Study identifies three genes linked to dapagliflozin's effects in type 2 diabetes

Dapagliflozin has become a standard treatment for type 2 diabetes, improving blood sugar control and protecting the heart and kidneys. Yet doctors and researchers have long wondered exactly how the drug works at the molecular level—what switches it flips inside cells, what genes it activates or silences. A new study from researchers in Guiyang, China, offers some answers by tracking gene expression changes in diabetes patients before and after taking the medication.

The team collected blood samples from type 2 diabetes patients at two points: before starting dapagliflozin and after treatment began. They then sequenced the RNA in those samples to see which genes were turned up or down in response to the drug. The analysis identified 137 genes that changed expression following treatment. But 137 genes is too many to study meaningfully, so the researchers used three different machine-learning algorithms—LASSO regression, random forest, and Boruta—to narrow the field and find the most important ones. All three methods converged on the same three genes: TAS2R60, GPLD1, and GPR42. All three were upregulated, meaning the drug caused cells to produce more of their protein products.

Of the three, GPR42 emerged as the most influential player. When the researchers mapped out the regulatory network—the web of connections between genes, their protein products, and the molecules that control them—GPR42 sat at the center with the most connections to other elements in the system. This suggested it might be a key node through which dapagliflozin exerts some of its effects. The researchers also looked at immune cells, using a technique called single-sample gene set enrichment analysis to estimate how abundant different immune populations were in the patients' blood. They found that GPR42 expression correlated strongly with two types of immune cells: central memory CD4+ T cells and memory B cells. This hints that the drug may work partly by reshaping the immune landscape in ways that help control inflammation, a hallmark of type 2 diabetes.

The researchers used computational docking—a technique that predicts how molecules fit together—to model whether dapagliflozin might bind directly to the proteins made by these three genes. The models suggested plausible binding conformations, though the team was careful to note that computational predictions do not prove actual physical interactions occur in living cells. When they validated the findings using RT-qPCR, a gold-standard technique for measuring gene expression, they confirmed that GPLD1 and GPR42 were indeed upregulated. TAS2R60 showed an upward trend but the increase was not statistically significant.

The work was funded by three grants from health and science agencies in Guizhou Province, all overseen by researcher Zhaohui Jiang, and was approved by the ethics committee at The First People's Hospital of Guiyang. The study was registered with the Chinese Clinical Trial Registry in June 2026. The authors declare no competing financial interests.

What the study does not yet do is prove that these three genes are necessary for dapagliflozin to work, or that measuring them could help doctors predict which patients will respond well to the drug. The researchers themselves acknowledge that their findings need validation in larger, independent patient populations and in controlled laboratory experiments that directly test the biological roles of these genes. For now, the work is a molecular snapshot—a map of what changes when dapagliflozin does its job, and a starting point for understanding the mechanism.