Robert D. Malmstrom

Allostery through the computational microscope: cAMP activation of a canonical signalling domain

Ligand-induced protein allostery plays a central role in modulating cellular signaling pathways. Here, using the conserved cyclic-nucleotide binding domain of protein kinase A’s (PKA) regulatory subunit as a prototype signaling unit, we combine long-timescale, all-atom molecular dynamics simulations with Markov state models to elucidate the conformational ensembles of PKA’s cyclic-nucleotide binding domain A for the cAMP-free (apo) and cAMP-bound states. We find that both systems exhibit shallow free-energy landscapes that link functional states through multiple transition pathways. This observation suggests conformational selection as the general mechanism of allostery in this canonical signaling domain. Further, we expose the propagation of the allosteric signal through key structural motifs in the cyclic-nucleotide binding domain and explore the role of kinetics in its function. Our approach integrates disparate lines of experimental data into one cohesive framework to understand structure, dynamics, and function in complex biological systems.

Emerging Computational Methods for the Rational Discovery of Allosteric Drugs

Allosteric drug development holds promise for delivering medicines that are more selective and less toxic than those that target orthosteric sites. To date, the discovery of allosteric binding sites and lead compounds has been mostly serendipitous, achieved through high-throughput screening. Over the past decade, structural data has become more readily available for larger protein systems and more membrane protein classes (e.g., GPCRs and ion channels), which are common allosteric drug targets. In parallel, improved simulation methods now provide better atomistic understanding of the protein dynamics and cooperative motions that are critical to allosteric mechanisms. As a result of these advances, the field of predictive allosteric drug development is now on the cusp of a new era of rational structure-based computational methods. Here, we review algorithms that predict allosteric sites based on sequence data and molecular dynamics simulations, describe tools that assess the druggability of these pockets, and discuss how Markov state models and topology analyses provide insight into the relationship between protein dynamics and allosteric drug binding. In each section, we first provide an overview of the various method classes before describing relevant algorithms and software packages.

On the Application of Molecular-Dynamics Based Markov State Models to Functional Proteins.

Owing to recent developments in computational algorithms and architectures, it is now computationally tractable to explore biologically relevant, equilibrium dynamics of realistically sized functional proteins using all-atom molecular dynamics simulations. Molecular dynamics simulations coupled with Markov state models is a nascent but rapidly growing technology that is enabling robust exploration of equilibrium dynamics. The objective of this work is to explore the challenges of coupling molecular dynamics simulations and Markov state models in the study of functional proteins. Using recent studies as a framework, we explore progress in sampling, model building, model selection, and coarse-grained analysis of models. Our goal is to highlight some of the current challenges in applying Markov state models to realistically sized proteins and spur discussion on advances in the field.

Bridging scales through multiscale modeling: a case study on protein kinase A

The goal of multiscale modeling in biology is to use structurally based physico-chemical models to integrate across temporal and spatial scales of biology and thereby improve mechanistic understanding of, for example, how a single mutation can alter organism-scale phenotypes. This approach may also inform therapeutic strategies or identify candidate drug targets that might otherwise have been overlooked. However, in many cases, it remains unclear how best to synthesize information obtained from various scales and analysis approaches, such as atomistic molecular models, Markov state models (MSM), subcellular network models, and whole cell models. In this paper, we use protein kinase A (PKA) activation as a case study to explore how computational methods that model different physical scales can complement each other and integrate into an improved multiscale representation of the biological mechanisms. Using measured crystal structures, we show how molecular dynamics (MD) simulations coupled with atomic-scale MSMs can provide conformations for Brownian dynamics (BD) simulations to feed transitional states and kinetic parameters into protein-scale MSMs. We discuss how milestoning can give reaction probabilities and forward-rate constants of cAMP association events by seamlessly integrating MD and BD simulation scales. These rate constants coupled with MSMs provide a robust representation of the free energy landscape, enabling access to kinetic, and thermodynamic parameters unavailable from current experimental data. These approaches have helped to illuminate the cooperative nature of PKA activation in response to distinct cAMP binding events. Collectively, this approach exemplifies a general strategy for multiscale model development that is applicable to a wide range of biological problems.

