Machine learning designs new GCGR/GLP-1R dual agonists with enhanced biological potency.

Several peptide dual agonists of the human glucagon receptor (GCGR) and the glucagon-like peptide-1 receptor (GLP-1R) are in development for the treatment of type 2 diabetes, obesity and their associated complications. Candidates must have high potency at both receptors, but it is unclear whether the limited experimental data available can be used to train models that accurately predict the activity at both receptors of new peptide variants. Here we use peptide sequence data labelled with in vitro potency at human GCGR and GLP-1R to train several models, including a deep multi-task neural-network model using multiple loss optimization. Model-guided sequence optimization was used to design three groups of peptide variants, with distinct ranges of predicted dual activity. We found that three of the model-designed sequences are potent dual agonists with superior biological activity. With our designs we were able to achieve up to sevenfold potency improvement at both receptors simultaneously compared to the best dual-agonist in the training set.

DLAB: deep learning methods for structure-based virtual screening of antibodies.

MOTIVATION: Antibodies are one of the most important classes of pharmaceuticals, with over 80 approved molecules currently in use against a wide variety of diseases. The drug discovery process for antibody therapeutic candidates however is time- and cost-intensive and heavily reliant on in vivo and in vitro high throughput screens. Here, we introduce a framework for structure-based deep learning for antibodies (DLAB) which can virtually screen putative binding antibodies against antigen targets of interest. DLAB is built to be able to predict antibody-antigen binding for antigens with no known antibody binders. RESULTS: We demonstrate that DLAB can be used both to improve antibody-antigen docking and structure-based virtual screening of antibody drug candidates. DLAB enables improved pose ranking for antibody docking experiments as well as selection of antibody-antigen pairings for which accurate poses are generated and correctly ranked. We also show that DLAB can identify binding antibodies against specific antigens in a case study. Our results demonstrate the promise of deep learning methods for structure-based virtual screening of antibodies. AVAILABILITY AND IMPLEMENTATION: The DLAB source code and pre-trained models are available at https://github.com/oxpig/dlab-public. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Bios2cor: an R package integrating dynamic and evolutionary correlations to identify functionally important residues in proteins.

SUMMARY: Both dynamic correlations in protein sidechain motions during molecular dynamics (MD) simulations and evolutionary correlations in multiple sequence alignments (MSAs) of homologous proteins may reveal functionally important residues. We developed the R package Bios2cor that provides a unique framework to investigate and, possibly, integrate both analyses. Bios2cor starts with an MSA or an MD trajectory and computes correlation/covariation scores between positions in the MSA or between sidechain dihedral angles or rotamers in the MD trajectory. In addition, Bios2cor provides a variety of tools for the analysis, the visualization and the interpretation of the data. AVAILABILITY AND IMPLEMENTATION: The R package Bios2cor is available from the Comprehensive R Archive Network, at https://CRAN.R-project.org/package=Bios2cor.

Ab-Ligity: identifying sequence-dissimilar antibodies that bind to the same epitope.

Solving the structure of an antibody-antigen complex gives atomic level information of the interactions between an antibody and its antigen, but such structures are expensive and hard to obtain. Alternative experimental sources include epitope mapping and binning experiments, which can be used as a surrogate to identify key interacting residues. However, their resolution is usually not sufficient to identify if two antibodies have identical interactions. Computational approaches to this problem have so far been based on the premise that antibodies with similar sequences behave similarly. Such approaches will fail to identify sequence-distant antibodies that target the same epitope. Here, we present Ab-Ligity, a structure-based similarity measure tailored to antibody-antigen interfaces. Using predicted paratopes on model antibody structures, we assessed its ability to identify those antibodies that target highly similar epitopes. Most antibodies adopting similar binding modes can be identified from sequence similarity alone, using methods such as clonotyping. In the challenging subset of antibodies whose sequences differ significantly, Ab-Ligity is still able to predict antibodies that would bind to highly similar epitopes (precision of 0.95 and recall of 0.69). We compared Ab-Ligity's performance to an existing tool for comparing general protein interfaces, InterComp, and showed improved performance on antibody cases achieved in a substantially reduced time. These results suggest that Ab-Ligity will allow the identification of diverse (sequence-dissimilar) antibodies that bind to the same epitopes from large datasets such as immune repertoires. The tool is available at http://opig.stats.ox.ac.uk/resources.

