Central nervous system (CNS) repair and regeneration suffer from tremendous clinical challenges due to current limitations in replacing lost neural tissues and restoring long-term neural circuits. Neural organoids, 3D lab-cultured neural tissues derived from stem cells, can recapitulate key cellular, structural, and physiological features of the human CNS, showing promising potential for neural regeneration. Here, we envision organoid brain-machine-interface (Organoid-BMI) devices as a new kind of neuroelectrical interface for CNS repair. The Organoid-BMI devices employ neural organoids and bioelectrodes as biohybrid bidirectional communication pathways to connect the human CNS and the external world. Acting as a biologically compatible intermediate, this approach may facilitate structural incorporation and functional alignment with host neural circuits for addressing persistent challenges of CNS repair including graft-host mismatch and long-term circuit stability. Through implementing adaptive and closed-loop strategies, this approach can modulate interaction and functional communication with the host for promoting CNS circuit remodeling and functional recovery. Together, this innovative technology may open new avenues for personalized regenerative medicine.
Degenerative eye diseases are major causes of irreversible vision loss worldwide, but effective treatments remain limited, partly due to the lack of effective human models. Retinal organoids derived from stem cells can recapitulate key structural and physiological features of the human retina, offering powerful tools to study disease mechanisms and develop new therapies. Here, we review recent progress in engineering retinal organoids and eye-on-a-chip models for modeling degenerative eye diseases, with a focus on engineering innovations. We first describe conventional methods for organoid differentiation and characterization along with current outstanding challenges. To better engineer retinal organoids, new strategies that leverage microfluidics and biomaterials have emerged to regulate dynamic and physiologically relevant environments for organoid differentiation. Moreover, the integration of artificial intelligence, multimodal sensing, and data analytics improves the monitoring and prediction of retinal function and therapeutic outcomes. Finally, we discuss future directions in innovating next-generation retinal organoid and eye-on-a-chip models for disease modeling, drug discovery, and vision restoration, highlighting their potential for precision ophthalmology.
Bronchopulmonary dysplasia (BPD) is a chronic lung disease of prematurity with no curative therapy, characterized by impaired alveologenesis and capillary formation. However, the molecular mechanisms underlying endothelial dysfunction, a key driver of BPD pathogenesis, remain poorly understood. Through multiomic profiling of endothelial cells isolated from human BPD lungs, we identified an expansion of general capillary endothelial cells (gCaps) marked by neurotrophic receptor tyrosine kinase 2 (NTRK2). Notably, we uncovered a critical isoform switch that governs gCap regeneration. Full-length NTRK2 (NTRK2-FL) promoted gCap repair after hyperoxic injury, whereas RBFOX2-mediated splicing of NTRK2-FL into a truncated isoform (NTRK2-T1) contributed to maladaptive responses and persistent alveolar simplification. Restoring NTRK2-FL using lipid nanoparticle-delivered mRNA promoted angiogenesis in vessel organoids and reversed alveolar simplification in hyperoxic mice. These findings identified NTRK2 isoform imbalance as a key driver of endothelial dysfunction and support isoform-specific RNA therapy as a promising strategy for vascular regeneration and repair.
The COVID-19 pandemic, caused by SARS-CoV-2, has underscored the urgency of understanding viral entry mechanisms to develop effective therapeutic strategies. SARS-CoV-2 primarily exploits angiotensin-converting enzyme 2 (ACE2) as its entry receptor and relies on the serine protease TMPRSS2 to prime its spike protein, enabling membrane fusion and infection. Traditionally, TMPRSS2 has been described as a cell surface protein, but our study reveals that in human lung epithelial cells, TMPRSS2 is largely absent from the plasma membrane and instead resides intracellularly. We show that TMPRSS2 is secreted together with ACE2 in extracellular vesicles (EVs) from lung epithelial cells, which are subsequently taken up by non-epithelial cells, specifically alveolar macrophages, endothelial cells, and pericytes, that do not express TMPRSS2 or ACE2 mRNAs under homeostatic conditions. This EV uptake deposits ACE2 and TMPRSS2 protein onto recipient cells, equipping them for SARS-CoV-2 entry. By transferring these viral entry proteins, EVs expand the spectrum of susceptible cell types in the lung, offering a new explanation for how the virus can infect diverse cell populations and cause widespread tissue damage. Identifying EVs as vehicles for delivering functional ACE2 and TMPRSS2 across cell types reveals a previously unrecognized pathway of viral entry with important implications for not only COVID-19 pathogenesis but also for other viral infections that exploit similar entry mechanisms. These findings open new avenues for therapeutic intervention aimed at disrupting EV-mediated protein transfer, potentially limiting viral dissemination and severity, and may also represent a generalizable mechanism exploited by other viral pathogens, highlighting the potential relevance of EV-mediated protein transfer beyond SARS-CoV-2.
