Sunghoon Kwon

Liver tissue engineering offers a promising alternative for end-stage liver disease, yet the recreation of functional vasculature remains a major bottleneck to clinical translation. Here, we developed vascularized liver organoids by integrating human induced pluripotent stem cell (iPSC)–derived hepatoblasts and endothelial cells into decellularized scaffolds functionalized with an anti-CD31 aptamer–based vascular coating agent (VCA). This facilitated spatially coordinated organization of vasculature and parenchyma. Spatial transcriptomic profiling and subsequent functional perturbation demonstrated IGF2-IGF1R-AKT/MAPK signaling as a key axis governing spatial organization and functional maturation of the liver organoids. Furthermore, exogenous IGF2 synergized with the VCA to augment the structural and functional refinement of liver organoids, which translated into markedly improved therapeutic outcomes following transplantation into a chronic liver failure mouse model. Collectively, these findings establish a comprehensive framework for generating physiologically relevant liver tissues from iPSCs and demonstrate the utility of spatial transcriptomics for uncovering regenerative mechanisms. This approach advances the feasibility of autologous, transplantable liver grafts for personalized regenerative therapy.

Gut microbiota and humoral immunity have been suggested as key players in the pathogenesis of Sjögren disease (SjD), but their mechanisms remain unclear. In this study, we transferred the gut microbiota of SjD-like autoimmune dry eye disease model mice to B6 mice, then characterized the resulting gut microbiome composition, clinical ocular phenotype, and B cell receptor (BCR) repertoire. Notable changes were observed in the gut microbiome of NOD-FMT mice, accompanied by SjD-like clinical features, including elevated corneal fluorescein staining scores, reduced tear production, increased IL-6 mRNA levels, and decreased MUC5AC mRNA levels. Additionally, stereotypic B cell receptor (BCR) clonotypes were shared at significantly higher frequencies in NOD-FMT mice than in controls. The majority of B cell clones encoding these stereotypic clonotypes developed and expanded locally in the lacrimal gland, and some also achieved systemic presence. These results uncover a gut–ocular immune axis in which microbiota transfer induces stereotyped, systemically disseminating BCR clonotypes that contribute to the immunopathogenesis of autoimmune dry eye disease.

The genetic architecture of the antibody repertoires in chronic autoimmunity, and its relationship to B cell malignancy risk, remains incompletely defined. Using deep B cell receptor (BCR) heavy-chain sequencing from 20 Sjögren's syndrome (SS) patients (> 7 million unique sequences), compared to 11 age- and sex-matched healthy controls and 330 blood samples from 64 healthy donors, we identified 1,942 SS-associated extended stereotypic BCR clonotypes (eSBCs), alongside reductions in diversity and expansion of class-switched clonotypes. Nineteen eSBCs were shared by ≥ 4 of 20 patients (≥ 20%). The most prevalent clonotype (eSBC1; IGHV4-59/IGHJ2), detected in 65% of SS patients, matches in IGHV/IGHJ gene usage, and CDR3 features to a canonical rheumatoid factor (RF) previously reported in non-Hodgkin lymphoma and chronic lymphocytic leukemia, linking SS-associated autoreactive B cells to established lymphomagenic populations. RF activity was confirmed by ELISA for 10 BCRs spanning five SS-associated eSBCs with shared IGHV-CDR3 features, including eSBC1. Eight of the 19 eSBCs, including four RF-positive clonotypes, were also detected in rheumatoid arthritis and inflammatory demyelinating disease, revealing convergent autoantibody pathways across immune-mediated disorders. These results indicate that public clonotypes populate chronically autoreactive B cell pools across autoimmune diseases and B cell malignancies, supporting a model in which sustained autoantigen drive of public B cell clonotypes predisposes to lymphomagenesis.

Artificial intelligence (AI) has shown promise in detecting and characterizing musculoskeletal diseases from radiographs. However, most existing models remain task-specific, annotation-dependent, and limited in their adaptability across diseases and anatomical regions. Although a comprehensive foundation model trained on large-scale musculoskeletal radiographs is clinically needed, publicly available datasets remain limited in size and lack sufficient diversity to enable training across a wide range of musculoskeletal conditions and anatomical sites. Here, we present SKELEX, a large-scale foundation model for musculoskeletal radiographs, trained using self-supervised learning on 1.2 million diverse, condition-rich images. The model was evaluated on 12 downstream diagnostic tasks and generally outperformed baselines in fracture detection, osteoarthritis grading, and bone tumor classification. Furthermore, SKELEX demonstrated unsupervised reconstruction-based anomaly localization, producing error maps that identified pathologic regions without task-specific training. Building on this capability, we developed an interpretable, region-guided bone tumor classifier. This model maintained robust performance on independent external datasets and was deployed as a publicly accessible web application, serving as a proof of concept for its potential clinical translation. Overall, SKELEX provides a scalable, label-efficient, and broadly applicable AI framework for musculoskeletal radiographs, with its robust external validity specifically demonstrated in bone tumor applications.

