搜索结果：Cell discovery

This commemorative article reflects on a research journey spanning neural development, stem cell biology, regenerative medicine, and iPSC-based drug discovery. My early work focused on RNA-mediated regulation in the nervous system, including studies on myelin basic protein gene regulation and the identification and functional characterization of the RNA-binding protein Musashi. These studies contributed to the conceptual foundation of neural stem cell biology and helped establish methods for identifying and isolating neural stem/progenitor cells, including those present in the adult human brain. Building on this foundation, my colleagues and I pursued translational research in spinal cord injury, ranging from analyses of injury pathophysiology and molecular interventions to preclinical studies using rodent and non-human primate models. These efforts ultimately led to the first-in-human clinical study of induced pluripotent stem cell-derived neural stem/progenitor cell transplantation for subacute spinal cord injury. In parallel, we developed patient-derived iPSC platforms for neurological disease modeling and drug discovery, particularly for amyotrophic lateral sclerosis, where iPSC-based screening identified Ropinirole as a therapeutic candidate and enabled reverse translational research linking cellular phenotypes with clinical responses. Looking ahead, I argue that the future of regenerative therapy will depend on the continued integration of developmental biology, stem cell science, disease modeling, rehabilitation, and clinical translation to address unmet medical needs in disorders of the central nervous system.

Chronic inflammatory demyelinating polyneuropathy (CIDP) is an immune-mediated peripheral neuropathy with heterogeneous and often incomplete responses to current immunotherapies, but the underlying immune basis remains poorly defined. Although CIDP shares features of immune-mediated demyelination with multiple sclerosis (MS), the two diseases affect distinct anatomical compartments and exhibit divergent therapeutic responses, suggesting fundamentally different underlying immune programs. Here, we address this gap by defining the peripheral immune architecture of CIDP using an integrated, multi-modal approach. Peripheral blood was obtained from 20 patients with CIDP and 20 age- and sex-matched healthy controls. Single-cell RNA sequencing was performed in a discovery subset and integrated with publicly available MS peripheral blood datasets to provide a cross-disease reference framework. The single-cell analysis was designed as an exploratory discovery step to identify candidate immune signatures. Transcriptomic, pathway, and ligand-receptor analyses were complemented by cytokine profiling and flow-cytometric validation in the full cohort. CIDP exhibited broad inflammatory activation with preferential enrichment of type I interferon and inflammasome-related programs compared with MS. Despite reduced B-cell frequencies, CIDP showed transcriptional enrichment of germinal center-associated programs, indicating a dissociation between cell number and activation state. In parallel, CD8 effector T cells demonstrated enhanced cytotoxicity and cytoskeletal remodeling programs, supported by increased expression of actin-regulatory genes and strengthened intercellular signaling interactions. In contrast, MS showed greater enrichment of integrin-talin-vinculin signaling pathways in B cells and CD4 T-cell subsets, consistent with trafficking-related immune mechanisms. Together, these findings indicate a coordinated immune axis linking B-cell dysregulation and cytotoxic CD8 T-cell activation in CIDP. Integrated peripheral immune profiling identified candidate CIDP-associated immune signatures including dysregulated B-cell activation despite numerical reduction and a prominent cytotoxic CD8 T-cell program within a type I interferon- and inflammasome-skewed inflammatory milieu. These findings provide an exploratory framework for understanding peripheral immune dysregulation in CIDP and warrant further translational studies in larger, treatment-stratified cohorts.

Substance use disorders (SUDs) impose major global morbidity and mortality, yet the cellular mechanisms linking genetic risk to neural vulnerability, disrupted neurodevelopment, and drug-induced neuroadaptations remain poorly understood. Human pluripotent stem cell (hPSC) technologies, including embryonic and induced pluripotent stem cell (ESC/iPSC)-derived neurons, three-dimensional (3D) brain organoids, and organoid-on-a-chip platforms, provide scalable, human-relevant models to address this gap. ESC/iPSC-derived neuronal cultures enable interrogation of genetic and epigenetic determinants of drug susceptibility and response; cerebral organoids recapitulate tissue architecture and emergent network dynamics; and microfluidic organoid-on-a-chip systems facilitate maturation, enhance reproducibility, and enable controlled exposure paradigms. In this review, we synthesize recent stem cell-based studies of alcohol, opioid, and stimulant exposure, highlighting insights into neurodevelopmental disruption, synaptic and signaling alterations, neuroinflammation, and network-level dysfunction. We critically evaluate limitations of stem cell-based studies, including ethical concerns, inter- and intra-line variability, incomplete cellular maturation, and difficulties in modeling complex circuitry and comorbid conditions. We propose strategies to enhance translational relevance, including standardized differentiation protocols, addition of patient-derived cells and vascular and immune components, and integration of multi-omics approaches (transcriptomics, epigenomics, and proteomics) with functional readouts to map molecular pathways underlying drug vulnerability and resilience. Finally, we outline the therapeutic and precision-medicine potential of stem cell platforms for target discovery, predictive toxicology, and individualized treatment modeling. Despite remaining challenges, stem cell-based approaches offer a powerful and increasingly tractable path from genetic association to mechanistic insight and therapeutic innovation in addiction research.

