Biology subjects

Bowman, G. R.

Publications and source records attributed to Bowman, G. R..

At least 19 recordsLinked to original sources

bioRxiv2026

Antibody maturation increases rigidity in protein-contacting regions and flexibility at glycan interfaces

Antibody design is a challenging task that could be improved by understanding the conformational changes accompanying affinity maturation. Antibody maturation is a critical immune system process by which antibodies gain mutations that improve affinity and specificity for antigen targets. However, how these mutations change antibody dynamics, specifically whether antibodies become more rigid as they mature, remains a contested topic. Using adaptive sampling molecular dynamics simulations and over 8.5 milliseconds of Folding@home simulations of seven lineages, we discover that affinity maturation selectively tunes the dynamics of antibody paratope regions depending on whether they contact protein residues or glycans. We find that antibody regions that contact glycans evolve to become more flexible and, consistent with other studies, antibody regions that contact protein residues on the antigen become rigid. This pattern holds regardless of whether one includes the constant region in the simulations, indicating that the computational cost of all-atom antibody MD simulations can be reduced by half without sacrificing the accuracy of the variable regions dynamics. We expect the principles identified in this study will enable precise, dynamics-based engineering of high-affinity antibodies and to inform immunogen design against challenging, glycan-shielded viral targets such as HIV. Significance StatementAntibodies are currently one of the most powerful therapeutics available. Our bodies evolve antibodies to bind a target (called an antigen) through a process called maturation. Understanding how antibody dynamics change during maturation would lead to improved antibody design strategies. Through extensive all-atom molecular dynamics simulations of multiple antibody lineages, we demonstrate that the dynamics accompanying affinity maturation are dependent on the antigen binding site. Antibody regions that contact protein residues on the antigen become rigid while regions that contact glycans gain flexibility with maturation. These findings can be leveraged for improved, dynamics-based antibody design.

biophysics↗

bioRxiv2026

AE-PocketMiner Uses Attention to Simultaneously Predict Cryptic Pockets and Their Allosteric Coupling

Finding and targeting cryptic pockets could dramatically expand the druggable proteome. However, discovering these sites remains challenging since they are only open a fraction of the time. It is also difficult to predict the functional relevance of a cryptic site as this often requires insight into allostery. Here we introduce attention enabled (AE-)PocketMiner, an artificial intelligence (AI) method that uses a graph neural network with an attention mechanism to simultaneously predict the locations of cryptic pockets and their allosteric coupling to the rest of the protein from a single input structure. We show that AE-PocketMiner outperforms past methods for identifying cryptic pockets and recapitulates known allosteric interactions. Moreover, we experimentally confirm newly predicted cryptic pockets and mutations that allosterically control pocket opening. AE-PocketMiner thus provides a powerful framework for multiple steps of the drug discovery process--including pocket identification, prioritization, and assay design--that will help expand the druggable proteome.

biophysics↗

bioRxiv2026

Landscape of reversion alterations in homologous recombination genes reveals evolutionary constraints

Reversion mutations (REVs) restore homologous recombination repair (HRR) and confer resistance to PARP inhibitors (PARPi) in HR-deficient cancers. Yet, their prevalence, mechanisms, and biological constraints remain undefined. We analyzed genomic profiling of 609,464 tissue and liquid biopsy samples across multiple cancer types to delineate the pan-cancer landscape of REVs. REVs were identified in eight HRR genes, most frequently BRCA2 and BRCA1 and notably never in ATM or CHEK2. REVs exclusively impacted truncating pathogenic variants, predominantly through large in-frame and exon-level deletions, associated with repetitive sequences. Conserved functional domains are relatively depleted of REVs. Structural modeling and functional studies support that exon-level deletions preserve critical domain architecture and confer PARPi resistance. The study establishes HRR reversion as a structurally permissive, yet evolutionarily constrained, resistance mechanism with implications on response, monitoring, and therapeutic strategy. One Sentence SummaryHomologous recombination reversion evolves through structurally tolerated, microhomology-driven alterations that restore DNA repair under therapeutic selection across cancers.