Dynamic transitions of mature osteoblasts between active and quiescent states are essential for bone homeostasis and present a promising target for osteoanabolic therapy. However, these transitions remain poorly understood due to cellular heterogeneity and limited spatial context. Here, we employed spatially resolved osteoblast-traced transcriptomics, integrating an osteoblast-specific lineage tracing study and spatially resolved laser-activated cell sorting (SLACS), to profile osteoblast states on quiescent bone surfaces. This approach identified transforming growth factor-beta (TGF-β) signaling as a regulator of osteoblast activation. We further validated this role using single-cell RNA sequencing, in vitro functional assays, and in vivo. In a hindlimb unloading mouse model, dual inhibition of TGF-β and sclerostin enhanced bone mass and mitigated bone loss more effectively than sclerostin inhibition alone. These findings reveal a mechanistic role for TGF-β in regulating osteoblast dynamics and propose a dual-target therapeutic strategy that enhances the efficacy of anti-sclerostin treatment in osteoporosis.

Artificial intelligence (AI) has shown promise in detecting and characterizing musculoskeletal diseases from radiographs. However, most existing models remain task-specific, annotation-dependent, and limited in generalizability across diseases and anatomical regions. Although a generalizable foundation model trained on large-scale musculoskeletal radiographs is clinically needed, publicly available datasets remain limited in size and lack sufficient diversity to enable training across a wide range of musculoskeletal conditions and anatomical sites. Here, we present SKELEX, a large-scale foundation model for musculoskeletal radiographs, trained using self-supervised learning on 1.2 million diverse, condition-rich images. The model was evaluated on 12 downstream diagnostic tasks and generally outperformed baselines in fracture detection, osteoarthritis grading, and bone tumor classification. Furthermore, SKELEX demonstrated zero-shot abnormality localization, producing error maps that identified pathologic regions without task-specific training. Building on this capability, we developed an interpretable, region-guided model for predicting bone tumors, which maintained robust performance on independent external datasets and was deployed as a publicly accessible web application. Overall, SKELEX provides a scalable, label-efficient, and generalizable AI framework for musculoskeletal imaging, establishing a foundation for both clinical translation and data-efficient research in musculoskeletal radiology.

Correction for ‘Sequential trench well based microfluidic platform to isolate bacteria from whole blood with large volume processing’ by Cheonggyu Lee et al. , Lab Chip , 2025, 25 , 6650–6661, https://doi.org/10.1039/d5lc00931f.

Using a microfluidic trench-based structure, bacteria were effectively isolated from blood, significantly improving diagnostic speed.

Background: Although radiographs are the first-line imaging modality, differentiating between neoplastic pathologic fractures and nonpathologic fractures on radiographs can sometimes be challenging. In this study, we aimed to develop and evaluate a deep learning model capable of distinguishing neoplastic pathologic fractures from nonpathologic fractures on hip radiographs to enhance diagnostic accuracy. Methods: This retrospective, multicenter study analyzed anteroposterior hip radiographs from patients who visited the emergency department at 4 different institutions. The deep learning model was trained on, and tested using, data from 338 patients at a single institution and externally validated on data from 488 patients across 3 additional institutions. Results: The model achieved an overall accuracy of 0.880, with a sensitivity of 0.882 and a specificity of 0.879, on the internal test set. It was then externally validated using the data of 488 patients (67 with neoplastic pathologic fracture and 421 with nonpathologic fracture) from institutions separate from where the model was developed. The model achieved an overall accuracy of 0.848, sensitivity of 0.910, and specificity of 0.786. Its performance was comparable with that of general orthopaedic surgeons. Conclusions: The developed deep learning model is a reliable and valid tool for distinguishing neoplastic pathologic fractures from nonpathologic fractures on hip radiographs. It has the potential to assist orthopaedic surgeons in resource-limited settings, where optimizing the interpretation of radiographs is critical for patient care. The model is publicly available at https://pathfxdx.org. Level of Evidence: Diagnostic Level III . See Instructions for Authors for a complete description of levels of evidence.

Biopharmaceuticals are emerging as viable alternatives to chemically synthesized drugs for potential treatment to various diseases. However, many of these human-derived components cannot withstand terminal sterilization procedures, and the duration of conventional sterility testing methods often exceeds their limited shelf life. Consequently, biopharmaceuticals are now frequently administered to patients before sterility confirmation. Here we present a nanoparticle-based enrichment and rapid sterility test that can determine product sterility within a single day, mitigating clinical risks of biopharmaceuticals and maintaining therapeutic efficacies during delivery. The assay incorporates synthetic beta-2-glycoprotein I peptides for selective isolation and purification of a broad spectrum of microorganisms and a microfluidic chip designed to automatically monitor their metabolic activities via fluorescence imaging, which are inferred from the reduction of a non-toxic dye as they grow. Compared with conventional approaches, the turnaround time was substantially reduced by >58 h with 100% accuracy and a limit of detection down to a concentration of 1 colony forming unit per millilitre. We validate our approach using various forms of clinical-grade biopharmaceutical products.