Natural variants modify Klebsiella pneumoniae carbapenemase (KPC) acyl-enzyme conformational dynamics to extend antibiotic resistance

Class A serine β-lactamases (SBLs) are key antibiotic resistance determinants in Gram-negative bacteria. SBLs neutralize β-lactams via a hydrolytically labile covalent acyl-enzyme intermediate. Klebsiella pneumoniae carbapenemase (KPC) is a widespread SBL that hydrolyzes carbapenems, the most potent β-lactams; known KPC variants differ in turnover of expanded-spectrum oxyimino-cephalosporins (ESOCs), for example, cefotaxime and ceftazidime. Here, we compare ESOC hydrolysis by the parent enzyme KPC-2 and its clinically observed double variant (P104R/V240G) KPC-4. Kinetic analyses show that KPC-2 hydrolyzes cefotaxime more efficiently than the bulkier ceftazidime, with improved ESOC turnover by KPC-4 resulting from enhanced turnover (kcat), rather than altered KM values. High-resolution crystal structures of ESOC acyl-enzyme complexes with deacylation-deficient (E166Q) KPC-2 and KPC-4 mutants show that ceftazidime acylation causes rearrangement of three loops; the Ω, 240, and 270 loops, which border the active site. However, these rearrangements are less pronounced in the KPC-4 than the KPC-2 ceftazidime acyl-enzyme and are not observed in the KPC-2:cefotaxime acyl-enzyme. Molecular dynamics simulations of KPC:ceftazidime acyl-enyzmes reveal that the deacylation general base E166, located on the Ω loop, adopts two distinct conformations in KPC-2, either pointing "in" or "out" of the active site; with only the "in" form compatible with deacylation. The "out" conformation was not sampled in the KPC-4 acyl-enzyme, indicating that efficient ESOC breakdown is dependent upon the ordering and conformation of the KPC Ω loop. The results explain how point mutations expand the activity spectrum of the clinically important KPC SBLs to include ESOCs through their effects on the conformational dynamics of the acyl-enzyme intermediate.

Dynamical nonequilibrium molecular dynamics reveals the structural basis for allostery and signal propagation in biomolecular systems

ABSTRACT

A dynamical approach to nonequilibrium molecular dynamics (D-NEMD), proposed in the 1970s by Ciccotti et al., is undergoing a renaissance and is having increasing impact in the study of biological macromolecules. This D-NEMD approach, combining MD simulations in stationary (in particular, equilibrium) and nonequilibrium conditions, allows for the determination of the time-dependent structural response of a system using the Kubo-Onsager relation. Besides providing a detailed picture of the system's dynamic structural response to an external perturbation, this approach also has the advantage that the statistical significance of the response can be assessed. The D-NEMD approach has been used recently to identify a general mechanism of inter-domain signal propagation in nicotinic acetylcholine receptors, and allosteric effects in β -lactamase enzymes, for example. It complements equilibrium MD and is a very promising approach to identifying and analysing allosteric effects. Here, we review the D-NEMD approach and its application to biomolecular systems, including transporters, receptors, and enzymes.

Interactive Molecular Dynamics in Virtual Reality Is an Effective Tool for Flexible Substrate and Inhibitor Docking to the SARS-CoV-2 Main Protease

The main protease (Mpro) of the SARS-CoV-2 virus is one focus of drug development efforts for COVID-19. Here, we show that interactive molecular dynamics in virtual reality (iMD-VR) is a useful and effective tool for creating Mpro complexes. We make these tools and models freely available. iMD-VR provides an immersive environment in which users can interact with MD simulations and so build protein complexes in a physically rigorous and flexible way. Recently, we have demonstrated that iMD-VR is an effective method for interactive, flexible docking of small molecule drugs into their protein targets (Deeks et al. PLoS One 2020, 15, e0228461). Here, we apply this approach to both an Mpro inhibitor and an oligopeptide substrate, using experimentally determined crystal structures. For the oligopeptide, we test against a crystallographic structure of the original SARS Mpro. Docking with iMD-VR gives models in agreement with experimentally observed (crystal) structures. The docked structures are also tested in MD simulations and found to be stable. Different protocols for iMD-VR docking are explored, e.g., with and without restraints on protein backbone, and we provide recommendations for its use. We find that it is important for the user to focus on forming binding interactions, such as hydrogen bonds, and not to rely on using simple metrics (such as RMSD), in order to create realistic, stable complexes. We also test the use of apo (uncomplexed) crystal structures for docking and find that they can give good results. This is because of the flexibility and dynamic response allowed by the physically rigorous, atomically detailed simulation approach of iMD-VR. We make our models (and interactive simulations) freely available. The software framework that we use, Narupa, is open source, and uses commodity VR hardware, so these tools are readily accessible to the wider research community working on Mpro (and other COVID-19 targets). These should be widely useful in drug development, in education applications, e.g., on viral enzyme structure and function, and in scientific communication more generally.

