Protein Structure Modeling Research Articles

Morphological and molecular phylogenetic characterization of Sarcocystis kani sp. nov. and other novel, closely related Sarcocystis spp. infecting small mammals and colubrid snakes in Asia

We investigated the morphology and phylogenetic relationships of novel and previously recognized Sarcocystis spp. infecting small mammals and colubrid snakes in Asia. The nuclear 18S rRNA and mitochondrial cox1 of Sarcocystis sp.1 from mangrove snakes (Boiga dendrophila) in Thailand and Sarcocystis sp.2 from a ricefield rat (Rattus argentiventer) in Sumatra were partially sequenced. Sporocysts of Sarcocystis sp.1 induced development of sarcocysts in experimentally infected rats, which showed a unique ultrastructure that was observed previously by S.P. Kan in rats from Malaysia; therefore, we describe this species as Sarcocystis kani sp. nov. Its integration into the 18S rRNA phylogeny of Sarcocystis spp. cycling between small mammals and colubrid snakes helped clarify relationships among the so-called S. zuoi-complex of molecularly cryptic species: Sarcocystis kani sp. nov., S. sp.2, S. attenuati, S. scandentiborneensis, and S. zuoi were all included in this clade. Tree topology was resolved into dichotomies congruent with the morphological disparities between the taxa. However, cox1 gene sequencing (including newly sequenced S. singaporensis and S. zamani) revealed that Sarcocystis kani, S. attenuati, and S. scandentiborneensis were identical suggesting a recent, common ancestry. To identify other distinctive features, lineage-specific molecular patterns within both genes were examined revealing that all 18S rRNA sequences of the S. zuoi – complex possess a unique, 7-nt long motif in helix 38 of domain V7 that was different in S. clethrionomyelaphis which branched off basally from the complex. Three-dimensional homology modelling of COX1 protein structure identified amino acid substitutions within the barcode area specific for the S. zuoi-complex and substantial divergence in structurally important amino acids between Sarcocystis species of snakes as definitive hosts and other lineages of the Sarcocystidae. We discuss the utility of selected genes for species delimitation of the Sarcocystis spp. under investigation, which probably evolved during recent radiations of their intermediate and definitive hosts.

Phosphate starvation regulates cellulose synthesis to modify root growth.

In the model plant Arabidopsis (Arabidopsis thaliana), the absence of the essential macro-nutrient phosphate reduces primary root growth through decreased cell division and elongation, requiring alterations to the polysaccharide-rich cell wall surrounding the cells. Despite its importance, the regulation of cell wall synthesis in response to low phosphate levels is not well understood. In this study, we show that plants increase cellulose synthesis in roots under limiting phosphate conditions, which leads to changes in the thickness and structure of the cell wall. These changes contribute to the reduced growth of primary roots in low-phosphate conditions. Furthermore, we found that the cellulose synthase complex (CSC) activity at the plasma membrane increases during phosphate deficiency. Moreover, we show that this increase in the activity of the CSC is likely due to alterations in the phosphorylation status of cellulose synthases in low-phosphate conditions. Specifically, phosphorylation of CELLULOSE SYNTHASE 1 (CESA1) at the S688 site decreases in low-phosphate conditions. Phosphomimic versions of CESA1 with an S688E mutation showed significantly reduced cellulose induction and primary root length changes in low-phosphate conditions. Protein structure modeling suggests that the phosphorylation status of S688 in CESA1 could play a role in stabilizing and activating the CSC. This mechanistic understanding of root growth regulation under limiting phosphate conditions provides potential strategies for changing root responses to soil phosphate content.

