Nature Biomedical Engineering

Body surface gastric mapping (BSGM) enables non-invasive measurement of gastric electrophysiology, but the signals are approximately 100 times weaker than cardiac potentials and overlap spectrally with motion artifacts, necessitating labor-intensive manual review that limits clinical scalability. We present an uncertainty-aware deep learning framework combining a signal reconstruction network with a parallel uncertainty estimation network to automate artifact correction in high-resolution BSGM. Models were trained on 2,398 multihour, 64-channel recordings from 27 international clinical sites using weak supervision, a physiology-aware loss function, and uncertainty-gated quality control. In an independent cohort of 127 patients, the system achieved relative reductions of 39% in signal reconstruction error, 9% in total data removed, and 23% in amplitude--movement correlation compared with the industry-standard Wiener filter. Improved signal fidelity altered automated clinical phenotyping in 7% of patients by recovering previously obscured gastric rhythms. Uncertainty-aware deep learning enables reliable automated artifact correction in body-surface gastric mapping, improving signal fidelity and enabling scalable clinical interpretation. The system is FDA-cleared (510(k) K252504) and deployed in clinical practice, demonstrating that data-driven artifact correction can meet regulatory requirements for medical devices and reduce dependence on specialist manual review.

The use of cellular systems to advance cancer therapeutics has expanded rapidly, spanning cell therapies to patient-specific tumor models. Platforms that recapitulate key features of the tumor microenvironment, including vascular and immune components, hold significant potential to improve the predictive power and translational relevance of preclinical models. Here, we report a vascularized tumor organoid platform that combines self-organizing microvascular networks with patient-derived tumor organoids and tumor-infiltrating lymphocytes. To minimize non-specific endothelial immunogenicity and enable broader compatibility across patient samples, we engineered the vasculature using {beta}2-microglobulin-knockout endothelial cells. Leveraging this system, we established patient-specific, lymphocyte-incorporated tumor models that enabled quantitative assessment of T cell infiltration. In conjunction with immune checkpoint blockade, this platform distinguishes responder and non-responder patient samples, consistent with the clinical observations. Single-cell RNA-sequencing revealed tumor-intrinsic and immune-associated programs underlying this stratification, identifying tumor-driven hyperangiogenic signaling as a barrier to T cell extravasation. Pharmacological co-targeting of PD1 and VEGF restored T cell infiltration in non-responder organoids, shifting them from an immune-excluded to an immune-inflamed state. Together, this vascularized tumor organoid platform provides a predictive and mechanistic framework for modeling patient-specific immunotherapy responses and design of combination therapies.

Medical imaging models depend on large, shareable datasets, yet privacy constraints limit data dissemination. Current text-conditioned diffusion models fail to preserve subtle, distributed clinical signals, such as continuous physiological biomarkers, rendering synthetic data insufficient for robust downstream physiological modeling. Here, we evaluate image-to-image (I2I) diffusion as a tunable, privacy-preserving transformation that produces a synthetic counterpart of real images while preserving downstream-relevant information. We fine-tune Stable Diffusion with low-rank adapters on retinal fundus photographs and chest radiographs, assessing fidelity, clinical signal preservation, cross-site transfer, and empirical re-identification risk. I2I consistently outperforms text-to-image generation in image fidelity and in preserving biomarker information. In cross-cohort transfer to an external retinal dataset from the UK Biobank, pretraining on I2I synthetic data performs comparably to real-image pretraining and surpasses it in the smallest fine-tuning sets. Varying I2I strength reveals that the privacy-utility tradeoff is highly modality-dependent: while retinal images achieve practical de-identification, chest X-rays exhibit structural combinatorics that leave them substantially re-identifiable even at high noise strengths, exposing critical boundaries for diffusion-based anonymization. These results position image-conditioned diffusion as a practical approach for generating shareable medical images with tunable de-identification.

