Large-scale clustering of AlphaFold2 3D models shines light on the structure and function of proteins.

Two recent studies exploited ultra-fast structural aligners and deep-learning approaches to cluster the protein structure space in the AlphaFold Database. Barrio-Hernandez et al.1 and Durairaj et al.2 uncovered fascinating new protein functions and structural features previously unknown.

Ultrafast end-to-end protein structure prediction enables high-throughput exploration of uncharacterized proteins.

Deep learning-based prediction of protein structure usually begins by constructing a multiple sequence alignment (MSA) containing homologs of the target protein. The most successful approaches combine large feature sets derived from MSAs, and considerable computational effort is spent deriving these input features. We present a method that greatly reduces the amount of preprocessing required for a target MSA, while producing main chain coordinates as a direct output of a deep neural network. The network makes use of just three recurrent networks and a stack of residual convolutional layers, making the predictor very fast to run, and easy to install and use. Our approach constructs a directly learned representation of the sequences in an MSA, starting from a one-hot encoding of the sequences. When supplemented with an approximate precision matrix, the learned representation can be used to produce structural models of comparable or greater accuracy as compared to our original DMPfold method, while requiring less than a second to produce a typical model. This level of accuracy and speed allows very large-scale three-dimensional modeling of proteins on minimal hardware, and we demonstrate this by producing models for over 1.3 million uncharacterized regions of proteins extracted from the BFD sequence clusters. After constructing an initial set of approximate models, we select a confident subset of over 30,000 models for further refinement and analysis, revealing putative novel protein folds. We also provide updated models for over 5,000 Pfam families studied in the original DMPfold paper.

Deuteros 2.0: peptide-level significance testing of data from hydrogen deuterium exchange mass spectrometry.

SUMMARY: Hydrogen deuterium exchange mass spectrometry (HDX-MS) is becoming increasing routine for monitoring changes in the structural dynamics of proteins. Differential HDX-MS allows comparison of protein states, such as in the absence or presence of a ligand. This can be used to attribute changes in conformation to binding events, allowing the mapping of entire conformational networks. As such, the number of necessary cross-state comparisons quickly increases as additional states are introduced to the system of study. There are currently very few software packages available that offer quick and informative comparison of HDX-MS datasets and even fewer which offer statistical analysis and advanced visualization. Following the feedback from our original software Deuteros, we present Deuteros 2.0 which has been redesigned from the ground up to fulfill a greater role in the HDX-MS analysis pipeline. Deuteros 2.0 features a repertoire of facilities for back exchange correction, data summarization, peptide-level statistical analysis and advanced data plotting features. AVAILABILITY AND IMPLEMENTATION: Deuteros 2.0 can be downloaded for both Windows and MacOS from https://github.com/andymlau/Deuteros_2.0 under the Apache 2.0 license.

Moderated Test Statistics to Detect Differential Deuteration in Hydrogen/Deuterium Exchange Mass Spectrometry Experiments.

With differential hydrogen/deuterium exchange, differences in the structure and dynamics of protein states can be studied. Detecting statistically significant differentially deuterated peptides is crucial to draw meaningful conclusions about the distinct conformations and dynamics of the protein under study. Here, we introduced a linear model in combination with an empirical Bayes approach to detect differentially deuterated peptides. Using a linear model allows one to test for differences in deuteration between two (two-sample t-test) or more groups (F-statistic), while potentially controlling for the effects of other variables that are not of interest. The empirical Bayes approach improves the estimation of deuteration-level variances, especially in experiments with a low number of replicates. As a consequence, the two sample t-tests and the F-statistic become moderated, resulting in a lower number of false positive and false negative findings. Furthermore, we introduce a thresholded-moderated t-statistic to test if the observed deuteration differences are larger than a specified, biologically relevant difference. Finally, we underline the importance of having a sufficient number of replicates, and the effect of the number of replicates on the power of the statistical significance tests. The R-code for the proposed moderated test statistics is available upon request.

Deuteros: software for rapid analysis and visualization of data from differential hydrogen deuterium exchange-mass spectrometry.

