Glucosamine links hyperglycemia to mTORC1 activation and glucose toxicity in diabetes Authors: Yael Riahi, Aviram Kogot-Levin, Ziv Teselpapa, Elisheva Zemelman, Fatema Gamal, Tamar Cohen, Abed Nasereddin, Idit Shiff, Ifat Abramovich, Bella Agranovich, Dana Avrahami, Liad Hinden, Erol Cerasi, Daljeet Kaur, Lihi Grinberg, Ron Piran, Joseph Tam, Ernesto Bernal-Mizrachi, Erez Dror, Gil Leibowitz Published: 2026 ABSTRACT Hyperglycemia is a principal driver of β cell failure and multiple-organ complications in diabetes. Chronic exposure to hyperglycemia overstimulates mTORC1, disrupting glucose metabolism and promoting ER stress, oxidative stress, and inflammation; however, the upstream metabolic signal(s) linking glucose to mTORC1 activation remains unclear. Here, we identified glucosamine as a key metabolite connecting elevated glucose to mTORC1 signaling in pancreatic islets and kidney, both major targets of hyperglycemic damage. Using 13C6-glucose metabolic labeling in diabetic rodents treated with or without the SGLT2 inhibitor dapagliflozin or insulin, combined with targeted metabolomics and metabolic flux analysis, we found that tissue glucose concentrations strongly correlated with glucosamine. A similar correlation with plasma glucose was conserved in humans with or without type 2 diabetes, and inversely associated with β cell function. In vitro, low-dose glucosamine stimulated mTORC1 in islets and kidney proximal tubule cells in an O-GlcNAcylation–dependent manner. Broad phosphoproteomics and transcriptomics analyses in β cells showed that glucosamine activated mTORC1-regulating pathways, induced oxidative stress, ER stress, and dedifferentiation. Genetic inhibition of β cell mTORC1 via heterozygous Raptor knockout, as well as pharmacologic inhibition of the glucosamine/mTORC1 axis through SGLT2 inhibition, alleviated β cell stress, improved glycemic control, and restored β cell function. These findings identified the glucosamine/mTORC1 pathway as an important mediator of β cell and kidney dysfunction in diabetes. Introduction Diabetes is a major health problem resulting from the progressive decline in β cell function, leading to hyperglycemia and subsequent end-organ complications (1). Diabetic kidney disease (DKD) is a common complication of diabetes, occurring in 20%–40% of patients (2). It is the leading cause of end-stage renal disease in the Western world and markedly increases mortality risk (3). Clinical trials have consistently shown that reducing hyperglycemia, i.e., glucotoxicity, improves β cell function and prevents diabetes complications, including DKD (4–6). Hyperglycemia may promote ER stress, oxidative stress, and inflammation observed in islets as well as in other tissues through sustained activation of the key nutrient sensor mTORC1 (7, 8). Identifying the mechanisms underlying glucose regulation of mTORC1 is challenging because glucose is metabolized via glycolysis and the tricarboxylic acid (TCA) cycle, affecting ATP generation, redox state, and the levels of various glycolytic intermediates and amino acids (AAs). The ATP/AMP ratio can modulate mTORC1 activity through AMPK, whereas AAs, and likely carbohydrates, signal to mTORC1 via the Rag family of GTPases (9–11). Glucose metabolism varies across tissues and cell types, complicating the question of how glucose regulates mTORC1. We and others have previously shown that mTORC1 activity in islet cells and in kidney proximal tubule cells (KPTCs) plays a central role in mediating β cell, α cell, and kidney dysfunction in diabetes (7, 8, 12–16). Pancreatic β cells and KPTCs represent 2 complementary regulators of whole-body glucose homeostasis; while β cells control glycemia through insulin secretion, KPTCs regulate glucose clearance through reabsorption. Both cell types are highly sensitive to nutrient availability and metabolic stress, making them ideal models for investigating mTOR signaling and nutrient sensing pathways. In β cells, glucose transport is mediated by GLUT2 (GLUT1/3 in humans) and followed by phosphorylation by glucokinase, which has a low affinity for glucose and is not inhibited by its product, glucose-6-phosphate (G6P). These properties allow tight coupling between plasma glucose levels, glycolysis, and glucose oxidation for energy production, thereby regulating insulin secretion (17). In contrast, in KPTCs, glucose transport occurs through sodium-glucose cotransporters (SGLTs, primarily SGLT2) on the luminal side and various glucose transporters on the basolateral side. Glucose is phosphorylated by hexokinase, which has high affinity for glucose and is inhibited by G6P. Glucose is poorly oxidized in the TCA cycle, and most of the glucose exits the cells via GLUT2 (18, 19). Studying the mechanisms of mTORC1 activation by hyperglycemia in cells with distinct glucose utilization and oxidation patterns may help to identify novel regulators of mTORC1 that act independently of glycolysis and glucose oxidation in the TCA cycle. Therefore, we investigated how glucose regulates mTORC1 in these cell types with the goal of identifying a common mechanism underlying mTORC1 activation during chronic hyperglycemia. Here, we performed in vivo metabolic labeling using 13C6-glucose in normoglycemic and diabetic animals treated with or without glucose-lowering agents, followed by broad, unbiased targeted metabolomics across multiple organs. We analyzed correlations between tissue glucose levels and various mTORC1-regulating metabolites and tested their effects on diabetes and β cell function. Our results identified glucosamine as a key regulatory metabolite. This amino sugar, which serves as the biochemical precursor to all nitrogen-containing sugars (20), is essential for the synthesis of glycosylated proteins and lipids. We found that glucosamine is a ubiquitous sensor of tissue glucose levels and partially mediates glucose toxicity in diabetes through sustained stimulation of mTORC1 and induction of ER stress and oxidative stress. Genetic inhibition of mTORC1 in adult β cells ameliorated ER stress–induced diabetes. We propose that SGLT2 inhibitors (SGLT2is) may improve diabetes outcomes and β cell function by suppressing the glucosamine/mTORC1 pathway. Results Glucosamine synthesis and metabolism in diabetes. We treated WT mice and diabetic