Large-scale determination of previously unsolved protein structures using evolutionary information

The Rise of Bacterial G-Quadruplexes in Current Antimicrobial Discovery

Antimicrobial resistance (AMR) is a silent critical issue that poses several challenges to health systems. While the discovery of novel antibiotics is currently stalled and prevalently focused on chemical variations of the scaffolds of available drugs, novel targets and innovative strategies are urgently needed to face this global threat. In this context, bacterial G-quadruplexes (G4s) are emerging as timely and profitable targets for the design and development of antimicrobial agents. Indeed, they are expressed in regulatory regions of bacterial genomes, and their modulation has been observed to provide antimicrobial effects with translational perspectives in the context of AMR. In this work, we review the current knowledge of bacterial G4s as well as their modulation by small molecules, including tools and techniques suitable for these investigations. Finally, we critically analyze the needs and future directions in the field, with a focus on the development of small molecules as bacterial G4s modulators endowed with remarkable drug-likeness.

Combined Computational-Biochemical Approach Offers an Accelerated Path to Membrane Protein Solubilization

Membrane proteins are difficult to isolate and purify due to their dependence on the surrounding lipid membrane for structural stability. Detergents are often used to solubilize these proteins, with this approach requiring a careful balance between protein solubilization and denaturation. Determining which detergent is most appropriate for a given protein has largely been done empirically through screening, which requires large amounts of membrane protein and associated resources. Here, we describe an alternative to conventional detergent screening using a computational modeling approach to identify the most likely candidate detergents for solubilizing a protein of interest. We demonstrate our approach using ghrelin O-acyltransferase (GOAT), a member of the membrane-bound O-acyltransferase family of integral membrane enzymes that has not been solubilized or purified in active form. A computationally derived GOAT structural model provides the only structural information required for this approach. Using computational analysis of detergent ability to penetrate phospholipid bilayers and stabilize the GOAT structure, a panel of common detergents were rank-ordered for their proposed ability to solubilize GOAT. The simulations were performed at all-atom resolution for a combined simulation time of 24 μs. Independently, we biologically screened these detergents for their solubilization of fluorescently tagged GOAT constructs. We found computational prediction of protein structural stabilization was the better predictor of detergent solubilization ability, but neither approach was effective for predicting detergents that would support GOAT enzymatic function. The current rapid expansion of membrane protein computational models lacking experimental structural information and our computational detergent screening approach can greatly improve the efficiency of membrane protein detergent solubilization, supporting downstream functional and structural studies.

Learning sequence, structure, and function representations of proteins with language models

The sequence-structure-function relationships that ultimately generate the diversity of extant observed proteins is complex, as proteins bridge the gap between multiple informational and physical scales involved in nearly all cellular processes. One limitation of existing protein annotation databases such as UniProt is that less than 1% of proteins have experimentally verified functions, and computational methods are needed to fill in the missing information. Here, we demonstrate that a multi-aspect framework based on protein language models can learn sequence-structure-function representations of amino acid sequences, and can provide the foundation for sensitive sequence-structure-function aware protein sequence search and annotation. Based on this model, we introduce a multi-aspect information retrieval system for proteins, Protein-Vec, covering sequence, structure, and function aspects, that enables computational protein annotation and function prediction at tree-of-life scales.

Co-evolution at protein-protein interfaces guides inference of stoichiometry of oligomeric protein complexes by de novo structure prediction

The quaternary structure with specific stoichiometry is pivotal to the specific function of protein complexes. However, determining the structure of many protein complexes experimentally remains a major bottleneck. Structural bioinformatics approaches, such as the deep learning algorithm Alphafold2-multimer (AF2-multimer), leverage the co-evolution of amino acids and sequence-structure relationships for accurate de novo structure and contact prediction. Pseudo-likelihood maximization direct coupling analysis (plmDCA) has been used to detect co-evolving residue pairs by statistical modeling. Here, we provide evidence that combining both methods can be used for de novo prediction of the quaternary structure and stoichiometry of a protein complex. We achieve this by augmenting the existing AF2-multimer confidence metrics with an interpretable score to identify the complex with an optimal fraction of native contacts of co-evolving residue pairs at intermolecular interfaces. We use this strategy to predict the quaternary structure and non-trivial stoichiometries of Bacillus subtilis spore germination protein complexes with unknown structures. Co-evolution at intermolecular interfaces may therefore synergize with AI-based de novo quaternary structure prediction of structurally uncharacterized bacterial protein complexes.

Development of anti-PD-L1 antibody based on structure prediction of AlphaFold2

Accurate structural information plays a crucial role in comprehending biological processes and designing drugs. Indeed, the remarkable precision of the AlphaFold2 has facilitated significant advancements in predicting molecular structures, encompassing antibodies and antigens. This breakthrough has paved the way for rational drug design, ushering in new possibilities in the field of pharmaceutical development. Within this study, performing analysis and humanization guided by the structures predicted by AlphaFold2. Notably, the resulting humanized antibody, h3D5-hIgG1, demonstrated exceptional binding affinity to the PD-L1 protein. The KD value of parental antibody 3D5-hIgG1 was increased by nearly 7 times after humanization. Both h3D5-hIgG1 and 3D5-hIgG1 bound to cells expressing human PD-L1 with EC50 values of 5.13 and 9.92nM, respectively. Humanization resulted in a twofold increase in the binding capacity of the antibody, with h3D5-hIgG1 exhibiting superior performance compared to the parental antibody 3D5-hIgG1. Furthermore, h3D5-hIgG1 promoted cytokine secretion of T cells, and significantly suppressed MC38-hPD-L1 tumor growth. This study highlights the potential for artificial intelligence-assisted drug development, which is poised to become a prominent trend in the future.

Enantioselectivity in the enzymatic dehydration of malate and tartrate: Mirror image specificities of structurally similar dehydratases

Malate (2-hydroxysuccinic acid) and tartrate (2,3-dihydroxysuccinic acid) are chiral substrates; the former existing in two enantiomeric forms (R and S) while the latter exists as three stereoisomers (R,R; S,S; and R,S). Dehydration by stereospecific hydrogen abstraction and antielimination of the hydroxyl group yield the achiral products fumarate and oxaloacetate, respectively. Class-I fumarate hydratase (FH) and L-tartrate dehydratase (L-TTD) are two highly conserved enzymes belonging to the iron-sulfur cluster hydrolyase family of enzymes that catalyze reactions on specific stereoisomers of malate and tartrate. FH from Methanocaldococcus jannaschii accepts only (S)-malate and (S,S)-tartrate as substrates while the structurally similar L-TTD from Escherichia coli accepts only (R)-malate and (R,R)-tartrate as substrates. Phylogenetic analysis reveals a common evolutionary origin of L-TTDs and two-subunit archaeal FHs suggesting a divergence during evolution that may have led to the switch in substrate stereospecificity preference. Due to the high conservation of their sequences, a molecular basis for switch in stereospecificity is not evident from analysis of crystal structures of FH and predicted structure of L-TTD. The switch in enantiomer preference may be rationalized by invoking conformational plasticity of the amino acids interacting with the substrate, together with substrate reorientation and conformer selection about the C2C3 bond of the dicarboxylic acid substrates. Although classical models of enzyme-substrate binding are insufficient to explain such a phenomenon, the enantiomer superposition model suggests that a minor reorientation in the active site residues could lead to the switch in substrate stereospecificity.

Insights into DNA cleavage by MutL homologs from analysis of conserved motifs in eukaryotic Mlh1

MutL family proteins contain an N-terminal ATPase domain (NTD), an unstructured interdomain linker, and a C-terminal domain (CTD), which mediates constitutive dimerization between subunits and often contains an endonuclease active site. Most MutL homologs direct strand-specific DNA mismatch repair by cleaving the error-containing daughter DNA strand. The strand cleavage reaction is poorly understood; however, the structure of the endonuclease active site is consistent with a two- or three-metal ion cleavage mechanism. A motif required for this endonuclease activity is present in the unstructured linker of Mlh1 and is conserved in all eukaryotic Mlh1 proteins, except those from metamonads, which also lack the almost absolutely conserved Mlh1 C-terminal phenylalanine-glutamate-arginine-cysteine (FERC) sequence. We hypothesize that the cysteine in the FERC sequence is autoinhibitory, as it sequesters the active site. We further hypothesize that the evolutionary co-occurrence of the conserved linker motif with the FERC sequence indicates a functional interaction, possibly by linker motif-mediated displacement of the inhibitory cysteine. This role is consistent with available data for interactions between the linker motif with DNA and the CTDs in the vicinity of the active site.

