MiMiC: A high-performance framework for multiscale molecular dynamics simulations.

MiMiC is a framework for performing multiscale simulations in which loosely coupled external programs describe individual subsystems at different resolutions and levels of theory. To make it highly efficient and flexible, we adopt an interoperable approach based on a multiple-program multiple-data (MPMD) paradigm, serving as an intermediary responsible for fast data exchange and interactions between the subsystems. The main goal of MiMiC is to avoid interfering with the underlying parallelization of the external programs, including the operability on hybrid architectures (e.g., CPU/GPU), and keep their setup and execution as close as possible to the original. At the moment, MiMiC offers an efficient implementation of electrostatic embedding quantum mechanics/molecular mechanics (QM/MM) that has demonstrated unprecedented parallel scaling in simulations of large biomolecules using CPMD and GROMACS as QM and MM engines, respectively. However, as it is designed for high flexibility with general multiscale models in mind, it can be straightforwardly extended beyond QM/MM. In this article, we illustrate the software design and the features of the framework, which make it a compelling choice for multiscale simulations in the upcoming era of exascale high-performance computing.

Impact of quantum and neuromorphic computing on biomolecular simulations: Current status and perspectives.

New high-performance computing architectures are becoming operative, in addition to exascale computers. Quantum computers (QC) solve optimization problems with unprecedented efficiency and speed, while neuromorphic hardware (NMH) simulates neural network dynamics. Albeit, at the moment, both find no practical use in all atom biomolecular simulations, QC might be exploited in the not-too-far future to simulate systems for which electronic degrees of freedom play a key and intricate role for biological function, whereas NMH might accelerate molecular dynamics simulations with low energy consumption. Machine learning and artificial intelligence algorithms running on NMH and QC could assist in the analysis of data and speed up research. If these implementations are successful, modular supercomputing could further dramatically enhance the overall computing capacity by combining highly optimized software tools into workflows, linking these architectures to exascale computers.

Effective data-driven collective variables for free energy calculations from metadynamics of paths.

A variety of enhanced sampling (ES) methods predict multidimensional free energy landscapes associated with biological and other molecular processes as a function of a few selected collective variables (CVs). The accuracy of these methods is crucially dependent on the ability of the chosen CVs to capture the relevant slow degrees of freedom of the system. For complex processes, finding such CVs is the real challenge. Machine learning (ML) CVs offer, in principle, a solution to handle this problem. However, these methods rely on the availability of high-quality datasets-ideally incorporating information about physical pathways and transition states-which are difficult to access, therefore greatly limiting their domain of application. Here, we demonstrate how these datasets can be generated by means of ES simulations in trajectory space via the metadynamics of paths algorithm. The approach is expected to provide a general and efficient way to generate efficient ML-based CVs for the fast prediction of free energy landscapes in ES simulations. We demonstrate our approach with two numerical examples, a 2D model potential and the isomerization of alanine dipeptide, using deep targeted discriminant analysis as our ML-based CV of choice.

All-Atom Biomolecular Simulation in the Exascale Era.

Exascale supercomputers have opened the door to dynamic simulations, facilitated by AI/ML techniques, that model biomolecular motions over unprecedented length and time scales. This new capability holds the potential to revolutionize our understanding of fundamental biological processes. Here we report on some of the major advances that were discussed at a recent CECAM workshop in Pisa, Italy, on the topic with a primary focus on atomic-level simulations. First, we highlight examples of current large-scale biomolecular simulations and the future possibilities enabled by crossing the exascale threshold. Next, we discuss challenges to be overcome in optimizing the usage of these powerful resources. Finally, we close by listing several grand challenge problems that could be investigated with this new computer architecture.

Unexpected Single-Ligand Occupancy and Negative Cooperativity in the SARS-CoV-2 Main Protease.