How allelic variants in lineage-regulating transcription factors drive diverging human developmental outcomes remains poorly understood. This is partly due to the lack of human model systems. Here, we used vessel organoids from human induced pluripotent stem cells (hiPSCs) to resolve variant-specific functions of Forkhead Box F1 (FOXF1), a critical regulator of mesoderm and vascular development. Using three patient-derived hiPSC lines harboring unique FOXF1 variants, we show that heterozygous variants cause capillary maldevelopment of varying severity. Single-nucleus multiomic analysis revealed variant-specific mechanisms - a severe variant impairs differentiation of nascent mesoderm to lateral plate mesoderm and disrupts vascular progenitor specification, while moderate variants permit mesoderm differentiation but rewire vascular progenitor states and function. Restoration of wild-type FOXF1 via lipid nanoparticle-mediated mRNA delivery rescued capillary formation in a variant- and developmental-stage-dependent manner. Together, these findings demonstrate that different variants disrupt stage-specific FOXF1 functions in human mesoderm-to-vascular development, underscoring the importance of variant-specific therapeutic strategies. HIGHLIGHTS Human vessel organoids reveal variant-specific roles of FOXF1 in mesoderm patterning and capillary development. Severe FOXF1 variant c.253T>A (p.F85I) impairs nascent mesoderm-to-lateral plate mesoderm differentiation and disrupts vascular progenitor specification. 'Moderate' FOXF1 variants differentially rewire endothelial and mural progenitor cell states and function. Lipid nanoparticle-mediated FOXF1 mRNA delivery rescues capillary formation in a variant- and developmental-stage-dependent manner.
Human brain organoids recapitulate key physiological features and functions of the human brain and hold remarkable potential for studying neurological diseases. Despite clinical evidence suggesting that neurodegenerative diseases impair the information-processing ability of the human brain, the capacity of brain organoids for information processing and its relationship to neural network function remain largely unexplored. Here, we test and quantify information-processing-like properties of human cortical organoids using a task-based functional phenotyping framework (Brainopheno). We demonstrate the pattern-processing-like phenotype of cortical organoids through the representation and classification of evoked neural activities in response to distinct spatial input stimulation patterns via a microelectrode array (MEA) system. Moreover, this functional phenotype emerges with network maturation and is disrupted by pharmacological perturbations, linking classification performance to the functional integrity of organoid neural networks (ONNs). Importantly, this functional phenotype also reveals functional deficits in ONNs altered by a familial Alzheimer's disease (AD)-associated gene mutation (APP) and by monocytes from patients with sporadic AD. This work may establish a new quantifiable functional phenotype of neural organoids and provide a framework to bridge molecular and cellular profiles with neural circuit function for basic neurology, disease phenotyping, and therapeutic development.
Increasing evidence suggests that Alzheimer's disease (AD) pathogenesis strongly correlates with neuroinflammation. Peripheral monocytes are crucial components of the human immune system that may play a role in neuroinflammation, but their contribution to AD pathogenesis is largely understudied partially due to the lack of appropriate human models. Here, we present human cortical organoid microphysiological systems (hCO-MPSs) for modeling dynamic AD neuroinflammation mediated by monocytes. By incorporating 3D printed devices into an existing cortical organoid protocol, 96 hCO-MPSs can be established with significantly reduced necrosis and hypoxia as well as enhanced viability within a commonly used 96 well plate, and each hCO-MPS consists of a doughnut-shaped hCO and a 3D printed device per well. Using this approach, monocytes from AD patients exhibit higher infiltration, decreased amyloid-beta (Aβ) clearance, and stronger inflammatory responses compared to monocytes from age-matched control donors. Moreover, pro-inflammatory effects such as elevated astrocyte activation and neuronal apoptosis were observed to be induced by AD monocytes. Furthermore, the significant increase in the expression of IL1B and CCL3, both at the transcriptional and protein levels, indicated the pivotal role of these cytokine and chemokine in monocyte-mediated AD neuroinflammation. Our findings provide insight for understanding monocytes' role in AD pathogenesis, and the user-friendly MPS models we present are compatible with existing laboratory settings, highlighting their potential for modeling neuroinflammation and developing new therapeutics for various neuroinflammatory diseases.
