Recent advancements in artificial intelligence (AI) have revolutionized the application of electrocardiography (ECG) in cardiovascular diagnostics. This review highlights the transformative impact of AI on traditional ECG analysis, detailing how deep learning algorithms are overcoming the limitations of human interpretation and conventional diagnostic criteria. Historically, ECG interpretation has relied on well-established, physiologically-based criteria. The advancement of AI-ECG is marked by its capacity to process complex high-dimensional data directly from raw signals, revealing patterns often missed by conventional methods. Notably, AI models have identified signs of asymptomatic low ejection fraction and paroxysmal atrial fibrillation during normal sinus rhythm, enabling earlier clinical intervention. In addition to improved diagnostic utility, AI-ECG offers promising applications in risk stratification and community screening. Several randomized controlled trials (RCTs) have shown that integrating AI into clinical workflows not only reduces critical intervention times but also identifies patients at elevated risk of adverse outcomes. Future directions involve integrating additional clinical data sources, improving model interpretability through explainable AI, and developing unified platforms to manage outputs from multiple models.

Relapse in melanoma after targeted or immune therapy necessitates the rapid identification of effective alternatives. To address this gap, we investigated whether the timely generation of preclinical models for functional drug testing could reveal additional therapeutic options. Our study focused on: (i) the feasibility of generating in vivo and in vitro models from melanoma lymph node (LN)-derived disseminated cancer cells (DCCs) before relapse, (ii) the implementation of preclinical models to identify therapeutic alternatives, and (iii) the ability to detect patients who could benefit from early functional in vitro drug testing. Successful model generation was significantly associated with DCC quantity, LN origin, and mortality risk. All patient-derived xenograft models were available before patient death and, in 82% of cases, before relapse. Proof-of-concept in vitro drug screening using 315 anti-cancer drugs identified additional candidates, and coculture of DCCs and LN cells revealed specific T-cell activation and responses to immunotherapy. Our data establish a process for selecting melanoma patients at high risk of progression, enabling the timely generation of patient-derived models to support functionally guided treatment decisions at relapse.

Brain amyloid-β (Aβ) pathology is a core feature of Alzheimer disease (AD) and can be quantified using positron emission tomography (PET). Cerebrospinal fluid (CSF) and plasma biomarkers detect abnormal Aβ, but it is unclear to what degree they can predict quantitative Aβ-PET. We explored plasma and CSF biomarkers in relation to Aβ-PET in the BioFINDER-2 study (N = 1053), and the BioFINDER-1 study (N = 238). We developed a machine learning pipeline to predict Aβ-PET using CSF and plasma measures. The best models achieved R = 0.79. Plasma P-tau217 and CSF Aβ42/Aβ40 contributed the most. CSF Aβ42/Aβ40 contributed most to identify Aβ-positivity, while continuous Aβ-PET load within the positive range was best predicted by plasma P-tau217. Models using only plasma measures approached performance of CSF models. Altered metabolism of soluble Aβ may be highly associated with presence of Aβ plaques, while soluble P-tau217 levels may continue to change during build-up of Aβ pathology.

Salmonella Typhi secretes typhoid toxin that activates cellular DNA damage responses (DDR) during acute typhoid fever. Human infection challenge studies revealed that the toxin suppresses bacteraemia via unknown mechanisms. Using quantitative proteomic analysis on the plasma of bacteraemic participants, we demonstrate that wild-type toxigenic Salmonella induced secretion of lysozyme (LYZ) and apolipoprotein C3 (APOC3). Recombinant typhoid toxin or Salmonella infection recapitulated LYZ and APOC3 secretion in cultured cells, which involved ATM/ATR-dependent DDRs and confirmed observations in typhoid fever. LYZ caused spheroplast formation, inhibited the Salmonella type 3 secretion system, and intracellular infections. LYZ expression was regulated by p53 in a cell type-specific manner and driven by mitochondrial oxidative stress that caused nuclear DDRs and p53-mediated senescence responses. Addition of LYZ inhibited oxidative DNA damage and resulting senescence responses caused by typhoid toxin. Our findings may indicate that toxin-induced DDRs elicit antimicrobial responses, which suppress Salmonella bacteraemia during typhoid fever.

