Establishing a Framework of Using Residue–Residue Interactions in Protein Difference Network Analysis

Deciphering S100B Allosteric Signaling: The Role of a Peptide Target, TRTK-12, as an Ensemble Modulator

Allostery is an essential biological phenomenon in which perturbation at one site in a biomolecule elicits a functional response at a distal location(s). It is integral to biological processes, such as cellular signaling, metabolism, and transcription regulation. Understanding allostery is also crucial for rational drug discovery. In this work, we focus on an allosteric S100B protein that belongs to the S100 class of EF-hand Ca2+-binding proteins. The Ca2+-binding affinity of S100B is modulated allosterically by TRTK-12 peptide binding 25 Å away from the Ca2+-binding site. We investigated S100B allostery by carrying out nuclear magnetic resonance (NMR) measurements along with microsecond-long molecular dynamics (MD) simulations on S100B/Ca2+ with/without TRTK-12 at different NaCl salt concentrations. NMR HSQC results show that TRTK-12 reorganizes how S100B/Ca2+ responds to different salt concentrations at both orthosteric and allosteric sites. The MD data suggest that TRTK-12 breaks the dynamic aromatic and hydrogen-bond interactions (not observed in X-ray crystallographic structures) between the hinge/helix and Ca2+-binding EF-hand loop of the two subunits in the homodimeric protein. This triggers rearrangement in the protein network architectures and leads to allosteric communication. Finally, computational studies of S100B at distinct ionic strengths suggest that ligand-bound species are more robust to the changing environment relative to the S100B/Ca2+ complex.

Unsupervised deep learning for molecular dynamics simulations: a novel analysis of protein-ligand interactions in SARS-CoV-2 Mpro

Molecular dynamics (MD) simulations, which are central to drug discovery, offer detailed insights into protein-ligand interactions. However, analyzing large MD datasets remains a challenge. Current machine-learning solutions are predominantly supervised and have data labelling and standardisation issues. In this study, we adopted an unsupervised deep-learning framework, previously benchmarked for rigid proteins, to study the more flexible SARS-CoV-2 main protease (Mpro). We ran MD simulations of Mpro with various ligands and refined the data by focusing on binding-site residues and time frames in stable protein conformations. The optimal descriptor chosen was the distance between the residues and the center of the binding pocket. Using this approach, a local dynamic ensemble was generated and fed into our neural network to compute Wasserstein distances across system pairs, revealing ligand-induced conformational differences in Mpro. Dimensionality reduction yielded an embedding map that correlated ligand-induced dynamics and binding affinity. Notably, the high-affinity compounds showed pronounced effects on the protein's conformations. We also identified the key residues that contributed to these differences. Our findings emphasize the potential of combining unsupervised deep learning with MD simulations to extract valuable information and accelerate drug discovery.

Toward a Benchmark for Markov State Models: The Folding of HP35

Adopting a 300 μs long MD trajectory of the folding of villin headpiece (HP35) by D. E. Shaw Research, we recently constructed a Markov state model (MSM) based on inter-residue contacts. The model reproduces the folding time and predicts that the native basin and unfolded region consist of metastable substates that are structurally well-characterized. Recognizing the need to establish well-defined benchmark problems, we study to what extent and in what sense this MSM can be employed as a reference model. Hence, we test the robustness of the MSM by comparing it to models that use alternative combinations of features, dimensionality reduction methods, and clustering schemes. The study suggests some main characteristics of the folding of HP35 that should be reproduced by other competitive models. Moreover, the discussion reveals which parts of the MSM workflow matter most for the considered problem and illustrates the promises and pitfalls of state-based models for the interpretation of biomolecular simulations.

