We present a deep learning approach to extract physical parameters (e.g., mobility, Schottky contact barrier height, and defect profiles) of two-dimensional (2D) transistors from electrical measurements, enabling automated parameter extraction and technology computer-aided design (TCAD) fitting. To facilitate this task, we implement a simple data augmentation and pretraining approach by training a secondary neural network to approximate a physics-based device simulator. This method enables high-quality fits after training the neural network on electrical data generated from physics-based simulations of ~500 devices, a factor >40× fewer than other recent efforts. Consequently, fitting can be achieved by training on physically rigorous TCAD models, including complex geometry, self-consistent transport, and electrostatic effects, and is not limited to computationally inexpensive compact models. We apply our approach to reverse-engineer key parameters from experimental monolayer WS₂ transistors, achieving a median coefficient of determination (R²) = 0.99 when fitting measured electrical data. We also demonstrate that this approach generalizes and scales well by reverse-engineering electrical data from high-electron-mobility transistors while fitting 35 parameters simultaneously. To facilitate future research on deep learning approaches for the inverse design of transistors, we have published our code and sample datasets online.

Achieving a prescribed 1- or 2-dimensional curvature from a flat shape, and their active modulation through far-field intelligent control, is crucial for a range of technologically demanding applications. In this article, we endeavor to exploit the bistable mechanics of waterbomb origami units, their strategic assembly, and spatially sequential alteration of local stability states to develop a new class of active metasurfaces. Based on a series of computational and experimental investigations, we propose a sequentially programmable approach of far-field stimuli-responsive activation, exploiting locally controllable bistability for achieving target curvatures that can hold their shapes without any external energy supply. A magnetic beam amplification system is developed and tested, demonstrating contactless stability state inversion of waterbomb units, which would essentially lead to prescribed curvatures in an origami-architected 1- or 2-dimensional domain. We further demonstrate that algorithmic control of localized stability states and their dynamic inversion can lead to robotic locomotion. The presented concept of spatially sequential stability state alteration, leading to the multifunctionality of actively reconfigurable metasurfaces, gripper function, and robotic locomotion, will find critical engineering applications in the fields of active and reconfigurable antennas, solar panels, aerodynamically adaptive aerofoils, soft robotics, energy-efficient grippers, biomedical apparatus, and programmable optical and acoustic devices.

Volumetric muscle loss (VML) causes irreversible loss of contractile tissue and creates a hostile regenerative niche marked by sustained inflammation, oxidative stress, fibrosis, and poor functional recovery. Here, we develop an injectable and photocurable microsphere-in-hydrogel platform that couples structural support with sustained small-molecule immunoregulation. To counteract this hostile microenvironment, we utilized emodin, a natural anthraquinone recognized for its potent anti-inflammatory and reactive-oxygen-species-scavenging properties. Emodin-loaded sodium alginate microspheres were generated via ionic crosslinking and embedded within a gelatin methacryloyl matrix to form E-AMs@GM. The composite hydrogel exhibited defect-conforming moldability, porous microarchitecture, tunable swelling/degradation, and broad interfacial adhesion. In vitro, E-AMs@GM showed excellent cytocompatibility and attenuated intracellular reactive oxygen species in human bone-marrow-derived mesenchymal stem cells under oxidative challenge. In macrophages, E-AMs@GM reduced pro-inflammatory activation while enhancing pro-regenerative programs, accompanied by decreased inflammatory cytokines and increased interleukin-10. E-AMs@GM also promoted C2C12 myogenic differentiation and myotube maturation. In a murine VML model, E-AMs@GM alleviated inflammation and fibrotic remodeling, increased myogenic progenitor activity and myofiber regeneration, and improved locomotor performance by CatWalk analysis. Mechanistically, scRNA sequencing revealed that E-AMs@GM enriches a reparative Mmp12⁺ macrophage subset and rewires macrophage–muscle satellite cell interactions through SPP1–CD44 and SPP1–integrin signaling, consistent with accelerated transition from inflammatory clearance to tissue reconstruction. Together, this depot-enabled hydrogel provides an instructive biomaterial strategy for functional VML repair.

Background: Glioblastoma, IDH-wild-type, is a highly invasive tumor, and prognosis largely depends on the extent of resection. Contrast-enhanced (CE) MRI often underestimates tumor borders, resulting in incomplete resection. Prospective multicenter evidence on positron emission tomography (PET)/magnetic resonance (MR)-guided resection planning in newly diagnosed glioblastoma remains limited. This study is the first nationwide, multicenter evaluation of ¹⁸F-fluoroethyltyrosine (FET) PET-guided glioma resection. Methods: We conducted a multicenter, prospective cohort study to evaluate the efficacy of a novel ¹⁸F-FET PET-based surgical planning system (PET/MR cross-modal tumor delineation system [PCMDS]) in glioblastoma, IDH-wild-type resection. The study comprised 239 patients: 115 underwent ¹⁸F-FET PET-guided surgery and 124 underwent CE MRI-guided surgery. The PCMDS integrated multimodal image registration and automated metabolic segmentation to delineate surgical margins. Resection outcomes were intraoperatively validated using ultrasound, MRI, and ultramicroscopic cellular imaging. Results: Compared with the CE MRI group, the ¹⁸F-FET PET group achieved a significantly higher gross total resection rate (91.3% versus 72.7%, P < 0.05) and improved survival outcomes (median overall survival: 19.7 versus 16.0 months, P = 0.0099; median progression-free survival: 12.5 versus 9.3 months, P = 0.0078). ¹⁸F-FET PET-defined margins extended beyond CE MRI-defined boundaries, and both histopathological analysis and single-cell RNA sequencing confirmed the presence of infiltrative tumor cells in ¹⁸F-FET PET-positive but MRI-negative regions. Conclusion: This multicenter study demonstrates that ¹⁸F-FET PET-based surgical planning significantly increases the extent of glioblastoma, IDH-wild-type resection and prolongs patient survival. Integrating ¹⁸F-FET PET into routine surgical practice could substantially improve clinical outcomes for patients with glioblastoma, IDH-wild-type.

