Significance of Molecular Dynamics Simulations for Life Sciences

Advancing biomolecular simulation through exascale HPC, AI and quantum computing

Biomolecular simulation can act as both a digital microscope and a crystal ball; offering the potential for a deeper understanding of experimental observations whilst also presenting a forward-looking avenue for the in silico design and evaluation of hitherto unsynthesized compounds. Indeed, as the intricacy of our scientific inquiries has grown, so too has the computational prowess we seek to deploy in our pursuit of answers. As we enter the Exascale era, this mini-review surveys the computational landscape from both the point of view of the development of new and ever more powerful systems, and the simulations that are run on them. Moreover, as we stand on the cusp of a transformative phase in computational biology, this article offers a contemplative glance into the future, speculating on the profound implications of artificial intelligence and quantum computing for large-scale biomolecular simulations.

Fluoride Transport and Inhibition Across CLC Transporters

The Chloride Channel (CLC) family includes proton-coupled chloride and fluoride transporters. Despite their similar protein architecture, the former exchange two chloride ions for each proton and are inhibited by fluoride, whereas the latter efficiently transport one fluoride in exchange for one proton. The combination of structural, mutagenesis, and functional experiments with molecular simulations has pinpointed several amino acid changes in the permeation pathway that capitalize on the different chemical properties of chloride and fluoride to fine-tune protein function. Here we summarize recent findings on fluoride inhibition and transport in the two prototypical members of the CLC family, the chloride/proton transporter from Escherichia coli (CLC-ec1) and the fluoride/proton transporter from Enterococcus casseliflavus (CLCF-eca).

Modeling the Homologous Recombination Process: Methods, Successes and Challenges

Homologous recombination (HR) is a fundamental process common to all species. HR aims to faithfully repair DNA double strand breaks. HR involves the formation of nucleoprotein filaments on DNA single strands (ssDNA) resected from the break. The nucleoprotein filaments search for homologous regions in the genome and promote strand exchange with the ssDNA homologous region in an unbroken copy of the genome. HR has been the object of intensive studies for decades. Because multi-scale dynamics is a fundamental aspect of this process, studying HR is highly challenging, both experimentally and using computational approaches. Nevertheless, knowledge has built up over the years and has recently progressed at an accelerated pace, borne by increasingly focused investigations using new techniques such as single molecule approaches. Linking this knowledge to the atomic structure of the nucleoprotein filament systems and the succession of unstable, transient intermediate steps that takes place during the HR process remains a challenge; modeling retains a very strong role in bridging the gap between structures that are stable enough to be observed and in exploring transition paths between these structures. However, working on ever-changing long filament systems submitted to kinetic processes is full of pitfalls. This review presents the modeling tools that are used in such studies, their possibilities and limitations, and reviews the advances in the knowledge of the HR process that have been obtained through modeling. Notably, we will emphasize how cooperative behavior in the HR nucleoprotein filament enables modeling to produce reliable information.

Using classifiers to understand coarse-grained models and their fidelity with the underlying all-atom systems

Bottom-up coarse-grained (CG) molecular dynamics models are parameterized using complex effective Hamiltonians. These models are typically optimized to approximate high dimensional data from atomistic simulations. However, human validation of these models is often limited to low dimensional statistics that do not necessarily differentiate between the CG model and said atomistic simulations. We propose that classification can be used to variationally estimate high dimensional error and that explainable machine learning can help convey this information to scientists. This approach is demonstrated using Shapley additive explanations and two CG protein models. This framework may also be valuable for ascertaining whether allosteric effects at the atomistic level are accurately propagated to a CG model.

