Membrane lipids drive formation of KRAS4b-RAF1 RBDCRD nanoclusters on the membrane.

The oncogene RAS, extensively studied for decades, presents persistent gaps in understanding, hindering the development of effective therapeutic strategies due to a lack of precise details on how RAS initiates MAPK signaling with RAF effector proteins at the plasma membrane. Recent advances in X-ray crystallography, cryo-EM, and super-resolution fluorescence microscopy offer structural and spatial insights, yet the molecular mechanisms involving protein-protein and protein-lipid interactions in RAS-mediated signaling require further characterization. This study utilizes single-molecule experimental techniques, nuclear magnetic resonance spectroscopy, and the computational Machine-Learned Modeling Infrastructure (MuMMI) to examine KRAS4b and RAF1 on a biologically relevant lipid bilayer. MuMMI captures long-timescale events while preserving detailed atomic descriptions, providing testable models for experimental validation. Both in vitro and computational studies reveal that RBDCRD binding alters KRAS lateral diffusion on the lipid bilayer, increasing cluster size and decreasing diffusion. RAS and membrane binding cause hydrophobic residues in the CRD region to penetrate the bilayer, stabilizing complexes through ß-strand elongation. These cooperative interactions among lipids, KRAS4b, and RAF1 are proposed as essential for forming nanoclusters, potentially a critical step in MAP kinase signal activation.

Revealing the mechanism of action of a first-in-class covalent inhibitor of KRASG12C (ON) and other functional properties of oncogenic KRAS by 31P NMR.

Individual oncogenic KRAS mutants confer distinct differences in biochemical properties and signaling for reasons that are not well understood. KRAS activity is closely coupled to protein dynamics and is regulated through two interconverting conformations: state 1 (inactive, effector binding deficient) and state 2 (active, effector binding enabled). Here, we use 31P NMR to delineate the differences in state 1 and state 2 populations present in WT and common KRAS oncogenic mutants (G12C, G12D, G12V, G13D, and Q61L) bound to its natural substrate GTP or a commonly used nonhydrolyzable analog GppNHp (guanosine-5'-[(ß,γ)-imido] triphosphate). Our results show that GppNHp-bound proteins exhibit significant state 1 population, whereas GTP-bound KRAS is primarily (90% or more) in state 2 conformation. This observation suggests that the predominance of state 1 shown here and in other studies is related to GppNHp and is most likely nonexistent in cells. We characterize the impact of this differential conformational equilibrium of oncogenic KRAS on RAF1 kinase effector RAS-binding domain and intrinsic hydrolysis. Through a KRAS G12C drug discovery, we have identified a novel small-molecule inhibitor, BBO-8956, which is effective against both GDP- and GTP-bound KRAS G12C. We show that binding of this inhibitor significantly perturbs state 1-state 2 equilibrium and induces an inactive state 1 conformation in GTP-bound KRAS G12C. In the presence of BBO-8956, RAF1-RAS-binding domain is unable to induce a signaling competent state 2 conformation within the ternary complex, demonstrating the mechanism of action for this novel and active-conformation inhibitor.

Clustering Protein Binding Pockets and Identifying Potential Drug Interactions: A Novel Ligand-Based Featurization Method.

Protein-ligand interactions are essential to drug discovery and drug development efforts. Desirable on-target or multitarget interactions are the first step in finding an effective therapeutic, while undesirable off-target interactions are the first step in assessing safety. In this work, we introduce a novel ligand-based featurization and mapping of human protein pockets to identify closely related protein targets and to project novel drugs into a hybrid protein-ligand feature space to identify their likely protein interactions. Using structure-based template matches from PDB, protein pockets are featured by the ligands that bind to their best co-complex template matches. The simplicity and interpretability of this approach provide a granular characterization of the human proteome at the protein-pocket level instead of the traditional protein-level characterization by family, function, or pathway. We demonstrate the power of this featurization method by clustering a subset of the human proteome and evaluating the predicted cluster associations of over 7000 compounds.

Evaluating point-prediction uncertainties in neural networks for protein-ligand binding prediction.

