Protein structures are stabilized by several types of chemical interactions between amino acids, which can compete with each other. This is the case of chalcogen and hydrogen bonds formed by the thiol group of cysteine, which can form three hydrogen bonds with one hydrogen acceptor and two hydrogen donors and a chalcogen bond with a nucleophile along the extension of the C-S bond. A survey of the Protein Data Bank shows that hydrogen bonds are about 40-50 more common than chalcogen bonds, suggesting that they are stronger and, consequently, prevail, though not always. It is also observed that frequently a thiol group that forms a chalcogen bond is also involved, as a hydrogen donor, in a hydrogen bond.

The revelation of protein folding is a challenging subject in both discovery and description. Except acquirement of accurate 3D structure for protein stable state, another big hurdle is how to discover structural flexibility for protein innate character. Even if a huge number of flexible conformations are known, difficulty is how to describe these conformations. A novel approach, protein structure fingerprint, has been developed to expose the comprehensive local folding variations, and then construct folding conformations for entire protein. The backbone of 5 amino acid residues was identified as a universal folden, and then a set of Protein Folding Shape Code (PFSC) was derived for completely covering folding space in alphabetic description. Sequentially, a database was created to collect all possible folding shapes of local folding variations for all permutation of 5 amino acids. Successively, Protein Folding Variation Matrix (PFVM) assembled all possible local folding variations along sequence for a protein, which possesses several prominent features. First, it showed the fluctuation with certain folding patterns along sequence which revealed how the protein folding was related the order of amino acids in sequence. Second, all folding variations for an entire protein can be simultaneously apprehended at a glance within PFVM. Third, all conformations can be determined by local folding variations from PFVM, so total number of conformations is no longer ambiguous for any protein. Finally, the most possible folding conformation and its 3D structure can be acquired according PFVM for protein structure prediction. Therefore, the protein structure fingerprint approach provides a significant means for investigation of protein folding problem.

Processing of CASP15 targets into evaluation units (EUs) and assigning them to evolutionary-based prediction classes is presented in this study. The targets were first split into structural domains based on compactness and similarity to other proteins. Models were then evaluated against these domains and their combinations. The domains were joined into larger EUs if predictors’ performance on the combined units was similar to that on individual domains. Alternatively, if most predictors performed better on the individual domains, then they were retained as EUs. As a result, 112 evaluation units were created from 77 tertiary structure prediction targets. The EUs were assigned to four prediction classes roughly corresponding to target difficulty categories in previous CASPs: TBM (template-based modeling, easy or hard), FM (free modeling), and the TBM/FM overlap category. More than a third of CASP15 EUs were attributed to the historically most challenging FM class, where homology or structural analogy to proteins of known fold cannot be detected.

Cysteine (Cys) is the most reactive amino acid participating in a wide range of biological functions. In-silico predictions complement the experiments to meet the need of functional characterization. Multiple Cys function prediction algorithm is scarce, in contrast to specific function prediction algorithms. Here we present a deep neural network-based multiple Cys function prediction, available on web-server (DeepCys) (https://deepcys.herokuapp.com/). DeepCys model was trained and tested on two independent datasets curated from protein crystal structures. This prediction method requires three inputs, namely, PDB identifier (ID), chain ID and residue ID for a given Cys and outputs the probabilities of four cysteine functions, namely, disulphide, metal-binding, thioether and sulphenylation and predicts the most probable Cys function. The algorithm exploits the local and global protein properties, like, sequence and secondary structure motifs, buried fractions, microenvironments and protein/enzyme class. DeepCys outperformed most of the multiple and specific Cys function algorithms. This method can predict maximum number of cysteine functions. Moreover, for the first time, explicitly predicts thioether function. This tool was used to elucidate the cysteine functions on domains of unknown functions (DUFs) belonging to cytochrome C oxidase subunit-II (COX2) like transmembrane domains. Apart from the web-server, a standalone program is also available on GitHub (https://github.com/vam-sin/deepcys)

Protein domains are structural, functional, and evolutionary units. These domains bring out the diversity of functionality by means of interactions with other co-existing domains and provide stability. Hence, it is important to study intra-protein inter-domain interactions from the perspective of types of interactions. Domains within a chain could interact over short timeframes or permanently, rather like protein-protein interactions (PPIs). However, no systematic study has been carried out between two classes, namely permanent and transient domain-domain interactions (DDIs). In this work, we studied 264 two-domain proteins, belonging to either of these classes and their interfaces on the basis of several factors, such as interface area and details of interactions (number, strengths, and types of interactions). We also characterized them based on residue conservation at the interface, correlation of residue motions across domains, its involvement in repeat formation, and their involvement in particular molecular processes. Finally, we could analyse the interactions arising from domains in two-domain monomeric proteins, and we observed significant differences between these two classes of domain interactions and a few similarities. This study will help to obtain a better understanding of structure-function and folding principles of multi-domain proteins.