Epidermal growth factor-like domain 7 (EGFL7) was discovered as an extracellular matrix protein with an EGF-like domain and angiogenic and vasculogenic functions. It has also been shown to play a role in immunological evasion of tumor cells through attenuation of extravasation of immune cells by reducing the expression of cell adhesion molecules on vascular endothelial cells. Furthermore, microRNA-126 (miR-126), which is encoded within the intron of Egfl7, has been reported to be a vasculogenic factor. However, its immune-related functions in tumors remain unclear. Here, we examined the roles of tumor-derived EGFL7 and miR-126 in in vivo tumor progression using Egfl7/Mir-126 knockout and rescue tumor cell lines. Tumor growth was significantly suppressed in Egfl7/Mir-126-deficient cells and was restored by re-expression of miR-126, but not EGFL7. These differences in in vivo tumor growth were not observed in immunodeficient mice, suggesting the involvement of the adaptive immune system. CD4+ or CD25+ cell depletion suppressed the growth of miR-126-expressing tumors, whereas CD8+ cell depletion enhanced the growth of miR-126-deficient tumors. Histological analysis revealed an increased Foxp3+/CD8+ cell ratio in miR-126-expressing tumors during the early phase of tumor establishment. Bilateral tumor models further demonstrated that miR-126-expressing tumors promoted the growth of contralateral miR-126-deficient tumors, and a similar effect was observed using apoptotic miR-126-expressing tumor cells In vitro analyses using extracellular vesicles derived from dying tumor cells showed the transfer of miR-126 to CD4+ T cells and a higher proportion of Foxp3+ cells. Together, these findings suggest that tumor-derived miR-126 promotes tumor progression through non-local immune modulation, potentially involving maintenance of Treg-associated populations.

Covering: 2013 up to 2026The rapid escalation of antimicrobial resistance has outpaced the discovery of natural product (NP)-derived antibiotics, underscoring the need for new strategies to identify and characterize antibacterial agents. NPs have historically dominated antibiotic development due to their structural and mechanistic diversity, yet their chemical complexity, low abundance, laborious dereplication, and challenging mode of action (MoA) identification continue to limit discovery efficiency. Bacterial Cytological Profiling (BCP), an image-based phenotypic screening, has emerged as a powerful approach capable of capturing rich, single-cell-resolved responses, making it particularly well suited for investigating complex NP-derived antibiotic discovery. This review focuses on how BCP, a prokaryotic image-based strategy central to antibiotic MoA elucidation, addresses key challenges in NP-derived antibacterial discovery, including dereplication, deconvolution of multiple MoAs, and identification of novel mechanisms. We also discuss current biological and methodological limitations of the approach and provide practical perspectives for NP researchers seeking to implement BCP in discovery programs. Finally, we examine how lessons from the widely adopted eukaryotic image-based profiling method, the Cell Painting Assay, could inform future BCP development and propose a "BCP v1.0" framework to facilitate broader adoption and ultimately accelerate NP-derived antibiotic discovery.