cancer biology↗

bioRxiv2026

AI-Based Methods for Cryptic Pocket Detection Are Fast and Qualitative Compared to Quantitatively Predictive Simulations

Identifying and understanding cryptic pockets remains a compelling goal in drug discovery, as they can offer new avenues for targeting proteins that are otherwise challenging to modulate. While artificial intelligence (AI) methods for structure prediction have been revolutionary, they have not learned enough physics to characterize the rest of a proteins conformational ensemble, such as structures with cryptic pockets. This limitation has motivated the development of new AI-based methods for characterizing ensembles, or properties thereof, such as AlphaFlow, BioEmu, PocketMiner, and CryptoBank. Here, we benchmark how well these models and physics-based molecular dynamics (MD) simulations can recapitulate the thermodynamics of known cryptic pockets in Ebola VP35 and TEM {beta}-lactamase. These two pockets were chosen because the probability that they are open and the impact of point mutations have been well characterized experimentally. Multiple methods are remarkably successful at predicting whether a mutation will increase or decrease the probability of cryptic pocket opening. However, none can reliably predict the absolute probability of pocket opening. MD is very close for the two wild-type proteins but all the methods struggle for pockets with small probabilities of opening experimentally (e.g. less than 1%). BioEmu and PocketMiner capture some trends between variants with experimental populations over 1% but have systematic errors and poorer performance for rare pockets (e.g. <1% open experimentally). These results highlight the promise of AI- and simulation-based strategies for cryptic pocket characterization while pointing to the need for further improvements to achieve robust predictors.

biophysics↗

bioRxiv2025

Programmable cell differentiation in budding yeast uncouples reproductive and metabolic tasks

High-yield biomanufacturing requires large cell populations and a mechanism for directing metabolic resources towards product synthesis. However, the resources that support population growth are the same as those that drive productivity, creating a conflict that limits production yields. To overcome this fundamental limitation, we apply the principle of division of labor to separate reproductive and metabolic tasks into distinct cell types within an isogenic Saccharomyces cerevisiae culture. We introduce MiSTY (Microbial Stem Cell Technology in budding Yeast), a genetic platform that exploits natural asymmetric cues to control cell differentiation. Leveraging bud cell-specific transcription, a sequential series of recombinase-based genetic circuits generates Activated Stem Cells (ASCs) that divide asymmetrically into two cell types: bud cells that terminally differentiate into Factory Cells (FCs) and mother cells that remain self-renewing ASCs. Time-lapse microscopy demonstrated 100% differentiation fidelity across 97 cell divisions. Phenotypic and genotypic analyses showed that stem cell populations could be converted to over 95% FCs within 24 generations. By converting FCs into leucine auxotrophs, we inhibited FC proliferation while allowing continued ASC division, demonstrating complete uncoupling of cell growth from product synthesis. Because they continuously generate healthy new FCs, MiSTY cultures maintain high levels of productivity even under conditions that severely impair the growth and biosynthetic capacity of metabolically exhausted factory cells.

bioengineering↗

bioRxiv2024

Regulating IL-2 immune signaling function via a core allosteric structural network