Bronchopulmonary dysplasia (BPD) is a chronic lung disease of prematurity with no curative therapy, characterized by impaired alveologenesis and capillary formation. However, the molecular mechanisms underlying endothelial dysfunction, a key driver of BPD pathogenesis, remain poorly understood. Through multiomic profiling of endothelial cells isolated from human BPD lungs, we identified an expansion of general capillary endothelial cells (gCaps) marked by neurotrophic receptor tyrosine kinase 2 (NTRK2). Notably, we uncovered a critical isoform switch that governs gCap regeneration. Full-length NTRK2 (NTRK2-FL) promoted gCap repair after hyperoxic injury, whereas RBFOX2-mediated splicing of NTRK2-FL into a truncated isoform (NTRK2-T1) contributed to maladaptive responses and persistent alveolar simplification. Restoring NTRK2-FL using lipid nanoparticle-delivered mRNA promoted angiogenesis in vessel organoids and reversed alveolar simplification in hyperoxic mice. These findings identified NTRK2 isoform imbalance as a key driver of endothelial dysfunction and support isoform-specific RNA therapy as a promising strategy for vascular regeneration and repair.
Organoids, 3D organ-like tissue cultures derived from stem cells, show promising potential for developmental biology, drug discovery, and regenerative medicine. However, the function and phenotype of current organoids, especially neural organoids, are still limited by insufficient diffusion of oxygen, nutrients, metabolites, signaling molecules, and drugs. Herein, we present vascular network-inspired diffusible (VID) scaffolds to mimic physiological diffusion physics for generating functional organoids and phenotyping their drug response. Specifically, the VID scaffolds, 3D-printed meshed tubular channel networks, successfully engineer human midbrain organoids almost without necrosis and hypoxia in commonly used well plates. Compared with conventional organoids, these engineered organoids develop more physiologically relevant features and functions, including midbrain-specific identity, oxygen metabolism, neuronal maturation, and network activity. Moreover, these engineered organoids also better recapitulate pharmacological responses, such as neural activity changes to fentanyl exposure, compared with conventional organoids with significant diffusion limits. This platform may provide insights for organoid development and therapeutic innovation.
Bronchopulmonary dysplasia (BPD) is a chronic lung disease in premature infants with no curative therapy, characterized by impaired alveologenesis and capillary formation. However, the molecular mechanisms underlying endothelial dysfunction, a key driver of BPD pathogenesis in human, remain poorly understood. Here, through multiomic profiling of vascular endothelial cells isolated from control and BPD patient lungs, we uncovered an expansion of general capillary endothelial cells (gCap) with aberrant expression of the neurotrophic receptor tyrosine kinase 2 (NTRK2) in BPD. Importantly, we identified a pathological NTRK2 isoform switch that dictates the regenerative capacity of gCap cells. Full-length NTRK2 (NTRK2-FL) promoted gCap regeneration in response to hyperoxic injury, whereas RBFOX2-mediated splicing of NTRK2-FL into a truncated isoform (NTRK2-T1) contributed to maladaptive responses and irreversible alveolar simplification in severe BPD cases. Restoring NTRK2-FL using lipid nanoparticle-delivered mRNA promoted angiogenesis and reversed alveolar simplification in vessel organoids and BPD-like mice. These findings identified NTRK2 isoform imbalance as a key driver of endothelial dysfunction and support isoform-specific RNA therapy as a promising strategy for vascular regeneration and repair.
