Autism spectrum disorder (ASD), attention-deficit/hyperactivity disorder (ADHD), and schizophrenia (SCZ) represent major neurodevelopmental disorders with distinct typical ages of onset. These disorders exhibit substantial genetic and phenotypic overlap, yet their shared and disorder-specific neurobiological mechanisms remain unclear. We analyzed resting-state functional magnetic resonance imaging data from 2,176 participants (ASD, ADHD, SCZ, and healthy controls). Using heterogeneous matrix factorization, we extracted meta-blood-oxygen-level-dependent signals to reduce individual heterogeneity and constructed functional connectivity networks. Partial least squares identified a shared transdiagnostic abnormal connectivity pattern (STACP) and disorder-specific connectivity deviations (DSCDs). We annotated edges with transcriptomic, neurotransmitter, and mitochondrial maps for biological interpretation. The STACP involved connections linking deep regulatory systems (cerebellum, brain stem, and subcortical network) and cortical perceptual–executive networks (default mode, visual, frontoparietal, and somatomotor). The DSCDs of ASD and ADHD implicated overlapping networks with opposite functional connectivity directions (decreased in ASD and increased in ADHD), while SCZ showed more widespread desynchronization. STACP-related genes were enriched for synaptic development, cytoskeletal remodeling, and lipid metabolism, expressed in midbrain and deep-layer cortical neurons, and associated with serotonin transporter and cytochrome c oxidase. DSCDs were linked to glutamatergic plasticity and immune activation in ASD, dopaminergic regulation and glia–neuron interactions in ADHD, and broad synaptic plus immune–metabolic dysregulation in SCZ. Together, these findings provide a systems-level characterization of shared and disorder-specific neurobiological features across major neurodevelopmental disorders observed at different life stages.

Alkaline zinc–ferricyanide flow batteries (AZFFBs) emerge as promising candidates for long-duration energy storage. However, at cryogenic temperatures, these systems suffer from electrolyte solidification, anodic zinc dendrite formation, zinc-related side reactions, and cathodic Fe(CN)₆⁴⁻ precipitation-induced capacity decay. Herein, we propose a synergistic solvation strategy in which Li⁺ and Cl⁻ jointly inhibit the formation of tetrahedral hydrogen bond networks, thereby lowering the liquid–solid transition peak temperature of both the anolyte and catholyte. Meanwhile, Cl⁻ is utilized to construct a water-poor solvation structure around Zn(OH)₄²⁻ to optimize zinc deposition and inhibit the side reactions, while Li⁺ enhances the solubility of Fe(CN)₆⁴⁻ by incorporating additional water molecules into its solvation structure through strong ion–dipole interactions. The optimized AZFFB exhibits outstanding low-temperature performance, achieving stable cycling at −20 °C with an average coulombic efficiency of 99.54%. It also demonstrates excellent stability at room temperature, sustaining over 500 cycles at 28 °C with an average coulombic efficiency of 99.79%, representing more than a 22-fold extension in cycle life. Additionally, the AZFFB exhibits robust stability under fluctuating temperature conditions. These breakthroughs markedly enhance the potential of AZFFBs as viable solutions for extreme-environment energy storage, particularly in polar region microgrids, cold-climate off-grid power systems, and subsea power applications.

Protein secretion plays a crucial role in numerous biological processes, yet its underlying mechanisms remain incompletely understood. This study investigates the role of Sec72, a component of the Sec complex in Saccharomyces cerevisiae, in protein targeting and translocation to the endoplasmic reticulum. We discovered that deleting SEC72 significantly enhances the secretion of proteins with strongly hydrophobic signal peptides (SPs), accompanied by observable changes in cellular functions, such as iron homeostasis, cell wall assembly, and protein synthesis. Importantly, we identified specific gene modifications that, in combination with SEC72 deletion, enable a yeast strain to secrete α-amylase up to 6.5 g/l in fed-batch fermentation. These findings deepen our understanding of SP-mediated protein translocation and provide a basis for optimizing yeast hosts for more effective protein production.

Lithium-ion batteries (LIBs) are central to sustainable energy systems but are vulnerable to thermal runaway (TR), necessitating robust safety monitoring. Conventional single-sensor systems cannot decouple the coupled electrochemical, thermal, and mechanical processes, while existing multimodal sensors suffer from signal cross-talk and large footprints in practical applications. Here, we introduce an insect-inspired, self-decoupling multimodal sensor, fabricated via maskless laser direct writing, that leverages distinct sensing mechanisms and orthogonal output signals for simultaneous strain, temperature, and gas detection. The sensor achieves reliable intrinsic decoupling of strain and temperature over 20 to 110 °C, along with an independent gas response. Seamlessly integrated onto lithium iron phosphate (LiFePO₄) cells, it captures real-time multimodal data during both normal and TR events. Coupled with a bespoke multiphysics model, this platform reconstructs the thermal–mechanical evolution of LIBs. Our work provides a compact and durable strategy for precise real-time monitoring and early warning of battery failure.