Akita mice with dapagliflozin or insulin for 5–7 days followed by metabolic labeling with 13C6-glucose, and performed metabolomics and metabolic flux analysis in kidney cortex extracts, liver, heart, and plasma. The full list of metabolites in the different tissues of WT and diabetic animals and their response to treatment has been published and is available online (16). Glucosamine was one of the top metabolites elevated in diabetic animals, showing 1.5- to 2-fold higher levels in kidney, liver, and heart extracts; treatment with dapagliflozin or insulin decreased the glucosamine levels (Figure 1, A–C), suggesting that this elevation is a consequence of hyperglycemia. In humans, glucosamine is present in the form of glucosamine-6-phosphate (Gn6P). The current view is that Gn6P is synthesized from fructose-6-phosphate (F6P) and glutamine by glutamine-fructose-6-phosphate transaminase (GFAT) as the first step of the hexosamine pathway (Figure 1D) (21). The end-product of this pathway is uridine diphosphate–N-acetylglucosamine (UDP-GlcNAc), which is used for generating glycosaminoglycans, proteoglycans, and glycolipids, and for posttranslational modification by glycosylation (22). Surprisingly, 13C6-glucose labeling showed that all 6 glucosamine carbons were labeled with the 13C-isotopologue, indicating that glucosamine is directly synthesized from glucose; the fraction of M+6 constituted 94.3%–100% of all labeled glucosamine (Figure 1, A–C). A summary of the full mass isotopologue distributions from all experiments is provided in Supplemental Table 1. Glucosamine levels in different tissues strongly correlated with the glucose concentration (r = 0.86–0.98) (Figure 1E). Of note, the correlation between tissue glucose and glycolytic intermediates, including F6P and dihydroxyacetone-phosphate (DHAP), branched-chain AAs and additional AAs that regulate mTORC1, was weaker than that of glucosamine and varied between tissues (Figure 1E). To explore the relevance of altered glucosamine metabolism to type 2 diabetes (T2D) in humans, we analyzed previously reported plasma metabolomics data of patients with impaired glucose tolerance (IGT) and/or impaired fasting glucose (IFG) and T2D (23). The glucosamine metabolite GlcNAc-1P was increased in the plasma of patients with IFG and T2D (Figure 1F). GlcNAc-1P correlated with fasting glucose, HbA1c, and homeostatic model assessment for insulin resistance (HOMA-IR) (Figure 1, G, H, and J), and was inversely correlated with HOMA-B, a marker of β cell function (Figure 1I). Collectively, these data suggest that glucosamine/GlcNAc is associated with both β cell dysfunction and insulin resistance. We further studied glucosamine synthesis and metabolism in primary KPTCs (Figure 1K) and human islets (Figure 1L and Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.197331DS1) cultured under chronic low glucose (LG) or high glucose (HG). Consistent with the in vivo data, exposure of both islets and KPTCs to HG markedly increased glucosamine levels and approximately 100% of the glucosamine molecules were labeled with the 13C-isotopologue at all 6 carbons (M+6), indicating endogenous synthesis from glucose. Exposure of KPTCs to HG increased 13C6-glucosamine levels by approximately 3-fold; metabolic labeling with L-glutamine-13C5,15N2 demonstrated that the amine group is derived from glutamine (Figure 1K). Similarly, in human islets, exposure to HG increased 13C6-labeled and total glucosamine by 6-fold and UDP-GlcNAc levels by 2-fold (Figure 1L and Supplemental Figure 1). Notably, chronic incubation of mouse islets under HG increased the expression of different hexosamine pathway enzymes, including Gfat1, Pgm3, and Oga, suggesting that chronic exposure to HG also modulates the activity of this pathway (Figure 1M). We next investigated glucosamine metabolism in KPTCs and human islets treated with glucosamine followed by liquid chromatography–mass spectrometry (LCMS). Glucosamine markedly increased levels of Gn6P and UDP-GlcNAc in both KPTCs and human islets, indicating that glucosamine enters the hexosamine pathway via phosphorylation to Gn6P with subsequent generation of the end-product UDP-GlcNAc (Figure 2A and Supplemental Figure 1A). Importantly the area under the curve (AUC) of glucosamine levels in human islets incubated with 0.3 mM glucosamine was like that of human islets incubated with HG (25 mM; Supplemental Figure 1A). In extracts of human islets cultured under LG conditions, the intracellular glucose concentration was 25.4 nM/mg of protein. The corresponding glucosamine concentration was 0.98 nM/mg of protein, representing approximately 4% of the intracellular glucose level (Supplemental Figure 1B). This proportion is consistent with the established fraction of glucose metabolized through the hexosamine pathway (24), indicating that the exogenous glucosamine levels used in this study effectively mimic the intracellular environment seen in HG conditions in vitro and in vivo. It has been previously suggested that glucosamine is phosphorylated to Gn6P by hexokinases (25). Consistently, we found that treatment of KPTCs with the hexokinase inhibitor 2-deoxyglucose (2-DG) effectively inhibited glucosamine phosphorylation to Gn6P, although F6P levels increased (Figure 2B). 2-DG inhibits G6P isomerase, which may explain the accumulation of F6P despite hexokinase inhibition. In any case, the marked reduction of Gn6P by 2-DG under conditions where F6P is increased supports the hypothesis that Gn6P can be generated by direct phosphorylation of glucosamine, rather than via GFAT. To rigorously test this hypothesis and distinguish between the canonical GFAT-dependent pathway and this non-canonical route, we inhibited GFAT in MIN6 cells cultured under HG conditions using both pharmacological inhibitors and genetic knockdown. Treatment with the GFAT inhibitor azaserine or 6-diazo-5-oxo-L-norleucine (DON) did not alter the levels of 13C6-glucose–derived hexosamine metabolites (GlcN, Gn6P, GlcNAc, and UDP-GlcNAc) (Figure 2C). Additionally, siRNA-mediated knockdown of Gfat1 reduced its expression by approximately 40%; to compensate for the partial knockdown, we inhibited residual GFAT activity by cotreatment with azaserine. Neither Gfat1 knockdown alone nor in combination with azaserine affected