Selecting Features for Markov Modeling: A Case Study on HP35

Markov state models represent a popular means to interpret molecular dynamics trajectories in terms of memoryless transitions between metastable conformational states. To provide a mechanistic understanding of the considered biomolecular process, these states should reflect structurally distinct conformations and ensure a time scale separation between fast intrastate and slow interstate dynamics. Adopting the folding of villin headpiece (HP35) as a well-established model problem, here we discuss the selection of suitable input coordinates or "features", such as backbone dihedral angles and interresidue distances. We show that dihedral angles account accurately for the structure of the native energy basin of HP35, while the unfolded region of the free energy landscape and the folding process are best described by tertiary contacts of the protein. To construct a contact-based model, we consider various ways to define and select contact distances and introduce a low-pass filtering of the feature trajectory as well as a correlation-based characterization of states. Relying on input data that faithfully account for the mechanistic origin of the studied process, the states of the resulting Markov model are clearly discriminated by the features, describe consistently the hierarchical structure of the free energy landscape, and─as a consequence─correctly reproduce the slow time scales of the process.

Non-active Site Residue in Loop L4 Alters Substrate Capture and Product Release in d-Arginine Dehydrogenase

Numerous studies demonstrate that enzymes undergo multiple conformational changes during catalysis. The malleability of enzymes forms the basis for allosteric regulation: residues located far from the active site can exert long-range dynamical effects on the active site residues to modulate catalysis. The structure of Pseudomonas aeruginosa d-arginine dehydrogenase (PaDADH) shows four loops (L1, L2, L3, and L4) that span the substrate and the FAD-binding domains. Loop L4 comprises residues 329-336, spanning over the flavin cofactor. The I335 residue on loop L4 is ∼10 Å away from the active site and ∼3.8 Å from N(1)-C(2)═O atoms of the flavin. In this study, we used molecular dynamics and biochemical techniques to investigate the effect of the mutation of I335 to histidine on the catalytic function of PaDADH. Molecular dynamics showed that the conformational dynamics of PaDADH are shifted to a more closed conformation in the I335H variant. In agreement with an enzyme that samples more in a closed conformation, the kinetic data of the I335H variant showed a 40-fold decrease in the rate constant of substrate association (k₁), a 340-fold reduction in the rate constant of substrate dissociation from the enzyme-substrate complex (k₂), and a 24-fold decrease in the rate constant of product release (k₅), compared to that of the wild-type. Surprisingly, the kinetic data are consistent with the mutation having a negligible effect on the reactivity of the flavin. Altogether, the data indicate that the residue at position 335 has a long-range dynamical effect on the catalytic function in PaDADH.

A new approach for extracting information from protein dynamics

Increased ability to predict protein structures is moving research focus towards understanding protein dynamics. A promising approach is to represent protein dynamics through networks and take advantage of well-developed methods from network science. Most studies build protein dynamics networks from correlation measures, an approach that only works under very specific conditions, instead of the more robust inverse approach. Thus, we apply the inverse approach to the dynamics of protein dihedral angles, a system of internal coordinates, to avoid structural alignment. Using the well-characterized adhesion protein, FimH, we show that our method identifies networks that are physically interpretable, robust, and relevant to the allosteric pathway sites. We further use our approach to detect dynamical differences, despite structural similarity, for Siglec-8 in the immune system, and the SARS-CoV-2 spike protein. Our study demonstrates that using the inverse approach to extract a network from protein dynamics yields important biophysical insights.

The Importance of Charge Transfer and Solvent Screening in the Interactions of Backbones and Functional Groups in Amino Acid Residues and Nucleotides

Quantum mechanical (QM) calculations at the level of density-functional tight-binding are applied to a protein–DNA complex (PDB: 2o8b) consisting of 3763 atoms, averaging 100 snapshots from molecular dynamics simulations. A detailed comparison of QM and force field (Amber) results is presented. It is shown that, when solvent screening is taken into account, the contributions of the backbones are small, and the binding of nucleotides in the double helix is governed by the base–base interactions. On the other hand, the backbones can make a substantial contribution to the binding of amino acid residues to nucleotides and other residues. The effect of charge transfer on the interactions is also analyzed, revealing that the actual charge of nucleotides and amino acid residues can differ by as much as 6 and 8% from the formal integer charge, respectively. The effect of interactions on topological models (protein -residue networks) is elucidated.

