Huaxiao Yang

Understanding how cardiomyocyte structure governs contractile function is fundamental to cardiac biology and disease modeling, yet current approaches rely on time‐resolved imaging and computationally intensive analysis. Here, we present a generative artificial intelligence (AI) framework that directly predicts contractile behavior of human‐induced pluripotent stem cell‐derived cardiomyocytes (hiPSC‐CMs) from single static images. Our approach integrates a U‐Net‐based generator with a patch‐based generative adversarial network (GAN) discriminator to translate morphological and sarcomere structural features into pixel‐resolved contraction heatmaps. To further enhance performance and generalizability, we incorporated synthetic cell–function pairs generated via a StyleGAN2 framework, improving prediction accuracy and perceptual similarity. Region‐specific and whole‐cell analyses revealed that AI predictions capture biologically meaningful structure–function relationships, with sarcomere organization strongly associated with both contractile output and prediction fidelity. Reconstruction error emerged as an interpretable metric reflecting localized inefficiencies in sarcomere‐to‐contraction coupling. Together, this framework establishes a scalable, interpretable strategy for inferring cardiomyocyte function from static morphology, eliminating the need for time‐lapse imaging, and positions generative AI as a powerful tool for bridging cellular structure and function to advance high‐throughput phenotyping and in vitro cardiac modeling.

Accurately forecasting the spatiotemporal dynamics of biological systems, such as human pluripotent stem cell (hPSC)-derived cardiac organoids, from microscopy time-series is a critical challenge in biomedicine with profound implications for drug discovery. Existing generative models often fail to capture the intricate dynamics of organoid development, struggling with their irregular morphology, indistinct boundaries, and complex spatiotemporal patterns. To overcome these limitations, we introduce OrgaCast, a novel multimodal conditional diffusion model for high-fidelity organoid forecasting. OrgaCast uniquely conditions the generative process on three synergistic modalities: (i) historical image sequences, captured by a dedicated spatiotemporal control module; (ii) structured numerical metadata defining experimental conditions; and (iii) descriptive text captions summarizing the biological context. This comprehensive conditioning enables the generation of forecasts with high visual accuracy and biological plausibility. Furthermore, to enhance the model's utility in critical research settings, we introduce a post-hoc uncertainty quantification method that produces intuitive confidence maps, bolstering the interpretability and trustworthiness of predictions. Extensive experiments on a challenging cardiac organoid dataset demonstrate that OrgaCast outperforms baselines in metrics such as SSIM, PSNR, and LPIPS. Our framework presents a robust solution for biological forecasting, promising to accelerate research discovery while minimizing experimental costs and manual effort.

Cardiotoxicity remains a major cause of drug attrition and postmarket withdrawal, yet the vast majority of environmental chemicals to which humans may be exposed remain uncharacterized for cardiotoxicity risk. Human induced pluripotent stem cell (hiPSC)-based testing has been proposed to address this gap. Here, we present an unsupervised deep learning framework for multi-donor cardiotoxicity screening using high-throughput calcium transient recordings from hiPSC-derived cardiomyocytes (hiPSC-CMs). We analyzed data from a library of 1,029 compounds tested in hiPSC-CMs from 5 donors across a concentration range. An autoencoder trained exclusively on baseline signals quantified chemically induced functional perturbations through reconstruction error, bypassing the need for labeled training data while capturing the full spectrum of calcium-handling disruptions. Aggregation of donor-specific scores revealed substantial inter-individual variability in potential cardiotoxicity, underscoring the value of this approach for multi-donor risk prediction. We identified microbiocides, dyes, and pesticides as chemical classes of potential concern, characterized by high toxicity scores and low interdonor variability. This framework establishes a scalable, human-relevant, and genetically diverse platform for cardiotoxicity surveillance across both pharmacological and environmental chemical spaces.

The phenotypic immaturity of human induced pluripotent stem cell‐derived cardiomyocytes (hiPSC‐CMs) remains a critical barrier to their effective use in disease models, drug screening, and cardiac regeneration. Current culture platforms still find it hard to provide a physiologically relevant, stable, and scalable microenvironment to support cardiomyocyte maturation. We report a reproducible heart tissue‐derived cardiac extracellular matrix (ECM) microsphere that integrates native biochemical cues with a porous three‐dimensional (3D) architecture to support cardiac cell culture and maturation. These microspheres supported the attachment and proliferation of C2C12, HL‐1, H9c2, and hiPSC‐CMs culture, suggesting the broad applicability of multiple cell types. Compared with the conventional two‐dimensional (2D) culture system, decellularized cardiac ECM microspheres provided a spherical 3D culture interface and significantly enhanced hiPSC‐CM maturation, as indicated by upregulation of cardiac genes ( ACTA2, TNNT2, GJA1 ), rapid calcium cycling, and synchronized calcium transients. Long‐term culture (up to 8 months) on these ECM microspheres supported hiPSC‐CM maturation, as evidenced by enhanced sarcomere alignment, robust α‐actinin expression, contractile phenotype, elevated gap junction (connexin 43, CX‐43) protein expression, and increased binucleation. This ECM microsphere system represents a scalable and bioactive platform for long‐term cardiomyocyte culture and maturation and broader applications in cardiac tissue engineering.