The biomechanical signature is directly correlated with the progression of disease in multiple soft tissues. However, their variations and roles, particularly during the initiation period of the disease, remain unclear. Here, we report that PM2.5 exposure induces corneal biomechanical cues alterations prior to corneal injury, as evidenced by increased corneal hysteresis in humans, thickened corneal thickness in rats, and enhanced tensile stress and cortical stiffness in HCECs. Specifically, intracellular PAI-2 is identified as a crucial mediator of the biomechanical responses in HCECs. It modulates PM2.5-induced autophagy and inflammation through a PAI-2/myosin II/F-actin/YAP-positive feedback loop, which ultimately drives HCEC injury. Furthermore, extracellular secretory PAI-2 levels in tears reflect PM2.5-related corneal damage in real time, making it a specific biomarker for the early diagnosis when combined with biomechanical cues. Early intervention for PM2.5-induced ocular damage could be achieved by developing an LNP-siPAI-2 ocular local delivery system targeting intracellular PAI-2. Overall, we propose that biomechanical cues in conjunction with specific biomarkers may serve as targets for the early diagnosis and intervention of soft tissue diseases.

Circulating blood proteomics enables minimally invasive biomarker discovery. Nanoparticle-based circulating plasma proteomics studies have reported varying number of proteins (ca 2000-7000), but it remains unclear whether a higher protein number is more informative. Here, we first develop OmniProt-a silica-nanoparticle workflow optimized through a systematic evaluation of nanoparticle types and protein corona formation parameters. Next, we present an Astral spectral library for 10,109 protein groups. Using the Astral with 60 sample-per-day throughput, OmniProt identifies ca 3000 to 6000 protein groups from human plasma. Platelet/erythrocyte/coagulation-related contamination artificially inflates protein identifications and compromises quantification accuracy in nanoparticle-enriched samples. Through controlled contamination experiments, we identified biomarkers for platelet/erythrocyte/coagulation-related contamination in nanoparticle-based plasma proteomics. We developed open-access software Baize for contamination assessment. We validated the pipeline in 193 patients with CT-indistinct benign nodules or early-stage lung cancers, flagging five contaminated samples. This study reveals that contamination alters protein identification/quantification in nanoparticle-based plasma proteomics and presents Baize software to evaluate it.

Coordination of cellular and physiological development by signaling is required for normal brain structure and function. Mutations in OCRL, a phosphatidylinositol 4,5 bisphosphate [PI(4,5)P], 5-phosphatase leads to Lowe Syndrome (LS). However, the mechanism by which mutations in OCRL leads to the neurodevelopmental phenotypes of LS is not understood. We find that on differentiation of LS patient iPSC, neural cultures show reduced excitability and enhanced GFAP levels. Multiomic single-nucleus RNA and ATACseq analysis of neural stem cells revealed enhanced numbers of cells with a gliogenic cell state. Analysis of snRNA seq revealed increased levels of DLK1, a Notch ligand in LS patient NSC associated increased levels of cleaved Notch and elevation of its transcriptional target HES5, indicating upregulated Notch signaling. Treatment of iPSC derived brain organoid with an inhibitor of PIP5K, the lipid kinase that synthesizes PI(4,5)P, was able to restore neuronal excitability and rescue Notch signaling defects in OCRL deficient organoids. Overall, our results demonstrate a role for PI(4,5)P dependent regulation of Notch signaling, cell fate specification and neuronal excitability regulated by OCRL.

KRAS mutations are responsible for a quarter of all lung adenocarcinomas. However, the molecular mechanisms linking these mutations and their frequent secondary dosage amplification to tumor formation are still not fully understood. While ample evidence supports a crucial role for the MAPK pathway in tumor development, the primary effectors targeted by this pathway remain largely unexplored. Here we identify the transcriptional repressor Capicua (CIC) as a key target inactivated by KRAS/MAPK signaling in lung adenocarcinoma. We show that genetic loss of CIC recapitulates the phenotypic consequences of amplified KRAS signaling. Genetic disruption of CIC suppressed the requirement for Kras allelic imbalances and accelerated the transformation of bronchiolar Club cells. We also demonstrate that restoring CIC repressor activity impaired proliferation of CIC-deficient tumor cells and reverted resistance to MAPK pathway inhibitors. These results highlight the key role of CIC during lung tumor formation and suggest that selective pressure for effective CIC inactivation favors secondary amplification of KRAS/MAPK signaling in tumor cells.

