The target was MscL-G22S, a gain-of-function variant of the bacterial mechanosensitive channel MscL. MscL responds to membrane tension, while the G22S mutation lowers its mechanical activation threshold and enhances responsiveness to ultrasound without causing spontaneous opening. Researchers delivered the engineered channel to the pancreas with an adeno-associated viral vector. Fluorescence imaging showed expression in insulin-producing beta cells, and RT-qPCR confirmed increased pancreatic MscL messenger RNA. Because the vector was not strictly specific to beta cells, the in vivo metabolic response cannot be assigned exclusively to direct stimulation of those cells.

The team mapped acoustic dose against biological and metabolic outcomes rather than treating ultrasound as a single undifferentiated exposure. In cell experiments using 293T cells and MIN6 mouse beta cells, MscL-G22S localized predominantly to the plasma membrane and did not impair basal viability. Ultrasound-treated cells expressing the channel showed greater viability than negative controls, increased intracellular calcium and stronger glucose-stimulated insulin secretion. The responses were non-linear: intracellular calcium and insulin secretion peaked at 11.76 arbitrary units, while higher exposures produced a plateau or slight decline rather than additional improvement. The in vitro dose series used 5.88, 11.76, 17.64 and 23.52 arbitrary units, with average intensities of 200, 400, 600 and 800 mW/cm², respectively, a 20% duty cycle, a frequency of 0.36 MHz and 30-second irradiations.

A corresponding pattern emerged in the mouse dose-response experiment. After the first 10-day treatment round, glucose reductions were modest and generally not statistically significant. A second round produced clearer dose-dependent differences, and locally estimated scatterplot smoothing showed an inverted U-shaped relationship between acoustic dose and the change in blood glucose. The largest reductions occurred at intermediate doses of approximately 3.5–5.9 arbitrary units, represented by groups G2–G4. These groups received 600 mW/cm² at a 20% duty cycle, 400 mW/cm² at a 40% duty cycle, or 1,000 mW/cm² at a 20% duty cycle. Each was applied for 10 minutes per day across 10 treatment days at 0.36 MHz.

Longitudinal measurements supported the persistence of the metabolic difference. Mice in the three intermediate-dose groups had lower random and fasting blood-glucose levels than mock-treated db/db mice, with significant differences emerging from 14 to 16 weeks of age. Their daily water intake also declined, consistent with alleviation of the polydipsia associated with diabetes, while body weight rose progressively in all diabetic groups without significant differences among treatments. These observations indicate that the optimized exposure was not accompanied by an apparent suppression of overall growth in the reported experiment.

At the endpoint, four weeks after the LIPUS intervention ended, fasting insulin tended to be lower in ultrasound-treated animals, and the homeostatic model assessment for insulin resistance decreased significantly, with the greatest improvements in groups G3 and G4. Glucose- and insulin-tolerance tests were also reported as confirming enhanced glucose handling in moderate-dose groups, although the detailed tolerance results appeared in supplementary material. These are physiological measures in a diabetic mouse model and do not demonstrate restored human pancreatic function or clinical diabetes control.

The optimized stimulation was also associated with changes beyond blood glucose. High-density lipoprotein cholesterol increased significantly in groups G2–G4, while low-density lipoprotein cholesterol, total cholesterol and triglycerides were moderately reduced. Glycosylated serum protein, which the study used as a time-integrated measure of average glycemic exposure over the preceding two to three weeks, was lower in treated animals than in mock controls. By contrast, the ratio of epididymal white adipose tissue to body weight did not differ significantly across groups, so the measured metabolic changes were not associated with a detectable difference in fat mass. Histological examination of major organs in the pilot experiment showed no obvious pathological alterations, but that observation does not establish the long-term safety of repeated stimulation.

Untargeted serum metabolomics provided a wider view of the response. Principal component analysis separated mice receiving ultrasound alone from those receiving ultrasound combined with MscL-G22S, with the first component accounting for 26.62% of total variance. Hierarchical clustering also revealed distinct metabolite patterns. At thresholds of variable importance in projection greater than 1 and a Benjamini-Hochberg false discovery rate below 5%, the analysis identified 237 differentially abundant metabolites: 75 were higher and 162 were lower in the combined-treatment group than in the ultrasound-only group. Pathway classification associated these changes with lipid and sphingolipid metabolism, arachidonic acid metabolism, insulin resistance, inflammatory signaling and broader metabolic and signal-transduction pathways. A separate comparison with an MscL-only group showed limited, less coordinated alterations, consistent with the interpretation that the pronounced remodeling depended on the combination of channel expression and ultrasound activation rather than expression alone.

Histology from the pilot experiment suggested that the physiological response was accompanied by partial preservation of pancreatic islets. Under a conventional 10-day ultrasound protocol, the combination of LIPUS and MscL-G22S improved glucose clearance and reduced fasting glucose relative to mock db/db controls, but individual mice varied considerably in their response. Some islet hypertrophy and structural disorganization remained compared with non-diabetic controls. Immunofluorescence showed a greater insulin-positive beta-cell area and a lower ratio of glucagon-positive alpha-cell area to beta-cell area in the channel-expressing group. These morphological observations cannot establish direct beta-cell activation as the sole mechanism because the viral expression pattern was not cell-restricted and native mechanosensitive channels, including Piezo and TRP-family members, could also respond to ultrasound.

Several limitations constrain the study’s reach. Acoustic dose was expressed partly in relative arbitrary units rather than fully standardized physical measures such as acoustic intensity or mechanical index, which limits direct comparison with other ultrasound experiments. The work used one frequency, 0.36 MHz, without spatial focusing, and employed cultured cells and genetically modified mice rather than humans. The pancreas is located deep in the body, and attenuation of ultrasound energy may make delivery less efficient or precise in larger animals and humans. The durability and safety of repeated stimulation remain uncertain, the cellular signaling cascades connecting channel opening to systemic metabolic remodeling were not fully elucidated, and the sample sizes were generally four to six mice per group. The supported conclusion is therefore bounded: in db/db mice, combining pancreatic MscL-G22S expression with LIPUS within the identified acoustic range produced metabolic improvements that were weaker, more variable or absent outside that range. The study establishes a quantitative framework for further testing, not evidence of a ready clinical intervention.

Subject of Research: Preclinical sonogenetic regulation of glucose metabolism using ultrasound-activated MscL-G22S in diabetic mice

Article Title: Non-invasive glycemic control leveraging ultrasound dose window activation of MscL in diabetic mice

Article References: Zhang, X., Zhan, J., Mao, W., Han, X., Chen, Y., Chen, J., & Wang, Y. (2026). Non-invasive glycemic control leveraging ultrasound dose window activation of MscL in diabetic mice. iScience, 29(11), Article 117962. https://doi.org/10.1016/j.isci.2026.117962

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117962

Keywords: type 2 diabetes, sonogenetics, low-intensity pulsed ultrasound, MscL-G22S, mechanosensitive channels, pancreatic beta cells, db/db mice, glucose regulation, metabolomics, acoustic dose window, Non-invasive, glycemic