Deep generative models of LDLR protein structure to predict variant pathogenicity

The complex structure and function of low density lipoprotein receptor (LDLR) makes classification of protein-coding missense variants challenging. Deep generative models, including Evolutionary model of Variant Effect (EVE), Evolutionary Scale Modeling (ESM), and AlphaFold 2 (AF2), have enabled significant progress in the prediction of protein structure and function. ESM and EVE directly estimate the likelihood of a variant sequence but are purely data-driven and challenging to interpret. AF2 predicts LDLR structures, but variant effects are explicitly modeled by estimating changes in stability. We tested the effectiveness of these models for predicting variant pathogenicity compared to established methods. AF2 produced two distinct conformations based on a novel hinge mechanism. Within ESM’s hidden space, benign and pathogenic variants had different distributions. In EVE, these distributions were similar. EVE and ESM were comparable to Polyphen-2, SIFT, REVEL, and Primate AI for predicting binary classifications in ClinVar. However, they were more strongly correlated with experimental measures of LDL uptake. AF2 poorly performed in these tasks. Using the UK Biobank to compare association with clinical phenotypes, ESM and EVE were more strongly associated with serum LDL-C than Polyphen-2. ESM was able to identify variants with more extreme LDL-C levels than EVE and had a significantly stronger association with atherosclerotic cardiovascular disease. In conclusion, AF2 predicted LDLR structures do not accurately model variant pathogenicity. ESM and EVE are competitive with prior scoring methods for prediction based on binary classifications in ClinVar but are superior based on correlations with experimental assays and clinical phenotypes.

EasyModel: a user-friendly web-based interface based on MODELLER

Three-dimensional protein structures are invaluable sources of information for the functional annotation of protein molecules. Describing the function of a protein sequence is one of the most common problems in biology. Generally, this problem can be facilitated by studying the tertiary structure of proteins. In the lack of protein structures, comparative modeling often provides a useful three-dimensional model of the protein associated with at least one known protein structure. Comparative modeling predicts the tertiary structure of a certain protein sequence (target) mainly based on its homological sequence to the sequence of one or more proteins with known structures (templates). MODELLER is one of the most widely used tools for homology or comparative modeling of three-dimensional protein structures. However, most users find it challenging to start with MODELLER as it is a command line based and requires knowledge of basic Python scripting to use it efficiently. In this study, a web-based interface has been designed to predict the tertiary structure of proteins based on Modeller, which does the comparative modeling automatically, and uses PHP and Python programming languages. This tool is called “EasyModel” and is available at http://bioinf.modares.ac.ir/software/easymodel/. EasyModel provides a straightforward graphical interface for Modeller that can be used in only one browser.

Inferring mechanical properties of the SARS-CoV-2 virus particle with nano-indentation tests and numerical simulations

The pandemic caused by the SARS-CoV-2 virus has claimed more than 6.5 million lives worldwide. This global challenge has led to accelerated development of highly effective vaccines tied to their ability to elicit a sustained immune response. While numerous studies have focused primarily on the spike (S) protein, less is known about the interior of the virus. Here we propose a methodology that combines several experimental and simulation techniques to elucidate the internal structure and mechanical properties of the SARS-CoV-2 virus. The mechanical response of the virus was analyzed by nanoindentation tests using a novel flat indenter and evaluated in comparison to a conventional sharp tip indentation. The elastic properties of the viral membrane were estimated by analytical solutions, molecular dynamics (MD) simulations on a membrane patch and by a 3D Finite Element (FE)-beam model of the virion's spike protein and membrane molecular structure. The FE-based inverse engineering approach provided a reasonable reproduction of the mechanical response of the virus from the sharp tip indentation and was successfully verified against the flat tip indentation results. The elastic modulus of the viral membrane was estimated in the range of 7–20 MPa. MD simulations showed that the presence of proteins significantly reduces the fracture strength of the membrane patch. However, FE simulations revealed an overall high fracture strength of the virus, with a mechanical behavior similar to the highly ductile behavior of engineering metallic materials. The failure mechanics of the membrane during sharp tip indentation includes progressive damage combined with localized collapse of the membrane due to severe bending. Furthermore, the results support the hypothesis of a close association of the long membrane proteins (M) with membrane-bound hexagonally packed ribonucleoproteins (RNPs). Beyond improved understanding of coronavirus structure, the present findings offer a knowledge base for the development of novel prevention and treatment methods that are independent of the immune system.

Selection of mutant Listeria phages under food-relevant conditions can enhance application potential.