GlcNAc or UDP-GlcNAc levels (Figure 2D). Taken together, these results provide strong evidence for a non-canonical, GFAT-independent hexosamine synthesis pathway. Collectively, we show the presence of a non-canonical hexosamine pathway, in which glucosamine is generated de novo from glucose and glutamine followed by its phosphorylation to Gn6P to feed the hexosamine pathway (Figure 1D). Phosphoproteomic analysis of the glucosamine-OGT effects on signal transduction focusing on mTORC1. O-GlcNAcylation is a posttranslational modification, responsive to nutrient availability and stress, involving the attachment of O-linked GlcNAc moieties to serine and threonine residues of cytoplasmic and mitochondrial proteins (26, 27). This process is regulated by a single pair of enzymes — O-GlcNAcyl transferase (OGT) and O-GlcNAcase (OGA) — that tightly control the dynamic cycling of this modification (28). Balanced O-GlcNAcylation is required for β cell and kidney function (29–34). Consistently, both hypo-O-GlcNAcylation by Ogt knockout (KO) or hyper-GlcNAcylation by Oga KO are deleterious (29–34). Mechanistically, OGT utilizes UDP-GlcNAc as a substrate, which is generated via the hexosamine biosynthetic pathway (HBP). Exogenous glucosamine enters the HBP downstream of the rate-limiting enzyme GFAT, rapidly elevating UDP-GlcNAc levels and driving global O-GlcNAcylation. To study the downstream effect of this metabolic flux in β cells, we sought to investigate the changes in the immediate response to glucosamine and how it is affected by inhibition of OGT. Thus, we performed a phosphoproteomic analysis in cells treated with low-dose glucosamine with or without an OGT inhibitor (OGTi; ST 045849, 10 μM). Interestingly, gene set enrichment analysis (GSEA) revealed that glucosamine treatment significantly enriched the mTORC1 signaling pathway, an effect that was prevented by OGTi (Figure 3A). Focused analysis of this pathway discovered that the phosphorylation of key regulatory nodes of mTORC1 is regulated by glucosamine in an OGT-dependent manner (Figure 3B). Glucosamine increased the phosphorylation of upstream regulators of mTORC1, including the MAPK/ERK protein kinase Raf1 (35, 36) and the PKA-scaffolding protein Akap13, which mediates PKA signaling to mTORC1 (37). Additionally, phosphorylation of the phosphoinositide-producing kinase PIKFYVE (38, 39) was activated by glucosamine through O-GlcNAcylation. PIKFYVE regulates mTORC1 primarily through enzymatic production of phosphatidylinositol 3,5-biphsphate, which facilitates mTORC1 translocation to lysosomes, where mTORC1 is activated in response to nutrients. Glucosamine increased the phosphorylation of presenilin-1 (Psen1) at T370, a conserved residue within its cytosolic loop (40). Psen1 mutations account for the most common cause of familial Alzheimer disease. Of note, Psen1 knockin mutations have been shown to induce sustained activation of mTORC1 (31). Downstream effectors of mTORC1, including S6K1 and 4E-BP1, along with subunits of the EIF3 initiation factor (41), were hyperphosphorylated by glucosamine, whereas OGT inhibition abrogated it. Western blot analysis confirmed that glucosamine acutely stimulated mTORC1 activity in both MIN6 β cells and islets (Figure 3, C and D, and Supplemental Figure 2, A–C). Specifically, acute treatment with glucosamine or glucose increased the phosphorylation of S6 and 4EBP1 (without affecting ULK1 and AMPK phosphorylation; Figure 3, C and D). The OGTi inhibited mTORC1 (4EBP1 phosphorylation) in β cells treated with glucosamine (Figure 3E). Importantly, the GFAT inhibitor azaserine and the siRNA-mediated Gfat1 knockdown did not affect the mTORC1 response to glucose and glucosamine (Figure 3C and Supplemental Figure 3A). To evaluate potential changes in the transcriptome of β cells, we profiled dispersed islets at single-nucleus resolution (snRNA-seq). This approach enabled us to specifically analyze β cells following chronic glucosamine treatment, identifying transcriptional changes that may contribute to sustained activation of mTORC1. Of note, glucosamine increased the expression of mTOR, as well as genes involved in growth factor signaling, e.g., Igf1r (42), Pikfye (38), Fnip1 (folliculin-interacting peptide) (43, 44), and Lars1 (leucyl-tRNA synthetase 1, LARS1), which regulates nutrient sensing via the RAG protein complex (45) (Figure 3F). Western blot analysis confirmed that chronic treatment with glucosamine stimulated mTORC1, evident by increased S6 and 4EBP1 phosphorylation and protein levels (Figure 3G and Supplemental Figure 2C). Phosphorylation of 4E-BP1 causes its release from eIF4E to allow cap-dependent translation. We hypothesized that sustained phosphorylation of 4EBP1 requires an adaptive increase in 4EBP1 level to allow adequate regulation of protein synthesis under conditions of increased demand. We tested this hypothesis by incubating MIN6 cells with glucosamine with and without low-dose (10 nM) rapamycin for 48 hours. Treatment with rapamycin markedly decreased 4EBP1 levels, indicating that both 4EBP1 synthesis and phosphorylation are regulated by mTORC1 (Supplemental Figure 3B). Treatment with HG or glucosamine promoted mTORC1 localization to the lysosome, as evidenced by the increased colocalization of mTORC1 with LAMP2 (Supplemental Figure 3C), further suggesting mTORC1 activation. Similarly, Western blotting in HK2 cells (human KPTC line) and in primary KPTCs and immunostaining showed that both HG and glucosamine stimulated mTORC1 in an OGT-dependent manner, whereas azaserine did not affect the glucose stimulation of mTORC1 (Figure 4, A–E). The OGA inhibitor TMG increased mTORC1 activity under LG, indicating that induction of O-linked GlcNAcylation is sufficient to activate mTORC1 (Figure 4F). We found that inhibiting hexokinase with 2-DG prevented glucosamine metabolism in the hexosamine pathway (Figure 2B). Consistently, 2-DG abrogated the stimulation of mTORC1 by glucosamine (Figure 4G). Collectively, we show that glucosamine induces sustained activation of mTORC1 in an OGT-dependent manner. The glucosamine regulation of mTORC1 is probably mediated through multiple effects on protein posttranslational modifications and gene expression. The glucosamine/mTORC1 pathway regulates glucose metabolism and mitochondrial