SUMMARY: Hydrogen deuterium exchange-mass spectrometry (HDX-MS) has emerged as a powerful technique for interrogating the conformational dynamics of proteins and their complexes. Currently, analysis of HDX-MS data remains a laborious procedure, mainly due to the lack of streamlined software to process the large datasets. We present Deuteros which is a standalone software designed to be coupled with Waters DynamX HDX data analysis software, allowing the rapid analysis and visualization of data from differential HDX-MS. AVAILABILITY AND IMPLEMENTATION: Deuteros is open-source and can be downloaded from https://github.com/andymlau/Deuteros, under the Apache 2.0 license. Written in MATLAB and supported on both Windows and MacOS. Requires the MATLAB runtime library. According to the Wellcome Trust and UK research councils' Common Principles on Data Policy on data, software and materials management and sharing, all data supporting this study will be openly available from the software repository.

Structural predictions of the functions of membrane proteins from HDX-MS.

HDX-MS has emerged as a powerful tool to interrogate the structure and dynamics of proteins and their complexes. Recent advances in the methodology and instrumentation have enabled the application of HDX-MS to membrane proteins. Such targets are challenging to investigate with conventional strategies. Developing new tools are therefore pertinent for improving our fundamental knowledge of how membrane proteins function in the cell. Importantly, investigating this central class of biomolecules within their native lipid environment remains a challenge but also a key goal ahead. In this short review, we outline recent progresses in dissecting the conformational mechanisms of membrane proteins using HDX-MS. We further describe how the use of computational strategies can aid the interpretation of experimental data and enable visualisation of otherwise intractable membrane protein states. This unique integration of experiments with computations holds significant potential for future applications.

Crystal structure of zinc-α2-glycoprotein in complex with a fatty acid reveals multiple different modes of protein-lipid binding.

Human zinc-α2-glycoprotein (ZAG) is a 42 kDa adipokine which regulates body fat mass and is associated with cachexia and obesity. ZAG belongs to the major histocompatibility complex class I protein family and binds long-chain polyunsaturated fatty acids in its groove formed from the α1 and α2 domains. To identify the molecular basis of its lipid-binding function, we determined the first crystal structure at 2.49 Šresolution for fatty acid-bound ZAG, where the ligand was the fluorescent 11-(dansylamino)undecanoic acid (DAUDA). The 192 kDa crystallographic asymmetric unit contained six ZAG and eight fatty acid molecules in unique conformations. Six fatty acid molecules were localised to the ZAG grooves, where their tails were bound in two distinct conformations. The carboxylate groups of three fatty acids projected out of the groove, while the fourth was hydrogen bonded with R73 inside the groove. Other ligand-residue contacts were primarily hydrophobic. A new fatty acid site was revealed for two further DAUDA molecules at the ZAG α3 domains. Following conformational changes from unbound ZAG, the α3 domains formed tetrameric ß-barrel structures lined by fatty acid molecules that doubled the binding capacity of ZAG. Analytical ultracentrifugation revealed that ZAG in solution was a monomer in the absence of DAUDA, but formed small amounts of tetramers with DAUDA. By showing that ZAG binds fatty acids in different locations, we demonstrate an augmented mechanism for fatty acid binding in ZAG that is distinct from other known fatty acid binding proteins, and may be relevant to cachexia.

Mechanistic insight into the assembly of the HerA-NurA helicase-nuclease DNA end resection complex.

The HerA-NurA helicase-nuclease complex cooperates with Mre11 and Rad50 to coordinate the repair of double-stranded DNA breaks. Little is known, however, about the assembly mechanism and activation of the HerA-NurA. By combining hybrid mass spectrometry with cryo-EM, computational and biochemical data, we investigate the oligomeric formation of HerA and detail the mechanism of nucleotide binding to the HerA-NurA complex from thermophilic archaea. We reveal that ATP-free HerA and HerA-DNA complexes predominantly exist in solution as a heptamer and act as a DNA loading intermediate. The binding of either NurA or ATP stabilizes the hexameric HerA, indicating that HerA-NurA is activated by substrates and complex assembly. To examine the role of ATP in DNA translocation and processing, we investigated how nucleotides interact with the HerA-NurA. We show that while the hexameric HerA binds six nucleotides in an 'all-or-none' fashion, HerA-NurA harbors a highly coordinated pairwise binding mechanism and enables the translocation and processing of double-stranded DNA. Using molecular dynamics simulations, we reveal novel inter-residue interactions between the external ATP and the internal DNA binding sites. Overall, here we propose a stepwise assembly mechanism detailing the synergistic activation of HerA-NurA by ATP, which allows efficient processing of double-stranded DNA.