Deciphering collaborative sidechain motions in proteins during molecular dynamics simulations.

The dynamic structure of proteins is essential for their functions and may include large conformational transitions which can be studied by molecular dynamics (MD) simulations. However, details of these transitions are difficult to automatically track. To facilitate their analysis, we developed two scores of correlation between sidechain dihedral angles. The CIRCULAR and OMES scores are computed from, respectively, dihedral angle values and rotamer distributions. As a case study, we applied our methods to an activation-like transition of the chemokine receptor CXCR4, observed during accelerated MD simulations. The principal component analysis of the correlation matrices was consistent with the networking structure of the top ranking pairs. Both scores identify a set of residues whose "collaborative" sidechain rotamerization immediately preceded or accompanied the conformational transition of CXCR4. Detailed analysis of the sequential order of these rotamerizations suggests that an allosteric mechanism, involving the outward motion of an asparagine residue in transmembrane helix 3, might be a prerequisite to the large scale conformational transition of CXCR4. This case study provides the proof-of-concept that the correlation methods developed here are valuable exploratory techniques to help decipher complex reactional pathways.

Thera-SAbDab: the Therapeutic Structural Antibody Database.

The Therapeutic Structural Antibody Database (Thera-SAbDab; http://opig.stats.ox.ac.uk/webapps/therasabdab) tracks all antibody- and nanobody-related therapeutics recognized by the World Health Organisation (WHO), and identifies any corresponding structures in the Structural Antibody Database (SAbDab) with near-exact or exact variable domain sequence matches. Thera-SAbDab is synchronized with SAbDab to update weekly, reflecting new Protein Data Bank entries and the availability of new sequence data published by the WHO. Each therapeutic summary page lists structural coverage (with links to the appropriate SAbDab entries), alignments showing where any near-matches deviate in sequence, and accompanying metadata, such as intended target and investigated conditions. Thera-SAbDab can be queried by therapeutic name, by a combination of metadata, or by variable domain sequence - returning all therapeutics that are within a specified sequence identity over a specified region of the query. The sequences of all therapeutics listed in Thera-SAbDab (461 unique molecules, as of 5 August 2019) are downloadable as a single file with accompanying metadata.

Antibodies binding the head domain of P2X4 inhibit channel function and reverse neuropathic pain.

P2X4 is a ligand-gated ion channel implicated in neuropathic pain. Drug discovery efforts targeting P2X4 have been unsuccessful largely because of the difficulty in engineering specificity and selectivity. Here, we describe for the first time the generation of a panel of diverse monoclonal antibodies (mAbs) to human and mouse P2X4, capable of both positive and negative modulation of channel function. The affinity-optimised anti-P2X4 mAb IgG#151-LO showed exquisite selectivity for human P2X4 and induced potent and complete block of P2X4 currents. Site-directed mutagenesis of P2X4 revealed the head domain as a key interaction site for inhibitory mAbs. Inhibition of spinal P2X4 either by intrathecal delivery of an anti-P2X4 mAb or by systemic delivery of an anti-P2X4 bispecific mAb with enhanced blood-spinal cord barrier permeability produced long-lasting (>7 days) analgesia in a mouse model of neuropathic pain. We therefore propose that inhibitory mAbs binding the head domain of P2X4 have therapeutic potential for the treatment of neuropathic pain.

Five computational developability guidelines for therapeutic antibody profiling.

