Structures Of Protein Complexes Research Articles

Cross-linking mass spectrometry discovers, evaluates, and corroborates structures and protein–protein interactions in the human cell

Significant recent advances in structural biology, particularly in the field of cryoelectron microscopy, have dramatically expanded our ability to create structural models of proteins and protein complexes. However, many proteins remain refractory to these approaches because of their low abundance, low stability, or-in the case of complexes-simply not having yet been analyzed. Here, we demonstrate the power of using cross-linking mass spectrometry (XL-MS) for the high-throughput experimental assessment of the structures of proteins and protein complexes. This included those produced by high-resolution but invitro experimental data, as well as in silico predictions based on amino acid sequence alone. We present the largest XL-MS dataset to date, describing 28,910 unique residue pairs captured across 4,084 unique human proteins and 2,110 unique protein-protein interactions. We show that models of proteins and their complexes predicted by AlphaFold2, and inspired and corroborated by the XL-MS data, offer opportunities to deeply mine the structural proteome and interactome and reveal mechanisms underlying protein structure and function.

Structure of the heterotrimeric membrane protein complex FtsB-FtsL-FtsQ of the bacterial divisome

The synthesis of the cell-wall peptidoglycan during bacterial cell division is mediated by a multiprotein machine, called the divisome. The essential membrane protein complex of FtsB, FtsL and FtsQ (FtsBLQ) is at the heart of the divisome assembly cascade in Escherichia coli. This complex regulates the transglycosylation and transpeptidation activities of the FtsW-FtsI complex and PBP1b via coordination with FtsN, the trigger for the onset of constriction. Yet the underlying mechanism of FtsBLQ-mediated regulation is largely unknown. Here, we report the full-length structure of the heterotrimeric FtsBLQ complex, which reveals a V-shaped architecture in a tilted orientation. Such a conformation could be strengthened by the transmembrane and the coiled-coil domains of the FtsBL heterodimer, as well as an extended β-sheet of the C-terminal interaction site involving all three proteins. This trimeric structure may also facilitate interactions with other divisome proteins in an allosteric manner. These results lead us to propose a structure-based model that delineates the mechanism of the regulation of peptidoglycan synthases by the FtsBLQ complex.

Cross-linking mass spectrometry for mapping protein complex topologies in situ.

Cross-linking mass spectrometry has become an established technology to provide structural information on the topology and dynamics of protein complexes. Readily accessible workflows can provide detailed data on simplified systems, such as purified complexes. However, using this technology to study the structure of protein complexes in situ, such as in organelles, cells, and even tissues, is still a technological frontier. The complexity of these systems remains a considerable challenge, but there have been dramatic improvements in sample handling, data acquisition, and data processing. Here, we summarise these developments and describe the paths towards comprehensive and comparative structural interactomes by cross-linking mass spectrometry.

Determining protein structures in cellular lamella at pseudo-atomic resolution by GisSPA

Cryo-electron tomography is a major tool used to study the structure of protein complexes in situ. However, the throughput of tilt-series image data collection is still quite low. Here, we show that GisSPA, a GPU accelerated program, can translationally and rotationally localize the target protein complex in cellular lamellae, as prepared with a focused ion beam, using single cryo-electron microscopy images without tilt-series, and reconstruct the protein complex at near-atomic resolution. GisSPA allows high-throughput data collection without the acquisition of tilt-series images and reconstruction of the tomogram, which is essential for high-resolution reconstruction of asymmetric or low-symmetry protein complexes. We demonstrate the power of GisSPA with 3.4-Å and 3.9-Å resolutions of resolving phycobilisome and tetrameric photosystem II complex structures in cellular lamellae, respectively. In this work, we present GisSPA as a practical tool that facilitates high-resolution in situ protein structure determination.

SCARdock: A Web Server and Manually Curated Resource for Discovering Covalent Ligands.