Emerging and re-emerging infectious diseases ranging from the 1918 H1N1 influenza pandemic to the recent SARS-CoV-2 and monkeypox virus outbreaks continue to pose profound threats to global public health. These crises underscore the critical need for high-fidelity and human-relevant infection models. Organoid technology has emerged as a cornerstone platform for pathogen research by faithfully recapitulating the 3-dimensional architecture and physiological microenvironment of native human tissues in vitro. This review systematically examines the development and structural refinement of organoid-based infection models with an emphasis on evidence-based strategies for stem cell source selection, extracellular matrix optimization, and dynamic culture system engineering. Such advancements enable the robust generation of multi-organ models including respiratory, intestinal, and neural organoids tailored for investigating viral tropism, spatiotemporal infection kinetics, and host immune responses. Furthermore, we evaluate the translational utility of organoids in high-throughput antiviral drug screening and preclinical vaccine assessment. To further enhance physiological relevance and functional fidelity, organoid platforms are being increasingly combined with advanced engineering strategies, including coculture approaches, CRISPR–Cas9-mediated genetic perturbation, engineered microphysiological systems (such as organ-on-a-chip), and 3D bioprinting. These integrated technologies improve biomimicry while expanding experimental controllability and scalability. In addition, we critically examine the major bottlenecks limiting clinical translation and discuss emerging frontiers driven by artificial intelligence and synthetic biology. Through iterative technological refinement and cross-disciplinary convergence, organoids have evolved beyond reductionist in vitro surrogates into physiologically informed and mechanism-driven platforms that advance our understanding of host–pathogen interactions while enhancing global preparedness against emerging pathogens.

Early-life adversity, including prenatal stress exposure, has enduring effects on stress responsivity later in life. Yet the impact of maternal stress on offspring susceptibility to stress and its underlying biological mechanisms remain to be elucidated. Here, we established a “2-hit” stress model to investigate whether maternal chronic unpredictable stress (E9.5 to E18) increases offspring susceptibility to social isolation during adolescence (P28 to P49). Our study reveals that maternal stress increases susceptibility to adolescent social isolation in male—but not female—offspring, manifested as increased anxiety- and depressive-like behaviors. This vulnerability is mediated by the priming of hippocampal dentate gyrus microglia through the complement C3–C3aR signaling pathway, which promotes the aberrant phagocytosis of excitatory synapses and reduces glutamatergic neuronal activity. Notably, pharmacological blockade of C3aR or chemogenetic activation of glutamatergic neurons in the dentate gyrus during adolescence effectively alleviated stress susceptibility. Therefore, our findings identify a targetable immune-mediated mechanism in the hippocampus that underlies stress vulnerability in offspring exposed to gestational maternal stress, offering new avenues for preventing adolescent-onset mood disorders.

The strong interaction is the fundamental force that holds quarks and the gluon force carriers together to form protons and neutrons and also binds the atomic nucleus. The theory governing quark–gluon interactions is quantum chromodynamics (QCD). A wide variety of experimental data teaches us that quarks and gluons cannot be observed in isolation, a phenomenon known as confinement that is unique to QCD. But no one has used QCD to mathematically prove confinement. Here, we show how to define and measure the force on quarks in the proton using available experimental data. Direct evidence for confinement is obtained because the force is found to be attractive and constant for a wide range of quark positions. This work guides future experimental efforts at future electron-ion colliders aimed at obtaining a rigorous quantitative understanding of confinement and the origin of nuclear mass.

Immune evasion in lung cancer is tightly regulated by epigenetic mechanisms. This study identifies key epigenetic drivers of immune escape in lung cancer and elucidates how they remodel the tumor microenvironment to induce CD8⁺ T cell exhaustion. We performed a CRISPR-negative screen targeting 159 epigenetic regulators (mEpi driver library) in a mouse model of lung metastasis and used single-cell sequencing, cell-based experiments, and chromatin analyses to identify mechanisms of immune escape. SET domain-containing protein 2 (SETD2) emerged as a critical suppressor of immune evasion. Loss of SETD2 markedly increased lung colonization and metastasis in mice with intact T cells, whereas this effect disappeared after T cell depletion. Mechanistically, SETD2 recruited adenine–thymine-rich interaction domain-containing protein 2 (ARID2) to maintain a repressive chromatin state at genes encoding the chemokines C-X-C motif chemokine ligand 1, 2, and 5. Loss of the SETD2–ARID2 axis increased secretion of these chemokines and recruited polymorphonuclear myeloid-derived suppressor cells, which inhibit T cell responses. These cells increased expression of free fatty acid receptor 2 (FFAR2), depleted arginine from the tumor environment, and impaired signaling required for T cell activation, driving cytotoxic T cells into terminal exhaustion. Depleting polymorphonuclear myeloid-derived suppressor cells or conditionally deleting FFAR2 in these cells restored T cell function and reduced metastasis of SETD2-deficient tumors. These findings show that loss of SETD2 creates an arginine-depleted, immunosuppressive tumor environment through a chemokine–FFAR2 pathway, revealing an epigenetic–metabolic immune checkpoint that may inform combination treatments for lung cancer.

During early embryogenesis, epiblast cells ingress through the primitive streak (PS) to adopt a mesendoderm (MES) fate, giving rise to both mesodermal and endodermal lineages. Robust systems that accurately recapitulate MES specification—and thereby enable a comprehensive understanding of cell fate transitions during PS formation—remain limited. Here, we describe 3-dimensional MES organoids derived from human induced pluripotent stem cells that faithfully model in vivo human MES specification, particularly from the anterior PS. MES specification requires PTEN expression, as PTEN loss profoundly impairs mesoderm and definitive endoderm derivatives. Integrative multi-omics analyses—including transcriptomics, chromatin accessibility, metabolomics, and phosphoproteomics—on wild-type and PTEN^−/− MES cells revealed that PTEN suppresses retinoic acid (RA) signaling during MES commitment. Furthermore, we identified the RA-degrading enzyme CYP26A1 as a downstream effector of PTEN. Notably, while excessive RA induced by PTEN ablation is detrimental to MES cell generation, physiological levels of RA are necessary. Collectively, the human MES organoids offer a valuable model for dissecting early human development, and our findings identify PTEN as a key regulator of MES fate commitment through inhibition of RA signaling.

Precision theranostics for castration-resistant prostate cancer remain a major challenge because of tumor heterogeneity, drug toxicity, and resistance to conventional therapies. Here, we report a multifunctional theranostic nanoplatform that integrates multimodal imaging guidance, dual-targeted delivery, photothermal therapy, and immune activation for the precision treatment of prostate cancer. By encapsulating aggregation-induced emission luminogens and magnetic nanoparticles, this system achieves superior second near-infrared fluorescence and magnetic resonance imaging fusion for deep-tissue visualization. The dual-targeting strategy, which integrates magnetic navigation with simultaneous engagement of six-transmembrane epithelial antigen of the prostate 1 and prostate specific membrane antigen (PSMA), effectively addresses tumor heterogeneity by enabling externally guided and biologically specific delivery to both PSMA⁺ and PSMA⁻ prostate cancer cells. Upon targeted delivery, the synergistic combination of photothermal therapy and chemotherapy achieves potent tumor growth inhibition, activates CD8⁺ T cells and macrophages to promote robust antitumor immune responses, and elicits durable immunological memory to prevent tumor recurrence. In vitro and in vivo studies demonstrate superior tumor targeting specificity, enhanced imaging contrast, and minimal off-target toxicity. This study provides novel insights into the construction of a multifunctional theranostic nanoplatform in precision oncology, with promising translational potential for clinical management of castration-resistant prostate cancer.