Blind assessment of monomeric AlphaFold2 protein structure models with experimental NMR data

Recent advances in molecular modeling of protein structures are changing the field of structural biology. AlphaFold-2 (AF2), an AI system developed by DeepMind, Inc., utilizes attention-based deep learning to predict models of protein structures with high accuracy relative to structures determined by X-ray crystallography and cryo-electron microscopy (cryoEM). Comparing AF2 models to structures determined using solution NMR data, both high similarities and distinct differences have been observed. Since AF2 was trained on X-ray crystal and cryoEM structures, we assessed how accurately AF2 can model small, monomeric, solution protein NMR structures which (i) were not used in the AF2 training data set, and (ii) did not have homologous structures in the Protein Data Bank at the time of AF2 training. We identified nine open-source protein NMR data sets for such "blind" targets, including chemical shift, raw NMR FID data, NOESY peak lists, and (for 1 case) 15N-1H residual dipolar coupling data. For these nine small (70-108 residues) monomeric proteins, we generated AF2 prediction models and assessed how well these models fit to these experimental NMR data, using several well-established NMR structure validation tools. In most of these cases, the AF2 models fit the NMR data nearly as well, or sometimes better than, the corresponding NMR structure models previously deposited in the Protein Data Bank. These results provide benchmark NMR data for assessing new NMR data analysis and protein structure prediction methods. They also document the potential for using AF2 as a guiding tool in protein NMR data analysis, and more generally for hypothesis generation in structural biology research.

Evolution of the Immunoglobulin Isotypes-Variations of Biophysical Properties among Animal Classes

The adaptive immune system arose around 500 million years ago in jawed fish, and, since then, it has mediated the immune defense against pathogens in all vertebrates. Antibodies play a central role in the immune reaction, recognizing and attacking external invaders. During the evolutionary process, several immunoglobulin isotypes emerged, each having a characteristic structural organization and dedicated function. In this work, we investigate the evolution of the immunoglobulin isotypes, in order to highlight the relevant features that were preserved over time and the parts that, instead, mutated. The residues that are coupled in the evolution process are often involved in intra- or interdomain interactions, meaning that they are fundamental to maintaining the immunoglobulin fold and to ensuring interactions with other domains. The explosive growth of available sequences allows us to point out the evolutionary conserved residues and compare the biophysical properties among different animal classes and isotypes. Our study offers a general overview of the evolution of immunoglobulin isotypes and advances the knowledge of their characteristic biophysical properties, as a first step in guiding protein design from evolution.

SARS-CoV2 billion-compound docking

This dataset contains ligand conformations and docking scores for 1.4 billion molecules docked against 6 structural targets from SARS-CoV2, representing 5 unique proteins: MPro, NSP15, PLPro, RDRP, and the Spike protein. Docking was carried out using the AutoDock-GPU platform on the Summit supercomputer and Google Cloud. The docking procedure employed the Solis Wets search method to generate 20 independent ligand binding poses per compound. Each compound geometry was scored using the AutoDock free energy estimate, and rescored using RFScore v3 and DUD-E machine-learned rescoring models. Input protein structures are included, suitable for use by AutoDock-GPU and other docking programs. As the result of an exceptionally large docking campaign, this dataset represents a valuable resource for discovering trends across small molecule and protein binding sites, training AI models, and comparing to inhibitor compounds targeting SARS-CoV-2. The work also gives an example of how to organize and process data from ultra-large docking screens.

Structure and mechanism of a tripartite ATP-independent periplasmic TRAP transporter

In bacteria and archaea, tripartite ATP-independent periplasmic (TRAP) transporters uptake essential nutrients. TRAP transporters receive their substrates via a secreted soluble substrate-binding protein. How a sodium ion-driven secondary active transporter is strictly coupled to a substrate-binding protein is poorly understood. Here we report the cryo-EM structure of the sialic acid TRAP transporter SiaQM from Photobacterium profundum at 2.97 Å resolution. SiaM comprises a "transport" domain and a "scaffold" domain, with the transport domain consisting of helical hairpins as seen in the sodium ion-coupled elevator transporter VcINDY. The SiaQ protein forms intimate contacts with SiaM to extend the size of the scaffold domain, suggesting that TRAP transporters may operate as monomers, rather than the typically observed oligomers for elevator-type transporters. We identify the Na⁺ and sialic acid binding sites in SiaM and demonstrate a strict dependence on the substrate-binding protein SiaP for uptake. We report the SiaP crystal structure that, together with docking studies, suggest the molecular basis for how sialic acid is delivered to the SiaQM transporter complex. We thus propose a model for substrate transport by TRAP proteins, which we describe herein as an 'elevator-with-an-operator' mechanism.

Sterol-regulated transmembrane protein TMEM86a couples LXR signaling to regulation of lysoplasmalogens in macrophages

Lysoplasmalogens are a class of vinyl ether bioactive lipids that have a central role in plasmalogen metabolism and membrane fluidity. The liver X receptor (LXR) transcription factors are important determinants of cellular lipid homeostasis owing to their ability to regulate cholesterol and fatty acid metabolism. However, their role in governing the composition of lipid species such as lysoplasmalogens in cellular membranes is less well studied. Here, we mapped the lipidome of bone marrow-derived macrophages (BMDMs) following LXR activation. We found a marked reduction in the levels of lysoplasmalogen species in the absence of changes in the levels of plasmalogens themselves. Transcriptional profiling of LXR-activated macrophages identified the gene encoding transmembrane protein 86a (TMEM86a), an integral endoplasmic reticulum protein, as a previously uncharacterized sterol-regulated gene. We demonstrate that TMEM86a is a direct transcriptional target of LXR in macrophages and microglia and that it is highly expressed in TREM2+/lipid-associated macrophages in human atherosclerotic plaques, where its expression positively correlates with other LXR-regulated genes. We further show that both murine and human TMEM86a display active lysoplasmalogenase activity that can be abrogated by inactivating mutations in the predicted catalytic site. Consequently, we demonstrate that overexpression of Tmem86a in BMDM markedly reduces lysoplasmalogen abundance and membrane fluidity, while reciprocally, silencing of Tmem86a increases basal lysoplasmalogen levels and abrogates the LXR-dependent reduction of this lipid species. Collectively, our findings implicate TMEM86a as a sterol-regulated lysoplasmalogenase in macrophages that contributes to sterol-dependent membrane remodeling.

Machine learning in computational modelling of membrane protein sequences and structures: From methodologies to applications

Membrane proteins mediate a wide spectrum of biological processes, such as signal transduction and cell communication. Due to the arduous and costly nature inherent to the experimental process, membrane proteins have long been devoid of well-resolved atomic-level tertiary structures and, consequently, the understanding of their functional roles underlying a multitude of life activities has been hampered. Currently, computational tools dedicated to furthering the structure-function understanding are primarily focused on utilizing intelligent algorithms to address a variety of site-wise prediction problems (e.g., topology and interaction sites), but are scattered across different computing sources. Moreover, the recent advent of deep learning techniques has immensely expedited the development of computational tools for membrane protein-related prediction problems. Given the growing number of applications optimized particularly by manifold deep neural networks, we herein provide a review on the current status of computational strategies mainly in membrane protein type classification, topology identification, interaction site detection, and pathogenic effect prediction. Meanwhile, we provide an overview of how the entire prediction process proceeds, including database collection, data pre-processing, feature extraction, and method selection. This review is expected to be useful for developing more extendable computational tools specific to membrane proteins.

