Wenyu Shi

Plant–virus protein–protein interactions (PPIs) play crucial roles in viral infection and host immune responses. However, large-scale experimental characterization of plant–virus PPIs is costly and labor-intensive, suggesting the need to develop computational methods. Recent advances in biological sequence language models are offering unprecedented opportunities to develop machine learning-based plant virus PPI predictors. In this study, we collected publicly available PPI data between Arabidopsis thaliana (A. thaliana) and viruses, establishing a robust dataset for method development. Based on this dataset, we developed AraVirusPPI, a machine learning framework for predicting A. thaliana virus PPIs. AraVirusPPI employs the protein language model ESM Cambrian (ESMC) to encode sequence features and combines these representations with Extreme Gradient Boosting (XGBoost) to build the prediction model. Furthermore, we systematically evaluated multiple protein and genomic language models combined with various classification algorithms, thereby justifying the optimal model architecture adopted in AraVirusPPI. We applied AraVirusPPI to predict PPIs in the A. thaliana–Cabbage leaf curl virus (CabLCV) system. Predicted viral target proteins were compared with experimental datasets from A. thaliana–Turnip mosaic virus (TuMV) and A. thaliana–Pseudomonas syringae (Psy) systems. Network topology and functional enrichment analyses showed that the predicted CabLCV targets in A. thaliana were more similar to TuMV than Psy targets. These results suggest potential shared and distinct host-targeting patterns among pathogens and provide computational evidence supporting the biological significance of the proposed AraVirusPPI predictor.

Dysglycaemia and periodontal inflammation frequently co-occur during pregnancy, but the microbial mechanisms linking these conditions and their potential for intervention remain incompletely understood. Here, we establish prospective pregnancy cohorts including more than 2500 volunteers and longitudinally profile oral microbiome dynamics in 534 pregnant women. We show that gestational diabetes mellitus (GDM) is associated with a progressive shift from Streptococcus-dominated oral microbiota to Prevotella/Porphyromonas-enriched dysbiosis. In mouse and cellular models, this dysbiotic oral microbiota induces periodontal inflammation, systemic IL-17 and IL-1β responses, suppression of glucagon-like peptide-1 and insulin, and exacerbation of hyperglycemia. Conversely, oral microbiota remodeling through transplantation of Streptococcus-dominated bacteria attenuates periodontal inflammation, restores glucagon-like peptide-1 and insulin levels, and improves glycaemic status in mice. Salivary metabolomics identifies docosahexaenoic acid (DHA) depletion in GDM, and in vitro assays show selective suppression of dysbiosis-associated oral pathogens by DHA. We therefore test topical gingival DHA in a double-blind randomized controlled trial of 40 pregnant women with GDM (ChiCTR2400080741), with probing depth and fasting blood glucose as primary endpoints and gingival index, attachment loss and plaque index as secondary endpoints. Daily gingival DHA application for six weeks improves probing depth and attenuates fasting glucose increase compared with placebo, with median fasting glucose changes from baseline of 0.10 versus 0.27 mmol/L. Together, these findings identify oral dysbiosis as a microbial driver of periodontal and glycaemic deterioration during pregnancy and support oral microbiome modulation as a potential adjunctive strategy for pregnancy care, although the clinical findings remain preliminary and require validation in larger trials with broader glycaemic endpoints.

Recycling thermoset polyurethanes is hindered by their cross-linked structures and chemically stable urethane bonds. Although chemo-enzymatic approaches offer promise, known urethanases remain inefficient under industrial glycolysis conditions. Here, we present GRASE [graph neural network (GNN)–based recommendation of active and stable enzymes], a GNN-based framework that integrates self-supervised and supervised learning to identify efficient, glycolysis-compatible urethanases. Among these, Ab PURase exhibited two orders of magnitude greater activity than previously known enzymes in 6 molar diethylene glycol, enabling near-complete depolymerization of commercial polyurethane at kilogram scale within 8 hours. Structural analysis revealed that a tightly packed hydrophobic core and proline-stabilized lid loop may confer Ab PURase’s stability and efficiency in harsh solvents. This work highlights how deep learning accelerates the discovery of biocatalysts with industrial potential and addresses a critical barrier in polyurethane recycling.