Human interleukin-2 (IL-2) is a crucial cytokine for T cell regulation, with therapeutic potential in cancer and autoimmune diseases. However, IL-2s pleiotropic effects across different immune cell types often lead to toxicity and limited efficacy. Previous efforts to enhance IL-2s therapeutic profile have focused on modifying its receptor binding sites. Yet, the underlying dynamics and intramolecular networks contributing to IL-2 receptor recognition remain unexplored. This study presents a detailed characterization of IL-2 dynamics compared to two engineered IL-2 mutants, "superkines" S15 and S1, which exhibit biased signaling towards effector T cells. Using NMR spectroscopy and molecular dynamics simulations, we demonstrate significant variations in core dynamic pathways and conformational exchange rates across these three IL-2 variants. We identify distinct allosteric networks and excited state conformations in the superkines, despite their structural similarity to wild-type IL-2. Furthermore, we rationally design a mutation (L56A) in the S1 superkines core network, which partially reverts its dynamics, receptor binding affinity, and T cell signaling behavior towards that of wild-type IL-2. Our results reveal that IL-2 superkine core dynamics play a critical role in their enhanced receptor binding and function, suggesting that modulating IL-2 dynamics and core allostery represents an untapped approach for designing immunotherapies with improved immune cell selectivity profiles. HighlightsO_LINMR and molecular dynamics simulations revealed distinct conformational dynamics and allosteric networks in computationally re-designed IL-2 superkines compared to wild-type IL-2, despite their similar crystal structures. C_LIO_LIThe superkines S1 and S15 exhibit altered sampling of excited state conformations at an intermediate timescale, with slower conformational exchange rates compared to wild-type IL-2. C_LIO_LIA rationally designed mutation (L56A) in the S1 superkines core allosteric network partially reverted its dynamics, receptor binding affinity, and T cell signaling behavior towards that of wild-type IL-2. C_LIO_LIOur study demonstrates that IL-2 core dynamics play a critical role in receptor binding and signaling function, providing a foundation for engineering more selective IL-2-based immunotherapies. C_LI Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=152 HEIGHT=200 SRC="FIGDIR/small/617024v1_ufig1.gif" ALT="Figure 1"> View larger version (30K): org.highwire.dtl.DTLVardef@1e298adorg.highwire.dtl.DTLVardef@195b602org.highwire.dtl.DTLVardef@1e06205org.highwire.dtl.DTLVardef@4351e5_HPS_FORMAT_FIGEXP M_FIG C_FIG

biochemistry↗

bioRxiv2024

Deep Mining of the Human Antibody Repertoire Identifies Frequent and Immunogenetically Diverse CDRH3 Topologies Targetable by Vaccination

Germline targeting vaccination strategies against highly variable pathogens such as HIV aim to elicit broadly neutralizing antibodies (bnAbs) with particular immunogenetic or structural features. The V2 apex of the HIV Env protein is a promising target for a class of bnAbs that contain conserved structural motifs in the heavy chain complementarity determining region 3 (CDRH3). Here, we show that these structural motifs are targetable by vaccination by characterizing V2 apex axe-like CDRH3s in the human repertoire and developing new immunogens capable of engaging them. We determined the frequency and diversity of axe-like CDHR3s in healthy human donors using a series of structural informatics approaches, finding these precursors in nearly 90% of donors. Axe-targeting immunogens based on the HIV Env Q23.17 bound axe-like precursors in cryo-EM structures, induced V2 apex-specific antibody responses in humanized mice, and induced axe-like heterologous neutralizing antibodies in rhesus macaques infected with a germline-targeted simian-human immunodeficiency virus. These results illustrate a new structure-guided immunoinformatic vaccine design paradigm that can be employed to elicit immunogenetically diverse yet structurally conserved classes of antibodies. Significance StatementMany broadly neutralizing antibodies (bnAbs) utilize modes of epitope recognition dominated by the antibody complementarity determining region 3 (CDRH3). The CDRH3 is the most diverse part of the antibody, posing a challenge for germline targeting vaccine designs that aim to elicit antibodies with particular immunogenetic features. Vaccine design strategies that accommodate CDRH3 variability are therefore needed. Many HIV Env V2 apex bnAbs share "axe-like" CDRH3 microfolds that arise from diverse immunogenetic origins. Here we determined the frequency in humans of B cells with such CDRH3 topologies and designed immunogens to engage their precursors. This work opens a path toward vaccines that engage specific structural classes of B cells, thereby advancing the rational design of immunogens for HIV and other pathogens.

immunology↗

bioRxiv2024

The probability of cryptic pocket opening controls functional tradeoffs in filovirus immune evasion.

