Automated multiconformer model building for X-ray crystallography and cryo-EM.

In their folded state, biomolecules exchange between multiple conformational states that are crucial for their function. Traditional structural biology methods, such as X-ray crystallography and cryogenic electron microscopy (cryo-EM), produce density maps that are ensemble averages, reflecting molecules in various conformations. Yet, most models derived from these maps explicitly represent only a single conformation, overlooking the complexity of biomolecular structures. To accurately reflect the diversity of biomolecular forms, there is a pressing need to shift toward modeling structural ensembles that mirror the experimental data. However, the challenge of distinguishing signal from noise complicates manual efforts to create these models. In response, we introduce the latest enhancements to qFit, an automated computational strategy designed to incorporate protein conformational heterogeneity into models built into density maps. These algorithmic improvements in qFit are substantiated by superior Rfree and geometry metrics across a wide range of proteins. Importantly, unlike more complex multicopy ensemble models, the multiconformer models produced by qFit can be manually modified in most major model building software (e.g., Coot) and fit can be further improved by refinement using standard pipelines (e.g., Phenix, Refmac, Buster). By reducing the barrier of creating multiconformer models, qFit can foster the development of new hypotheses about the relationship between macromolecular conformational dynamics and function.

Uncovering Protein Ensembles: Automated Multiconformer Model Building for X-ray Crystallography and Cryo-EM.

In their folded state, biomolecules exchange between multiple conformational states that are crucial for their function. Traditional structural biology methods, such as X-ray crystallography and cryogenic electron microscopy (cryo-EM), produce density maps that are ensemble averages, reflecting molecules in various conformations. Yet, most models derived from these maps explicitly represent only a single conformation, overlooking the complexity of biomolecular structures. To accurately reflect the diversity of biomolecular forms, there is a pressing need to shift towards modeling structural ensembles that mirror the experimental data. However, the challenge of distinguishing signal from noise complicates manual efforts to create these models. In response, we introduce the latest enhancements to qFit, an automated computational strategy designed to incorporate protein conformational heterogeneity into models built into density maps. These algorithmic improvements in qFit are substantiated by superior Rfree and geometry metrics across a wide range of proteins. Importantly, unlike more complex multicopy ensemble models, the multiconformer models produced by qFit can be manually modified in most major model building software (e.g. Coot) and fit can be further improved by refinement using standard pipelines (e.g. Phenix, Refmac, Buster). By reducing the barrier of creating multiconformer models, qFit can foster the development of new hypotheses about the relationship between macromolecular conformational dynamics and function.

SARS-CoV-2 main protease mutation analysis via a kinematic method.

The Severe Acute Respiratory Syndrome CoronaVirus 2 (SARS-CoV-2) is the virus responsible for the COVID-19 pandemic. COVID-19 continues to cause millions of deaths globally in part due to immune-evading mutations. SARS-CoV-2 main protease (Mpro) is an important enzyme for viral replication and potentially an effective drug target. Mutations affect the dynamics of enzymes and thereby their activity and ability to bind ligands. Here, we use kinematic flexibility analysis (KFA) to identify how mutations and ligand binding changes the conformational flexibility of Mpro. KFA decomposes macromolecules into regions of different flexibility near-instantly from a static structure, allowing conformational dynamics analysis at scale. Altogether, we analyzed 47 mutation sites across 69 Mpro-ligand complexes resulting in more than 3300 different structures which includes 69 mutated structures with all 47 sites mutated simultaneously and 3243 single residue mutated structures. We found that mutations generally increased the conformational flexibility of the protein. Understanding the impact of mutations on the flexibility of Mpro is essential for identifying potential drug targets in the treatment of SARS-CoV-2. Further studies in this area can offer valuable insights into the mechanisms of molecular recognition.

Discovery of a cryptic pocket in the AI-predicted structure of PPM1D phosphatase explains the binding site and potency of its allosteric inhibitors.