Ticks are arthropod vectors capable of transmitting a wide spectrum of pathogens affecting humans and animals. However, we have relatively limited information of their genomic characteristics and the diversity of associated microbiomes. Here we used long- and short-read sequencing on 1,479 samples from 48 tick species across eight genera from China to determine their genome and associated pathogens and microbiome. Through de novo assembly, we reconstructed 7,783 bacterial genomes representing 1,373 bacterial species, of which, 712 genomes represented 32 potentially pathogenic species. Computational analysis found nutritional endosymbionts to be prevalent and highly specific to tick genera. The microbiome genome-wide association study revealed host genetic variants linked to pathogen diversity, abundance and key biological pathways essential to tick biology, including blood-feeding and pathogen invasion. These findings provide a resource for studying the host–microbe interactions within ticks, paving the way for strategies to control tick populations and tick-borne diseases.

Three novel facultatively methylotrophic bacteria, strains 3C^T, 1A, 8P, were isolated from activated sludges. The isolates were aerobic, Gram-stain-negative, non-motile, non-spore forming rods multiplying by binary fission. The predominant polar lipids were phosphatidylcholine, phosphatidylglycerol, phosphatidylethylethanolamine, phosphatidylmonomethylethanolamine, and diphosphatidylglycerol. The major fatty acids of cells were С_18:1ω7c, C_19:0ω8c cyclo and C_16:0. Levels of 16S rRNA gene similarity indicates that the closely relatives are representatives of the genera Starkeya, Ancylobacter, Angulomicrobium and Methylorhabdus (96.4–99.4%). Genomic comparisons of 3C^T and its closest relatives, S. novella DSM 506^T and S. koreensis Jip08^T, shared 87.3 and 86.8% nucleotide identity and 28.3 and 26.8% digital DNA–DNA hybridization values, respectively. The average amino acid identities between the strain 3C^T and representatives of Starkeya, Ancylobacter and Angulomicrobium were in the range of 75.6–84.3%, which combines these strains into a single genus and gives rise to their reclassification. Based on polyphasic analyses, the strains 3C^T, 1A, 8P represents a novel species of the genus Ancylobacter, for which the name Ancylobacter moscoviensis sp. nov. is proposed. The type strain is 3C^T (= VKM B-3218^T = KCTC 62336^T). Furthermore, we also suggested the reclassification of Starkeya novella as Ancylobacter novellus comb. nov., Starkeya koreensis as Ancylobacter koreensis comb. nov., Angulomicrobium tetraedrale as Ancylobacter tetraedralis comb. nov., Angulomicrobium amanitiforme as Ancylobacter amanitiformis comb. nov. and Methylorhabdus multivorans as Ancylobacter multivorans comb. nov. with the emended description of the genus Ancylobacter.

Pan‐genomics is one of the most powerful means to study genomic variation and obtain a sketch of genes within a defined clade of species. Though there are a lot of computational tools to achieve this, an integrated framework to evaluate their performance and offer the best choice to users has never been achieved. To ease the process of large‐scale prokaryotic genome analysis, we introduce Integrated Prokaryotes Genome and pan‐genome Analysis (IPGA), a one‐stop web service to analyze, compare, and visualize pan‐genome as well as individual genomes, that rids users of installing any specific tools. IPGA features a scoring system that helps users to evaluate the reliability of pan‐genome profiles generated by different packages. Thus, IPGA can help users ascertain the profiling method that is most suitable for their data set for the following analysis. In addition, IPGA integrates several downstream comparative analysis and genome analysis modules to make users achieve diverse targets.

Two carbon dioxide-requiring, gliding, Gram-stain-negative strains, designated p1a2^T and 051621, were isolated from subgingival plaque in association with severe periodontitis. The 16S rRNA gene sequence analysis revealed that they represented members of the genus Capnocytophaga and had less than 96.4 % pairwise similarity with species with validly published names in this genus. The whole-genome sequences of those strains had less than 91.9 % average nucleotide identity and 48.4 % digital DNA–DNA hybridization values with the other type strains of species of the genus Capnocytophaga , both below the species delineation threshold. The results of pan-genomic analysis indicated that p1a2^T and 051621 shared 765 core gene families with the other ten species in this genus, and the numbers of strain-specific gene families were 493 and 455, respectively. The major fatty acids were iso-C_15 : 0 and C_16 : 0. A combination of phenotypic, chemotaxonomic, phylogenetic and genotypic data clearly indicate that p1a2^T and 051621 should be considered to represent a novel species of the genus Capnocytophaga , for which the name Capnocytophaga periodontitidis sp. nov. is proposed. The type strain is p1a2^T (=CGMCC 1.17337^T=JCM 34126^T).