function in KPTCs. Incubation of KPTCs in medium containing a high glucosamine concentration (5 mM) effectively stimulated mTORC1 (Figure 4, C, E, and G). Under these conditions, glycolysis and glucose oxidation downstream of phosphofructokinase (PFK) were inhibited, evident by decreased fructose 1,6-biphosphate (F1,6-BP), DHAP/glyceraldehyde-3-phosphate (G3P), phosphoenolpyruvate (PEP), and lactate, and of TCA cycle metabolites (Supplemental Figure 4, A and B), as was previously shown (46). Seahorse experiments further showed that lactate production (extracellular acidification rate) decreased, along with impaired basal and ATP-linked respiration (Supplemental Figure 4, C and D). These findings indicate that glucosamine stimulates mTORC1 even under conditions where glycolysis and glucose oxidation are inhibited. In contrast, chronic exposure of KPTCs to HG or low-dose glucosamine (HG equivalent) increased glycolysis and basal and ATP-linked respiration, without affecting proton leak (Figure 5, A and B). Rapamycin prevented glucosamine stimulation of glycolysis and inhibited mitochondrial activity (Figure 5B). Metabolomics after 13C6-glucose labeling showed that exposure to HG increased the levels of labeled glucose and downstream glycolytic metabolites, including G6P, F6P, DHAP/G3P, and PEP, whereas the levels of pyruvate and lactate were unchanged, and certain TCA cycle intermediates (α-ketoglutarate and succinate) were reduced (Figure 5, C and D), suggesting dysregulation of glycolysis and glucose oxidation. Rapamycin markedly decreased G6P, F6P, pyruvate, and lactate, as well as glucose-derived acetyl CoA and malate, indicating inhibition of glycolysis and of glucose oxidation. Consistently, the levels of all glucose-derived AAs (serine, glycine, asparagine, aspartate, glutamate, and proline) were reduced (Supplemental Figure 5A). Treatment with rapamycin increased the levels of free fatty acids (FFAs) and glycerol, suggesting breakdown of triacylglycerols, along with increased availability of FFAs for energy production (Supplemental Figure 5B). Strikingly, treatment with rapamycin decreased NAD+ levels along with an increased NADH/NAD+ ratio (Figure 5E). A decreased NAD+ pool inhibits metabolic pathways, including glycolysis and the TCA cycle, resulting in decreased metabolic activity. Collectively, our findings suggest that inhibition of mTORC1 in KPTCs induces a protective metabolic reprogramming to manage the increased workload imposed by hyperglycemia. This adaptive shift is characterized by an inhibition of glycolysis and glucose oxidation, an enhanced reliance on fatty acid metabolism, and an overall reduction in mitochondrial work. Glucosamine regulation of mTORC1, insulin secretion, and mitochondrial activity in β cells. We further studied the link between the HBP, mTORC1 activation, and overall metabolic function in human islets that were incubated at LG or HG for 48 hours followed by metabolic labeling with ¹³C-glucose. Metabolomics analysis showed that the abundance of 13C-labeled glycolysis and TCA cycle metabolites increased (Supplemental Figure 6). The levels of 13C6-glucosamine and its downstream product, UDP-GlcNAc, were also increased (Figure 1L). Collectively, these findings indicate that chronic exposure to HG enhances glycolysis, glucose oxidation, and flux through the hexosamine pathway. Next, we studied the effects of glucosamine on insulin secretion. Acute exposure to glucosamine did not affect glucose-stimulated insulin secretion (Figure 6A). On the contrary, chronic islet incubation under HG with glucosamine markedly increased basal and glucose-stimulated insulin secretion (Figure 6B). Strikingly, treatment with rapamycin markedly reduced the amplification of insulin secretion by glucosamine (Figure 6B), indicating that this amplification is mTORC1 dependent. We further studied the effect of glucosamine on mitochondrial function. Seahorse experiments showed that islets incubated at HG with glucosamine tended to increase basal oxygen consumption rate and ATP-linked respiration (not significant), along with increased proton leak and decreased mitochondrial spare capacity (Figure 6C). Treatment with OGTi, but not with rapamycin, attenuated these changes. Collectively, these findings show that in islets exposed to HG, glucosamine augments mitochondrial respiration in an OGT-dependent manner, along with development of mitochondrial dysfunction, evident by increased proton leak. Rapamycin inhibited the amplification of insulin secretion by glucosamine without affecting mitochondrial respiration, suggesting that mTORC1 modulates insulin secretion downstream of mitochondrial respiration/ATP generation. Effects of prolonged exposure to glucosamine on cellular stress and dysfunction. Hyperglycemia and glucosamine enhanced insulin secretion, mitochondrial respiration, and glucose metabolism, indicating increased β cell workload, which could eventually lead to cellular stress and dysfunction. Consistently, treatment with glucosamine increased mitochondrial ROS generation in MIN6 cells (Figure 7A), along with increased expression of Txnip, a central mediator of oxidative stress and inflammation in islets (47), as well as the proinflammatory genes Ccl2 and Il6 (Figure 7B). Next, we employed analyses of the transcriptome (snRNA-seq) of β cells to clarify the mechanisms underlying the glucosamine effects. Gene expression analysis showed that treatment with glucosamine enriched the expression of genes associated with calcium and cAMP signaling, cytoskeleton organization, ER stress, and β cell function and differentiation (Figure 7C). Treatment with glucosamine increased the expression of Ppp3ca and Ppp3cb, which encode the catalytic subunits (α and β, respectively) of calcineurin, a Ca2+/calmodulin-dependent serine/threonine phosphatase that is essential for Ca2+-dependent signal transduction (48) (Figure 7D and Supplemental Figure 7A). The expression of Ca2+/calmodulin-activated kinase 1d (Camk1d) was also upregulated. Both calcineurin- and Ca2+/calmodulin-activated kinases are central regulators of insulin secretion, ER calcium homeostasis, and cellular stress adaptation, and play a role in diabetes pathophysiology (49). In contrast, the expression of Ca2+/calmodulin-dependent protein kinase II inhibitor 