Neural Network (NN) models provide potential to speed up the drug discovery process and reduce its failure rates. The success of NN models requires uncertainty quantification (UQ) as drug discovery explores chemical space beyond the training data distribution. Standard NN models do not provide uncertainty information. Some methods require changing the NN architecture or training procedure, limiting the selection of NN models. Moreover, predictive uncertainty can come from different sources. It is important to have the ability to separately model different types of predictive uncertainty, as the model can take assorted actions depending on the source of uncertainty. In this paper, we examine UQ methods that estimate different sources of predictive uncertainty for NN models aiming at protein-ligand binding prediction. We use our prior knowledge on chemical compounds to design the experiments. By utilizing a visualization method we create non-overlapping and chemically diverse partitions from a collection of chemical compounds. These partitions are used as training and test set splits to explore NN model uncertainty. We demonstrate how the uncertainties estimated by the selected methods describe different sources of uncertainty under different partitions and featurization schemes and the relationship to prediction error.

OFraMP: a fragment-based tool to facilitate the parametrization of large molecules.

An Online tool for Fragment-based Molecule Parametrization (OFraMP) is described. OFraMP is a web application for assigning atomic interaction parameters to large molecules by matching sub-fragments within the target molecule to equivalent sub-fragments within the Automated Topology Builder (ATB, atb.uq.edu.au) database. OFraMP identifies and compares alternative molecular fragments from the ATB database, which contains over 890,000 pre-parameterized molecules, using a novel hierarchical matching procedure. Atoms are considered within the context of an extended local environment (buffer region) with the degree of similarity between an atom in the target molecule and that in the proposed match controlled by varying the size of the buffer region. Adjacent matching atoms are combined into progressively larger matched sub-structures. The user then selects the most appropriate match. OFraMP also allows users to manually alter interaction parameters and automates the submission of missing substructures to the ATB in order to generate parameters for atoms in environments not represented in the existing database. The utility of OFraMP is illustrated using the anti-cancer agent paclitaxel and a dendrimer used in organic semiconductor devices. OFraMP applied to paclitaxel (ATB ID 35922).

Machine Learning-Driven Multiscale Modeling: Bridging the Scales with a Next-Generation Simulation Infrastructure.

Interdependence across time and length scales is common in biology, where atomic interactions can impact larger-scale phenomenon. Such dependence is especially true for a well-known cancer signaling pathway, where the membrane-bound RAS protein binds an effector protein called RAF. To capture the driving forces that bring RAS and RAF (represented as two domains, RBD and CRD) together on the plasma membrane, simulations with the ability to calculate atomic detail while having long time and large length- scales are needed. The Multiscale Machine-Learned Modeling Infrastructure (MuMMI) is able to resolve RAS/RAF protein-membrane interactions that identify specific lipid-protein fingerprints that enhance protein orientations viable for effector binding. MuMMI is a fully automated, ensemble-based multiscale approach connecting three resolution scales: (1) the coarsest scale is a continuum model able to simulate milliseconds of time for a 1 µm2 membrane, (2) the middle scale is a coarse-grained (CG) Martini bead model to explore protein-lipid interactions, and (3) the finest scale is an all-atom (AA) model capturing specific interactions between lipids and proteins. MuMMI dynamically couples adjacent scales in a pairwise manner using machine learning (ML). The dynamic coupling allows for better sampling of the refined scale from the adjacent coarse scale (forward) and on-the-fly feedback to improve the fidelity of the coarser scale from the adjacent refined scale (backward). MuMMI operates efficiently at any scale, from a few compute nodes to the largest supercomputers in the world, and is generalizable to simulate different systems. As computing resources continue to increase and multiscale methods continue to advance, fully automated multiscale simulations (like MuMMI) will be commonly used to address complex science questions.

The confluence of machine learning and multiscale simulations.

Multiscale modeling has a long history of use in structural biology, as computational biologists strive to overcome the time- and length-scale limits of atomistic molecular dynamics. Contemporary machine learning techniques, such as deep learning, have promoted advances in virtually every field of science and engineering and are revitalizing the traditional notions of multiscale modeling. Deep learning has found success in various approaches for distilling information from fine-scale models, such as building surrogate models and guiding the development of coarse-grained potentials. However, perhaps its most powerful use in multiscale modeling is in defining latent spaces that enable efficient exploration of conformational space. This confluence of machine learning and multiscale simulation with modern high-performance computing promises a new era of discovery and innovation in structural biology.

PDBspheres: a method for finding 3D similarities in local regions in proteins.

