We describe the photoisomerization of the retinylidene protonated Schiff base in human retinol-binding protein II (hCRBPII) and the role of water molecules in this process. We characterize the photoisomerization of the 15-cis/all-trans retinylidene protonated Schiff base in this system using UV-visible spectroscopy and atomic-resolution X-ray crystallography. We further demonstrate a process where the pKa of the protonated Schiff base is substantially altered by light-induced dehydration of the binding pocket, suggesting novel pathways of photoswitching that rely not on isomerization or conformational change of the chromophore but rather on light-induced reorganization of the protein environment.
Structural biology has fundamentally influenced pharmaceutical research and development at Merck Sharp & Dohme LLC, Rahway, New Jersey, USA, progressing from pioneering macromolecular crystallography in the 1980s to a fully integrated platform that today encompasses X-ray crystallography, cryo-electron microscopy, micro-electron diffraction and cryo-electron tomography. In this review, we present a comprehensive overview of how these complementary methods have advanced drug discovery across diverse therapeutic areas. We illustrate how atomic to cellular structural insights inform drug discovery and development, from target identification and validation, hit finding, lead identification through lead optimization and clinical progression. Furthermore, we describe how structural biology techniques aid in formulation strategies of antibodies and vaccines. Finally, we highlight the growing integration of ex situ and in situ approaches as a paradigm shift towards elucidating drug mechanisms in native cellular contexts, a transition poised to accelerate the discovery and development of next-generation therapeutics.
Nucleosomes are the repeating unit of chromatin. They act as recognition platforms for chromatin-binding factors that coordinate genome maintenance. The chromatin remodeler Amplified in Liver Cancer 1 (ALC1) is a key component of the DNA-damage response and a promising therapeutic target in cancer. Through extensive classification of our previously deposited cryo-electron microscopy dataset, we identified a previously unresolved ALC1-nucleosome complex characterized by a more open conformation of ALC1. This is the first structure of ALC1 in which all domains are visualized in the context of a nucleosome complex, including the regulatory macro domain and a single α-helix motif within the linker. This newly identified conformation may represent an intermediate between the auto-inhibited and active states, and provides new structural insights into the conformational transitions that regulate the activity of ALC1.
Structure determination of G protein-coupled receptors (GPCRs) plays an important role in accelerating drug development against this medically important protein family. This study outlines the development of a new fusion tool to enable structure determination of GPCRs in inactive conformations by cryo-EM. Initially, a PDB mining approach was applied to select eight naturally occurring proteins with the intention of fusing them into the intracellular loop 3 (ICL3) of GPCRs to create a suitable fiducial marker for cryo-EM workflows. During the selection process, candidates with known high-affinity protein binders were prioritized to enable a further increase in the protein mass of the fiducial marker. Fusion constructs were generated with adenosine receptor A2A (A2AR) and were assessed for expression and aggregation levels. For the two best-performing new fusion constructs, ligand binding was characterized to ensure that the fusion tag did not significantly affect protein behaviour. A2AR with a β-lactamase fusion in ICL3 and binding partner β-lactamase inhibitory protein II (BLIPII) was then selected to solve an antagonist-bound structure. The overall map was resolved to an average of 3.2 Å resolution with continuous helices connecting the β-lactamase to helices 5 and 6 of A2AR. Focused refinement of the A2AR region improved the local resolution and map detail in the orthosteric site, thereby allowing confident modelling of the antagonist ligand, ZM241385, which matches previously described X-ray crystallographic structures. This new fusion provides an alternative option for GPCR structure determination, with several potential benefits compared with existing tools, such as a more favourable position relative to the GPCR to reduce potential clashes.
Gasdermin D (GSDMD) has been identified as a critical component of the inflammasome, an important intracellular signaling multi-protein complex. Abnormal activation of GSDMD has been linked to a variety of inflammatory diseases, including non-alcoholic steatohepatitis, inflammatory bowel disease and COVID-19, and has been linked to potential pathogenesis of septic shock. While small molecules have been identified to inhibit N-terminal domain (NTD) pore formation and membrane translocation, further clinical application of these molecules is currently limited due to narrow specificity. Here, we report a peptide from a structure-based computational screen that selectively inhibits caspase-dependent cleavage of GSDMD. We have determined the structure of the GSDMD-peptide complex, indicating that peptide binding is facilitated by negatively charged residues from the peptide and the hydrophobic exosite of the protein, inhibiting caspase binding and activation. The experimental data suggest the potential to target the exosite for therapeutic intervention.