Mechanism of inhibition of SARS-CoV-2 Mpro by peptidyl Michael acceptor explained by QM/MM simulations and design of new derivatives with tunable chemical reactivity

The SARS-CoV-2 main protease (M^pro) is essential for replication of the virus responsible for the COVID-19 pandemic, and one of the main targets for drug design. Here, we simulate the inhibition process of SARS-CoV-2 M^pro with a known Michael acceptor (peptidyl) inhibitor, N3. The free energy landscape for the mechanism of the formation of the covalent enzyme-inhibitor product is computed with QM/MM molecular dynamics methods. The simulations show a two-step mechanism, and give structures and calculated barriers in good agreement with experiment. Using these results and information from our previous investigation on the proteolysis reaction of SARS-CoV-2 M^pro, we design two new, synthetically accessible N3-analogues as potential inhibitors, in which the recognition and warhead motifs are modified. QM/MM modelling of the mechanism of inhibition of M^pro by these novel compounds indicates that both may be promising candidates as drug leads against COVID-19, one as an irreversible inhibitor and one as a potential reversible inhibitor.

Enhanced sampling molecular dynamics simulations correctly predict the diverse activities of a series of stiff-stilbene G-quadruplex DNA ligands

Ligands with the capability to bind G-quadruplexes (G4s) specifically, and to control G4 structure and behaviour, offer great potential in the development of novel therapies, technologies and functional materials. Most known ligands bind to a pre-formed topology, but G4s are highly dynamic and a small number of ligands have been discovered that influence these folding equilibria. Such ligands may be useful as probes to understand the dynamic nature of G4 in vivo, or to exploit the polymorphism of G4 in the development of molecular devices. To date, these fascinating molecules have been discovered serendipitously. There is a need for tools to predict such effects to drive ligand design and development, and for molecular-level understanding of ligand binding mechanisms and associated topological perturbation of G4 structures. Here we study the G4 binding mechanisms of a family of stiff-stilbene G4 ligands to human telomeric DNA using molecular dynamics (MD) and enhanced sampling (metadynamics) MD simulations. The simulations predict a variety of binding mechanisms and effects on G4 structure for the different ligands in the series. In parallel, we characterize the binding of the ligands to the G4 target experimentally using NMR and CD spectroscopy. The results show good agreement between the simulated and experimentally observed binding modes, binding affinities and ligand-induced perturbation of the G4 structure. The simulations correctly predict ligands that perturb G4 topology. Metadynamics simulations are shown to be a powerful tool to aid development of molecules to influence G4 structure, both in interpreting experiments and to help in the design of these chemotypes.

Biomolecular Simulations in the Time of COVID19, and After

COVID19 has changed life for people worldwide. Despite lockdowns globally, computational research has pressed on, working remotely and collaborating virtually on research questions in COVID19 and the virus it is caused by, SARS-CoV-2. Molecular simulations can help to characterize the function of viral and host proteins and have the potential to contribute to the search for vaccines and treatments. Changes in the modus operandi of research groups include broader adoption of the use of preprint servers, earlier and more open sharing of methods, models, and data, the use of social media to rapidly disseminate information, online seminars, and cloud-based virtual collaboration. Research funders and computing providers worldwide recognized the need to provide rapid and significant access to computational architectures. In this review, we discuss how the interplay of all of these factors is influencing the impact - both potential and realized - of biomolecular simulations in the fight against SARS-CoV-2.