Increasing evidence strongly links neuroinflammation to Alzheimer's disease (AD) pathogenesis. Peripheral monocytes are crucial components of the human immune system, but their contribution to AD pathogenesis is still largely understudied partially due to limited human models. Here, we introduce human cortical organoid microphysiological systems (hCO-MPSs) to study AD monocyte-mediated neuroinflammation. By culturing doughnut-shape organoids on 3D-printed devices within standard 96-well plates, we generate hCO-MPSs with reduced necrosis, minimized hypoxia, and improved viability. Using these models, we found that monocytes from AD patients exhibit increased infiltration ability, decreased amyloid-β clearance capacity, and stronger inflammatory response than monocytes from age-matched control donors. Moreover, we observed that AD monocytes induce pro-inflammatory effects such as elevated astrocyte activation and neuronal apoptosis. Furthermore, the marked increase in IL1B and CCL3 expression underscores their pivotal role in AD monocyte-mediated neuroinflammation. Our findings provide insight into understanding monocytes' role in AD pathogenesis, and our lab-compatible MPS models may offer a promising way for studying various neuroinflammatory diseases.
Alzheimer's disease (AD) is a progressive and neurodegenerative disease, predominantly causing dementia. Despite increasing clinical evidence suggesting the involvement of peripheral immune cells such as monocytes in AD pathology, the dynamic penetration and infiltration of monocytes crossing blood-brain barrier (BBB) and inducing neuroinflammation is largely understudied in an AD brain. Herein, we engineer BBB-like microphysiological system (BBB-MPS) models for recapitulating the dynamic penetration and infiltration of monocytes in an AD patient's brain. Each BBB-MPS model can be engineered by integrating a functional BBB-like structure on a human cortical organoid using a 3D-printed device within a well of a plate. By coculturing these BBB-MPS models with monocytes from AD patients and age-matched healthy donors, we found that AD monocytes exhibit significantly greater BBB penetration and brain infiltration compared to age-matched control monocytes. Moreover, we also tested the interventions including Minocycline and Bindarit, and found they can effectively inhibit AD monocyte infiltration, subsequently reducing neuroinflammation and neuronal apoptosis. We believe these scalable and user-friendly BBB-MPS models may hold promising potential in modeling and advancing therapeutics for neurodegenerative and neuroinflammatory diseases.
The vasculature and mesenchyme exhibit distinct organ-specific characteristics adapted to local physiological needs, shaped by microenvironmental and cell-cell interactions from early development. To recapitulate this entire process, we co-differentiated mesoderm and endoderm within the same spheroid to vascularize lung and intestinal organoids from induced pluripotent stem cells (iPSCs). Bone morphogenetic protein (BMP) signaling fine-tuned the endoderm-to-mesoderm ratio, a critical step in generating appropriate proportions of endothelial and epithelial progenitors with tissue specificity. Single-cell RNA sequencing (scRNA-seq) revealed organ-specific gene signatures of endothelium and mesenchyme and identified key ligands driving endothelial specification. The endothelium exhibited tissue-specific barrier function, enhanced organoid maturation, cellular diversity, and alveolar formation on the engineered lung scaffold. Upon transplantation into mice, the organoid vasculature integrated with the host circulation while preserving organ specificity, further promoting organoid maturation. Leveraging these vascularized organoids, we uncovered abnormal endothelial-epithelial crosstalk in patients with forkhead box F1 (FOXF1) mutations. Multilineage organoids provide an advanced platform to study intricate cell-to-cell communications in human organogenesis and disease.