The phosphatidylinositol 5-phosphate 4-kinases (PIP4Ks) are an evolutionarily conserved family of lipid kinases that phosphorylate phosphatidylinositol 5-phosphate to generate phosphatidylinositol 4,5-bisphosphate. In mammals, the catalytically active α and β isoforms, encoded by PIP4K2A and PIP4K2B, respectively, localize to distinct cellular compartments and have been implicated in metabolism, immune regulation and tumorigenesis, prompting interest in their pharmacological inhibition. Notably, most reported small-molecule inhibitors display substantially higher potency towards the α isoform than the β isoform, suggesting intrinsic structural features that limit the effective targeting of PIP4K2B. Here, we report the crystal structure of PIP4K2A in complex with 422A, a potent dual α/β inhibitor with improved metabolic stability. The structure reveals an unexpected, water-mediated interaction in which a pyridyl nitrogen of the inhibitor engages a conserved structured water molecule in the roof of the specificity pocket, constraining the orientation of the pyridylmethyl side chain and stabilizing a rigid, high-affinity binding mode. Comparative structural analysis with the PIP4K2A-selective inhibitor BAY-091 shows that deeper penetration into the specificity pocket enhances PIP4K2A binding but is accompanied by local steric constraints that are likely to be less well tolerated in PIP4K2B. Together, these findings define structural determinants of isoform-dependent inhibitor binding within the PIP4K family and provide a framework for structure-guided optimization of lipid kinase inhibitors with improved isoform balance.
Glycoside hydrolases (GHs) achieve glycan breakdown through glycosidic bond hydrolysis. Some retaining GHs can also transglycosylate, which could be useful for glycosynthesis. Improving GHs for either glycan degradation or synthesis requires a deep understanding of the residues involved in catalysis and substrate binding. This study characterizes in detail the activity and structure of Ta_Cel5A, a cellulase in glycoside hydrolase family 5, subfamily 5 (GH5_5) from the thermophilic ascomycete Thermoascus aurantiacus. While its hydrolytic activity was confirmed, Ta_Cel5A was also found to exhibit a weak transglycosylase activity with cellopentaose as a substrate. Transglycosylation products were detected within the first minutes of reaction at 25°C, far below its optimal temperature. The structures of catalytically impaired variants were solved in complex with oligosaccharides. The entire catalytic cleft was defined, consisting of seven glucose-binding subsites, five negative subsites and two positive subsites, from the nonreducing end to the reducing end. The fifth negative subsite could not be inferred in silico, showing the limitation in predicting distal subsites based on structural analogy. The structure of the glycosyl-enzyme intermediate was also obtained, revealing the displacement of key residues in the active site. The covalent binding of a glycosidic molecule triggers a major displacement of the nucleophilic residue, Glu244, changing its interaction network. The acid/base residue, Glu133, and a conserved tyrosine residue, Tyr201, are also displaced during glycosyl-enzyme intermediate formation, hinting at their role in the deglycosylation step.
Scientific equipment, such as high-end electron microscopes, entails considerable costs for acquisition, maintenance and operation. Efficient use of these instruments requires thoughtful allocation of access among multiple users. This task may be aided by a thorough statistical analysis that combines cost modelling and queueing theory. In this work, we propose a framework for estimating the hourly access cost and the average waiting time, based on equipment characteristics (price, lifespan and operating hours), user patterns (cycle and acquisition times) and staffing assumptions. In this way, we derive expressions that help estimate the optimal number of users and service congestion levels. The methodology is general and applicable to many types of shared scientific infrastructure. We illustrate its use with realistic cryo-electron microscopy scenarios. To facilitate adoption, we provide an open-access online calculator implementing the model at https://i2pc.es/coss/Programs/equipmentCost.html. This tool can support evidence-based decision-making in equipment planning and policy design.
Urocanate reductase (UrdA) is a bacterial enzyme that converts urocanic acid to imidazole propionate. Its catalytic residue Arg411 undergoes a large conformational change in the substrate-bound versus product-bound states. In contrast to previously studied cryo-conditions, the room-temperature X-ray data of UrdA presented here show that the occupancy distribution of Arg411 is affected by crystal cryocooling. We further provide evidence that a phosphate ion stabilizes the substrate complex and can bias the conformation of Arg411. Both room-temperature and cryogenic X-ray datasets were essential to elucidate the dynamic nature of the UrdA active site, highlighting the importance of collecting data at both temperatures, as each may reveal distinct conformational states.