Linking protein structural and functional change to mutation using amino acid networks

The function of a protein is strongly dependent on its structure. During evolution, proteins acquire new functions through mutations in the amino-acid sequence. Given the advance in deep mutational scanning, recent findings have found functional change to be position dependent, notwithstanding the chemical properties of mutant and mutated amino acids. This could indicate that structural properties of a given position are potentially responsible for the functional relevance of a mutation. Here, we looked at the relation between structure and function of positions using five proteins with experimental data of functional change available. In order to measure structural change, we modeled mutated proteins via amino-acid networks and quantified the perturbation of each mutation. We found that structural change is position dependent, and strongly related to functional change. Strong changes in protein structure correlate with functional loss, and positions with functional gain due to mutations tend to be structurally robust. Finally, we constructed a computational method to predict functionally sensitive positions to mutations using structural change that performs well on all five proteins with a mean precision of 74.7% and recall of 69.3% of all functional positions.

pyProGA-A PyMOL plugin for protein residue network analysis

The field of protein residue network (PRN) research has brought several useful methods and techniques for structural analysis of proteins and protein complexes. Many of these are ripe and ready to be used by the proteomics community outside of the PRN specialists. In this paper we present software which collects an ensemble of (network) methods tailored towards the analysis of protein-protein interactions (PPI) and/or interactions of proteins with ligands of other type, e.g. nucleic acids, oligosaccharides etc. In parallel, we propose the use of the network differential analysis as a method to identify residues mediating key interactions between proteins. We use a model system, to show that in combination with other, already published methods, also included in pyProGA, it can be used to make such predictions. Such extended repertoire of methods allows to cross-check predictions with other methods as well, as we show here. In addition, the possibility to construct PRN models from various kinds of input is so far a unique asset of our code. One can use structural data as defined in PDB files and/or from data on residue pair interaction energies, either from force-field parameters or fragment molecular orbital (FMO) calculations. pyProGA is a free open-source software available from https://gitlab.com/Vlado_S/pyproga.

Influence of Disease-Causing Mutations on Protein Structural Networks

The interactions between residues in a protein tertiary structure can be studied effectively using the approach of protein structure network (PSN). A PSN is a node-edge representation of the structure with nodes representing residues and interactions between residues represented by edges. In this study, we have employed weighted PSNs to understand the influence of disease-causing mutations on proteins of known 3D structures. We have used manually curated information on disease mutations from UniProtKB/Swiss-Prot and their corresponding protein structures of wildtype and disease variant from the protein data bank. The PSNs of the wildtype and disease-causing mutant are compared to analyse variation of global and local dissimilarity in the overall network and at specific sites. We study how a mutation at a given site can affect the structural network at a distant site which may be involved in the function of the protein. We have discussed specific examples of the disease cases where the protein structure undergoes limited structural divergence in their backbone but have large dissimilarity in their all atom networks and vice versa, wherein large conformational alterations are observed while retaining overall network. We analyse the effect of variation of network parameters that characterize alteration of function or stability.

p53 Is Potentially Regulated by Cyclophilin D in the Triple-Proline Loop of the DNA Binding Domain

The multifunctional protein p53 is the central molecular sensor of cellular stresses. The canonical function of p53 is to transcriptionally activate target genes in response to, for example, DNA damage that may trigger apoptosis. Recently, p53 was also found to play a role in the regulation of necrosis, another type of cell death featured by the mitochondrial permeability transition (mPT). In this process, p53 directly interacts with the mPT regulator cyclophilin D, the detailed mechanism of which however remains poorly understood. Here, we report a comprehensive computational investigation of the p53-cyclophilin D interaction using molecular dynamics simulations and associated analyses. We have identified the specific cyclophilin D binding site on p53 that is located at proline 151 in the DNA binding domain. As a peptidyl-prolyl isomerase, cyclophilin D binds p53 and catalyzes the cis-trans isomerization of the peptide bond preceding proline 151. We have also characterized the effect of such an isomerization and found that the p53 domain in the cis state is overall more rigid than the trans state except for the local region around proline 151. Dynamical changes upon isomerization occur in both local and distal regions, indicating an allosteric effect elicited by the isomerization. We present potential allosteric communication pathways between proline 151 and distal sites, including the DNA binding surface. Our work provides, for the first time, a model for how cyclophilin D binds p53 and regulates its activity by switching the configuration of a specific site.