Free fatty acid binding pocket in the locked structure of SARS-CoV-2 spike protein

Many efforts to develop therapies against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are focused on the spike (S) protein trimer that binds to the host receptor. Structures of trimeric S protein show its receptor-binding domain in either an up or a down conformation. Toelzer et al. produced SARS-CoV-2 S in insect cells and determined the structure by cryo–electron microscopy. In their dataset, the closed form was predominant and was stabilized by binding linoleic acid, an essential fatty acid. A similar binding pocket appears to be present in previous highly pathogenic coronaviruses, and past studies suggested links between viral infection and fatty acid metabolism. The pocket could be exploited to develop inhibitors that trap S protein in the closed conformation.

Science, this issue p. 725

Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), represents a global crisis. Key to SARS-CoV-2 therapeutic development is unraveling the mechanisms that drive high infectivity, broad tissue tropism, and severe pathology. Our 2.85-angstrom cryo–electron microscopy structure of SARS-CoV-2 spike (S) glycoprotein reveals that the receptor binding domains tightly bind the essential free fatty acid linoleic acid (LA) in three composite binding pockets. A similar pocket also appears to be present in the highly pathogenic severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV). LA binding stabilizes a locked S conformation, resulting in reduced angiotensin-converting enzyme 2 (ACE2) interaction in vitro. In human cells, LA supplementation synergizes with the COVID-19 drug remdesivir, suppressing SARS-CoV-2 replication. Our structure directly links LA and S, setting the stage for intervention strategies that target LA binding by SARS-CoV-2.

Electronic structure benchmark calculations of CO2 fixing elementary chemical steps in RuBisCO using the projector-based embedding approach

Ribulose 1,5-bisphosphate carboxylase-oxygenase (RuBisCO) is the main enzyme involved in atmospheric carbon dioxide (CO₂ ) fixation in the biosphere. This enzyme catalyzes a set of five chemical steps that take place in the same active-site within magnesium (II) coordination sphere. Here, a set of electronic structure benchmark calculations have been carried out on a reaction path proposed by Gready et al. by means of the projector-based embedding approach. Activation and reaction energies for all main steps catalyzed by RuBisCO have been calculated at the MP2, SCS-MP2, CCSD, and CCSD(T)/aug-cc-pVDZ and cc-pVDZ levels of theory. The treatment of the magnesium cation with post-HF methods is explored to determine the nature of its involvement in the mechanism. With the high-level ab initio values as a reference, we tested the performance of a set of density functional theory (DFT) exchange-correlation (xc) functionals in reproducing the reaction energetics of RuBisCO carboxylase activity on a set of model fragments. Different DFT xc-functionals show large variation in activation and reaction energies. Activation and reaction energies computed at the B3LYP level are close to the reference SCS-MP2 results for carboxylation, hydration and protonation reactions. However, for the carbon-carbon bond dissociation reaction, B3LYP and other functionals give results that differ significantly from the ab initio reference values. The results show the applicability of the projector-based embedding approach to metalloenzymes. This technique removes the uncertainty associated with the selection of different DFT xc-functionals and so can overcome some of inherent limitations of DFT calculations, complementing, and potentially adding to modeling of enzyme reaction mechanisms with DFT methods.

Simulations support the interaction of the SARS-CoV-2 spike protein with nicotinic acetylcholine receptors