Cancer cachexia is a debilitating syndrome characterized by the progressive loss of skeletal muscle mass with or without fat loss. Recent studies have implicated dysregulation of the endoplasmic reticulum (ER) stress-induced unfolded protein response (UPR) pathways in skeletal muscle under various conditions, including cancer. In this study, we demonstrate that the IRE1α/XBP1 branch of the UPR promotes activation of the ubiquitin-proteasome system, autophagy, JAK-STAT3 signaling, and fatty acid metabolism in the skeletal muscle of the KPC mouse model of pancreatic cancer cachexia. Moreover, we show that the IRE1α/XBP1 pathway is a key contributor to muscle wasting. Skeletal muscle-specific deletion of the XBP1 transcription factor significantly attenuates tumor-induced muscle atrophy. Mechanistically, transcriptionally active XBP1 binds to the promoter regions of genes such as Map1lc3b, Fbxo32, and Il6, which encode proteins known to drive muscle proteolysis. Pharmacological inhibition of IRE1α using 4µ8C in KPC tumor-bearing mice attenuates cachexia-associated molecular changes and improves muscle mass and strength. Collectively, our findings suggest that targeting IRE1α/XBP1 pathway may offer a therapeutic strategy to counteract muscle wasting during pancreatic cancer-induced cachexia.

Rhabdomyosarcoma (RMS) is the most common pediatric soft tissue sarcoma, with poor outcomes in high-risk and relapsed patients. Here, we identify de novo cholesterol biosynthesis as a critical metabolic vulnerability in RMS. The transcription factor PROX1, previously implicated in RMS growth, acts as an upstream regulator of cholesterol biosynthesis, promoting expression of key pathway genes. Inhibition of cholesterol biosynthesis, either genetically or pharmacologically, impaired RMS cell proliferation, caused a broad halt of cell cycle progression, and activated ER stress-mediated apoptosis through the PERK-ATF4-CHOP axis. Notably, RMS cells could not be rescued by exogenous LDL cholesterol, indicating a unique reliance on endogenous cholesterol production, whereas normal cells, including myoblasts and astrocytes, largely relied on extracellular cholesterol uptake. Clinical and single-cell RNA-seq analyses further revealed that high expression of cholesterol biosynthesis genes correlate with poor survival and enrichment of cell cycle-related gene signatures across RMS subtypes. Together, these findings mechanistically link cholesterol biosynthesis to proliferative signaling and ER stress response in RMS and highlight this pathway as a promising, non-redundant therapeutic target.

Cytotoxic T cell (CTL) dysfunction is a hallmark of immune paralysis after major surgery, increasing susceptibility to severe nosocomial infections and contributing to mortality in critically ill patients. The mechanisms remain poorly understood. We demonstrate that reactive oxygen species (ROS) released by myeloid-derived suppressor cells (MDSC) transiently emerging after surgery, drive perioperative CTL immunoparalysis. These ROS damage CTL mitochondria, triggering secondary mitochondrial ROS amplification and overwhelming antioxidant defenses. The resulting oxidative cascade impairs oxidative phosphorylation and suppresses CTL effector function. Concurrently, stress-induced mitochondrial hyperfusion disrupts fission-dependent translocation to the immunological synapse, exacerbating bioenergetic failure. MitoTEMPO, a mitochondria-targeted antioxidant, partially mitigates these effects, highlighting mitochondrial stabilization as a potential strategy to prevent perioperative immune dysfunction.

Obesity is a major global health challenge, and promoting the browning of white adipose tissue (WAT) represents a promising therapeutic strategy. However, pharmacological approaches to induce adipose thermogenesis remain limited. Through a Connectivity Map-based screen, we identified isomeranzin (ISM) as a potent small-molecule activator of WAT browning. ISM enhances thermogenesis in adipocytes by activating the AMP-activated protein kinase (AMPK) pathway. Integrated limited proteolysis-mass spectrometry, cellular thermal shift assays, and molecular docking identified guanine nucleotide-binding protein G(s) alpha subunit (Gnas) as the direct binding target of ISM. Mechanistic studies further revealed that ISM induces WAT browning through the Gnas-dependent activation of cAMP-AMPK signaling cascade. These findings elucidate the molecular mechanism underlying ISM activity and highlight its potential as a lead compound for enhancing energy expenditure and combating obesity.