Many homodimeric enzymes tune their functions by exploiting either negative or positive cooperativity between subunits. In the SARS-CoV-2 Main protease (Mpro) homodimer, the latter has been suggested by symmetry in most of the 500 reported protease/ligand complex structures solved by macromolecular crystallography (MX). Here we apply the latter to both covalent and noncovalent ligands in complex with Mpro. Strikingly, our experiments show that the occupation of both active sites of the dimer originates from an excess of ligands. Indeed, cocrystals obtained using a 1:1 ligand/protomer stoichiometry lead to single occupation only. The empty binding site exhibits a catalytically inactive geometry in solution, as suggested by molecular dynamics simulations. Thus, Mpro operates through negative cooperativity with the asymmetric activity of the catalytic sites. This allows it to function with a wide range of substrate concentrations, making it resistant to saturation and potentially difficult to shut down, all properties advantageous for the virus' adaptability and resistance.

Kinetic and structural details of urease inactivation by thiuram disulphides.

This paper reports on the molecular details of the reactivity of urease, a nickel-dependent enzyme that catalyses the last step of organic nitrogen mineralization, with thiuram disulphides, a class of molecules known to inactivate the enzyme with high efficacy but for which the mechanism of action had not been yet established. IC50 values of tetramethylthiuram disulphide (TMTD or Thiram) and tetraethylthiuram disulphide (TETD or Disulfiram) in the low micromolar range were determined for plant and bacterial ureases. The X-ray crystal structure of Sporosarcina pasteurii urease inactivated by Thiram, determined at 1.68 Å resolution, revealed the presence of a covalent modification of the catalytically essential cysteine residue. This is located on the flexible flap that modulates the size of the active site channel and cavity. Formation of a Cys-S-S-C(S)-N(CH3)2 functionality responsible for enzyme inactivation was observed. Quantum-mechanical calculations carried out to rationalise the large reactivity of the active site cysteine support the view that a conserved histidine residue, adjacent to the cysteine in the active site flap, modulates the charge and electron density along the thiol SH bond by shifting electrons towards the sulphur atom and rendering the thiol proton more reactive. We speculate that this proton could be transferred to the nickel-coordinated urea amide group to yield a molecule of ammonia from the generated Curea-NH3+ functionality during catalysis.

Physical Chemistry of Chloroquine Permeation through the Cell Membrane with Atomistic Detail.

We provide a molecular-level description of the thermodynamics and mechanistic aspects of drug permeation through the cell membrane. As a case study, we considered the antimalaria FDA approved drug chloroquine. Molecular dynamics simulations of the molecule (in its neutral and protonated form) were performed in the presence of different lipid bilayers, with the aim of uncovering key aspects of the permeation process, a fundamental step for the drug's action. Free energy values obtained by well-tempered metadynamics simulations suggest that the neutral form is the only permeating protomer, consistent with experimental data. H-bond interactions of the drug with water molecules and membrane headgroups play a crucial role for permeation. The presence of the transmembrane potential, investigated here for the first time in a drug permeation study, does not qualitatively affect these conclusions.

Free energies at QM accuracy from force fields via multimap targeted estimation.

Accurate predictions of ligand binding affinities would greatly accelerate the first stages of drug discovery campaigns. However, using highly accurate interatomic potentials based on quantum mechanics (QM) in free energy methods has been so far largely unfeasible due to their prohibitive computational cost. Here, we present an efficient method to compute QM free energies from simulations using cheap reference potentials, such as force fields (FFs). This task has traditionally been out of reach due to the slow convergence of computing the correction from the FF to the QM potential. To overcome this bottleneck, we generalize targeted free energy methods to employ multiple maps-implemented with normalizing flow neural networks (NNs)-that maximize the overlap between the distributions. Critically, the method requires neither a separate expensive training phase for the NNs nor samples from the QM potential. We further propose a one-epoch learning policy to efficiently avoid overfitting, and we combine our approach with enhanced sampling strategies to overcome the pervasive problem of poor convergence due to slow degrees of freedom. On the drug-like molecules in the HiPen dataset, the method accelerates the calculation of the free energy difference of switching from an FF to a DFTB3 potential by three orders of magnitude compared to standard free energy perturbation and by a factor of eight compared to previously published nonequilibrium calculations. Our results suggest that our method, in combination with efficient QM/MM calculations, may be used in lead optimization campaigns in drug discovery and to study protein-ligand molecular recognition processes.

