Distinguishing crystallographic from biological interfaces in protein complexes: role of intermolecular contacts and energetics for classification

Early Events in β2AR Dimer Dynamics Mediated by Activation-Related Microswitches

G-Protein-Coupled Receptors (GPCRs) make up around 3-4% of the human genome and are the targets of one-third of FDA-approved drugs. GPCRs typically exist as monomers but also aggregate to form higher-order oligomers, including dimers. β2AR, a pharmacologically relevant GPCR, is known to be targeted for the treatment of asthma and cardiovascular diseases. The activation of β2AR at the dimer level remains under-explored. In the current study, molecular dynamics (MD) simulations have been performed to understand activation-related structural changes in β2AR at the dimer level. The transition from inactive to active and vice versa has been studied by starting the simulations in the apo, agonist-bound, and inverse agonist-bound β2AR dimers for PDB ID: 2RH1 and PDB ID: 3P0G, respectively. A cumulative total of around 21-μs simulations were performed. Residue-based distances, RMSD, and PCA calculations suggested that either of the one monomer attained activation-related features for the apo and agonist-bound β2AR dimers. The TM5 and TM6 helices within the two monomers were observed to be in significant variation in all the simulations. TM5 bulge and proximity of TM2 and TM7 helices may be contributing to one of the early events in activation. The dimeric interface between TM1 and helix 8 were observed to be well maintained in the apo and agonist-bound simulations. The presence of inverse agonists favored inactive features in both the monomers. These key features of activation known for monomers were observed to have an impact on β2AR dimers, thereby providing an insight into the oligomerization mechanism of GPCRs.

Residue-Specific Epitope Mapping of the PD-1/Nivolumab Interaction Using X-ray Footprinting Mass Spectrometry

X-ray footprinting coupled with mass spectrometry (XFMS) presents a novel approach in structural biology, offering insights into protein conformation and dynamics in the solution state. The interaction of the cancer-immunotherapy monoclonal antibody nivolumab with its antigen target PD-1 was used to showcase the utility of XFMS against the previously published crystal structure of the complex. Changes in side-chain solvent accessibility, as determined by the oxidative footprint of free PD-1 versus PD-1 bound to nivolumab, agree with the binding interface side-chain interactions reported from the crystal structure of the complex. The N-linked glycosylation sites of PD-1 were confirmed through an LC-MS/MS-based deglycosylation analysis of asparagine deamidation. In addition, subtle changes in side-chain solvent accessibility were observed in the C'D loop region of PD-1 upon complex formation with nivolumab.

The oligomeric states of dye-decolorizing peroxidases from Streptomyces lividans and their implications for mechanism of substrate oxidation

A common evolutionary mechanism in biology to drive function is protein oligomerization. In prokaryotes, the symmetrical assembly of repeating protein units to form homomers is widespread, yet consideration in vitro of whether such assemblies have functional or mechanistic consequences is often overlooked. Dye-decolorizing peroxidases (DyPs) are one such example, where their dimeric α + β barrel units can form various oligomeric states, but the oligomer influence, if any, on mechanism and function has received little attention. In this work, we have explored the oligomeric state of three DyPs found in Streptomyces lividans, each with very different mechanistic behaviors in their reactions with hydrogen peroxide and organic substrates. Using analytical ultracentrifugation, we reveal that except for one of the A-type DyPs where only a single sedimenting species is detected, oligomer states ranging from homodimers to dodecamers are prevalent in solution. Using cryo-EM on preparations of the B-type DyP, we determined a 3.02 Å resolution structure of a hexamer assembly that corresponds to the dominant oligomeric state in solution as determined by analytical ultracentrifugation. Furthermore, cryo-EM data detected sub-populations of higher-order oligomers, with one of these formed by an arrangement of two B-type DyP hexamers to give a dodecamer assembly. Our solution and structural insights of these oligomer states provide a new framework to consider previous mechanistic studies of these DyP members and are discussed in terms of long-range electron transfer for substrate oxidation and in the "storage" of oxidizable equivalents on the heme until a two-electron donor is available.