Ryanodine receptor 1 (RyR1) is an intracellular calcium ion (Ca2+) release channel required for skeletal muscle contraction. Although cryo-electron microscopy identified binding sites of three coactivators Ca2+, ATP and caffeine (CFF), the mechanism of co-regulation and synergy of these activators is unknown. Here, we report allosteric connections among the three ligand binding sites and pore region in (i) Ca2+ bound-closed, (ii) ATP/CFF bound- closed, (iii) Ca2+/ATP/CFF bound-closed, and (iv) Ca2+/ATP/CFF bound-open RyR1 states. We identified two dominant interactions that mediate interactions between the Ca2+ binding site and pore region in Ca2+ bound-closed state, which partially overlapped with the pore communications in ATP/CFF bound-closed RyR1 state. In Ca2+/ATP/CFF bound-closed and -open RyR1 states, co-regulatory interactions were analogous to communications in the Ca2+ bound-closed and ATP/CFF bound- closed states. Both ATP- and CFF- binding sites mediate communication between the Ca2+ binding site and the pore region in Ca2+/ATP/CFF bound - open RyR1 structure. We conclude that Ca2+, ATP, and CFF propagate their effects to the pore region through a network of overlapping interactions that mediate allosteric control and molecular synergy in channel regulation.

Lipid transporters play an important role in most if not all organisms, ranging from bacteria to humans. For example, in Mycobacterium tuberculosis, the trehalose monomycolate transporter MmpL3 is involved in cell wall biosynthesis, while in humans, cholesterol transporters are involved in normal cell function as well as in disease. Here, using structural and bioinformatics information, we propose that there are proteins that also contain “MmpL3-like” (MMPL) transmembrane (TM) domains in many protozoa, including Trypanosoma cruzi, as well as in the bacterium Staphylococcus aureus, where the fatty acid transporter FarE has the same set of “active-site” residues as those found in the mycobacterial MmpL3s, and in T. cruzi. We also show that there are strong sequence and predicted structural similarities between the TM proton-translocation domain seen in the X-ray structures of mycobacterial MmpL3s and several human as well as fungal lipid transporters, leading to the proposal that there are similar proteins in apicomplexan parasites, and in plants. The animal, fungal, apicomplexan and plant proteins have larger extra-membrane domains than are found in the bacterial MmpL3, but they have a similar TM domain architecture, with the introduction of a (catalytically essential) Phe>His residue change, and a Ser/Thr H-bond network, involved in H +-transport. Overall, the results are of interest since they show that MMPL-family proteins are present in essentially all life-forms: archaea, bacteria, protozoa, fungi, plants and animals and, where known, they are involved in “lipid” (glycolipid, phospholipid, sphingolipid, fatty acid, cholesterol, ergosterol) transport, powered by transmembrane molecular pumps having similar structures.

Substantial progresses in protein structure prediction have been made by utilizing deep-learning and residue-residue distance prediction since CASP13. Inspired by the advances, we improve our CASP14 MULTICOM protein structure prediction system in three main aspects: (1) a new deep learning based protein inter-residue distance predictor (DeepDist) to improve template-free (ab initio) tertiary structure prediction, (2) an enhanced template-based tertiary structure prediction method, and (3) distance-based model quality assessment methods empowered by deep learning. In the 2020 CASP14 experiment, MULTICOM predictor was ranked 7th out of 146 predictors in protein tertiary structure prediction and ranked 3rd out of 136 predictors in inter-domain structure predic-tion. The results of MULTICOM demonstrate that the template-free modeling based on deep learning and residue-residue distance prediction can predict the correct topology for almost all template-based modeling targets and a majority of hard targets (template-free targets or targets whose templates cannot be recognized), which is a significant improvement over the CASP13 MULTICOM predictor. The performance of template-free tertiary structure prediction largely depends on the accuracy of distance pre-dictions that is closely related to the quality of multiple sequence alignments. The structural model quality assessment works reasonably well on targets for which a sufficient number of good models can be predicted, but may perform poorly when only a few good models are predicted for a hard target and the distribution of model quality scores is highly skewed.