Low Molecular Weight Inhibitors Targeting the RNA-Binding Protein HuR.

The RNA-binding protein human antigen R (HuR) regulates stability, translation, and nucleus-to-cytoplasm shuttling of its target mRNAs. This protein has been progressively recognized as a relevant therapeutic target for several pathologies, like cancer, neurodegeneration, as well as inflammation. Inhibitors of mRNA binding to HuR might thus be beneficial against a variety of diseases. Here, we present the rational identification of structurally novel HuR inhibitors. In particular, by combining chemoinformatic approaches, high-throughput virtual screening, and RNA-protein pulldown assays, we demonstrate that the 4-(2-(2,4,6-trioxotetrahydropyrimidin-5(2H)-ylidene)hydrazineyl)benzoate ligand exhibits a dose-dependent HuR inhibition effect in binding experiments. Importantly, the chemical scaffold is new with respect to the currently known HuR inhibitors, opening up a new avenue for the design of pharmaceutical agents targeting this important protein.

AI-based identification of therapeutic agents targeting GPCRs: introducing ligand type classifiers and systems biology.

Identifying ligands targeting G protein coupled receptors (GPCRs) with novel chemotypes other than the physiological ligands is a challenge for in silico screening campaigns. Here we present an approach that identifies novel chemotype ligands by combining structural data with a random forest agonist/antagonist classifier and a signal-transduction kinetic model. As a test case, we apply this approach to identify novel antagonists of the human adenosine transmembrane receptor type 2A, an attractive target against Parkinson's disease and cancer. The identified antagonists were tested here in a radio ligand binding assay. Among those, we found a promising ligand whose chemotype differs significantly from all so-far reported antagonists, with a binding affinity of 310 ± 23.4 nM. Thus, our protocol emerges as a powerful approach to identify promising ligand candidates with novel chemotypes while preserving antagonistic potential and affinity in the nanomolar range.

Fluoride permeation mechanism of the Fluc channel in liposomes revealed by solid-state NMR.

Solid-state nuclear magnetic resonance (ssNMR) methods can probe the motions of membrane proteins in liposomes at the atomic level and propel the understanding of biomolecular processes for which static structures cannot provide a satisfactory description. In this work, we report our study on the fluoride channel Fluc-Ec1 in phospholipid bilayers based on ssNMR and molecular dynamics simulations. Previously unidentified fluoride binding sites in the aqueous vestibules were experimentally verified by 19F-detected ssNMR. One of the two fluoride binding sites in the polar track was identified as a water molecule by 1H-detected ssNMR. Meanwhile, a dynamic hotspot at loop 1 was observed by comparing the spectra of wild-type Fluc-Ec1 in variant buffer conditions or with its mutants. Therefore, we propose that fluoride conduction in the Fluc channel occurs via a "water-mediated knock-on" permeation mechanism and that loop 1 is a key molecular determinant for channel gating.

Drug Design in the Exascale Era: A Perspective from Massively Parallel QM/MM Simulations.

The initial phases of drug discovery - in silico drug design - could benefit from first principle Quantum Mechanics/Molecular Mechanics (QM/MM) molecular dynamics (MD) simulations in explicit solvent, yet many applications are currently limited by the short time scales that this approach can cover. Developing scalable first principle QM/MM MD interfaces fully exploiting current exascale machines - so far an unmet and crucial goal - will help overcome this problem, opening the way to the study of the thermodynamics and kinetics of ligand binding to protein with first principle accuracy. Here, taking two relevant case studies involving the interactions of ligands with rather large enzymes, we showcase the use of our recently developed massively scalable Multiscale Modeling in Computational Chemistry (MiMiC) QM/MM framework (currently using DFT to describe the QM region) to investigate reactions and ligand binding in enzymes of pharmacological relevance. We also demonstrate for the first time strong scaling of MiMiC-QM/MM MD simulations with parallel efficiency of ∼70% up to >80,000 cores. Thus, among many others, the MiMiC interface represents a promising candidate toward exascale applications by combining machine learning with statistical mechanics based algorithms tailored for exascale supercomputers.