Genetic evidence implicates immune dysfunction in Alzheimers disease (AD), yet human-specific neuroimmune mechanisms remain poorly defined. Here, we establish a modular human forebrain organoid platform that systematically integrates iPSC-derived microglia and CD8+ T cells to reconstitute multicellular Alzheimers disease pathology. This system enables functional interrogation of both innate and adaptive immune components in a human-relevant context. Using this platform, we demonstrate that microglia mediate amyloid-{beta} clearance and neuronal maturation but also drive inflammatory activation and recruit CD8+ T cells through CCL4/5-CXCL10 signaling via CCR1/5 and CXCR3, establishing a neuroinflammatory feedback loop. Pharmacological targeting of CCR5 or CXCR3 blocks T cell recruitment and modulates autophagy in a microglia-dependent manner. This modular organoid platform provides a versatile tool for dissecting neuron-immune interactions and enables cell-type-specific therapeutic screening in human neuroinflammatory disease models. HighlightsO_LIA modular human forebrain organoid platform integrating innate (microglia) and adaptive (CD8+ T cells) immunity recapitulates key features of the human neuroimmune environment C_LIO_LIEnables mechanistic dissection of multicellular interactions underlying Alzheimers disease pathology C_LIO_LIOvercomes limitations of traditional animal models by resolving human-specific, cell-type-specific neuroimmune mechanisms C_LIO_LIEstablishes a new approach methodology for studying neuroimmune disorders and advancing drug discovery C_LI

In vivo CAR-T cell therapy eliminates manufacturing complexities associated with ex vivo autologous approaches, but safety concerns have limited adoption. We developed viroVbot, a next-generation in vivo CAR-T platform, by combining computational immunogenicity prediction (CIMMEXTM) with envelope engineering. Screening 22,562 glycoprotein sequences, we identified 641 vesiculovirus homologs, from which we selected Piry virus glycoprotein (PIRYV) as the optimal candidate. PIRYV exhibited lower MHC-epitope density, reduced human seroprevalence, with decreased T cell activation compared to VSV-G. To enhance targeting specificity, we engineered receptor-binding-deficient PIRYV (ePIRYVRBD) displaying CD3/CD7 nanobodies for T cell-selective transduction. To maximize safety, we engineered CAR-TRAP producer cells to eliminate unwanted B cell transduction and incorporated machine learning-optimized T cell-specific promoters that restrict CAR activation exclusively to lymphocytes. Additional modifications suppressed hepatocyte expression and prevented phagocytic uptake. In humanized xenograft models, viroVbot3 generated potent BCMA/CD19 specific CAR-T responses against multiple myeloma and Claudin18.2-targeting gastric cancer, demonstrating sequential redosing with alternative envelopes. Critically, viroVbot3 exhibited minimal off-target organ biodistribution with CAR expression restricted to T lymphocytes. These findings establish viroVbot as a low-immunogenic platform for scalable in vivo CAR-T manufacturing with capability for sequential redosing across hematologic and solid tumors.

Insulin therapy revolutionized the care of patients with diabetes starting [~]100 years ago, yet insulin-induced hypoglycemia remains a serious life-threatening complication of insulin therapy. Glucagon is a highly effective treatment; however current dosage forms remain under-utilized due to poor patient compliance. The development of improved and situation-specific glucagon therapies remains challenging due to the poor drug stability and incomplete knowledge of the kinetics of different hypoglycemic events. Thus, we analyzed continuous glucose monitor (CGM) data from 1135 patients with type 1 diabetes (T1D) representing 246.18 patient years. We show that a surprisingly large proportion of hypoglycemic episodes (20-30%) are follow-on events resulting from under-treatment of prior events, and that the average duration of independent hypoglycemic events can last up to 79 to 108 minutes. We further show that the kinetics of hypoglycemic onset and persistence varies significantly by patient history, severity, time of occurrence. Guided by these findings, we recognize the opportunity to develop high-density, readily-soluble, and thermostable (ReST) solid glucagon formulations, and painless application-specific microneedle-patches that are in line with the timing needs of T1D patients who are awake and asleep. Thus, we demonstrate (1) on-demand patches for rapid prevention or treatment of mild hypoglycemia during the day, and (2) enzyme-driven hypoglycemia-responsive patches supporting autonomous glucagon release during the night. We show excellent in vitro glucagon stability, loading, and release kinetics of both systems and demonstrate their ability to treat hypoglycemia in diabetic animals. The engineering of these delivery systems demonstrates the potential of human CGM data and solid glucagon formulations to enable new modes of glucagon therapy, thereby expanding the clinical role of glucagon beyond the emergency setting.

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. This U-Net-GAN model achieved high predictive accuracy, with structural similarity index (SSIM) values up to 0.84 using combined morphological and structural inputs. To further enhance performance and generalizability, we incorporated synthetic cell-function pairs generated via a generative AI StyleGAN2 framework, improving prediction accuracy and perceptual similarity. Importantly, 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 and interpretable strategy for inferring cardiomyocyte function from static morphology, eliminating the need for time-lapse imaging. More broadly, this work positions generative AI as a powerful tool for bridging cellular structure and function, enabling high-throughput functional phenotyping and advancing in vitro cardiac modeling.