1 (CamK2n1) that acts as an endogenous inhibitor of CaMKII and of protein phosphatase 1 regulatory subunit 1A (Ppp1r1a) was downregulated. PPP1R1A acts as a cAMP/PKA-responsive inhibitor of protein phosphatase 1 (PP1), and its expression in β cells is required for optimal glucose-stimulated insulin secretion, biosynthesis, and for maintenance of β cell identity (50, 51). Overall, the expression of central genes regulating calcium fluxes and signaling were altered, probably leading to β cell dysfunction, ER stress, and dedifferentiation. Consistently, the expression of key genes involved in the ER stress response/unfolded protein response (UPR), including Hspa5 (BIP), Ern1 (Ire1a), and Atf6, was increased (Figure 7E and Supplemental Figure 7B); this was accompanied by decreased expression of multiple ribosome subunits, probably as an adaptive response to ER stress (Figure 7F). In addition, genes involved in mitochondria function, redox state (Supplemental Figure 7C), and cytoskeleton remodeling (Supplemental Figure 7D) were upregulated. Strikingly, key genes regulating β cell function and differentiation, including Pdx1, Nkx6.1, Mafa, Glp1r, and Ins1/2, were downregulated (Figure 7G and Supplemental Figure 7E). In summary, prolonged exposure to glucosamine stimulates mTORC1, along with dysregulation of mitochondrial function and the expression of key genes regulating calcium homeostasis, resulting in oxidative stress, ER stress, and β cell dysfunction, possibly via dedifferentiation. Effects of inhibiting β cell mTORC1 activity in diabetes. We have previously shown that inhibiting mTORC1 by heterozygous KO of Raptor in KPTCs prevents kidney fibrosis in the Akita model of DKD, in which kidney dysfunction develops due to prolonged exposure to hyperglycemia (12). Here, we further studied the role of mTORC1 in mediating β cell dysfunction in this ER stress–induced neonatal diabetes model. mTORC1 plays a role in disease pathophysiology; it is inhibited during the neonatal period, thereby preventing β cell expansion in early life, followed by reactivation during adulthood in parallel with the development of hyperglycemia (15). We generated heterozygous β cell Raptor-KO (βRaptor+/–) Akita mice and confirmed successful recombination (Figure 8A); we then assessed the effects on diabetes and β cell function (Figure 8, B–F). Fasting blood glucose levels were lower in βRaptor+/– Akita mice when compared with their βRaptor+/+ littermates (Figure 8B). Intraperitoneal glucose tolerance test (IPGTT) showed that glucose tolerance was improved (Figure 8C), along with a tendency to increased islet insulin content (P = 0.1) (Figure 8D). Inhibition of mTORC1 increased the protein level of PDX-1 (Figure 8E), which plays an important role in the regulation of insulin synthesis in adult β cells. The expression of the ER stress marker BIP was decreased (Figure 8F). Collectively, these findings suggest that inhibiting mTORC1 improved diabetes and β cell function by reducing ER stress. The SGLT2i dapagliflozin effectively inhibited the glucosamine/mTORC1 pathway, evident by decreased glucosamine levels, along with inhibition of mTORC1. As previously shown (15), treatment with dapagliflozin decreased mTORC1 activity in Akita β cells (Supplemental Figure 8A). We treated Akita mice with dapagliflozin for 2–6 weeks, followed by washout leading to complete elimination of the drug (confirmed by LCMS) (Supplemental Figure 8B); glucose tolerance was improved (Figure 9A). Plasma insulin levels were increased in Akita mice treated with dapagliflozin compared with Akita controls (Figure 9B). β Cell area was decreased in Akita mice and was not affected by treatment with dapagliflozin (Supplemental Figure 8C). In Akita mice, pancreatic insulin content was decreased by approximately 90%, whereas treatment with dapagliflozin induced 2- and 3-fold increases in insulin and proinsulin content, respectively, indicating partial restoration of insulin production (Figure 9, C–E). Treatment with dapagliflozin did not affect the proinsulin/insulin ratio (Supplemental Figure 8D), suggesting that it mainly affected (pro)insulin synthesis without affecting the efficiency of proinsulin processing. PDX-1 and NKX6.1 protein levels were slightly decreased in Akita β cells, whereas treatment with dapagliflozin restored the expression of these transcription factors, which are required for adult β cell function (Supplemental Figure 8, E and F). Treatment with dapagliflozin decreased the expression of the ER stress marker BIP, along with increased pancreatic (pro)insulin content (Figure 9, C–F), indicating that dapagliflozin enhanced insulin production by alleviating ER stress. We performed bulk RNA-seq on Akita islets treated with and without dapagliflozin for 2 weeks and compared gene expression to WT islets. GSEA showed that genes involved in metabolism (oxidative phosphorylation/aerobic respiration/ATP synthesis–coupled electron transport), ER stress, DNA damage repair, apoptosis, inflammation, and mTORC1 signaling were enriched in Akita compared with WT islets, whereas genes associated with autophagy were downregulated (Figure 9G). Treatment with dapagliflozin enriched the expression of genes regulating exocytosis and actin filament network organization, whereas genes involved in metabolism (oxidative phosphorylation/aerobic respiration) and translation, along with DNA damage response, inflammation, and senescence were downregulated (Figure 9H). The expression of several genes involved in the ER stress response, including Sec61b, which is involved in ER protein translocation, Cops5, a positive regulator of E3 ubiquitin ligases that are involved in the degradation of cyclin inhibitor kinase CDKN1B/p27Kip1, Ufc1, Ude2J2, Ube2k (ubiquitination), and Psmc6 (proteasome function) were increased in Akita islets and decreased in response to treatment with dapagliflozin (Supplemental Figure 8G). In addition, the protein disulfide isomerase Erp44 that promotes protein folding in the ER, Ptpn2, a phosphatase that regulates cell growth, proliferation, and differentiation, and Xbp1 that plays a role in the regulation of the UPR and of the immune system were upregulated in Akita islets and decreased with dapagliflozin treatment (Supplemental Figure 8G). Interestingly, the expression of