We present a structure-based method for finding and evaluating structural similarities in protein regions relevant to ligand binding. PDBspheres comprises an exhaustive library of protein structure regions ('spheres') adjacent to complexed ligands derived from the Protein Data Bank (PDB), along with methods to find and evaluate structural matches between a protein of interest and spheres in the library. PDBspheres uses the LGA (Local-Global Alignment) structure alignment algorithm as the main engine for detecting structural similarities between the protein of interest and template spheres from the library, which currently contains >2 million spheres. To assess confidence in structural matches, an all-atom-based similarity metric takes side chain placement into account. Here, we describe the PDBspheres method, demonstrate its ability to detect and characterize binding sites in protein structures, show how PDBspheres-a strictly structure-based method-performs on a curated dataset of 2528 ligand-bound and ligand-free crystal structures, and use PDBspheres to cluster pockets and assess structural similarities among protein binding sites of 4876 structures in the 'refined set' of the PDBbind 2019 dataset.

Deep generative molecular design reshapes drug discovery.

Recent advances and accomplishments of artificial intelligence (AI) and deep generative models have established their usefulness in medicinal applications, especially in drug discovery and development. To correctly apply AI, the developer and user face questions such as which protocols to consider, which factors to scrutinize, and how the deep generative models can integrate the relevant disciplines. This review summarizes classical and newly developed AI approaches, providing an updated and accessible guide to the broad computational drug discovery and development community. We introduce deep generative models from different standpoints and describe the theoretical frameworks for representing chemical and biological structures and their applications. We discuss the data and technical challenges and highlight future directions of multimodal deep generative models for accelerating drug discovery.

Exploring CRD mobility during RAS/RAF engagement at the membrane.

During the activation of mitogen-activated protein kinase (MAPK) signaling, the RAS-binding domain (RBD) and cysteine-rich domain (CRD) of RAF bind to active RAS at the plasma membrane. The orientation of RAS at the membrane may be critical for formation of the RAS-RBDCRD complex and subsequent signaling. To explore how RAS membrane orientation relates to the protein dynamics within the RAS-RBDCRD complex, we perform multiscale coarse-grained and all-atom molecular dynamics (MD) simulations of KRAS4b bound to the RBD and CRD domains of RAF-1, both in solution and anchored to a model plasma membrane. Solution MD simulations describe dynamic KRAS4b-CRD conformations, suggesting that the CRD has sufficient flexibility in this environment to substantially change its binding interface with KRAS4b. In contrast, when the ternary complex is anchored to the membrane, the mobility of the CRD relative to KRAS4b is restricted, resulting in fewer distinct KRAS4b-CRD conformations. These simulations implicate membrane orientations of the ternary complex that are consistent with NMR measurements. While a crystal structure-like conformation is observed in both solution and membrane simulations, a particular intermolecular rearrangement of the ternary complex is observed only when it is anchored to the membrane. This configuration emerges when the CRD hydrophobic loops are inserted into the membrane and helices α3-5 of KRAS4b are solvent exposed. This membrane-specific configuration is stabilized by KRAS4b-CRD contacts that are not observed in the crystal structure. These results suggest modulatory interplay between the CRD and plasma membrane that correlate with RAS/RAF complex structure and dynamics, and potentially influence subsequent steps in the activation of MAPK signaling.

Asynchronous Reciprocal Coupling of Martini 2.2 Coarse-Grained and CHARMM36 All-Atom Simulations in an Automated Multiscale Framework.

The appeal of multiscale modeling approaches is predicated on the promise of combinatorial synergy. However, this promise can only be realized when distinct scales are combined with reciprocal consistency. Here, we consider multiscale molecular dynamics (MD) simulations that combine the accuracy and macromolecular flexibility accessible to fixed-charge all-atom (AA) representations with the sampling speed accessible to reductive, coarse-grained (CG) representations. AA-to-CG conversions are relatively straightforward because deterministic routines with unique outcomes are achievable. Conversely, CG-to-AA conversions have many solutions due to a surge in the number of degrees of freedom. While automated tools for biomolecular CG-to-AA transformation exist, we find that one popular option, called Backward, is prone to stochastic failure and the AA models that it does generate frequently have compromised protein structure and incorrect stereochemistry. Although these shortcomings can likely be circumvented by human intervention in isolated instances, automated multiscale coupling requires reliable and robust scale conversion. Here, we detail an extension to Multiscale Machine-learned Modeling Infrastructure (MuMMI), including an improved CG-to-AA conversion tool called sinceCG. This tool is reliable (∼98% weakly correlated repeat success rate), automatable (no unrecoverable hangs), and yields AA models that generally preserve protein secondary structure and maintain correct stereochemistry. We describe how the MuMMI framework identifies CG system configurations of interest, converts them to AA representations, and simulates them at the AA scale while on-the-fly analyses provide feedback to update CG parameters. Application to systems containing the peripheral membrane protein RAS and proximal components of RAF kinase on complex eight-component lipid bilayers with ∼1.5 million atoms is discussed in the context of MuMMI.