Amphiphilic Nanointerface: Inducing the Interfacial Activation for Lipase

Graphene-based materials are widely used in the field of immobilized enzymes due to their easily tunable interfacial properties. We designed amphiphilic nanobiological interfaces between graphene oxide (GO) and lipase TL (Thermomyces lanuginosus) with tunable reduction degrees through molecular dynamics simulations and a facile chemical modulation, thus revealing the optimal interface for the interfacial activation of lipase TL and addressing the weakness of lipase TL, which exhibits weak catalytic activity due to an inconspicuous active site lid. It was demonstrated that the reduced graphene oxide (rGO) after 4 h of ascorbic acid reduction could boost the relative enzyme activity of lipase TL to reach 208%, which was 48% higher than the pristine GO and 120% higher than the rGO after 48 h of reduction. Moreover, TL-GO-4 h's tolerance against heat, organic solvent, and long-term storage environment was higher than that of free TL. The drawbacks of strong hydrophobic nanomaterials on lipase production were explored in depth with the help of molecular dynamics simulations, which explained the mechanism of enzyme activity enhancement. We demonstrated that nanomaterials with certain hydrophilicity could facilitate the lipase to undergo interfacial activation and improve its stability and protein loading rate, displaying the potential of the extensive application.

Accelerating COVID-19 Research Using Molecular Dynamics Simulation

The COVID-19 pandemic has emerged as a global medico-socio-economic disaster. Given the lack of effective therapeutics against SARS-CoV-2, scientists are racing to disseminate suggestions for rapidly deployable therapeutic options, including drug repurposing and repositioning strategies. Molecular dynamics (MD) simulations have provided the opportunity to make rational scientific breakthroughs in a time of crisis. Advancements in these technologies in recent years have become an indispensable tool for scientists studying protein structure, function, dynamics, interactions, and drug discovery. Integrating the structural data obtained from high-resolution methods with MD simulations has helped in comprehending the process of infection and pathogenesis, as well as the SARS-CoV-2 maturation in host cells, in a short duration of time. It has also guided us to identify and prioritize drug targets and new chemical entities, and to repurpose drugs. Here, we discuss how MD simulation has been explored by the scientific community to accelerate and guide translational research on SARS-CoV-2 in the past year. We have also considered future research directions for researchers, where MD simulations can help fill the existing gaps in COVID-19 research.

Biomolecular Modeling and Simulation: A Prospering Multidisciplinary Field

We reassess progress in the field of biomolecular modeling and simulation, following up on our perspective published in 2011. By reviewing metrics for the field's productivity and providing examples of success, we underscore the productive phase of the field, whose short-term expectations were overestimated and long-term effects underestimated. Such successes include prediction of structures and mechanisms; generation of new insights into biomolecular activity; and thriving collaborations between modeling and experimentation, including experiments driven by modeling. We also discuss the impact of field exercises and web games on the field's progress. Overall, we note tremendous success by the biomolecular modeling community in utilization of computer power; improvement in force fields; and development and application of new algorithms, notably machine learning and artificial intelligence. The combined advances are enhancing the accuracy andscope of modeling and simulation, establishing an exemplary discipline where experiment and theory or simulations are full partners.

Binding Mode Multiplicity and Multiscale Chirality in the Supramolecular Assembly of DNA and a π-Conjugated Polymer

Water-soluble π-conjugated polymers are increasingly considered for DNA biosensing. However, the conformational rearrangement, supramolecular organization and dynamics upon interaction with DNA have been overlooked, which prevents the rational design of such detection tools. To elucidate the binding of a cationic polythiophene (CPT) to DNA with atomistic resolution, we performed molecular simulations of their supramolecular assembly. Comparison of replicated simulations show a multiplicity of CPT binding geometries that contribute to the wrapping of CPT around DNA. The different binding geometries are stabilized by both electrostatic interactions between CPT lateral cations and DNA phosphodiesters and van der Waals interactions between the CPT backbone and the DNA grooves. Simulated circular dichroism (CD) spectra show that the induced CD signal stems from a conserved geometrical feature across the replicated simulations, i. e. the presence of segments of syn configurations between thiophene units along the CPT chain. At the macromolecular scale, we inspected the different shapes related to the CPT binding modes around the DNA through symmetry metrics. Altogether, molecular dynamics (MD) simulations, model Hamiltonian calculations of the CD spectra, and symmetry indices provide insights into the origin of induced chirality from the atomic to the macromolecular scale. Our multidisciplinary approach points out the hierarchical aspect of CPT chiral organization induced by DNA.

