Iron(III)-quercetin complex (IronQ) has previously been shown to support the ex vivo expansion of peripheral blood mononuclear cell (PBMC)-derived proangiogenic cells with CAC-like features, enable MRI-based cell tracking, and enhance proangiogenic activity. However, the intracellular proteomic changes associated with IronQ treatment remained uncharacterized. In this study, we aimed to define the intracellular protein programs of IronQ-treated PBMC-derived CAC-like proangiogenic adherent cells at days 7, 14, and 21 using LC-MS/MS-based quantitative proteomics, with targeted validation of selected proteins as an exploratory ex vivo model. IronQ treatment was associated with significant upregulation of ferritin heavy chain (FTH1) and ferritin light chain (FTL), consistent with intracellular iron sequestration. Stress-response and exocytosis-related proteins were concurrently downregulated, suggesting adaptive maintenance of cellular homeostasis during prolonged culture. Pathway analysis identified stage-specific upregulation of proteins implicated in vascular development at day 7 (8 proteins) and day 14 (14 proteins), including regulators of cell migration, blood vessel morphogenesis, and extracellular matrix remodeling. Overall, these findings provide a proteomic framework for understanding IronQ-mediated proangiogenic cell conditioning and identify candidate protein targets for subsequent functional validation.

Soil salinity severely constrains rice productivity worldwide. As an essential phytohormone, ethylene participates in the developmental processes and stress adaptation of plants; however, the regulatory mechanisms linking ethylene biosynthesis to salt tolerance in rice remain incompletely understood. Here, we demonstrate that the ethylene biosynthesis gene OsACO1 and the AP2/ERF transcription factor OsERF101 negatively regulate salinity tolerance in seedlings. The CRISPR/Cas9-mediated knockout mutants, osaco1 and oserf101, exhibited significantly enhanced survival under 150 mM NaCl treatment, along with reduced reactive oxygen species (ROS) accumulation and enhanced antioxidative enzymatic activity. Transcriptome profiling revealed that salt-treated osaco1 seedlings displayed extensive transcriptional reprogramming, including upregulation of genes involved in ion homeostasis, ROS detoxification, hormone signaling (JA/ABA/ET), transcriptional regulation, and stress defense. A salt-responsive module associated with the osaco1 genotype was detected using weighted gene co-expression network analysis. Promoter motif enrichment and intersection analyses highlighted AP2/ERF transcription factors as key regulatory candidates. The direct interaction between OsERF101 and the GCC-box element within the OsACO1 promoter and the transcriptional activation of OsACO1 expression were confirmed through yeast one-hybrid, dual-luciferase reporter, and electrophoretic mobility shift assays. Collectively, our results support a model in which OsERF101 positively regulates OsACO1 expression, thereby modulating ethylene-associated ROS homeostasis and hormone crosstalk to fine-tune salt stress responses. Our findings elucidate the mechanisms underlying the ethylene-driven regulation of salinity resistance in rice and identify potential targets for molecular breeding of salinity-tolerant cultivars.

Adaptive bioelectronics that autonomously adjust to environmental changes represent an emerging paradigm, enabling reliable operation across diverse conditions. Hypoxic microenvironments are prevalent across numerous pathological conditions including chronic wounds, tumors, and ischemic tissues, creating a fundamental challenge: oxygen-dependent enzymatic biosensors fail precisely when monitoring is most critical, while oxygen deficiency simultaneously impairs tissue regeneration. We present a soft wireless Hypoxia-Adaptive Sensing and Therapeutic (HAST) system that maintains reliable biosensing functionality in oxygen-deficient environments through integrated oxygen management. Using engineered poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS)/polydopamine (PDA)/enzyme biocomposites, HAST integrates multiplexed biosensing (glucose, uric acid, lactate) with dual-oxygen provision: wound exudate-triggered dissolved oxygen generation restores biosensor functionality while electrical stimulation promotes vascular regeneration for sustained tissue oxygenation. This hypoxia-adaptive design achieves about 10-fold biosensor sensitivity enhancement under oxygen-deficient conditions while promoting tissue repair. In preclinical diabetic wound models, HAST enabled accurate continuous monitoring with ∼30% accelerated wound closure, demonstrating environment-adaptive bioelectronics for precision medicine in oxygen-deficient pathological conditions.