Virtual screening is a widely used tool for drug discovery, but its predictive power can vary dramatically depending on how much structural data is available. In the best case, crystal structures of a ligand-bound protein can help find more potent ligands. However, virtual screens tend to be less predictive when only ligand-free crystal structures are available, and even less predictive if a homology model or other predicted structure must be used. Here, we explore the possibility that this situation can be improved by better accounting for protein dynamics, as simulations started from a single structure have a reasonable chance of sampling nearby structures that are more compatible with ligand binding. As a specific example, we consider the cancer drug target PPM1D/Wip1 phosphatase, a protein that lacks crystal structures. High-throughput screens have led to the discovery of several allosteric inhibitors of PPM1D, but their binding mode remains unknown. To enable further drug discovery efforts, we assessed the predictive power of an AlphaFold-predicted structure of PPM1D and a Markov state model (MSM) built from molecular dynamics simulations initiated from that structure. Our simulations reveal a cryptic pocket at the interface between two important structural elements, the flap and hinge regions. Using deep learning to predict the pose quality of each docked compound for the active site and cryptic pocket suggests that the inhibitors strongly prefer binding to the cryptic pocket, consistent with their allosteric effect. The predicted affinities for the dynamically uncovered cryptic pocket also recapitulate the relative potencies of the compounds (τb = 0.70) better than the predicted affinities for the static AlphaFold-predicted structure (τb = 0.42). Taken together, these results suggest that targeting the cryptic pocket is a good strategy for drugging PPM1D and, more generally, that conformations selected from simulation can improve virtual screening when limited structural data is available.

Discovery of a cryptic pocket in the AI-predicted structure of PPM1D phosphatase explains the binding site and potency of its allosteric inhibitors.

Virtual screening is a widely used tool for drug discovery, but its predictive power can vary dramatically depending on how much structural data is available. In the best case, crystal structures of a ligand-bound protein can help find more potent ligands. However, virtual screens tend to be less predictive when only ligand-free crystal structures are available, and even less predictive if a homology model or other predicted structure must be used. Here, we explore the possibility that this situation can be improved by better accounting for protein dynamics, as simulations started from a single structure have a reasonable chance of sampling nearby structures that are more compatible with ligand binding. As a specific example, we consider the cancer drug target PPM1D/Wip1 phosphatase, a protein that lacks crystal structures. High-throughput screens have led to the discovery of several allosteric inhibitors of PPM1D, but their binding mode remains unknown. To enable further drug discovery efforts, we assessed the predictive power of an AlphaFold-predicted structure of PPM1D and a Markov state model (MSM) built from molecular dynamics simulations initiated from that structure. Our simulations reveal a cryptic pocket at the interface between two important structural elements, the flap and hinge regions. Using deep learning to predict the pose quality of each docked compound for the active site and cryptic pocket suggests that the inhibitors strongly prefer binding to the cryptic pocket, consistent with their allosteric effect. The predicted affinities for the dynamically uncovered cryptic pocket also recapitulate the relative potencies of the compounds (τ b =0.70) better than the predicted affinities for the static AlphaFold-predicted structure (τ b =0.42). Taken together, these results suggest that targeting the cryptic pocket is a good strategy for drugging PPM1D and, more generally, that conformations selected from simulation can improve virtual screening when limited structural data is available.

Kinematic Vibrational Entropy Assessment and Analysis of SARS CoV-2 Main Protease.

The three-dimensional conformations of a protein influence its function and select for the ligands it can interact with. The total free energy change during protein-ligand complex formation includes enthalphic and entropic components, which together report on the binding affinity and conformational states of the complex. However, determining the entropic contribution is computationally burdensome. Here, we apply kinematic flexibility analysis (KFA) to efficiently estimate vibrational frequencies from static protein and protein-ligand structures. The vibrational frequencies, in turn, determine the vibrational entropies of the structures and their complexes. Our estimates of the vibrational entropy change caused by ligand binding compare favorably to values obtained from a dynamic Normal Mode Analysis (NMA). Higher correlation factors can be achieved by increasing the distance cutoff in the potential energy model. Furthermore, we apply our new method to analyze the entropy changes of the SARS CoV-2 main protease when binding with different ligand inhibitors, which is relevant for the design of potential drugs.

AtomNet PoseRanker: Enriching Ligand Pose Quality for Dynamic Proteins in Virtual High-Throughput Screens.