Efficient delivery of genetic cargo to primary human T cells remains a critical barrier for cell-based immunotherapies. AAVs are widely used but limited by immune neutralization and poor T cell transduction efficiency. We present NEO-AAV, an engineered fusion protein (PH-ALG2-PKD12) that recruits AAV capsids into endogenous extracellular vesicles (EVs) via PI(4,5)P2-directed membrane targeting, multivalent PKD12-capsid clustering, and ALG2-mediated ESCRT machinery recruitment, yielding EV-enveloped particles with improved AAV loading compared with passive EV-AAV controls. NEO-AAV displayed enhanced resistance to an anti-AAV6 neutralizing antibody (ADK6), maintaining transduction at concentrations that neutralized naked AAV6 and outperformed passively formed EV-AAV6. Surface display of a CD7/CD3/CD28 tri-chimera enabled single-step activation and transduction of CD7 T cells within PBMCs, reaching [~]38% eGFP cells without exogenous pre-activation. As a proof-of-concept, NEO-AAV generated functional CAR-T cells from primary human PBMCs, exhibiting antigen-specific IFN-{gamma} release and cytotoxicity against CD19+ target cells. CAR expression peaked at day 5 and declined by day 15, consistent with episomal AAV kinetics, framing NEO-AAV as an activation-coupled delivery module rather than a durable solution. Together, NEO-AAV provides a programmable EV-enveloped AAV platform with improved immune shielding and single-step T cell transduction, offering a building block for immune cell engineering with in vivo potential.

Adaptive deep brain stimulation (aDBS) relies on physiological biomarkers to infer motor state and guide therapeutic stimulation in Parkinson's disease. However, neural biomarkers may themselves be altered by stimulation, potentially limiting their utility for closed-loop control. We address this limitation by testing whether DBS device-embedded accelerometers can accurately track Parkinsonian motor state across stimulation conditions. We analysed over 1,900 hours of chronic recordings of subthalamic nucleus (STN), sensorimotor cortical and device-embedded accelerometry signals acquired before and during continuous STN stimulation, alongside continuous wearable assessments of bradykinesia and dyskinesia. Across stimulation conditions, accelerometry-derived features robustly tracked motor symptom severity and outperformed neural features for symptom decoding. Mechanistically, total STN beta power - a widely used biomarker for aDBS - proved less informative because it conflates periodic and aperiodic neural processes with opposing relationships to motor state. Under active stimulation, periodic beta activity showed reduced coupling to symptom severity, whereas STN aperiodic activity, cortical periodic activity and cortico-subthalamic coherence remained comparatively stable. Together, these findings demonstrate that neural and behavioural biomarkers exhibit differential robustness during deep brain stimulation and identify device-embedded accelerometry as a robust behavioural biomarker of motor state, motivating its use in next-generation adaptive DBS systems.

Background and Aims: Endoscopic artificial intelligence is commonly validated on selected single images, whereas gastric cancer interpretation requires integrating whole examinations. We developed GutCore and evaluated whether whole-case endoscopic images could be used for patient-level assessment of gastric cancer depth, biomarkers, and prognosis. Methods: GutCore was pretrained on 5.6 million de-identified endoscopic images from more than ten hospitals. We compared it with five general, medical, and endoscopy-specific foundation models using open image-level datasets and an internal tertiary-center cohort of 11,035 de-identified endoscopic examinations (2019-2023): 8,049 with early or advanced gastric cancer and 2,986 with benign gastritis or intestinal metaplasia. All examination images were aggregated for patient-level assessment of cancer status, invasion depth, molecular biomarkers, and overall survival. Results: Aggregating all stored images from each examination enabled patient-level gastric cancer assessment without selecting representative frames. GutCore achieved AUCs of 0.995 for cancer detection, 0.960 for muscularis propria invasion, and 0.804 for SM2-or-deeper invasion. Prediction of tissue-defined biomarker status was strongest for Epstein-Barr virus status and MLH1 loss (AUC, 0.831 and 0.854), with lower HER2 performance (AUC, 0.673). In the held-out advanced gastric cancer test set, GutCore-derived risk groups showed marked survival separation (log-rank P < .0001; high-risk vs low-risk hazard ratio, 13.18; 95% CI, 6.06-28.66), with stratification persisting within pathological stage II and III disease. External frame-level benchmarks showed strong performance for anatomical landmark recognition, disease grading, and segmentation. Conclusions: GutCore supported whole-case patient-level gastric cancer assessment using routinely stored endoscopic images. Further validation in independent clinical cohorts is needed to establish generalizability and clinical utility.