Changeux et al. recently suggested that the SARS-CoV-2 spike (S) protein may interact with nicotinic acetylcholine receptors (nAChRs). Such interactions may be involved in pathology and infectivity. Here, we use molecular simulations of validated atomically detailed structures of nAChRs, and of the S protein, to investigate this 'nicotinic hypothesis'. We examine the binding of the Y674-R685 loop of the S protein to three nAChRs, namely the human α4β2 and α7 subtypes and the muscle-like αβγδ receptor from Tetronarce californica. Our results indicate that Y674-R685 has affinity for nAChRs and the region responsible for binding contains the PRRA motif, a four-residue insertion not found in other SARS-like coronaviruses. In particular, R682 has a key role in the stabilisation of the complexes as it forms interactions with loops A, B and C in the receptor's binding pocket. The conformational behaviour of the bound Y674-R685 region is highly dependent on the receptor subtype, adopting extended conformations in the α4β2 and α7 complexes and more compact ones when bound to the muscle-like receptor. In the α4β2 and αβγδ complexes, the interaction of Y674-R685 with the receptors forces the loop C region to adopt an open conformation similar to other known nAChR antagonists. In contrast, in the α7 complex, Y674-R685 penetrates deeply into the binding pocket where it forms interactions with the residues lining the aromatic box, namely with TrpB, TyrC1 and TyrC2. Estimates of binding energy suggest that Y674-R685 forms stable complexes with all three nAChR subtypes. Analyses of the simulations of the full-length S protein show that the Y674-R685 region is accessible for binding, and suggest a potential binding orientation of the S protein with nAChRs.

Benchmarking quantum mechanical methods for calculating reaction energies of reactions catalyzed by enzymes

To assess the accuracy of different quantum mechanical methods for biochemical modeling, the reaction energies of 20 small model reactions (chosen to represent chemical steps catalyzed by commonly studied enzymes) were calculated. The methods tested included several popular Density Functional Theory (DFT) functionals, second-order Møller Plesset perturbation theory (MP2) and its spin-component scaled variant (SCS-MP2), and coupled cluster singles and doubles and perturbative triples (CCSD(T)). Different basis sets were tested. CCSD(T)/aug-cc-pVTZ results for all 20 reactions were used to benchmark the other methods. It was found that MP2 and SCS-MP2 reaction energy calculation results are similar in quality to CCSD(T) (mean absolute error (MAE) of 1.2 and 1.3 kcal mol−1, respectively). MP2 calculations gave a large error in one case, and are more subject to basis set effects, so in general SCS-MP2 calculations are a good choice when CCSD(T) calculations are not feasible. Results with different DFT functionals were of reasonably good quality (MAEs of 2.5–5.1 kcal mol−1), whereas popular semi-empirical methods (AM1, PM3, SCC-DFTB) gave much larger errors (MAEs of 11.6–14.6 kcal mol−1). These results should be useful in guiding methodological choices and assessing the accuracy of QM/MM calculations on enzyme-catalyzed reactions.

Temperature, Dynamics, and Enzyme-Catalyzed Reaction Rates

We review the adaptations of enzyme activity to different temperatures. Psychrophilic (cold-adapted) enzymes show significantly different activation parameters (lower activation enthalpies and entropies) from their mesophilic counterparts. Furthermore, there is increasing evidence that the temperature dependence of many enzyme-catalyzed reactions is more complex than is widely believed. Many enzymes show curvature in plots of activity versus temperature that is not accounted for by denaturation or unfolding. This is explained by macromolecular rate theory: A negative activation heat capacity for the rate-limiting chemical step leads directly to predictions of temperature optima; both entropy and enthalpy are temperature dependent. Fluctuations in the transition state ensemble are reduced compared to the ground state. We show how investigations combining experiment with molecular simulation are revealing fundamental details of enzyme thermoadaptation that are relevant for understanding aspects of enzyme evolution. Simulations can calculate relevant thermodynamic properties (such as activation enthalpies, entropies, and heat capacities) and reveal the molecular mechanisms underlying experimentally observed behavior.

Taming the Reactivity of Monoterpene Synthases To Guide Regioselective Product Hydroxylation