Adversarial-residual-coarse-graining: Applying machine learning theory to systematic molecular coarse-graining

We utilize connections between molecular coarse-graining (CG) approaches and implicit generative models in machine learning to describe a new framework for systematic molecular CG. Focus is placed on the formalism encompassing generative adversarial networks. The resulting method enables a variety of model parameterization strategies, some of which show similarity to previous CG methods. We demonstrate that the resulting framework can rigorously parameterize CG models containing CG sites with no prescribed connection to the reference atomistic system (termed virtual sites); however, this advantage is offset by the lack of a closed-form expression for the CG Hamiltonian at the resolution obtained after integration over the virtual CG sites. Computational examples are provided for cases in which these methods ideally return identical parameters as relative entropy minimization CG but where traditional relative entropy minimization CG optimization equations are not applicable.

Application of molecular dynamics simulation to improve the theoretical prediction for collisional cross section of aromatic compounds with long alkyl chains in crude oils

RATIONALE

Molecular dynamics (MD) simulations with finite temperature were performed to improve the theoretical prediction of collisional cross section (CCS) values, especially for aromatic compounds containing long alkyl chains.

METHODS

In this study, the CCS values of 11 aromatic compounds with long alkyl chains were calculated by MD simulations while considering internal energy at 300, 500, and 700 K, and the results were compared with experimentally determined values.

RESULTS

The CCS values calculated at higher energies showed better agreement with the experimental values. Polycyclic aromatic hydrocarbons (PAHs) such as pentacene and benz[b]anthracene were also investigated, and better agreement between the theoretical and experimental results was observed when higher temperature (or higher internal energy) was considered.

CONCLUSIONS

The data presented in this study show that the internal degrees of freedom of ions must be considered to accurately predict the CCS values of aromatic compounds with a flexible structure measured by ion mobility mass spectrometry.

Modeling and Simulation of Oligonucleotide Hybrids: Outlining a Strategy

Molecular dynamics simulations with a state-of-the-art force field provide an atomistic detailed description of the structural and thermodynamic features of biomolecules. Effects of chemical modifications and of the environment such as sequence, solvent, and ionic strength can explicitly be taken into account. Molecular simulation techniques can also provide insight in change in binding affinity, in protonation (pK_a shift) and tautomeric propensity due to changes in the environment or in the molecular system. The quality and reliability of a simulation depend on the quality of the force field and on the reproducibility of the data, and validation depends on the availability of suitable experimental data. Here, we describe the workflow to investigate oligonucleotide hybrids using molecular simulation including hardware and software information.

The Enigma of Amyloid Forming Proteins: Insights From Molecular Simulations

Molecular level insight into the interplay between protein sequence, structure, and conformational dynamics is crucial for the comprehensive understanding of protein folding, misfolding, and aggregation phenomena that are pertinent to the formation of amyloid fibrils implicated in several degenerative diseases. Computational modelling provides insight into protein behaviour at spatial and temporal resolution still largely outside the reach of experiments. Herein we present an account of our theoretical modelling research conducted in collaboration with several experimental groups where we explored the effects of local environment on the structure and aggregation propensity of several types of amyloidogenic peptides and proteins, including apolipoprotein C-II, insulin, amylin, and amyloid-β using a variety of computational approaches.

Recent advances in automated protein design and its future challenges

INTRODUCTION

Protein function is determined by protein structure which is in turn determined by the corresponding protein sequence. If the rules that cause a protein to adopt a particular structure are understood, it should be possible to refine or even redefine the function of a protein by working backwards from the desired structure to the sequence. Automated protein design attempts to calculate the effects of mutations computationally with the goal of more radical or complex transformations than are accessible by experimental techniques. Areas covered: The authors give a brief overview of the recent methodological advances in computer-aided protein design, showing how methodological choices affect final design and how automated protein design can be used to address problems considered beyond traditional protein engineering, including the creation of novel protein scaffolds for drug development. Also, the authors address specifically the future challenges in the development of automated protein design. Expert opinion: Automated protein design holds potential as a protein engineering technique, particularly in cases where screening by combinatorial mutagenesis is problematic. Considering solubility and immunogenicity issues, automated protein design is initially more likely to make an impact as a research tool for exploring basic biology in drug discovery than in the design of protein biologics.

