This research identifies the deubiquitinase USP10 as a vital regulator of T cell development and a key driver of T-cell acute lymphoblastic leukemia (T-ALL). By analyzing mouse models and human clinical samples, the authors demonstrate that USP10 stabilizes the transcription factor SOX4, preventing its degradation to ensure the rapid proliferation of thymocytes. The study reveals a genetic circuit where MYC signaling induces USP10 expression, which in turn sustains the cellular expansion necessary for immune maturation or, when dysregulated, malignant transformation. Crucially, the authors found that pharmacological inhibition of USP10 effectively slows the progression of leukemia in mice. These findings position USP10 as a promising therapeutic target for treating aggressive blood cancers while offering new insights into the molecular mechanics of the immune system. References: * Zhang M, Wu H, Lin X, et al. MYC-induced USP10 stabilizes SOX4 to promote thymocyte proliferation and leukemia onset in mice[J]. Nature Communications, 2026.

This research established a comprehensive multi-omics atlas of the perinatal brain using Bama miniature pigs to bridge critical gaps in human neurodevelopmental knowledge. By analyzing transcriptomes and proteomes across various brain regions and time points, the study identified a pivotal developmental transition occurring between embryonic days 94 and 104. The authors demonstrated that pigs serve as a superior large-animal model compared to rodents, mirroring primate-specific processes like gliogenesis and synaptic refinement. Their findings revealed a "rise-and-fall" pattern of regional heterogeneity and synchronized molecular surges essential for functional brain maturation. Furthermore, the study integrated disease-risk genes, showing that perinatal pig brain organization transitions toward an adult-like state relevant to neurodevelopmental disorders. Ultimately, this resource provides a translational framework for decoding the molecular mechanisms of the birth transition and the origins of pediatric brain pathologies. References: * Wang Z, Chu F, Wang X, et al. Perinatal brain developmental transition revealed by transcriptomic and proteomic analyses of Bama miniature pigs[J]. Nature Communications, 2026.

This study introduces a comprehensive placental transcriptome reference developed through long-read RNA sequencing, uncovering a level of transcriptional complexity previously unrecognized in this organ. By identifying over 37,000 high-confidence isoforms, including thousands of previously unannotated transcripts, the researchers challenge the notion that the placenta is a genomic void. The authors demonstrate that this tissue-specific reference significantly reduces quantification uncertainty in existing datasets compared to generic adult-tissue annotations. Utilizing this refined resolution, the study reveals that specific placental isoforms—particularly of the CSH1 gene—mechanistically mediate the effects of gestational diabetes on newborn birth weight. These findings underscore the critical importance of isoform-level analysis for understanding maternal-fetal health and the developmental origins of disease. Overall, the resource provides a powerful framework for re-evaluating how environmental exposures and genetic factors shape pregnancy outcomes. References: * Bresnahan S T, Yong H E J, Nemani A, et al. Long-read assembly reveals vast transcriptional complexity in the placenta associated with metabolic and endocrine function[J]. Nature Communications, 2026.

This research explores how Staphylococcus aureus circumvents barriers to horizontal gene transfer to facilitate microbial evolution. While clonal complexes typically restrict the exchange of mobile genetic elements through Type I restriction-modification systems, the study identifies lateral transduction as a uniquely efficient mechanism for transferring chromosomal DNA across different lineages. Crucially, the authors discovered that certain immune-deficient bacteria act as gateways, accepting foreign genetic material and subsequently sharing it with other members of their group. These hsdR mutants are prevalent in nature due to an evolutionary trade-off where the risk of viral infection is balanced by the acquisition of advantageous traits like antibiotic resistance. Ultimately, these "promiscuous" recipient strains play a pivotal role in the emergence of novel virulence factors and the global spread of pathogenicity. References: * Figueroa W, Sabnis A, Ibarra-Chávez R, et al. Immune-deficient bacteria serve as gateways to genetic exchange and microbial evolution[J]. Nature Communications, 2026.

This article introduces GluePCA, a high-throughput method used to map how thousands of mutations affect the PYL1 hormone receptor, a molecular switch in plants. By analyzing over 40,000 measurements, the researchers discovered that nearly 90% of mutations alter the receptor’s response to chemicals, primarily through changes in protein stability. The study reveals a modular genetic architecture where different regions of the protein independently control sensitivity and activity levels. Notably, specific rare mutations were found to create innovative signaling behaviors, such as inverted or "band-stop" responses. Ultimately, the data illustrates the evolutionary flexibility of allosteric proteins and provides a scalable framework for engineering future biosensors. References: * Stammnitz M R, Lehner B. The genetic architecture of an allosteric hormone receptor[J]. bioRxiv, 2025: 2025.05. 30.656975.