Background: Pulmonary arterial hypertension (PAH) has a high comorbidity burden. Our Large Language Models on ‘real-world’ data revealed several tightly correlated comorbidities in patients with drug-induced PAH. Diabetes, for instance, was one of the top comorbidities in PAH (p < 0.0001). Since comedications can play a critical role in defining patient outcomes, we analyzed the incidence rates of PAH and time-to-PAH in patients on antidiabetic medications. We hypothesize that differential rates of PAH and PAH onset patterns may offer insights into potential modulatory effects of antidiabetic medications. Methods: We analyzed 571,215 medical records of patients with diabetes and prior exposure to PAH-associated drugs. Drugs included dasatinib, methamphetamine, thiotepa, carboplatin, busulfan, cyclophosphamide, sofosbuvir, ribavirin, daclatasvir, interferon beta-1a, and sertraline. Patients were classified into 7 drug groups based on their antidiabetic medications: biguanides (n = 118,895), sulfonylureas (23,447), Glucagon-like peptide-1 (GLP-1) analogues (12,379), dipeptidyl peptidase 4 (DPP-4) inhibitors (8,942), Sodium-glucose co-transporter 2 (SGLT2) inhibitors (8,837), thiazolidinediones (2,871), and alpha glucosidase (AG) inhibitors (971). Patients on any antidiabetic combinations, a history of PAH, or conditions associated with PAH were excluded. Study groups were mutually exclusive. Analysis was performed using the TriNetX Research Network. Time to onset analysis was conducted using Weibull goodness-of-fit test. Results: Baseline rate of PAH in the control group was 3.95% (8,793/222,644) with a PAH onset of 2.42 weeks. Incidences of PAH were significantly lower in the GLP-1 (2.04%, OR: 0.546) and biguanides (2.78%, OR: 0.723) groups, as compared to control (all p < 0.0001). Time-to-PAH was 8.98 weeks in the GLP-1 group compared to 3.39 weeks in the biguanides group (p< 0001). PAH developed in 4.18% of SGLT2is (OR: 1.01; p = 0.8), 5% of thiazolidinediones (OR: 1.1; p = 0.4), 5.29% of sulfonylureas (OR: 1.3; p < 0.0001), and 5.6% of DPP4is (OR: 1.4; p < 0.0001) groups. Time-to-PAH varied across the treatment groups: 1.82 weeks for SGLT2is, 3.75 weeks for thiazolidinediones, 4.3 weeks for sulfonylureas, and 5.72 weeks for DPP4is. In the AG inhibitors group, 3.57% patients developed PAH. Time-to-PAH was longer in all antidiabetic groups, except SGLT2is, when compared to controls. Conclusion: Patients on GLP-1 and Biguanides presented a lower incidence of PAH and a longer (delayed) time-to-PAH when compared to controls. Longer time-to-PAH may indicate slower PAH progression and characterize the modulatory effects of antidiabetics. Understanding the differential outcomes and onset patterns in PAH could lead to novel combination therapies in the highly comorbid PAH landscape.
Heart valves are living structures whose sophisticated functions are mediated by a specialized population of mesenchymal cells known as valvular interstitial cells (VICs). Given their central role in valve homeostasis, VICs represent a promising cell population for studying heart valve diseases and developing novel therapies to treat them. Here, we describe a strategy for generating VICs from human pluripotent stem cells (hPSCs) by stage-specific manipulation of developmental signalling pathways. Our results demonstrate that hPSC-derived VICs show a high transcriptional similarity to primary human fetal VICs and can secrete key proteins of the valve extracellular matrix. We further investigate the heterogeneity of hPSC-derived VICs and identify two major subpopulations with distinct molecular and functional properties, mirroring the cellular diversity observed in vivo . Finally, we utilize an in vitro model of Noonan syndrome to demonstrate that hPSC-derived VICs can accurately recapitulate key aspects of valve disease. Collectively, these findings provide a reproducible method for the scaled generation of bona fide hPSC-derived VICs and establish their utility in disease modelling and tissue engineering applications. What is new? What are the clinical implications? ### Competing Interest Statement The authors have declared no competing interest.
Synthetic biology offers control over cellular and tissue functions. As it moves beyond microbes into humans, synthetic biology enables precise control over gene expression, cell fate, and tissue organization across heart, lung, blood, and sleep systems. By integrating genome engineering, dynamic gene circuits, and high-dimensional biosensors, these advances support scalable, quantitative models of multicellular biology, expanding the need for systems-level models and integration. We highlight emerging systems such as tunable transcriptional regulators, synthetic organizers, and feedback circuits that bridge molecular control with functional outcomes. Furthermore, by combining omics data with artificial intelligence (AI)-guided circuit design, synthetic biology enables high-resolution cellular and tissue-scale models of development, cellular interactions, drug development, gene therapy, and therapeutic response. Key challenges remain—including delivery, transgene stability, and robust spatiotemporal control in physiologically relevant models. This perspective synthesizes field-spanning progress and defines shared priorities for engineering cells and tissues that function reliably across dynamic, multi-organ environments.