EGGNet, a Generalizable Geometric Deep Learning Framework for Protein Complex Pose Scoring

Computational prediction of molecule-protein interactions has been key for developing new molecules to interact with a target protein for therapeutics development. Previous work includes two independent streams of approaches: (1) predicting protein-protein interactions (PPIs) between naturally occurring proteins and (2) predicting binding affinities between proteins and small-molecule ligands [also known as drug-target interaction (DTI)]. Studying the two problems in isolation has limited the ability of these computational models to generalize across the PPI and DTI tasks, both of which ultimately involve noncovalent interactions with a protein target. In this work, we developed Equivariant Graph of Graphs neural Network (EGGNet), a geometric deep learning (GDL) framework, for molecule-protein binding predictions that can handle three types of molecules for interacting with a target protein: (1) small molecules, (2) synthetic peptides, and (3) natural proteins. EGGNet leverages a graph of graphs (GoG) representation constructed from the molecular structures at atomic resolution and utilizes a multiresolution equivariant graph neural network to learn from such representations. In addition, EGGNet leverages the underlying biophysics and makes use of both atom- and residue-level interactions, which improve EGGNet's ability to rank candidate poses from blind docking. EGGNet achieves competitive performance on both a public protein-small-molecule binding affinity prediction task (80.2% top 1 success rate on CASF-2016) and a synthetic protein interface prediction task (88.4% area under the precision-recall curve). We envision that the proposed GDL framework can generalize to many other protein interaction prediction problems, such as binding site prediction and molecular docking, helping accelerate protein engineering and structure-based drug development.

Using a novel structure/function approach to select diverse swine major histocompatibility complex 1 alleles to predict epitopes for vaccine development

MOTIVATION

Swine leukocyte antigens (SLAs) (i.e. swine major histocompatibility complex proteins) conduct a fundamental role in swine immunity. To generate a protective vaccine across an outbred species, such as pigs, it is critical that epitopes that bind to diverse SLA alleles are used in the vaccine development process. We introduced a new strategy for epitope prediction.

RESULTS

We employed molecular dynamics simulation to identify key amino acids for interactions with epitopes. We developed an algorithm wherein each SLA-1 is compared to a crystalized reference allele with unique weighting for non-conserved amino acids based on R group and position. We then performed homology modeling and electrostatic contact mapping to visualize how relatively small changes in sequences impacted the charge distribution in the binding site. We selected eight diverse SLA-1 alleles and performed homology modeling followed, by protein-peptide docking and binding affinity analyses, to identify porcine reproductive and respiratory syndrome virus matrix protein epitopes that bind with high affinity to these alleles. We also performed docking analysis on the epitopes identified as strong binders using NetMHCpan 4.1. Epitopes predicted to bind to our eight SLA-1 alleles had equivalent or higher energetic interactions than those predicted to bind to the NetMHCpan 4.1 allele repertoire. This approach of selecting diverse SLA-1 alleles, followed by homology modeling, and docking simulations, can be used as a novel strategy for epitope prediction that complements other available tools and is especially useful when available tools do not offer a prediction for SLAs/major histocompatibility complex.

AVAILABILITY AND IMPLEMENTATION

The data underlying this article are available in the online Supplementary Material.

The World of GPCR dimers - Mapping dopamine receptor D2 homodimers in different activation states and configuration arrangements

G protein-coupled receptors (GPCRs) are known to dimerize, but the molecular and structural basis of GPCR dimers is not well understood. In this study, we developed a computational framework to generate models of symmetric and asymmetric GPCR dimers using different monomer activation states and identified their most likely interfaces with molecular details. We chose the dopamine receptor D2 (D2R) homodimer as a case study because of its biological relevance and the availability of structural information. Our results showed that transmembrane domains 4 and 5 (TM4 and TM5) are mostly found at the dimer interface of the D2R dimer and that these interfaces have a subset of key residues that are mostly nonpolar from TM4 and TM5, which was in line with experimental studies. In addition, TM2 and TM3 appear to be relevant for D2R dimers. In some cases, the inactive configuration is unaffected by the partnered protomer, whereas in others, the active protomer adopts the properties of an inactive receptor. Additionally, the β-arrestin configuration displayed the properties of an active receptor in the absence of an agonist, suggesting that a switch to another meta-state during dimerization occurred. Our findings are consistent with the experimental data, and this method can be adapted to study heterodimers and potentially extended to include additional proteins such as G proteins or β-arrestins. In summary, this approach provides insight into the impact of the conformational status of partnered protomers on the overall quaternary GPCR macromolecular structure and dynamics.