Plant-derived extracellular vesicles are promising candidates for oral drug delivery, yet their clinical translation is hindered by limited targeting precision and inconsistent systemic absorption. While surface engineering can enhance tissue accumulation, strategies that preserve biocompatibility and enable scalable production remain limited. Here, we introduce boiling as a simple thermal processing approach that structurally reconfigures ginger extracellular vesicles (GEVs) into functionally enhanced, thermally reassembled GEVs (T-GEVs). The surface architecture of T-GEVs is enriched with key vesicle trafficking regulators, including V-type proton adenosine triphosphatase subunit G, ARF1, and β-adaptin-like protein. This specific composition drives their tissue-specific accumulation in the intestine and liver and potentiates clathrin-dependent cellular uptake in intestinal cells by 8.57-fold. Beyond superior intrinsic anti-inflammatory activity through NLRP3 inflammasome suppression, T-GEVs function as an efficient oral delivery platform. When loaded with tumor necrosis factor-α (TNF-α) small interfering RNA, they enable a synergistic therapy that combines innate anti-inflammatory activity with targeted gene silencing of TNF-α, showing potent efficacy in colitis. Our findings position boiling as a natural strategy for enhancing the bioactivity and targeted oral delivery potential of GEVs.

Cell-based therapeutics hold immense promise for clinical intervention, yet their practical translation is severely bottlenecked by suboptimal delivery kinetics, safety risks, and low patient adherence. This review provides a comprehensive overview of recent innovations in microneedle (MN)-based cell delivery systems and their applications across multiple organs. We initiate the discussion by evaluating progress in regenerative medicine, specifically targeting stem and immune-derived lineages. The article further classifies MN systems for cell therapy delivery according to material composition, fabrication techniques, and functional characteristics, with an in-depth discussion of innovative designs, highlighting their unique advantages. Additionally, we categorize and analyze recent studies on MN-based cell therapy for skin diseases, endometrial repair, cardiac regeneration, and oral disorders. Beyond summarizing current achievements, we critically examine the translational barriers that limit clinical adoption, including manufacturing scalability, storage stability, and regulatory considerations. This review provides a systematic and organ-specific overview that may help guide the design of improved and more clinically translatable cell delivery platforms.

Cancer-related lymphedema is a chronic progressive side effect of cancer treatments followed by lymph node dissection or radiotherapy. Clinicians often identify lymphedema through limb swelling, while the disease begins earlier than that. Lymphatic injury is the original cause, where elevated interstitial fluid pressure and distorted tissue mechanics will lead to immune activation and fibrofatty remodeling. Recently, mechanobiology studies have deepened our understanding by linking lymph stasis to T helper 2/transforming growth factor β signaling, fibroblast mechanotransduction, and YAP/TAZ activity that together lock tissues into a stiff, poorly draining state. Simultaneously, emerging artificial intelligence (AI) in the field are being explored, from proof-of-concept image classification to much more diagnostic models that integrate elastography, indocyanine green lymphography, radiomics, clinical variables, and wearable signals to detect preclinical mechanical signatures and predict risk. These advances are driving the development of promising mechanomedical approaches, such as adaptive compression systems, AI-assisted plans for lymphatic reconstruction, anti-fibrotic strategies, and lymphangiogenic regeneration, although most remain at preclinical or early clinical feasibility stages of translation. We discuss the strength of current evidence, challenges for clinical translation, and standards for reporting. We propose that cancer-related lymphedema can be understood as a measurable and targetable mechano-immune-fibrotic disease, for which AI may eventually support earlier diagnosis, risk prediction, and personalized mechanotherapy.

Spinal cord injury often causes permanent disability because the body’s own repair mechanisms are limited, and the molecules that control damage and healing are not fully understood. One such molecule, FK506-binding protein 5 (FKBP5), is known to rise sharply after injury, but whether it only drives harmful inflammation or also participates in later recovery has been unclear. In this study, we investigated how FKBP5 affects microglia—the brain’s immune cells—at different stages after spinal cord injury in mice. We found that FKBP5 plays a dual role. In the first few days, it works together with another protein, GPR84, to boost the production of an inflammatory signal called interleukin-1β. This signal pushes microglia into a destructive state and triggers a coordinated form of neuronal cell death that involves multiple death pathways. However, as FKBP5 levels continue to rise over time, it switches its function. It binds to and modifies an enzyme called LDHA, changing how microglia process lactate. This lactate then acts as a signal to add chemical tags (lactylation) onto histones, which turns on a protective gene, Fxyd5, and its partner Lgals1. These changes convert microglia from a harmful to a healing state, reduce neuronal death, and improve the local environment for tissue repair. Our results reveal that FKBP5 is a double-edged sword—first worsening damage, then promoting repair. This discovery suggests that precisely timing therapies that target FKBP5 could offer a new way to improve recovery after spinal cord injury.

Uterine fibroids pose a major global health burden, affecting approximately 120 million women worldwide. While minimally invasive surgeries, such as electromechanical morcellation, preserve fertility, they are severely hampered by the risk of unpredictable tissue dissemination—a hazard that prompted a U.S. Food and Drug Administration black-box warning. Resolving this clinical dilemma requires a paradigm shift in instrument design. This review identifies a persistent “mechanical blind spot”: Current surgical tools operate with static parameters that fundamentally clash with the profound multi-scale mechanical heterogeneity of fibroids. Driven by FIGO-defined microenvironments and degeneration-driven remodeling, this extreme variability in tissue stiffness shapes the dynamic—and often hazardous—cutting challenges during surgery. To bridge this critical gap, we review the mechanobiological origins of this heterogeneity and analyze its direct impact on morcellation-induced fragmentation. We then synthesize multiscale mechanical quantification methods (ex vivo and in vivo) and evaluate the evolution of instrument-optimization strategies. Moving beyond conventional structural refinements, we highlight the potential of bio-inspired end-effectors for safe tissue interaction. Crucially, we propose a technological roadmap integrating these physical mechanisms with artificial intelligence—enabling preoperative “mechanical mapping” and real-time adaptive control. Taken together, this work offers new perspectives for the clinical management of this prevalent condition and provides a comprehensive theoretical framework for developing next-generation, mechanics-aware surgical ecosystems.