The ability to treat Helicobacter pylori (H. pylori) infection and eliminate its associated gastric cancer risk is highly desirable but has proven to be extremely difficult. In this study, pilot proteomic screening of clinical gastric mucosal samples suggested a progressive decline in Sirtuin 1 abundance along the H. pylori-associated pathological cascade. Based on these findings, gastric epithelial cells-localizable oral nanomedicines (GLONs) are developed, whereby H. pylori eradication and reversal of precancerous intestinal metaplasia (IM) are simultaneously achieved via Sirtuin 1 restoration. GLONs are constructed by coating resveratrol, lactoferrin, and disulfide modified-fucoidan (DFu) co-assembled nanoparticles (RLF) with engineered mucin-overexpressing gastric epithelial cell membranes. The shell of GLONs resists gastric acid, enhances mucus penetration and epithelial cells uptake. After internalization, DFu undergoes oxidative destabilization in the H2O2-enriched infectious microenvironment induced by H. pylori, thereby triggering RLF core dissociation and subsequent component release. The resveratrol restored H. pylori infection-induced impairment of Sirtuin 1, thereby activating autophagy. Meanwhile, lactoferrin promoted antimicrobial peptide production and synergized with fucoidan-mediated enhancement of antigen presentation, ultimately enabling the clearance of both intracellular and extracellular H. pylori. In metaplastic gastric cells, Sirtuin 1 repairs damaged DNA, and inhibits malignant proliferation. In mouse models, under the tested 7-day regimen, GLONs produced greater reductions in gastric H. pylori burden and more pronounced improvements in IM-related phenotypes than the abbreviated triple-therapy. GLONs represent an innovative and highly efficient therapeutic platform for H. pylori infection and its complications.

Antiphospholipid syndrome (APS) is an acquired autoimmune disorder characterized by recurrent vascular events in large, medium, or small vessels. These events contribute to cardiovascular disease primarily through thrombosis and atherosclerosis (AS). Carotid atherosclerosis (CAS) represents a particularly high-risk manifestation of subclinical AS in patients with APS. However, the shared molecular signatures linking APS and CAS remain unclear. Bulk transcriptome datasets from Gene Expression Omnibus (GEO) were analyzed to identify differentially expressed genes (DEGs) in APS and CAS. Common DEGs were characterized by Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment and protein-protein interaction analyses. Candidate hub genes were prioritized by integrating Least Absolute Shrinkage and Selection Operator (LASSO), random forest, weighted gene co-expression network analysis (WGCNA), and MCODE, followed by diagnostic evaluation in independent datasets. Upstream regulatory networks and immune infiltration were assessed using in silico approaches. Single-cell RNA-seq was used to determine cell-type specificity. Genome-Wide Association Study (GWAS) summary statistics were integrated via MAGMA and intersected with expression signatures and OMIM-curated genes to prioritize additional candidates. A total of 4,264 DEGs were identified in APS and 838 DEGs in CAS, including 52 common DEGs (43 upregulated and nine downregulated). These common DEGs were enriched in plasma-membrane and actin-cytoskeleton-related functions, with nominal KEGG signals involving oxytocin, Jak-STAT, and PI3K-Akt pathways. Cross-method prioritization highlighted CLEC4A and P2RY13, which showed consistent dysregulation and exploratory diagnostic performance across discovery and validation datasets and exhibited stage-associated patterns in carotid plaques. Immune deconvolution suggested that CLEC4A/P2RY13 tracked with myeloid- and mastcell-related signals in CAS and with neutrophil-related signals in APS. Single-cell analysis indicated predominant expression in the mDC/cDC compartment and positive associations with mast-cell proportions in CAS. MAGMA/OMIM integration further implicated TLR8 (APS) and IL18/HAND2 (CAS) as additional candidates. Collectively, our results nominate CLEC4A and P2RY13 as candidate genes potentially involved in shared APS-CAS immune-related pathways, with expression changes associated with CAS progression among APS patients.

A library of N6-modified adenosine derivatives (compounds 7-36) was synthesized via a one-step reaction of 6-chloro- or 2-amino-6-chloro-(β-D-ribofuranosyl)-9H-purine with primary or secondary amines. The structures of selected compounds (14, 31, 34) were confirmed by single-crystal X-ray diffraction, revealing diverse crystal packing motifs and hydrogen-bonding networks. Biological evaluation demonstrated a broad range of cytotoxic activities across a panel of cancer and non-cancer cell lines, ranging from highly potent to non-toxic derivatives. Structure-activity relationship analysis revealed that the biological properties of the synthesized compounds were strongly influenced by the nature of the N6 substituent, with distinct structural features governing anticancer and antiviral activities. Several compounds (21, 22, 28, 30, and 32) markedly reduced cancer cell viability, with compound 28 exhibiting the highest potency across multiple solid tumor models. Importantly, compound 28 induced apoptosis, suppressed proliferation, and rapidly decreased cell viability in both HCT116 and HCT116 p53-/- colorectal carcinoma cells, indicating p53-independent activity. In contrast, cladribine displayed pronounced p53 dependence, promoting apoptosis primarily in p53-proficient cells and inducing a senescence-like phenotype rather than rapid cell death. Antiviral screening identified compounds 22, 25, and 35 as promising inhibitors of human parainfluenza virus type 3 (HPIV-3), whereas compound 12 exhibited the highest activity and selectivity against human adenovirus type 5 (AdV5). Collectively, these findings identify N6-modified adenosines as a promising class of compounds for developing anticancer agents with p53-independent activity and reveal several derivatives as attractive scaffolds for further optimization toward antiviral agents targeting respiratory viruses.