Covalent drugs have been intentionally discarded historically due to the concern of off-target side effects, but the past decade has seen a fast resurgence of the discovery of covalent drugs. Compared to noncovalent ligands, covalent ligands might have better biochemical efficiency, lower patient burden, less dosing frequency, less drug resistance, and improved target specificity. Previously, we proposed the steric-clashes alleviating receptor (SCAR) strategy for screening and repurposing covalent inhibitors. To help the discovery of covalent ligands targeting protein targets, we have developed a web server dedicated to providing the SCARdock protocol to general users. Along with this server, we presented three high-quality data sets for the discovery of covalent ligands: a manually curated data set containing 954 high-quality complex structures of covalent ligands and proteins, a manually curated data set of 68 experimentally confirmed covalent warheads targeting 11 different residues, and a prefiltered, classified, and ready-to-use data set of 690,018 entries of purchasable virtual compounds containing these experimentally verified warheads. The SCARdock server and the accompanied data sets would be of great value to the discovery of covalent ligands and are available freely at http://www.liugroup.site/scardock/ or https://scardock.com.

Limited genetic variations of the Rh5-CyRPA-Ripr invasion complex in Plasmodium falciparum parasite population in selected malaria-endemic regions, Kenya.

The invasion of human erythrocytes by Plasmodium falciparum merozoites requires interaction between parasite ligands and host receptors. Interaction of PfRh5-CyRPA-Ripr protein complex with basigin, an erythrocyte surface receptor, via PfRh5 is essential for erythrocyte invasion. Antibodies raised against each antigen component of the complex have demonstrated erythrocyte invasion inhibition, making these proteins potential blood-stage vaccine candidates. Genetic polymorphisms present a significant challenge in developing efficacious vaccines, leading to variant-specific immune responses. This study investigated the genetic variations of the PfRh5 complex proteins in P. falciparum isolates from Lake Victoria islands, Western Kenya. Here, twenty-nine microscopically confirmed P. falciparum field samples collected from islands in Lake Victoria between July 2014 and July 2016 were genotyped by whole genome sequencing, and results compared to sequences mined from the GenBank database, from a study conducted in Kilifi, as well as other sequences from the MalariaGEN repository. We analyzed the frequency of polymorphisms in the PfRh5 protein complex proteins, PfRh5, PfCyRPA, PfRipr, and PfP113, and their location mapped on the 3D protein complex structure. We identified a total of 58 variants in the PfRh5 protein complex. PfRh5 protein was the most polymorphic with 30 SNPs, while PfCyRPA was relatively conserved with 3 SNPs. The minor allele frequency of the SNPs ranged between 1.9% and 21.2%. Ten high-frequency alleles (>5%) were observed in PfRh5 at codons 147, 148, 277, 410, and 429 and in PfRipr at codons 190, 255, 259, and 1003. A SNP was located in protein-protein interaction region C203Y and F292V of PfRh5 and PfCyRPA, respectively. Put together, this study revealed low polymorphisms in the PfRh5 invasion complex in the Lake Victoria parasite population. However, the two mutations identified on the protein interaction regions prompts for investigation on their impacts on parasite invasion process to support the consideration of PfRh5 components as potential malaria vaccine candidates.

Mutant structure of metabolic switch protein in complex with monomeric c-di-GMP reveals a potential mechanism of protein-mediated ligand dimerization

Bacterial second messengers c-di-GMP and (p)ppGpp have broad functional repertoires ranging from growth and cell cycle control to the regulation of biofilm formation and virulence. The recent identification of SmbA, an effector protein from Caulobacter crescentus that is jointly targeted by both signaling molecules, has opened up studies on how these global bacterial networks interact. C-di-GMP and (p)ppGpp compete for the same SmbA binding site, with a dimer of c-di-GMP inducing a conformational change that involves loop 7 of the protein that leads to downstream signaling. Here, we report a crystal structure of a partial loop 7 deletion mutant, SmbA∆loop in complex with c-di-GMP determined at 1.4 Å resolution. SmbA∆loop binds monomeric c-di-GMP indicating that loop 7 is required for c-di-GMP dimerization. Thus the complex probably represents the first step of consecutive c-di-GMP binding to form an intercalated dimer as has been observed in wild-type SmbA. Considering the prevalence of intercalated c-di-GMP molecules observed bound to proteins, the proposed mechanism may be generally applicable to protein-mediated c-di-GMP dimerization. Notably, in the crystal, SmbA∆loop forms a 2-fold symmetric dimer via isologous interactions with the two symmetric halves of c-di-GMP. Structural comparisons of SmbA∆loop with wild-type SmbA in complex with dimeric c-di-GMP or ppGpp support the idea that loop 7 is critical for SmbA function by interacting with downstream partners. Our results also underscore the flexibility of c-di-GMP, to allow binding to the symmetric SmbA∆loop dimer interface. It is envisaged that such isologous interactions of c-di-GMP could be observed in hitherto unrecognized targets.