Blind Assessment of Monomeric AlphaFold2 Protein Structure Models with Experimental NMR Data

Recent advances in molecular modeling of protein structures are changing the field of structural biology. AlphaFold-2 (AF2), an AI system developed by DeepMind, Inc., utilizes attention-based deep learning to predict models of protein structures with high accuracy relative to structures determined by X-ray crystallography and cryo-electron microscopy (cryoEM). Comparing AF2 models to structures determined using solution NMR data, both high similarities and distinct differences have been observed. Since AF2 was trained on X-ray crystal and cryoEM structures, we assessed how accurately AF2 can model small, monomeric, solution protein NMR structures which (i) were not used in the AF2 training data set, and (ii) did not have homologous structures in the Protein Data Bank at the time of AF2 training. We identified nine open source protein NMR data sets for such "blind" targets, including chemical shift, raw NMR FID data, NOESY peak lists, and (for 1 case) 15 N- 1 H residual dipolar coupling data. For these nine small (70 - 108 residues) monomeric proteins, we generated AF2 prediction models and assessed how well these models fit to these experimental NMR data, using several well-established NMR structure validation tools. In most of these cases, the AF2 models fit the NMR data nearly as well, or sometimes better than, the corresponding NMR structure models previously deposited in the Protein Data Bank. These results provide benchmark NMR data for assessing new NMR data analysis and protein structure prediction methods. They also document the potential for using AF2 as a guiding tool in protein NMR data analysis, and more generally for hypothesis generation in structural biology research.

Highlights

AF2 models assessed against NMR data for 9 monomeric proteins not used in training.AF2 models fit NMR data almost as well as the experimentally-determined structures. RPF-DP, PSVS , and PDBStat software provide structure quality and RDC assessment. RPF-DP analysis using AF2 models suggests multiple conformational states.

Cumulative Millisecond-Long Sampling for a Comprehensive Energetic Evaluation of Aqueous Ionic Liquid Effects on Amino Acid Interactions

The interactions of amino acid side-chains confer diverse energetic contributions and physical properties to a protein's stability and function. Various computational tools estimate the effect of changing a given amino acid on the protein's stability based on parametrized (free) energy functions. When parametrized for the prediction of protein stability in water, such energy functions can lead to suboptimal results for other solvents, such as ionic liquids (IL), aqueous ionic liquids (aIL), or salt solutions. However, to our knowledge, no comprehensive data are available describing the energetic effects of aIL on intramolecular protein interactions. Here, we present the most comprehensive set of potential of mean force (PMF) profiles of pairwise protein-residue interactions to date, covering 50 relevant interactions in water, the two biotechnologically relevant aIL [BMIM/Cl] and [BMIM/TfO], and [Na/Cl]. These results are based on a cumulated simulation time of >1 ms. aIL and salt ions can weaken, but also strengthen, specific residue interactions by more than 3 kcal mol^-1, depending on the residue pair, residue-residue configuration, participating ions, and concentration, necessitating considering such interactions specifically. These changes originate from a complex interplay of competitive or cooperative noncovalent ion-residue interactions, changes in solvent structural dynamics, or unspecific charge screening effects and occur at the contact distance but also at larger, solvent-separated distances. This data provide explanations at the atomistic and energetic levels for complex IL effects on protein stability and should help improve the prediction accuracies of computational tools that estimate protein stability based on (free) energy functions.

Contingency and Entrenchment of Drug-Resistance Mutations in HIV Viral Proteins

The ability of HIV-1 to rapidly mutate leads to antiretroviral therapy (ART) failure among infected patients. Drug-resistance mutations (DRMs), which cause a fitness penalty to intrinsic viral fitness, are compensated by accessory mutations with favorable epistatic interactions which cause an evolutionary trapping effect, but the kinetics of this overall process has not been well characterized. Here, using a Potts Hamiltonian model describing epistasis combined with kinetic Monte Carlo simulations of evolutionary trajectories, we explore how epistasis modulates the evolutionary dynamics of HIV DRMs. We show how the occurrence of a drug-resistance mutation is contingent on favorable epistatic interactions with many other residues of the sequence background and that subsequent mutations entrench DRMs. We measure the time-autocorrelation of fluctuations in the likelihood of DRMs due to epistatic coupling with the sequence background, which reveals the presence of two evolutionary processes controlling DRM kinetics with two distinct time scales. Further analysis of waiting times for the evolutionary trapping effect to reverse reveals that the sequences which entrench (trap) a DRM are responsible for the slower time scale. We also quantify the overall strength of epistatic effects on the evolutionary kinetics for different mutations and show these are much larger for DRM positions than polymorphic positions, and we also show that trapping of a DRM is often caused by the collective effect of many accessory mutations, rather than a few strongly coupled ones, suggesting the importance of multiresidue sequence variations in HIV evolution. The analysis presented here provides a framework to explore the kinetic pathways through which viral proteins like HIV evolve under drug-selection pressure.

Aeromonas species isolated from aquatic organisms, insects, chicken, and humans in India show similar antimicrobial resistance profiles

Aeromonas species are Gram-negative bacteria that infect various living organisms and are ubiquitously found in different aquatic environments. In this study, we used whole genome sequencing (WGS) to identify and compare the antimicrobial resistance (AMR) genes, integrons, transposases and plasmids found in Aeromonas hydrophila, Aeromonas caviae and Aeromonas veronii isolated from Indian major carp (Catla catla), Indian carp (Labeo rohita), catfish (Clarias batrachus) and Nile tilapia (Oreochromis niloticus) sampled in India. To gain a wider comparison, we included 11 whole genome sequences of Aeromonas spp. from different host species in India deposited in the National Center for Biotechnology Information (NCBI). Our findings show that all 15 Aeromonas sequences examined had multiple AMR genes of which the Ambler classes B, C and D β-lactamase genes were the most dominant. The high similarity of AMR genes in the Aeromonas sequences obtained from different host species point to interspecies transmission of AMR genes. Our findings also show that all Aeromonas sequences examined encoded several multidrug efflux-pump proteins. As for genes linked to mobile genetic elements (MBE), only the class I integrase was detected from two fish isolates, while all transposases detected belonged to the insertion sequence (IS) family. Only seven of the 15 Aeromonas sequences examined had plasmids and none of the plasmids encoded AMR genes. In summary, our findings show that Aeromonas spp. isolated from different host species in India carry multiple AMR genes. Thus, we advocate that the control of AMR caused by Aeromonas spp. in India should be based on a One Health approach.

C4-dicarboxylate metabolons: interaction of C4-dicarboxylate transporters of Escherichia coli with cytosolic enzymes

Metabolons represent the structural organization of proteins for metabolic or regulatory pathways. Here the interaction of fumarase FumB, aspartase AspA, and L-tartrate dehydratase TtdAB with the C4-dicarboxylate (C4-DC) transporters DcuA, DcuB, DcuC, and the L-tartrate transporter TtdT of Escherichia coli was tested by a bacterial two-hybrid (BACTH) assay in situ, or by co-chromatography using mSPINE (membrane Streptavidin protein interaction experiment). From the general C4-DC transporters, DcuB interacted with FumB and AspA, DcuA with AspA, whereas DcuC interacted with neither FumB nor AspA. Moreover, TtdT did not interact with TtdAB. The fumB-dcuB, the dcuA-aspA, and the ttdAB-ttdT genes encoding the respective proteins co-localize on the genome and each pair of genes forms co-transcripts whereas the dcuC gene lies alone. The data suggest the formation of DcuB/FumB and DcuB/AspA metabolons for the uptake of L-malate, or L-aspartate, and their conversion to fumarate for fumarate respiration and excretion of the product succinate. The DcuA/AspA metabolon catalyzes uptake and conversion of L-Asp to fumarate coupled to succinate excretion. The DcuA/AspA metabolon provides ammonia at the same time for nitrogen assimilation (ammonia shuttle). On the other hand, TtdT and TtdAB are not organized in a metabolon. Reasons for the formation (DcuA/AspA, DcuB/FumB, DcuB/AspA) or non-formation (DcuC, TtdT and TtdAB) of metabolons are discussed based on their metabolic roles.