Cryptic pockets are of growing interest as potential drug targets, particularly to control protein-nucleic acid interactions that often occur via flat surfaces. However, it remains unclear whether cryptic pockets contribute to protein function or if they are merely happenstantial features that can easily be evolved away to achieve drug resistance. Here, we explore whether a cryptic pocket in the Interferon Inhibitory Domain (IID) of viral protein 35 (VP35) of Zaire ebolavirus aids its ability to bind double-stranded RNA (dsRNA). We use simulations and experiments to study the relationship between cryptic pocket opening and dsRNA binding of the IIDs of two other filoviruses, Reston and Marburg. These homologs have nearly identical structures but block different interferon pathways due to different affinities for blunt ends and backbone of the dsRNA. Simulations and thiol-labeling experiments demonstrate that the homologs have varying probabilities of pocket opening. Subsequent dsRNA-binding assays suggest that closed conformations preferentially bind dsRNA blunt ends while open conformations prefer binding the backbone. Point mutations that modulate pocket opening proteins further confirm this preference. These results demonstrate the open cryptic pocket has a function, suggesting cryptic pockets are under selective pressure and may be difficult to evolve away to achieve drug resistance.

biophysics↗

bioRxiv2024

Accounting for fast vs slow exchange in single molecule FRET experiments reveals hidden conformational states

Proteins are dynamic systems whose structural preferences determine their function. Unfortunately, building atomically detailed models of protein structural ensembles remains challenging, limiting our understanding of the relationships between sequence, structure, and function. Combining single molecule Forster resonance energy transfer (smFRET) experiments with molecular dynamics simulations could provide experimentally grounded, all-atom models of a proteins structural ensemble. However, agreement between the two techniques is often insufficient to achieve this goal. Here, we explore whether accounting for important experimental details like averaging across structures sampled during a given smFRET measurement is responsible for this apparent discrepancy. We present an approach to account for this time-averaging by leveraging the kinetic information available from Markov state models of a proteins dynamics. This allows us to accurately assess which timescales are averaged during an experiment. We find this approach significantly improves agreement between simulations and experiments in proteins with varying degrees of dynamics, including the well-ordered protein T4 lysozyme, the partially disordered protein apolipoprotein E (ApoE), and a disordered amyloid protein (A{beta}40). We find evidence for hidden states that are not apparent in smFRET experiments because of time averaging with other structures, akin to states in fast exchange in NMR, and evaluate different force fields. Finally, we show how remaining discrepancies between computations and experiments can be used to guide additional simulations and build structural models for states that were previously unaccounted for. We expect our approach will enable combining simulations and experiments to understand the link between sequence, structure, and function in many settings.

biophysics↗

bioRxiv2023

G protein activation occurs via a largely universal mechanism

Understanding how signaling proteins like G proteins are allosterically activated is a long-standing challenge with significant biological and medical implications. Because it is difficult to directly observe such dynamic processes, much of our understanding is based on inferences from a limited number of static snapshots of relevant protein structures, mutagenesis data, and patterns of sequence conservation. Here, we use computer simulations to directly interrogate allosteric coupling in six G protein -subunit isoforms covering all four G protein families. To analyze this data, we introduce automated methods for inferring allosteric networks from simulation data and assessing how allostery is conserved or diverged among related protein isoforms. We find that the allosteric networks in these six G protein -subunits are largely conserved and consist of two pathways, which we call pathway-I and pathway-II. This analysis predicts that pathway-I is generally dominant over the second pathway, which we experimentally confirmed by showing that mutations to pathway-I perturb nucleotide exchange more than mutations to pathway-II. In the future, insights into unique elements of each G protein family could inform the design of isoform-specific drugs. More broadly, our tools should also be useful for studying allostery in other proteins and assessing the extent to which this allostery is conserved in related proteins.

biophysics↗

bioRxiv2023

Phospho-signaling couples polar asymmetry and proteolysis within a membraneless microdomain in C. crescentus.