Structure-based, virtual High-Throughput Screening (vHTS) methods for predicting ligand activity in drug discovery are important when there are no or relatively few known compounds that interact with a therapeutic target of interest. State-of-the-art computational vHTS necessarily relies on effective methods for pose sampling and docking and generating an accurate affinity score from the docked poses. However, proteins are dynamic; in vivo ligands bind to a conformational ensemble. In silico docking to the single conformation represented by a crystal structure can adversely affect the pose quality. Here, we introduce AtomNet PoseRanker (ANPR), a graph convolutional network trained to identify and rerank crystal-like ligand poses from a sampled ensemble of protein conformations and ligand poses. In contrast to conventional vHTS methods that incorporate receptor flexibility, a deep learning approach can internalize valid cognate and noncognate binding modes corresponding to distinct receptor conformations, thereby learning to infer and account for receptor flexibility even on single conformations. ANPR significantly enriched pose quality in docking to cognate and noncognate receptors of the PDBbind v2019 data set. Improved pose rankings that better represent experimentally observed ligand binding modes improve hit rates in vHTS campaigns and thereby advance computational drug discovery, especially for novel therapeutic targets or novel binding sites.

Ligand binding remodels protein side-chain conformational heterogeneity.

While protein conformational heterogeneity plays an important role in many aspects of biological function, including ligand binding, its impact has been difficult to quantify. Macromolecular X-ray diffraction is commonly interpreted with a static structure, but it can provide information on both the anharmonic and harmonic contributions to conformational heterogeneity. Here, through multiconformer modeling of time- and space-averaged electron density, we measure conformational heterogeneity of 743 stringently matched pairs of crystallographic datasets that reflect unbound/apo and ligand-bound/holo states. When comparing the conformational heterogeneity of side chains, we observe that when binding site residues become more rigid upon ligand binding, distant residues tend to become more flexible, especially in non-solvent-exposed regions. Among ligand properties, we observe increased protein flexibility as the number of hydrogen bonds decreases and relative hydrophobicity increases. Across a series of 13 inhibitor-bound structures of CDK2, we find that conformational heterogeneity is correlated with inhibitor features and identify how conformational changes propagate differences in conformational heterogeneity away from the binding site. Collectively, our findings agree with models emerging from nuclear magnetic resonance studies suggesting that residual side-chain entropy can modulate affinity and point to the need to integrate both static conformational changes and conformational heterogeneity in models of ligand binding.

Proteins are the workhorses of our cells. They are large molecules that 'fold' into specific, often highly complex, three-dimensional configurations. These structures are not static, but rather dynamic and flexible. In other words, proteins can shift between different three-dimensional shapes to perform their tasks within the cell. To perform their roles, many proteins have to bind to small molecule ligands. Many ligands are drugs, which means that their effectiveness depends on their ability to bind to and impact the proteins involved in the disease they are treating. When a ligand binds to a protein, it can reshape the protein. For example, certain conformations of the protein, which were difficult for the protein to be in on its own, may become more stable when the ligand binds. Additionally, upon ligand binding, some parts of the protein may move relative to each other. Previous studies have shown that these movements can affect the interaction between ligand and protein. However, these studies only examined a small number of proteins. Therefore, Wankowicz et al. set out to determine, in greater detail, what happens to protein flexibility upon ligand binding. First, a pipeline was created to model alternative configurations of the protein both with and without ligands attached. These models measured flexibility within protein structures. The models revealed that when ligands bind to proteins, the flexibility of different regions of the protein changes ­ and does so in a consistent way. Proteins that become more rigid in the region interacting with their ligands become less rigid in other, distant regions, and vice versa. In other words, the rest of the protein is able to compensate for any changes in flexibility caused by ligand binding, which may contribute to how well a ligand binds to a protein. This study demonstrates the ability of ligands to affect the entire structure of the proteins they bind to, and therefore sheds new light on the role of proteins' innate conformational flexibility during this process. These results will contribute to our understanding of how the ligands and proteins involved in different cellular processes interact with each other ­ and, potentially, how these interactions can be manipulated.

