Urea, the world’s predominant nitrogen fertilizer, has supported human population growth for the past 60 years, yet its effects on freshwater ecosystems are largely unknown. Here urea additions at ecologically relevant rates tripled summer phytoplankton abundance in replicate agricultural reservoirs with significant responses by most eukaryotic algae, but not cyanobacteria or their toxins. Mass budgets reveal that fertilized reservoirs did not become limited by phosphorus due to its continuous release from sediments. Further, most added nitrogen did not accumulate in reservoirs but was lost to the atmosphere, probably as NH3. Sub-continental spatial analysis shows that study reservoirs are characteristic of shallow water bodies within Canada’s largest agricultural region, where >40% of surface waters are vulnerable to degradation by urea. Similar degrees of water quality loss by urea are expected in other global agricultural regions (for example, China, India, North America) where elevated urea use interacts with phosphorus-rich surface waters to induce extreme eutrophication. Urea, the most widely used nitrogen fertilizer, has unknown impacts on freshwater ecosystems. This study demonstrates that urea additions in Canadian prairie agricultural reservoirs triple phytoplankton abundance without increasing cyanobacterial toxins, revealing considerable nitrogen loss to the atmosphere and highlighting potential global water quality degradation in phosphorus-rich agricultural regions.

Transcription factors (TFs) regulate gene expression by binding to specific DNA sites on genome, making accurate TF binding site prediction critical for understanding gene regulation and downstream phenotypes. Almost all current deep learning methods use only DNA-related information to predict TF binding sites, ignoring the fact that different TF protein sequences and structures recognize distinct DNA patterns. Not leveraging TF information not only limits prediction accuracy but also makes the methods not generalizable to predicting binding sites of new TFs that do not exist in the training data. Here, we present TransBind, a protein-aware deep learning architecture that integrates DNA sequence information with protein embeddings containing both sequence and structural information derived from a protein language model pretrained on DNA-binding proteins, to improve TF binding site prediction. Through the cross-attention, a TF embedding selectively attends to genomic regions according to its unique binding properties. Evaluated on the data of 690 ChIP-seq experiments spanning 161 TFs across 91 human cell types, TransBind achieves an AUROC of 0.9508 and AUPR of 0.3741—representing a 11.8% relative AUPR improvement over state-of-the-art methods including TBiNet, EPBDXDNABERT-2, DanQ, and DeepSEA. The model outperformed existing methods in 98% of TF–cell type combinations. It also recovered 160 known TF binding motifs in the JASPAR database, providing the biological interpretability of the model. Moreover, the approach enables label-zero-shot prediction for unseen TFs, demonstrating its potential of generalizing to new, poorly characterized TFs. The source code of TransBind is available at https://github.com/jianlin-cheng/TransBind. The version used in this work is archived at https://doi.org/10.5281/zenodo.19462292.

Optogenetic tools—light-responsive proteins that enable to regulate specific cellular activities, study biological processes, and develop new therapies—are attractive approaches for achieving endogenous gene regulation under minimally invasive conditions. Our first step in constructing an optogenetic system to regulate endogenous Drosophila gene expression was to identify inhibitory anti-CRISPR (Acr) proteins that block CRISPRa-mediated activation. Next, we inserted optogenetic protein LOV2 into these Acrs, tested for their ability to optogenetically modulate endogenous gene upregulation through the CRISPRa-based flySAM system in Drosophila, and found that the photoswitchability of these prototypes was weak. We therefore engineered an optimized Acr–LOV2 fusion module by refining length of intrinsically disordered and ordered regions (IDR and IOR) of Acrs. This optimization yielded a variant with significantly greater sensitivity to blue-light-induced endogenous gene upregulation than the prototypes, leading to new in vivo discoveries. In addition, this work provides insights for in vivo functional characterization of the IDR and the IOR of these small-sized proteins. Together, these findings establish a robust optogenetic toolbox for precise, light-controlled endogenous gene regulation in Drosophila.

Despite considerable powers, the application of CRISPR activation (CRISPRa) screens in primary human cells remains a formidable challenge. Here, we develop dCas12f1-SAM, a compact SAM-based transcriptional activation platform, that outperforms existing systems in both immortalized cell lines and primary human T cells and hematopoietic stem/progenitor cells (HSPCs). Using dCas12f1-SAM, we perform a pooled CRISPRa screen targeting 1559 human transcription factors (TFs) in primary human T cells and identify multiple positive regulators of IL-2 expression. We further implement a single-cell CRISPRa screen via our miCROP-seq construct, resolving how these genetic perturbations reshape T cell activation dynamics and drive functionally distinct cellular states. Among the top-ranking genes, we spotlight KLF12 and LHX5, whose overexpression significantly improves antigen-specific responses of chimeric antigen receptor T (CAR-T) cells. Collectively, these findings establish dCas12f1-SAM as a robust transcriptional activation tool, highlighting its potential to advance applications in cellular engineering and immunotherapy. CRISPRa screening in primary human cells has remained a formidable technical challenge. Here, the authors developed dCas12f1-SAM and successfully implemented a gain-of-function screen in primary human T cells, identifying key regulators of T cell activation.

