Home / Resources / Clinical Gems / International Textbook of Diabetes Mellitus, 4th Ed., Excerpt #148: Glucose Toxicity Part 4

International Textbook of Diabetes Mellitus, 4th Ed., Excerpt #148: Glucose Toxicity Part 4

Oct 23, 2018

The hexosamine/O-linked N-acetyl glucosamine pathway

The hexosamine biosynthetic pathway and insulin resistance


One metabolic fate of glucose is the hexosamine biosynthetic pathway (HBP), and high flux through this pathway, such as occurs in diabetes or overnutrition, has been convincingly linked to insulin resistance and β-cell failure. In this pathway a relatively small fraction of cellular glucose flux—a few percent in most tissues—is converted to UDP-N-acetylglucosamine (UDP-GlcNAc) and other amino sugars (Figure 27.5). The rate limiting step is catalyzed by the enzyme glutamine: fructose-6-phosphate amidotransferase (GFA) that catalyzes the synthesis of glucosamine-6-P from fructose-6-P and glutamine. The product of the HBP, UDP-GlcNAc is well placed to serve a nutrient-sensing function, as its production has been shown to be responsive not only to cellular glucose flux [27] but also to the levels and availability of nucleotides [41], fatty acids [41], and amino acids [27]. It is a high-energy intermediate and is therefore also reflective of cellular energy status.

The first evidence of the involvement of the HBP in metabolic regulation was its mediation of glucose toxicity in models of insulin resistance. High concentrations of glucose downregulate the HBP serves a nutrient-sensing function in multiple tissues and affects metabolism in a wide-ranging manner. Specifically:

  • overexpression in muscle plus fat, or fat alone results in insulin resistance, downregulation of glucose transport [44–46], and;
  • overexpression in β cells results in hyperinsulinemia [47].

These phenotypes mimicked and therefore validated those that had been seen in the infusion models and with glucosamine treatment of explanted tissues [48–50]. Importantly, the effects of increased HBP flux are nonadditive with hyperglycemia [51] and high-fat feeding [52], in inducing insulin resistance suggesting that both hyperglycemia and overnutrition in general may induce insulin resistance through the HBP.

O-linked glycosylation of nuclear and cytosolic proteins as the mechanism for HBP effects on glucose homeostasis

What is the mediator of the effects of the HBP? UDP-GlcNAc is a substrate for enzymatic glycosylation of proteins, suggesting a mechanism for HBP signaling. Since neither N-linked glycosylation nor O-linked glycosylation of secreted proteoglycans are responsive to changes in glucose fluxes, interest in HBP signaling originally focused on the then recently discovered pathway of O-linked glycosylation wherein single O-GlcNAc residues are enzymatically added to serines and threonines of nuclear and cytosolic proteins (Figure 27.5). Discovered by Hart in 1984, this form of glycosylation is dynamic, is commonly found on proteins involved in signal transduction, and is responsive to changes in glucose flux [53,54]. O-glycosylation is often reciprocal with phosphorylation and in some cases occurs on the same residues that are otherwise phosphorylated (e.g. c-myc [55]), further suggesting a potential role in signaling. Glycosylation is accomplished by a single O-linked GlcNAc transferase (OGT) [56] and deglycosylation is by a single known O-GlcNAcase [57]. The enzymologic and biochemical aspects of the pathway involved in protein O-GlcNAc modification have been reviewed recently and will not be described here in further detail [58]. Importantly and consistent with playing a role in glucose homeostasis, the pathway is highly feedback-regulated by substrates and transcriptional control of OGT and O-GlcNAcase such that in cultured cell lines the minimum levels of glycosylation of at least some proteins insulin-stimulated glucose uptake, and Marshall made a breakthrough in understanding this phenomenon when he demonstrated in adipocytes that glucose metabolism to glucosamine-6-P was required for this effect [27]. Since then, numerous studies have shown that increased hexosamine flux can induce insulin resistance in cultured cells and whole animals (reviewed in [42]). Many of these studies employed infusion of glucosamine that enters the HBP directly after uptake by glucose transporters and phosphorylation by hexokinase. Interpretations of these studies were criticized because pharmacologic concentrations of glucosamine can compete for glucose uptake, deplete cellular ATP, and even trigger apoptosis [43]. Furthermore, short-term infusion studies may not reflect the full range of posttranslational and transcriptional changes that can result from chronic activation of the HBP.More definitive evidence for the involvement of the HBP in insulin resistance was obtained using animal models wherein chronic but physiologic changes in hexosamine flux were induced by overexpression of GFA in a tissue-specific manner. The results of these studies showed that happens to occur at the level of normoglycemia in mammals, 5mM glucose [59]. Consistent with that finding, the pathway regulates insulin signaling even in the normoglycemic range, demonstrating that it is a physiologic and not solely pathophysiologic (“toxic”) regulator [60]. Finally, the pathway’s highly evolutionarily conserved role in nutrient homeostasis and hormone responsiveness in C. elegans [47,61], Drosophila [62], and plants [63] speaks to its importance in metabolic regulation.