Objective To verify the reliability of a self developed bowel sound monitoring device under real biological tissue acoustic propagation conditions using a controllable sound source, and to establish quantitative evidence for its translational applicability. Methods Freshly euthanized six month old Bama miniature pigs were used as an experimental model. A high fidelity Bluetooth audio playback device was implanted into the abdominal cavity to deliver manually annotated bowel sound recordings as controllable acoustic stimuli. A self developed bowel sound monitoring device was fixed on the abdominal surface for continuous signal acquisition. Playback timestamps were defined as the ground truth, and event level matching was performed within a predefined temporal tolerance window. Four performance indicators were evaluated: (1) bowel sound acquisition and energy amplification, (2) event matching accuracy, (3) acoustic feature consistency, and (4) subjective agreement assessed by blinded auscultation from gastroenterologists with different levels of clinical experience. Results The monitoring device exhibited stable detection capability and effectively covered the full spectral range of the original signals. It significantly enhanced bowel sound energy while preserving temporal and spectral characteristics, demonstrating high consistency in time and frequency domain features. Blinded clinician assessments showed a subjective agreement rate of 88.9% between original and surface recorded bowel sound events. Conclusions Under real tissue acoustic propagation conditions, the self-developed bowel sound monitoring device reliably captures bowel sound events with high temporal accuracy, acoustic fidelity, and clinical perceptual consistency. This controllable sound source based validation provides robust technical evidence for subsequent in vivo studies and clinical translation, supporting the development of objective and continuous gastrointestinal function monitoring.

Synthetic gene-regulation logic is established in immortalized cell lines but remains largely aspirational in human induced pluripotent stem cells (hiPSCs) and derivatives. This gap constrains both mechanistic discovery and translational engineering in physiologically relevant models. We developed CIRI (Combinatorial Inducible CRISPR in IPSCs), an isogenic, safe-harbor-engineered platform in which tetracycline-responsive single guide RNAs (sgR-NAs) carry modular RNA aptamers that recruit RNA-binding proteins and effector domains. This design enables multimodal regulation from a single catalytically inactive Cas9 (dCas9), exemplified by orthogonal CRISPR activation and interference (CRISPRa/i). After optimizing sgRNA-aptamer architectures, we achieved robust CRISPRa and CRISPRi in hiPSCs and hiPSC-derived cardiac organoids. CIRI rapidly channels hiPSC forward programming into skeletal myocytes by activating MYOD1 while repressing NANOG, POU5F1/OCT4, and SOX2. Combinatorial pooled dual-guide single-cell RNA sequencing screens identify ID3 as a road-block and KDM6B and SMARCD3 as synergistic enhancers of myogenic maturation. Together, CIRI establishes a programmable synthetic biology framework in human stem cell models. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=177 SRC="FIGDIR/small/729073v1_ufig1.gif" ALT="Figure 1"> View larger version (53K): org.highwire.dtl.DTLVardef@8723dforg.highwire.dtl.DTLVardef@440adforg.highwire.dtl.DTLVardef@125ce33org.highwire.dtl.DTLVardef@104b25b_HPS_FORMAT_FIGEXP M_FIG C_FIG

IgG-degrading enzymes have emerged as innovative therapeutic agents for treating conditions driven by pathogenic antibodies. Here, we used structure-guided rational design to engineer IdeSM33, a double mutant (K167R/D226E) of the IgG-specific bacterial protease IdeS from Streptococcus pyogenes, with improved catalytic efficiency. Biolayer interferometry revealed a fourfold increase in binding affinity relative to wild-type IdeS (IdeSWT). This enhancement is likely attributable to mutations that strengthen hydrogen bonding at the enzyme-IgG Fc interface. In vitro, IdeSM33 has higher performance than IdeSWT in cleaving serum IgG. In vivo studies in rabbits demonstrated that IdeSM33 effectively depleted circulating IgG and showed better performance at a dose of 0.005 mg/kg than the IdeSWT. Although doses greater than 0.2 mg/kg demonstrated higher plasma concentrations of IdeS and a larger AUC 0 to last, they did not show a significant enhancement in the pharmacodynamics of IgG degradation. Importantly, a single dose of IdeSM33 (0.2 mg/kg) potently degraded binding and neutralizing antibodies against AAV9 within 1-2 days and restored hepatic AAV9 transduction in pre-immunized animals. Together, these findings highlight IdeSM33 as a potent and safe engineered enzyme with therapeutic potential for autoimmune disorders, transplant rejection, and overcoming pre-existing humoral immunity in gene therapy.