Microtubules are cytoskeletal filaments that are typically characterized by a discontinuous helical lattice of α/β-tubulin heterodimers. Microtubules can also adopt variable lattice architectures both in vitro and in cellular contexts. Pseudo-helical averaging processing strategies have been developed to generate cryo-EM reconstructions of microtubules with and without decorating protein-binding partners, but these pipelines can be difficult to implement for the average user, especially for undecorated filaments. Here, we describe MiCSPARC, a cryo-EM processing pipeline developed around CryoSPARC [Punjani et al. (2017), Nat. Methods, 14, 290-296], which leverages automated particle picking and fast 3D refinement times in CryoSPARC to determine the structures of both decorated and undecorated microtubules. We generate reconstructions of undecorated GDP microtubules, as well as kinesin-1 motor domain-decorated GMPCPP filaments, at resolutions of up to 2.8 Å, demonstrating the robustness of the pipeline. Based on its convenient implementation and its ability to routinely generate high-resolution, seam-corrected microtubule reconstructions, MiCSPARC should provide a valuable tool for understanding microtubule dynamics, microtubule-associated proteins and microtubule-targeting agents.
The rise of antimicrobial resistance in Mycobacterium tuberculosis underscores the urgent need for novel therapeutic strategies. Mycobacterial membrane protein large 3 (MmpL3) has emerged as a promising drug target, given its essential role in trehalose monomycolate transport and cell-wall biosynthesis. We determined the crystal structure of M. smegmatis MmpL3 in complex with a potent indolecarboxamide inhibitor, UPAR-1109, at 2.15 Å resolution, the highest reported to date. This structure provided unprecedented insights into the binding mode of the inhibitor, highlighting strong polar interactions and extensive hydrophobic contacts, which disrupt proton translocation. Speculative analysis about the molecular mechanism of inhibition has been proposed based on this model. Computational studies, including docking, molecular-dynamics and enhanced sampling simulations, revealed the remarkable plasticity of the MmpL3 binding site and confirmed the crystallographic orientation of UPAR-1109 as the most stable and biologically relevant binding mode. Together, these findings advance our understanding of the function and inhibition of MmpL3, providing valuable information for the rational design of next-generation antituberculosis agents, also with potential applications against nontuberculous mycobacteria.
Many biological processes rely on the activity of multiple proteins acting in concert as part of higher-order oligomeric complexes. However, the majority of structural studies have routinely simplified these systems to examine only one or a few protein partners of a multi-component system. Artificial intelligence (AI) methods implemented in AlphaFold, RoseTTAFold, ESMFold and others have revolutionized our ability to not only predict the fold of individual proteins, but also to predict that of complex assemblies. A major bottleneck in exploiting AI-based protein fold prediction for multiprotein complexes, however, is identifying likely interacting partners from a given interactome. Here, we present a protein complex prediction pipeline based on AlphaFold, called Alphafuser, that combines experimental interaction data with systematic querying of possible combinations of protein partners in a computationally parsimonious manner. This versatile protein complex prediction pipeline creates a queue of all potential complexes, up to a user defined number of partners. Because a complete search of all permutations is not typically practical, we implemented a simple dead-end trimming algorithm from the dimer step onwards based on the interface probability template modeling (ipTM) score to remove low-probability subcomplexes from the queue of higher-order complexes. Experimentally structurally characterized multiprotein complexes in the Protein Data Bank, PP2A-B56γ1 holoenzyme-PME-1 complex, the human γ-secretase complex, the TFIIIC complex and the TRAPP I complex, were used to obtain a general ipTM cutoff parameter for the pipeline and to confirm that complexes without extensive direct contacts between all subunits could be identified. We applied Alphafuser to two test cases starting from yeast two-hybrid and co-immunoprecipitation/mass spectrometry (co-IP/MS) interaction data. These predictions were tested experimentally using pull-down assays, confirming direct interactions of the proteins identified computationally by the Alphafuser pipeline.