Derived iPSCs airway epithelium are challenging given their dependency on the mesenchymal compartment. We hypothesized that growing vAFE cells on well-organized stiff matrix precolonized by adult pulmonary fibroblasts would improve epithelial differentiation yield and maturity. Collagen-1/chitosan matrix were engineered to reach stiffness and scaffolding characteristics of subepithelial compartments. Primary fibroblasts derived from human lung samples were seeded for 45 days before the addition of vAFE cells differentiated from iPSCs, and comparisons made with iPSC-derived fibroblasts. Beads tracking was used to assess cilia beating efficiency. Primary human bronchial fibroblasts were able to enrich the CC Matrix with extracellular matrix components such as collagen, decorin and vimentin. In turn, iALI cultures performed in primary fibroblasts enriched CC matrix successfully led to high level of epithelial differentiation including rare cells (club, basal, neuroendocrine, ciliated, secretory). Large apical surfaces were covered by approximately 60% of ciliated cells able to generate mucociliary vortex. Primary human bronchial fibroblasts seeded into collagen-chitosan matrix dramatically improved iALI epithelial differentiation from vAFE cells, related to highly specific transcriptomic signatures when compared to iPSC derived fibroblasts.

Inflammatory diseases of connective tissues such as periodontitis and rheumatoid arthritis exhibit localized destruction of matrix collagen and loss of tissue function. These diseases are driven by interconnected signaling pathways that determine disease progression and severity. Both periodontitis and rheumatoid arthritis involve the release of extracellular Vimentin (ECV) from stromal and immune cells at diseased sites, but the processes by which ECV binds to cells and promotes inflammatory signaling are not well defined. Recent data point to several putative ECV receptor proteins, one of which is Leucine-Rich Repeat Containing 15, an orphan receptor and cancer-associated fibroblast marker that contributes to inflammation and matrix destruction. Here, we consider the roles of ECV and LRRC15 in connective tissue diseases and discuss how their interactions may promote matrix destruction. We propose that ECV and LRRC15 signal through β1-integrin/FAK, Wnt/β-catenin, and NF-κB to promote cell adhesion, migration and matrix degradation by local fibroblast and immune cell populations. To obtain further insights into ECV-LRRC15 engagement, we used in silico modelling to predict the most likely binding conformation of the ECV-LRRC15 interaction with MEGADOCK. The most probable model suggests that the convex face of the LRRC15 leucine-rich repeat loop binds to two sites on the α-helices of Vimentin rod domains and to one site on Vimentin's N-terminal head domain, potentially signaling through this interface. In this review, we provide an in-depth overview of the functional links between ECV and LRRC15 and discuss their potential roles as drivers of matrix destruction in periodontitis and rheumatoid arthritis.

Aurora kinase A (AURKA) is a pivotal driver of malignant progression and poor prognosis in triple-negative breast cancer (TNBC). In this study, we developed a cascaded AI-driven virtual screening pipeline, integrating sequence-based affinity prediction (PSICHIC), equivariant deep learning docking (KarmaDock), and geometric rescoring (DeepDock) to identify novel AURKA inhibitor candidates. From an in-house 160,000-compound screening library assembled from commercially available collections, three leads (compounds 3, 5, and 8) were selected and subsequently validated via HTRF biochemical assays, exhibiting potent enzymatic inhibition with IC50 values of 157 nM, 21.64 nM, and 46.03 nM, respectively. Cell-based assays demonstrated that compound 3 produced stronger short-term cell-growth inhibition in MDA-MB-231 (TNBC) cells compared to clinical benchmarks MLN8237 and CCT241736, whereas compounds 3 and 5 showed cell-growth inhibition in NIH/3T3 cells within the same concentration range as the reference inhibitors. Triplicate 500 ns molecular dynamics simulations supported stable binding modes of the identified leads in the AURKA binding pocket. Additional computational analyses further provided supportive information for subsequent lead optimization. This study provides a transparent and open-source workflow for AI-assisted identification of AURKA-active chemotypes.