Dissecting Dynamic and Hydration Contributions to Sequence-Dependent DNA Minor Groove Recognition

Sequence selectivity is a critical attribute of DNA-binding ligands and underlines the need for detailed molecular descriptions of binding in representative sequence contexts. We investigated the binding and volumetric properties of DB1976, a model bis(benzimidazole)-selenophene diamidine compound with emerging therapeutic potential in acute myeloid leukemia, debilitating fibroses, and obesity-related liver dysfunction. To sample the scope of cognate DB1976 target sites, we evaluated three dodecameric duplexes spanning >10³-fold in binding affinity. The attendant changes in partial molar volumes varied substantially, but not in step with binding affinity, suggesting distinct modes of interactions in these complexes. Specifically, whereas optimal binding was associated with loss of hydration water, low-affinity binding released more hydration water. Explicit-atom molecular dynamics simulations showed that minor groove binding perturbed the conformational dynamics and hydration at the termini and interior of the DNA in a sequence-dependent manner. The impact of these distinct local dynamics on hydration was experimentally validated by domain-specific interrogation of hydration with salt, which probed the charged axial surfaces of oligomeric DNA preferentially over the uncharged termini. Minor groove recognition by DB1976, therefore, generates dynamically distinct domains that can make favorable contributions to hydration release in both high- and low-affinity binding. Because ligand binding at internal sites of DNA oligomers modulates dynamics at the termini, the results suggest both short- and long-range dynamic effects along the DNA target that can influence their effectiveness as low-MW competitors of protein binding.

Insight into the structures of Interleukin-18 systems

Structure-based molecular designs play a critical role in the context of next generation drug development. Besides their fundamental scientific aspects, the findings established in this approach have significant implications in the expansions of target-based therapies and vaccines. Interleukin-18 (IL-18), also known as interferon gamma (IFN-γ) inducing factor, is a pro-inflammatory cytokine. The IL-18 binds first to the IL-18α receptor and forms a lower affinity complex. Upon binding with IL-18β a hetero-trimeric complex with higher affinity is formed that initiates the signal transduction process. The present study, including structural and molecular dynamics simulations, takes a close look at the structural stabilities of IL-18 and IL-18 receptor-bound ligand structures as functions of time. The results help to identify the conformational changes of the ligand due to receptor binding, as well as the structural orders of the apo and holo IL-18 protein complexes.

Large-Scale Structure-Based Prediction of Stable Peptide Binding to Class I HLAs Using Random Forests

Prediction of stable peptide binding to Class I HLAs is an important component for designing immunotherapies. While the best performing predictors are based on machine learning algorithms trained on peptide-HLA (pHLA) sequences, the use of structure for training predictors deserves further exploration. Given enough pHLA structures, a predictor based on the residue-residue interactions found in these structures has the potential to generalize for alleles with little or no experimental data. We have previously developed APE-Gen, a modeling approach able to produce pHLA structures in a scalable manner. In this work we use APE-Gen to model over 150,000 pHLA structures, the largest dataset of its kind, which were used to train a structure-based pan-allele model. We extract simple, homogenous features based on residue-residue distances between peptide and HLA, and build a random forest model for predicting stable pHLA binding. Our model achieves competitive AUROC values on leave-one-allele-out validation tests using significantly less data when compared to popular sequence-based methods. Additionally, our model offers an interpretation analysis that can reveal how the model composes the features to arrive at any given prediction. This interpretation analysis can be used to check if the model is in line with chemical intuition, and we showcase particular examples. Our work is a significant step toward using structure to achieve generalizable and more interpretable prediction for stable pHLA binding.