Therapeutic mAbs must not only bind to their target but must also be free from "developability issues" such as poor stability or high levels of aggregation. While small-molecule drug discovery benefits from Lipinski's rule of five to guide the selection of molecules with appropriate biophysical properties, there is currently no in silico analog for antibody design. Here, we model the variable domain structures of a large set of post-phase-I clinical-stage antibody therapeutics (CSTs) and calculate in silico metrics to estimate their typical properties. In each case, we contextualize the CST distribution against a snapshot of the human antibody gene repertoire. We describe guideline values for five metrics thought to be implicated in poor developability: the total length of the complementarity-determining regions (CDRs), the extent and magnitude of surface hydrophobicity, positive charge and negative charge in the CDRs, and asymmetry in the net heavy- and light-chain surface charges. The guideline cutoffs for each property were derived from the values seen in CSTs, and a flagging system is proposed to identify nonconforming candidates. On two mAb drug discovery sets, we were able to selectively highlight sequences with developability issues. We make available the Therapeutic Antibody Profiler (TAP), a computational tool that builds downloadable homology models of variable domain sequences, tests them against our five developability guidelines, and reports potential sequence liabilities and canonical forms. TAP is freely available at opig.stats.ox.ac.uk/webapps/sabdab-sabpred/TAP.php.

SCALOP: sequence-based antibody canonical loop structure annotation.

MOTIVATION: Canonical forms of the antibody complementarity-determining regions (CDRs) were first described in 1987 and have been redefined on multiple occasions since. The canonical forms are often used to approximate the antibody binding site shape as they can be predicted from sequence. A rapid predictor would facilitate the annotation of CDR structures in the large amounts of repertoire data now becoming available from next generation sequencing experiments. RESULTS: SCALOP annotates CDR canonical forms for antibody sequences, supported by an auto-updating database to capture the latest cluster information. Its accuracy is comparable to that of a standard structural predictor but it is 800 times faster. The auto-updating nature of SCALOP ensures that it always attains the best possible coverage. AVAILABILITY AND IMPLEMENTATION: SCALOP is available as a web application and for download under a GPLv3 license at opig.stats.ox.ac.uk/webapps/scalop. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Evolution of chemokine receptors is driven by mutations in the sodium binding site.

Chemokines and their receptors (members of the GPCR super-family) are involved in a wide variety of physiological processes and diseases; thus, understanding the specificity of the chemokine receptor family could help develop new receptor specific drugs. Here, we explore the evolutionary mechanisms that led to the emergence of the chemokine receptors. Based on GPCR hierarchical classification, we analyzed nested GPCR sets with an eigen decomposition approach of the sequence covariation matrix and determined three key residues whose mutation was crucial for the emergence of the chemokine receptors and their subsequent divergence into homeostatic and inflammatory receptors. These residues are part of the allosteric sodium binding site. Their structural and functional roles were investigated by molecular dynamics simulations of CXCR4 and CCR5 as prototypes of homeostatic and inflammatory chemokine receptors, respectively. This study indicates that the three mutations crucial for the evolution of the chemokine receptors dramatically altered the sodium binding mode. In CXCR4, the sodium ion is tightly bound by four protein atoms and one water molecule. In CCR5, the sodium ion is mobile within the binding pocket and moves between different sites involving from one to three protein atoms and two to five water molecules. Analysis of chemokine receptor evolution reveals that a highly constrained sodium binding site characterized most ancient receptors, and that the constraints were subsequently loosened during the divergence of this receptor family. We discuss the implications of these findings for the evolution of the chemokine receptor functions and mechanisms of action.

Molecular Insights into the Transmembrane Domain of the Thyrotropin Receptor.

The thyrotropin receptor (TSHR) is a G protein-coupled receptor (GPCR) that is member of the leucine-rich repeat subfamily (LGR). In the absence of crystal structure, the success of rational design of ligands targeting the receptor internal cavity depends on the quality of the TSHR models built. In this subfamily, transmembrane helices (TM) 2 and 5 are characterized by the absence of proline compared to most receptors, raising the question of the structural conformation of these helices. To gain insight into the structural properties of these helices, we carried out bioinformatics and experimental studies. Evolutionary analysis of the LGR family revealed a deletion in TM5 but provided no information on TM2. Wild type residues at positions 2.58, 2.59 or 2.60 in TM2 and/or at position 5.50 in TM5 were substituted to proline. Depending on the position of the proline substitution, different effects were observed on membrane expression, glycosylation, constitutive cAMP activity and responses to thyrotropin. Only proline substitution at position 2.59 maintained complex glycosylation and high membrane expression, supporting occurrence of a bulged TM2. The TSHR transmembrane domain was modeled by homology with the orexin 2 receptor, using a protocol that forced the deletion of one residue in the TM5 bulge of the template. The stability of the model was assessed by molecular dynamics simulations. TM5 straightened during the equilibration phase and was stable for the remainder of the simulations. Our data support a structural model of the TSHR transmembrane domain with a bulged TM2 and a straight TM5 that is specific of glycoprotein hormone receptors.