Bacteriophages are viruses that infect and kill bacteria. Currently, phage products are available for the control of the pathogen Listeria monocytogenes in food products in the United States. In this study, we explore whether experimental evolution can be used to generate phages with improved abilities to function under specific food-relevant conditions. Ultra-pasteurized oat and whole milk were chosen as test matrices as they represent different food groups, yet have similar physical traits and macronutrient composition. We showed that (i) wild-type phage LP-125 infection kinetics are different in the two matrices and (ii) LP-125 has a significantly higher burst size in oat milk. From this, we attempted to evolve LP-125 to have improved infection kinetics in whole milk. Ancestral LP-125 was passaged through 10 rounds of amplification in milk conditions. Plaque-purified DNA samples from milk-selected phages were isolated and sequenced, and mutations present in the isolated phages were identified. We found two nonsynonymous substitutions in LP125_108 and LP125_112 genes, which encode putative baseplate-associated glycerophosphoryl diester phosphodiesterase and baseplate protein, respectively. Protein structural modeling showed that the substituted amino acids in the mutant phages are predicted to localize to surface-exposed helices on the corresponding structures, which might affect the surface charge of proteins and their interaction with the bacterial cell. The phage containing the LP125_112 mutation adsorbed significantly faster than the ancestral phage in both oat and whole milk. Follow-up experiments suggest that fat content may be a key factor for the expression of the phenotype of this mutation. IMPORTANCE Bacteriophages are one of the tools available to control the foodborne pathogen, Listeria monocytogenes. Phage products must work under a broad range of food conditions to be an effective control for L. monocytogenes. Here, we show that the experimental evolution of phages can be used to generate new phages with phenotypes useful under specific conditions. We used this approach to select for a mutant phage that more efficiently binds to L. monocytogenes that is grown in whole milk and oat milk. We show that the fat content of these milks is necessary for the expression of this phenotype. Our findings show that experimental evolution can be used to select for improved phages with better performance under specific conditions. This approach has the potential to support the development of condition-specific phage-based biocontrols in the food industry.

Identification of A Novel Variant of Filamin A Destroying the Attraction Between Benzene Rings and Sulfhydryl in Developmental Dysplasia of the Hip.

Developmental dysplasia of the hip (DDH), characterized by acetabular deformity that manifests from loose ligaments to complete dislocation of the hip, can cause notable pain and dysfunction and lead to hip dislocation, secondary fractures, scoliosis, and osteoarthritis of hip. Variants in FLNA may produce a spectrum of malformations in multiple organs, especially the skeleton. This study aimed to identify the genetic etiologies of DDH patients and provide genetic testing information for further diagnosis and treatment of DDH. We recruited a Chinese woman with DDH and her family members. Whole-exome sequencing was used to identify the patient's genetic etiologies. Protein models were used to analyze the pathogenic mechanism of the identified variants. A novel variant (c.3493T>G, p.C1165G) of FLNA was detected. The structural models of the mutant FLNA protein indicated that the variant would lose its sulfhydryl side chain and destroy the attraction between benzene rings and sulfhydryl. We reported a novel variant (c.3493T>G, p.C1165G) of FLNA in a Chinese woman with DDH. Our research outcome enriches the gene pool for hip dysplasia and emphasizes the pathogenicity of sulfhydryl side chain disruption in FLNA.

ECL 3.0: a sensitive peptide identification tool for cross-linking mass spectrometry data analysis