Upon tumor metastasis, lymph nodes (LNs) undergo mechanical stiffening, yet how this change influences T cell activation within the microenvironment remains incompletely understood. In particular, the dynamic mechanical forces during activation are transduced by cell–extracellular matrix (ECM) interactions, while cell–cell interactions persist. Here, we established a novel T cell culture platform using hydrogels with tunable stiffness and decoupled presentation of RGD peptide and anti-CD3 monoclonal antibody, separately mimicking ECM–T cell and T cell–antigen-presenting cell interactions. This platform closely mimics the LN microenvironment during T cell activation. By integrating experiments with mathematical modeling, we revealed that T cells sensed mechanical changes in the microenvironment requiring RGD/integrin ligation, while stiff matrix up-regulated F-actin aggregation instead of myosin contraction, deforming the nucleus and promoting yes-associated protein nucleus translocation, resulting in interleukin-2 expression and T cell activation. Our findings shed light on the mechanobiological mechanism underlying the potential benefits of immunotherapy in patients with LN metastases and provide an optimized mechanical platform for studying T cell activation and expansion in vitro.

Humanoid robots possess the potential to solve complex problems across diverse environments, such as nuclear-contaminated zones, epidemic-affected areas, and extraterrestrial missions. However, a humanoid robot is an inherently complex system that integrates multiple disciplines, including perception, mechanical design, materials science, and motion control, each of which requires comprehensive and in-depth investigation. Among these aspects, motion control plays a crucial role, as it directly determines the robot’s motion accuracy, stability, and flexibility. In recent years, with the rapid evolution of graphics-processing-unit-based parallel computing and high-fidelity simulation environments, various deep-reinforcement-learning (DRL)-based approaches have been proposed to achieve precise and robust motion control due to its flexibility and adaptability in uncertain and dynamic environments. However, the inherent complexity and uncertainty of real-world tasks pose substantial challenges when designing effective reward functions for DRL agents. Most current methods typically rely on manually engineered or externally tuned reward signals and therefore require considerable domain expertise, associated with considerable human efforts and a long convergence time; these issues may even trigger mission failure. This work proposes an automatic reward learning method to derive reward functions for DRL in humanoid robot locomotion control. Specifically, a bilevel optimization framework is developed to enable automatic reward learning during policy learning. The reward learning mechanism in the upper level adaptively constructs and optimizes the reward function. The DRL framework in the lower level learns the locomotion control policy using the learned reward function. Three sets of experiments are conducted to verify the effectiveness of the proposed approach: training soft actor-critic and proximal policy optimization agents in MuJoCo environments, training the proximal policy optimization agent in a humanoid robot environment built with Isaac Lab, and transferring the agent to the real-world Unitree G1 humanoid robot through sim-to-real. The experimental results demonstrate that the proposed automatic reward learning method substantially improves learning efficiency and achieves superior performance to manually designed reward functions in both simulation and real-world deployment. By enhancing the success rate of transferring control policies from simulation to real-world humanoid robots, this approach provides a promising pathway toward accelerating the deployment of stable and adaptive humanoid robots in practical applications.

In response to the disparity between the notable biological efficacy and systemic toxicity of celastrol (Cel), this study established a synergistic delivery system (Cel/Lipo+/BHMs) that integrates Bletilla striata polysaccharide (BSP) and hyaluronic acid biomimetic lubricating double-cross-linked microgels (BHMs) infused with Cel cationic liposomes (Cel/Lipo+). The objective is to collaboratively attain attenuation, sustained retention, and reconfiguration of the immunological milieu for the treatment of osteoarthritis. Cel/Lipo+ selectively targets chondrocytes via electrostatic interactions, hence improving Cel delivery and reducing off-target damage. Biolubrication by BHMs reduces friction and markedly extends the drug’s retention duration within the joint cavity. In vitro studies have shown that Cel/Lipo+/BHMs possess excellent biocompatibility, enhance the proliferation and migration of chondrocytes, diminish oxidative stress, and work in conjunction with Cel/BSP to modulate macrophage reprogramming while suppressing the release of proinflammatory factors tumor necrosis factor-α and interleukin-1β. In a rat osteoarthritis model induced by monosodium iodoacetate, Cel/Lipo+/BHMs effectively mitigated cartilage degeneration by synergistically suppressing the inflammatory response, diminishing the expression of the cartilage degradation enzyme matrix metalloproteinase 13, enhancing the synthesis of type II collagen and aggrecan, and ameliorating subchondral bone microstructure. The therapeutic efficacy markedly exceeds that of either Cel/Lipo+ or BHMs individually, thereby emphasizing the primary benefits of the multitiered collaborative mechanism of the delivery system.