Metadynamics simulations of ligands binding to protein surfaces: a novel tool for rational drug design.

Structure-based drug design protocols may encounter difficulties to investigate poses when the biomolecular targets do not exhibit typical binding pockets. In this study, by providing two concrete examples from our labs, we suggest that the combination of metadynamics free energy methods (validated against affinity measurements), along with experimental structural information (by X-ray crystallography and NMR), can help to identify the poses of ligands on protein surfaces. The simulation workflow proposed here was implemented in a widely used code, namely GROMACS, and it could straightforwardly be applied to various drug-design campaigns targeting ligands' binding to protein surfaces.

Molecular determinants of acrylamide neurotoxicity through covalent docking.

Acrylamide (ACR) is formed during food processing by Maillard reaction between sugars and proteins at high temperatures. It is also used in many industries, from water waste treatment to manufacture of paper, fabrics, dyes and cosmetics. Unfortunately, cumulative exposure to acrylamide, either from diet or at the workplace, may result in neurotoxicity. Such adverse effects arise from covalent adducts formed between acrylamide and cysteine residues of several neuronal proteins via a Michael addition reaction. The molecular determinants of acrylamide reactivity and its impact on protein function are not completely understood. Here we have compiled a list of acrylamide protein targets reported so far in the literature in connection with neurotoxicity and performed a systematic covalent docking study. Our results indicate that acrylamide binding to cysteine is favored in the presence of nearby positively charged amino acids, such as lysines and arginines. For proteins with more than one reactive Cys, docking scores were able to discriminate between the primary ACR modification site and secondary sites modified only at high ACR concentrations. Therefore, docking scores emerge as a potential filter to predict Cys reactivity against acrylamide. Inspection of the ACR-protein complex structures provides insights into the putative functional consequences of ACR modification, especially for non-enzyme proteins. Based on our study, covalent docking is a promising computational tool to predict other potential protein targets mediating acrylamide neurotoxicity.

MiMiCPy: An Efficient Toolkit for MiMiC-Based QM/MM Simulations.

MiMiC is a highly flexible, extremely scalable multiscale modeling framework. It couples the CPMD (quantum mechanics, QM) and GROMACS (molecular mechanics, MM) codes. The code requires preparing separate input files for the two programs with a selection of the QM region. This can be a tedious procedure prone to human error, especially when dealing with large QM regions. Here, we present MiMiCPy, a user-friendly tool that automatizes the preparation of MiMiC input files. It is written in Python 3 with an object-oriented approach. The main subcommand PrepQM can be used to generate MiMiC inputs directly from the command line or through a PyMOL/VMD plugin for visually selecting the QM region. Many other subcommands are also provided for debugging and fixing MiMiC input files. MiMiCPy is designed with a modular structure that allows seamless extensions to new program formats depending on the requirements of MiMiC.

Mechanism of Calcium Permeation in a Glutamate Receptor Ion Channel.

The α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) are neurotransmitter-activated cation channels ubiquitously expressed in vertebrate brains. The regulation of calcium flux through the channel pore by RNA-editing is linked to synaptic plasticity while excessive calcium influx poses a risk for neurodegeneration. Unfortunately, the molecular mechanisms underlying this key process are mostly unknown. Here, we investigated calcium conduction in calcium-permeable AMPAR using Molecular Dynamics (MD) simulations with recently introduced multisite force-field parameters for Ca2+. Our calculations are consistent with experiment and explain the distinct calcium permeability in different RNA-edited forms of GluA2. For one of the identified metal binding sites, multiscale Quantum Mechanics/Molecular Mechanics (QM/MM) simulations further validated the results from MD and revealed small but reproducible charge transfer between the metal ion and its first solvation shell. In addition, the ion occupancy derived from MD simulations independently reproduced the Ca2+ binding profile in an X-ray structure of an NaK channel mimicking the AMPAR selectivity filter. This integrated study comprising X-ray crystallography, multisite MD, and multiscale QM/MM simulations provides unprecedented insights into Ca2+ permeation mechanisms in AMPARs, and paves the way for studying other biological processes in which Ca2+ plays a pivotal role.