Modeling of protein conformational changes with Rosetta guided by limited experimental data

Conformational changes are an essential component of functional cycles of many proteins, but their characterization often requires an integrative structural biology approach. Here, we introduce and benchmark ConfChangeMover (CCM), a new method built into the widely used macromolecular modeling suite Rosetta that is tailored to model conformational changes in proteins using sparse experimental data. CCM can rotate and translate secondary structural elements and modify their backbone dihedral angles in regions of interest. We benchmarked CCM on soluble and membrane proteins with simulated Cα-Cα distance restraints and sparse experimental double electron-electron resonance (DEER) restraints, respectively. In both benchmarks, CCM outperformed state-of-the-art Rosetta methods, showing that it can model a diverse array of conformational changes. In addition, the Rosetta framework allows a wide variety of experimental data to be integrated with CCM, thus extending its capability beyond DEER restraints. This method will contribute to the biophysical characterization of protein dynamics.

Structural insights into the role of the WW2 domain on tandem WW/PPxY-motif interactions of oxidoreductase WWOX

Class I WW domains are present in many proteins of various functions and mediate protein interactions by binding to short linear PPxY motifs. Tandem WW domains often bind peptides with multiple PPxY motifs, but the interplay of WW–peptide interactions is not always intuitive. The WW domain–containing oxidoreductase (WWOX) harbors two WW domains: an unstable WW1 capable of PPxY binding and stable WW2 that cannot bind PPxY. The WW2 domain has been suggested to act as a WW1 domain chaperone, but the underlying mechanism of its chaperone activity remains to be revealed. Here, we combined NMR, isothermal calorimetry, and structural modeling to elucidate the roles of both WW domains in WWOX binding to its PPxY-containing substrate ErbB4. Using NMR, we identified an interaction surface between these two domains that supports a WWOX conformation compatible with peptide substrate binding. Isothermal calorimetry and NMR measurements also indicated that while binding affinity to a single PPxY motif is marginally increased in the presence of WW2, affinity to a dual-motif peptide increases 10-fold. Furthermore, we found WW2 can directly bind double-motif peptides using its canonical binding site. Finally, differential binding of peptides in mutagenesis experiments was consistent with a parallel N- to C-terminal PPxY tandem motif orientation in binding to the WW1–WW2 tandem domain, validating structural models of the interaction. Taken together, our results reveal the complex nature of tandem WW-domain organization and substrate binding, highlighting the contribution of WWOX WW2 to both protein stability and target binding.

AlphaFold Models of Small Proteins Rival the Accuracy of Solution NMR Structures

Recent advances in molecular modeling using deep learning have the potential to revolutionize the field of structural biology. In particular, AlphaFold has been observed to provide models of protein structures with accuracies rivaling medium-resolution X-ray crystal structures, and with excellent atomic coordinate matches to experimental protein NMR and cryo-electron microscopy structures. Here we assess the hypothesis that AlphaFold models of small, relatively rigid proteins have accuracies (based on comparison against experimental data) similar to experimental solution NMR structures. We selected six representative small proteins with structures determined by both NMR and X-ray crystallography, and modeled each of them using AlphaFold. Using several structure validation tools integrated under the Protein Structure Validation Software suite (PSVS), we then assessed how well these models fit to experimental NMR data, including NOESY peak lists (RPF-DP scores), comparisons between predicted rigidity and chemical shift data (ANSURR scores), and ¹⁵N-¹H residual dipolar coupling data (RDC Q factors) analyzed by software tools integrated in the PSVS suite. Remarkably, the fits to NMR data for the protein structure models predicted with AlphaFold are generally similar, or better, than for the corresponding experimental NMR or X-ray crystal structures. Similar conclusions were reached in comparing AlphaFold2 predictions and NMR structures for three targets from the Critical Assessment of Protein Structure Prediction (CASP). These results contradict the widely held misperception that AlphaFold cannot accurately model solution NMR structures. They also document the value of PSVS for model vs. data assessment of protein NMR structures, and the potential for using AlphaFold models for guiding analysis of experimental NMR data and more generally in structural biology.

Mycobacterium tuberculosis encodes a YhhN family membrane protein with lysoplasmalogenase activity that protects against toxic host lysolipids

The pathogen Mycobacterium tuberculosis (M.tb) resides in human macrophages, wherein it exploits host lipids for survival. However, little is known about the interaction between M.tb and macrophage plasmalogens, a subclass of glycerophospholipids with a vinyl ether bond at the sn-1 position of the glycerol backbone. Lysoplasmalogens, produced from plasmalogens by hydrolysis at the sn-2 carbon by phospholipase A2, are potentially toxic but can be broken down by host lysoplasmalogenase, an integral membrane protein of the YhhN family that hydrolyzes the vinyl ether bond to release a fatty aldehyde and glycerophospho-ethanolamine or glycerophospho-choline. Curiously, M.tb encodes its own YhhN protein (MtbYhhN), despite having no endogenous plasmalogens. To understand the purpose of this protein, the gene for MtbYhhN (Rv1401) was cloned and expressed in Mycobacterium smegmatis (M.smeg). We found the partially purified protein exhibited abundant lysoplasmalogenase activity specific for lysoplasmenylethanolamine or lysoplasmenylcholine (pLPC) (Vmax∼15.5 μmol/min/mg; Km∼83 μM). Based on cell density, we determined that lysoplasmenylethanolamine, pLPC, lysophosphatidylcholine, and lysophosphatidylethanolamine were not toxic to M.smeg cells, but pLPC and LPC were highly toxic to M.smeg spheroplasts, which are cell wall-deficient mycobacterial forms. Importantly, spheroplasts prepared from M.smeg cells overexpressing MtbYhhN were protected from membrane disruption/lysis by pLPC, which was rapidly depleted from the media. Finally, we found that overexpression of full-length MtbYhhN in M.smeg increased its survival within human macrophages by 2.6-fold compared to vector controls. These data support the hypothesis that MtbYhhN protein confers a growth advantage for mycobacteria in macrophages by cleaving toxic host pLPC into potentially energy-producing products.

Sampling alternative conformational states of transporters and receptors with AlphaFold2

Equilibrium fluctuations and triggered conformational changes often underlie the functional cycles of membrane proteins. For example, transporters mediate the passage of molecules across cell membranes by alternating between inward- and outward-facing states, while receptors undergo intracellular structural rearrangements that initiate signaling cascades. Although the conformational plasticity of these proteins has historically posed a challenge for traditional de novo protein structure prediction pipelines, the recent success of AlphaFold2 (AF2) in CASP14 culminated in the modeling of a transporter in multiple conformations to high accuracy. Given that AF2 was designed to predict static structures of proteins, it remains unclear if this result represents an underexplored capability to accurately predict multiple conformations and/or structural heterogeneity. Here, we present an approach to drive AF2 to sample alternative conformations of topologically diverse transporters and G-protein-coupled receptors that are absent from the AF2 training set. Whereas models of most proteins generated using the default AF2 pipeline are conformationally homogeneous and nearly identical to one another, reducing the depth of the input multiple sequence alignments by stochastic subsampling led to the generation of accurate models in multiple conformations. In our benchmark, these conformations spanned the range between two experimental structures of interest, with models at the extremes of these conformational distributions observed to be among the most accurate (average template modeling score of 0.94). These results suggest a straightforward approach to identifying native-like alternative states, while also highlighting the need for the next generation of deep learning algorithms to be designed to predict ensembles of biophysically relevant states.

