Lymphatic vessels are essential for tissue homoeostasis and their growth is regulated by vascular endothelial growth factor C (VEGF-C) signalling through VEGFR3. However, how VEGF-C balances lymphatic endothelial cells (LECs) proliferation versus sprouting to ensure functional vessel formation has remained unclear. Using high-fidelity conditional genetics and receptor-specific ligands, we uncover a requirement for the alternative receptor VEGFR2 in VEGF-C-VEGFR3–driven lymphatic vessel sprouting. While activation of VEGFR2 alone fails to induce lymphangiogenesis, VEGFR2 loss abolishes LEC sprouting, but not proliferation, in response to VEGF-C. In contrast, deletion of the VEGFR3 downstream effector PI3Kα completely abrogates lymphangiogenesis. VEGFR2 is activated and found in proximity to VEGFR3 in LECs in vivo, with PI3Kα controlling their relative cell-surface availability and VEGF-C increasing VEGFR2 relative to VEGFR3, thereby priming LECs for sprouting. This receptor coordination balances VEGF-C-driven proliferative and sprouting responses, coupling LEC expansion to vessel growth, ensuring the formation of functional lymphatic networks.

The molecular drivers and fitness trade-offs controlling the transition of environmental bacteria into human pathogens remain enigmatic. Here, we describe a small RNA (sRNA) with a unique modular structure that shapes the evolution of toxigenic Vibrio cholerae, the agent of cholera. The sRNA comprises a variable 5' module located within the ompU ORF and a conserved 3' region downstream. This atypical location confers a distinct bimodular structure to the OmpU-encoded sRNA (OueS), generating allelic variants with unique functions. Unlike environmental counterparts, the OueS allele from toxigenic strains controls phenotypes essential during host colonization in a domain-dependent manner: a) inhibits biofilm formation by suppressing RyhB, b) confers resistance against intestinal bacteriophages by activating the CBASS system, and c) drives mucus penetration. Toxigenic OueS is essential for successful intestinal colonization and acts as a functional surrogate of the master virulence regulator ToxR, controlling over 84% of its regulome. However, this allele leads to reduced competitive fitness during colonization of natural reservoirs when compared to environmental ones. Finally, we determined that the acquisition of the ompU ORF is modular and controlled by exogenous conditions, providing critical insights into the evolution of toxigenic V. cholerae and the emergence of pathogenic traits in bacteria.

Controlling the intrinsic activity and long-term stability of active sites is essential to advance the formulation of catalysts. The hydrogenation of CO2 to methanol over indium oxide (In2O3) is believed to proceed at oxygen vacancies (VO∙∙) formed in situ. Here, we study how the structural dynamics of c-In2O3 are altered through doping with Sn or Zr, affecting the local structure, catalytic activity, and stability. We find that VO∙∙ sites in Sn-doped c-In2O3 are unreactive towards their replenishment by CO2, leading to catalyst deactivation by the formation of In0 and Sn0. Conversely, VO∙∙ sites in Zr-doped c-In2O3 show a high reactivity towards CO2, translating into a high catalytic activity and stability against over-reduction-induced deactivation. The diverging properties originate from the distinct defect dynamics in these two materials. The balance between VO∙∙ formation and its replenishment during CO2 hydrogenation is the key characteristic for both activity and stability of In2O3-based catalysts.

The ever-growing demand of soft strain sensors in ubiquitous electronics poses urgent need of device durability innovations to reduce resource waste. As most soft sensors use non-degradable materials, it is important to ensure durable sensor performance with reliable sensing signals over its lifecycle to reduce device waste, yet facing pervasive signal predicaments of nonlinearity, hysteresis, cycling attenuation, and batch inconsistency. Herein, instead of empirical experiments via material or structural engineering, we propose an efficient computational calibration framework to comprehensively address these signal predicaments by utilizing hierarchical physics-guided machine learning (ML) models. Using an eco-friendly carbon waste-based strain sensor as a case study, physics-guided ML models are developed and show both high computation efficiency and learning accuracy in terms of real-time sensing signal calibration of the developed sensor, which automatically calibrates its unsatisfied sensing performance to equivalently linear, non-hysteresis, long-term stable, and batch consistent one without trial-and-error experiments. As final demonstrations, the ML-driven calibration models are capable of expanding sensor's reliable working lifetime for >3000 times than its own counterpart for multiple robotic tasks, facilitating their long-term usages during practical applications.

Little is known about how three-dimensional chromatin topology shapes mammalian craniofacial development. In mouse cranial neural crest cells, a Polycomb Repressive Complex 2 (PRC2)-dependent chromatin architecture is established before migration. This configuration maintains craniofacial gene promoters poised and connects them with distal Polycomb tethering elements, positioning promoters in spatial proximity to future long-range enhancers. Deletion of Ezh2 disrupts this early topology, causing inappropriate gene derepression in post-migratory craniofacial subpopulations where these genes are normally silenced, and failure of long-range enhancer recruitment where activation is required, thereby impairing proper gene expression. We further identify a distal Polycomb tethering element essential for Hoxa2 enhancer recruitment across topologically associating domains. Thus, Polycomb acts not only as a transcriptional repressor, but also as a chromatin-folding organizer that prepares developmental genes for later activation, by facilitating subsequent recruitment of distal active enhancers previously not in contact. Polycomb-mediated topology therefore orchestrates the transition from progenitor plasticity to precise spatiotemporal control of morphogenetic gene programs during neural crest development and face formation.