Diabetic kidney disease, the leading cause of end-stage kidney disease worldwide, involves complex interactions beyond classical hemodynamic and oxidative stress pathways. Recent advances emphasize metabolic reprogramming in renal cells—characterized by mitochondrial dysfunction, impaired fatty acid oxidation, lipotoxicity, and glycolytic shifts—as upstream drivers of cellular injury and fibrosis. Single-cell RNA sequencing reveals profound immunometabolic heterogeneity, including dynamic macrophage subpopulations (e.g., proinflammatory early states transitioning to TREM2^hi/MRC1^hi lipid-associated phenotypes) and T helper 17/regulatory T imbalance, which amplify inflammation via bidirectional crosstalk with podocytes, tubular cells, and mesangial cells. Interorgan axes, particularly gut dysbiosis and uremic toxin accumulation, further perpetuate immune dysregulation. This review integrates these insights to propose precision strategies targeting mitochondrial homeostasis, ferroptosis inhibition, glycolytic blockade in immune cells, and multimodal therapies (e.g., combination strategies integrating sodium–glucose cotransporter 2 inhibitors with immunometabolic modulators). Multi-omics integration and spatial transcriptomics hold promise for individualized and mechanism-guided interventions to halt diabetic kidney disease progression.

Colorectal cancer (CRC) remains a global health challenge with rising incidence in younger populations and limited efficacy of current chemotherapies. In this study, we investigated the anti-tumor mechanisms of metochalcone (MET), a natural chalcone derivative, in CRC. Using a combination of transcriptomic, proteomic, and targeted in vitro and in vivo assays, we found that MET substantially reduced viability, migration, and invasion and arrested the cell cycle at the G₀/G₁ phase in HCT-116 CRC cells. Integrative analysis identified heme oxygenase-1 (HO-1) as a key target, and molecular docking and cellular thermal shift assay confirmed that MET directly binds to and stabilizes the HO-1 protein. Mechanistically, MET treatment led to mitochondrial dysfunction, characterized by increased reactive oxygen species and decreased membrane potential, which in turn activated PTEN induced kinase 1 (PINK1)/Parkin-mediated mitophagy. In 2 distinct mouse models of CRC, MET potently suppressed tumor growth. Furthermore, 16S ribosomal RNA gene sequencing revealed that MET treatment was associated with changes in gut microbiota composition in mice, including an increased relative abundance of beneficial Lactobacillus and a decreased abundance of pro-inflammatory Desulfovibrionaceae. Our findings demonstrate that MET exerts multifaceted anti-tumor effects, including direct targeting of HO-1 and activation of PINK1/Parkin-mediated mitophagy, accompanied by alterations in gut microbiota composition.

Driven by the demands of intelligent manufacturing, medical rehabilitation, and human–robot collaboration, anthropomorphic dexterous hands have become a key direction for complex manipulation. This paper reviews their evolution from structural biomimetics to intelligent control, emphasizing structural optimization, task generalization, and perception–control coupling. First, it summarizes structural design principles, including structural parameterization, underactuated and compliant mechanisms, functional materials, and multimaterial fabrication. Second, it reviews perception technologies, including angle and strain sensing, tactile and pressure detection, and compact vision–tactile fusion modules. Third, it compares classical model-based control, data-driven learning control, and hybrid model–data control, with attention to sample efficiency, robustness, and transferability. Finally, it discusses deployment-oriented strategies, including modular design, standardized interfaces, integrated packaging, and unified communication frameworks for hardware–software coordination. In summary, this paper establishes a system-level technical framework for dexterous hand development and deployment, organized around 4 tightly coupled dimensions: structural embodiment, multimodal perception, control, and deployment-oriented integration. By clarifying the interactions and constraints across these dimensions, the paper provides a system-level basis for the design, evaluation, and deployment of dexterous hands in complex manipulation and human–robot collaboration.

Engineered exosomes are increasingly recognized as promising vehicles for precision drug delivery. Genetic display, chemical modification, physical loading, and hybrid vesicle construction have expanded their applications, while introducing additional variables to the interpretation of delivery outcomes. Because engineering can alter vesicle properties, receptor engagement, endocytic routing, intracellular trafficking, and functional readouts, tissue accumulation or cellular uptake alone is insufficient to define effective delivery. This review traces engineered exosome delivery from cellular uptake to therapeutic function, covering engineering strategies, targeting validation, endocytic routing, intracellular fate, and functional readouts. We further discuss the strengths and limitations of commonly used evaluation methods. Across biomedical applications, we emphasize that delivery evidence should be judged according to cargo type, site of action, disease model, and therapeutic mechanism. By linking engineering design, assay selection, and evidence strength, this review supports a more careful interpretation of engineered exosome delivery.

Hypoxia, immunosuppression, and pronounced heterogeneity within the tumor microenvironment (TME) hinder the effectiveness of cancer therapies. Engineered bacteria–nanomaterial hybrid systems have emerged as a promising approach to address these challenges. Bacterial chassis provide active tumor targeting, deep tissue penetration, and in situ proliferation, facilitating the precise delivery of immunomodulators. Concurrently, nanomaterials interfaced with these living carriers can be activated by external physical stimuli, inducing photothermal, photodynamic, sonodynamic, and magnetothermal effects within solid tumors. These interactions promote immunogenic cell death (ICD) and enable real-time monitoring. Recent advances in synthetic biology and nanotechnology have led to the development of an expanding range of preclinical biohybrid platforms, while several related components, including bacterial therapeutics, bacterial derivatives, and physically activated nanomedicine platforms, have progressed into clinical evaluation. This review first explores the origins and roles of tumor-associated bacteria. It then summarizes strategies for engineering bacteria–nanomaterial hybrid systems. Subsequently, this review examines how physical stimuli enhance targeting, remodel the TME, and amplify antitumor immunity. Finally, safety, manufacturing, and regulatory challenges impacting clinical translation are discussed. Overall, these platforms offer a potentially powerful framework for precision cancer immunotherapy. However, successful clinical translation will require stronger evidence regarding safety, controllability, manufacturing consistency, and therapeutic efficacy.

Stretchable elastomers require a stretchability exceeding 200% and a Young’s modulus ranging from 0.1 to 10 MPa to balance strength and extensibility for dynamic applications. Conventional petrochemical polymers meeting these criteria rely on covalent networks, which lack mechanical tunability and face challenges in terms of repairability and recyclability. Here, we introduce a highly dynamic covalent branched network to fabricate a polyimine elastomer (PIE) featuring extensive mechanical tunability, controllable self-healing properties, reprocessability, and recyclability. This PIE is synthesized at room temperature from a lignocellulose-derived dialdehyde prepolymer (dialdehyde-terminated prepolymer) and cross-linked with a triamine via imine bonds. The resulting PIE, characterized by dynamic covalent networks, presents a skin-friendly Young’s modulus of less than 3 MPa and an impressive stretchability of strain exceeding 1,110%. Additionally, PIE exhibits direct thermal reprocessability at 100 °C within 15 min, enabling transformation into various forms, along with self-healing capability. Flexible electronics utilizing the PIE substrate demonstrate appealing stretchability, high sensitivity, and recyclability. This biomass-derived highly branched cross-linking network likely provides a promising approach for designing high-performance bio-based synthetic elastomers, contributing to the advancement of sustainable flexible electronic devices.