Gamma delta (γδ) T cells are innate-like lymphocytes with potent anti-tumor properties. Herein, we show that immune checkpoint receptors (ICRs) display differential expression and regulation by the JAK-STAT pathway in Vδ1 and Vδ2 cells and identify constitutive (e.g. TIGIT, PD-1) and inducible (e.g. TIM-3, LAG-3, CTLA-4) ICRs. In melanoma, all γδ T cell subsets downregulated AP-1 transcription factors, but Vδ1 cells specifically expressed high levels of ICR, TOX and inhibitory killer Ig-like receptor (KIR) transcripts, reminiscent of exhaustion. However, patient-derived cells were functionally competent, although induction of LAG-3 and CTLA-4 was impaired. During anti-PD-1 monotherapy, Vδ1 cells specifically bound high levels of therapeutic antibody but only in patients who responded to treatment, revealing a potential new prognostic marker for evaluating the efficacy of IC blockade (ICB) therapy. Finally, expression of KIR genes in Vδ1 cells was downregulated in response to successful ICB therapy. Collectively, our data indicate an intricate relationship between ICRs and γδ T cells and reveal novel approaches by which these cells can be harnessed in order to discern or improve cancer immunotherapy.

Cardiac hypertrophy is one of the significant causes of heart failure and is closely related to the rising rate of hospitalization and readmissions. Given the diverse regulatory roles of alternative splicing in cardiovascular diseases, RNA-binding proteins have attracted increasing research attention. Here, for the first time, we discovered elevated expression of RBMS1 in heart tissues of patients with dilated cardiomyopathy and in mice with cardiac hypertrophy. We demonstrated that RBMS1 activated the PI3K/AKT signaling pathway by promoting the splicing CTTN to generate CTTN-Δe11 splicing isoform, resulting in cytoskeleton and sarcomere damage in cardiomyocytes. Additionally, pharmacological inhibition of RBMS1 by nortriptyline alleviated cardiac hypertrophy and heart failure. These results provide a new perspective for developing novel therapeutic approaches for cardiac hypertrophy and establish a theoretical basis for targeting RBMS1 in the clinical treatment of cardiac hypertrophy.

Cetuximab, an EGFR-targeting monoclonal antibody, provides beneficial yet limited clinical improvement in KRAS wild-type metastatic colorectal cancer (mCRC). While circRNA dysregulation has been implicated in various cancers, the role of circ-EGFR in response to EGFR-targeted therapy in mCRC remains largely unexplored. Here, we identified circ-EGFR as a promising predictive biomarker for cetuximab response. Clinically, we first determined that tissue-based circ-EGFR biomarker effectively stratified responders from non-responders to cetuximab in mCRC, with an Area under the Curve (AUC) of 76.8%. Functional assays demonstrated that circ-EGFR enhances the sensitivity to cetuximab, whereas its depletion induces resistance in CRC. Mechanistically, we revealed that circ-EGFR functions as a sponge for miR-942-3p, resulting in the upregulation of GAS1, which activates the Hedgehog signaling pathway and promotes the efficacy of cetuximab in CRC. Importantly, we effectively translated this tissue-based biomarker into a liquid biopsy predictor for anti-EGFR response (AUC: 76.9%), highlighting its non-invasive potential. In conclusion, circ-EGFR is a significant predictor of cetuximab efficacy in mCRC, potentially aiding in patient selection and treatment management, especially for patients with low circ-EGFR expression.

Vascular malformations (VMs) are congenital disorders characterized by structurally abnormal blood and lymphatic vessels. Advances in genetics have revealed that most sporadic VMs result from post-zygotic variants in genes involved in key endothelial signaling pathways, including the phosphoinositide-3-kinase (PI3K) and the mitogen-associated proliferation kinase (MAPK) pathways. As these variants are shared with cancer, genetics now have theragnostic impact by helping predict relevant targeted therapies. mTOR and PI3Kα inhibitors such as sirolimus and alpelisib have shown promising efficacy in slow-flow VMs, while reports have suggested that MAPK inhibitors such as trametinib may improve arteriovenous malformations. Despite these advances, several challenges remain, including obtaining accurate genetic diagnosis, enhancing treatment efficacy while mitigating drug-related toxicities, and personalizing multimodal treatment strategies. Emerging approaches such as mutant-selective inhibitors, proteolysis-targeting chimeras, and gene therapy hold promises for improving treatment specificity and minimizing adverse effects. This review provides an overview of the genetic bases of VMs, recent advances in targeted therapies, and future directions in the field, highlighting the ongoing evolution of precision medicine for VMs.