Many diseases arise from dysfunction of defined cell populations, yet most therapeutics distribute broadly, limiting efficacy and causing toxicity. We developed ExACT, a platform for cell-type-selective intracellular delivery that exploits membrane transporters. In vivo screening of combinatorial fluorescent small-molecule libraries in mouse brain identified chemistries whose uptake is dictated by endogenous transporter expression, yielding compounds with preferential entry into neurons, astrocytes, pericytes and endothelial cells. One series showed strong selectivity for brain and retinal endothelium, where Slco1a4 mediated uptake. This selectivity principle extended to the human orthologue SLCO1A2, highly expressed in brain endothelium and oligodendrocytes, where it mediated selective uptake in a humanized mouse model and human iPSC-derived oligodendrocytes. Ectopic expression of SLCO1A2 in neurons via gene therapy created a synthetic entry port, conferring ExACT conjugate uptake on otherwise inaccessible cells. Bifunctional compounds linking transporter-targeting motifs to antisense oligonucleotides or small-molecule drugs retained pharmacological activity while conferring transporter-dependent cell-type selectivity, illustrating how transporter diversity can be harnessed for precision pharmacotherapy.

Spatial transcriptomics couples hematoxylin and eosin (H&E) tissue morphology with spatially resolved gene expression (GE). However, generative models that exploit this coupling to synthesize tissue images from transcriptomic profiles remain scarce. We present STMDiT (Spatial Transcriptomics and Morphology Diffusion Transformer), a diffusion transformer that synthesizes H&E histopathology patches conditioned jointly on morphological embeddings and transcriptomic profiles. Building on PixCell (Yellapragada et al., 2025), we integrate gene expression from a frozen CancerFoundation encoder (Theus et al., 2024) through adaptive layer normalization and per-block cross-attention, and we train under dual classifier-free guidance with independent modality dropout. On the 10x TuPro Visium melanoma cohort, GE conditioning improves both image quality over the no-GE PixCell-B baseline (best FID = 252.9 vs 330.7) and transcriptomic fidelity (best AUC = 0.267 vs 0.229, reaching 82% of the real-tile ceiling). Training with DeepSpots predicted-transcriptomics pseudo-labels (PTPL) uniquely transfers zero-shot to TCGA SKCM, an out-of-distribution (OOD) H&E-only melanoma cohort: PTPL-XAttn-PMA-B reaches FID = 690.0, a 57-point improvement over the no-GE baseline (747.1), with a within-model GE-ablation effect of {Delta}OOD = +309.5, enabling virtual tissue synthesis beyond native spatial-transcriptomics coverage. Our results indicate that gene-expression conditioning produces morphologically distinct tissue images and supports virtual tissue simulation for hypothesis testing in computational pathology.

Over 200 monoclonal antibodies (mAbs) are approved for clinical use, yet their therapeutic potential is constrained by dependence on repeated injections or infusions that drive non-adherence, limit access in low-resource settings, and generate peak-trough pharmacokinetics linked to adverse effects and reduced efficacy. Here, we developed an immunomodulatory encapsulated cell-based biologics factory that overcomes mAb instability, immunogenicity, and the fibrotic foreign body response that have limited previous approaches, enabling continuous in situ production of therapeutic antibodies from a single administration. Screening chemically modified alginate biomaterials in immunocompetent mice identified a lead immunomodulatory alginate formulation that sustains stable serum titers of the HIV-neutralizing mAb 3BNC117 for one year. Single-cell RNA sequencing revealed that this formulation promotes a local anti-inflammatory, pro-resolving immune niche that attenuates fibrosis. The platforms versatility was demonstrated by production of thirteen diverse mAbs from an allogeneic cell chassis, with sustained in vivo delivery of a subset including ipilimumab, pembrolizumab, adalimumab, and PGT121. Integration into a retrievable macrodevice enabled on-demand therapeutic termination and re-implantation for dose-proportional tuning. In a non-human primates, subcutaneous implantation maintained stable ipilimumab titers for over six months with no detectable toxicity, anti-drug antibodies, or adverse events, and dose-dependent exposure was confirmed across a three-dose escalation. These results demonstrate a clinically translatable platform offering a practical strategy to replace frequent injections with single-administration therapy.