Bacteria span Earth’s ecosystems, coupling ecological versatility with genome-architectural reconfiguration across shifting physicochemical conditions. Yet the genomic routes by which free-living lineages cross ecosystem boundaries, and the consequences for genome architecture, remain poorly understood. Here, we use comparative and evolutionary genomics to investigate a soil-to-sediment-to-freshwater transition in Limnocylindria, an abundant clade within the Chloroflexota phylum. Two sister families show contrasting strategies. CSP1-4 expands genomes through niche-specific gene acquisition, whereas Limnocylindraceae undergoes genome reduction and metabolic simplification—revealing alternative evolutionary routes to similar ecological outcomes. In Limnocylindraceae, the loss of key DNA glycosylases coincides with degradation of base excision repair and is consistent with a hypermutator state that may have accelerated genomic erosion during freshwater specialization, potentially facilitating ecological expansion. This reductive genome trajectory is associated with a freshwater-adapted lineage with unexpectedly high GC content, challenging canonical links between base composition and genome size. While mutational processes appear to dominate genome restructuring, proteome-level patterns suggest selection favoring carbon- and nitrogen-efficient amino acid usage, implying that adaptive refinement can emerge alongside primarily non-adaptive dynamics. Overall, our findings are consistent with mutation-driven genome reduction and proteome optimization acting in concert to support cross-ecosystem boundary crossing and freshwater specialization in a free-living Chloroflexota lineage. References: * Serra Moncadas L, Shakurova A, Hofer C, et al. Deep-branching Chloroflexota lineages illuminate the eco-evolutionary foundation of cross-ecosystem colonization[J]. Nature Communications, 2026.

Genome annotation currently requires performing dozens of molecular assays in hundreds of cell and tissue samples, an expensive endeavor which is impractical to replicate across all species and conditions of interest. Here, we introduce BioSeq2Seq, a deep learning framework that infers cell-line-specific molecular assays widely used for genome annotation by leveraging a tri-modal input: evolutionarily conserved DNA sequence features, together with cell-line-specific transcriptional activity and directionality captured by a single run-on sequencing assay. BioSeq2Seq enables flexible genome annotation tasks through parameterized configurations of input features and output targets, combined with gradient-guided architectural refinement for specific biological objectives. Our model demonstrates high accuracy across four downstream tasks, showing improvements of 14.27% in histone modification prediction, 2.50% in functional element identification, and 2.90% in gene expression prediction compared to state-of-the-art methods. In transcription factor binding site (TFBS) prediction, it maintains performance comparable to that of leading existing approaches. By achieving competitive performance across tasks with single-cell-line input data, BioSeq2Seq provides an efficient and low-cost alternative for genome annotation. References: * Zhang Z, Fan X, Zhong J, et al. An end-to-end generalizable deep learning framework to comprehensively analyze transcriptional regulation[J]. Nature Communications, 2026.

The research explores how extracellular vesicles (EVs) and their microRNA cargo serve as a critical communication bridge between the immune system and the brain to regulate social behavior. Scientists discovered that transferring blood-borne EVs from healthy mice into models with social deficits, such as immunodeficient Rag1-/- mice, successfully restored normal sociability levels. These EVs travel through the bloodstream, enter the brain, and target neurons in the prefrontal cortex, where they deliver specific miRNAs like miR-23a-3p. This molecular delivery system improves inhibitory synaptic signaling by regulating the protein PKCε, which in turn stabilizes GABAA receptors to normalize brain activity. Ultimately, the study identifies T cells as the primary source of these pro-social vesicles, suggesting that EV-based therapies could offer a non-canonical pathway for treating neurodevelopmental and psychiatric conditions involving social impairment. References: * Matoba K, Dohi E, Garcia P A, et al. Circulating extracellular vesicle microRNAs mediate immune modulation of social behavior in male mice[J]. Nature Communications, 2026.