The first direct evidence that the O-glycosylation mediated the observed effects of the HBP on insulin resistance was demonstrated by overexpression of the enzyme responsible for this glycosylation, O-linked GlcNAc transferase (OGT), in skeletal muscle and fat under control of the GLUT4 promoter [64]. This maneuver recapitulated the insulin resistance and hyperleptinemia seen with overexpression of GFA in the same tissues [44]. Other early evidence for the role of OGT came from demonstrations in Hart’s laboratory that enzymatic inhibition of the enzyme O-GlcNAcase in 3T3L1 cells leads to insulin resistance in parallel with increased O-linked glycosylation of proteins [65]. It should also be pointed out, however, that there are models of insulin resistance that do not involve increased levels of O-GlcNAc [66], and conversely, there are circumstances in which increased O-GlcNAc may not lead to insulin resistance [67]. In both of the latter cases, extremely high and unphysiologic levels of protein O-GlcNAc modification were achieved pharmacologically.

Modification of metabolic enzymes and transcription factors by O-GlcNAc has been shown to alter their function in a way that is consistent with the phenotypes observed in animals. Glycogen synthase (GS), the first protein shown to be affected by phosphorylation, is also a target of the HBP/O-GlcNAc pathway [68]. Exposure of cultured adipocytes to high glucose or glucosamine, or streptozotocin-induced diabetes in mice renders the enzyme insulin resistant and less sensitive to its principal positive allosteric regulator, glucose-6-phosphate [69], and this is related not to changes in phosphorylation but rather to levels of O-glycosylation. O-glycosylation of GS therefore mimics phosphorylation and inhibits the enzyme, but in a way that cannot be reversed by insulin action.Thus, O-GlcNAc acts as a dominant negative signal that renders GS insulin resistant and limits glycogen accumulation in situations of chronic hyperinsulinemia and nutrient excess.

Many other components of the insulin signal transduction pathway are modified by O-GlcNAc, most often in an inhibitory fashion. Rats infused with glucosamine, for example, exhibit insulin resistance of skeletal muscle through decreased phosphorylation of insulin receptor substrate-1 (IRS-1) and its association with phosphatidylinositol 3-kinase (PI3K) [56]. IRS-1 [65], the downstream insulin signaling kinase Akt [70], proteins of glucose transporter vesicles [71], the crucial transcriptional metabolic regulator FOXO1 [72], and the nutrient sensor AMP-activated protein kinase (AMPK) [73] are also modified by O-GlcNAc, with proven functional consequences. A recent and particularly exciting insight into the crosstalk between hormone signaling and O-GlcNAc is that OGT contains a domain that recognizes phosphatidylinositol 3,4,5-trisphosphate [74]. This results in the recruitment of OGT to the plasma membrane after activation of PI3K through the insulin/IR/IRS-1 pathway. Thus, even as the insulin signal transduction pathway is being activated, steps are also being taken by the cell to attenuate that acute signal through increased O-glycosylation of many of the same proteins.

These effects of O-GlcNAc on insulin signaling, particularly in rodent models wherein the levels can be effectively and chronically modulated in the physiologic range, are very consistent with the insulin resistance associated with “glucose toxicity” seen in humans. The direct evidence that the HBP may be directly involved in human diabetes, however, is limited and as yet inconclusive, largely because of experimental limitations. In humans, glucose and insulin infusions increase levels of O-GlcNAc in skeletal muscle [75]. Human studies, however, have revealed modest [76] or no [77] effects of glucosamine infusion on insulin resistance. These studies are insufficient to refute the hexosamine hypothesis, however. The glucosamine infusions may not have reached threshold concentrations for sufficient lengths of time or may not have targeted key tissues such as liver or adipose tissue. For example, in the negative study cited earlier, glucosamine infusion into the forearm did not lead to insulin resistance [78] but the study was limited in its sensitivity to detect a modest effect, and moreover the principal target for producing insulin resistance in muscle may be the adipocyte rather than muscle itself [45]. Other evidence in humans does support a role for the HBP. We showed, for example, that insulin resistance in type 2 diabetes was correlated with GFA levels in skeletal muscle [30,54]. Leptin mRNA and protein levels are also correlated with adipocyte UDP-GlcNAc levels that are in turn well correlated with body mass index in humans [79].