Predictions of the Poses and Affinity of a Ligand over the Entire Surface of a NEET Protein: The Case of Human MitoNEET.

Human NEET proteins contain two [2Fe-2S] iron-sulfur clusters, bound to three Cys residues and one His residue. They exist in two redox states. Recently, these proteins have revealed themselves as attractive drug targets for mitochondrial dysfunction-related diseases, such as type 2 diabetes, Wolfram syndrome 2, and cancers. Unfortunately, the lack of information and mechanistic understanding of ligands binding to the whole functional, cytoplasmatic domain has limited rational drug design approaches. Here, we use an enhanced sampling technique, volume-based metadynamics, recently developed by a team involving some of us, to predict the poses and affinity of the 2-benzamido-4-(1,2,3,4-tetrahydronaphthalen-2-yl)-thiophene-3-carboxylate ligand to the entire surface of the cytoplasmatic domain of the human NEET protein mitoNEET (mNT) in an aqueous solution. The calculations, based on the recently published X-ray structure of the complex, are consistent with the measured affinity. The calculated free energy landscape revealed that the ligand can bind in multiple sites and with poses other than the one found in the X-ray. This difference is likely to be caused by crystal packing effects that allow the ligand to interact with multiple adjacent NEET protein copies. Such extra contacts are of course absent in the solution; therefore, the X-ray pose is only transient in our calculations, where the binding free energy correlates with the number of contacts. We further evaluated how the reduction and protonation of the Fe-bound histidine, as well as temperature, can affect ligand binding. Both such modifications introduce the possibility for the ligand to bind in an area of the protein other than the one observed in the X-ray, with no or little impact on affinity. Overall, our study can provide insights on the molecular recognition mechanisms of ligand binding to mNT in different oxidative conditions, possibly helping rational drug design of NEET ligands.

Molecular Dynamics and Structural Studies of Zinc Chloroquine Complexes.

Chloroquine (CQ) is a first-choice drug against malaria and autoimmune diseases. It has been co-administered with zinc against SARS-CoV-2 and soon dismissed because of safety issues. The structural features of Zn-CQ complexes and the effect of CQ on zinc distribution in cells are poorly known. In this study, state-of-the-art computations combined with experiments were leveraged to solve the structural determinants of zinc-CQ interactions in solution and the solid state. NMR, ESI-MS, and X-ray absorption and diffraction methods were combined with ab initio molecular dynamics calculations to address the kinetic lability of this complex. Within the physiological pH range, CQ binds Zn2+ through the quinoline ring nitrogen, forming [Zn(CQH)Clx(H2O)3-x](3+)-x (x = 0, 1, 2, and 3) tetrahedral complexes. The Zn(CQH)Cl3 species is stable at neutral pH and at high chloride concentrations typical of the extracellular medium, but metal coordination is lost at a moderately low pH as in the lysosomal lumen. The pentacoordinate complex [Zn(CQH)(H2O)4]3+ may exist in the absence of chloride. This in vitro/in silico approach can be extended to other metal-targeting drugs and bioinorganic systems.

Proton Transfers to DNA in Native Electrospray Ionization Mass Spectrometry: A Quantum Mechanics/Molecular Mechanics Study.

Native electrospray ionization-ion mobility mass spectrometry (N-ESI/IM-MS) is a powerful approach for low-resolution structural studies of DNAs in the free state and in complex with ligands. Solvent vaporization is coupled with proton transfers from ammonium ions to the DNA, resulting in a reduction of the DNA charge. Here we provide insight into these processes by classical molecular dynamics and quantum mechanics/molecular mechanics free energy calculations on the d(GpCpGpApApGpC) heptamer, for which a wealth of experiments is available. Our multiscale simulations, consistent with experimental data, reveal a highly complex scenario. The proton either sits on one of the molecules or is fully delocalized on both, depending on the level of hydration of the analytes and the size of the droplets formed during the electrospray experiments. This work complements our previous study of the intramolecular proton transfer on the same heptamer occurring after the processes studied here, and together, they provide a first molecular view of proton transfer in N-ESI/IM-MS.