Reproducibility of protein x-ray diffuse scattering and potential utility for modeling atomic displacement parameters.

Protein structure and dynamics can be probed using x-ray crystallography. Whereas the Bragg peaks are only sensitive to the average unit-cell electron density, the signal between the Bragg peaks-diffuse scattering-is sensitive to spatial correlations in electron-density variations. Although diffuse scattering contains valuable information about protein dynamics, the diffuse signal is more difficult to isolate from the background compared to the Bragg signal, and the reproducibility of diffuse signal is not yet well understood. We present a systematic study of the reproducibility of diffuse scattering from isocyanide hydratase in three different protein forms. Both replicate diffuse datasets and datasets obtained from different mutants were similar in pairwise comparisons (Pearson correlation coefficient ≥0.8). The data were processed in a manner inspired by previously published methods using custom software with modular design, enabling us to perform an analysis of various data processing choices to determine how to obtain the highest quality data as assessed using unbiased measures of symmetry and reproducibility. The diffuse data were then used to characterize atomic mobility using a liquid-like motions (LLM) model. This characterization was able to discriminate between distinct anisotropic atomic displacement parameter (ADP) models arising from different anisotropic scaling choices that agreed comparably with the Bragg data. Our results emphasize the importance of data reproducibility as a model-free measure of diffuse data quality, illustrate the ability of LLM analysis of diffuse scattering to select among alternative ADP models, and offer insights into the design of successful diffuse scattering experiments.

Modeling of Hidden Structures Using Sparse Chemical Shift Data from NMR Relaxation Dispersion.

NMR relaxation dispersion measurements report on conformational changes occurring on the µs-ms timescale. Chemical shift information derived from relaxation dispersion can be used to generate structural models of weakly populated alternative conformational states. Current methods to obtain such models rely on determining the signs of chemical shift changes between the conformational states, which are difficult to obtain in many situations. Here, we use a "sample and select" method to generate relevant structural models of alternative conformations of the C-terminal-associated region of Escherichia coli dihydrofolate reductase (DHFR), using only unsigned chemical shift changes for backbone amides and carbonyls (1H, 15N, and 13C'). We find that CS-Rosetta sampling with unsigned chemical shift changes generates a diversity of structures that are sufficient to characterize a minor conformational state of the C-terminal region of DHFR. The excited state differs from the ground state by a change in secondary structure, consistent with previous predictions from chemical shift hypersurfaces and validated by the x-ray structure of a partially humanized mutant of E. coli DHFR (N23PP/G51PEKN). The results demonstrate that the combination of fragment modeling with sparse chemical shift data can determine the structure of an alternative conformation of DHFR sampled on the µs-ms timescale. Such methods will be useful for characterizing alternative states, which can potentially be used for in silico drug screening, as well as contributing to understanding the role of minor states in biology and molecular evolution.

qFit 3: Protein and ligand multiconformer modeling for X-ray crystallographic and single-particle cryo-EM density maps.

New X-ray crystallography and cryo-electron microscopy (cryo-EM) approaches yield vast amounts of structural data from dynamic proteins and their complexes. Modeling the full conformational ensemble can provide important biological insights, but identifying and modeling an internally consistent set of alternate conformations remains a formidable challenge. qFit efficiently automates this process by generating a parsimonious multiconformer model. We refactored qFit from a distributed application into software that runs efficiently on a small server, desktop, or laptop. We describe the new qFit 3 software and provide some examples. qFit 3 is open-source under the MIT license, and is available at https://github.com/ExcitedStates/qfit-3.0.

Comparing serial X-ray crystallography and microcrystal electron diffraction (MicroED) as methods for routine structure determination from small macromolecular crystals.

Innovative new crystallographic methods are facilitating structural studies from ever smaller crystals of biological macromolecules. In particular, serial X-ray crystallography and microcrystal electron diffraction (MicroED) have emerged as useful methods for obtaining structural information from crystals on the nanometre to micrometre scale. Despite the utility of these methods, their implementation can often be difficult, as they present many challenges that are not encountered in traditional macromolecular crystallography experiments. Here, XFEL serial crystallography experiments and MicroED experiments using batch-grown microcrystals of the enzyme cyclophilin A are described. The results provide a roadmap for researchers hoping to design macromolecular microcrystallography experiments, and they highlight the strengths and weaknesses of the two methods. Specifically, we focus on how the different physical conditions imposed by the sample-preparation and delivery methods required for each type of experiment affect the crystal structure of the enzyme.