several genes that were not affected by diabetes in Akita mice were increased by dapagliflozin, e.g. Edem3, which regulates the degradation of misfolded glycoproteins in the ER, and the phospholipases Pla2g6 and Bhlha15, which are involved in several processes, including cellular response to glucose starvation and UPR (Supplemental Figure 8G). The expression of proinflammatory genes, including Tnfa and Il6 and the chemokines Cxcl1 and Ccl2, were increased in Akita islets and decreased with dapagliflozin treatment (Supplemental Figure 8H). Key regulators of the cell cycle, including Cdkn1a (p21, a cyclin-dependent kinase inhibitor involved in senescence), as well as Cdkn1b and Ccna2 were increased in Akita and decreased with dapagliflozin treatment (Supplemental Figure 8I). Several genes involved in calcium signaling and secretory granule trafficking and fusion were downregulated in Akita islets and rescued by dapagliflozin treatment (Supplemental Figure 8J). In summary, treatment with an SGLT2i induces sustained improvement of diabetes and β cell function, along with modulation of the ER stress response, despite persistent proinsulin misfolding. Improved β cell function is explained by recovery of proinsulin synthesis and increased expression of genes regulating exocytosis, along with downregulation of the proinflammatory genes. The genetic pathways that were altered in diabetic Akita β cells partially mimicked those induced by glucosamine. This aligns with our finding that intracellular glucosamine levels were elevated in Akita islets and normalized with dapagliflozin treatment (Figure 1A). Consistently, the beneficial effects of dapagliflozin resembled those observed by genetic inhibition of mTORC1 in diabetic β cells. Therefore, we suggest that the glucosamine/mTORC1 pathway contributes to glucotoxicity in diabetes. SGLT2is may exert their beneficial effects in different organs through inhibition of the glucosamine/mTORC1 pathway. Discussion Our findings demonstrate that glucosamine transduces mTORC1 activation in response to chronic hyperglycemia and contributes to glucotoxic effects in diabetes. Several lines of evidence support this conclusion: (a) glucosamine levels are increased in diabetes, whereas glucose-lowering treatments (insulin and SGLT2is) reduce glucosamine across multiple tissues; (b) the glucosamine metabolite GlcNAc-1-P is increased in the plasma of individuals with IFG and T2D and inversely correlates with β cell function; (c) tissue glucosamine levels closely correlate with tissue glucose levels; (d) incubation of KPTCs and islets in HG induces de novo glucosamine generation; (e) treatment of KPTCs, islets, and β cells with glucosamine at concentrations comparable to those measured in HG-exposed tissues activates mTORC1; (f) glucosamine enters the hexosamine pathway and stimulates mTORC1 in an OGT/OGA-dependent manner; (g) glucosamine enhances mitochondrial respiration in islets and KPTCs (in β cells, glucosamine dysregulates the expression of key genes involved in calcium homeostasis, induces ER stress, and promotes β cell dedifferentiation, thereby mimicking β cell dysfunction seen in diabetes); and (h) inhibition of mTORC1 by generating a heterozygous Raptor loss-of-function mutation in β cells ameliorates diabetes and reduces β cell stress and inflammation. Previous studies proposed that upstream glycolytic intermediates and enzymes may act as sensors linking glucose metabolism to mTORC1 activity in various cell types. For example, in mouse embryonic fibroblasts, F1,6-BP binds aldolase, enabling interaction with the lysosomal v-ATPase complex, thereby activating mTORC1 and inhibiting AMPK (52–54). Others reported that PFK and PFKFB3 associate with the RagB GTPase–Ragulator complex at the lysosome (55). The Sabatini group identified DHAP as an mTORC1 activator in HEK cells (56), whereas other studies showed that G3P, also generated by aldolase, prevents GAPDH binding to Rheb, thereby inhibiting mTORC1 (57). Conversely, in INS-1 β cells, knockdown or pharmacologic inhibition of GAPDH activated, rather than suppressed, mTORC1, suggesting that the relevant mTORC1-stimulating metabolite(s) lies upstream of GAPDH (9). While glucose-mTORC1 coupling may be cell-type specific, previous efforts relied on genetic or pharmacologic perturbation of glycolytic enzymes, which produces broad effects on metabolism, energy generation, and redox state, all of which can influence mTORC1 activity. To elucidate underlying mechanisms, we performed in vivo 13C6-glucose labeling in normoglycemic and diabetic animals, with or without glucose-lowering therapies, followed by broad, unbiased metabolomics, without manipulating glycolytic enzymes. Unexpectedly, glucosamine, but not other known mTORC1-stimulating metabolites such as branched-chain AAs, methionine, S-adenosylmethionine, or glycolytic intermediates, showed tight correlation with tissue glucose. Glucosamine was elevated in kidney cortex, liver, and heart from hyperglycemic animals, and in KPTCs and islets exposed to HG. We further show that glucosamine is synthesized de novo through glucose amination, likely via amine transfer from glutamine (and possibly additional amino acids). Because glucosamine is not abundant in most foods, its generation is largely independent of dietary intake, making it a plausible intracellular sensor of glucose metabolism. In mammals, glucosamine is predominantly present as GlcN-6-P, produced from F6P by GFAT, the first committed step of the hexosamine pathway (22, 23). Here, we describe what we believe is a novel, non-canonical pathway that predominates in cells exposed to HG. Consistent with previous reports (25), we found that glucosamine phosphorylation to GlcN-6-P is mediated by hexokinases. Inhibition of hexokinase/glucokinase with 2-DG decreased GlcN-6-P levels and suppressed glucosamine-induced mTORC1 activation in KPTCs. O-GlcNAcylation is a dynamic form of glycosylation in which GlcNAc cycles on and off serine and threonine residues. Since its discovery 40 years ago (26), more than 16,000 O-GlcNAc–modified proteins have been identified, spanning transcription, translation, nutrient sensing, immunity, cell cycle, circadian regulation, and cell signaling (58). In HEK cells, HG and glucosamine stimulate mTORC1 via O-GlcNAcylation