Deep Learning Enables Automatic Correction of Experimental HDX-MS Data with Applications in Protein Modeling

Observed mass shifts associated with deuterium incorporation in hydrogen-deuterium exchange mass spectrometry (HDX-MS) frequently deviate from the initial signals due to back and forward exchange. In typical HDX-MS experiments, the impact of these disparities on data interpretation is generally low because relative and not absolute mass changes are investigated. However, for more advanced data processing including optimization, experimental error correction is imperative for accurate results. Here the potential for automatic HDX-MS data correction using models generated by deep neural networks is demonstrated. A multilayer perceptron (MLP) is used to learn a mapping between uncorrected HDX-MS data and data with mass shifts corrected for back and forward exchange. The model is rigorously tested at various levels including peptide level mass changes, residue level protection factors following optimization, and ability to correctly identify native protein folds using HDX-MS guided protein modeling. AI is shown to demonstrate considerable potential for amending HDX-MS data and improving fidelity across all levels. With access to big data, online tools may eventually be able to predict corrected mass shifts in HDX-MS profiles. This should improve throughput in workflows that require the reporting of real mass changes as well as allow retrospective correction of historic profiles to facilitate new discoveries with these data.

Mapping HDX-MS Data to Protein Conformations through Training Ensemble-Based Models

An original approach that adopts machine learning inference to predict protein structural information using hydrogen-deuterium exchange mass spectrometry (HDX-MS) is described. The method exploits an in-house optimization program that increases the resolution of HDX-MS data from peptides to amino acids. A system is trained using Gradient Tree Boosting as a type of machine learning ensemble technique to assign a protein secondary structure. Using limited training data we generate a discriminative model that uses optimized HDX-MS data to predict protein secondary structure with an accuracy of 75%. This research could form the basis for new methods exploiting artificial intelligence to model protein conformations by HDX-MS.

Computational Modeling of Molecular Structures Guided by Hydrogen-Exchange Data

Data produced by hydrogen-exchange monitoring experiments have been used in structural studies of molecules for several decades. Despite uncertainties about the structural determinants of hydrogen exchange itself, such data have successfully helped guide the structural modeling of challenging molecular systems, such as membrane proteins or large macromolecular complexes. As hydrogen-exchange monitoring provides information on the dynamics of molecules in solution, it can complement other experimental techniques in so-called integrative modeling approaches. However, hydrogen-exchange data have often only been used to qualitatively assess molecular structures produced by computational modeling tools. In this paper, we look beyond qualitative approaches and survey the various paradigms under which hydrogen-exchange data have been used to quantitatively guide the computational modeling of molecular structures. Although numerous prediction models have been proposed to link molecular structure and hydrogen exchange, none of them has been widely accepted by the structural biology community. Here, we present as many hydrogen-exchange prediction models as we could find in the literature, with the aim of providing the first exhaustive list of its kind. From purely structure-based models to so-called fractional-population models or knowledge-based models, the field is quite vast. We aspire for this paper to become a resource for practitioners to gain a broader perspective on the field and guide research toward the definition of better prediction models. This will eventually improve synergies between hydrogen-exchange monitoring and molecular modeling.

Quantitative Evaluation of Native Protein Folds and Assemblies by Hydrogen Deuterium Exchange Mass Spectrometry (HDX-MS)

Hydrogen deuterium exchange mass spectrometry (HDX-MS) has significant potential for protein structure initiatives but its relationship with protein conformations is unclear. We report on the efficacy of HDX-MS to distinguish between native and non-native proteins using a popular approach to calculate HDX protection factors (PFs) from protein structures. The ability of HDX-MS to identify native protein conformations is quantified by binary structural classification such that merits of the approach for protein modelling can be quantified and better understood. We show that highly accurate PF calculations are not a prerequisite for HDX-MS simulations that are capable of effectively discriminating between native and non-native protein folds. The simulations can also be performed directly on unique structures facilitating high-throughput evaluation of many alternate conformations. The ability of HDX-MS to classify the conformations of homo-protein assemblies is also investigated. In contrast to protein monomers, we show a significant lack of correspondence between the simulated and experimental HDX-MS data for these systems with a subsequent decrease in the ability of HDX-MS to identify native states. However, we demonstrate surprisingly high diagnostic ability of the simulated data for assemblies in which a significant proportion of the individual chains occupy protein-protein interfaces. We relate this to the number of peptides that can sample alternate subunit orientations and discuss these observations within the larger context of applying HDX-MS to evaluate protein structures. Graphical Abstract.