This article reports and analyzes the results of protein complex model accuracy estimation by our methods (DeepUMQA3 and GraphGPSM) in the 15 th Critical Assessment of techniques for protein Structure Prediction (CASP15). The new deep learning-based multimeric complex model accuracy estimation methods are proposed based on the ensemble of three level features coupling with deep residual/graph neural networks. For the input multimeric complex model, we describe it from three levels: overall complex features, intra-monomer features, and inter-monomer features. We designed an overall ultrafast shape recognition (USR) to characterize the relationship between local residues and the overall complex topology, and an inter-monomer USR to characterize the relationship between the residues of one monomer and the topology of other monomers. On the 39 complex targets of CASP15, DeepUMQA3 (Group name: GuijunLab-RocketX) ranked first in the assessment of interface residue accuracy. The Pearson correlation coefficient (PCC) between the interface residues lDDT predicted by DeepUMQA3 and the real lDDT is 0.570, and DeepUMQA3 achieved the highest PCC on 29 out of 39 targets. GraphGPSM (Group name: GuijunLab-PAthreader) had a TM-score PCC>0.9 on 14 targets, showing a good ability to estimate the overall fold accuracy.

This paper aims to understand the binding strategies of a nanobody-protein pair by studying known complexes. Rigid body protein-ligand docking programs produce several complexes, called decoys, which are good candidates with high scores of shape complementarity, electrostatic interactions, desolvation, buried surface area, and Lennard-Jones potentials. It is not known which decoy represents the true structure. We studied thirty-seven nanobody-protein complexes from the Single Domain Antibody Database, sd-Ab DB, [http://www.sdab-db.ca/](http://www.sdab-db.ca/). For each structure, a large number of decoys are generated using the Fast Fourier Transform algorithm of the software ZDOCK. The decoys were ranked according to their target protein-nanobody interaction energies, calculated by using the Dreiding Force Field, with rank 1 having the lowest interaction energy. Out of thirty-six PDB structures, twenty-five true structures were predicted as rank 1. Eleven of the remaining structures required Ångstrom size rigid body translations of the nanobody relative to the protein to match the given PDB structure. After the translation the Dreiding interaction (DI) energies of all complexes decreased and became rank 1. In one case, rigid body rotations as well as translations of the nanobody were required for matching the crystal structure. We used a Monte Carlo algorithm that randomly translates and rotates the nanobody of a decoy and calculates the DI energy. Results show that rigid body translations and the DI energy are sufficient for determining the correct binding location and pose of ZDOCK created decoys. A survey of the sd-Ab DB showed that each nanobody makes at least one salt bridge with its partner protein, indicating that salt bridge formation is an essential strategy in nanobody-protein recognition. Based on the analysis of the thirty-six crystal structures and evidence from existing literature, we propose a set of principles that could be used in the design of nanobodies.

The human C14orf166 protein, also known as RTRAF, shows positive modulatory activity on the cellular RNA polymerase II enzyme. This protein is a component of the tRNA-splicing ligase complex and is involved in RNA metabolism. It also functions in the nucleo-cytoplasmic transport of RNA molecules. The C14orf166 protein has been reported to be associated with some types of cancer. It has been shown that the C14orf166 protein binds to the influenza A virus RNA polymerase PA subunit and has a stimulating effect on viral replication. In this study, candidate interactor proteins for influenza A virus PA protein were screened with a Y2H assay using HEK293 Matchmaker cDNA. The C14orf166 protein fragments in different sizes were found to interact with the PA. The three-dimensional structures of the viral PA and C14orf166 proteins interacting with the PA were generated using the I-TASSER algorithm. The interaction models between these proteins were predicted with the ClusPro protein docking algorithm and analyzed with PyMol software. The results revealed that the carboxy-terminal end of the C14orf166 protein is involved in this interaction, and it is highly possible that it binds to the carboxy-terminal of the PA protein. Although amino acid residues in the interaction area of the PA protein with the C14orf166 showed distribution from 450th to 700th position, the intense interaction region was revealed to be at amino acid positions 610 to 630.