Sequence-based deep learning has advanced genome interpretation, yet most models remain task-specific and rely on retraining, limiting scalability across biological contexts. Here we present SUCCEED, a supervised multi-task DNA foundation model pretrained on 6,389 ENCODE functional genomics tracks to learn transferable regulatory representations. By integrating convolutional layers with a Transformer architecture, SUCCEED captures both local sequence motifs and long-range regulatory dependencies, achieving performance comparable to or exceeding Enformer across benchmark tasks. Through transfer learning, it predicts cell-type-specific epigenomic profiles, denoises sparse chromatin accessibility signals, and predicts three-dimensional chromatin contacts without CTCF input across data scales and cell types. Across diverse genomics tasks, SUCCEED performs comparably to supervised foundation models such as Sei and outperforms self-supervised models trained solely on DNA sequence. Overall, SUCCEED is a transferable and scalable foundation model that provides a unified framework for genome-scale regulatory modeling in complex biological contexts.

Eukaryotic DNA replication requires the precise assembly of MCM2-7 single hexamers (SHs) into head-to-head double hexamers (DHs) at replication origins. While DH formation is well-characterized in budding yeast, the underlying mechanisms in human cells remain poorly understood. Here, we report cryo-electron microscopy structures of endogenous human MCM2-7 SH isolated from G1-phase cells. In these structures, human MCM2-7 adopts a latched spiral conformation in an autoinhibited state where the carboxyl-terminal extension (CTE) of MCM5 occupies the central channel, and MCM3-CTE is capable of locking the MCM2-5 gate to occlude DNA entry. Systematic functional analysis demonstrates that the six CTEs of MCM2-7 play distinct roles in SH stability, MCM loading, and DH formation on chromatin. Surprisingly, unlike in yeast, the human MCM3-CTE is dispensable for cell viability but ensures efficient genome-wide replication initiation. Our findings establish how human MCM2-7 enables flexible yet precise MCM loading via its CTEs, providing a framework for understanding the regulation of DNA replication initiation in higher eukaryotes.

Harvesting hot carriers before they lose energy to the lattice is a critical route toward surpassing the conventional thermodynamic limit in optical-to-electrical (O-E) conversion. However, photocurrent from such hot carriers has remained challenging to directly detect because they equilibrate on picosecond timescales, outpacing conventional electronic measurement. Here, by employing terahertz electronics with sub-picosecond temporal resolution, we directly monitor ultrafast O-E conversion in black phosphorus (BP). Photoexcitation near the metal contact under zero source-drain bias generates an ultrafast photocurrent with a decay time of ~ 400 fs-orders of magnitude faster than the typical sub-nanosecond energy relaxation in BP-demonstrating a measured 3 dB bandwidth of 260 GHz with an intrinsic limit of ~ 600 GHz. Notably, this photocurrent flows via energetic holes toward the contact electrode, regardless of the equilibrium carrier type. We propose super-diffusive hot-carrier transport as the microscopic origin of the ultrafast photocurrent. Furthermore, we show that the ultrafast hot-carrier contribution can coexist with the much slower cold-carrier contribution based on the photovoltaic effect, demonstrating that hot carriers can be harvested without discarding lower-energy carriers. These findings highlight the potential of sub-picosecond hot-carrier extraction to expand the O-E conversion bandwidth without sacrificing efficiency, bridging fundamental hot-carrier physics with ultrahigh-speed technological applications.

Anion exchange membrane water electrolysis is an emerging technology for cost-effective hydrogen production, but its practical deployment is limited by inefficient electrode architectures that suffer from poor mass transport, inadequate interfacial contact within the assembled cell, and insufficient reliability. Here we show a high-performance and durable oxygen evolution electrode, aNiFe-based functionalized porous transport layer (NiFe-f-PTL), that integrates catalytic activity and porous transport functionality into a single structure with interfacial compatibility. Fabricated via a simple tape-casting process, this architecture minimizes interfacial resistance and enhances mass transport, resulting in a high current density of 6.73 A cm-2 at 1.8 Vcell under steady-state operation. It further demonstrates over 2142 hours of stable operation, maintaining structural integrity and catalytic activity. These results highlight the importance of electrode architecture at the membrane electrode assembly level and provide a scalable strategy for developing reliable and high-performance anion exchange membrane water electrolysis systems.

Human-driven freshwater change relates to elevated Earth system risks, which motivates analysis to better understand its global characteristics. Here, we analyse global and regional patterns of anomalous conditions and their drivers in streamflow (blue water) and soil moisture (green water), building on the recently updated planetary boundary for freshwater change (PB-FW). Our data consist of updated scenario simulations from a large ensemble of global hydrological models covering years 1901-2019. During the early twenty-first century, PB-FW transgression has increased across its blue and green water components. Climate has increasingly become the dominant global influence on dry and wet streamflow and soil moisture deviations from pre-industrial-like baseline conditions. Amongst the regionally variable change in blue and green water, direct human forcings (encompassing land and water use changes) intensify particularly dry deviations, whereas wet deviations are mainly climate-driven. Regional unpacking of the global PB-FW transgression improves understanding of the extent, degree and drivers of global freshwater change, guiding mitigation and adaptation strategies in response to it, and is a notable advancement for analysing PBs across scales.