Background: Left ventricular (LV) mass and left atrial (LA) strain are physiologically coupled and individually predict adverse cardiovascular outcomes. Their imbalance may reflect early atrioventricular uncoupling and atrial myopathy before overt structural disease. Aims: We evaluated whether a cardiac magnetic resonance (CMR)-derived LV mass-to-LA strain ratio identifies early atrioventricular mechanical uncoupling and predicts cardiovascular events. Methods: Participants from the Multi-Ethnic Study of Atherosclerosis (MESA) underwent CMR at Exams 1 (2000-2002) and 5 (2010-2012). The LV mass-to-LA strain ratio was calculated as indexed end-diastolic LV mass divided by peak LA strain. Classification and regression tree analysis identified cut points for baseline ratio and longitudinal change, defining low (<1.04), reference (1.04-3.26), and high (>3.26) groups, and lesser-(≤130%) versus greater-change (>130%) groups. Multivariable Cox models assessed associations with heart failure (HF), atrial fibrillation (AF), myocardial infarction (MI), and all-cause mortality. Results: Among 4,232 participants (mean age 61.5±10.1 years; 52.8% female), the mean ratio was 1.95±0.72. Over a median 18-year follow-up, 6.0% developed HF, 5.2% MI, 18.7% AF, and 25.6% died. A high LV mass-to-LA strain ratio independently predicted HF (HR 2.85), AF (HR 2.00), MI (HR 1.96), and death (HR 1.70; all P≤.003). With repeat CMR, an increasing ratio also predicted cardiovascular events, including among individuals with normal LV mass and LA strain. Higher baseline ratios were associated with progression toward abnormal volumetric atrioventricular coupling over time. Conclusion: A higher LV mass-to-LA strain ratio and its increase independently predict cardiovascular events before overt chamber abnormalities, supporting impaired atrial adaptation as an early stage of atrial myopathy.

The pig (Sus scrofa) is an important model for evolutionary, comparative, and translational research, however, current functional genomic resources in pigs remain largely limited to one-dimensional genomic annotation and are therefore insufficient for systematically resolving regulatory region-gene relationships, particularly distal ones. Here, we present the Pig Matrix database, a comprehensive 3D regulatory genomics database for pigs, available at https://pigmatrix.kiz.ac.cn/. Built on a standardized experimental framework, Pig Matrix integrates matched multi-omics datasets across tissues, developmental stages, and porcine cell lines, including genomic, transcriptomic, epigenomic, and 3D genome information. In total, it contains 16 library types across 7 omics layers and 7,959 processed files from 1,170 libraries. By integrating epigenomic and 3D genome information, Pig Matrix links cis-regulatory elements to putative proximal and distal target genes, thereby facilitating interpretation of non-coding variants and genomic signals. This database provides modules for genes, candidate cis-regulatory elements, 3D genome architecture, genome browsing, and single cell transcriptomics, together with dedicated Evolution and Comparative resources and user-oriented Genome Annotation and LiftOver tools. A representative use case illustrates how 3D regulatory annotation extends interpretation beyond linear annotation alone, recovering additional candidate genes in domestication-related signals, notably including the classical domestication gene KIT. Pig Matrix also incorporates xenotransplantation-related resources and may support benchmarking of AI models for regulatory genomics. Together, Pig Matrix provides an integrated platform for regulatory interpretation, evolutionary analysis, and comparative genomics in pigs.