BackgroundCross-linking mass spectrometry (XL-MS) is a powerful technique for detecting protein–protein interactions (PPIs) and modeling protein structures in a high-throughput manner. In XL-MS experiments, proteins are cross-linked by a chemical reagent (namely cross-linker), fragmented, and then fed into a tandem mass spectrum (MS/MS). Cross-linkers are either cleavable or non-cleavable, and each type requires distinct data analysis tools. However, both types of cross-linkers suffer from imbalanced fragmentation efficiency, resulting in a large number of unidentifiable spectra that hinder the discovery of PPIs and protein conformations. To address this challenge, researchers have sought to improve the sensitivity of XL-MS through invention of novel cross-linking reagents, optimization of sample preparation protocols, and development of data analysis algorithms. One promising approach to developing new data analysis methods is to apply a protein feedback mechanism in the analysis. It has significantly improved the sensitivity of analysis methods in the cleavable cross-linking data. The application of the protein feedback mechanism to the analysis of non-cleavable cross-linking data is expected to have an even greater impact because the majority of XL-MS experiments currently employs non-cleavable cross-linkers.ResultsIn this study, we applied the protein feedback mechanism to the analysis of both non-cleavable and cleavable cross-linking data and observed a substantial improvement in cross-link spectrum matches (CSMs) compared to conventional methods. Furthermore, we developed a new software program, ECL 3.0, that integrates two algorithms and includes a user-friendly graphical interface to facilitate wider applications of this new program.ConclusionsECL 3.0 source code is available at https://github.com/yuweichuan/ECL-PF.git. A quick tutorial is available at https://youtu.be/PpZgbi8V2xI.

Purifying selection leads to low protein diversity of the mitochondrial cyt b gene in avian malaria parasites

BackgroundMitochondrial respiration plays a central role in the survival of many eukaryotes, including apicomplexan parasites. A 479-bp fragment from the mitochondrial cytochrome b gene is widely used as a barcode to identify genetic lineages of avian malaria parasites Plasmodium and related haemosporidians. Here we looked for evidence of selection in the avian Plasmodium cyt b gene, using tests of selection and protein structure modeling. We also tested for the association between cyt b polymorphism and the host specificity of these parasites.ResultsBased on 1,089 lineages retrieved from the Malavi database, we found that the frequency of the most conserved amino acids in most sites was more than 90%, indicating that the protein diversity of the avian Plasmodium cyt b barcode was low. The exceptions were four amino acid sites that were highly polymorphic, though the substitutions had only slight functional impacts on the encoded proteins. The selection analyses revealed that avian Plasmodium cyt b was under strong purifying selection, and no positively selected sites were detected. Besides, lineages with a wide host range tend to share cyt b protein haplotypes.ConclusionsOur research indicates that purifying selection is the dominant force in the evolution of the avian Plasmodium cyt b lineages and leads to its low diversity at the protein level. Host specificity may also play a role in shaping the low mitochondrial diversity in the evolution of avian malaria parasites. Our results highlight the importance of considering selection pressure on the cyt b barcode region and lay a foundation for further understanding the evolutionary pattern of mitochondrial genes in avian malaria.

Exploring Proteus mirabilis Methionine tRNA Synthetase Active Site: Homology Model Construction, Molecular Dynamics, Pharmacophore and Docking Validation.

Currently, the treatment of Proteus mirabilis infections is considered to be complicated as the organism has become resistant to numerous antibiotic classes. Therefore, new inhibitors should be developed, targeting bacterial molecular functions. Methionine tRNA synthetase (MetRS), a member of the aminoacyl-tRNA synthetase family, is essential for protein biosynthesis offering a promising target for novel antibiotics discovery. In the context of computer-aided drug design (CADD), the current research presents the construction and analysis of a comparative homology model for P. mirabilis MetRS, enabling development of novel inhibitors with greater selectivity. Molecular Operating Environment (MOE) software was used to build a homology model for P. mirabilis MetRS using Escherichia coli MetRS as a template. The model was evaluated, and the active site of the target protein predicted from its sequence using conservation analysis. Molecular dynamic simulations were performed to evaluate the stability of the modeled protein structure. In order to evaluate the predicted active site interactions, methionine (the natural substrate of MetRS) and several inhibitors of bacterial MetRS were docked into the constructed model using MOE. After validation of the model, pharmacophore-based virtual screening for a systemically prepared dataset of compounds was performed to prove the feasibility of the proposed model, identifying possible parent compounds for further development of MetRS inhibitors against P. mirabilis.