Machine learning-driven multiscale modeling reveals lipid-dependent dynamics of RAS signaling proteins.

RAS is a signaling protein associated with the cell membrane that is mutated in up to 30% of human cancers. RAS signaling has been proposed to be regulated by dynamic heterogeneity of the cell membrane. Investigating such a mechanism requires near-atomistic detail at macroscopic temporal and spatial scales, which is not possible with conventional computational or experimental techniques. We demonstrate here a multiscale simulation infrastructure that uses machine learning to create a scale-bridging ensemble of over 100,000 simulations of active wild-type KRAS on a complex, asymmetric membrane. Initialized and validated with experimental data (including a new structure of active wild-type KRAS), these simulations represent a substantial advance in the ability to characterize RAS-membrane biology. We report distinctive patterns of local lipid composition that correlate with interfacially promiscuous RAS multimerization. These lipid fingerprints are coupled to RAS dynamics, predicted to influence effector binding, and therefore may be a mechanism for regulating cell signaling cascades.

Interpretable artificial intelligence and exascale molecular dynamics simulations to reveal kinetics: Applications to Alzheimer's disease.

The rapid increase in computing power, especially with the integration of graphics processing units, has dramatically increased the capabilities of molecular dynamics simulations. To date, these capabilities extend from running very long simulations (tens to hundreds of microseconds) to thousands of short simulations. However, the expansive data generated in these simulations must be made interpretable not only by the investigator who performs them but also by others as well. Here, we demonstrate how integrating learning techniques, such as artificial intelligence, machine learning, and neural networks, into analysis pipelines can reveal the kinetics of Alzheimer's disease (AD) protein aggregation. We review select AD targets, describe current simulation methods, and introduce learning concepts and their application in AD, highlighting limitations and potential solutions.

Discovery of Small-Molecule Inhibitors of SARS-CoV-2 Proteins Using a Computational and Experimental Pipeline.

A rapid response is necessary to contain emergent biological outbreaks before they can become pandemics. The novel coronavirus (SARS-CoV-2) that causes COVID-19 was first reported in December of 2019 in Wuhan, China and reached most corners of the globe in less than two months. In just over a year since the initial infections, COVID-19 infected almost 100 million people worldwide. Although similar to SARS-CoV and MERS-CoV, SARS-CoV-2 has resisted treatments that are effective against other coronaviruses. Crystal structures of two SARS-CoV-2 proteins, spike protein and main protease, have been reported and can serve as targets for studies in neutralizing this threat. We have employed molecular docking, molecular dynamics simulations, and machine learning to identify from a library of 26 million molecules possible candidate compounds that may attenuate or neutralize the effects of this virus. The viability of selected candidate compounds against SARS-CoV-2 was determined experimentally by biolayer interferometry and FRET-based activity protein assays along with virus-based assays. In the pseudovirus assay, imatinib and lapatinib had IC50 values below 10 µM, while candesartan cilexetil had an IC50 value of approximately 67 µM against Mpro in a FRET-based activity assay. Comparatively, candesartan cilexetil had the highest selectivity index of all compounds tested as its half-maximal cytotoxicity concentration 50 (CC50) value was the only one greater than the limit of the assay (>100 µM).

Development of a CNS-permeable reactivator for nerve agent exposure: an iterative, multi-disciplinary approach.

Nerve agents have experienced a resurgence in recent times with their use against civilian targets during the attacks in Syria (2012), the poisoning of Sergei and Yulia Skripal in the United Kingdom (2018) and Alexei Navalny in Russia (2020), strongly renewing the importance of antidote development against these lethal substances. The current standard treatment against their effects relies on the use of small molecule-based oximes that can efficiently restore acetylcholinesterase (AChE) activity. Despite their efficacy in reactivating AChE, the action of drugs like 2-pralidoxime (2-PAM) is primarily limited to the peripheral nervous system (PNS) and, thus, provides no significant protection to the central nervous system (CNS). This lack of action in the CNS stems from their ionic nature that, on one end makes them very powerful reactivators and on the other renders them ineffective at crossing the Blood Brain Barrier (BBB) to reach the CNS. In this report, we describe the use of an iterative approach composed of parallel chemical and in silico syntheses, computational modeling, and a battery of detailed in vitro and in vivo assays that resulted in the identification of a promising, novel CNS-permeable oxime reactivator. Additional experiments to determine acute and chronic toxicity are ongoing.