Plasma p-tau217/Aβ42 accurately captures the systemic molecular risk of Alzheimer’s disease (AD) but lacks the spatial resolution necessary to predict individualized clinical trajectories. Here, we combined plasma biomarker stratification with normative connectome mapping to identify macroscale neurodegenerative epicenters and developed a personalized prognostic tool, the Network Vulnerability Index (NVI). Across the Alzheimer’s Disease Neuroimaging Initiative and independent China ADNI cohorts, plasma p-tau217/Aβ42-positive individuals exhibited highly reproducible epicenters tightly anchored to the default mode network and limbic axis. Multiscale analyses revealed that this spatial vulnerability aligned with transcriptomic signatures of synaptic and mitochondrial dysfunction, monoaminergic receptor density gradients, and memory-related cognitive domains. Longitudinally, baseline epicenter centrality strictly dictated future localized atrophy rates. To translate these group-level topological constraints into a personalized prognostic metric, we utilized least absolute shrinkage and selection operator regression to formulate the NVI. Cross-sectionally, the NVI robustly tracked progressive tau-positron emission tomography accumulation (meta-temporal r = 0.547) and hippocampal atrophy (r = −0.455). Crucially, the NVI demonstrated robust, stage-dependent prognostic utility. When evaluated across the continuous disease spectrum, incorporating the NVI into a fully adjusted baseline model comprising plasma p-tau217/Aβ42 and APOE-ε4, and clinical scores significantly improved the prediction of conversion from mild cognitive impairment to dementia (hazard ratio = 1.47, P = 0.004), providing essential incremental prognostic value. Collectively, our spatially contextualized framework demonstrates that mapping systemic molecular risk onto structural network vulnerability supports a highly scalable “plasma pre-screen plus standard MRI” triage pathway for precision staging in Alzheimer’s disease.

Despite their widespread use, synthetic adhesives raise substantial concerns regarding environmental impact, toxicity, recyclability, and volatile organic compound emissions, while their predominantly liquid nature often requires energy-intensive curing processes. Here, we report an all-cellulose film adhesive derived from renewable cellulose through a simple oxidation–reduction modification that transforms cellulose into a linear heat-softenable structure. The resulting material can be activated by either water or heat to function as an effective wood adhesive. Upon water activation, the cellulose film bonds thick wood panels into structural composites without external heating, achieving lap shear and compressive shear strengths of 3.1 and 6.5 MPa, respectively. This room-temperature, water-triggered adhesion reduces processing energy demand and is well suited for manufacturing engineered wood products. Owing to its thermoplastic behavior, the adhesive also enables heat-activated bonding of wood and paper, including thermal sealing of paper bags. This work establishes a renewable and low-emission adhesive platform, advancing sustainable alternatives to conventional petrochemical adhesives.

Electroactive biomaterials represent a promising strategy for reconstructing the electrobiological microenvironment of bone and enhancing tissue regeneration. Among these materials, poly(3,4-ethylenedioxythiophene) (PEDOT) and its composites have attracted considerable attention because of their mixed electronic and ionic conductivity and compatibility with soft and porous scaffolds. However, existing reviews rarely address how fabrication strategies govern the relationships between structure, properties, and translational performance. This review establishes a fabrication, performance, and translation framework for PEDOT-based bone repair systems. Fabrication strategies are categorized into interfacial polymerization, bulk matrix and solution-processed conductive networks, patterned and fibrous conductive architectures, and porous and 3-dimensional scaffold fabrication and are correlated with conductive network topology, mechanical performance, and cytocompatibility. The mechanistic roles of PEDOT in osteogenesis, angiogenesis, immunomodulation, and electroresponsive drug release are further summarized. In addition, this review discusses the key trade-offs that limit practical applications, including the balance between conductivity and degradability, mechanical strength and porosity, as well as multifunctionality and manufacturability. Overall, this review provides a framework-oriented perspective to guide the rational design and clinical translation of PEDOT-based bioelectronic materials for bone tissue engineering.

Hearing loss is becoming a common disease worldwide, affecting individuals across all age groups, from infants to older adults. However, the unique anatomical barriers of the inner ear limit the effectiveness of conventional systemic therapies, highlighting the need for advanced drug delivery strategies. Engineered inner ear drug delivery systems have emerged as promising approaches for improving drug targeting, retention, and therapeutic efficacy. In this review, we provide a comprehensive overview of the advances by integrating 3 major aspects: major administration routes, diverse therapeutic agents, and delivery carriers employed in inner ear treatment. We first introduce systemic administration, trans-tympanic drug delivery, intracochlear drug delivery, and cochlear implantation (CI). We then summarize anti-inflammatory drugs and antioxidants for the treatment of inner ear disorders. Subsequently, we discuss multiple nanocarriers, viral vectors, injectable hydrogels, and other innovative carriers used in inner ear drug delivery, highlighting their design principles and therapeutic applications. We also illustrate recent progress in clinical trials and discuss the difficulties associated with translating experimental systems into clinical applications. Finally, we outline current limitations and future opportunities, together with potential strategies to address these challenges.

Quantum secure direct communication (QSDC), a paradigm-shifting breakthrough in quantum communication, exploits quantum states for unmediated information transmission. Rooted in the inviolable fundamental laws of quantum mechanics, QSDC enables ultrasensitive detection of even the faintest eavesdropping attempts, guaranteeing true communication security solely when no interference exists. However, its practical scalability remains constrained by insufficient transmission rates. Semantic communication, which drastically boosts transmission efficiency by extracting core information features, nevertheless stays vulnerable to malicious intrusions. Integrating these paradigms promises to simultaneously enhance the equivalent data rate and security. Herein, we propose and experimentally validate a quantum semantic communication scheme, applying it to 3-dimensional point clouds. It achieves a 46.30-fold efficiency gain over direct transmission in the meaning domain, beyond the Wyner and Shannon capacity limits of syntactic communication. This breakthrough not only broadens the application scope of QSDC with limited bandwidth, but also marks a pivotal milestone in quantum information science.