Structural profile of ligand-based inhibition of bacterial FtsZ

Aim. The idea of the study was to compare and generalize RCSB Protein Data Bank and ChEMBL data in order to establish the structural and biological relationship of experimentaly proved effectors of FtsZ with binding sites. Methods. Literature and database search. Comparison of protein and ligand structures. Protein structure modeling, MD, structural superimposition, etc. Results. The experimental protein-ligand complexes structures of bacterial FtsZ were revised. The structural superimposition of experinental PDB and full-atomic AlphaFold2 models of bacterial FtsZs confirmed their significant structural similarity. Three protein-ligand binding sites were identified by structural alignment. The rating based on database (RCSB Protein Data Bank, ChEMBL, DrugBank, BindingDB, PubChem), patente and literature information on FtsZ-ligand interactions identify perspective sites and main reference compounds. Сonclusions. It was identifyd 3 main protein-ligand binding regions in FtsZ: I. Nucleotide Binding Domain (Ia. Site of GTP/GDP and Ib. MB3 site); II. Site of inter-domain cleft (IDC) and III. Site of coumarin bindig (4HC = 4-hydroxycoumarin). It was indicated that benzamide-binding site, located in the region of inter-domain cleft of FtsZ, demonstrate highest site- and target-specificity.

Differences in amino acid composition of carrot α-tubulin potentially confer the resistance to dinitroaniline herbicides

Aim. To reveal the features of amino acid composition of carrot α-tubulin isotypes that potentially determine natural tolerance to dinitroaniline herbicides. Methods. Literature and database search. Comparison of protein sequences and structures: multiple sequence alignment, phylogenetic profiling, protein and ligand structure modeling, etc. Results. Genomic and proteomic analysis of Daucus carota has revealed at least eight unique isotypes of α-tubulin that differ in amino acid sequences and gene loci. Remarkable differences in amino acid composition of the dinitroanilinebinding-like (DBL) region of analyzed α-tubulin have been revealed, which may be the reason of its natural resistance to these compounds. Сonclusions. Differences in amino acids at positions of canonical mutations – Cys4 (TBA1, 2, 3, 6, 7 and 8), Thr53 (TBA6), Ile202 (TBA1 and 7) and Met202 (TBA5), as well as previously undescribed non-canonical substitutions – Ile4 (TBA4 and 5), Cys52 (TBA6), Ser201 (TBA1, 2, 3 and 8) and Val194 (TBA4 and 5), were noted as potentially associated with natural tolerance of the carrot to dinitroaniline herbicides.

Molecular modelling and site-directed mutagenesis provide insight into saccharide pyruvylation by the Paenibacillus alvei CsaB enzyme

Pyruvylation is a biologically versatile but mechanistically unexplored saccharide modification. 4,6-Ketal pyruvylated N-acetylmannosamine within bacterial secondary cell wall polymers serves as a cell wall anchoring epitope for proteins possessing a terminal S-layer homology domain trimer. The pyruvyltransferase CsaB from Paenibacillus alvei served as a model to investigate the structural basis of the pyruvyltransfer reaction by a combination of molecular modelling and site-directed mutagenesis together with an enzyme assay using phosphoenolpyruvate (PEP; donor) and synthetic β-D-ManNAc-(1 → 4)-α-D-GlcNAc-diphosphoryl-11-phenoxyundecyl (acceptor). CsaB protein structure modelling was done using Phyre2 and I-Tasser based on the partial crystal structure of the Schizosaccharomyces pombe pyruvyltransferase Pvg1p and by AlphaFold. The models informed the construction of twelve CsaB mutants targeted at plausible PEP and acceptor binding sites and KM and kcat values were determined to evaluate the mutants, indicating the importance of a loop region for catalysis. R148, H308 and K328 were found to be critical to PEP binding and insight into acceptor binding was obtained from an analysis of Y14 and F16 mutants, confirming the modelled binding sites and interactions predicted using Molecular Operating Environment. These data lay the basis for future mechanistic studies of saccharide pyruvylation as a novel target for interference with bacterial cell wall assembly.

Application of conformational space annealing to the protein structure modeling using cryo-EM maps.