Unique features of different classes of G-protein-coupled receptors revealed from sequence coevolutionary and structural analysis

G-protein-coupled receptors (GPCRs) are the largest family of human membrane proteins and represent the primary targets of about one third of currently marketed drugs. Despite the critical importance, experimental structures have been determined for only a limited portion of GPCRs and functional mechanisms of GPCRs remain poorly understood. Here, we have constructed novel sequence coevolutionary models of the A and B classes of GPCRs and compared them with residue contact frequency maps generated with available experimental structures. Significant portions of structural residue contacts were successfully detected in the sequence-based covariational models. "Exception" residue contacts predicted from sequence coevolutionary models but not available structures added missing links that were important for GPCR activation and allosteric modulation. Moreover, we identified distinct residue contacts involving different sets of functional motifs for GPCR activation, such as the Na⁺ pocket, CWxP, DRY, PIF, and NPxxY motifs in the class A and the HETx and PxxG motifs in the class B. Finally, we systematically uncovered critical residue contacts tuned by allosteric modulation in the two classes of GPCRs, including those from the activation motifs and particularly the extracellular and intracellular loops in class A GPCRs. These findings provide a promising framework for rational design of ligands to regulate GPCR activation and allosteric modulation.

A reproducibility analysis-based statistical framework for residue-residue evolutionary coupling detection

Direct coupling analysis (DCA) has been widely used to infer evolutionary coupled residue pairs from the multiple sequence alignment (MSA) of homologous sequences. However, effectively selecting residue pairs with significant evolutionary couplings according to the result of DCA is a non-trivial task. In this study, we developed a general statistical framework for significant evolutionary coupling detection, referred to as irreproducible discovery rate (IDR)-DCA, which is based on reproducibility analysis of the coupling scores obtained from DCA on manually created MSA replicates. IDR-DCA was applied to select residue pairs for contact prediction for monomeric proteins, protein-protein interactions and monomeric RNAs, in which three different versions of DCA were applied. We demonstrated that with the application of IDR-DCA, the residue pairs selected using a universal threshold always yielded stable performance for contact prediction. Comparing with the application of carefully tuned coupling score cutoffs, IDR-DCA always showed better performance. The robustness of IDR-DCA was also supported through the MSA downsampling analysis. We further demonstrated the effectiveness of applying constraints obtained from residue pairs selected by IDR-DCA to assist RNA secondary structure prediction.

Structural model of human PORCN illuminates disease-associated variants and drug-binding sites

Wnt signaling is essential for normal development and is a therapeutic target in cancer. The enzyme PORCN, or porcupine, is a membrane-bound O-acyltransferase (MBOAT) that is required for the post-translational modification of all Wnts, adding an essential mono-unsaturated palmitoleic acid to a serine on the tip of Wnt hairpin 2. Inherited mutations in PORCN cause focal dermal hypoplasia, and therapeutic inhibition of PORCN slows the growth of Wnt-dependent cancers. Based on homology to mammalian MBOAT proteins, we developed and validated a structural model of human PORCN. The model accommodates palmitoleoyl-CoA and Wnt hairpin 2 in two tunnels in the conserved catalytic core, shedding light on the catalytic mechanism. The model predicts how previously uncharacterized human variants of uncertain significance can alter PORCN function. Drugs including ETC-159, IWP-L6 and LGK-974 dock in the PORCN catalytic site, providing insights into PORCN pharmacologic inhibition. This structural model enhances our mechanistic understanding of PORCN substrate recognition and catalysis, as well as the inhibition of its enzymatic activity, and can facilitate the development of improved inhibitors and the understanding of disease-relevant PORCN mutants. This article has an associated First Person interview with the joint first authors of the paper.

Recruitment of the TolA protein to cell constriction sites in Escherichia coli via three separate mechanisms, and a critical role for FtsWI activity in recruitment of both TolA and TolQ

The Tol-Pal system of Gram-negative bacteria helps maintain integrity of the cell envelope and ensures that invagination of the envelope layers during cell fission occurs in a well-coordinated manner. In E. coli, the five Tol-Pal proteins (TolQ, R, A, B and Pal) accumulate at cell constriction sites in a manner that normally requires the activity of the cell constriction initiation protein FtsN. While septal recruitment of TolR, TolB and Pal also requires the presence of TolQ and/or TolA, each of the the latter two can recognize constriction sites independently of the other system proteins. What attracts TolQ or TolA to these sites is unclear. We show that FtsN attracts both proteins in an indirect fashion, and that PBP1A, PBP1B and CpoB are dispensable for their septal recruitment. However, the β-lactam aztreonam readily interferes with septal accumulation of both TolQ and TolA, indicating that FtsN-stimulated production of septal peptidoglycan by the FtsWI synthase is critical to their recruitment. We also discovered that each of TolA's three domains can recognize division sites in a separate fashion. Notably, the middle domain (TolAII) is responsible for directing TolA to constriction sites in the absence of other Tol-Pal proteins and CpoB, while recruitment of TolAI and TolAIII requires TolQ and a combination of TolB, Pal, and CpoB, respectively. Additionally, we describe the construction and use of functional fluorescent sandwich fusions of the ZipA division protein, which should be more broadly valuable in future studies of the E. coli cell division machinery. IMPORTANCE Cell division (cytokinesis) is a fundamental biological process that is incompletely understood for any organism. Division of bacterial cells relies on a ring-like machinery called the septal ring or divisome that assembles along the circumference of the mother cell at the site where constriction will eventually occur. In the well-studied bacterium Escherichia coli, this machinery contains over thirty distinct proteins. We studied how two such proteins, TolA and TolQ, which also play a role in maintaining integrity of the outer-membrane, are recruited to the machinery. We find that TolA can be recruited by three separate mechanisms, and that both proteins rely on the activity of a well-studied cell division enzyme for their recruitment.

Improved protein contact prediction using dimensional hybrid residual networks and singularity enhanced loss function

Deep residual learning has shown great success in protein contact prediction. In this study, a new deep residual learning-based protein contact prediction model was developed. Comparing with previous models, a new type of residual block hybridizing 1D and 2D convolutions was designed to increase the effective receptive field of the residual network, and a new loss function emphasizing the easily misclassified residue pairs was proposed to enhance the model training. The developed protein contact prediction model referred to as DRN-1D2D was first evaluated on 105 CASP11 targets, 76 CAMEO hard targets and 398 membrane proteins together with two in house-developed reference models based on either the standard 2D residual block or the traditional BCE loss function, from which we confirmed that both the dimensional hybrid residual block and the singularity enhanced loss function can be employed to improve the model performance for protein contact prediction. DRN-1D2D was further evaluated on 39 CASP13 and CASP14 free modeling targets together with the two reference models and six state-of-the-art protein contact prediction models including DeepCov, DeepCon, DeepConPred2, SPOT-Contact, RaptorX-Contact and TripleRes. The result shows that DRN-1D2D consistently achieved the best performance among all these models.

Virus-inspired designs of antimicrobial nanocapsules

Antimicrobial resistance is becoming a serious burden for drug design. The challenges are in finding novel approaches for effectively targeting a number of different bacterial strains, and in delivering these to the site of action. We propose here a novel approach that exploits the assembly of antimicrobial peptidic units in nanocapsules that can penetrate and rupture the bacterial membrane. Additionally, the chemical versatility of the designed units can be tailored to specific targets and to the delivery of genetic material in the cell. The proposed design exploits a β-annulus (sequence ITHVGGVGGSIMAPVAVSRQLVGS) triskelion unit from the Tomato Bushy Stunt Virus, able to self assemble in solution, and functionalised with antimicrobial sequences to form dodecahedral antimicrobial nanocapsules. The stability and the activity of the antimicrobial β-annulus capsule is measured by molecular dynamics simulations in water and in the presence of model membranes.