While recent studies have established links between metabolic reprogramming and inflammatory senescence, the specific metabolic drivers in vascular aging remain incompletely defined. Here, we systematically characterized senescent phenotypes and targeted metabolomic profiles in primary aging endothelial cells, identifying a pyruvate dehydrogenase E1 component subunit alpha (PDHA1)-dependent metabolic shift as a hallmark of cellular senescence. Using a D-galactose-induced senescence model, we demonstrated that endothelial-specific Pdha1 knockdown alleviated pulmonary vascular endothelial senescence and associated functional decline. Further investigation revealed that PDHA1 hyperactivation disrupts mitochondrial homeostasis, leading to excessive mitochondrial reactive oxygen species production, oxidative mitochondrial DNA damage, and subsequent cytosolic mitochondrial DNA release, thereby triggering cyclic GMP-AMP synthase-mediated senescence. Mechanistically, decreased lactylation of PDHA1 at lysine 336 potentiated its activity and promoted dephosphorylation at serine 293. This posttranslational cross talk enhanced PDHA1 activation and drove a prosenescent metabolic shift. Together, our results elucidate that a previously unrecognized PDHA1 hyperactivation promotes endothelial senescence.

Photocatalytic biomass refining offers a promising strategy for the sustainable co-production of high-value chemicals and clean fuels. Herein, a coral-like bifunctional g-C₃N₄/Mn_0.7Cd_0.3S S-scheme heterojunction (CNMCS) was rationally constructed for simultaneous xylan photoreforming and hydrogen evolution. X-ray diffraction and transmission electron microscopy analyses confirm the formation of an interconnected hierarchical structure with intimate interfacial contact between g-C₃N₄ nanotubes and Mn_0.7Cd_0.3S nanoparticles, while x-ray photoelectron spectroscopy and density functional theory calculations reveal an S-scheme charge transfer pathway driven by the internal electric field. The optimized 6CNMCS photocatalyst achieves a glyceric acid yield of 77.6% and an H₂ evolution rate of 2.89 mmol•g_cat⁻¹•h⁻¹ without sacrificial agents. The enhanced performance is attributed to efficient charge separation and synergistic interfacial interactions that promote xylan adsorption and selective C–H bond activation. Mechanistic studies indicate that carbon-centered radicals and reactive oxygen species cooperatively drive selective C–C bond cleavage toward glyceric acid formation. This work provides a new strategy for integrating biomass valorization with clean hydrogen production.

Pathological cardiac hypertrophy is definitely identified as an adverse deterioration of hemodynamically stressful overload, ultimately heightening risk of sudden death as heart failure (HF) ensues. However, molecular mechanisms underlying cardiac hypertrophy pathogenesis have not been fully described. Herein, we employed targeted genetic approaches and in vivo functional imaging to describe a potent cardiac pro-hypertrophic role of serine arginine protein kinase 3 (SRPK3), one of RNA splicing protein kinases family, which serves as a regulator in pathological hypertrophy and HF. Our study revealed heightened level of SRPK3 in the myocardium of transverse aortic constriction (TAC)-administrated mice and patients who suffer from hypertrophic hearts. Adult mammalian extended expression of Srpk3 in cardiomyocytes (CMs) led to mitochondrial defects and dysfunction, inducing spontaneous concentric hypertrophy and HF, and ultimately premature mortality. Conversely, conditional deletion of Srpk3 in CMs attenuated TAC-induced mitochondrial deficits, consequently ameliorating cardiac hypertrophy and HF. Mechanistically, in canonical splicing regulation, we identified SRPK3- serine/arginine splicing-rich factor 2 (SRSF2)-mediated RNA alternative splicing affecting genes related to mitochondrial function. Also, a noncanonical role of SRPK3 involving binding to RPS3 at Ser¹⁴⁹ to destabilize and degrade PGC1α mRNA was unveiled. Using genetically edited human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs), we also showed that SRPK3-induced dual pathway coordinately resulted in defective mitochondrial biogenesis and bioenergetics, thus triggering pathological cardiac hypertrophy, as evidenced by the SRPK3-induced rewiring of mitochondrial proteome and impaired heart fueling. In preclinical treatment, intravein injection of adeno-associated virus 9–Srpk3 short hairpin RNA attenuated concentric hypertrophy and HF progression in mice. Together, our findings help to elucidate the mechanism of mitochondrial proteome depletion in triggering hypertrophic hearts, establishing SRPK3 suppression as a long-term and improved strategy for treating pathological cardiac hypertrophy and HF.

The pancreatic adenosine triphosphate (ATP)-sensitive potassium (K_ATP) channel acts as a crucial metabolic sensor, regulating insulin secretion to maintain whole-body energy homeostasis. Gain-of-function mutations in this channel lead to neonatal diabetes mellitus, a rare disorder in which certain mutants exhibit resistance to standard sulfonylurea therapy. Recent studies have identified a centipede toxin, SpTx1, as a potent inhibitor of both human pancreatic K_ATP channels and their gain-of-function mutants. This toxin stimulates insulin secretion, offering a promising therapeutic strategy for neonatal diabetes. However, the molecular mechanism by which SpTx1 inhibits the K_ATP channel has remained unclear. Here, we report the crystal structure of SpTx1 and the cryo-electron microscopy structure of the K_ATP channel in complex with SpTx1. The structure reveals that a single SpTx1 molecule binds to the extracellular surface of the Kir6.2 tetramer. Multiple interactions at the SpTx1–Kir6.2 interface underpin the high affinity and specificity of SpTx1 for human Kir6.2. We demonstrate that SpTx1 inhibits potassium currents by physically blocking the ion conduction pore. Furthermore, a structure-guided search identified another centipede toxin, Sm3a, as a novel K_ATP channel blocker. Together, these findings provide key insights into the inhibitory mechanism of centipede toxins against the human Kir6.2-containing K_ATP channel and establish a foundation for developing these toxins into potential therapeutics for K_ATP channel-related diseases.

Deep learning models are increasingly used to analyze medical images, but their “black box” nature makes it hard to understand the underlying biology and slows down the development of targeted treatments. To tackle this, we built a multi-step approach that combines deep learning analysis of breast magnetic resonance imaging (MRI) with several types of molecular data, including gene activity, protein levels, and genetic information, along with laboratory experiments. Our MRI-based deep learning model accurately predicted whether breast cancer had spread to lymph nodes, and it performed consistently across 3 separate groups of patients. Causal inference using double least absolute shrinkage and selection operator (LASSO) and causal forest double machine learning established a significant effect of NBPF4 expression on the imaging-defined high-risk phenotype, independent of genomic confounders. When we looked at which genes were linked to the imaging-defined high-risk pattern, one gene called NBPF4 stood out because it was supported by all 4 kinds of evidence: imaging features, gene expression, protein data, and genetic association studies. Follow-up experiments in cells and animals showed that boosting NBPF4 activity made tumor cells grow faster, move more, form new lymphatic vessels, and spread to lymph nodes. Mechanistically, NBPF4 worked by activating the mitogen-activated protein kinase (MAPK) signaling pathway and triggering a process known as epithelial mesenchymal transition (EMT). Interestingly, tumors with high NBPF4 were sensitive to drugs that block one part of the MAPK pathway (JNK/p38) but resistant to another part (ERK), suggesting that the pathway had been rewired. Using this insight, computer-based drug screening and further testing identified MK-886 as a promising compound that could suppress NBPF4-promoted MAPK activation and tumor growth. Together, this work traces a complete path from a noninvasive imaging finding to a specific gene (NBPF4) and a potential treatment (MK-886). It establishes the NBPF4–MAPK–EMT axis as a key player in breast cancer metastasis and provides a general framework for turning imaging-based risk predictions into biological understanding and possible therapies.