Primary human pancreatic ductal organoids (HPDO) have emerged as a model to study pancreas biology and model disease like pancreatitis and pancreatic cancer. Yet, donor material availability, genetic variability and a lack of extensive benchmarking to healthy and disease pancreas limits the range of applications. To address this gap, we established porcine pancreatic ductal organoids (PPDO) as a system from a reliable, genetically defined and easily obtainable source to model pancreatic ductal/progenitor biology. We benchmarked PPDO to HPDO and primary porcine pancreas using single-cell RNA sequencing (scRNA-Seq). We observed no overt phenotypic differences in PPDO derived from distinct developmental stages using extensive proteomics profiling, with a WNT/basal cell signaling enriched population characterizing PPDO. PPDO exhibited differentiation potential towards mature ductal cells and limited potential towards endocrine lineages. We used PPDO as a chemical screening platform to assess the safety of FDA-approved drugs and showed conserved toxicity of statins and α-adrenergic receptor inhibitors between PPDO and HPDO cultures. Overall, our results highlight the PPDO as a model for mammalian duct/progenitor applications.

Estrogen Receptor alpha (ER)-positive, HER2-negative breast cancers are less aggressive than other subtypes and show good patient clinical outcome because they are likely to respond to endocrine therapies. Unfortunately, therapy-resistant metastases may develop and start an inexorable downhill course. ESR1 mutations leading to resistance to endocrine therapy are prevalent in 20-55% of patients with ER+ metastatic breast cancer. Here, we found that glucocorticoid receptor (GR) activation by dexamethasone in ESR1 mutant metastases-bearing mice decreases liver metastases and prolongs survival. Transcriptomic and proteomic profiling revealed that GR activation not only downregulates estrogen response signature but also induces dramatic loss of ER itself. ChIP-Seq analyses show that prolonged dexamethasone treatment almost completely abrogates ER chromatin binding and that GR binds a subset of ER-related genes, including ESR1. Finally, the GR activity signature predicts a good outcome in patients with ER+ breast cancer. In summary, we show that dexamethasone inhibits ER+ metastatic growth by depleting ER, and hence could be tested for treating patients with ER+ metastatic breast cancer, particularly those suffering from refractory ESR1 mutant metastases.

Fungi are estimated to cause the death of almost 4 million people annually, and we urgently need new drug targets to overcome antifungal resistance. We found that four-stranded nucleic acid structures called G-quadruplexes (G4s) could form within the critical priority fungal pathogen Aspergillus fumigatus. Sequences with the potential to form G4s could be found in genes involved in fungal growth, virulence, and drug resistance. This included cyp51A, which encodes the target of azoles. Notably, we observed the formation of both canonical and unusual acid-stabilised G4s in these sequences. We found that PhenDC3 (a G4-stabilising ligand) could refold DNA into antiparallel G4 structures in cyp51A that were associated with decreased transcription. PhenDC3 also had potent fungistatic activity, prevented germination, synergised with the antifungal amphotericin B in vitro and in vivo, and displayed low genotoxicity and cytotoxicity towards human cells. Interestingly, PhenDC3 had greater antifungal activity towards the pan-azole-resistant A. fumigatus TR34/L98H isolate, and another G4-stabiliser, pyridostatin, killed multi-drug-resistant Candida auris. Taken together, G4s represent a promising target for the development of antifungals with novel mechanisms of action.

High-throughput screening studies provide an additional approach to discovering repurposed drugs for antimicrobial treatments. In this work, we report the identification of ENOblock, an anticancer drug, as an antimicrobial agent. We computationally and experimentally validated that ENOblock synergizes with colistin, the last resort antibiotic. Additionally, we identified enolase as the potential bacterial target for ENOblock. The in silico and in vitro antibacterial activity of ENOblock translated into potent in vivo efficacy in an animal infection model. Collectively, the preclinical data support the selection of ENOblock as a promising candidate for antimicrobial development, with the potential to address the urgent threat of infections caused by Acinetobacter baumannii.