Progress towards Automated Kepler Scientific Workflows for Computer-aided Drug Discovery and Molecular Simulations

We describe the development of automated workflows that support computed-aided drug discovery (CADD) and molecular dynamics (MD) simulations and are included as part of the National Biomedical Computational Resource (NBCR). The main workflow components include: file-management tasks, ligand force field parameterization, receptor-ligand molecular dynamics (MD) simulations, job submission and monitoring on relevant high-performance computing (HPC) resources, receptor structural clustering, virtual screening (VS), and statistical analyses of the VS results. The workflows aim to standardize simulation and analysis and promote best practices within the molecular simulation and CADD communities. Each component is developed as a stand-alone workflow, which allows easy integration into larger frameworks built to suit user needs, while remaining intuitive and easy to extend.

Electrostatic Interactions as Mediators in the Allosteric Activation of Protein Kinase A RIα

Close-range electrostatic interactions that form salt bridges are key components of protein stability. Here we investigate the role of these charged interactions in modulating the allosteric activation of protein kinase A (PKA) via computational and experimental mutational studies of a conserved basic patch located in the regulatory subunits B/C helix. Molecular dynamics simulations evidenced the presence of an extended network of fluctuating salt bridges spanning the helix and connecting the two cAMP binding domains in its extremities. Distinct changes in the flexibility and conformational free energy landscape induced by the separate mutations of Arg239 and Arg241 suggested alteration of cAMP-induced allosteric activation and were verified through in vitro fluorescence polarization assays. These observations suggest a mechanical aspect to the allosteric transition of PKA, with Arg239 and Arg241 acting in competition to promote the transition between the two protein functional states. The simulations also provide a molecular explanation for the essential role of Arg241 in allowing cooperative activation, by evidencing the existence of a stable interdomain salt bridge with Asp267. Our integrated approach points to the role of salt bridges not only in protein stability but also in promoting conformational transition and function.

A Kepler Workflow Tool for Reproducible AMBER GPU Molecular Dynamics

With the drive toward high throughput molecular dynamics (MD) simulations involving ever-greater numbers of simulation replicates run for longer, biologically relevant timescales (microseconds), the need for improved computational methods that facilitate fully automated MD workflows gains more importance. Here we report the development of an automated workflow tool to perform AMBER GPU MD simulations. Our workflow tool capitalizes on the capabilities of the Kepler platform to deliver a flexible, intuitive, and user-friendly environment and the AMBER GPU code for a robust and high-performance simulation engine. Additionally, the workflow tool reduces user input time by automating repetitive processes and facilitates access to GPU clusters, whose high-performance processing power makes simulations of large numerical scale possible. The presented workflow tool facilitates the management and deployment of large sets of MD simulations on heterogeneous computing resources. The workflow tool also performs systematic analysis on the simulation outputs and enhances simulation reproducibility, execution scalability, and MD method development including benchmarking and validation.

Seeing the Unseen: Sampling the Excited State of T4 Lysozyme L99A with Simulations on the Anton Supercomputer

Proteins are the workhorses of the cell, converting small molecules into energy, harnessing energy for macromolecular synthesis, and interacting with one another to relay important cellular messages. All of these tasks are performed through the coordinated movement of a proteins atoms, which guide a protein from one conformational state to another. While experimental techniques provide insight into protein atomic motions, experiments cannot track all atomic-level motions across the nanosecond (ns) to millisecond (ms) time regimes. Even for proteins like T4 lysozyme, which have been extensively characterized (1-3), mysteries remain about the mechanism of converting between different protein states. The Leu99⇒Ala (L99A) mutant of T4 lysozyme is a model system for studying rarely seen, hidden excited states, even though the structure of its excited state has never been solved (4-9). Here we have sampled the excited state of L99A at atomic resolution with computational simulation on the Anton supercomputer (10). In this simulation, phenylalanine 114 (F114) and helices F and G undergo conformational changes that are indicative of a transition to the excited state. This MD-generated excited state also reproduces known excited state chemical shifts (R2 = 0.95) and agrees with multiple lines of experiment (4, 7, 11, 12). Our results of the L99A excited state cohere decades of data on this invisible conformation. We anticipate that this structure will provide experimentalists studying this protein with the tools to understand the atomic level details underlying L99A dynamics.