Red blood cell (RBC) transfusion is a core clinical intervention. However, hypothermic storage induces progressive biochemical, structural, and functional impairments collectively termed the RBC storage lesion (RSL), which compromises post-transfusion efficacy and safety. Conventional RSL detection relies on bulk population-averaged indicators with low physiological relevance and no single-cell resolution, failing to capture cellular heterogeneity and microcirculatory dysfunction. Lab-on-a-chip (LOC) platforms are established with microfluidic technology, and organ-on-a-chip represents a biomimetic and advanced extension of such systems. Together, they enable biomimetic reconstruction of the in vivo microcirculatory microenvironment, high-throughput single-cell analysis, and quantitative assessment of RBC mechanical phenotypes under physiological shear conditions. Distinct from fragmented prior reviews that separate microfluidic engineering from transfusion clinical demands, this work systematically outlines the molecular mechanisms and clinical impacts of RSL alongside unresolved detection bottlenecks, and builds an integrated LOC–endothelium-on-a-chip technical framework covering structural design, biocompatible material screening, and standardized fabrication workflows. It further summarizes microfluidic core applications including single-cell deformability quantification, stiffness evaluation, hemolysis susceptibility testing, microvascular occlusion simulation, and RBC–endothelial interaction analysis. Uniquely, it constructs a tiered translational roadmap integrating microfluidics with multi-omics, artificial intelligence, vascularized multi-organ chips, and function-centered pre-transfusion surveillance, and proposes microfluidic strategies to optimize RBC storage regimens. In summary, microfluidic technology overcomes the limitations of conventional assays for RSL quality control, provides an emerging technical platform for RSL mechanistic research, pretransfusion quality evaluation, and donor-specific precise matching, and promotes the transformation of transfusion medicine from time-based empirical management to function-oriented precision practice.

We investigate thermodynamic crossovers for black holes in the supercritical regime beyond the critical point, where small and large black holes become indistinguishable from the conventional viewpoint. We establish a refined supercritical phase diagram that comprehensively characterizes the phases of small, large, and indistinguishable black holes, delineated by 2 supercritical crossover lines. The universal scaling laws of these crossover lines are fully verified using the thermodynamics of Reissner–Nordström-anti-de Sitter black holes in both the standard framework and the extended thermodynamic phase space, where the cosmological constant is treated as pressure, as well as in 4 other black hole systems. Analogies with supercritical crossovers observed in liquid–gas and liquid–liquid phase transitions are discussed. This work can be extended to more complex black hole backgrounds and offers valuable insights into the fundamental nature of black hole thermodynamics.

Sepsis frequently leaves patients with persistent immune impairment that contributes to late mortality, yet the anatomical routes that transmit this response to the lung remain poorly defined. Here, we identify the spleen as a key extramedullary hub orchestrating pulmonary immune paralysis through a spleen–lung axis. Mechanistically, splenic programmed death-ligand 1 (PD-L1)⁺ polymorphonuclear myeloid-derived suppressor cells undergo TFDP1-driven expansion and migrate to the lungs via CXCL2/CXCR2 signaling, establishing an interorgan immunosuppressive circuit. In the lung, these cells reprogram alveolar macrophages through programmed death-1 (PD-1)-dependent checkpoint signaling, inducing a dysfunctional state characterized by impaired antibacterial responses and enrichment of complement and immune checkpoint pathways. Genetic or pharmacologic disruption of the PD-L1–PD-1 axis restores macrophage function and improves pulmonary host defense in experimental sepsis. Together, these findings define a mechanistic link between splenic myelopoiesis and distal lung immune paralysis.

Hair loss is common and can markedly impair quality of life, yet the contribution of the scalp microbiome to alopecia remains fragmented across disease subtypes and mechanistic levels. This review aims to construct a clinically relevant and mechanistically coherent framework for the scalp microbiome–hair axis. We first summarize the composition and physiological functions of the scalp microbiome, including its roles in barrier integrity, immune regulation, and nutrient metabolism. We then compare dysbiosis patterns across androgenetic alopecia, alopecia areata, seborrheic dermatitis-associated hair loss, and folliculitis/scarring alopecia. This comparison identifies a shared pathogenic layer—barrier disruption, innate immune activation, altered lipid or metabolite processing, and impaired anagen support—together with subtype-selective mechanisms, including immune-privilege collapse in alopecia areata, androgen–lipid–microbiome coupling in androgenetic alopecia, Malassezia-centered inflammatory ecology in seborrheic dermatitis, and infection/biofilm-driven injury with fibrosis in scarring disease. We further organize the available evidence into a 3-tier signaling cascade linking upstream microbial triggers to midstream NF-κB, JAK–STAT, PI3K/AKT, and Wnt/β-catenin integration and downstream outcomes such as stem-cell quiescence, dermal papilla dysfunction, vascular regression, immune-privilege loss, and fibrosis. We also evaluate probiotics, postbiotics, and engineered microbial systems, whose potential benefits may arise from coordinated anti-inflammatory, ecological, regenerative, and pro-angiogenic effects. However, translation is limited by methodological heterogeneity, weak causal evidence, strain stability, targeted delivery, manufacturing quality control, and regulatory uncertainty. Overall, this framework clarifies shared and subtype-specific roles of scalp dysbiosis and defines priorities for precision, microbiome-guided alopecia therapy.