The clinical success of gene therapy depends critically on the development of delivery platforms capable of achieving precise, efficient and tissue-specific delivery of genetic payloads in vivo. A diverse array of carriers, including viral, non-viral, synthetic and natural vectors, have been explored to address this challenge. Among them, adeno-associated viruses, lipid nanoparticles and extracellular vesicles have emerged as leading candidates, each offering distinct advantages and translational hurdles. Here we provide a comparative analysis of these delivery modalities, highlighting their respective design principles, targeting capabilities, immunogenicity profiles and clinical progress. We survey preclinical and clinically adopted delivery strategies and explore how the three delivery platforms can be tailored for gene therapeutics in different diseases. Finally, we discuss emerging strategies to overcome current limitations and outline future directions for the rational design of next-generation gene delivery platforms that combine safety, scalability and functional precision. This Review offers a comparative analysis of platforms for the tissue-specific delivery of genetic payloads, discusses how carriers such as adeno-associated viruses, lipid nanoparticles and extracellular vesicles can be tailored for use in different diseases, and charts a path for engineering improved platforms for clinical use.

Nanoplastics (NPs), as persistent and unregulated aquatic contaminants, evade conventional removal technologies, posing severe health risks and urgently demanding innovative, effective solutions. Here we introduce recyclable magnetic biohybrid nanonets (LAF-IONPs), engineered through in situ growth of iron oxide nanoparticles (IONPs) on lysozyme amyloid fibrils (LAFs). Stabilized by synergistic interfacial interactions, LAF-IONPs efficiently capture NPs over a wide range of sizes (30–1,000 nm) and chemical compositions across environmentally relevant concentrations and conditions (pH 7–9, high salinity and co-existing pollutants). The superior removal efficiency arises from magnetic active motion and abundant fibril binding sites. Remarkably, LAF-IONPs achieved 98.0–99.9% NPs removal from various real water samples and maintained >95% efficiency over 100 recycling cycles when using a custom alternating magnetic field system. Critically, LAF-IONPs treatment reduced in vivo NPs bioaccumulation by 91.5%. This work establishes a blueprint for designing recyclable biohybrid adsorbents for active, efficient and sustainable removal of NPs and potentially other emerging contaminants. Nanoplastics are persistent water contaminants that evade conventional removal methods. Magnetic biohybrid nanonets are found to capture up to 99.9% of them, remain highly recyclable and greatly reduce in vivo accumulation.

Elucidating protein–ligand interactions is pivotal for understanding biological mechanisms and accelerating drug discovery. Blind docking, which identifies binding sites without prior knowledge, has become an indispensable computational strategy for analyzing the surge of protein structures generated by Cryo-EM and AI-based prediction tools like AlphaFold3. Our previous server, CB-Dock2, has been widely adopted by the global research community, averaging over 1000 daily submissions since July 2022 due to its accuracy and user-friendliness. Building on this foundation and incorporating extensive user feedback, we present CB-Dock3, a substantially enhanced platform. Key upgrades include a refined docking engine, an expanded template library, and support for diverse file formats. Benchmark evaluations on CASF-2016 demonstrate that CB-Dock3 achieves a success rate of 67.4% (RMSD ≤ 2.0 Å), representing a 10.6 percentage-point absolute improvement over its predecessor and outperforming other popular blind docking tools. Additionally, CB-Dock3 introduces critical new features driven by community needs: support for user-defined docking regions to handle large complexes, and a metal-aware protocol that explicitly retains essential metal ions and cofactors during simulation.

Syntrophic interactions based on reciprocal metabolite exchange are widespread in microbial communities, yet the factors determining their stability remain unclear. Using synthetic E. coli consortia composed of lysine and arginine auxotrophs, we show that lower initial metabolite production promotes, rather than limits, syntrophic stability. During serial propagation, replicate cocultures diverged sharply: a minority maintained sustained growth, whereas most became extinct. This divergence was associated with phenotypic differences in metabolite production among founding isolates. Consortia founded by low-producing strains recovered reliably after dilution and were more resistant to invasion by non-producing mutants. By contrast, high-producing founder generated diminishing returns for consortium growth, and increased extracellular metabolite availability that favored exploitation by non-producer. Although we detected no consistent coding-region variations between high- and low-producing isolates, expression differences suggest that outside coding regions may influence these production traits. These results identify constrained initial metabolite production as a key determinant of syntrophic stability. Low initial metabolite production stabilizes syntrophic E. coli consortia by constraining metabolite availability, promoting localized exchange, and resisting invasion by non-producers