Structural organization of RDGB (retinal degeneration B), a multi-domain lipid transfer protein: a molecular modelling and simulation based approach

Lipid transfer proteins (LTPs) that shuttle lipids at membrane contact sites (MCS) play an important role in maintaining cellular homeostasis. One such important LTP is the Retinal Degeneration B (RDGB) protein. RDGB is localized at the MCS formed between the endoplasmic reticulum (ER) and the apical plasma membrane (PM) in Drosophila photoreceptors where it transfers phosphatidylinositol (PI) during G-protein coupled phospholipase C signalling. Previously, the C-terminal domains of RDGB have been shown to be essential for its function and accurate localization. In this study, using in-silico integrative modelling we predict the structure of entire RDGB protein in complex with the ER membrane protein VAP. The structure of RDGB has then been used to decipher the structural features of the protein important for its orientation at the contact site. Using this structure, we identify two lysine residues in the C-terminal helix of the LNS2 domain important for interaction with the PM. Using molecular docking, we also identify an unstructured region USR1, immediately c-terminal to the PITP domain that is important for the interaction of RDGB with VAP. Overall the 10.06 nm length of the predicted RDGB-VAP complex spans the distance between the PM and ER and is consistent with the cytoplasmic gap between the ER and PM measured by transmission electron microscopy in photoreceptors. Overall our model explains the topology of the RDGB-VAP complex at this ER-PM contact site and paves the way for analysis of lipid transfer function in this setting. Communicated by Ramaswamy H. Sarma

Improved inter-protein contact prediction using dimensional hybrid residual networks and protein language models.

The knowledge of contacting residue pairs between interacting proteins is very useful for the structural characterization of protein-protein interactions (PPIs). However, accurately identifying the tens of contacting ones from hundreds of thousands of inter-protein residue pairs is extremely challenging, and performances of the state-of-the-art inter-protein contact prediction methods are still quite limited. In this study, we developed a deep learning method for inter-protein contact prediction, which is referred to as DRN-1D2D_Inter. Specifically, we employed pretrained protein language models to generate structural information-enriched input features to residual networks formed by dimensional hybrid residual blocks to perform inter-protein contact prediction. Extensively bechmarking DRN-1D2D_Inter on multiple datasets, including both heteromeric PPIs and homomeric PPIs, we show DRN-1D2D_Inter consistently and significantly outperformed two state-of-the-art inter-protein contact prediction methods, including GLINTER and DeepHomo, although both the latter two methods leveraged the native structures of interacting proteins in the prediction, and DRN-1D2D_Inter made the prediction purely from sequences. We further show that applying the predicted contacts as constraints for protein-protein docking can significantly improve its performance for protein complex structure prediction.

ComMap: a software to perform large-scale structure-based mapping for cross-linking mass spectrometry.

Chemical cross-linking combined with mass spectrometry (CXMS) is now a well-established method for profiling existing protein-protein interactions (PPIs) with partially known structures. It is expected to map the results of CXMS with existing structure databases to study the protein dynamic profile in the structure analysis. However, currently available structure-based analysis software suffers from the difficulty of achieving large-scale analysis. Besides, it is infeasible for structure analysis and data mining on a large scale, since of lacking global measurement of dynamic structure mapping results. ComMap (protein complex structure mapping) is a software designed to perform large-scale structure-based mapping by integrating CXMS data with existing structures. It allows complete the distance calculation of PPIs with existing structures in batch within minutes and provides scores for different PPI-structure pairs of testable hypothetical structural dynamism via a global view. Supplementary data are available at Bioinformatics online.