Structure and dynamics of a protein-surfactant assembly studied by ion-mobility mass spectrometry and molecular dynamics simulations

The structure and dynamics of a protein-surfactant assembly studied by ion-mobility mass spectrometry (IMS) and vacuum molecular dynamics (MD) simulations is reported. Direct evidence is provided for the ability of the surfactant dodecyl-β-D-maltoside (DDM) to prevent charge-induced unfolding of the membrane protein (PagP) in the gas-phase. Restraints obtained by IMS are used to map the surfactant positions onto the protein surface. Surfactants occupying more exposed positions at the apexes of the β-barrel structure are most in-line with the experimental observations. MD simulations provide additional evidence for this assembly organization through surfactant inversion and migration on the protein structure in the absence of solvent. Surfactant migration entails a net shift from apolar membrane spanning regions to more polar regions of the protein structure with the DDM molecule remaining attached to the protein via headgroup interactions. These data provide evidence for the role of protein-DDM headgroup interactions in stabilizing membrane protein structure from gas-phase unfolding.

Charge reduction stabilizes intact membrane protein complexes for mass spectrometry

The study of intact soluble protein assemblies by means of mass spectrometry is providing invaluable contributions to structural biology and biochemistry. A recent breakthrough has enabled similar study of membrane protein complexes, following their release from detergent micelles in the gas phase. Careful optimization of mass spectrometry conditions, particularly with respect to energy regimes, is essential for maintaining compact folded states as detergent is removed. However, many of the saccharide detergents widely employed in structural biology can cause unfolding of membrane proteins in the gas phase. Here, we investigate the potential of charge reduction by introducing three membrane protein complexes from saccharide detergents and show how reducing their overall charge enables generation of compact states, as evidenced by ion mobility mass spectrometry. We find that charge reduction stabilizes the oligomeric state and enhances the stability of lipid-bound complexes. This finding is significant since maintaining native-like membrane proteins enables ligand binding to be assessed from a range of detergents that retain solubility while protecting the overall fold.

Detergent release prolongs the lifetime of native-like membrane protein conformations in the gas-phase

Recent studies have suggested that detergents can protect the structure of membrane proteins during their transition from solution to the gas-phase. Here we provide mechanistic insights into this process by interrogating the structures of membrane protein-detergent assemblies in the gas-phase using ion mobility mass spectrometry. We show a clear correlation between the population of native-like protein conformations and the degree of detergent attachment to the protein in the gas-phase. Interrogation of these protein-detergent assemblies, by tandem mass spectrometry, enables us to define the mechanism by which detergents preserve native-like protein conformations in a solvent free environment. We show that the release of detergent is more central to the survival of these conformations than the physical presence of detergent bound to the protein. We propose that detergent release competes with structural collapse for the internal energy of the ion and permits the observation of transient native-like membrane protein conformations that are otherwise lost to structural rearrangement in the gas-phase.

The 'sticky business' of cleaning gas-phase membrane proteins: a detergent oriented perspective

In recent years the properties of gas-phase detergent clusters have come under close scrutiny due in part to their participation in the analysis of intact membrane protein complexes by mass spectrometry. The detergent molecules that cover the protein complex are removed in the gas-phase by thermally agitating the ions by collision-induced dissociation. This process however, is not readily controlled and can frequently result in the disruption of protein structure. Improved methods of releasing proteins from detergent clusters are clearly required. To facilitate this the structural properties of detergent clusters along with the mechanistic details of their dissociation need to be understood. Pivotal to understanding the properties of gas-phase detergent clusters is the technique of ion mobility mass spectrometry. This technique can be used to assign polydisperse detergent clusters and provide information about their geometries and packing densities. In this article we consider the shapes of detergent clusters and show that these clusters possess geometries that are inconsistent with those in solution. We analyse the distributions of clusters in detail using tandem mass spectrometry and suggest that the mean charge of clusters formed from certain detergents is governed by electrostatic repulsion. We discuss the dissociation of detergent clusters and propose that detergent evaporation it a key process in the protection of protein complexes during high energy collisions in the gas-phase.