Strong non-ideality effects at low protein concentrations: considerations for elongated proteins

A recent investigation was aimed at obtaining structural information on a highly extended protein via SEC-MALS-SAXS. Significantly broadened elution peaks were observed, reminiscent of a phenomenon known as viscous fingering. This phenomenon is usually observed above 50 mg/mL for proteins like bovine serum albumin (BSA). Interestingly, the highly extended protein (Brpt5.5) showed viscous fingering at concentrations lower than 5 mg/mL. The current study explores this and other non-ideal behavior, emphasizing the presence of these effects at relatively low concentrations for extended proteins. BSA, Brpt5.5, and a truncated form of Brpt5.5 referred to as Brpt1.5 are studied systematically using size-exclusion chromatography (SEC), sedimentation velocity analytical ultracentrifugation (AUC), and viscosity. The viscous fingering effect is quantified using two approaches and is found to correlate well with the intrinsic viscosity of the proteins-Brpt5.5 exhibits the most severe effect and is the most extended protein tested in the study. By AUC, the hydrodynamic non-ideality was measured for each protein via global analysis of a concentration series. Compared to BSA, both Brpt1.5 and Brpt5.5 showed significant non-ideality that could be easily visualized at concentrations at or below 5 mg/mL and 1 mg/mL, respectively. A variety of relationships were examined for their ability to differentiate the proteins by shape using information from AUC and/or viscosity. Furthermore, these relationships were also tested in the context of hydrodynamic modeling. The importance of considering non-ideality when investigating the structure of extended macromolecules is discussed.

The protein common assembly database (ProtCAD)-a comprehensive structural resource of protein complexes

Proteins often act through oligomeric interactions with other proteins. X-ray crystallography and cryo-electron microscopy provide detailed information on the structures of biological assemblies, defined as the most likely biologically relevant structures derived from experimental data. In crystal structures, the most relevant assembly may be ambiguously determined, since multiple assemblies observed in the crystal lattice may be plausible. It is estimated that 10-15% of PDB entries may have incorrect or ambiguous assembly annotations. Accurate assemblies are required for understanding functional data and training of deep learning methods for predicting assembly structures. As with any other kind of biological data, replication via multiple independent experiments provides important validation for the determination of biological assembly structures. Here we present the Protein Common Assembly Database (ProtCAD), which presents clusters of protein assembly structures observed in independent structure determinations of homologous proteins in the Protein Data Bank (PDB). ProtCAD is searchable by PDB entry, UniProt identifiers, or Pfam domain designations and provides downloads of coordinate files, PyMol scripts, and publicly available assembly annotations for each cluster of assemblies. About 60% of PDB entries contain assemblies in clusters of at least 2 independent experiments. All clusters and coordinates are available on ProtCAD web site (http://dunbrack2.fccc.edu/protcad).

Antibody Interfaces Revealed Through Structural Mining

Antibodies are fundamental effectors of humoral immunity, and have become a highly successful class of therapeutics. There is increasing evidence that antibodies utilize transient homotypic interactions to enhance function, and elucidation of such interactions can provide insights into their biology and new opportunities for their optimization as drugs. Yet the transitory nature of weak interactions makes them difficult to investigate. Capitalizing on their rich structural data and high conservation, we have characterized all the ways that antibody fragment antigen-binding (Fab) regions interact crystallographically. This approach led to the discovery of previously unrealized interfaces between antibodies. While diverse interactions exist, β-sheet dimers and variable-constant elbow dimers are recurrent motifs. Disulfide engineering enabled interactions to be trapped and investigated structurally and functionally, providing experimental validation of the interfaces and illustrating their potential for optimization. This work provides first insight into previously undiscovered oligomeric interactions between antibodies, and enables new opportunities for their biotherapeutic optimization.