We have created a new data-analysis pipeline for the discovery of host-derived candidate biomarkers in blood cell-free DNA sequencing data. Unlike approaches that rely on specific molecular or genetic signatures, our method leverages the coverage distribution of cell-free DNA sequences mapped to a reference genome, applying statistical analyses to identify informative short genomic regions for biomarker discovery. The pipeline is applicable to diverse diseases and can be used to analyze cell-free DNA sequences from plasma or serum to identify candidate biomarkers that are characteristic of disease states in mammals. Core functionalities were developed in Java and integrated with open-source software tools for the preprocessing of raw sequencing data, complemented by Python scripts for the machine-learning analysis and statistical validation. The pipeline is designed for HPC use and users can access the pipeline through a Galaxy workflow, which offers a user-friendly web interface for input selection prior to execution and analysis progress monitoring. Performance tests, carried out using duplicate sets of COVID-19 samples and controls, showed linear scalability of execution time with an increasing dataset size, as well as a substantial reduction in execution time through parallelized computation, whereby each HPC node is used to process the data of one chromosome. Further statistical tests confirmed the quality of the pipeline's results by showing that the set of identified candidate biomarkers remained stable across varying dataset sizes.

Primary liver cancer is a major global cause of cancer death, and hepatocellular carcinoma (HCC) is the predominant histological subtype. This literature review synthesizes current evidence on the exposome, multi-omics landscape, and artificial intelligence (AI)-based integration strategies relevant to biomarker discovery in liver cancer, with a focus on biological rationale, emerging clinical applications, and translational limitations. Key etiologic drivers include viral hepatitis, alcohol-related liver disease, and metabolic dysfunction-associated steatotic liver disease, all of which interact with environmental exposures across the life course. Biomarker discovery increasingly relies on integrated assessment of exposure-related signals together with genomic, epigenomic, transcriptomic, proteomic, metabolomic, and spatially resolved data. Hepatocarcinogenesis involves a complex interplay of chronic liver injury, environmentally patterned molecular perturbation, and dynamic tumor-host interactions. We emphasize an exposome-informed, multimodal strategy in which interpretable AI models identify clinically relevant signatures for early detection, prognostic stratification, and treatment guidance. Critical limitations of current evidence include incomplete exposure assessment, heterogeneous data platforms, retrospective study design, limited external validation, and insufficient model transparency. Emerging approaches, including proteogenomic, lipidomic, single-cell, and digital pathology-based modeling, show promise but require further validation in etiologically diverse cohorts. The purpose of this review is to critically examine how AI can integrate exposome-related information with multi-omics data for biomarker discovery in liver cancer. Here, particular attention is given to the exposure-to-biomarker sequence, immune-metabolic remodeling, liquid-biopsy translation, and the reduction of high-dimensional signatures into clinically deployable assays.

To advance next-generation risk assessment of non-genotoxic carcinogens, robust mechanism-based assays are essential. A recognized mode of action for non-genotoxic carcinogens is induction of oxidative stress leading to regenerative proliferation. Most of the currently available New Approach Methodologies (NAMs) rely on simple high-throughput cell models with limited biological complexity and often lack metabolic capacity. In this study, we quantify chemically-induced oxidative stress in zebrafish embryos (ZFE), Danio rerio, to evaluate the added value of a whole-organism model with functional metabolism over a simple high-throughput hepatocyte cell line. Four-day-old ZFE were exposed to a set of 22 chemicals, including fifteen chemicals inducing oxidative stress and seven with another primary mode of action. Following 24hours of exposure, reactive oxygen species (ROS) were quantified in the ZFE using the dichloro-dihydro-fluorescein (DCFH) assay. Using analytically determined internal concentrations, chemicals were ranked based on their ROS-inducing potential. Results were compared with ROS induction in maturated HepG2 cells, where DCFH fluorescence was quantified during the first hour of chemical exposure. Both models identified nine chemicals with ROS-inducing potency, although the identified chemicals did not completely overlap. Of the fifteen chemicals reported to primarily induce oxidative stress, only four were not detected by either model. In ZFE, three of the seven chemicals reported to have another primary mode of action than oxidative stress were flagged for ROS production, whereas none of these seven were flagged in maturated HepG2 cells. We explore potential explanations for discrepancies between the models and discuss their applicability in a regulatory context.