Asymmetric cell division in bacteria is achieved through cell polarization, where regulatory proteins are directed to specific cell poles. Curiously, both poles contain a membraneless microdomain, established by the polar assembly hub PopZ, through most of the cell cycle, yet many PopZ clients are unipolar and transiently localized. We find that PopZs interaction with the response regulator CpdR is controlled by phosphorylation, via the histidine kinase CckA. Phosphorylated CpdR does not interact with PopZ and is not localized to cell poles. At poles where CckA acts as a phosphatase, de-phosphorylated CpdR binds directly with PopZ and subsequently recruits ClpX, substrates, and other members of a protease complex to the cell pole. We also find that co-recruitment of protease components and substrates to polar microdomains enhances their coordinated activity. This study connects phosphosignaling with polar assembly and the activity of a protease that triggers cell cycle progression and cell differentiation.

molecular biology↗

bioRxiv2023

POPSHIFT: A THERMODYNAMICALLY SOUND APPROACH TO ESTIMATE BINDING FREE ENERGIES BY ACCOUNTING FOR LIGAND-INDUCED POPULATION SHIFTS FROM A LIGAND-FREE MSM

Obtaining accurate binding free energies from in silico screens has been a longstanding goal for the computational chemistry community. However, accuracy and computational cost are at odds with one another, limiting the utility of methods that perform this type of calculation. Many methods achieve massive scale by explicitly or implicitly assuming that the target protein adopts a single structure, or undergoes limited fluctuations around that structure, to minimize computational cost. Others simulate each protein-ligand complex of interest, accepting lower throughput in exchange for better predictions of binding affinities. Here, we present the PopShift framework for accounting for the ensemble of structures a protein adopts and their relative probabilities. Protein degrees of freedom are enumerated once, and then arbitrarily many molecules can be screened against this ensemble. Specifically, we use Markov state models (MSMs) as a compressed representation of a proteins thermodynamic ensemble. We start with a ligand-free MSM and then calculate how addition of a ligand shifts the populations of each protein conformational state based on the strength of the interaction between that protein conformation and the ligand. In this work we use docking to estimate the affinity between a given protein structure and ligand, but any estimator of binding affinities could be used in the PopShift framework. We test PopShift on the classic benchmark pocket T4 Lysozyme L99A. We find that PopShift is more accurate than common strategies, such as docking to a single structure and traditional ensemble docking--producing results that compare favorably with alchemical binding free energy calculations in terms of RMSE but not correlation--and may have a more favorable computational cost profile in some applications. In addition to predicting binding free energies and ligand poses, PopShift also provides insight into how the probability of different protein structures is shifted upon addition of various concentrations of ligand, providing a platform for predicting affinities and allosteric effects of ligand binding. Therefore, we expect PopShift will be valuable for hit finding and for providing insight into phenomena like allostery.

biophysics↗

bioRxiv2023

Divergent Molecular Phenotypes in Point Mutations at the Same Residue in Beta-Myosin Heavy Chain Lead to Distinct Cardiomyopathies

In genetic cardiomyopathies, a frequently described phenomenon is how similar mutations in one protein can lead to discrete clinical phenotypes. One example is illustrated by two mutations in beta myosin heavy chain ({beta}-MHC) that are linked to hypertrophic cardiomyopathy (HCM) (Ile467Val, I467V) and left ventricular non-compaction (LVNC) (Ile467Thr, I467T). To investigate how these missense mutations lead to independent diseases, we studied the molecular effects of each mutation using recombinant human {beta}-MHC Subfragment 1 (S1) in vitro assays. Both HCM-I467V and LVNC-I467T S1 mutations exhibited similar mechanochemical functions, including unchanged ATPase and enhanced actin velocity but had distinct effects on the basal activity of myosin. HCM-I467V S1 showed no change in basal ATPase activity of myosin while LVNC-I467T reduced the basal ATPase activity by 50%. Molecular dynamics simulations reveal that I467T allosterically disrupts nucleotide binding of myosin, which may contribute to the uncoupled reduced basal activity and enhanced actin velocity observed in this mutation. These contrasting molecular effects may lead to contractile dysregulation that initiates LVNC-associated signaling pathways that progress the phenotype. Together, analysis of these mutations provides evidence that phenotypic complexity originates at the molecular level and is critical to understanding disease progression and developing therapies.