of Raptor at threonine 700, promoting activity under nutrient-replete conditions when AMPK is inactive (35). Glucose availability also regulates mTORC1 via the intracellular leucine sensor LARS1 (59); glucose starvation induces O-GlcNAcylation of LARS1, inhibiting its interaction with RagD and decreasing its affinity for leucine, thereby integrating glucose and AA signals. Our phosphoproteome and transcriptome analyses showed that glucosamine altered the phosphorylation and gene expression of multiple regulators of mTORC1, including RAF1 (MAP/ERK signaling), PIKFYVE (mTOR attachment to the lysosome), and AKAP13, which modulates mTORC1 activity in response to cAMP/PKA. In addition, glucosamine increased the expression of mTOR itself, as well as genes involved in growth factor signaling to mTORC1, e.g., Igf1r, Pikfye, and Fnip1 (folliculin-interacting peptide), and Lars1. These changes may explain the sustained activation of mTORC1 by glucosamine. Both O-GlcNAcylation and mTORC1 activity oscillate with feeding-fasting cycles (60, 61). Chronic hyperglycemia may disrupt this rhythmicity, increasing cellular workload and contributing to diabetes and its complications. Our findings support this model; prolonged HG or glucosamine exposure induced sustained mTORC1 activation in KPTCs and β cells, accompanied by increased basal and ATP-linked mitochondrial respiration. In KPTCs, mTORC1 inhibition by rapamycin triggered metabolic reprogramming, including reduced glycolysis and oxidative phosphorylation, and a substrate shift from glucose to fatty acids, evidenced by decreased glucose oxidation and glycogenic AA synthesis, and increased FFAs and glycerol. In β cells, chronic exposure to HG and glucosamine increased glycolysis, glucose oxidation, mitochondrial respiration, and insulin secretion, the latter being inhibited by rapamycin. Glucosamine-stimulated β cell respiration was associated with oxidative stress and increased expression of the key mediators of ER stress, along with a robust decrease in β cell genes that ultimately impair β cell function and differentiation. In diabetes, reducing cellular workload may alleviate oxidative stress, ER stress, and inflammation, but excessive inhibition could be detrimental, particularly in β cells, which must sustain insulin output. Given the dual (“Yin-Yang”) roles of mTORC1, moderate, fine-tuned inhibition may be optimal. Indeed, we previously showed that partial mTORC1 inhibition via heterozygous Raptor deletion in KPTCs prevented DKD in Akita mice (12). We used a similar strategy to study the effects of moderate mTORC1 reduction in ER-stressed β cells in diabetes. Akita mice develop diabetes due to a proinsulin missense mutation that causes irreversible misfolding and lifelong β cell ER stress, modeling the neonatal diabetes syndrome mutant-insulin diabetes of the young (MIDI) (62). mTORC1 is suppressed during the neonatal period due to ER stress, impairing β cell expansion and differentiation and reducing insulin secretion (15). In Akita mice, hyperglycemia after weaning is associated with mTORC1 reactivation (15). Remarkably, heterozygous β cell Raptor deletion in adult male Akita mice improved diabetes and partially preserved insulin secretion. Paradoxically, dietary D-glucosamine supplementation extends lifespan in nematodes and mice (63) and ameliorates insulin resistance and inflammation in metabolic dysfunction–associated steatotic liver disease (MASLD) models (64). Oral glucosamine inhibits hexokinases and glycolysis and activates AMPK, mimicking caloric restriction similarly to 2-DG (65). We hypothesize that exogenous glucosamine competes with glucose for transporter binding and inhibits hexokinase/glucokinase, thereby suppressing both glycolysis and conversion of glucosamine to GlcN-6-P. In contrast, our findings underscore the importance of intracellular glucosamine synthesis and its metabolism through the hexosamine pathway in mediating hyperglycemia-induced mTORC1 activation and glucotoxicity. We propose that glucosamine functions primarily as a rheostat for intracellular glucose availability, thereby regulating mTORC1 activity and mediating glucotoxicity. At the translational level, we show that the SGLT2i dapagliflozin, widely used for diabetes, kidney disease, and heart failure, reduces glucosamine levels and inhibits mTORC1. This was associated with sustained improvement in diabetes and β cell function, attenuation of ER stress responses, and decreased inflammatory gene expression. Chronic kidney injury in diabetes and other kidney diseases enhances glucose uptake in KPTCs, promoting glycolysis, increased O-GlcNAcylation, and mTORC1 activation (33, 66–68). Hyperactivation of the hexosamine pathway also contributes to pathological cardiac hypertrophy and heart failure via persistent mTORC1 activation (69). SGLT2is consistently improve DKD, non-DKDs, and heart failure (70–72). We propose that inhibition of the glucosamine/mTORC1 pathway may partially mediate these robust renal and cardiac benefits. Furthermore, SGLT2is may provide durable improvement in β cell function even under severe ER stress, thereby slowing diabetes progression. Study limitations. Our data suggest that in diabetes, glucosamine is a central mediator of mTORC1 activation in different tissues. However, we can not exclude the possibility that AAs and/or other glycolytic intermediates regulate mTORC1 in a tissue- or context-dependent manner. Hyperglycemia may induce cellular stress and tissue dysfunction through additional mechanisms independent of glucosamine/mTORC1 signaling. Moreover, OGT regulates O-GlcNAcylation of numerous proteins, which may contribute to glucotoxicity. Nonetheless, we show that glucosamine is consistently increased in multiple diabetic tissues and that glucosamine activates mTORC1 in an OGT-dependent manner. Partial mTORC1 inhibition in adult diabetic β cells and KPTCs improved diabetes, β cell function, and DKD, supporting a role for this pathway in disease pathogenesis. However, future studies are warranted to extend these findings to other models of diabetes (e.g. obesity-associated T2D) and to further dissect the precise molecular interplay between O-GlcNAcylation and specific components of the mTORC1 signaling complex. In summary, we identify endogenous