Do plants contain g protein-coupled receptors?

Whether G protein-coupled receptors (GPCRs) exist in plants is a fundamental biological question. Interest in deorphanizing new GPCRs arises because of their importance in signaling. Within plants, this is controversial, as genome analysis has identified 56 putative GPCRs, including G protein-coupled receptor1 (GCR1), which is reportedly a remote homolog to class A, B, and E GPCRs. Of these, GCR2 is not a GPCR; more recently, it has been proposed that none are, not even GCR1. We have addressed this disparity between genome analysis and biological evidence through a structural bioinformatics study, involving fold recognition methods, from which only GCR1 emerges as a strong candidate. To further probe GCR1, we have developed a novel helix-alignment method, which has been benchmarked against the class A-class B-class F GPCR alignments. In addition, we have presented a mutually consistent set of alignments of GCR1 homologs to class A, class B, and class F GPCRs and shown that GCR1 is closer to class A and/or class B GPCRs than class A, class B, or class F GPCRs are to each other. To further probe GCR1, we have aligned transmembrane helix 3 of GCR1 to each of the six GPCR classes. Variability comparisons provide additional evidence that GCR1 homologs have the GPCR fold. From the alignments and a GCR1 comparative model, we have identified motifs that are common to GCR1, class A, B, and E GPCRs. We discuss the possibilities that emerge from this controversial evidence that GCR1 has a GPCR fold.

Identifying subset errors in multiple sequence alignments.

Multiple sequence alignment (MSA) accuracy is important, but there is no widely accepted method of judging the accuracy that different alignment algorithms give. We present a simple approach to detecting two types of error, namely block shifts and the misplacement of residues within a gap. Given a MSA, subsets of very similar sequences are generated through the use of a redundancy filter, typically using a 70-90% sequence identity cut-off. Subsets thus produced are typically small and degenerate, and errors can be easily detected even by manual examination. The errors, albeit minor, are inevitably associated with gaps in the alignment, and so the procedure is particularly relevant to homology modelling of protein loop regions. The usefulness of the approach is illustrated in the context of the universal but little known [K/R]KLH motif that occurs in intracellular loop 1 of G protein coupled receptors (GPCR); other issues relevant to GPCR modelling are also discussed.

The role of ECL2 in CGRP receptor activation: a combined modelling and experimental approach.

The calcitonin gene-related peptide (CGRP) receptor is a complex of a calcitonin receptor-like receptor (CLR), which is a family B G-protein-coupled receptor (GPCR) and receptor activity modifying protein 1. The role of the second extracellular loop (ECL2) of CLR in binding CGRP and coupling to Gs was investigated using a combination of mutagenesis and modelling. An alanine scan of residues 271-294 of CLR showed that the ability of CGRP to produce cAMP was impaired by point mutations at 13 residues; most of these also impaired the response to adrenomedullin (AM). These data were used to select probable ECL2-modelled conformations that are involved in agonist binding, allowing the identification of the likely contacts between the peptide and receptor. The implications of the most likely structures for receptor activation are discussed.

Modeling active GPCR conformations.

The most significant advance in modeling GPCR active states has been the ß(2)-adrenergic receptor-Gs complex as this essentially transforms active-state modeling into homology modeling. Various different molecular dynamics-based approaches for modeling active states are presented, and a number of key applications discussed. These simulations have given insights into the activation pathway, conformational changes, dimerization, hydration, the ionic lock, ligand binding, protonation, and sodium binding. Crystallography and simulations have shown that the presence of agonist alone is unlikely to be sufficient to form the active state and that restraints applied to the G protein-binding region are required. The role of various microswitches in activation is discussed, including the controversial rotamer toggle switch. The importance of explicitly simulating experimental molecular probes to understand activation is highlighted, along with the need to ensure that such molecules are well parameterized. Approaches to loop modeling are discussed. We argue that the role of successful virtual screening against active models should not be overestimated as the main conformational changes on activation occur in the intracellular region.