Liver cancer, particularly hepatocellular carcinoma (HCC), remains a major global health burden due to late diagnosis, limited therapeutic options, and frequent resistance to drugs such as sorafenib, regorafenib, and lenvatinib. Natural products continue to attract attention as a rich source of structurally diverse anticancer candidates. Plant-derived secondary metabolites, including terpenoids, flavonoids, alkaloids, and phenolics, have shown activity against HCC, yet their organisation by plant organ and mechanism of action remains insufficiently addressed. In this review, anti-HCC natural products reported over the past four decades were collected from the Dictionary of Natural Products, PubMed, and Google Scholar. More than 116 compounds were classified by plant sources, including leaves, fruits, seeds, flowers, bark, and subterranean organs. These compounds mainly act by inhibiting proliferation, inducing apoptosis, modulating the cell cycle, suppressing telomerase activity, and regulating key signalling pathways in HCC cell lines, supporting natural products as valuable leads for anti-HCC drug discovery.

Cancer is a complex disease driven by genetic, metabolic, and environmental alterations, whose investigation is often constrained by the limited tractability of mammalian systems. The budding yeast Saccharomyces cerevisiae has emerged as a powerful eukaryotic model to study conserved cellular processes relevant to tumor biology in a simplified and scalable context and as a versatile platform for translational and biotechnological applications. Through genetic manipulation and heterologous expression, yeast allows systematic analysis of human cancer genes and variants, providing quantitative insights into their functional impact. In parallel, yeast reproduces fundamental features of cancer cell metabolism and stress adaptation, offering a controlled system to investigate cellular responses to environmental constraints. The conservation of major DNA repair and autophagy pathways further supports the use of yeast to study genome stability and survival mechanisms in cancer. Beyond its role in basic research, S. cerevisiae represents a scalable platform for anticancer drug discovery, enabling systematic identification of drug targets, resistance mechanisms, and genotype-specific vulnerabilities through high-throughput and engineered strain-based approaches. Continued development of yeast platforms, together with synthetic biology, functional genomics, and advanced genomic technologies, is expected to accelerate therapeutic innovation and improve our understanding of cancer biology. This review discusses recent advances in yeast-based cancer research, highlighting the contribution of engineered yeast platforms to the investigation of oncogenic signaling, metabolic rewiring, stress adaptation, DNA repair, and autophagy, reinforcing the role of yeast at the interface between cancer research and biotechnology.

Human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) offer a powerful platform for disease modeling, drug discovery, and regenerative therapies. However, their clinical and research utility remains limited by their immature, fetal-like phenotype. In the human heart, postnatal metabolic maturation involves a critical switch from glycolysis to fatty acid β-oxidation, enabling efficient ATP production via oxidative phosphorylation. In this study, we investigated whether mimicking this metabolic shift in vitro by culturing hiPSC-CMs in a fatty acid-based maturation medium (FAM) could enhance their structural and functional development compared to a conventional glucose-based medium (GLM). hiPSC-CMs cultured in FAM for two weeks exhibited significant improvements in morphological, metabolic, and mechanical maturation markers. Morphologically, FAM-cultured CMs exhibited enhanced sarcomeric organization, increased cellular alignment, and a more elongated and rod-like shape, which are characteristics typically associated with mature CMs. Ultrastructural analysis further confirmed improved maturation, revealing more organized sarcomeres and densely packed mitochondria compared to GLM-cultured CMs. Metabolically, FAM-cultured CMs demonstrated a clear shift from glycolytic to oxidative metabolism, as evidenced by higher mitochondrial membrane potential, increased oxidative phosphorylation capacity, elevated ATP production, and reduced glycolytic activity. These metabolic adaptations indicate a more adult-like energy profile, consistent with enhanced fatty acid β-oxidation. Mechanically, FAM-cultured cardiomyocytes exhibited enhanced functional maturity, as evidenced by faster calcium transients and greater contraction amplitude, indicating improvements in specific electrophysiological properties. In conclusion, fatty acid supplementation effectively promotes the structural, metabolic, and mechanical maturation of hiPSC-CMs, resulting in a more adult-like phenotype. This strategy provides a robust and straightforward approach to enhance the physiological relevance of hiPSC-CMs for preclinical applications in disease modeling, drug testing, and regenerative medicine.