The basal ganglia integrate cortical inputs with dopaminergic signals to potentiate and select actions. The reward-related activity of dopamine neurons is well-studied, but the coding properties of cortical inputs to the basal ganglia remain largely unknown. We examined the activity of neurons in the frontal eye field of monkeys that were optogenetically identified as projecting to the basal ganglia. We found that the projecting neurons contained information about expected rewards and selected actions. The reward-related signal and modulations independent of task condition were stronger in optogenetically identified projecting neurons than in other neurons in the same area. These findings indicate that reward, choice, and sensorimotor information are already integrated into the cortical inputs to the basal ganglia, suggesting that the basal ganglia network integrates reward from both cortical and dopaminergic inputs rather than relying on a dopaminergic source alone. References: * Hovav-Lixenberg A, Henshke Y, Kreisel T, et al. Enhanced reward coding and condition-independent dynamics in optogenetically identified corticostriatal neurons in monkeys[J]. Nature Communications, 2026.

This research outlines a comprehensive spatial tumor evolution panorama of high-grade serous ovarian cancer using the innovative STARLETS framework. By integrating whole-genome sequencing, single-cell RNA-seq, and spatial transcriptomics, the authors illustrate how hypoxia and immune pressure drive a Darwinian selection of malignant clones that eventually metastasize. The study identifies a unique tripartite ensemble consisting of SPP1+ macrophages, MMP11+ myCAFs, and epithelial cells that facilitate cancer spread within the peritoneal cavity and ascites. Experimental evidence shows that inhibiting the SPP1-CD44 axis can disaggregate these metastatic units and reduce tumor burden in vivo. Furthermore, the presence of SPP1+ macrophages serves as a vital biomarker for predicting patient responses to oncolytic virus and PARP inhibitor therapies. These findings offer deep insights into site-specific tumor-host dynamics and present new opportunities for targeted therapeutic intervention. References: * Feng C, Yang Y, Li G, et al. Spatial tumor evolution panorama of ovarian cancer[J]. Cell Reports Medicine, 2026, 7(3).

This research presents a detailed spatial transcriptomic atlas of small cell lung cancer (SCLC) to identify how the tumor microenvironment evolves during lymph node metastasis. By examining over 600,000 individual cells from 75 patients, the study identifies three specific malignant subclusters that appear during metastasis and exhibit unique metabolic and immune-evading properties. The authors describe a dynamic vascular-immune crosstalk, noting that immune cells within the vascular niche are functionally reprogrammed to become more cytotoxic as the disease spreads. Through cellular neighborhood analysis, the researchers characterized distinct multicellular structures, such as the pan-immune hotspot (PIHs-1), which serves as a significant predictor of patient survival. Ultimately, these findings reveal how the spatial organization of cancer and immune cells contributes to tumor progression and provides new prognostic biomarkers for clinical use. These insights offer a high-resolution map of the biological architectures that facilitate SCLC growth and adaptation within the lymphatic system. References: * Zhang Z, Wu D, Chen R, et al. Single-cell spatial transcriptomics reveals tumor microenvironment heterogeneity in primary and lymph node-metastatic small cell lung cancer[J]. Cell Reports Medicine, 2026.

The paper introduces SMART, an unsupervised deep learning framework designed to integrate spatial multi-omics data into a unified representation. By combining graph neural networks with metric learning, the model effectively captures both local spatial contexts and long-distance biological similarities across various omics layers. The researchers developed a variant called SMART-MS to align data across multiple tissue sections while correcting for batch effects. Benchmarking demonstrates that SMART offers superior computational efficiency, scalability, and accuracy in identifying anatomical structures compared to existing methods. Ultimately, this tool provides a versatile solution for analyzing complex tissue microenvironments at various spatial resolutions. References: * Du Z, Chen Q, Huang W, et al. SMART: spatial multi-omic aggregation using graph neural networks and metric learning[J]. Nature Communications, 2026, 17(1): 2876.

They present PocketXMol, an atom-level model that unifies generative tasks related to protein pocket interactions. Using atomic prompts as task specifications, PocketXMol supports various molecular tasks, including structure prediction and de novo design of small molecules and peptides, without task-specific fine-tuning. PocketXMol achieved strong performance on 11 of 13 computational benchmarks and remained competitive on the remaining two, outperforming 55 baseline models. We applied PocketXMol to design caspase-9-inhibiting small molecules, achieving efficacy comparable with commercial pan-caspase inhibitors. We also adopted PocketXMol to generate PD-L1-binding peptides, resulting in a success rate that largely exceeds library screening. Three representative peptides underwent further experiments, which validated their cellular specificity and confirmed their potential for molecular probing and therapeutics. PocketXMol provides a general platform for AI-aided drug discovery and enables a wide range of future applications. References: * Peng X, Guo R, Guo F, et al. Unified modeling of 3D molecular generation via atomic interactions with PocketXMol[J]. Cell, 2026.