Assessment of prediction methods for protein structures determined by NMR in CASP14: Impact of AlphaFold2

NMR studies can provide unique information about protein conformations in solution. In CASP14, three reference structures provided by solution NMR methods were available (T1027, T1029, and T1055), as well as a fourth data set of NMR‐derived contacts for an integral membrane protein (T1088). For the three targets with NMR‐based structures, the best prediction results ranged from very good (GDT_TS = 0.90, for T1055) to poor (GDT_TS = 0.47, for T1029). We explored the basis of these results by comparing all CASP14 prediction models against experimental NMR data. For T1027, NMR data reveal extensive internal dynamics, presenting a unique challenge for protein structure prediction methods. The analysis of T1029 motivated exploration of a novel method of “inverse structure determination,” in which an AlphaFold2 model was used to guide NMR data analysis. NMR data provided to CASP predictor groups for target T1088, a 238‐residue integral membrane porin, was also used to assess several NMR‐assisted prediction methods. Most groups involved in this exercise generated similar beta‐barrel models, with good agreement with the experimental data. However, as was also observed in CASP13, some pure prediction groups that did not use any NMR data generated models for T1088 that better fit the NMR data than the models generated using these experimental data. These results demonstrate the remarkable power of modern methods to predict structures of proteins with accuracies rivaling solution NMR structures, and that it is now possible to reliably use prediction models to guide and complement experimental NMR data analysis.

Protein sequence-to-structure learning: Is this the end(-to-end revolution)?

The potential of deep learning has been recognized in the protein structure prediction community for some time, and became indisputable after CASP13. In CASP14, deep learning has boosted the field to unanticipated levels reaching near-experimental accuracy. This success comes from advances transferred from other machine learning areas, as well as methods specifically designed to deal with protein sequences and structures, and their abstractions. Novel emerging approaches include (i) geometric learning, that is, learning on representations such as graphs, three-dimensional (3D) Voronoi tessellations, and point clouds; (ii) pretrained protein language models leveraging attention; (iii) equivariant architectures preserving the symmetry of 3D space; (iv) use of large meta-genome databases; (v) combinations of protein representations; and (vi) finally truly end-to-end architectures, that is, differentiable models starting from a sequence and returning a 3D structure. Here, we provide an overview and our opinion of the novel deep learning approaches developed in the last 2 years and widely used in CASP14.

Highly accurate protein structure prediction for the human proteome

Protein structures can provide invaluable information, both for reasoning about biological processes and for enabling interventions such as structure-based drug development or targeted mutagenesis. After decades of effort, 17% of the total residues in human protein sequences are covered by an experimentally determined structure1. Here we markedly expand the structural coverage of the proteome by applying the state-of-the-art machine learning method, AlphaFold2, at a scale that covers almost the entire human proteome (98.5% of human proteins). The resulting dataset covers 58% of residues with a confident prediction, of which a subset (36% of all residues) have very high confidence. We introduce several metrics developed by building on the AlphaFold model and use them to interpret the dataset, identifying strong multi-domain predictions as well as regions that are likely to be disordered. Finally, we provide some case studies to illustrate how high-quality predictions could be used to generate biological hypotheses. We are making our predictions freely available to the community and anticipate that routine large-scale and high-accuracy structure prediction will become an important tool that will allow new questions to be addressed from a structural perspective.

The evolutionary history of topological variations in the CPA/AT transporters

CPA/AT transporters are made up of scaffold and a core domain. The core domain contains two non-canonical helices (broken or reentrant) that mediate the transport of ions, amino acids or other charged compounds. During evolution, these transporters have undergone substantial changes in structure, topology and function. To shed light on these structural transitions, we create models for all families using an integrated topology annotation method. We find that the CPA/AT transporters can be classified into four fold-types based on their structure; (1) the CPA-broken fold-type, (2) the CPA-reentrant fold-type, (3) the BART fold-type, and (4) a previously not described fold-type, the Reentrant-Helix-Reentrant fold-type. Several topological transitions are identified, including the transition between a broken and reentrant helix, one transition between a loop and a reentrant helix, complete changes of orientation, and changes in the number of scaffold helices. These transitions are mainly caused by gene duplication and shuffling events. Structural models, topology information and other details are presented in a searchable database, CPAfold (cpafold.bioinfo.se).

Structural discrimination analysis for constraint selection in protein modeling

MOTIVATION

Protein structure modeling can be improved by the use of distance constraints between amino acid residues, provided such data reflects-at least partially-the native tertiary structure of the target system. In fact, only a small subset of the native contact map is necessary to successfully drive the model conformational search, so one important goal is to obtain the set of constraints with the highest true-positive rate, lowest redundancy, and greatest amount of information. In this work, we introduce a constraint evaluation and selection method based on the point-biserial correlation coefficient, which utilizes structural information from an ensemble of models to indirectly measure the power of each constraint in biasing the conformational search towards consensus structures.

RESULTS

Residue contact maps obtained by direct coupling analysis are systematically improved by means of discriminant analysis, reaching in some cases accuracies often seen only in modern deep-learning based approaches. When combined with an iterative modeling workflow, the proposed constraint classification optimizes the selection of the constraint set and maximizes the probability of obtaining successful models. The use of discriminant analysis for the valorization of the information of constraint data sets is a general concept with possible applications to other constraint types and modeling problems.

AVAILABILITY AND IMPLEMENTATION

scripts and procedures to implement the methodology presented herein are available at https://github.com/m3g/2021_Bottino_Biserial.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Molecular structure of the intact bacterial flagellar basal body

The bacterial flagellum is a macromolecular protein complex that enables motility in many species. Bacterial flagella self-assemble a strong, multicomponent drive shaft that couples rotation in the inner membrane to the micrometre-long flagellar filament that powers bacterial swimming in viscous fluids1-3. Here, we present structures of the intact Salmonella flagellar basal body4, encompassing the inner membrane rotor, drive shaft and outer-membrane bushing, solved using cryo-electron microscopy to resolutions of 2.2-3.7 Å. The structures reveal molecular details of how 173 protein molecules of 13 different types assemble into a complex spanning two membranes and a cell wall. The helical drive shaft at one end is intricately interwoven with the rotor component with both the export gate complex and the proximal rod forming interactions with the MS-ring. At the other end, the drive shaft distal rod passes through the LP-ring bushing complex, which functions as a molecular bearing anchored in the outer membrane through interactions with the lipopolysaccharide. The in situ structure of a protein complex capping the drive shaft provides molecular insights into the assembly process of this molecular machine.

A selective transmembrane recognition mechanism by a membrane-anchored ubiquitin ligase adaptor

While it is well-known that E3 ubiquitin ligases can selectively ubiquitinate membrane proteins in response to specific environmental cues, the underlying mechanisms for the selectivity are poorly understood. In particular, the role of transmembrane regions, if any, in target recognition remains an open question. Here, we describe how Ssh4, a yeast E3 ligase adaptor, recognizes the PQ-loop lysine transporter Ypq1 only after lysine starvation. We show evidence of an interaction between two transmembrane helices of Ypq1 (TM5 and TM7) and the single transmembrane helix of Ssh4. This interaction is regulated by the conserved PQ motif. Strikingly, recent structural studies of the PQ-loop family have suggested that TM5 and TM7 undergo major conformational changes during substrate transport, implying that transport-associated conformational changes may determine the selectivity. These findings thus provide critical information concerning the regulatory mechanism through which transmembrane domains can be specifically recognized in response to changing environmental conditions.

In vitro reconstitution reveals major differences between human and bacterial cytochrome c synthases

Cytochromes c are ubiquitous heme proteins in mitochondria and bacteria, all possessing a CXXCH (CysXxxXxxCysHis) motif with covalently attached heme. We describe the first in vitro reconstitution of cytochrome c biogenesis using purified mitochondrial (HCCS) and bacterial (CcsBA) cytochrome c synthases. We employ apocytochrome c and peptide analogs containing CXXCH as substrates, examining recognition determinants, thioether attachment, and subsequent release and folding of cytochrome c. Peptide analogs reveal very different recognition requirements between HCCS and CcsBA. For HCCS, a minimal 16-mer peptide is required, comprised of CXXCH and adjacent alpha helix 1, yet neither thiol is critical for recognition. For bacterial CcsBA, both thiols and histidine are required, but not alpha helix 1. Heme attached peptide analogs are not released from the HCCS active site; thus, folding is important in the release mechanism. Peptide analogs behave as inhibitors of cytochrome c biogenesis, paving the way for targeted control.