Conformational space annealing (CSA), a global optimization method, has been applied to various protein structure modeling tasks. In this paper, we applied CSA to the cryo-EM structure modeling task by combining the python subroutine of CSA (PyCSA) and the fast relax (FastRelax) protocol of PyRosetta. Refinement of initial structures generated from two methods, rigid fitting of predicted structures to the Cryo-EM map and de novo protein modeling by tracing the Cryo-EM map, was performed by CSA. In the refinement of the rigid-fitted structures, the final models showed that CSA can generate reliable atomic structures of proteins, even when large movements of protein domains were required. In the de novo modeling case, although the overall structural qualities of the final models were rather dependent on the initial models, the final models generated by CSA showed improved MolProbity scores and cross-correlation coefficients to the maps. These results suggest that CSA can accomplish flexible fitting and refinement together by sampling diverse conformations effectively and thus can be utilized for cryo-EM structure modeling.

Modeling orientational features via geometric algebra for 3D protein coordinates prediction

Protein structure prediction (PSP) is the prediction of the three‐dimensional (3D) folding of a protein (its tertiary structure) starting from its amino acid sequence (its primary structure). The state of the art in PSP is achieved by deep learning pipelines that require several input features extracted from amino acid sequences. It has been demonstrated that features that grasp the relative orientation of amino acids positively impact the prediction accuracy of the 3D coordinates of atoms in the protein backbone. In this paper, we demonstrate the relevance of geometric algebra (GA) in instantiating orientational features for PSP problems. We do so by proposing two novel GA‐based metrics which contain information on relative orientations of amino acid residues. We then employ these metrics as additional input features to a graph transformer (GT) architecture to aid the prediction of the 3D coordinates of a protein, and compare them to classical angle‐based metrics. We show how our GA features yield comparable results to angle maps in terms of accuracy of the predicted coordinates. This is despite being constructed from less initial information about the protein backbone. The features are also fewer and more informative and can be (i) closely associated to protein secondary structures and (ii) more easily predicted compared to angle maps. We hence deduce that GA can be employed as a tool to simplify the modeling of protein structures and pack orientational information in a more natural and meaningful way.

Enhancing cryo-EM maps with 3D deep generative networks for assisting protein structure modeling.

The tertiary structures of an increasing number of biological macromolecules have been determined using cryo-electron microscopy (cryo-EM). However, there are still many cases where the resolution is not high enough to model the molecular structures with standard computational tools. If the resolution obtained is near the empirical borderline (3-4.5 Å), improvement in the map quality facilitates structure modeling. We report EM-GAN, a novel approach that modifies an input cryo-EM map to assist protein structure modeling. The method uses a 3D generative adversarial network (GAN) that has been trained on high- and low-resolution density maps to learn the density patterns, and modifies the input map to enhance its suitability for modeling. The method was tested extensively on a dataset of 65 EM maps in the resolution range of 3-6 Å and showed substantial improvements in structure modeling using popular protein structure modeling tools. https://github.com/kiharalab/EM-GAN, Google Colab: https://tinyurl.com/3ccxpttx.

Molecular Docking Study of Chalcone Analogue Compounds with Hydroxy and Methoxy Subtituents as Bcl-2 Inhibitors

Molecular docking study of 6 chalcone analogues with protein target from the crystallographic structure modeling of Bcl-2 protein with PDB code 2W3L was carried out using computational media using the Molecular Operating Environment (MOE) program. The aim of this study is to determine the potentiality of the 6 chalcone analogue compounds as Bcl-2 inhibitors using molecular docking studies. In this study, venetoclax was used as positive control. Based on docking results, binding free energy was used as information to know which wheather chalcone analogue compounds are active or not as Bcl-2 inhibitors. According to the docking results that have been carried out, it showed that the 6 chalcone analogue compounds have no potential as Bcl-2 inhibitors. Due to the superimposition of the 6 compounds that did not stick to the positive control and most importantly the binding free energy values ​​(S) of the 6 chalcone analogue compounds were higher than the binding free energy values ​​of the positive control (Venetoclax).

VoroIF-GNN: Voronoi tessellation-derived protein-protein interface assessment using a graph neural network.