biophysics↗

bioRxiv2023

Discovery of a cryptic pocket in the AI-predicted structure of PPM1D phosphatase explains the binding site and potency of its allosteric inhibitors

Virtual screening is a widely used tool for drug discovery, but its predictive power can vary dramatically depending on how much structural data is available. In the best case, crystal structures of a ligand-bound protein can help find more potent ligands. However, virtual screens tend to be less predictive when only ligand-free crystal structures are available, and even less predictive if a homology model or other predicted structure must be used. Here, we explore the possibility that this situation can be improved by better accounting for protein dynamics, as simulations started from a single structure have a reasonable chance of sampling nearby structures that are more compatible with ligand binding. As a specific example, we consider the cancer drug target PPM1D/Wip1 phosphatase, a protein that lacks crystal structures. High-throughput screens have led to the discovery of several allosteric inhibitors of PPM1D, but their binding mode remains unknown. To enable further drug discovery efforts, we assessed the predictive power of an AlphaFold-predicted structure of PPM1D and a Markov state model (MSM) built from molecular dynamics simulations initiated from that structure. Our simulations reveal a cryptic pocket at the interface between two important structural elements, the flap and hinge regions. Using deep learning to predict the pose quality of each docked compound for the active site and cryptic pocket suggests that the inhibitors strongly prefer binding to the cryptic pocket, consistent with their allosteric effect. The predicted affinities for the dynamically uncovered cryptic pocket also recapitulate the relative potencies of the compounds ({tau}b=0.70) better than the predicted affinities for the static AlphaFold-predicted structure ({tau}b=0.42). Taken together, these results suggest that targeting the cryptic pocket is a good strategy for drugging PPM1D and, more generally, that conformations selected from simulation can improve virtual screening when limited structural data is available.

biophysics↗

bioRxiv2022

Deep Generative Design of Epitope-Specific Binding Proteins by Latent Conformation Optimization

AO_SCPLOWBSTRACTC_SCPLOWDesigning de novo binding proteins against arbitrary epitopes using a single scaffold, as seen with natural antibodies, remains an unsolved challenge in protein design. Current design methods are unable to capture the structural dynamics of flexible loops nor search loop conformational space in a principled way. Here we present Sculptor, a deep generative design algorithm that creates epitope-specific protein binders. The Sculptor algorithm constitutes a joint search over the positions, interactions, and generated conformations of a fold, and crafts a backbone to complement a user-specified epitope. Sequences are designed onto generated backbones using a combination of a residue-wise interaction database, a convolutional sequence design module, and Rosetta. Instead of relying on static structures, we capture the local conformational landscape of a single fold using molecular dynamics, and demonstrate that a model trained on such dense conformational data can generate backbones tailor-fit to an epitope. We use Sculptor to design binders against a conserved epitope on venom toxins implicated in neuromuscular paralysis, and obtain a multi-toxin binder from a small naive library - a promising step towards creating broadly neutralizing binders. This study constitutes a novel application of deep generative modeling for epitope-targeted design, leveraging conformational dynamics to achieve function.