AlphaFold accurately predicts distinct conformations based on the oligomeric state of a de novo designed protein

Using the molecular modeling program Rosetta, we designed a de novo protein, called SEWN0.1, which binds the heterotrimeric G protein Gαq. The design is helical, well-folded, and primarily monomeric in solution at a concentration of 10 μM. However, when we solved the crystal structure of SEWN0.1 at 1.9 Å, we observed a dimer in a conformation incompatible with binding Gαq . Unintentionally, we had designed a protein that adopts alternate conformations depending on its oligomeric state. Recently, there has been tremendous progress in the field of protein structure prediction as new methods in artificial intelligence have been used to predict structures with high accuracy. We were curious if the structure prediction method AlphaFold could predict the structure of SEWN0.1 and if the prediction depended on oligomeric state. When AlphaFold was used to predict the structure of monomeric SEWN0.1, it produced a model that resembles the Rosetta design model and is compatible with binding Gαq , but when used to predict the structure of a dimer, it predicted a conformation that closely resembles the SEWN0.1 crystal structure. AlphaFold's ability to predict multiple conformations for a single protein sequence should be useful for engineering protein switches.

DeepRank: a deep learning framework for data mining 3D protein-protein interfaces

Three-dimensional (3D) structures of protein complexes provide fundamental information to decipher biological processes at the molecular scale. The vast amount of experimentally and computationally resolved protein-protein interfaces (PPIs) offers the possibility of training deep learning models to aid the predictions of their biological relevance. We present here DeepRank, a general, configurable deep learning framework for data mining PPIs using 3D convolutional neural networks (CNNs). DeepRank maps features of PPIs onto 3D grids and trains a user-specified CNN on these 3D grids. DeepRank allows for efficient training of 3D CNNs with data sets containing millions of PPIs and supports both classification and regression. We demonstrate the performance of DeepRank on two distinct challenges: The classification of biological versus crystallographic PPIs, and the ranking of docking models. For both problems DeepRank is competitive with, or outperforms, state-of-the-art methods, demonstrating the versatility of the framework for research in structural biology.

Getting to know each other: PPIMem, a novel approach for predicting transmembrane protein-protein complexes

Because of their considerable number and diversity, membrane proteins and their macromolecular complexes represent the functional units of cells. Their quaternary structure may be stabilized by interactions between the α-helices of different proteins in the hydrophobic region of the cell membrane. Membrane proteins equally represent potential pharmacological targets par excellence for various diseases. Unfortunately, their experimental 3D structure and that of their complexes with other intramembrane protein partners are scarce due to technical difficulties. To overcome this key problem, we devised PPIMem, a computational approach for the specific prediction of higher-order structures of α-helical transmembrane proteins. The novel approach involves proper identification of the amino acid residues at the interface of molecular complexes with a 3D structure. The identified residues compose then nonlinear interaction motifs that are conveniently expressed as mathematical regular expressions. These are efficiently implemented for motif search in amino acid sequence databases, and for the accurate prediction of intramembrane protein-protein complexes. Our template interface-based approach predicted 21,544 binary complexes between 1,504 eukaryotic plasma membrane proteins across 39 species. We compare our predictions to experimental datasets of protein-protein interactions as a first validation method. The online database that results from the PPIMem algorithm with the annotated predicted interactions are implemented as a web server and can be accessed directly at https://transint.univ-evry.fr.

Modeling and Structure Determination of Homo-Oligomeric Proteins: An Overview of Challenges and Current Approaches

Protein homo-oligomerization is a very common phenomenon, and approximately half of proteins form homo-oligomeric assemblies composed of identical subunits. The vast majority of such assemblies possess internal symmetry which can be either exploited to help or poses challenges during structure determination. Moreover, aspects of symmetry are critical in the modeling of protein homo-oligomers either by docking or by homology-based approaches. Here, we first provide a brief overview of the nature of protein homo-oligomerization. Next, we describe how the symmetry of homo-oligomers is addressed by crystallographic and non-crystallographic symmetry operations, and how biologically relevant intermolecular interactions can be deciphered from the ordered array of molecules within protein crystals. Additionally, we describe the most important aspects of protein homo-oligomerization in structure determination by NMR. Finally, we give an overview of approaches aimed at modeling homo-oligomers using computational methods that specifically address their internal symmetry and allow the incorporation of other experimental data as spatial restraints to achieve higher model reliability.