Improved Protein-Ligand Binding Affinity Prediction with Structure-Based Deep Fusion Inference.

Predicting accurate protein-ligand binding affinities is an important task in drug discovery but remains a challenge even with computationally expensive biophysics-based energy scoring methods and state-of-the-art deep learning approaches. Despite the recent advances in the application of deep convolutional and graph neural network-based approaches, it remains unclear what the relative advantages of each approach are and how they compare with physics-based methodologies that have found more mainstream success in virtual screening pipelines. We present fusion models that combine features and inference from complementary representations to improve binding affinity prediction. This, to our knowledge, is the first comprehensive study that uses a common series of evaluations to directly compare the performance of three-dimensional (3D)-convolutional neural networks (3D-CNNs), spatial graph neural networks (SG-CNNs), and their fusion. We use temporal and structure-based splits to assess performance on novel protein targets. To test the practical applicability of our models, we examine their performance in cases that assume that the crystal structure is not available. In these cases, binding free energies are predicted using docking pose coordinates as the inputs to each model. In addition, we compare these deep learning approaches to predictions based on docking scores and molecular mechanic/generalized Born surface area (MM/GBSA) calculations. Our results show that the fusion models make more accurate predictions than their constituent neural network models as well as docking scoring and MM/GBSA rescoring, with the benefit of greater computational efficiency than the MM/GBSA method. Finally, we provide the code to reproduce our results and the parameter files of the trained models used in this work. The software is available as open source at https://github.com/llnl/fast. Model parameter files are available at ftp://gdo-bioinformatics.ucllnl.org/fast/pdbbind2016_model_checkpoints/.

Comprehensive characterization of protein-protein interactions perturbed by disease mutations.

Technological and computational advances in genomics and interactomics have made it possible to identify how disease mutations perturb protein-protein interaction (PPI) networks within human cells. Here, we show that disease-associated germline variants are significantly enriched in sequences encoding PPI interfaces compared to variants identified in healthy participants from the projects 1000 Genomes and ExAC. Somatic missense mutations are also significantly enriched in PPI interfaces compared to noninterfaces in 10,861 tumor exomes. We computationally identified 470 putative oncoPPIs in a pan-cancer analysis and demonstrate that oncoPPIs are highly correlated with patient survival and drug resistance/sensitivity. We experimentally validate the network effects of 13 oncoPPIs using a systematic binary interaction assay, and also demonstrate the functional consequences of two of these on tumor cell growth. In summary, this human interactome network framework provides a powerful tool for prioritization of alleles with PPI-perturbing mutations to inform pathobiological mechanism- and genotype-based therapeutic discovery.

Atomistic Characterization of Gramicidin Channel Formation.

We investigated gramicidin A (gA) subunit dimerization in lipid bilayers using microsecond-long replica-exchange umbrella sampling simulations, millisecond-long unbiased molecular dynamics simulations, and machine learning. Our simulations led to a dimer structure that is indistinguishable from the experimentally determined gA channel structures, with the two gA subunits joined by six hydrogen bonds (6HB). The simulations also uncovered two additional dimer structures, with different gA-gA stacking orientations that were stabilized by four or two hydrogen bonds (4HB or 2HB). When examining the temporal evolution of the dimerization, we found that two bilayer-inserted gA subunits can form the 6HB dimer directly, with no discernible intermediate states, as well as through paths that involve the 2HB and 4HB dimers.

Assessing the Perturbing Effects of Drugs on Lipid Bilayers Using Gramicidin Channel-Based In Silico and In Vitro Assays.

Partitioning of bioactive molecules, including drugs, into cell membranes may produce indiscriminate changes in membrane protein function. As a guide to safe drug development, it therefore becomes important to be able to predict the bilayer-perturbing potency of hydrophobic/amphiphilic drugs candidates. Toward this end, we exploited gramicidin channels as molecular force probes and developed in silico and in vitro assays to measure drugs' bilayer-modifying potency. We examined eight drug-like molecules that were found to enhance or suppress gramicidin channel function in a thick 1,2-dierucoyl-sn-glycero-3-phosphocholine (DC22:1PC) but not in thin 1,2-dioleoyl-sn-glycero-3-phosphocholine (DC18:1PC) lipid bilayer. The mechanism underlying this difference was attributable to the changes in gramicidin dimerization free energy by drug-induced perturbations of lipid bilayer physical properties and bilayer-gramicidin interactions. The combined in silico and in vitro approaches, which allow for predicting the perturbing effects of drug candidates on membrane protein function, have implications for preclinical drug safety assessment.