Artificial intelligence accelerates anticancer peptides (ACPs) discovery. However, existing computational methods lack integration of identification with activity-based candidate prioritization. Here, we present DeepACPred, a three-stage pipeline encompassing ACP binary classification model, ACP multilabel classification model, and ACP IC50 prediction model, leveraging multimodal features from ESM2 protein language model embeddings, AAindex physicochemical descriptors, and sequence composition. On 5712 benchmark sequences, the binary classifier achieved 95.10% accuracy (AUC = 0.9913), with performance remaining stable under CD-HIT cluster-aware splitting at 40%-90% identity thresholds. Multilabel cancer-type prediction yielded macro-F1 = 0.9124 across seven cancer types, and log10(IC50) regression achieved Spearman ρ = 0.8602 under 5-fold cross-validation. Ablation experiments showed task-dependent feature contributions rather than uniformly additive multimodal effects. Applied to 260 000 motif-enriched 18-mer candidates, DeepACPred selected 12 peptides predicted to be active against breast cancer cells, all of which showed measurable in vitro cytotoxic activity against murine 4T1 cells in OD-derived dose-response assays (IC50: 0.88-36.83 μg/ml). Although prospective IC50 ranking showed limited fine-grained resolution, these results support the use of the regression module for coarse candidate enrichment. In conclusion, DeepACPred provides a systematic framework for ACP candidate enrichment and prioritization.

Polymyxins are often last-resort antibiotics against high-priority Gram-negative pathogens, particularly Acinetobacter baumannii. However, most mechanistic studies to date have overlooked the heterogeneity within bacterial populations during polymyxin treatment. Using time-lapse imaging and propidium iodide (PI) staining, we observed that a subset of PI-positive (PI+) cells, which are traditionally considered non-viable, were capable of regrowth. This unexpected scenario promoted further investigation into the distinct metabolic responses of PI+ or PI-negative (PI-) subpopulations following polymyxin exposure. By combining a synthetic fluorescent polymyxin probe, FADDI-043, with fluorescence-activated cell sorting (FACS), we isolated PI+ and PI- cells and profiled their metabolic responses. PI+ cells exhibited increased levels of phosphatidylethanolamine, likely to compensate for the severe membrane damage by polymyxins. In contrast, PI- cells, where polymyxin interacted with bacterial membranes without causing extensive damage, demonstrated broader metabolic adaptations. Notably, arginine metabolism was uniquely upregulated in the PI- group, and exogenous arginine supplementation conferred protection against polymyxin treatment. Collectively, this is the first study to demonstrate subpopulation-specific metabolic responses to polymyxins, highlighting dynamic and heterogeneous bacterial adaptations. Our findings underscore the importance of single-cell analysis to unravel antibiotic mechanisms and may inform novel metabolic reprogramming strategies to enhance antibiotic efficacy and minimize resistance emergence.IMPORTANCEMultidrug-resistant Acinetobacter baumannii is designated a World Health Organization "critical priority" pathogen, and polymyxins remain the few effective treatment options, particularly in low- and middle-income countries. However, polymyxin heteroresistance poses a major global clinical challenge, and its mechanistic basis remains poorly defined. Most antimicrobial studies rely on population-level measurements, obscuring the biological consequences of phenotypic heterogeneity. Here, we demonstrate that a subset of polymyxin-treated, propidium iodide-positive (PI+) A. baumannii cells, typically classified as non-viable, retain the capacity to regrow. By isolating PI+ and PI-negative (PI-) subpopulations, we uncover distinct metabolic adaptations under polymyxin exposure: PI+ cells exhibit elevated phosphatidylethanolamine levels, whereas the PI- cells activate their arginine metabolism. These findings reveal an unrecognized layer of metabolic heterogeneity underlying polymyxin exposure. Our work challenges conventional interpretations of viability assays and highlights the importance of subpopulation-resolved analyses for identifying metabolic vulnerabilities to enhance polymyxin efficacy while limiting resistance emergence.