Climatic adaptation of plants involves the adjustment of carbon transport to achieve an optimal balance between carbon assimilation and carbon utilization. This study describes a potential molecular mechanism of this adjustment by investigating the climate-dependent regulation of SUCROSE TRANSPORTER 2 (SUC2) in the model plant Arabidopsis thaliana. SUC2 is responsible for loading photosynthetically-produced sucrose into the leaf vascular system. Its protein and mRNA abundance in A. thaliana natural accessions were found to correlate with leaf sucrose export and summer temperature. Exploration of methylome data and epigenome editing experiments suggested that epigenetic marks on SUC2's exon 2 contributed to the climate-dependent expression of SUC2. Two siRNAs were identified as mediators of SUC2 methylations and their respective origins localized to the SUC5 and SUC8 genes. The analysis of plants with reduced or enhanced SUC5 or SUC8 transcriptional activity corroborated the link to SUC2 methylation. Besides illustrating a novel regulatory strategy of carbon export from leaves, our results establish exon methylation mediated by gene-derived siRNAs as yet another potential mechanism for plant adaptation.

Rainfall substantially degrades highway traffic safety by reducing pavement friction, shortening driver perception distance, and amplifying the adverse effects of longitudinal grade on vehicle dynamics. To support refined speed management under rainy-weather conditions, this study develops a dynamic safe speed-limit model that jointly accounts for pavement friction, longitudinal slope, and perception distance. An improved Intelligent Driver Model (IDM) is further proposed to represent rainy-weather car-following behavior and speed adaptation. The proposed model was evaluated using the Hangzhou West-Fuxi highway section, and simulation experiments were conducted under different rainfall intensities and slope conditions. Model validation showed that the simulated speeds were consistent with observed speed characteristics, with RMSE values of 2.13-3.08 km/h and MAPE values of 2.31%-4.12%. The safety evaluation results indicated that the proposed speed-limit strategy reduced conflict rates by 7.1%, 73.6%, and 82.4% across the three rainfall scenarios compared with the unrestricted-speed condition. Sensitivity analysis further showed that increasing rainfall intensity reduced the pavement friction coefficient, increased braking distance, and narrowed the safety boundary, particularly on downhill sections. The results indicate that no additional speed restriction is generally required for rainfall intensities of 0-1.0 mm/min, whereas rainfall intensities of 1.0-5.0 mm/min require progressively stricter speed limits. A longitudinal slope of +1% provides the most favorable operating condition, while a -3% slope represents the most adverse case. These findings provide a quantitative basis for adaptive highway speed-limit control under rainy-weather conditions.

For medical image segmentation, accurately balancing local details and global long-range dependencies is critical to tackling thyroid nodule challenges (variable sizes, ambiguous boundaries, complex context). Traditional CNNs excel at local feature extraction but are constrained by local receptive fields, hindering efficient global dependency modeling. To address this, we propose a Parallel Mamba Dual-U Network (PM-DUNet). It adopts a cascaded dual U-Net encoder-decoder for two-stage "coarse-to-fine" segmentation refinement. We design a Multi-Path Parallel Mamba (MPM) module-using State Space Models (SSMs)-to efficiently model global context with linear complexity. Additionally, Squeeze-Excitation Downsampling (SED) and Spatial Attention Upsampling (SAU) modules are integrated to adaptively enhance key features in encoding/decoding. Results show PM-DUNet achieves highly competitive performance and outperforms state-of-the-art methods on most core metrics, verifying its effectiveness and robustness for complex medical image segmentation. Our code is available on https://github.com/Andrevict/MPDUNet.