Monoterpenoids are industrially important natural products with applications in the flavours, fragrances, fuels and pharmaceutical industries. Most monoterpenoids are produced by plants, but recently two bacterial monoterpene synthases have been identified, including a cineole synthase (bCinS). Unlike plant cineole synthases, bCinS is capable of producing nearly pure cineole from geranyl diphosphate in a complex cyclisation cascade that is tightly controlled. Here we have used a multidisciplinary approach to show that Asn305 controls water attack on the α-terpinyl cation and subsequent cyclisation and deprotonation of the α-terpineol intermediate, key steps in the cyclisation cascade which direct product formation towards cineole. Mutation of Asn305 results in variants that no longer produce α-terpineol or cineole. Molecular dynamics simulations revealed that water coordination is disrupted in all variants tested. Quantum mechanics calculations indicate that Asn305 is most likely a (transient) proton acceptor for the final deprotonation step. Our synergistic approach gives unique insight into how a single residue, Asn305, tames the promiscuous chemistry of monoterpene synthase cyclisation cascades. It does this by tightly controlling the final steps in cineole formation catalysed by bCinS to form a single hydroxylated monoterpene product.

Cyclic boronates as versatile scaffolds for KPC-2 β-lactamase inhibition

Klebsiella pneumoniae carbapenemase-2 (KPC-2) is a serine-β-lactamase (SBL) capable of hydrolysing almost all β-lactam antibiotics. We compare KPC-2 inhibition by vaborbactam, a clinically-approved monocyclic boronate, and VNRX-5133 (taniborbactam), a bicyclic boronate in late-stage clinical development. Vaborbactam inhibition is slowly reversible, whereas taniborbactam has an off-rate indicating essentially irreversible complex formation and a 15-fold higher on-rate, although both potentiate β-lactam activity against KPC-2-expressing K. pneumoniae. High resolution X-ray crystal structures reveal closely related binding modes for both inhibitors to KPC-2, with differences apparent only in positioning of the endocyclic boronate ester oxygen. The results indicate the bicyclic boronate scaffold as both an efficient, long-lasting, KPC-2 inhibitor and capable of supporting further iterations that may improve potency against specific enzyme targets and pre-empt the emergence of inhibitor resistant KPC-2 variants.

Multiscale simulation approaches to modeling drug-protein binding

Simulations can provide detailed insight into the molecular processes involved in drug action, such as protein-ligand binding, and can therefore be a valuable tool for drug design and development. Processes with a large range of length and timescales may be involved, and understanding these different scales typically requires different types of simulation methodology. Ideally, simulations should be able to connect across scales, to analyze and predict how changes at one scale can influence another. Multiscale simulation methods, which combine different levels of treatment, are an emerging frontier with great potential in this area. Here we review multiscale frameworks of various types, and selected applications to biomolecular systems with a focus on drug-ligand binding.

Visualizing the protons in a metalloenzyme electron proton transfer pathway

In redox metalloenzymes, the process of electron transfer often involves the concerted movement of a proton. These processes are referred to as proton-coupled electron transfer, and they underpin a wide variety of biological processes, including respiration, energy conversion, photosynthesis, and metalloenzyme catalysis. The mechanisms of proton delivery are incompletely understood, in part due to an absence of information on exact proton locations and hydrogen bonding structures in a bona fide metalloenzyme proton pathway. Here, we present a 2.1-Å neutron crystal structure of the complex formed between a redox metalloenzyme (ascorbate peroxidase) and its reducing substrate (ascorbate). In the neutron structure of the complex, the protonation states of the electron/proton donor (ascorbate) and all of the residues involved in the electron/proton transfer pathway are directly observed. This information sheds light on possible proton movements during heme-catalyzed oxygen activation, as well as on ascorbate oxidation.

IMPRESSION - prediction of NMR parameters for 3-dimensional chemical structures using machine learning with near quantum chemical accuracy

The IMPRESSION (Intelligent Machine PREdiction of Shift and Scalar information Of Nuclei) machine learning system provides an efficient and accurate method for the prediction of NMR parameters from 3-dimensional molecular structures. Here we demonstrate that machine learning predictions of NMR parameters, trained on quantum chemical computed values, can be as accurate as, but computationally much more efficient (tens of milliseconds per molecular structure) than, quantum chemical calculations (hours/days per molecular structure) starting from the same 3-dimensional structure. Training the machine learning system on quantum chemical predictions, rather than experimental data, circumvents the need for the existence of large, structurally diverse, error-free experimental databases and makes IMPRESSION applicable to solving 3-dimensional problems such as molecular conformation and stereoisomerism.

