Diabetes is, at its root, a shortage of working insulin-producing beta cells. Researchers have found a strikingly simple way to make more: switch off a single gene in a different type of pancreatic cell.

Scientists at Harvard Medical School and the Joslin Diabetes Center, led by Peng Yi, reported in Science Translational Medicine that disabling ALDH3B2 converts human pancreatic duct cells into functional, glucose-responsive beta-like cells.

Why beta cells are the whole problem

In type 1 diabetes, the immune system destroys beta cells, and patients depend on injected insulin for life. In type 2, beta cells initially compensate for insulin resistance by working harder, then progressively falter. Different diseases, converging on the same deficit.

Adult humans regenerate beta cells very poorly. Once they are gone, they largely stay gone — which is why replacing them has been a central goal of diabetes research for decades.

Why duct cells are a sensible starting point

The pancreas contains several cell types with a shared developmental origin. Duct cells — which form the plumbing carrying digestive enzymes to the intestine — are relatively abundant, accessible, and, crucially, closely related to beta cells in developmental lineage.

Converting one adult cell type into another, transdifferentiation, is far easier between related cells than between distant ones. Prior work in this field established that pancreatic cells can be pushed toward a beta-cell state using combinations of developmental transcription factors. What is notable here is the simplicity: one gene, switched off.

How they found it

The team ran a genome-wide CRISPR screen — systematically disabling genes across the genome and looking for any whose loss shifted cell identity. That is an unbiased approach: rather than testing a hypothesis about which genes should matter, it lets the screen surface candidates nobody would have predicted.

ALDH3B2 emerged. Losing it switches on the insulin gene and shifts the cell’s overall programme toward a beta-cell state — suggesting the gene normally acts as a brake holding duct cells in their assigned identity.

What happened in animals

Transformed human cells were transplanted into diabetic mice under the kidney capsule — a standard site for testing cell grafts because it is well vascularised and the graft can be retrieved. The cells lowered blood glucose to near-normal for about six weeks and showed strongly increased beta-cell marker gene expression.

Near-normal glucose control from transplanted human cells is a meaningful functional result, not merely a change in gene expression.

The limitation the authors are explicit about

The converted cells produced much less insulin than natural beta cells when stimulated by glucose.

This is the gap between proof of concept and therapy, and it should not be minimised. Beta cells are specialised secretory machines: they sense glucose continuously and release insulin in a precisely graded, rapid response. A cell that expresses insulin and responds somewhat to glucose has taken on the identity but not the full performance. In a mouse with modest metabolic demand, partial function can normalise glucose; in a human, the requirement is considerably greater.

Where this sits against the alternatives

Beta-cell replacement is a crowded field. The most advanced approach derives islet cells from stem cells, and has produced patients achieving insulin independence — a genuine milestone, but one requiring immunosuppression to protect the graft.

The appeal of converting a patient’s own duct cells is that it could sidestep the rejection problem entirely, since the cells would be autologous. That is a real advantage over donor-derived or allogeneic stem-cell products.

The autoimmunity problem nobody escapes

One caveat applies to every beta-cell replacement strategy and deserves stating plainly. In type 1 diabetes, the immune system destroyed the original beta cells because it recognises them as targets. New beta cells — whether grown from stem cells or converted from a patient’s own duct cells — present the same targets.

Being autologous solves transplant rejection. It does not solve autoimmunity. Any durable therapy for type 1 will need to pair cell replacement with immune protection, whether through encapsulation, engineered immune evasion, or tolerance induction.

For type 2 diabetes, where beta cells fail from exhaustion rather than immune attack, that particular obstacle does not apply — which may make it the more tractable near-term target for this approach.

What optimisation would have to achieve

The insulin-output shortfall is a specific, addressable engineering problem rather than a vague caveat, and it is worth understanding what closing it involves.

A mature beta cell does not simply contain insulin. It maintains large reserves of pre-made hormone in secretory granules, senses glucose through a tightly calibrated metabolic pathway, and releases insulin in two phases — a rapid initial burst followed by sustained secretion. Converted cells that express insulin and respond partially to glucose have acquired the identity without the full apparatus.

Routes forward include combining ALDH3B2 loss with additional factors that drive maturation, extending the time cells spend maturing before transplantation, or improving the graft environment so cells continue developing in place. Stem-cell-derived islet programmes faced exactly this maturation problem and spent years solving it, which is both a caution about the timeline and evidence that the problem yields to sustained work.

It is a promising proof of concept requiring substantial optimisation, not a therapy. Preclinical research; not medical advice.