Insulin injections and GLP-1 drugs are daily routines for hundreds of millions of people with diabetes. The problem is that this dosing is open-loop — regardless of your current blood glucose level, the drug enters the body at a fixed rhythm. When glucose drops, you risk hypoglycemia; when it rises, the fixed dose may not be enough. Could a drug behave like the pancreas — sensing that glucose is high, producing medicine, then stopping when levels fall?
On August 12, 2026, the team of Ye Haifeng and Guan Ningzi at East China Normal University published a platform called GIFT in *Nature* that does exactly this. The approach is a standard synthetic biology move: decompose a complex biological system into standardized modules, then reassemble them with genetic circuits. They chose *Escherichia coli* Nissle 1917 (EcN), a probiotic with a century-long safety record, as the engineering chassis, and implanted it with a "sense–compute–actuate" genetic circuit.
The molecular switch
The sensing module is a "molecular lock": the HexR transcriptional regulator from *Pseudomonas putida*. When blood glucose rises, the bacteria metabolize glucose and produce a small metabolite called KDPG. KDPG acts as the "molecular key" that fits into the lock, lifting repression and triggering expression of the therapeutic gene GLP-1. When glucose falls back, KDPG decreases, the lock closes, and drug production stops. This is the so-called Sense-and-Respond mode — essentially rebuilding a feedback loop similar to the body's own pancreatic islets.
Validation across two species
Validation spanned two species:
- In db/db diabetic mice, 30 consecutive days of oral GIFT produced serum GLP-1 that rose and fell precisely with blood glucose fluctuations. Fasting glucose, glucose tolerance, insulin sensitivity, and HbA1c all improved significantly, with parallel relief of hepatic steatosis, kidney injury, and intestinal inflammation.
- The tougher hurdle was non-human primates: in a spontaneous type 2 diabetes cynomolgus monkey model, a single oral dose of GIFT lowered blood glucose for up to 3 days.
Why it matters
The background numbers support the urgency: over 530 million people worldwide now live with diabetes. Even GLP-1 drugs like semaglutide have improved efficacy, but the hypoglycemia risk and side effects of open-loop continuous dosing remain a real pain point. If an oral probiotic could produce medicine in a glucose-dependent closed loop, "daily injections" could be replaced by "culturing a colony of microbes that work on their own."
Remaining hurdles
1. Colonization and safety: the persistence of oral bacteria in the gut and their long-term safety have not yet been verified in humans. 2. Generalization: GLP-1 is only one pathway in diabetes; whether this Sense-and-Respond framework can extend to other hormones and metabolic diseases requires more genetic circuits. 3. Delivery and regulation: dosing, individualized control precision, and the regulatory and manufacturing gates of engineered bacteria still stand between this and clinical reality.
What GIFT changes is the paradigm: shifting "drug delivery" from external timed dosing to an adaptive in-vivo biofeedback device. The endpoint of this path is not necessarily diabetes — any chronic disease requiring real-time response to physiological state could, in theory, use the same framework. Three-day efficacy in a monkey model is already enough to move "living drugs" one step from concept toward the clinic.