Cancer-induced bone osteolysis is a common complication of multiple malignancies and may actively contribute to bone metastasis. Its core pathology is closely associated with mitochondrial metabolic dysfunction during osteoclast differentiation. In this study, a mitochondrial transplantation strategy based on in situ biomineralization (Mito@ZIF@RGD) was developed to overcome multiple delivery barriers in differentiating osteoclast precursor cells. A zeolitic imidazolate framework-8 (ZIF-8) shell was formed via the in situ self-assembly of Zn²⁺ and 2-methylimidazole on the mitochondrial membrane, thereby enhancing mitochondrial stability. Meanwhile, cyclic RGD (arginine–glycine–aspartic acid) peptides were coordinated with exposed Zn²⁺ sites on the outer shell to promote αvβ3-mediated uptake during osteoclast differentiation. Furthermore, the sustained Zn²⁺ release from the ZIF-8 biomineralization reshaped intracellular ionic homeostasis, thereby improving the durability of therapeutic efficacy following mitochondrial transplantation. In vitro experiments demonstrated that ZIF-8 encapsulation stabilized mitochondria and enabled sustained adenosine triphosphate production for more than 48 h. RGD modification improved cellular uptake efficiency by approximately 55% in differentiating osteoclast precursors, while the mildly acidic microenvironment triggered the coordinated release of mitochondria and Zn²⁺, effectively reducing intracellular reactive oxygen species levels and osteoclast formation. In vivo, Mito@ZIF@RGD treatment promoted the recovery of bone mineral density, suppressed osteoclast surface area formation by approximately 30%, and preserved bone microstructural integrity. Therefore, as a stable, specific, and durable mitochondrial transplantation platform modulating cellular metabolism during osteoclast differentiation, Mito@ZIF@RGD represents a stable and promising platform for the treatment of cancer-induced osteolysis and other metabolic imbalance-associated bone diseases.

Biological groups exhibit an intrinsic capacity for self-organization, engendering complex and intelligent cooperative behaviors—a capability highly sought after for robotic swarms. This paper proposes an architecture for constructing swarm intelligence in cooperative tasks, termed Cooperation Emergence and Strategy Generation (CE&SG). The architecture empowers robot swarms with the ability to dynamically synthesize composite strategies through self-organized interactions, setting it apart from traditional passive or fixed approaches. CE&SG is inspired by biological observations of cooperation in primate groups, from which we extract an explanatory mechanism for self-organized strategy generation and implement it in robot swarms using a hyper-heuristic algorithm. Through this approach, we demonstrate that sophisticated swarm cooperation can emerge via continuous strategy generation. Concurrently, the framework addresses long-standing challenges in robotics, including adaptability to dynamic changes, system scalability, and strategic interpretability. In a proof-of-concept task of air defense suppression, the proposed approach outperforms several existing advanced methods. The task is executed collaboratively by robots with diverse functionalities. We tested scalability and fault tolerance mechanisms against multiple disruptive disturbances at scales of up to 250 robots and 1,500 tasks in scheduling-oriented discrete-event simulations. Physical feasibility is further validated in a high-fidelity environment with continuous kinematics and collision constraints. The evolutionary patterns of strategies are analyzed through case studies.

Genetically engineered probiotic bacteria are promising living therapeutics. However, their therapeutic potential is frequently confined to their colonization sites, and limitations in efficacy, controllability, and robustness remain major barriers to translation. Anti-inflammatory cytokines play a crucial role in mitigating the pathogenesis of inflammation-related diseases, yet their application is limited by their vulnerability in the mammalian gastrointestinal tract. In this regard, we first found that interleukin-37b (IL-37b) mitigated inflammatory responses at picogram concentrations in primary intestinal cells, supporting its translational potential for the oral treatment of inflammatory diseases. Then, we harnessed the probiotic bacterium Escherichia coli Nissle 1917 as a chassis and further engineered it into a hypervesiculating strain, ΔECIL-37b, capable of efficiently secreting IL-37b-bearing bacterial outer membrane vesicles (OMVs) in situ. These OMVs facilitated IL-37b penetration across the intestinal barrier, enabling it to reach the inflamed mucosa and interact with target cells. In both acute and chronic murine colitis models, we demonstrated that ΔECIL-37b achieved anti-inflammatory efficacy comparable to direct IL-37b injection while enabling local intestinal delivery. Mechanistically, ΔECIL-37b exerted anti-inflammatory effects by attenuating myeloid differentiation primary response 88 (MyD88)-related downstream extracellular-signal-regulated kinase/nuclear factor κB signaling and inflammatory-immune cross-talk. Together, these findings provide a proof-of-concept strategy to harness the anti-inflammatory activity of IL-37b via probiotic-derived OMV delivery and offer insights into a precision anti-inflammatory approach integrating cytokine and probiotic functions.

While critical for human–machine interfaces, on-skin touch sensors struggle to replicate human skin’s curved deformation mechanics and achieve simultaneous high sensitivity, linearity, robustness, and stability. Existing capacitive, piezoresistive, piezoelectric, or triboelectric mechanisms cannot effectively harness touch-induced curved indentation. We introduce a skin-inspired flexoelectret touch sensor that exploits flexoelectricity—converting strain gradients into electricity—to mimic skin’s curved deformation geometry. This enables self-powered, self-calibrated tactile imaging through a novel mechanism establishing linear correlation between inhomogeneous deformation and electrical signal. The sensor achieves high sensitivity (the highest sensitivity up to ~835 mV/N), good linearity, exceptional robustness (>80,000 s continuous operation; >8,000 cycles), and immunity to temperature fluctuations, ultraviolet light illumination, and humidity interference. A 7 × 7-pixel array demonstrates shape-adaptive and touch-trajectory imaging capabilities. This technology unlocks transformative potential for wearable electronics and advanced human–machine interactions.

D-peptides are attractive therapeutic modalities because they are generally more resistant to proteolysis than their L-counterparts, yet systematic design of target-binding D-peptides remains nontrivial. Here, we report Mirror-Peptidizer, an end-to-end in silico mirror-image screening workflow that generates D-peptide binders without requiring chemical synthesis of D-protein targets. The workflow mirrors an L-protein structure to a virtual D-protein, designs L-peptide backbones in the presence of the mirrored target using a diffusion-based backbone generator, selects sequences with a neural sequence design model, and explores local sequence neighborhoods via Bayesian multi-objective optimization balancing sequence-backbone compatibility and a solubility heuristic. Mirroring the resulting complex yields the corresponding D-peptide predicted to bind the native L-target. Using MDM2, PD-L1, and interleukin-23 receptor (IL-23R) as test cases, we identified D-peptides spanning α-helical, β-rich, and mixed conformations with affinity from 11.9 nM to sub-μM. For the MDM2 system, the ¹H-¹⁵N HSQC (heteronuclear single quantum coherence) perturbations and protein mutagenesis support the designed interface, and cell-penetrating conjugates show p53-dependent cancer growth inhibition. Similarly, PD-L1-targeting D-peptides potently inhibited PD-1/PD-L1 interactions in competitive binding assays in vitro, and IL-23R-targeting D-peptides inhibited IL-2/IL-12/IL-23-induced interferon-γ production in human peripheral blood mononuclear cells. Mirror-Peptidizer is provided as an open-source implementation to facilitate rapid generation of experimentally testable D-peptide starting points.