Postmenopausal osteoporosis and depression often occur together, but a single treatment that improves both conditions is currently lacking. The loss of estrogen can trigger oxidative stress, damage mitochondria, and drive dysregulated autophagy with impaired flux, simultaneously harming bone and the brain. We evaluated whether total flavonoids from Drynaria fortunei (TFDF) could counter these problems by activating sirtuin-1 (SIRT1), a protein that supports autophagy and mitochondrial health. In menopausal and chronic stress model mice and in cultured bone-forming cells and hippocampal neurons exposed to oxidative injury, we measured bone structure and strength indicators, mood-related behaviors, mitochondrial function, and gene activity patterns. The flavonoids preserved bone density and fine bone structure, shifted bone turnover toward formation, and improved depression-like behaviors (greater exploration and sucrose preference, less immobility). Across bone and the brain, TFDF modulated SIRT1–FOXO3–DEPP1 signaling and FOXO-linked oxidative stress and autophagy programs, thereby normalizing autophagic recycling and mitochondrial function. In cellular models, TFDF preserved mitochondrial function and restored autophagic recycling, and the loss or gain of SIRT1 function abolished or enhanced these benefits, respectively, indicating that SIRT1 activity is necessary for the effects of TFDF. These findings identify TFDF as a single, mechanism-based strategy that addresses both skeletal deterioration and depressive symptoms after menopause by engaging SIRT1-dependent stress autophagy pathways to restore cellular recycling and energy control.

Brain metastasis (BrM) is a common occurrence in lung cancer and substantially worsens the prognosis due to the blood–brain barrier (BBB), which restricts drug entry into the brain. Here, we found that exosomes secreted by lung cancer cells that had acquired epidermal growth factor receptor tyrosine kinase inhibitor resistance and undergone epithelial–mesenchymal transition (osimertinib- and WZ4002-resistant H1975) exhibited enhanced brain-specific distribution and a concomitant increase in BrM compared with exosomes from parental H1975 cells. To identify exosomal mediators of this phenotype, liquid chromatography-tandem mass spectrometry-based proteomic analysis was performed. Exosomes derived from resistant cell lines exhibited distinct protein profiles relative to parental cells, with 744 exosomal proteins significantly altered (fold change ≥ 2; P ≤ 0.05). Prioritization of membrane proteins and ligand–receptor interaction analysis identified ITGAV (integrin αV), ITGB3 (intergrin β3), and L1CAM (L1 cell adhesion molecule) as candidates interacting with brain-specific ligands, including neural cell adhesion molecule 1 (NCAM1) and contactin 2. Validation of exosomal association by Western blotting identified ITGB3 and L1CAM as final candidates. Subsequent functional modulation studies demonstrated that exosomal L1CAM plays a dominant role in brain distribution and metastatic progression. Exosomal L1CAM increased BBB permeability by disrupting endothelial tight-junction integrity both in vitro and in vivo. This effect was associated with the involvement of NCAM1 on BBB endothelial cells, as suggested by an exosomal L1CAM masking experiment. Clinically, exosomal L1CAM demonstrated diagnostic potential for BrM (area under the curve [AUC] = 0.80), and a combined exosomal L1CAM/ITGB3 panel significantly improved diagnostic accuracy (AUC = 0.98). Collectively, these findings identify exosomal L1CAM as a key regulator of brain-specific metastasis and support the clinical utility of the L1CAM/ITGB3 panel as a noninvasive biomarker for BrM in lung cancer.

Direct generation (DG) technologies—comprising dielectric elastomer generators (DEGs) and dielectric fluid generators (DFGs)—offer a promising paradigm for ocean wave energy conversion by integrating transduction mechanisms directly into wave-responsive materials. This assessment provides a comprehensive analysis of DG systems, outlining their working principles, recent material innovations, and comparative performance in harsh marine environments. We examine advancements in dielectric materials, including silicone-based and emerging nonsilicone elastomers, and discuss their influence on energy density, electromechanical efficiency, and environmental resilience. Comparative assessments highlight the advantages of DFGs in long-term durability and energy conversion under complex wave dynamics, while DEGs remain competitive due to their mechanical flexibility and scalable fabrication. The review concludes with a discussion of hybrid system integration, challenges in large-scale deployment, and a roadmap toward commercialization. By synthesizing current research trajectories, this article aims to accelerate the transition from laboratory-scale prototypes to deployable, cost-effective ocean energy harvesting solutions.