Structural Biology in the Clouds: The WeNMR-EOSC Ecosystem

Structural biology aims at characterizing the structural and dynamic properties of biological macromolecules at atomic details. Gaining insight into three dimensional structures of biomolecules and their interactions is critical for understanding the vast majority of cellular processes, with direct applications in health and food sciences. Since 2010, the WeNMR project (www.wenmr.eu) has implemented numerous web-based services to facilitate the use of advanced computational tools by researchers in the field, using the high throughput computing infrastructure provided by EGI. These services have been further developed in subsequent initiatives under H2020 projects and are now operating as Thematic Services in the European Open Science Cloud portal (www.eosc-portal.eu), sending >12 millions of jobs and using around 4,000 CPU-years per year. Here we review 10 years of successful e-infrastructure solutions serving a large worldwide community of over 23,000 users to date, providing them with user-friendly, web-based solutions that run complex workflows in structural biology. The current set of active WeNMR portals are described, together with the complex backend machinery that allows distributed computing resources to be harvested efficiently.

DeepFrag: a deep convolutional neural network for fragment-based lead optimization

Machine learning has been increasingly applied to the field of computer-aided drug discovery in recent years, leading to notable advances in binding-affinity prediction, virtual screening, and QSAR. Surprisingly, it is less often applied to lead optimization, the process of identifying chemical fragments that might be added to a known ligand to improve its binding affinity. We here describe a deep convolutional neural network that predicts appropriate fragments given the structure of a receptor/ligand complex. In an independent benchmark of known ligands with missing (deleted) fragments, our DeepFrag model selected the known (correct) fragment from a set over 6500 about 58% of the time. Even when the known/correct fragment was not selected, the top fragment was often chemically similar and may well represent a valid substitution. We release our trained DeepFrag model and associated software under the terms of the Apache License, Version 2.0.

PRODIGY-crystal: a web-tool for classification of biological interfaces in protein complexes

Summary

Distinguishing biologically relevant interfaces from crystallographic ones in biological complexes is fundamental in order to associate cellular functions to the correct macromolecular assemblies. Recently, we described a detailed study reporting the differences in the type of intermolecular residue–residue contacts between biological and crystallographic interfaces. Our findings allowed us to develop a fast predictor of biological interfaces reaching an accuracy of 0.92 and competitive to the current state of the art. Here we present its web-server implementation, PRODIGY-CRYSTAL, aimed at the classification of biological and crystallographic interfaces. PRODIGY-CRYSTAL has the advantage of being fast, accurate and simple. This, together with its user-friendly interface and user support forum, ensures its broad accessibility.

Availability and implementation

PRODIGY-CRYSTAL is freely available without registration requirements at https://haddock.science.uu.nl/services/PRODIGY-CRYSTAL.

Biological vs. Crystallographic Protein Interfaces: An Overview of Computational Approaches for Their Classification

Complexes between proteins are at the basis of almost every process in cells. Their study, from a structural perspective, has a pivotal role in understanding biological functions and, importantly, in drug development. X-ray crystallography represents the broadest source for the experimental structural characterization of protein-protein complexes. Correctly identifying the biologically relevant interface from the crystallographic ones is, however, not trivial and can be prone to errors. Over the past two decades, computational methodologies have been developed to study the differences of those interfaces and automatically classify them as biological or crystallographic. Overall, protein-protein interfaces show differences in terms of composition, energetics and evolutionary conservation between biological and crystallographic ones. Based on those observations, a number of computational methods have been developed for this classification problem, which can be grouped into three main categories: Energy-, empirical knowledge- and machine learning-based approaches. In this review, we give a comprehensive overview of the training datasets and methods so far implemented, providing useful links and a brief description of each method.

Accurate Classification of Biological and non-Biological Interfaces in Protein Crystal Structures using Subtle Covariation Signals

Proteins often work as oligomers or multimers in vivo. Therefore, elucidating their oligomeric or multimeric form (quaternary structure) is crucially important to ascertain their function. X-ray crystal structures of numerous proteins have been accumulated, providing information related to their biological units. Extracting information of biological units from protein crystal structures represents a meaningful task for modern biology. Nevertheless, although many methods have been proposed for identifying biological units appearing in protein crystal structures, it is difficult to distinguish biological protein-protein interfaces from crystallographic ones. Therefore, our simple but highly accurate classifier was developed to infer biological units in protein crystal structures using large amounts of protein sequence information and a modern contact prediction method to exploit covariation signals (CSs) in proteins. We demonstrate that our proposed method is promising even for weak signals of biological interfaces. We also discuss the relation between classification accuracy and conservation of biological units, and illustrate how the selection of sequences included in multiple sequence alignments as sources for obtaining CSs affects the results. With increased amounts of sequence data, the proposed method is expected to become increasingly useful.