A Mass-Spectrometry-Based Modelling Workflow for Accurate Prediction of IgG Antibody Conformations in the Gas Phase.

Immunoglobulins are biomolecules involved in defence against foreign substances. Flexibility is key to their functional properties in relation to antigen binding and receptor interactions. We have developed an integrative strategy combining ion mobility mass spectrometry (IM-MS) with molecular modelling to study the conformational dynamics of human IgG antibodies. Predictive models of all four human IgG subclasses were assembled and their dynamics sampled in the transition from extended to collapsed state during IM-MS. Our data imply that this collapse of IgG antibodies is related to their intrinsic structural features, including Fab arm flexibility, collapse towards the Fc region, and the length of their hinge regions. The workflow presented here provides an accurate structural representation in good agreement with the observed collision cross section for these flexible IgG molecules. These results have implications for studying other nonglobular flexible proteins.

Surface Accessibility and Dynamics of Macromolecular Assemblies Probed by Covalent Labeling Mass Spectrometry and Integrative Modeling.

Mass spectrometry (MS) has become an indispensable tool for investigating the architectures and dynamics of macromolecular assemblies. Here we show that covalent labeling of solvent accessible residues followed by their MS-based identification yields modeling restraints that allow mapping the location and orientation of subunits within protein assemblies. Together with complementary restraints derived from cross-linking and native MS, we built native-like models of four heterocomplexes with known subunit structures and compared them with available X-ray crystal structures. The results demonstrated that covalent labeling followed by MS markedly increased the predictive power of the integrative modeling strategy enabling more accurate protein assembly models. We applied this strategy to the F-type ATP synthase from spinach chloroplasts (cATPase) providing a structural basis for its function as a nanomotor. By subjecting the models generated by our restraint-based strategy to molecular dynamics (MD) simulations, we revealed the conformational states of the peripheral stalk and assigned flexible regions in the enzyme. Our strategy can readily incorporate complementary chemical labeling strategies and we anticipate that it will be applicable to many other systems providing new insights into the structure and function of protein complexes.

Zinc-induced oligomerization of zinc α2 glycoprotein reveals multiple fatty acid-binding sites.

Zinc α2 glycoprotein (ZAG) is an adipokine with a class I MHC protein fold and is associated with obesity and diabetes. Although its intrinsic ligand remains unknown, ZAG binds the dansylated C11 fatty acid 11-(dansylamino)undecanoic acid (DAUDA) in the groove between the α1 and α2 domains. The surface of ZAG has approximately 15 weak zinc-binding sites deemed responsible for precipitation from human plasma. In the present study the functional significance of these metal sites was investigated. Analytical ultracentrifugation (AUC) and CD showed that zinc, but not other divalent metals, causes ZAG to oligomerize in solution. Thus ZAG dimers and trimers were observed in the presence of 1 and 2 mM zinc. Molecular modelling of X-ray scattering curves and sedimentation coefficients indicated a progressive stacking of ZAG monomers, suggesting that the ZAG groove may be occluded in these. Using fluorescence-detected sedimentation velocity, these ZAG-zinc oligomers were again observed in the presence of the fluorescent boron dipyrromethene fatty acid C16-BODIPY (4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-hexadecanoic acid). Fluorescence spectroscopy confirmed that ZAG binds C16-BODIPY. ZAG binding to C16-BODIPY, but not to DAUDA, was reduced by increased zinc concentrations. We conclude that the lipid-binding groove in ZAG contains at least two distinct fatty acid-binding sites for DAUDA and C16-BODIPY, similar to the multiple lipid binding seen in the structurally related immune protein CD1c. In addition, because high concentrations of zinc occur in the pancreas, the perturbation of these multiple lipid-binding sites by zinc may be significant in Type 2 diabetes where dysregulation of ZAG and zinc homoeostasis occurs.

Machine learning methods for predicting protein structure from single sequences.

Recent breakthroughs in protein structure prediction have increasingly relied on the use of deep neural networks. These recent methods are notable in that they produce 3-D atomic coordinates as a direct output of the networks, a feature which presents many advantages. Although most techniques of this type make use of multiple sequence alignments as their primary input, a new wave of methods have attempted to use just single sequences as the input. We discuss the make-up and operating principles of these models, and highlight new developments in these areas, as well as areas for future development.