From tiny bacteria to the tallest trees, most life on Earth carries a protein called cytochrome c, which helps to create the energy that powers up cells. Cytochrome c does so thanks to its heme, a molecule that enables the chemical reactions required for the energy-creating process.

Despite both relying on cytochrome c, animals and bacteria differ in the enzyme they use to attach the heme to the cytochrome. Spotting variations in how this ‘cytochrome c synthase’ works would help to find compounds that deactivate the enzyme in bacteria, but not in humans. However, studying cytochrome c synthase in living cells is challenging.

To bypass this issue, Sutherland, Mendez, Babbitt et al. successfully reconstituted cytochrome c synthases from humans and bacteria in test tubes. This allowed them to examine in detail which structures the enzymes recognize to spot where to attach the heme onto their target. The experiments revealed that human and bacterial synthases actually rely on different parts of the cytochrome c to orient themselves. Different short compounds could also block either the human or bacterial enzyme.

Variations between human and bacterial cytochrome c synthase could lead to new antibiotics which deactivate the cytochrome and kill bacteria while sparing patients. The next step is to identify molecules that specifically interfere with cytochrome c synthase in bacteria, and could be tested in clinical trials.

A two-track model for the spatiotemporal coordination of bacterial septal cell wall synthesis revealed by single-molecule imaging of FtsW

Synthesis of septal peptidoglycan (sPG) is crucial for bacterial cell division. FtsW, an indispensable component of the cell division machinery in all walled bacterial species, was recently identified in vitro as a peptidoglycan glycosyltransferase (PGTase). Despite its importance, the septal PGTase activity of FtsW has not been demonstrated in vivo. How its activity is spatiotemporally regulated in vivo has also remained elusive. Here, we confirmed FtsW as an essential septum-specific PGTase in vivo using an N-acetylmuramic acid analogue incorporation assay. Next, using single-molecule tracking coupled with genetic manipulations, we identified two populations of processively moving FtsW molecules: a fast-moving population correlated with the treadmilling dynamics of the essential cytoskeletal FtsZ protein and a slow-moving population dependent on active sPG synthesis. We further identified that FtsN, a potential sPG synthesis activator, plays an important role in promoting the slow-moving population. Our results suggest a two-track model, in which inactive sPG synthases follow the 'Z-track' to be distributed along the septum and FtsN promotes their release from the Z-track to become active in sPG synthesis on the slow 'sPG-track'. This model provides a mechanistic framework for the spatiotemporal coordination of sPG synthesis in bacterial cell division.

On the use of direct-coupling analysis with a reduced alphabet of amino acids combined with super-secondary structure motifs for protein fold prediction

Direct-coupling analysis (DCA) for studying the coevolution of residues in proteins has been widely used to predict the three-dimensional structure of a protein from its sequence. We present RADI/raDIMod, a variation of the original DCA algorithm that groups chemically equivalent residues combined with super-secondary structure motifs to model protein structures. Interestingly, the simplification produced by grouping amino acids into only two groups (polar and non-polar) is still representative of the physicochemical nature that characterizes the protein structure and it is in line with the role of hydrophobic forces in protein-folding funneling. As a result of a compressed alphabet, the number of sequences required for the multiple sequence alignment is reduced. The number of long-range contacts predicted is limited; therefore, our approach requires the use of neighboring sequence-positions. We use the prediction of secondary structure and motifs of super-secondary structures to predict local contacts. We use RADI and raDIMod, a fragment-based protein structure modelling, achieving near native conformations when the number of super-secondary motifs covers >30-50% of the sequence. Interestingly, although different contacts are predicted with different alphabets, they produce similar structures.

Genetic analysis of the septal peptidoglycan synthase FtsWI complex supports a conserved activation mechanism for SEDS-bPBP complexes

SEDS family peptidoglycan (PG) glycosyltransferases, RodA and FtsW, require their cognate transpeptidases PBP2 and FtsI (class B penicillin binding proteins) to synthesize PG along the cell cylinder and at the septum, respectively. The activities of these SEDS-bPBPs complexes are tightly regulated to ensure proper cell elongation and division. In Escherichia coli FtsN switches FtsA and FtsQLB to the active forms that synergize to stimulate FtsWI, but the exact mechanism is not well understood. Previously, we isolated an activation mutation in ftsW (M269I) that allows cell division with reduced FtsN function. To try to understand the basis for activation we isolated additional substitutions at this position and found that only the original substitution produced an active mutant whereas drastic changes resulted in an inactive mutant. In another approach we isolated suppressors of an inactive FtsL mutant and obtained FtsW^E289G and FtsI^K211I and found they bypassed FtsN. Epistatic analysis of these mutations and others confirmed that the FtsN-triggered activation signal goes from FtsQLB to FtsI to FtsW. Mapping these mutations, as well as others affecting the activity of FtsWI, on the RodA-PBP2 structure revealed they are located at the interaction interface between the extracellular loop 4 (ECL4) of FtsW and the pedestal domain of FtsI (PBP3). This supports a model in which the interaction between the ECL4 of SEDS proteins and the pedestal domain of their cognate bPBPs plays a critical role in the activation mechanism.

Bacterial cell division requires the synthesis of septal peptidoglycan by the widely conserved SEDS-bPBP protein complex FtsWI, but how the complex is activated during cell division is still poorly understood. Previous studies suggested that FtsN initiates a signaling cascade in the periplasm to activate FtsWI. Here we isolated and characterized activated FtsW and FtsI mutants and confirmed that the signaling cascade for FtsW activation goes from FtsN to FtsQLB to FtsI and then to FtsW. The residues corresponding to mutations affecting FtsWI activation are clustered to a small region of the interaction interface between the pedestal domain of FtsI and the extracellular loop 4 of FtsW, suggesting that this interaction mediates activation of FtsW. This is strikingly similar to the proposed activation mechanism for the RodA-PBP2 complex, another SEDS-bPBP complex required for cell elongation. Thus, the two homologous SEDS-bPBP complexes are activated similarly by completely unrelated activators that modulate the interaction interface between the SEDS proteins and the bPBPs.

Identification of potential regulatory domains within the MreC and MreD components of the cell elongation machinery

The bacterial peptidoglycan (PG) cell wall maintains cell shape and prevents osmotic lysis. During growth of rod-shaped cells, PG is incorporated along the cell cylinder by the RodA-PBP2 synthase of the multi-protein Rod system (elongasome). Filaments of the actin-like MreB protein orient synthesis of the new PG material. They are connected to the RodA-PBP2 synthase in part through the RodZ component. MreC and MreD are other conserved components of the system, but their function is not well understood. Amino acid changes in RodA-PBP2 were recently identified that bypass a requirement for MreC and MreD function, suggesting the Mre proteins act as activators of the synthase. To further investigate their function, we developed a genetic strategy to identify dominant-negative alleles of mreC and mreD in Escherichia coli Residues essential for Rod system function were identified at the junction of two subdomains within MreC and in a predicted ligand-binding pocket of MreD. Additionally, we found that although the proline-rich C-terminal domain of MreC is non-essential, substitutions within this region disrupt its function. Based on these results, we propose that the C-terminus of MreC and the putative ligand-binding domain of MreD play regulatory roles in controlling Rod system activity.IMPORTANCE: Cell shape in bacteria is largely determined by the cell wall structure that surrounds them. The multi-protein machine called the Rod system (elongasome) has long been implicated in rod-shape determination in bacilli. However, the functions of many of its conserved components remain unclear. Here, we describe a new genetic system to dissect the function of these proteins and how we used it to identify potential regulatory domains within them that may modulate the function of the shape-determining machinery.

Large-scale discovery of protein interactions at residue resolution using co-evolution calculated from genomic sequences

Increasing numbers of protein interactions have been identified in high-throughput experiments, but only a small proportion have solved structures. Recently, sequence coevolution-based approaches have led to a breakthrough in predicting monomer protein structures and protein interaction interfaces. Here, we address the challenges of large-scale interaction prediction at residue resolution with a fast alignment concatenation method and a probabilistic score for the interaction of residues. Importantly, this method (EVcomplex2) is able to assess the likelihood of a protein interaction, as we show here applied to large-scale experimental datasets where the pairwise interactions are unknown. We predict 504 interactions de novo in the E. coli membrane proteome, including 243 that are newly discovered. While EVcomplex2 does not require available structures, coevolving residue pairs can be used to produce structural models of protein interactions, as done here for membrane complexes including the Flagellar Hook-Filament Junction and the Tol/Pal complex.