In CASP15 there was a greater emphasis on multimeric modelling than in previous experiments, with assembly structures nearly doubling in number (41 up from 22) since the previous round. CASP15 also included a new estimation of model accuracy (EMA) category in recognition of the importance of objective quality assessment for quaternary structure models. ModFOLDdock is a multimeric model quality assessment server developed by the McGuffin group at the University of Reading, which brings together a range of single-model, clustering and deep learning methods to form a consensus of approaches. For CASP15 three variants of ModFOLDdock were developed to optimise for the different facets of the quality estimation problem. The standard ModFOLDdock variant produced predicted scores optimised for positive linear correlations with the observed scores. The ModFOLDdockR variant produced predicted scores optimised for ranking, i.e., the top-ranked models have highest accuracy. In addition, the ModFOLDdockS variant used a quasi-single model approach to score each model on an individual basis. The scores from all three variants achieved strongly positive Pearson correlation coefficients with the CASP observed scores (oligo-lDDT) in excess of 0.70, which were maintained across both homomeric and heteromeric model populations. In addition, at least one of the ModFOLDdock variants was consistently ranked in the top two methods across all three EMA categories. Specifically, for overall global fold prediction accuracy, ModFOLDdock placed second and ModFOLDdockR placed third; for overall interface quality prediction accuracy ModFOLDdockR, ModFOLDdock and ModFOLDdockS were placed above all other predictor methods, and ModFOLDdockR and ModFOLDdockS were placed second and third respectively for individual residue confidence scores. The ModFOLDdock server is available at: [https://www.reading.ac.uk/bioinf/ModFOLDdock/](https://www.reading.ac.uk/bioinf/ModFOLDdock/). ModFOLDdock is also available as part of the MultiFOLD docker package: [https://hub.docker.com/r/mcguffin/multifold](https://hub.docker.com/r/mcguffin/multifold)

Probing the structures of amyloid-beta (Aβ) peptides in the early steps of aggregation is extremely difficult experimentally and computationally. Yet, this knowledge is extremely important as small oligomers are the most toxic species. Experiments and simulations on Aβ42 monomer point to random coil conformations with either transient helical or β-strand content. Our current conformational description of small Aβ42 oligomers is funneled toward amorphous aggregates with some β-sheet content and rare excited states with well-ordered assemblies of β-sheets. In this study, we emphasize another view based on metastable α-helix bundle oligomers spanning the C-terminus residues which are predicted by the machine-learning AlphaFold2 method and supported indirectly by low-resolution experimental data on many amyloid polypeptides. This finding has consequences in designing drugs to reduce aggregation and toxicity.

In vertebrates, the mineralocorticoid receptor (MR) is a steroid-activated nuclear receptor (NR) that plays essential roles in water-electrolyte balance and blood pressure homeostasis. It belongs to the group of oxo-steroidian NRs, together with the glucocorticoid (GR), progesterone (PR), and androgen (AR) receptors. Classically, these oxo-steroidian NRs homodimerize and bind to specific genomic sequences to activate gene expression. NRs are multi-domain proteins, and dimerization is mediated by both the DNA (DBD) and ligand binding (LBD) domains, with the latter thought to provide the largest dimerization interface. However, at the structural level, the LBD dimerization of oxo-steroidian receptors has remained largely a matter of debate. This is linked to the receptor refractory expression, purification and crystallization. As a result, there is currently no consensus on a common homodimer assembly across the 4 receptors, i.e. GR, PR, AR and MR, despite their sequence homology. Examining the available MR LBD crystals and using widely plebiscited tools such as PISA, PRISM and EPPIC, and the MM/PBSA method, we have determined that an interface mediated by the helices H9 and H10 of the LBD as well as by the F domain presents the features of a biological protein-protein interaction surface. This interface which has been observed in both GR alpha and MR crystals, distinguished itself among other contacts and provided for the first time a homodimer architecture that is common to both oxo-steroidian receptors.