Water-mediated interactions enable smooth substrate transport in a bacterial efflux pump

BACKGROUND

Efflux pumps of the Resistance-Nodulation-cell Division superfamily confer multi-drug resistance to Gram-negative bacteria. The most-studied polyspecific transporter belonging to this class is the inner-membrane trimeric antiporter AcrB of Escherichia coli. In previous studies, a functional rotation mechanism was proposed for its functioning, according to which the three monomers undergo concerted conformational changes facilitating the extrusion of substrates. However, the molecular determinants and the energetics of this mechanism still remain unknown, so its feasibility must be proven mechanistically.

METHODS

A computational protocol able to mimic the functional rotation mechanism in AcrB was developed. By using multi-bias molecular dynamics simulations we characterized the translocation of the substrate doxorubicin driven by conformational changes of the protein. In addition, we estimated for the first time the free energy profile associated to this process.

RESULTS

We provided a molecular view of the process in agreement with experimental data. Moreover, we showed that the conformational changes occurring in AcrB enable the formation of a layer of structured waters on the internal surface of the transport channel. This water layer, in turn, allows for a fairly constant hydration of the substrate, facilitating its diffusion over a smooth free energy profile.

CONCLUSIONS

Our findings reveal a new molecular mechanism of polyspecific transport whereby water contributes by screening potentially strong substrate-protein interactions.

GENERAL SIGNIFICANCE

We provided a mechanistic understanding of a fundamental process related to multi-drug transport. Our results can help rationalizing the behavior of other polyspecific transporters and designing compounds avoiding extrusion or inhibitors of efflux pumps.

Molecular Modeling of Multidrug Properties of Resistance Nodulation Division (RND) Transporters

Efflux pumps of the resistance nodulation division (RND) superfamily are among the major contributors to intrinsic and acquired multidrug resistance in Gram-negative bacteria. Structural information on AcrAB-TolC and MexAB-OprM, major efflux pumps of Escherichia coli and Pseudomonas aeruginosa respectively, boosted intensive research aimed at understanding the molecular mechanisms ruling the active extrusion processes. In particular, several studies were devoted to the understanding of the determinants behind the extraordinary broad specificity of the RND transporters AcrB and MexB. In this chapter, we discuss the ever-growing role computational methods have been playing in deciphering key structural and dynamical features of these transporters and of their interaction with substrates and inhibitors. We further discuss and illustrate examples from our lab of how molecular docking, homology modeling, all-atom molecular dynamics simulations and in silico free energy estimations can all together give precious insights into the processes of recognition and extrusion of substrates, as well as on the possible inhibition strategies.

LNA effects on DNA binding and conformation: from single strand to duplex and triplex structures

The anti-gene strategy is based on sequence-specific recognition of double-strand DNA by triplex forming (TFOs) or DNA strand invading oligonucleotides to modulate gene expression. To be efficient, the oligonucleotides (ONs) should target DNA selectively, with high affinity. Here we combined hybridization analysis and electrophoretic mobility shift assay with molecular dynamics (MD) simulations to better understand the underlying structural features of modified ONs in stabilizing duplex- and triplex structures. Particularly, we investigated the role played by the position and number of locked nucleic acid (LNA) substitutions in the ON when targeting a c-MYC or FXN (Frataxin) sequence. We found that LNA-containing single strand TFOs are conformationally pre-organized for major groove binding. Reduced content of LNA at consecutive positions at the 3'-end of a TFO destabilizes the triplex structure, whereas the presence of Twisted Intercalating Nucleic Acid (TINA) at the 3'-end of the TFO increases the rate and extent of triplex formation. A triplex-specific intercalating benzoquinoquinoxaline (BQQ) compound highly stabilizes LNA-containing triplex structures. Moreover, LNA-substitution in the duplex pyrimidine strand alters the double helix structure, affecting x-displacement, slide and twist favoring triplex formation through enhanced TFO major groove accommodation. Collectively, these findings should facilitate the design of potent anti-gene ONs.