Mix-and-inject XFEL crystallography reveals gated conformational dynamics during enzyme catalysis.

How changes in enzyme structure and dynamics facilitate passage along the reaction coordinate is a fundamental unanswered question. Here, we use time-resolved mix-and-inject serial crystallography (MISC) at an X-ray free electron laser (XFEL), ambient-temperature X-ray crystallography, computer simulations, and enzyme kinetics to characterize how covalent catalysis modulates isocyanide hydratase (ICH) conformational dynamics throughout its catalytic cycle. We visualize this previously hypothetical reaction mechanism, directly observing formation of a thioimidate covalent intermediate in ICH microcrystals during catalysis. ICH exhibits a concerted helical displacement upon active-site cysteine modification that is gated by changes in hydrogen bond strength between the cysteine thiolate and the backbone amide of the highly strained Ile152 residue. These catalysis-activated motions permit water entry into the ICH active site for intermediate hydrolysis. Mutations at a Gly residue (Gly150) that modulate helical mobility reduce ICH catalytic turnover and alter its pre-steady-state kinetic behavior, establishing that helical mobility is important for ICH catalytic efficiency. These results demonstrate that MISC can capture otherwise elusive aspects of enzyme mechanism and dynamics in microcrystalline samples, resolving long-standing questions about the connection between nonequilibrium protein motions and enzyme catalysis.

Transport Properties of Nanoporous, Chemically Forced Biological Lattices.

Permselective nanochannels are ubiquitous in biological systems, controlling ion transport and maintaining a potential difference across a cell surface. Surface layers (S-layers) are proteinaceous, generally charged lattices punctuated with nanoscale pores that form the outermost cell envelope component of virtually all archaea and many bacteria. Ammonia oxidizing archaea (AOA) obtain their energy exclusively from oxidizing ammonia directly below the S-layer lattice, but how the charged surfaces and nanochannels affect availability of NH4+ at the reaction site is unknown. Here, we examine the electrochemical properties of negatively charged S-layers for asymmetrically forced ion transport governed by Michaelis-Menten kinetics at ultralow concentrations. Our 3-dimensional electrodiffusion reaction simulations revealed that a negatively charged S-layer can invert the potential across the nanochannel to favor chemically forced NH4+ transport, analogous to polarity switching in nanofluidic field-effect transistors. Polarity switching was not observed when only the interior of the nanochannels was charged. We found that S-layer charge, nanochannel geometry, and enzymatic turnover rate are finely tuned to elevate NH4+ concentration at the active site, potentially enabling AOA to occupy nutrient-poor ecological niches. Strikingly, and in contrast to voltage-biased systems, magnitudes of the co- and counterion currents in the charged nanochannels were nearly equal and amplified disproportionally to the NH4+ current. Our simulations suggest that engineered arrays of crystalline proteinaceous membranes could find unique applications in industrial energy conversion or separation processes.

Shining light on cysteine modification: connecting protein conformational dynamics to catalysis and regulation.

Cysteine is a rare but functionally important amino acid that is often subject to covalent modification. Cysteine oxidation plays an important role in many human disease processes, and basal levels of cysteine oxidation are required for proper cellular function. Because reactive cysteine residues are typically ionized to the thiolate anion (Cys-S-), their formation of a covalent bond alters the electrostatic and steric environment of the active site. X-ray-induced photo-oxidation to sulfenic acids (Cys-SOH) can recapitulate some aspects of the changes that occur under physiological conditions. Here we propose how site-specific cysteine photo-oxidation can be used to interrogate ensuing changes in protein structure and dynamics at atomic resolution. Although this powerful approach can connect cysteine covalent modification to global protein conformational changes and function, careful biochemical validation must accompany all such studies to exclude misleading artifacts. New types of X-ray crystallography experiments and powerful computational methods are creating new opportunities to connect conformational dynamics to catalysis for the large class of systems that use covalently modified cysteine residues for catalysis or regulation.