Conductive polymers (CPs) have gained increasing attention in cardiac tissue engineering (CTE) due to their ability to restore electrical conductivity, enhance cardiomyocyte (CM) function, and support tissue regeneration. Despite significant progress in the field, challenges related to variability in biocompatibility testing, including dopant-dependent cytotoxicity, poor reporting of biodegradability, unpredictable long-term stability, and regulatory uncertainty of CPs continue to hinder their applications. To address this, we reviewed the properties, applications, and biocompatibility of the three most studied CPs: polypyrrole (PPy), poly(3,4-ethylenedioxythiophene) (PEDOT), and polyaniline (PANI) in CTE, and contrast their advantages and safety challenges with inorganic electrodes and carbon-based materials. We critically assessed current methods for evaluating the biocompatibility of CPs, highlighting limitations in traditional in vitro and in vivo approaches. Our analysis revealed a significant gap in chronic implantation data beyond six months and provided dopant-centered assessment and toxicity risks across different CP platforms. A comprehensive roadmap was further suggested to guide the evaluation of the biocompatibility of CPs, including material characterization, in vitro cytotoxicity testing with particular emphasis on in vitro 3D human heart model testing platforms of human pluripotent stem cell (hPSC)-derived engineered heart tissues and cardiac organoids, and in vivo evaluation. Additionally, we discussed recent advances in improving the biocompatibility of CPs through hybrid scaffold development, molecular engineering, surface chemistry modifications, and the development of stimuli-responsive and targeted CP constructs. By establishing this standardized framework and highlighting critical regulatory requirements, this review aims to overcome current biocompatibility barriers and facilitate the improved implementation of CPs in CTE applications.

Background and Aims: Osimertinib is a third-generation tyrosine kinase inhibitor targeting activating mutations of epidermal growth factor receptor with remarkable therapeutic efficacy against non-small cell lung carcinoma. However, its use has been limited by associated cardiotoxicity, primarily with heart failure. Herein, this study aims to better understand the mechanisms underlying osimertinib cardiotoxicity and explore cardioprotective strategies. Methods: This study leverages an in vitro model of human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) and a clinically relevant in vivo mouse model of osimertinib cardiotoxicity by co-employing transverse aortic constriction to mimic haemodynamic stress in cancer patients. Results: Osimertinib treatment leads to significant contractile dysfunction in mice without cell death, inflammation, or fibrosis. By leveraging single-nucleus RNA sequencing of mouse heart tissues and in vitro assays of human iPSC-CMs, the study reveals significant downregulation of MYLK3 and a subsequent decrease in MYL2 phosphorylation with marked sarcomere disarray as the main mechanism of osimertinib cardiotoxicity. GATA4 is further identified as a putative target of osimertinib, connecting its decreased phosphorylation to repressed MYLK3 transcription. The reversibility of osimertinib-induced cardiac dysfunction upon discontinuation of osimertinib treatment supports the hypothesis that transient sarcomere disruption, rather than permanent cellular damage, serves as the key underlying mechanism. Finally, the myosin activator omecamtiv is shown to be effective in preventing osimertinib cardiotoxicity. Conclusions: These findings suggest that osimertinib causes reversible cardiac dysfunction by disrupting MYL2 phosphorylation via GATA4 dephosphorylation-mediated suppression of MYLK3 and highlight the potential of myosin activation as a preventive or rescue strategy for osimertinib cardiotoxicity.

Although model organisms have provided insight into the earliest stages of cardiac and hepatic vascularization, we know very little about this process in humans because of ethical restrictions and the technical difficulty of obtaining embryos during very early development. In this study, we demonstrate that micropatterned human pluripotent stem cell–derived gastruloids enable in vitro modeling of the earliest stages of vascularization. We identify a combination of vascular-inducing factors that give rise to cardiac vascularized organoids with a spatially organized and branched vascular network. To show the broader utility of our vascularization strategy, we use the same vascular-inducing factors to produce hepatic vascularized organoids. Our results suggest that a conserved developmental program generates the vasculature within different types of organs.

Human pluripotent stem cell (hPSC)‐derived cardiac organoids (COs) are the most recent three‐dimensional tissue structure that mimics the human heart's structure and functionality for modeling heart development and disease. Fluorescent labeling and imaging are commonly utilized to characterize the cellular information in COs. However, the additional step of fluorescence labeling and imaging is time‐consuming, inefficient, and typically for end‐timepoint characterization. Meanwhile, the COs are routinely examined by brightfield/phase contrast microscope to track live‐cell organoid formation in structure and morphology. Although the brightfield microscope provides essential information about COs, such as morphology and overall structure, it limits our understanding of cardiovascular cells (e.g., cardiomyocytes, CMs and endothelial cells, ECs) and corresponding quantifications in COs. Is it possible to overcome these limitations of bright‐field microscopic imaging and provide cardiovascular cell type‐specific information similar to the fluorescence‐labeled imaging acquisition in COs? This research addresses this limitation by proposing a generative AI system for colorizing phase contrast images of COs from bright‐field microscopic imaging using conditional generative adversarial networks (cGANs) to generate cardiovascular cell type‐specific fluorescence images of COs. By giving these phase contrast images with multichannel fluorescence colorization, this intelligence system unlocks cell type and quantifications of COs in high efficiency and accuracy.