biochemistry↗

bioRxiv2022

Accelerating cryptic pocket discovery using AlphaFold

Cryptic pockets, or pockets absent in ligand-free, experimentally determined structures, hold great potential as drug targets. However, cryptic pocket opening is often beyond the reach of conventional biomolecular simulations because certain cryptic pocket openings involve slow motions. Here, we investigate whether AlphaFold can be used to accelerate cryptic pocket discovery either by generating structures with open pockets directly or generating structures with partially open pockets that can be used as starting points for simulations. We use AlphaFold to generate ensembles for 10 known cryptic pocket examples, including 5 that were deposited after AlphaFolds training data was extracted from the PDB. We find that in 6 out of 10 cases AlphaFold samples the open state. For plasmepsin II, an aspartic protease from the causative agent of malaria, AlphaFold only captures partial pocket opening. As a result, we ran simulations from an ensemble of AlphaFold-generated structures and show that this strategy samples cryptic pocket opening, even though an equivalent amount of simulations launched from a ligand-free experimental structure fails to do so. Markov state models (MSMs) constructed from the AlphaFold-seeded simulations quickly yield a free energy landscape of cryptic pocket opening that is in good agreement with the same landscape generated with well-tempered metadynamics. Taken together, our results demonstrate that AlphaFold has a useful role to play in cryptic pocket discovery but that many cryptic pockets may remain difficult to sample using AlphaFold alone.

biochemistry↗

bioRxiv2022

Drug specificity and affinity are encoded in the probability of cryptic pocket opening in myosin motor domains

The design of compounds that can discriminate between closely related target proteins remains a central challenge in drug discovery. Specific therapeutics targeting the highly conserved myosin motor family are urgently needed as mutations in at least 6 of its members cause numerous diseases. Allosteric modulators, like the myosin-II inhibitor blebbistatin, are a promising means to achieve specificity. However, it remains unclear why blebbistatin inhibits myosin-II motors with different potencies given that it binds at a highly conserved pocket that is always closed in blebbistatin-free experimental structures. We hypothesized that the probability of pocket opening is an important determinant of the potency of compounds like blebbistatin. To test this hypothesis, we used Markov state models (MSMs) built from over 2 milliseconds of aggregate molecular dynamics simulations with explicit solvent. We find that blebbistatins binding pocket readily opens in simulations of blebbistatin-sensitive myosin isoforms. Comparing these conformational ensembles reveals that the probability of pocket opening correctly identifies which isoforms are most sensitive to blebbistatin inhibition and that docking against MSMs quantitatively predicts blebbistatin binding affinities (R2=0.82). To test our ability to make blind predictions, we predicted blebbistatins binding affinity for an isoform (Myh7b) whose blebbistatin sensitivity was unknown. Encouragingly, we find good agreement between the predicted and measured IC50 (0.67 {micro}M vs. 0.36 {micro}M). Therefore, we expect this framework to be useful for the development of novel specific drugs across numerous protein targets. SignificanceDrug development requires the discovery of compounds which specifically target one member of a protein family without triggering side effects that arise from interactions with other related proteins. Myosins are a family of motor proteins that are drug targets for heart diseases, cancer, and parasitic infections. Here, we investigate why the compound blebbistatin specifically inhibits some myosins more potently than others, even though its binding site is closed in all known experimental structures. We find that the blebbistatin binding pocket opens in molecular dynamics simulations of certain myosin motors, and that the probability of opening predicts how potently blebbistatin inhibits a particular motor. Our work suggests that differences in cryptic pocket formation can be exploited to develop specific therapeutics.

biophysics↗

bioRxiv2022

Predicting the locations of cryptic pockets from single protein structures using the PocketMiner graph neural network

Cryptic pockets expand the scope of drug discovery by enabling targeting of proteins currently considered undruggable because they lack pockets in their ground state structures. However, identifying cryptic pockets is labor-intensive and slow. The ability to accurately and rapidly predict if and where cryptic pockets are likely to form from a protein structure would greatly accelerate the search for druggable pockets. Here, we present PocketMiner, a graph neural network trained to predict where pockets are likely to open in molecular dynamics simulations. Applying PocketMiner to single structures from a newly-curated dataset of 39 experimentally-confirmed cryptic pockets demonstrates that it accurately identifies cryptic pockets (ROC-AUC: 0.87) >1,000-fold faster than existing methods. We apply PocketMiner across the human proteome and show that predicted pockets open in simulations, suggesting that over half of proteins thought to lack pockets based on available structures are likely to contain cryptic pockets, vastly expanding the druggable proteome.

biophysics↗