Pancreatic beta-cell loss through apoptosis represents a key factor in the pathogenesis of diabetes; however, no effective approaches to block this process and preserve endogenous beta-cell mass are currently available. To study the role of thioredoxin-interacting protein (TXNIP), a proapoptotic beta-cell factor we recently identified, we used HcB-19 (TXNIP nonsense mutation) and beta-cell-specific TXNIP knockout (bTKO) mice. Interestingly, HcB-19 mice demonstrate increased adiposity, but have lower blood glucose levels and increased pancreatic beta-cell mass (as assessed by morphometry). Moreover, HcB-19 mice are resistant to streptozotocin-induced diabetes. When intercrossed with obese, insulin-resistant, and diabetic mice, double-mutant BTBRlep(ob/ob)txnip(hcb/hcb) are even more obese, but are protected against diabetes and beta-cell apoptosis, resulting in a 3-fold increase in beta-cell mass. Beta-cell-specific TXNIP deletion also enhanced beta-cell mass (P<0.005) and protected against diabetes, and terminal deoxynucleotidyl transferase-mediated nick end labeling (TUNEL) revealed a approximately 50-fold reduction in beta-cell apoptosis in streptozotocin-treated bTKO mice. We further discovered that TXNIP deficiency induces Akt/Bcl-xL signaling and inhibits mitochondrial beta-cell death, suggesting that these mechanisms may mediate the beta-cell protective effects of TXNIP deficiency. These results suggest that lowering beta-cell TXNIP expression could serve as a novel strategy for the treatment of type 1 and type 2 diabetes by promoting endogenous beta-cell survival.

No abstract

Glucose effects on cellular functions such as gene expression require, in general, glucose metabolism at least to glucose-6-phosphate (G-6-P). However, the example of thioredoxin-interacting protein (TXNIP), a glucose-regulated gene involved in the cellular redox state and pancreatic beta cell apoptosis, demonstrates that this rule may not always apply. We found that aside form glucose, the nonmetabolizable sugars 2-deoxyglucose, which is still converted to G-6-P as well as 3-O-methylglucose (3-MG), which cannot be phosphorylated by glucokinase, stimulate TXNIP expression. In contrast, incubation of INS-1 beta cells with equimolar amounts (25 mM) of l-glucose or mannitol had no effect on TXNIP expression as measured by real-time RT-PCR, eliminating the possibility of an osmotic effect. Also, glucose uptake into the cell is critical because phloretin, an inhibitor of glucose transporter 2, blunted the glucose effects. Moreover, the 3-MG effect was not restricted to a cell line and was observed in 293 cells and primary human islets. Incubation of INS-1 cells with 30mM mannoheptulose, an inhibitor of glucose metabolism, blunted all glucose-induced gene expression but left the 3-MG effects unaltered. Using transient transfection studies and deletion constructs of the human TXNIP promoter, we found that the effects of glucose and 3-MG were dependent on the same region of the TXNIP promoter containing an E-box repeat carbohydrate response element (ChoRE). Thus, these findings provide the first evidence for regulation of gene expression by 3-MG, which is independent of glucose metabolism and suggest that glucose and 3-MG regulate transcription by two distinct pathways converging at a common ChoRE.

No abstract

No abstract