Merizo: a rapid and accurate protein domain segmentation method using invariant point attention.

The AlphaFold Protein Structure Database, containing predictions for over 200 million proteins, has been met with enthusiasm over its potential in enriching structural biological research and beyond. Currently, access to the database is precluded by an urgent need for tools that allow the efficient traversal, discovery, and documentation of its contents. Identifying domain regions in the database is a non-trivial endeavour and doing so will aid our understanding of protein structure and function, while facilitating drug discovery and comparative genomics. Here, we describe a deep learning method for domain segmentation called Merizo, which learns to cluster residues into domains in a bottom-up manner. Merizo is trained on CATH domains and fine-tuned on AlphaFold2 models via self-distillation, enabling it to be applied to both experimental and AlphaFold2 models. As proof of concept, we apply Merizo to the human proteome, identifying 40,818 putative domains that can be matched to CATH representative domains.

Identification of diverse lipid-binding modes in the groove of zinc α2 glycoprotein reveals its functional versatility.

ZAG is a multifunctional glycoprotein with a class I MHC-like protein fold and an α1-α2 lipid-binding groove. The intrinsic ZAG ligand is unknown. Our previous studies showed that ZAG binds the dansylated C11 fatty acid, DAUDA, differently to the boron dipyrromethane C16 fatty acid, C16 -BODIPY. Here, the molecular basis for this difference was elucidated. Multi-wavelength analytical ultracentrifugation confirmed that DAUDA and C16 -BODIPY individually bind to ZAG and compete for the same binding site. Molecular docking of lipid-binding in the structurally related Cluster of differentiation 1 proteins predicted nine conserved ligand contact residues in ZAG. Twelve mutants were accordingly created by alanine scanning site directed mutagenesis for characterisation. Mutation of Y12 caused ZAG to misfold. Mutation of K147, R157 and A158 abrogated C16 -BODIPY but not DAUDA binding. L69 and T169 increased the fluorescence emission intensity of C16 -BODIPY but not of DAUDA compared to wild-type ZAG and showed that C16 -BODIPY binds close to T169 and L69. Distance measurements of the crystal structure revealed K147 forms a salt bridge with D83. A range of bioactive bulky lipids including phospholipids and sphingolipids displaced DAUDA from the ZAG binding site but unexpectedly did not displace C16 -BODIPY. We conclude that the ZAG α1-α2 groove contains separate but overlapping sites for DAUDA and C16 -BODIPY and is involved in binding to a bulkier and wider repertoire of lipids than previously reported. This work suggested that the in vivo activity of ZAG may be dictated by its lipid ligand.

Integrative Mass Spectrometry-Based Approaches for Modeling Macromolecular Assemblies.

Mass spectrometry (MS)-based strategies have emerged as key elements for structural modeling of proteins and their assemblies. In particular, merging together complementary MS tools, through the so-called hybrid approaches, has enabled structural characterization of proteins in their near-native states. Here, we describe how different MS techniques, such as native MS, chemical cross-linking MS, and ion mobility MS, are brought together using sophisticated computational algorithms and modeling restraints. We demonstrate the applicability of the strategy by building accurate models of multimeric protein assemblies. These strategies can practically be applied to any protein complex of interest and be readily integrated with other structural approaches such as electron density maps from cryo-electron microscopy.

Analysis of protein-DNA interactions in chromatin by UV induced cross-linking and mass spectrometry.

Protein-DNA interactions are key to the functionality and stability of the genome. Identification and mapping of protein-DNA interaction interfaces and sites is crucial for understanding DNA-dependent processes. Here, we present a workflow that allows mass spectrometric (MS) identification of proteins in direct contact with DNA in reconstituted and native chromatin after cross-linking by ultraviolet (UV) light. Our approach enables the determination of contact interfaces at amino-acid level. With the example of chromatin-associated protein SCML2 we show that our technique allows differentiation of nucleosome-binding interfaces in distinct states. By UV cross-linking of isolated nuclei we determined the cross-linking sites of several factors including chromatin-modifying enzymes, demonstrating that our workflow is not restricted to reconstituted materials. As our approach can distinguish between protein-RNA and DNA interactions in one single experiment, we project that it will be possible to obtain insights into chromatin and its regulation in the future.

Hydrogen-deuterium exchange mass spectrometry captures distinct dynamics upon substrate and inhibitor binding to a transporter.