Noninvasive monitoring of plaque inflammatory dynamics remains an unmet need. We previously developed a monocyte-mimetic nanoprobe, termed MoNP-SPION, for MRI detection of atherosclerotic lesions. Here we demonstrate MoNP-SPION enables longitudinal tracking of plaque inflammatory status in a clinically relevant mouse model. Following 16 weeks of plaque induction, mice were maintained on high-fat diet or switched to chow for 6 weeks to model persistent versus resolving plaque inflammation. MoNP-SPION-enhanced MRI was performed at 3- and 6-weeks post-adjustment, and arterial tissue was collected for histological assessment. Mice maintained on high-fat diet exhibited persistent hypointense T2* signal at the carotid bifurcation and aortic root, whereas chow-transitioned mice showed progressive signal attenuation, consistent with histological evidence of reduced plaque burden and inflammation. These findings establish MoNP-SPION as an effective molecular MRI probe for longitudinal assessment of plaque inflammatory dynamics, supporting its potential for monitoring atherosclerosis progression and therapeutic response.

Adeno-associated virus (AAV) gene therapy holds immense promise for treating muscular dystrophies, yet its efficacy and safety are constrained by the suboptimal tissue tropism of natural serotypes. Here, we employed the REACH platform, which combines rational design and directed evolution, to engineer muscle targeting vectors. Systemic administration in non-human primates (NHPs) revealed that lead candidate M1 mediates a >10-fold increase in skeletal muscle transduction compared to the AAV9 and 2-3 fold higher than MyoAAV, while concurrently achieving a remarkable 183-fold reduction in liver distribution. Furthermore, M1 exhibited significant de-targeting from key off-target tissues, including dorsal root ganglia (11 fold), lung (27 fold), spleen (2 fold), and kidney (2 fold). These findings demonstrate that the REACH platform can generate AAV capsids with simultaneously enhanced muscle tropism and favorable safety profiles, addressing a critical bottleneck in muscle-directed gene therapy.

Extracellular vesicles (EVs) are promising vehicles for nucleic acid delivery, yet efficient delivery of circular RNA (circRNA) remains challenging due to inefficient loading and limited intracellular expression. Here, we establish an EV-based platform that enables efficient circRNA delivery via in situ biogenesis and sorting. By optimizing intracellular circularization and translation through vector design, we markedly enhance circRNA expression. By combining this with Snu13-mediated EV sorting and enhanced vesicle biogenesis, we achieve efficient packaging of circRNA without compromising vesicle integrity. This integrated strategy enables robust and sustained protein expression following EV-based circRNA delivery. By leveraging this platform, we demonstrate a dendritic cell-targeting circRNA vaccine that elicits strong antigen-specific CD8+ T cell responses and antitumor efficacy. We further show that systemic delivery of BNP-encoding circRNA attenuates doxorubicin-induced myocardial fibrosis. Together, this work establishes a generalizable platform for circRNA therapeutics by overcoming key barriers in circRNA expression and EV-mediated delivery. TeaserEngineered EVs enable efficient circRNA delivery for sustained protein expression and therapy.

We report the discovery of novel adeno-associated virus (AAV) capsid variants engineered for superior intravitreal (IVT) gene delivery to the primate retina. Utilizing the REACH platform, we constructed a diverse AAV variant library and employed a multi-stage screening strategy involving in vitro selection on human retinal pigment cells followed by direct in vivo screening in non-human primates (NHPs). Following IVT administration in NHPS of a barcoded variant pool, next-generation sequencing analysis of retinal tissues identified lead candidates (e.g., E52, E54, and E57) that achieved transduction levels in the neural retina and RPE 5-10 fold higher than the benchmark R100. Concurrently, these high-potency variants exhibited an exceptional ocular confinement profile, with minimal to undetectable vector genome distribution in systemic organs. This combination of markedly enhanced retinal transduction and stringent local tropism establishes these engineered capsids as promising next-generation vectors for the treatment of inherited and acquired retinal diseases via a minimally invasive IVT route.