Modeling Functional Motions of Biological Systems by Customized Natural Moves

Simulating the functional motions of biomolecular systems requires large computational resources. We introduce a computationally inexpensive protocol for the systematic testing of hypotheses regarding the dynamic behavior of proteins and nucleic acids. The protocol is based on natural move Monte Carlo, a highly efficient conformational sampling method with built-in customization capabilities that allows researchers to design and perform a large number of simulations to investigate functional motions in biological systems. We demonstrate the use of this protocol on both a protein and a DNA case study. Firstly, we investigate the plasticity of a class II major histocompatibility complex in the absence of a bound peptide. Secondly, we study the effects of the epigenetic mark 5-hydroxymethyl on cytosine on the structure of the Dickerson-Drew dodecamer. We show how our customized natural moves protocol can be used to investigate causal relationships of functional motions in biological systems.

Extra views on structure and dynamics of DNA loops on nucleosomes studied with molecular simulations

It has been shown experimentally that the action of the RSC chromatin remodeler leads to the formation of an irregular, partially remodeled nucleosome, termed a remosome. The remosome contains an extra 30-40 base pairs of DNA compared to a canonical nucleosome. Large-scale molecular simulations have provided information on the probable structure of remosomes and have explained why they remain stable in the absence of RSC. Here we explain how these simulations were carried out and what the resulting remosome models imply in terms of the mechanism of action of RSC. We notably show that local kinks within DNA are key in explaining how extra DNA can be in added to nucleosomes without unduly disturbing DNA-histone binding.

Mapping transiently formed and sparsely populated conformations on a complex energy landscape

Determining the structures, kinetics, thermodynamics and mechanisms that underlie conformational exchange processes in proteins remains extremely difficult. Only in favourable cases is it possible to provide atomic-level descriptions of sparsely populated and transiently formed alternative conformations. Here we benchmark the ability of enhanced-sampling molecular dynamics simulations to determine the free energy landscape of the L99A cavity mutant of T4 lysozyme. We find that the simulations capture key properties previously measured by NMR relaxation dispersion methods including the structure of a minor conformation, the kinetics and thermodynamics of conformational exchange, and the effect of mutations. We discover a new tunnel that involves the transient exposure towards the solvent of an internal cavity, and show it to be relevant for ligand escape. Together, our results provide a comprehensive view of the structural landscape of a protein, and point forward to studies of conformational exchange in systems that are less characterized experimentally.

DOI:http://dx.doi.org/10.7554/eLife.17505.001

Proteins are the workhorses of cells, where they perform a wide range of roles. To do so, they adopt specific three-dimensional structures that enable them to interact with other molecules as necessary. Often a protein needs to be able to shift between different states with distinct structures as it goes about its job. To fully understand how a protein works, it is important to be able to characterize these different structures and how the protein changes between them.

Many of the experimental techniques used to study protein structure rely on isolating the individual structural forms of a protein. Since many structures only exist briefly, this can be very difficult. To complement experimental results, computer simulations allow researchers to model how atoms behave within a molecule. However, a number of factors limit how well these models represent what happens experimentally, such as the accuracy of the physical description used for the modeling.

Wang et al. set out to test and benchmark how well computer simulations could model changes in structure for a protein called T4 lysozyme, which has been studied extensively using experimental techniques. T4 lysozyme exists in two different states that have distinct structures. By comparing existing detailed experimental measurements with the results of their simulations, Wang et al. found that the simulations could capture key aspects of how T4 lysozyme changes its shape.

The simulations described the structure of the protein in both states and accurately determined the relative proportion of molecules that are found in each state. They could also determine how long it takes for a molecule to change its shape from one state to the other. The findings allowed Wang et al. to describe in fine detail – down to the level of individual atoms – how the protein changes its shape and how mutations in the protein affect its ability to do so. A key question for future studies is whether these insights can be extended to other proteins that are less well characterized experimentally than T4 lysozyme.