Proton-coupled transporters use transmembrane proton gradients to power active transport of nutrients inside the cell. High-resolution structures often fail to capture the coupling between proton and ligand binding, and conformational changes associated with transport. We combine HDX-MS with mutagenesis and MD simulations to dissect the molecular mechanism of the prototypical transporter XylE. We show that protonation of a conserved aspartate triggers conformational transition from outward-facing to inward-facing state. This transition only occurs in the presence of substrate xylose, while the inhibitor glucose locks the transporter in the outward-facing state. MD simulations corroborate the experiments by showing that only the combination of protonation and xylose binding, and not glucose, sets up the transporter for conformational switch. Overall, we demonstrate the unique ability of HDX-MS to distinguish between the conformational dynamics of inhibitor and substrate binding, and show that a specific allosteric coupling between substrate binding and protonation is a key step to initiate transport.

Perturbed structural dynamics underlie inhibition and altered efflux of the multidrug resistance pump AcrB.

Resistance-nodulation-division efflux pumps play a key role in inherent and evolved multidrug resistance in bacteria. AcrB, a prototypical member of this protein family, extrudes a wide range of antimicrobial agents out of bacteria. Although high-resolution structures exist for AcrB, its conformational fluctuations and their putative role in function are largely unknown. Here, we determine these structural dynamics in the presence of substrates using hydrogen/deuterium exchange mass spectrometry, complemented by molecular dynamics simulations, and bacterial susceptibility studies. We show that an efflux pump inhibitor potentiates antibiotic activity by restraining drug-binding pocket dynamics, rather than preventing antibiotic binding. We also reveal that a drug-binding pocket substitution discovered within a multidrug resistant clinical isolate modifies the plasticity of the transport pathway, which could explain its altered substrate efflux. Our results provide insight into the molecular mechanism of drug export and inhibition of a major multidrug efflux pump and the directive role of its dynamics.

Integrating hydrogen-deuterium exchange mass spectrometry with molecular dynamics simulations to probe lipid-modulated conformational changes in membrane proteins.

Biological membranes define the boundaries of cells and are composed primarily of phospholipids and membrane proteins. It has become increasingly evident that direct interactions of membrane proteins with their surrounding lipids play key roles in regulating both protein conformations and function. However, the exact nature and structural consequences of these interactions remain difficult to track at the molecular level. Here, we present a protocol that specifically addresses this challenge. First, hydrogen-deuterium exchange mass spectrometry (HDX-MS) of membrane proteins incorporated into nanodiscs of controlled lipid composition is used to obtain information on the lipid species that are involved in modulating the conformational changes in the membrane protein. Then molecular dynamics (MD) simulations in lipid bilayers are used to pinpoint likely lipid-protein interactions, which can be tested experimentally using HDX-MS. By bringing together the MD predictions with the conformational readouts from HDX-MS, we have uncovered key lipid-protein interactions implicated in stabilizing important functional conformations. This protocol can be applied to virtually any integral membrane protein amenable to classic biophysical studies and for which a near-atomic-resolution structure or homology model is available. This protocol takes ~4 d to complete, excluding the time for data analysis and MD simulations, which depends on the size of the protein under investigation.

Structural basis of Cullin 2 RING E3 ligase regulation by the COP9 signalosome.

Cullin-Ring E3 Ligases (CRLs) regulate a multitude of cellular pathways through specific substrate receptors. The COP9 signalosome (CSN) deactivates CRLs by removing NEDD8 from activated Cullins. Here we present structures of the neddylated and deneddylated CSN-CRL2 complexes by combining single-particle cryo-electron microscopy (cryo-EM) with chemical cross-linking mass spectrometry (XL-MS). These structures suggest a conserved mechanism of CSN activation, consisting of conformational clamping of the CRL2 substrate by CSN2/CSN4, release of the catalytic CSN5/CSN6 heterodimer and finally activation of the CSN5 deneddylation machinery. Using hydrogen-deuterium exchange (HDX)-MS we show that CRL2 activates CSN5/CSN6 in a neddylation-independent manner. The presence of NEDD8 is required to activate the CSN5 active site. Overall, by synergising cryo-EM with MS, we identify sensory regions of the CSN that mediate its stepwise activation and provide a framework for understanding the regulatory mechanism of other Cullin family members.