glucosamine as a central mediator of mTORC1 activation by hyperglycemia. Glucosamine alters the phosphorylation and expression of upstream regulators of mTORC1, leading to its sustained activation, which contributes to β cell dysfunction in diabetes. SGLT2 inhibition reduces glucosamine levels, suppresses mTORC1, improves β cell function, and may prevent kidney disease in part through inhibition of the glucosamine/mTORC1 pathway. Methods Sex as a biological variable. Only male Akita mice and transgenic heterozygous Raptor-KO Akita mice were studied because female Akita mice develop only mild diabetes and to avoid leakiness of the CreER system due to activation of Cre by circulating estrogen in female mice. The findings are expected to be relevant to the mechanisms of glucotoxicity in both males and females. Animals. Mice were housed in the Hebrew University animal care unit with 12-hour light/dark cycles and fed standard chow diet and water ad libitum. Eight-week-old male WT C57BL/6 mice fed regular chow and diabetic male Akita (Ins2WT/C96Y) mice (The Jackson Laboratory) were treated with dapagliflozin (10 mg/kg/day in drinking water; Forxiga, AstraZeneca) for 1 week. When specified, Akita mice received a subcutaneous injection of insulin (2–4 U/day, degludec, Novo Nordisk) for 7 days. MIP-CreER mice were provided by Yuval Dor (The Hebrew University) and Raptorfl/fl (Rptortm1.1Dmsa) mice were obtained from The Jackson Laboratory (strain 013188). Mice were crossed to generate MIP-CreER+/–;Raptorfl/+ mice on the Akita background. Raptor deletion was induced by subcutaneous injection of 6-week-old transgenic mice with 2 daily doses of 8 mg tamoxifen (20 mg/mL in corn oil; Sigma-Aldrich). Littermate control mice (Akita; βRaptor+/– mice) included Akita mice bearing MIP-CreER+/–;Raptorfl/+ that were injected with oil, or Raptorfl/+ Akita mice injected with tamoxifen. When specified, Akita mice were treated with subcutaneous injection of degludec insulin (2–4 U/day; Novo Nordisk) or with dapagliflozin (10 mg/kg/day in drinking water; AstraZeneca) for 5 days followed by intraperitoneal injection of 13C6-D-glucose (1 mg/g; Sigma-Aldrich). After plasma was drawn, mice were euthanized by cervical dislocation and kidneys, liver, heart, pancreases, or islets were harvested for further analyses. Detailed information on the in vivo metabolic tests in mice, metabolomics analyses, primary islet and KPTC isolation and cell line culture, Seahorse analyses, Western blotting, immunofluorescence, mitochondrial labeling with MitoTracker Red CMXRos, flow cytometry, RNA extraction, quantitative real-time PCR, RNA-seq data processing and analysis, and phosphoproteomics is provided in the Supplemental Methods. Data availability. The sequencing data generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO). The snRNA-seq data from islets treated with glucosamine are accessible under accession number GSE325252. The bulk RNA-seq data from islets of WT and Akita mice treated with dapagliflozin are available under accession number GSE284078. The raw metabolomics and phosphoproteomic data generated during this study are available from the corresponding author upon request. Statistics. Data are expressed as the mean ± SEM, with statistical significance defined as a P value of less than 0.05. For general statistical analyses, 2-group comparisons were performed using unpaired, 2-tailed Student’s t tests unless specified otherwise, while multiple-group comparisons utilized 1-way or 2-way ANOVA with Bonferroni’s post hoc test. Pearson’s correlation coefficient (r) was determined via linear regression. All analyses were performed using Prism 8.0.2 (GraphPad Software Inc.). For RNA-seq experiments, analyses were conducted in R (version 4.5.0, https://cran.r-project.org/src/base/R-4/R-4.5.0.tar.gz); differentially expressed genes were identified using the Wilcoxon rank-sum test with Bonferroni’s correction, and pathway enrichment was assessed via a Gene Ontology (GO) analysis using a hypergeometric test with a Benjamini-Hochberg correction for the false discovery rate. Study approval. The present studies in animals were reviewed and approved by the Institutional Animal Care and Use Committee of The Hebrew University of Jerusalem. Human islets were obtained from Prodo, which holds an independent institutional review board exemption from Western Institutional Review Board for the “Use of Human Islets for Research” stating no approval from human subjects is required since all pancreases were from cadaver organ donors less than 45 years old (CFR 46.102h). Author contributions GL, ED, YR, and AKL conceived the study. GL, YR, AKL, EZ, TC, ZT, FG, IA, and BA designed experiments. YR, AKL, TC, AN, IS, IA, BA, LG, and DK performed experiments and analyzed results. EC, LH, JT, RP, DA, ED, and EBM were involved in the analysis and interpretation of the data. GL, YR, and AKL wrote the manuscript. Conflict of interest The authors have declared that no conflict of interest exists. Funding support This work is the result of NIH funding, in whole or in part, and is subject to the NIH Public Access Policy. Through acceptance of this federal funding, the NIH has been given a right to make the work publicly available in PubMed Central. National Institute of Diabetes and Digestive and Kidney Diseases grants R01-DK073716 (to EBM and GL), R01-DK132103 (to EBM), and R01-DK133183 (to EBM). National Institute of Diabetes and Digestive and Kidney Diseases grants R01-DK073716 (to EBM and GL), R01-DK132103 (to EBM), and R01-DK133183 (to EBM). US Department of Veterans Affairs Merit Award BX002728 (to EBM). US Department of Veterans Affairs Merit Award BX002728 (to EBM). European Foundation for the Study of Diabetes/Novo Nordisk A/S Programme for Diabetes Research in Europe (to GL). European Foundation for the Study of Diabetes/Novo Nordisk A/S Programme for Diabetes Research in Europe (to GL). Israel Science Foundation grant ISF-398/20 (to GL). Israel Science Foundation grant ISF-398/20 (to GL). Israel Science Foundation–Juvenile Diabetes Research Foundation grant ISF-2982/20 (to GL). Israel Science Foundation–Juvenile Diabetes Research Foundation grant ISF-2982/20 (to GL). Supplementary Material Supplemental data Unedited blot and gel images Supplemental table 1 Supporting data values