DOI:http://dx.doi.org/10.7554/eLife.17505.002

Structural effects of modified ribonucleotides and magnesium in transfer RNAs

Modified nucleotides are ubiquitous and important to tRNA structure and function. To understand their effect on tRNA conformation, we performed a series of molecular dynamics simulations on yeast tRNA^Phe and tRNA_init, Escherichia coli tRNA_init and HIV tRNA^Lys. Simulations were performed with the wild type modified nucleotides, using the recently developed CHARMM compatible force field parameter set for modified nucleotides (J. Comput. Chem.2016, 37, 896), or with the corresponding unmodified nucleotides, and in the presence or absence of Mg²⁺. Results showed a stabilizing effect associated with the presence of the modifications and Mg²⁺ for some important positions, such as modified guanosine in position 37 and dihydrouridines in 16/17 including both structural properties and base interactions. Some other modifications were also found to make subtle contributions to the structural properties of local domains. While we were not able to investigate the effect of adenosine 37 in tRNA_init and limitations were observed in the conformation of E. coli tRNA_init, the presence of the modified nucleotides and of Mg²⁺ better maintained the structural features and base interactions of the tRNA systems than in their absence indicating the utility of incorporating the modified nucleotides in simulations of tRNA and other RNAs.

Additive CHARMM force field for naturally occurring modified ribonucleotides

More than 100 naturally occurring modified nucleotides have been found in RNA molecules, in particular in tRNAs. We have determined molecular mechanics force field parameters compatible with the CHARMM36 all‐atom additive force field for all these modifications using the CHARMM force field parametrization strategy. Emphasis was placed on fine tuning of the partial atomic charges and torsion angle parameters. Quantum mechanics calculations on model compounds provided the initial set of target data, and extensive molecular dynamics simulations of nucleotides and oligonucleotides in aqueous solutions were used for further refinement against experimental data. The presented parameters will allow for computational studies of a wide range of RNAs containing modified nucleotides, including the ribosome and transfer RNAs. © 2016 The Authors. Journal of Computational Chemistry Published by Wiley Periodicals, Inc.

Surface-water Interface Induces Conformational Changes Critical for Protein Adsorption: Implications for Monolayer Formation of EAS Hydrophobin

The class I hydrophobin EAS is part of a family of small, amphiphilic fungal proteins best known for their ability to self-assemble into stable monolayers that modify the hydrophobicity of a surface to facilitate further microbial growth. These proteins have attracted increasing attention for industrial and biomedical applications, with the aim of designing surfaces that have the potential to maintain their clean state by resisting non-specific protein binding. To gain a better understanding of this process, we have employed all-atom molecular dynamics to study initial stages of the spontaneous adsorption of monomeric EAS hydrophobin on fully hydroxylated silica, a commonly used industrial and biomedical substrate. Particular interest has been paid to the Cys3-Cys4 loop, which has been shown to exhibit disruptive behavior in solution, and the Cys7-Cys8 loop, which is believed to be involved in the aggregation of EAS hydrophobin at interfaces. Specific and water mediated interactions with the surface were also analyzed. We have identified two possible binding motifs, one which allows unfolding of the Cys7-Cys8 loop due to the surfactant-like behavior of the Cys3-Cys4 loop, and another which has limited unfolding due to the Cys3-Cys4 loop remaining disordered in solution. We have also identified intermittent interactions with water which mediate the protein adsorption to the surface, as well as longer lasting interactions which control the diffusion of water around the adsorption site. These results have shown that EAS behaves in a similar way at the air-water and surface-water interfaces, and have also highlighted the need for hydrophilic ligand functionalization of the silica surface in order to prevent the adsorption of EAS hydrophobin.

A Database of Force-Field Parameters, Dynamics, and Properties of Antimicrobial Compounds

We present an on-line database of all-atom force-field parameters and molecular properties of compounds with antimicrobial activity (mostly antibiotics and some beta-lactamase inhibitors). For each compound, we provide the General Amber Force Field parameters for the major species at physiological pH, together with an analysis of properties of interest as extracted from µs-long molecular dynamics simulations in explicit water solution. The properties include number and population of structural clusters, molecular flexibility, hydrophobic and hydrophilic molecular surfaces, the statistics of intraand inter-molecular H-bonds, as well as structural and dynamical properties of solvent molecules within first and second solvation shells. In addition, the database contains several key molecular parameters, such as energy of the frontier molecular orbitals, vibrational properties, rotational constants, atomic partial charges and electric dipole moment, computed by Density Functional Theory. The present database (to our knowledge the first extensive one including dynamical properties) is part of a wider project aiming to build-up a database containing structural, physico-chemical and dynamical properties of medicinal compounds using different force-field parameters with increasing level of complexity and reliability. The database is freely accessible at http://www.dsf.unica.it/translocation/db/.