Gramm: A webserver for free and template-based protein docking.

Most biological functions depend on protein interactions. Structural modeling of proteins and protein assemblies is important for understanding the fundamental properties of biomolecular mechanisms and for our ability to manipulate them. Recently released artificial intelligence (AI) based programs, AlphaFold and RoseTTAFold, can accurately predict three-dimensional structures of proteins, even in the absence of the structures of homologous proteins. With the success of these AI-based approaches to protein structure prediction, the research community is anticipating methods/algorithms that will enable us to reliably predict the structures of protein complexes, including those in transient interactions, in the crowded environment inside a cell, accounting for often significant conformational flexibility of the interacting proteins. Prediction of the structure of protein complexes by docking methods is a well-established research field. The intermolecular energy landscape mapped by systematic docking approaches, like the FFT-based docking algorithms, can be used to refine docking predictions, to detect macro characteristics such as the binding funnel, and to generate ensembles of transient interactions for investigation of the protein interaction patterns. We present a user-friendly web interface to our popular GRAMM docking software for structural characterization of protein-protein interactions. Users can choose between the free and the template-based docking methodology, and select a number of advanced features, including mapping of the intermolecular energy landscape by distributions of docking poses, and the dynamic visualization of the docked models.

Applying AI methods to protein biology.

There is an increasing and productive overlap between the fields of artificial intelligence and biology. AI models have proven to be useful tools for making sense of complex data and generating hypotheses. This is exemplified by the progress made in predicting protein structures, and protein complex structures, from input amino acid sequences. In this talk, I'll discuss recent developments in the application of AI methods to protein biology. I'll also touch on what these approaches mean for high-throughput experimental work, where an explicit goal of the project may now be to use the data generated to train a model.

Understanding cell biology using cryo-electron microscopy

In order to fully understand biological processes, and how they fail in disease, it is vital to obtain structural information for the relevant biological machinery. Notably, it is becoming increasingly apparent that proteins, the key biological players in fundamental biology or disease mechanisms, often adopt multiple conformations or act in complexes with other proteins. These large and/or dynamic systems present a challenge to traditional methods of 3D structural determination as X-Ray crystallography or NMR. In recent years, single particle analysis (SPA) through cryo-EM has emerged as a mainstream structural biology technique, which can determine the 3D structure of proteins and protein complexes at near-atomic resolution. SPA allows determination of molecular details of purified and isolated proteins at near native conditions, albeit without the spatial and functional context of these proteins within the cell. Cryo Electron Tomography (Cryo-ET) fills this gap by visualizing proteins within their functional cellular environments. This allows for observation of their relationships and interactions with other cellular components and holds great promise for cell biology. In Cryo-EM, specimens are rapidly frozen (vitrified) so that their biologically relevant native states are preserved. This technique has transformed the field of structural biology, leading to new insight into numerous biological processes. Senior Scientist with Thermo Fisher Scientific, Natalia de Val, will discuss the latest advances in the SPA and cellular Cryo-ET workflows and will present the latest results in SPA and Tomography. These results will show how the field is changing our understanding of fundamental cell biology.

Breaking down the HAUS: Molecular insights into branching microtubule nucleation.

Accurate segregation of chromosomes during mitosis depends on the correct assembly of the mitotic spindle, a bipolar structure composed mainly of microtubules. The augmin complex, or homologous to augmin subunits (HAUS) complex, is an eight-subunit protein complex required for building robust mitotic spindles in metazoa. Augmin increases microtubule density within the spindle by recruiting the γ-tubulin ring complex (γ-TuRC) to pre-existing microtubules and nucleating branching microtubules in a manner that maintains spindle polarity. Here, we elucidate the molecular architecture of augmin by single particle cryo-electron microscopy (cryo-EM), computational methods, and crosslinking mass spectrometry (CLMS). Augmin's highly flexible structure contains a V-shaped head and a filamentous tail, with the head existing in either extended or contracted conformational states. Our work highlights how cryo-EM, complemented by computational advances and CLMS, can elucidate the structure of a challenging protein complex. We demonstrate how this approach has provided novel insights into the function of augmin in mediating microtubule branching nucleation.