Histone is a scaffold protein that constitutes nucleosomes with DNA in the cell nucleus. When forming histone, hetero octamer is assisted by histone chaperone proteins. As a histone chaperone protein, the crystal structure of yeast nucleosome assembly protein (yNap1) has been determined. For yNap1, a nuclear export signal/sequence (NES) has been identified as a part of the long -helix. Experimental evidence via mutagenesis on budding yeast suggests the NES is necessary for transport out from the cell nucleus. However, the NES is masked by a region defined as an accessory domain (AD). In addition, the role of the AD in nuclear transport has not been elucidated yet. To address the role of the AD, we focused on phosphorylation in the AD because proteome experiments have identified multiple phosphorylation sites of yNap1. To computationally treat phosphorylation, we performed all-atom molecular dynamics (MD) simulations for a set of non-phosphorylated and phosphorylated yNap1 (Nap1-nonP and Nap1-P). As an analysis, we addressed how the NES is exposed to the protein surface by measuring its solvent-access surface area (SASA). As a result, there was a difference in the SASA distributions between both systems. Quantitatively, the median of the SASA distribution of Nap1-P was greater than that of Nap1-nonP, meaning that phosphorylation in the AD exposed to the NES, resulting in increasing its accessibility. In conclusion, yNap1 might modulate the accessibility of the NES by dislocating the AD through phosphorylation.

The inversion from L- to D-stereochemistry endows peptides improved bioactivity and enhanced resistance to many proteases and peptidases. To strengthen the biostability and bioavailability of peptide drugs, enzymatic epimerization becomes an important way to incorporate D-amino acid into peptide backbones. Recently, a bifunctional thioesterase NocTE, which is responsible for the epimerization and hydrolysis of the C-terminal (p-hydroxyphenyl)glycine residue of β-lactam antibiotic nocardicin A, exclusively directs to the generation of D-diastereomers. Different from other epimerases, NocTE exhibits unique stereochemical selectivity. Herein, we investigated the catalytic mechanism of NocTE via molecular dynamic (MD) simulations and quantum mechanical/molecular mechanics (QM/MM) calculations. Through structural analyses, two key water molecules around the reaction site were found to serve as proton mediators in epimerization. The structural characteristics inspired us to propose a substrate-assisted mechanism for the epimerization, where multi-step proton transfers were mediated by water molecules and β-lactam ring, and the free energy barrier was calculated to be 20.3 kcal/mol. After that, the hydrolysis of D-configured substrate was energetically feasible with the energy barrier of 14.3 kcal/mol. As a comparison, the energy barrier for the direct hydrolysis of L-configured substrate was obtained to be 24.0 kcal/mol. Our study provides mechanistic insights into catalytic activities of bifunctional thioesterase NocTE, uncovers more clues to the molecular basis for stereochemical selectivity and paves the way for the directed biosynthesis of novel peptide drugs with various stereostructural characteristics by enzyme rational design.

To greatly expand the druggable genome, fast and accurate predictions of cryptic sites for small molecules binding in target proteins are in high demand. In this study, we have developed a fast and simple conformational sampling scheme guided by normal modes solved from the coarse-grained elastic models followed by atomistic backbone refinement and sidechain repacking. Despite the observations of complex and diverse conformational changes associated with ligand binding, we found that simply sampling along each of the lowest 30 modes is near optimal for adequately restructuring cryptic sites so they can be detected by existing pocket finding programs like fpocket and concavity. We further trained machine-learning protocols to optimize the combination of the sampling-enhanced pocket scores with other dynamic and conservation scores, which only slightly improved the performance. As assessed based on a training set of 84 known cryptic sites and a test set of 14 proteins, our method achieved high accuracy of prediction (with area under the receiver operating characteristic curve > 0.8) comparable to the CryptoSite server. Compared with CryptoSite and other methods based on extensive molecular dynamics simulation, our method is much faster (1-2 hours for an average-size protein) and simpler (using only pocket scores), so it is suitable for high-throughput processing of large datasets of protein structures at the genome scale.

The heterologous overexpression states of prion proteins play a critical role in understanding the mechanisms of prion-related diseases. We report herein the identification of soluble monomer and complex states for a bakers’ yeast prion, Sup35, when expressed in E. coli. Two peaks are apparent with the elution of His-tagged Sup35 by imidazole from a Ni 2+ affinity column. Peak I contains Sup35 in both monomer and aggregated states. Sup35 aggregate is abbreviated as C-aggregate and includes a non-fibril complex comprising Sup35 aggregate-HSP90-Dna K, ATP synthase β unit (chain D), 30S ribosome subunit, and Omp F. The purified monomer and C-aggregate can remain stable for an extended period of time. Peak II contains Sup35 also in both monomer and aggregated (abbreviated as S-aggregate) states, but the aggregated states are caused by the formation of inter-Sup35 disulfide bonds. This study demonstrates that further assembly of Sup35 non-fibril C-aggregate can be interrupted by the chaperone repertoire system in E. coli.