The activation of the CGRP receptor.

The CGRP (calcitonin gene-related peptide) receptor is a family B GPCR (G-protein-coupled receptor). It consists of a GPCR, CLR (calcitonin receptor-like receptor) and an accessory protein, RAMP1 (receptor activity modifying protein 1). RAMP1 is needed for CGRP binding and also cell-surface expression of CLR. CLR is an example of a family B GPCR. Unlike family A GPCRs, little is known about how these receptors are activated by their endogenous ligands. This review considers what is known about the activation of family B GPCRs and then considers how this might be applied to CLR, particularly in light of new knowledge of the crystal structures of family A GPCRs.

Similarity between class A and class B G-protein-coupled receptors exemplified through calcitonin gene-related peptide receptor modelling and mutagenesis studies.

Modelling class B G-protein-coupled receptors (GPCRs) using class A GPCR structural templates is difficult due to lack of homology. The plant GPCR, GCR1, has homology to both class A and class B GPCRs. We have used this to generate a class A-class B alignment, and by incorporating maximum lagged correlation of entropy and hydrophobicity into a consensus score, we have been able to align receptor transmembrane regions. We have applied this analysis to generate active and inactive homology models of the class B calcitonin gene-related peptide (CGRP) receptor, and have supported it with site-directed mutagenesis data using 122 CGRP receptor residues and 144 published mutagenesis results on other class B GPCRs. The variation of sequence variability with structure, the analysis of polarity violations, the alignment of group-conserved residues and the mutagenesis results at 27 key positions were particularly informative in distinguishing between the proposed and plausible alternative alignments. Furthermore, we have been able to associate the key molecular features of the class B GPCR signalling machinery with their class A counterparts for the first time. These include the [K/R]KLH motif in intracellular loop 1, [I/L]xxxL and KxxK at the intracellular end of TM5 and TM6, the NPXXY/VAVLY motif on TM7 and small group-conserved residues in TM1, TM2, TM3 and TM7. The equivalent of the class A DRY motif is proposed to involve Arg(2.39), His(2.43) and Glu(3.46), which makes a polar lock with T(6.37). These alignments and models provide useful tools for understanding class B GPCR function.

G-protein-coupled receptor dynamics: dimerization and activation models compared with experiment.

Our previously derived models of the active state of the ß2-adrenergic receptor are compared with recently published X-ray crystallographic structures of activated GPCRs (G-protein-coupled receptors). These molecular dynamics-based models using experimental data derived from biophysical experiments on activation were used to restrain the receptor to an active state that gave high enrichment for agonists in virtual screening. The ß2-adrenergic receptor active model and X-ray structures are in good agreement over both the transmembrane region and the orthosteric binding site, although in some regions the active model is more similar to the active rhodopsin X-ray structures. The general features of the microswitches were well reproduced, but with minor differences, partly because of the unexpected X-ray results for the rotamer toggle switch. In addition, most of the interacting residues between the receptor and the G-protein were identified. This analysis of the modelling has also given important additional insight into GPCR dimerization: re-analysis of results on photoaffinity analogues of rhodopsin provided additional evidence that TM4 (transmembrane helix 4) resides at the dimer interface and that ligands such as bivalent ligands may pass between the mobile helices. A comparison, and discussion, is also carried out between the use of implicit and explicit solvent for active-state modelling.

Bioinformatics and molecular modelling approaches to GPCR oligomerization.

The elusive nature of the structure and function of the G-protein coupled receptor (GPCR) dimer or oligomer has led to a variety of computational studies, most of which have been directed primarily towards understanding structure. Here we review some of the recent studies based on sequence analysis and docking experiments and the recent developments in GPCR structure that have underpinned dimerization studies. In addition, we review recent nanosecond molecular dynamics simulations and coarse-grained methods for investigating the dynamic consequences of dimerization. The strengths and weaknesses of these complementary methods are discussed. The consensus of a variety of studies is that several transmembrane helices are involved in the dimerization/oligomerization interface(s); computation has been particularly effective in elucidating the experiments that seem to indicate a key role for transmembrane helix 4.