Computational protein structure elucidation from covalent labeling mass spectrometry data.

Mass spectrometry (MS)-based methods such as covalent labeling MS are increasingly used to obtain information about protein structure. However, in contrast to other high-resolution structure determination methods, this information is not sufficient to deduce all atom coordinates and can only inform on certain elements of structure, such as solvent exposure of individual residues or location of protein-protein interfaces. However, easy and reliable translation of this information into accurate structural protein and protein complex models remains particularly challenging, due to a lack of standardized and automated methods. Computational methods are needed to predict high-resolution protein structures from the mass spectrometry (MS) data. My group develops algorithms within the Rosetta software package that use mass spectrometry data to guide protein structure prediction. We have successfully implemented the use of MS covalent labeling data into the Rosetta software to predict protein and protein complex structure guided by the experimental data. We developed algorithms to predict protein structure from hydroxyl radical and diethylpyrocarbonate (DEPC) labeling data. To evaluate the effectiveness of our covalent-labeling guided structure prediction algorithms, we benchmarked their performance using proteins of known structure for which experimental covalent labeling data exists. For all benchmark proteins, better quality models were built when MS data was used in the scoring. For modeling with hydroxyl radical labeling data, we measured exposure of labeled residues, while in the case of DEPC labeling we further assessed the local hydrophobic microenvironment of serine, threonine, and tyrosine residues. We are now extending this work to use MS data in deep learning-based predictions of protein structure.

Parameterization of a single H-bond in Orange Carotenoid Protein by atomic mutation reveals principles of evolutionary design of complex chemical photosystems.

Introduction: Dissecting the intricate networks of covalent and non-covalent interactions that stabilize complex protein structures is notoriously difficult and requires subtle atomic-level exchanges to precisely affect local chemical functionality. The function of the Orange Carotenoid Protein (OCP), a light-driven photoswitch involved in cyanobacterial photoprotection, depends strongly on two H-bonds between the 4-ketolated xanthophyll cofactor and two highly conserved residues in the C-terminal domain (Trp288 and Tyr201). Method: By orthogonal translation, we replaced Trp288 in Synechocystis OCP with 3-benzothienyl-L-alanine (BTA), thereby exchanging the imino nitrogen for a sulphur atom. Results: Although the high-resolution (1.8Å) crystal structure of the fully photoactive OCP-W288_BTA protein showed perfect isomorphism to the native structure, the spectroscopic and kinetic properties changed distinctly. We accurately parameterized the effects of the absence of a single H-bond on the spectroscopic and thermodynamic properties of OCP photoconversion and reveal general principles underlying the design of photoreceptors by natural evolution. Discussion: Such "molecular surgery" is superior over trial-and-error methods in hypothesis-driven research of complex chemical systems.

Perspective on the potential of tandem-ion mobility /mass spectrometry methods for structural proteomics applications.

Cellular processes are usually carried out collectively by the entirety of all proteins present in a biological cell, i.e. the proteome. Mass spectrometry-based methods have proven particularly successful in identifying and quantifying the constituent proteins of proteomes, including different molecular forms of a protein. Nevertheless, protein sequences alone do not reveal the function or dysfunction of the identified proteins. A straightforward way to assign function or dysfunction to proteins is characterization of their structures and dynamics. However, a method capable to characterize detailed structures of proteins and protein complexes in a large-scale, systematic manner within the context of cellular processes does not yet exist. Here, we discuss the potential of tandem-ion mobility / mass spectrometry (tandem-IM/MS) methods to provide such ability. We highlight the capability of these methods using two case studies on the protein systems ubiquitin and avidin using the tandem-TIMS/MS technology developed in our laboratory and discuss these results in the context of other developments in the broader field of tandem-IM/MS.

