I signal peptidase with lipopeptide inhibitor complex

A thorough investigation of different roles of Escherichia coli type I signal peptidase residues binding to lipopeptide inhibitor has been performed by a combination of computational alanine scanning mutagenesis and free energy decomposition methods. PB and GB models are both used to evaluate the binding free energy in computational alanine scanning method and only GB model can be used to decompose the binding free energy on a per-residue basis. The regression analysis between the PB and GB model and also between the computational alanine scanning and free energy decomposition have been reported with a correlation coefficient of 0.96 and 0.83, respectively, which suggest they are both in fair agreement with each other. Moreover, the contribution components from van der Waals, electrostatic interaction, non-polar and polar energy of solvation, have been determined as well as the effects of backbones and side-chains. The results indicate that Lys145 is the most important residue for the binding but also acts as a general base, activating Ser90 to increase its nucleophility, recognizing and stabilizing the binding of lipopeptide inhibitor to the E. coli SPase. The hydroxyl group of Ser88 plays a key role for the binding of the inhibitor. Ser90 contributes more to the intramolecular interaction than to the intermolecular interaction. Tyr143 and Phe84 contribute larger van der Waals interaction energies, indicating that these residues can be important for the selection based on the shape of the inhibitors. The contributions from other several interfacial residues of the E. coli SPase are also analyzed. This study can be a guide for the optimization of lipopeptide inhibitors and future design of new therapeutic agents for the treatment of bacterial infections.