Engineered Blood Vessels Unlock Functional Maturity in Lab-Grown Insulin-Producing Islets

Pancreatic beta-cells derived from human pluripotent stem cells — known as SC-islets — hold enormous promise for diabetes research and cell replacement therapy. However, a persistent problem has been that these lab-grown beta-cells remain functionally immature, failing to mount robust insulin responses to glucose stimulation the way native beta-cells do. A critical missing element is vasculature: in the body, each islet is densely woven with blood vessels that support glucose sensing, insulin release, and paracrine signaling. This study set out to build that vascular architecture directly into SC-islet organoids and measure the functional consequences.

The team developed two complementary vascularized models. In the first, SC-islet cells were combined with human primary endothelial cells (ECs) and fibroblasts in a 3D fibrin gel, with a carefully optimized medium-switching protocol (gradually shifting from 75% vascular to 100% islet medium) that preserved both the vascular network and SC-beta-cell insulin content. In the second model, this cellular mixture was loaded into a microfluidic organ-on-a-chip device where interstitial flow drove formation of a perfusable vascular network surrounding the islets over six days. Both platforms successfully generated vasculature that tightly wrapped around — though rarely penetrated — the SC-islets.

To quantify beta-cell function without relying on conventional insulin secretion assays (which were confounded by the fibrin gel), the team engineered a GCaMP6f calcium reporter into human embryonic stem cells. This allowed real-time fluorescence imaging of intracellular calcium influx — a direct proxy for insulin secretion. In non-vascularized SC-islets, only 17 ± 12.5% of beta-cells exhibited a dual calcium response to both high glucose and Exendin-4, while 51.5 ± 25% were completely non-responsive. In the static vascularized model, the proportion of dual-responsive cells increased significantly, and in the microfluidic perfused model, 26 ± 15.3% of beta-cells showed dual responsiveness along with substantially prolonged calcium influx duration — a hallmark of more mature beta-cell function not seen in non-perfused conditions.

In vivo transplantation experiments provided compelling validation. When a subtherapeutic dose of SC-islets (insufficient on its own to reverse diabetes) was implanted into diabetic mice, vascularized grafts reversed hyperglycemia significantly faster than non-vascularized grafts. Single-cell RNA sequencing of the organoids revealed that ECs deposit a rich basement membrane — including collagen IV, laminins, and fibronectin — that closely resembles native islet basement membrane composition, and blocking integrin-beta-1 signaling (the receptor for these ECM components) abrogated the vascularization-induced functional improvement. Additionally, ligand-receptor analysis predicted strong BMP2/4–BMPR2 signaling from ECs to beta-cells; exogenous BMP4 treatment alone was sufficient to enhance calcium responses and insulin secretion in non-vascularized SC-islets, confirming this as a direct paracrine mechanism.

This work establishes a mechanistic framework explaining why in vivo engraftment matures SC-beta-cells, and provides the first physiologically relevant 3D vascularized islet organoid platform for studying these interactions ex vivo. Limitations include the use of a subtherapeutic transplant dose (making definitive therapeutic conclusions premature), the fact that vasculature rarely penetrated the SC-islet core, and the absence of immune cells and neurons that also inhabit the native islet niche. Nonetheless, this platform represents a major advance for diabetes modeling, drug discovery, and the eventual development of cell-based therapies.