Critical Assessment of Methods for Predicting the 3D Structure of Proteins and Protein Complexes.

Advances in a scientific discipline are often measured by small, incremental steps. In this review, we report on two intertwined disciplines in the protein structure prediction field, modeling of single chains and modeling of complexes, that have over decades emulated this pattern, as monitored by the community-wide blind prediction experiments CASP and CAPRI. However, over the past few years, dramatic advances were observed for the accurate prediction of single protein chains, driven by a surge of deep learning methodologies entering the prediction field. We review the mainscientific developments that enabled these recent breakthroughs and feature the important role of blind prediction experiments in building up and nurturing the structure prediction field. We discuss how the new wave of artificial intelligence-based methods is impacting the fields of computational and experimental structural biology and highlight areas in which deep learning methods are likely to lead to future developments, provided that major challenges are overcome.

MetaScore: A Novel Machine-Learning-Based Approach to Improve Traditional Scoring Functions for Scoring Protein-Protein Docking Conformations.

Protein-protein interactions play a ubiquitous role in biological function. Knowledge of the three-dimensional (3D) structures of the complexes they form is essential for understanding the structural basis of those interactions and how they orchestrate key cellular processes. Computational docking has become an indispensable alternative to the expensive and time-consuming experimental approaches for determining the 3D structures of protein complexes. Despite recent progress, identifying near-native models from a large set of conformations sampled by docking-the so-called scoring problem-still has considerable room for improvement. We present MetaScore, a new machine-learning-based approach to improve the scoring of docked conformations. MetaScore utilizes a random forest (RF) classifier trained to distinguish near-native from non-native conformations using their protein-protein interfacial features. The features include physicochemical properties, energy terms, interaction-propensity-based features, geometric properties, interface topology features, evolutionary conservation, and also scores produced by traditional scoring functions (SFs). MetaScore scores docked conformations by simply averaging the score produced by the RF classifier with that produced by any traditional SF. We demonstrate that (i) MetaScore consistently outperforms each of the nine traditional SFs included in this work in terms of success rate and hit rate evaluated over conformations ranked among the top 10; (ii) an ensemble method, MetaScore-Ensemble, that combines 10 variants of MetaScore obtained by combining the RF score with each of the traditional SFs outperforms each of the MetaScore variants. We conclude that the performance of traditional SFs can be improved upon by using machine learning to judiciously leverage protein-protein interfacial features and by using ensemble methods to combine multiple scoring functions.

PIQLE: protein-protein interface quality estimation by deep graph learning of multimeric interaction geometries.

Accurate modeling of protein-protein interaction interface is essential for high-quality protein complex structure prediction. Existing approaches for estimating the quality of a predicted protein complex structural model utilize only the physicochemical properties or energetic contributions of the interacting atoms, ignoring evolutionarily information or inter-atomic multimeric geometries, including interaction distance and orientations. Here, we present PIQLE, a deep graph learning method for protein-protein interface quality estimation. PIQLE leverages multimeric interaction geometries and evolutionarily information along with sequence- and structure-derived features to estimate the quality of individual interactions between the interfacial residues using a multi-head graph attention network and then probabilistically combines the estimated quality for scoring the overall interface. Experimental results show that PIQLE consistently outperforms existing state-of-the-art methods including DProQA, TRScore, GNN-DOVE and DOVE on multiple independent test datasets across a wide range of evaluation metrics. Our ablation study and comparison with the self-assessment module of AlphaFold-Multimer repurposed for protein complex scoring reveal that the performance gains are connected to the effectiveness of the multi-head graph attention network in leveraging multimeric interaction geometries and evolutionary information along with other sequence- and structure-derived features adopted in PIQLE. An open-source software implementation of PIQLE is freely available at https://github.com/Bhattacharya-Lab/PIQLE. Supplementary data are available at Bioinformatics Advances online.