Analysis of allosteric communication in a multienzyme complex by ancestral sequence reconstruction

Tryptophan synthase (TS) is a heterotetrameric αββα complex. It is characterized by the channeling of the reaction intermediate indole and the mutual activation of the α-subunit TrpA and the β-subunit TrpB via a complex allosteric network. We have analyzed this allosteric network by means of ancestral sequence reconstruction (ASR), which is an in silico method to resurrect extinct ancestors of modern proteins. Previously, the sequences of TrpA and TrpB from the last bacterial common ancestor (LBCA) have been computed by means of ASR and characterized. LBCA-TS is similar to modern TS by forming a αββα complex with indole channeling taking place. However, LBCA-TrpA allosterically decreases the activity of LBCA-TrpB, whereas, for example, the modern ncTrpA from Neptuniibacter caesariensis allosterically increases the activity of ncTrpB. To identify amino acid residues that are responsible for this inversion of the allosteric effect, all 6 evolutionary TrpA and TrpB intermediates that stepwise link LBCA-TS with ncTS were characterized. Remarkably, the switching from TrpB inhibition to TrpB activation by TrpA occurred between 2 successive TS intermediates. Sequence comparison of these 2 intermediates and iterative rounds of site-directed mutagenesis allowed us to identify 4 of 413 residues from TrpB that are crucial for its allosteric activation by TrpA. The effect of our mutational studies was rationalized by a community analysis based on molecular dynamics simulations. Our findings demonstrate that ancestral sequence reconstruction can efficiently identify residues contributing to allosteric signal propagation in multienzyme complexes.

Detecting Functional Dynamics in Proteins with Comparative Perturbed-Ensembles Analysis

Recent advances have made all-atom molecular dynamics (MD) a powerful tool to sample the conformational energy landscape. There are still however three major challenges in the application of MD to biological systems: accuracy of force field, time scale, and the analysis of simulation trajectories. Significant progress in addressing the first two challenges has been made and extensively reviewed previously. This Account focuses on strategies of analyzing simulation data of biomolecules that also covers ways to properly design simulations and validate simulation results. In particular, we examine an approach named comparative perturbed-ensembles analysis, which we developed to efficiently detect dynamics in protein MD simulations that can be linked to biological functions. In our recent studies, we implemented this approach to understand allosteric regulations in several disease-associated human proteins. The central task of a comparative perturbed-ensembles analysis is to compare two or more conformational ensembles of a system generated by MD simulations under distinct perturbation conditions. Perturbations can be different sequence variations, ligand-binding conditions, and other physical/chemical modifications of the system. Each simulation is long enough (e.g., microsecond-long) to ensure sufficient sampling of the local substate. Then, sophisticated bioinformatic and statistical tools are applied to extract function-related information from the simulation data, including principal component analysis, residue-residue contact analysis, difference contact network analysis (dCNA) based on the graph theory, and statistical analysis of side-chain conformations. Computational findings are further validated with experimental data. By comparing distinct conformational ensembles, functional micro- to millisecond dynamics can be inferred. In contrast, such a time scale is difficult to reach in a single simulation; even when reached for a single condition of a system, it is elusive as to what dynamical motions are related to functions without, for example, comparing free and substrate-bound proteins at the minimum. We illustrate our approach with three examples. First, we discuss using the approach to identify allosteric pathways in cyclophilin A (CypA), a member of a ubiquitous class of peptidyl-prolyl cis-trans isomerase enzymes. By comparing side-chain torsion-angle distributions of CypA in wild-type and mutant forms, we identified three pathways: two are consistent with recent nuclear magnetic resonance experiments, whereas the third is a novel pathway. Second, we show how the approach enables a dynamical-evolution analysis of the human cyclophilin family. In the analysis, both conserved and divergent conformational dynamics across three cyclophilin isoforms (CypA, CypD, and CypE) were summarized. The conserved dynamics led to the discovery of allosteric networks resembling those found in CypA. A residue wise determinant underlying the unique dynamics in CypD was also detected and validated with additional mutational MD simulations. In the third example, we applied the approach to elucidate a peptide sequence-dependent allosteric mechanism in human Pin 1, a phosphorylation-dependent peptidyl-prolyl isomerase. We finally present our outlook of future directions. Especially, we envisage how the approach could help open a new avenue in drug discovery.