Molecular Recognition of M1-Linked Ubiquitin Chains by Native and Phosphorylated UBAN Domains.

Although the Ub-binding domain in ABIN proteins and NEMO (UBAN) is highly conserved, UBAN-containing proteins exhibit different Ub-binding properties, resulting in their diverse biological roles. Post-translational modifications further control UBAN domain specificity for poly-Ub chains. However, precisely, how the UBAN domain structurally confers such functional diversity remains poorly understood. Here we report crystal structures of ABIN-1 alone and in complex with one or two M1-linked di-Ub chains. ABIN-1 UBAN forms a homo-dimer that provides two symmetrical Ub-binding sites on either side of the coiled-coil structure. Moreover, crystal structures of ABIN1 UBAN in complex with di-Ub chains reveal a concentration-dependency of UBAN/di-Ub binding stoichiometry. Analysis of UBAN/M1-linked di-Ub binding characteristics indicates that phosphorylated S473 in OPTN and its corresponding phospho-mimetic residue in ABIN-1 (E484) are essential for high affinity interactions with M1-linked Ub chains. Also, a phospho-mimetic mutation of A303 in NEMO, corresponding to S473 of OPTN, increases binding affinity for M1-linked Ub chains. These findings are in line with the diverse physiological roles of UBAN domains, as phosphorylation of OPTN UBAN is required to enhance its binding to Ub during mitophagy.

Cryo-EM structures of Helicobacter pylori vacuolating cytotoxin A oligomeric assemblies at near-atomic resolution.

Human gastric pathogen Helicobacter pylori (H. pylori) is the primary risk factor for gastric cancer and is one of the most prevalent carcinogenic infectious agents. Vacuolating cytotoxin A (VacA) is a key virulence factor secreted by H. pylori and induces multiple cellular responses. Although structural and functional studies of VacA have been extensively performed, the high-resolution structure of a full-length VacA protomer and the molecular basis of its oligomerization are still unknown. Here, we use cryoelectron microscopy to resolve 10 structures of VacA assemblies, including monolayer (hexamer and heptamer) and bilayer (dodecamer, tridecamer, and tetradecamer) oligomers. The models of the 88-kDa full-length VacA protomer derived from the near-atomic resolution maps are highly conserved among different oligomers and show a continuous right-handed ß-helix made up of two domains with extensive domain-domain interactions. The specific interactions between adjacent protomers in the same layer stabilizing the oligomers are well resolved. For double-layer oligomers, we found short- and/or long-range hydrophobic interactions between protomers across the two layers. Our structures and other previous observations lead to a mechanistic model wherein VacA hexamer would correspond to the prepore-forming state, and the N-terminal region of VacA responsible for the membrane insertion would undergo a large conformational change to bring the hydrophobic transmembrane region to the center of the oligomer for the membrane channel formation.

qFit-ligand Reveals Widespread Conformational Heterogeneity of Drug-Like Molecules in X-Ray Electron Density Maps.

Proteins and ligands sample a conformational ensemble that governs molecular recognition, activity, and dissociation. In structure-based drug design, access to this conformational ensemble is critical to understand the balance between entropy and enthalpy in lead optimization. However, ligand conformational heterogeneity is currently severely underreported in crystal structures in the Protein Data Bank, owing in part to a lack of automated and unbiased procedures to model an ensemble of protein-ligand states into X-ray data. Here, we designed a computational method, qFit-ligand, to automatically resolve conformationally averaged ligand heterogeneity in crystal structures, and applied it to a large set of protein receptor-ligand complexes. In an analysis of the cancer related BRD4 domain, we found that up to 29% of protein crystal structures bound with drug-like molecules present evidence of unmodeled, averaged, relatively isoenergetic conformations in ligand-receptor interactions. In many retrospective cases, these alternate conformations were adventitiously exploited to guide compound design, resulting in improved potency or selectivity. Combining qFit-ligand with high-throughput screening or multitemperature crystallography could therefore augment the structure-based drug design toolbox.