This paper proposes a text clustering model based on mutual information-enhanced contrastive learning combined with Kolmogorov-Arnold Networks (KAN) to address challenges in text clustering, including insufficient robustness of text representations, feature redundancy, and the curse of dimensionality. Text clustering is essential for organizing unstructured textual data, yet existing methods often suffer from weak feature representations and limited scalability in high-dimensional spaces. The proposed model jointly addresses these three core issues by maximizing mutual information to capture nonlinear relationships among positive samples, expanding negative samples to enhance discriminability, and employing the KAN architecture to adapt to the complex structures of high-dimensional data. Experimental results on eight benchmark datasets demonstrate that the proposed model achieves state-of-the-art accuracy on seven out of eight datasets and leads normalized mutual information on six datasets, outperforming existing text clustering baselines. To validate its practical utility, the model was applied to cluster Weibo posts collected via web crawlers using "technology" as the keyword during January and February 2025. The case study results reveal that the model effectively identifies meaningful thematic clusters-such as AI applications, technology industry trends, and consumer electronics discussions-confirming its applicability to real-world social media data.

Older adults exhibit heightened vulnerability to heat-related morbidity, yet it remains unclear whether short-term heat acclimation modifies integrative physiological stress responses in aging. We tested whether seven days of passive heat acclimation via warm water immersion alters basal concentrations and acute exercise-heat stress responses of biomarkers reflecting fluid regulation, gastrointestinal epithelial integrity, renal stress, and inflammation in older males. Twelve habitually active older males (median [IQR] age: 68 [64-73] years; V̇O2peak: 34.1 [29.4-36.1] mL O2·kg-1·min-1) completed an exercise-heat stress test before and after seven consecutive days of warm water immersion (~40°C), with rectal temperature maintained at ~38.5°C for the final 60 min. Blood samples were collected before and after each exercise-heat stress test and analysed using linear mixed-effects models (Day x Time). Repeated warm water immersion increased resting renin by 68% [95% CI: 29, 120; P<0.001], while neutrophil gelatinase-associated lipocalin and lipopolysaccharide binding protein decreased by 24% [-40, -4; P=0.023] and 22% [-37, -3; P=0.026] respectively. Exercise heat stress increased aldosterone by 45% [12, 87%, P=0.004], renin by 41% [8, 84%, P=0.01], intestinal fatty acid-binding protein by 52% [26, 85%, P<0.001], and plasma osmolality by 5.6% [2.5, 8.7%, P<0.001]. However, there was no day x time interaction for any biomarker (all P>0.05), indicating that seven days of warm water immersion did not alter the magnitude of the exercise heat stress response. Short-term passive heat acclimation in older males induces selective fluid-regulatory endocrine adaptations consistent with preserved renin-angiotensin-aldosterone system plasticity, while gastrointestinal and renal stress responses to exercise-heat exposure remain unchanged.

Label noise remains a fundamental challenge in supervised deep learning, particularly as modern neural networks rely on large-scale datasets that frequently suffer from annotation errors due to cost and scalability constraints. While prior approaches address this issue through collaborative training on fully noisy data or meta-learning with access to a small clean subset, the paradigm of clean-label-free label purification remains insufficiently explored, which is particularly critical in expert-annotated domains such as remote sensing and medical imaging, where annotation inconsistencies persist even among trained experts. In this work, we propose a clean-label-free meta-learning framework for label purification. Our approach introduces a multi-round purification strategy that iteratively refines label quality by alternating between base-model training and meta-purification in a decoupled manner, thereby preserving training stability. We further analyze the multi-round purification process under a stochastic contraction perspective and introduce a KL-divergence-based stopping criterion to adaptively select the purification depth. As a result, our framework enables flexible downstream training without reliance on complex noisy-label learning pipelines. Moreover, it captures uncertainty through accumulated label distributions across rounds, improving the reliability of mislabeled instance detection and correction. Extensive experiments on benchmark datasets, including CIFAR and mini Clothing-1M, show that our method consistently outperforms state-of-the-art techniques. Notably, it achieves up to a 62.61% label accuracy improvement on expert-annotated remote sensing datasets (UC Merced and AID), demonstrating its robustness in real-world, noise-prone environments. The source code is available at https://github.com/mitsuyoshiohtsu/MRMP.