Your kidneys do far more than filter waste. Deep inside each kidney, microscopic tubes called proximal tubules constantly reclaim glucose, amino acids, and other molecules from the fluid that will eventually become urine. One particular stretch of those tubes, called the S3 segment, sits at the boundary between the outer and inner kidney layers and handles a surprisingly large share of glucose reabsorption. Under diabetic conditions, glucose floods this segment in amounts it was never designed to manage.
A research team published in the American Journal of Physiology: Renal Physiology used a technique called RNA sequencing to map which genes were switched on or off in S3 segments of mice. They looked at three different situations: type 1 diabetes, treatment with a drug that blocks glucose transporters in the kidney (an SGLT1/2 inhibitor), and treatment with a GLP-1 receptor agonist, which is a peptide-based compound that mimics a gut hormone. The goal was to understand, at a molecular level, what each condition does to the cells of that segment.
The findings are notable in part because of their scale. Out of nearly 19,000 detected genes, hundreds to more than a thousand showed meaningful changes depending on the condition studied. The results suggest that diabetes, SGLT inhibition, and GLP-1 receptor agonism each trigger partly overlapping but also distinctly different cellular programs in this underexamined kidney region.
Study design and methods
Researchers worked with two groups of male adult mice. One group was genetically normal. The other group carried the Akita mutation, a well-established mouse model of type 1 diabetes that causes chronically high blood glucose. Within those groups, some mice also lacked the gene for the SGLT1 transporter entirely, and some received either a vehicle (placebo) or two weeks of dapagliflozin, an SGLT2 inhibitor that also affects SGLT1 indirectly. A separate group of diabetic Akita mice received semaglutide, the generic peptide that acts as a GLP-1 receptor agonist.
To isolate the S3 segment specifically, the team used immunostaining to mark the cells of interest, then cut them out of thin kidney slices with a laser-guided microscope. This process, called laser-capture microdissection, allowed them to collect gene activity data from only those precise cells rather than the whole kidney. They then performed RNA sequencing on the collected material, producing a detailed snapshot of which genes were active and by how much.
Shared responses across diabetes and SGLT inhibition
Among the nearly 19,000 genes the researchers detected, 838 changed significantly in response to SGLT2 inhibition in normal mice, and 1,410 changed when comparing diabetic mice to normal mice. Roughly 34 percent of the genes sensitive to SGLT2 inhibition shifted in the same direction as in diabetes, pointing to a meaningful overlap in how these two conditions affect the S3 segment.
Both diabetes and SGLT2 inhibition turned up activity in pathways related to cellular proliferation. The researchers confirmed this with a separate staining technique that marks dividing cells. Both conditions also upregulated pathways involved in the cellular response to stress. On the other side of the ledger, both conditions dialed down gene pathways linked to immune and inflammatory signaling, cytokine production, and cell adhesion. The parallel pattern suggests that the S3 segment responds to increased glucose load, whether from disease or from drug-induced rerouting, with some consistent molecular signatures.
Unique responses specific to each condition
Despite the overlap, the study found that each condition also triggered its own distinct set of molecular changes. SGLT2 inhibition in normal mice uniquely increased activity in pathways for DNA demethylation and lysosomal acidification, a process by which cells degrade and recycle cellular components. It also reduced activity in a pathway related to the production of the amino acid valine. These responses did not appear in the diabetic mice to the same degree, which the researchers interpreted as evidence that the initiating mechanisms differ between the two conditions even when some downstream effects look similar.
The team also examined whether deleting the SGLT1 gene altered these patterns. They found that the extent to which SGLT1 was involved differed between the diabetic state and the drug-treated state, further supporting the idea that different upstream triggers are at work, even when the gene-activity readout appears similar on the surface.
GLP-1 receptor agonism and gene restoration
One of the more striking findings involved the GLP-1 receptor agonist. In diabetic mice, 25 percent of the genes that had shifted away from their normal pattern were pushed back toward normal by GLP-1 receptor agonism. By comparison, SGLT2 inhibition restored about 12 percent and SGLT1 knockout restored about 18 percent of the altered genes. Combining SGLT2 inhibition with SGLT1 knockout did not produce an additive effect, suggesting those two approaches work through overlapping mechanisms rather than complementary ones.
The GLP-1 receptor agonist specifically reduced markers of cellular stress and proliferation in the S3 segment of diabetic mice. Alongside these gene-level changes, the researchers observed that diabetic mice had lower levels of the SGLT1 protein at the kidney cell membrane, which the authors speculated may be a protective response by the cell to limit glucose uptake under conditions of chronic stress. GLP-1 receptor agonism was associated with a restoration of that membrane SGLT1 expression, a finding the authors described as noteworthy given the context of the broader gene-level changes.
Genes that did not respond to any treatment
Perhaps the most forward-looking aspect of the study was the identification of a subset of genes in diabetic mice that did not meaningfully respond to any of the three interventions tested. These genes represent molecular changes driven by diabetes in the S3 segment that SGLT inhibition, SGLT1 knockout, and GLP-1 receptor agonism all failed to address. The researchers proposed that this unresponsive cluster may point toward new therapeutic targets, areas where the field currently has no effective tools.
This finding underscores a broader theme in the study: even when drugs produce measurable improvements at the gene-activity level, a meaningful fraction of diabetes-driven changes in this kidney segment remain untouched. Understanding what drives that residual shift could become an important direction for future research.
Context and limitations
Several caveats are worth keeping in mind. This study was conducted entirely in male mice, and the diabetic model used represents a specific genetic form of type 1 diabetes rather than the more common type 2. Whether the same transcriptomic patterns would appear in female animals, in other diabetes models, or in humans is an open question. The researchers also studied relatively short treatment windows of two weeks, which may not capture long-term adaptive changes.
The technique itself, while powerful, captures a snapshot of gene activity rather than direct measurements of protein levels or cellular function. Changes in RNA do not always translate one-to-one into changes in what proteins are actually made or how the cell ultimately behaves. That said, the scale and specificity of the dataset, nearly 19,000 genes profiled from precisely isolated cells, gives this study an unusual level of molecular resolution for a kidney segment that has historically received less research attention than earlier parts of the proximal tubule.