Our understanding of the residue-level details of protein interactions remains incomplete. Here, the authors show sequence coevolution can be used to infer interacting proteins with residue-level details, including predicting 467 interactions de novo in the Escherichia coli cell envelope proteome.

Mi3-GPU: MCMC-based Inverse Ising Inference on GPUs for protein covariation analysis

Inverse Ising inference is a method for inferring the coupling parameters of a Potts/Ising model based on observed site-covariation, which has found important applications in protein physics for detecting interactions between residues in protein families. We introduce Mi3-GPU ("mee-three", for MCMC Inverse Ising Inference) software for solving the inverse Ising problem for protein-sequence datasets with few analytic approximations, by parallel Markov-Chain Monte-Carlo sampling on GPUs. We also provide tools for analysis and preparation of protein-family Multiple Sequence Alignments (MSAs) to account for finite-sampling issues, which are a major source of error or bias in inverse Ising inference. Our method is "generative" in the sense that the inferred model can be used to generate synthetic MSAs whose mutational statistics (marginals) can be verified to match the dataset MSA statistics up to the limits imposed by the effects of finite sampling. Our GPU implementation enables the construction of models which reproduce the covariation patterns of the observed MSA with a precision that is not possible with more approximate methods. The main components of our method are a GPU-optimized algorithm to greatly accelerate MCMC sampling, combined with a multi-step Quasi-Newton parameter-update scheme using a "Zwanzig reweighting" technique. We demonstrate the ability of this software to produce generative models on typical protein family datasets for sequence lengths L ~ 300 with 21 residue types with tens of millions of inferred parameters in short running times.

Structures of the stator complex that drives rotation of the bacterial flagellum

The bacterial flagellum is the prototypical protein nanomachine and comprises a rotating helical propeller attached to a membrane-embedded motor complex. The motor consists of a central rotor surrounded by stator units that couple ion flow across the cytoplasmic membrane to generate torque. Here, we present the structures of the stator complexes from Clostridium sporogenes, Bacillus subtilis and Vibrio mimicus, allowing interpretation of the extensive body of data on stator mechanism. The structures reveal an unexpected asymmetric A5B2 subunit assembly where the five A subunits enclose the two B subunits. Comparison to structures of other ion-driven motors indicates that this A5B2 architecture is fundamental to bacterial systems that couple energy from ion flow to generate mechanical work at a distance and suggests that such events involve rotation in the motor structures.

Structure of bacterial phospholipid transporter MlaFEDB with substrate bound

In double-membraned bacteria, phospholipid transport across the cell envelope is critical to maintain the outer membrane barrier, which plays a key role in virulence and antibiotic resistance. An MCE transport system called Mla has been implicated in phospholipid trafficking and outer membrane integrity, and includes an ABC transporter, MlaFEDB. The transmembrane subunit, MlaE, has minimal sequence similarity to other transporters, and the structure of the entire inner-membrane MlaFEDB complex remains unknown. Here, we report the cryo-EM structure of MlaFEDB at 3.05 Å resolution, revealing distant relationships to the LPS and MacAB transporters, as well as the eukaryotic ABCA/ABCG families. A continuous transport pathway extends from the MlaE substrate-binding site, through the channel of MlaD, and into the periplasm. Unexpectedly, two phospholipids are bound to MlaFEDB, suggesting that multiple lipid substrates may be transported each cycle. Our structure provides mechanistic insight into substrate recognition and transport by MlaFEDB.

αβDCA method identifies unspecific binding but specific disruption of the group I intron by the StpA chaperone

Chaperone proteins-the most disordered among all protein groups-help RNAs fold into their functional structure by destabilizing misfolded configurations or stabilizing the functional ones. But disentangling the mechanism underlying RNA chaperoning is challenging, mostly because of inherent disorder of the chaperones and the transient nature of their interactions with RNA. In particular, it is unclear how specific the interactions are and what role is played by amino acid charge and polarity patterns. Here, we address these questions in the RNA chaperone StpA. We adapted direct coupling analysis (DCA) into the αβDCA method that can treat in tandem sequences written in two alphabets, nucleotides and amino acids. With αβDCA, we could analyze StpA-RNA interactions and show consistency with a previously proposed two-pronged mechanism: StpA disrupts specific positions in the group I intron while globally and loosely binding to the entire structure. Moreover, the interactions are strongly associated with the charge pattern: Negatively charged regions in the destabilizing StpA amino-terminal affect a few specific positions in the RNA, located in stems and in the pseudoknot. In contrast, positive regions in the carboxy-terminal contain strongly coupled amino acids that promote nonspecific or weakly specific binding to the RNA. The present study opens new avenues to examine the functions of disordered proteins and to design disruptive proteins based on their charge patterns.

Deep Learning in Proteomics

Proteomics, the study of all the proteins in biological systems, is becoming a data-rich science. Protein sequences and structures are comprehensively catalogued in online databases. With recent advancements in tandem mass spectrometry (MS) technology, protein expression and post-translational modifications (PTMs) can be studied in a variety of biological systems at the global scale. Sophisticated computational algorithms are needed to translate the vast amount of data into novel biological insights. Deep learning automatically extracts data representations at high levels of abstraction from data, and it thrives in data-rich scientific research domains. Here, a comprehensive overview of deep learning applications in proteomics, including retention time prediction, MS/MS spectrum prediction, de novo peptide sequencing, PTM prediction, major histocompatibility complex-peptide binding prediction, and protein structure prediction, is provided. Limitations and the future directions of deep learning in proteomics are also discussed. This review will provide readers an overview of deep learning and how it can be used to analyze proteomics data.

Pair Potentials as Machine Learning Features

Atom pairwise potential functions make up an essential part of many scoring functions for protein decoy detection. With the development of machine learning (ML) tools, there are multiple ways to combine potential functions to create novel ML models and methods. Potential function parameters can be easily extracted; however, it is usually hard to directly obtain the calculated atom pairwise energies from scoring functions. Amber, as one of the most popular suites of modeling programs, has an extensive history and library of force field potential functions. In this work, we directly used the force field parameters in ff94 and ff14SB from Amber and encoded them to calculate atom pairwise energies for different interactions. Two sets of structures (single amino acid set and a dipeptide set) were used to evaluate the performance of our encoded Amber potentials. From the comparison results between energy terms obtained from our encoding and Amber, we find energy difference within ±0.06 kcal/mol for all tested structures. Previously we have shown that the Random Forest (RF) model can help to emphasize more important atom pairwise interactions and ignore insignificant ones [Pei, J.; Zheng, Z.; Merz, K. M. J. Chem. Inf. Model. 2019, 59, 1919-1929]. Here, as an example of combining ML methods with traditional potential functions, we followed the same work flow to combine the RF models with force field potential functions from Amber. To determine the performance of our RF models with force field potential functions, 224 different protein native-decoy systems were used as our training and testing sets We find that the RF models with ff94 and ff14SB force field parameters outperformed all other scoring functions (RF models with KECSA2, RWplus, DFIRE, dDFIRE, and GOAP) considered in this work for native structure detection, and they performed similarly in detecting the best decoy. Through inclusion of best decoy to decoy comparisons in building our RF models, we were able to generate models that outperformed the score functions tested herein both on accuracy and best decoy detection, again showing the performance and flexibility of our RF models to tackle this problem. Finally, the importance of the RF algorithm and force field parameters were also tested and the comparison results suggest that both the RF algorithm and force field potentials are important with the ML scoring function achieving its best performance only by combining them together. All code and data used in this work are available at https://github.com/JunPei000/FFENCODER_for_Protein_Folding_Pose_Selection.