We present VoroIF-GNN (Voronoi InterFace Graph Neural Network), a novel method for assessing inter-subunit interfaces in a structural model of a protein-protein complex, relying solely on the input structure without any additional information. Given a multimeric protein structural model, we derive interface contacts from the Voronoi tessellation of atomic balls, construct a graph of those contacts, and predict the accuracy of every contact using an attention-based GNN. The contact-level predictions are then summarized to produce whole interface-level scores. VoroIF-GNN was blindly tested for its ability to estimate the accuracy of protein complexes during CASP15 and showed strong performance in selecting the best multimeric model out of many. The method implementation is freely available at https://kliment-olechnovic.github.io/voronota/expansion_js/.

CryoEM structure and Alphafold molecular modelling of a novel molluscan hemocyanin.

Hemocyanins are multimeric oxygen transport proteins present in the blood of arthropods and molluscs, containing up to 8 oxygen-binding functional units per monomer. In molluscs, hemocyanins are assembled in decamer 'building blocks' formed of 5 dimer 'plates', routinely forming didecamer or higher-order assemblies with d5 or c5 symmetry. Here we describe the cryoEM structures of the didecamer (20-mer) and tridecamer (30-mer) forms of a novel hemocyanin from the slipper limpet Crepidula fornicata (SLH) at 7.0 and 4.7 Å resolution respectively. We show that two decamers assemble in a 'tail-tail' configuration, forming a partially capped cylinder, with an additional decamer adding on in 'head-tail' configuration to make the tridecamer. Analysis of SLH samples shows substantial heterogeneity, suggesting the presence of many higher-order multimers including tetra- and pentadecamers, formed by successive addition of decamers in head-tail configuration. Retrieval of sequence data for a full-length isoform of SLH enabled the use of Alphafold to produce a molecular model of SLH, which indicated the formation of dimer slabs with high similarity to those found in keyhole limpet hemocyanin. The fit of the molecular model to the cryoEM density was excellent, showing an overall structure where the final two functional units of the subunit (FU-g and FU-h) form the partial cap at one end of the decamer, and permitting analysis of the subunit interfaces governing the assembly of tail-tail and head-tail decamer interactions as well as potential sites for N-glycosylation. Our work contributes to the understanding of higher-order oligomer formation in molluscan hemocyanins and demonstrates the utility of Alphafold for building accurate structural models of large oligomeric proteins.

GraphGPSM: a global scoring model for protein structure using graph neural networks.

The scoring models used for protein structure modeling and ranking are mainly divided into unified field and protein-specific scoring functions. Although protein structure prediction has made tremendous progress since CASP14, the modeling accuracy still cannot meet the requirements to a certain extent. Especially, accurate modeling of multi-domain and orphan proteins remains a challenge. Therefore, an accurate and efficient protein scoring model should be developed urgently to guide the protein structure folding or ranking through deep learning. In this work, we propose a protein structure global scoring model based on equivariant graph neural network (EGNN), named GraphGPSM, to guide protein structure modeling and ranking. We construct an EGNN architecture, and a message passing mechanism is designed to update and transmit information between nodes and edges of the graph. Finally, the global score of the protein model is output through a multilayer perceptron. Residue-level ultrafast shape recognition is used to describe the relationship between residues and the overall structure topology, and distance and direction encoded by Gaussian radial basis functions are designed to represent the overall topology of the protein backbone. These two features are combined with Rosetta energy terms, backbone dihedral angles and inter-residue distance and orientations to represent the protein model and embedded into the nodes and edges of the graph neural network. The experimental results on the CASP13, CASP14 and CAMEO test sets show that the scores of our developed GraphGPSM have a strong correlation with the TM-score of the models, which are significantly better than those of the unified field score function REF2015 and the state-of-the-art local lDDT-based scoring models ModFOLD8, ProQ3D and DeepAccNet, etc. The modeling experimental results on 484 test proteins demonstrate that GraphGPSM can greatly improve the modeling accuracy. GraphGPSM is further used to model 35 orphan proteins and 57 multi-domain proteins. The results show that the average TM-score of the models predicted by GraphGPSM is 13.2 and 7.1% higher than that of the models predicted by AlphaFold2. GraphGPSM also participates in CASP15 and achieves competitive performance in global accuracy estimation.