Online UV-Vis absorption spectroscopy is often used during crystallographic data collection to characterize chromophores within proteins. This includes the study of electron-rich or redox-active enzyme intermediates and cofactors that are prone to specific radiation damage, necessitating suitable low-dose data-collection strategies. In crystallo spectroscopy enables the monitoring of X-ray-induced changes in the spectroscopic signatures of proteins and ligands; however, a comprehensive approach that also considers buffer components is required. Here, we present a mapping of X-ray-induced spectral changes in common crystallization chemicals and mixtures at cryogenic temperature, for which spectroscopic changes can be detected at doses as low as 1 kGy. A transient increase in absorption between 450 and 700 nm is frequently observed, arising from solvated electron absorption. Below 450 nm, several distinct absorption peaks were detected, for example for halides in buffers. In addition, Rayleigh scattering can lead to an increasing loss of photons from the optical path as the wavelength decreases, and thus to a significant elevation of the baseline at shorter wavelengths. In this study, we demonstrate the use of spectroscopic analyses to investigate the chromophoric enzyme intermediate I320 in vitamin B6 biosynthesis. In this case, X-ray-induced spectral changes were attributed to crystallization agents, cryoprotectants and light scattering, thereby excluding intrinsic alterations to the enzyme intermediate under low-dose conditions. Our study highlights the importance of monitoring spectral changes during diffraction data collection to ensure accurate interpretation of the electronic structure of chromophores.
Specific radiation damage frequently compromises structural analysis in macromolecular crystallography. However, it can also offer key mechanistic information. Here, we investigate the X-ray-induced radiolysis of the catalytic C5-peroxide adduct in crystals of the cofactor-independent enzyme urate oxidase. Using a top-hat X-ray beam to ensure homogeneous dose distribution, we monitored the occupancy of the peroxide species across extensive dose series at 100 K and room temperature (RT). We observe a fundamental kinetic phase transition between these thermal regimes. At RT, the peroxide decays rapidly following zero-order kinetics, consistent with a flux-limited regime where the radiolytically cleaved O2 product diffuses out of the active site to be replaced by water. Conversely, at 100 K the decay is markedly retarded and follows first-order kinetics. Bayesian kinetic modelling demonstrates that this cryoprotection arises from a recombination mechanism: the cleaved O2 molecule remains trapped in the active site with the organic species, enabling efficient recombination that competes with irreversible degradation, effectively resulting in radiation-induced in crystallo catalysis.
Sub-Ångström macromolecular crystallography allows the construction of structural models without any prior chemical knowledge and stereochemical restraints, making it possible to visualize deviations from peptide planarity, distortions of the planarity of aromatic ring systems and subtle alternate conformation networks. It also holds the promise of observing aspherical density features and other quantum-mechanical phenomena. However, the degree to which radiation damage might affect the details that can be discerned from biological structures at such resolutions has remained unknown. To address this, we report here the first study of radiation-damage effects at sub-Ångström resolution on Pyrococcus abyssi rubredoxin. Our data, collected at 100 K, indicate that the oxidized (Fe3+) state of the Fe atom is preserved to a large degree at a dose of 50 kGy, whereas at 1 MGy it is partially reduced to the Fe2+ state. Isomorphous difference maps reveal extensive conformational changes at 1 MGy that are most likely coupled to the reduction of Fe3+. At 1 MGy, hydrogen densities that are visible at 50 kGy are preserved, but are distinctly blurred by dose. These findings suggest that the `global' radiation damage in reciprocal space and the associated blurring of electron density in real space proceed through small structural changes that are much more extended than the local damage at specific sites which is normally seen at lower resolution. Thus, the `global' and `specific' damage could be viewed as two sides of the same coin. Our study highlights the unique benefits of using low-dose data-collection protocols on large crystals fully bathed in a large `top-hat' beam with a uniform fluence profile for accurate sub-Ångström structural investigations on macromolecules, as only this experimental configuration can deliver the uniform spatial distribution of dose necessary to resolve the fine details of progressive radiation damage: a capability for which we propose the term `resolution in dose'.
Analytical expressions for the damage-limited resolution (DLR) are developed and applied to X-ray and electron imaging of beam-sensitive specimens, allowing for variation of the characteristic radiation dose with spatial resolution. The dependence of DLR on specimen thickness is illustrated for the common modes of X-ray and transmission electron-microscope imaging. Similarities and differences between the radiolysis damage caused by electrons and X-rays are discussed. The meaning of a 'Bragg boost' in diffracted intensity is discussed.