Background Valve remodeling is a complex process involving extracellular matrix organization, development of trilaminar structures, and physical elongation of valve leaflets. However, the cellular and molecular mechanisms regulating valve remodeling and their roles in congenital valve disorders remain poorly understood. Methods Semilunar valves and atrioventricular valves from healthy and age-matched human fetal hearts with pulmonary stenosis (PS) were collected. Single-Cell RNA-sequencing (scRNA-seq) was performed to determine the transcriptomic landscape of multiple valvular cell subtypes in valve remodeling and disease. Spatial localization of newly-identified cell subtypes was determined via immunofluorescence and RNA in situ hybridization. The molecular mechanisms mediating valve development was investigated utilizing primary human fetal heart valve interstitial cells (VICs) and endothelial cells (VECs). Results scRNA-seq analysis of healthy human fetal valves identified a novel APOE + elastin-producing VIC subtype (Elastin-VICs) spatially located underneath VECs sensing the unidirectional flow. Knockdown of APOE in fetal VICs resulted in significant elastogenesis defects. In pulmonary valve with PS, we observed decreased expression of APOE and other genes regulating elastogenesis such as EMILIN1 and LOXL1 , as well as elastin fragmentation. These findings suggested the crucial role of APOE in regulating elastogenesis during valve remodeling. Furthermore, cell-cell interaction analysis revealed that JAG1 from unidirectional VECs activates NOTCH signaling in Elastin-VICs through NOTCH3. In vitro Jag1 treatment in VICs increased elastogenesis, while similar observations were found in VICs co-cultured with VECs in the presence of unidirectional flow. Notably, we found that the JAG1-NOTCH3 signaling pair was drastically reduced in the PS valves. Lastly, we demonstrated that APOE is indispensable for JAG1-induced NOTCH activation in VICs, reinforcing the presence of a synergistic intrinsic and external regulatory network involving APOE and NOTCH signaling that is responsible for regulating elastogenesis during human valve remodeling. Conclusion scRNA-seq analysis of human fetal valves identified a novel Elastin-VIC subpopulation, and revealed mechanism of intrinsic APOE and external NOTCH signaling in regulating elastogenesis during cardiac valve remodeling. These mechanisms may contribute to deciphering the pathogenesis of elastin malformation in congenital valve diseases. Clinical Perspective What Is New? High-resolution single-cell transcriptome atlas generated from healthy human fetal heart valves and valves affected by pulmonary stenosis during the early phase of valve remodeling prior to birth. A unique subset of valve interstitial cells (VICs) that produce elastin (Elastin-VICs) was identified. Elastin-VICs specifically located underneath the valve endothelial cells (VECs) sensing unidirectional flow, and played a crucial role in elastin maturation via the expression of APOE. Elastin-VICs communicated with adjacent VECs via the JAG1-NOTCH signaling, facilitating elastin formation and valve remodeling. What Are the Clinical Implications? Elastin-VICs from patient valvular tissues with Pulmonary Stenosis exhibit decreased APOE-NOTCH signaling and elastin fragmentation. Direct targeting of APOE and NOTCH signaling could be a novel approach to promote elastin fiber formation and valve remodeling in patients with valvular defects.
The human blood-brain barrier (hBBB) is a highly specialized structure that regulates passage across blood and central nervous system (CNS) compartments. Despite its critical physiological role, there are no reliable in vitro models that can mimic hBBB development and function. Here, we constructed hBBB assembloids from brain and blood vessel organoids derived from human pluripotent stem cells. We validated the acquisition of blood-brain barrier (BBB)-specific molecular, cellular, transcriptomic, and functional characteristics and uncovered an extensive neuro-vascular crosstalk with a spatial pattern within hBBB assembloids. When we used patient-derived hBBB assembloids to model cerebral cavernous malformations (CCMs), we found that these assembloids recapitulated the cavernoma anatomy and BBB breakdown observed in patients. Upon comparison of phenotypes and transcriptome between patient-derived hBBB assembloids and primary human cavernoma tissues, we uncovered CCM-related molecular and cellular alterations. Taken together, we report hBBB assembloids that mimic the core properties of the hBBB and identify a potentially underlying cause of CCMs.