Visible-light photoswitching of ligand binding mode suggests G-quadruplex DNA as a target for photopharmacology

A pyridinium-decorated photoresponsive dithienylethene selectively targets G-quadruplex DNA, allowing binding mode and toxicity to be controlled exclusively with visible light.

Resistance to the “last resort” antibiotic colistin: a single-zinc mechanism for phosphointermediate formation in MCR enzymes

Simulations show the mono-zinc form of MCR to be stable and competent for covalent phospho(ethanolamine) intermediate formation.

Interactive molecular dynamics in virtual reality for accurate flexible protein-ligand docking

Simulating drug binding and unbinding is a challenge, as the rugged energy landscapes that separate bound and unbound states require extensive sampling that consumes significant computational resources. Here, we describe the use of interactive molecular dynamics in virtual reality (iMD-VR) as an accurate low-cost strategy for flexible protein-ligand docking. We outline an experimental protocol which enables expert iMD-VR users to guide ligands into and out of the binding pockets of trypsin, neuraminidase, and HIV-1 protease, and recreate their respective crystallographic protein-ligand binding poses within 5-10 minutes. Following a brief training phase, our studies shown that iMD-VR novices were able to generate unbinding and rebinding pathways on similar timescales as iMD-VR experts, with the majority able to recover binding poses within 2.15 Å RMSD of the crystallographic binding pose. These results indicate that iMD-VR affords sufficient control for users to carry out the detailed atomic manipulations required to dock flexible ligands into dynamic enzyme active sites and recover crystallographic poses, offering an interesting new approach for simulating drug docking and generating binding hypotheses.

Discovery of New and Potent InhA Inhibitors as Antituberculosis Agents: Structure-Based Virtual Screening Validated by Biological Assays and X-ray Crystallography

The enoyl-acyl carrier protein reductase InhA of Mycobacterium tuberculosis is an attractive, validated target for antituberculosis drug development. Moreover, direct inhibitors of InhA remain effective against InhA variants with mutations associated with isoniazid resistance, offering the potential for activity against MDR isolates. Here, structure-based virtual screening supported by biological assays was applied to identify novel InhA inhibitors as potential antituberculosis agents. High-speed Glide SP docking was initially performed against two conformations of InhA differing in the orientation of the active site Tyr158. The resulting hits were filtered for drug-likeness based on Lipinski's rule and avoidance of PAINS-like properties and finally subjected to Glide XP docking to improve accuracy. Sixteen compounds were identified and selected for in vitro biological assays, of which two (compounds 1 and 7) showed MIC of 12.5 and 25 μg/mL against M. tuberculosis H37Rv, respectively. Inhibition assays against purified recombinant InhA determined IC₅₀ values for these compounds of 0.38 and 0.22 μM, respectively. A crystal structure of the most potent compound, compound 7, bound to InhA revealed the inhibitor to occupy a hydrophobic pocket implicated in binding the aliphatic portions of InhA substrates but distant from the NADH cofactor, i.e., in a site distinct from those occupied by the great majority of known InhA inhibitors. This compound provides an attractive starting template for ligand optimization aimed at discovery of new and effective compounds against M. tuberculosis that act by targeting InhA.

A General Mechanism for Signal Propagation in the Nicotinic Acetylcholine Receptor Family

Nicotinic acetylcholine receptors (nAChRs) modulate synaptic activity in the central nervous system. The α7 subtype, in particular, has attracted considerable interest in drug discovery as a target for several conditions, including Alzheimer's disease and schizophrenia. Identifying agonist-induced structural changes underlying nAChR activation is fundamentally important for understanding biological function and rational drug design. Here, extensive equilibrium and nonequilibrium molecular dynamics simulations, enabled by cloud-based high-performance computing, reveal the molecular mechanism by which structural changes induced by agonist unbinding are transmitted within the human α7 nAChR. The simulations reveal the sequence of coupled structural changes involved in driving conformational change responsible for biological function. Comparison with simulations of the α4β2 nAChR subtype identifies features of the dynamical architecture common to both receptors, suggesting a general structural mechanism for signal propagation in this important family of receptors.