Formation and dissociation processes of gas-phase detergent micelles

Growing interest in micelles to protect membrane complexes during the transition from solution to gas phase prompts a better understanding of their properties. We have used ion mobility mass spectrometry to separate and assign detergent clusters formed from the n-trimethylammonium bromide series of detergents. We show that cluster size is independent of detergent concentration in solution, increases with charge state, but surprisingly decreases with alkyl chain length. This relationship contradicts the thermodynamics of micelle formation in solution. However, the liquid drop model, which considers both the surface energy and charge, correlates extremely well with the experimental cluster size. To explore further the properties of gas-phase micelles, we have performed collision-induced dissociation on them during tandem mass spectrometry. We observed both sequential asymmetric charge separation and neutral evaporation from the precursor ion cluster. Interestingly, however, we also found markedly different dissociation pathways for the longer alkyl chain detergents, with significantly fewer intermediate ions formed than for those with a shorter alkyl chain. These experiments provide an essential foundation for understanding the process of the gas-phase analysis of membrane protein complexes. Moreover they imply valuable mechanistic details of the protection afforded to protein complexes by detergent clusters during gas-phase activation processes.

Rapid odorant release in mammalian odour binding proteins facilitates their temporal coupling to odorant signals

We have measured the effect of rat odorant-binding protein 1 on the rates of ligand uptake and liquid-to-air transfer rates with a set of defined odorous compounds. Comparison of observed rate constants (k(obs)) with data simulated over a wide range of different kinetic and thermodynamic regimes shows that the data do not agree with the previously held view of a slow off-rate regime (k(off) <0.0004 s(-1)). We propose that a rapid k(off) would be a necessary requirement for such a system, since slow odorant-release rates would result in significant decorrelation between the olfactory world and odour perception.

Structure of rat odorant-binding protein OBP1 at 1.6 A resolution

The nasal mucosa is a specialist interfacial region sandwiched between the olfactory system and the gaseous chemical milieu. In mammals and insects, this region is rich in odorant-binding proteins that are thought to aid olfaction by assisting mass transfer of the many different organoleptic compounds that make up the olfactory landscape. However, in mammals at least, our grasp on the exact function of odorant-binding proteins is tentative and better insight into the role of these proteins is warranted, not least because of their apparent significance in the olfactory systems of insects. Here, the crystal structure of rat odorant-binding protein 1 is reported at 1.6 A resolution. This protein is one of the best-characterized mammalian odorant-binding proteins and only the third such protein structure to be solved at high resolution. The protein was crystallized in the holo form and contains an unidentifiable ligand that is probably an artefact from the Pichia pastoris expression system. Comparisons are made between this structure and a modelled OBP1 structure produced using the crystal structure of aphrodisin as a template. Comparisons are also made between OBP1 and the other two rat OBP subtypes, for which crystallographic data are unavailable. Interestingly, we also show that OBP1 is monomeric, which is in contrast to its previous assignment.

Specific glycosaminoglycans promote unseeded amyloid formation from β2-microglobulin under physiological conditions

Dialysis-related amyloidosis (DRA) is a complication of hemodialysis where beta2-microglobulin (beta2m) forms plaques mainly in cartilaginous tissues. The tissue-specific deposition, along with a known intransigence of pure beta2m to form fibrils in vitro at neutral pH in the absence of preformed fibrillar seeds, suggests a role for factors within cartilage in enhancing amyloid formation from this protein. To identify these factors, we determined the ability of a derivative lacking the N-terminal six amino acids found in ex vivo beta2m amyloid deposits to form amyloid fibrils at pH 7.4 in the absence of fibrillar seeds. We show that the addition of the glycosaminoglycans (GAGs) chrondroitin-4 or 6-sulfate to fibril growth assays results in the spontaneous generation of amyloid-like fibrils. By contrast, no fibrils are observed over the same time course in the presence of hyaluronic acid, a nonsulfated GAG that is abundant in cartilaginous joints. Based on the observation that hyaluronic acid has no effect on fibril stability, while chrondroitin-6-sulfate decreases the rate of fibril disassembly, we propose that the latter GAG enhances amyloid formation by stabilizing the rare fibrils that form spontaneously. This leads to the accumulation of beta2m in fibrillar deposits. Our data rationalize the joint-specific deposition of beta2m amyloid in DRA, suggesting mechanisms by which amyloid formation may be promoted.