In this study, two methods are described for visualizing the H atoms in cryoEM maps of macromolecules. The first involves comparing experimental charge-density (CD) maps with calculated CD maps computed using structural models from which hydrogens are excluded. After the hydrogen-free model has been refined into the experimental map, the vector-difference Fourier map one computes, i.e. the (Fobs, αobs) - (Fcalc, αcalc) residual CD map, will reveal the H atoms missing from the model. The second method is model-independent and is known as a shoulder-peak decomposition method. It relies on the difference in the one-dimensional CD profile of a non-H atom between the side of the atom that adjoins a H atom and the side that does not. Here, the utility of both methods is demonstrated using both cryoEM CD maps derived from experimental electrostatic potential (ESP) maps for mouse heavy-chain ferritin at 1.09 Å resolution and X-ray crystallographic electron-density (ED) maps reported for the human enzyme at 1.06 Å resolution. A comparison shows that hydrogen signals in cryoEM maps are about 5-10 times stronger than those of X-ray crystallographic ED maps. Both methods should be applicable to cryoEM maps with a wide range of resolutions. The model-independent method is important for nucleoprotein complexes such as the ribosome because it bypasses the necessity of modeling the contribution that unscreened atomic partial charges make to cryoEM maps, which are difficult to model properly.
Flavodiiron proteins (FDPs) are NO and/or O-2 reductases which contain a diiron catalytic center. Interestingly, they exhibit different selectivities towards each one of these substrates, despite having the same ligands of the iron ions. Escherichia coli FDP is a selective NO reductase that protects this bacterium against nitric oxide by catalyzing two-electron reduction to the nontoxic N2O. Previously, based on kinetic studies, we explored the possible role of two amino acids located in the di-iron second coordination sphere, Lys53 and Tyr271, in modulation of the substrate selectivity of Entamoeba histolytica FDP, a selective O-2 reductase. In this work, we replaced the structurally equivalent residues in E. coli FDP, Asp52 and Ser262, by those present in the O-2-selective FDP and determined their crystal structures in both oxidized and reduced states. Furthermore, the molecular-substrate tunnels were experimentally identified using krypton pressurization of the crystals. The data obtained corroborated previous molecular-dynamics calculations on this FDP. The side chains of residues in both positions 52 and 262 of E. coli FDP variants and wild type are in the vicinity of the shorter intramolecular tunnel, which is suggested to be the exit route for the reaction products N2O and H2O. The E. coli FDP S262Y variant shows photoreduction of the di-iron center and partial loss of electron density in some of its coordinating ligands after X-ray exposure, and these effects are consistent with increased radiation sensitivity. The kinetic properties of the variants towards NO and O-2 were not significantly different from the wild type, contrary to what was observed previously for E. histolytica FDP.
A comparison is provided between the use of electrons and X-rays for collecting diffraction data from small protein crystals and imaging data from cells and tissues. The paper contains a review element written to enable an understanding of the relevant properties of electrons by those (including one of the authors) more used to X-ray imaging and diffraction. Radiation-damage mechanisms, sample thickness-dependent dose efficiency and energy-dependent scattering cross sections are discussed, together with contrast mechanisms in electron and X-ray imaging. Crossover points are calculated for diffraction from crystals where electrons and X-rays could yield equivalent data quality in the presence of radiation damage. Increasing the electron energy from 300 to 1000 keV results in a ∼43% rise in the electron/X-ray crossover point. However, the maximum information coefficient (useful signal/absorbed dose) for electrons alone occurs for a 250 nm crystal examined at around 800 keV. The impact of inelastic scattering on electron imaging and diffraction is examined, including its role in coherence loss, Bragg spot broadening and background elevation. The possibilities are investigated for locating regions of interest using X-rays for subsequent higher resolution imaging using electrons. For locating a 30 nm diameter protein or virus, the required X-ray dose would be much less than the tolerable dose for electron imaging at 0.5 or 0.25 nm. Overall, these findings are relevant for imaging at different length scales while minimizing dose-induced structural damage.
Metalloenzymes containing a heme cofactor catalyse a wide range of oxidative reactions critical to life. Understanding the structure and electronic states of the heme across the catalytic cycle is essential in understanding the oxidative chemistry performed on the substrate. This work demonstrates in crystallo manipulation of the heme-iron oxidation state in a B-type dye-decolourizing peroxidase from Streptomyces lividans (DtpB) using multiple, complementary, serial crystallography approaches. Fixed-target drop-on-chip serial femtosecond crystallography (SFX) together with dose-resolved serial synchrotron crystallography (SSX) allowed DtpB to be driven between multiple iron oxidation states. Drop-on-chip addition of hydrogen peroxide with fixed-target SFX is used to generate a ferryl [Fe(IV)=O] species, while the X-ray-driven approach modulates the iron oxidation state, with an apparent two-electron reduction leading to a return to a ferric state. The formation and dose response of the Fe(IV)-O state is highly variable between the chemically identical heme groups of the DtpB hexamer, highlighting the importance of understanding the effect of the crystalline lattice on observed changes in time- and dose-resolved crystallography.
