Glucosamine-induced activation of glycogen biosynthesis in isolated adipocytes. Evidence for a rapid allosteric control mechanism within the hexosamine biosynthesis pathway.

Enhanced flux through the hexosamine biosynthesis pathway (HBP) induces insulin resistance and facilitates lipid storage through the up-regulation of enzyme mRNA levels. Both actions occur over several hours and require gene expression. We now identify a regulatory arm of the HBP that involves rapid allosteric activation of glycogen synthase (GS) and stimulation of glycogen biosynthesis (GBS). When insulin-pretreated adipocytes were exposed to 2 mM GlcN, incorporation of [14C]glucose into glycogen doubled by 10 min (t(1/2) of <5 min), whereas UDP-glucose levels were concomitantly decreased during this time (t(1/2) of 1.4 min; >90% depletion). Stimulation of GBS and depletion of UDP-glucose both correlated with an early and rapid rise in the levels of glucosamine-6-phosphate (GlcN-6-P), a known activator of GS. The lowering of GlcN-6-P levels by removing extracellular GlcN (>80% reduction by 45 min) was accompanied by the restoration of UDP-glucose levels. Prolonged GlcN treatment (20 min to 2 h) inhibited GBS, which corresponded to a massive intracellular accumulation of GlcN-6-P (t(1/2) of approximately 32 min; >1,400 nmol/g). From these data, we conclude the following. 1) GlcN treatment elevated intracellular GlcN-6-P levels within minutes, resulting in allosteric activation of GS, stimulation of GBS, and a reduction in steady-state levels of UDP-glucose due to increased precursor utilization. 2) Prolonged treatment with high concentrations of GlcN caused massive accumulation of GlcN-6-P that adversely affected cellular metabolism and reduced GBS. 3) The biphasic actions of GlcN on GBS may explain many of the discrepant reports on the role of the HBP in glycogen metabolism.

In 1991 we discovered a unique metabolic pathway in isolated adipocytes that mediates glucose-induced desensitization of the insulin-responsive glucose transport system (GTS) 1 (1).
Because a major function of this pathway is the generation of hexosamine metabolites, we postulated that under hyperglycemic conditions the enhanced flux of incoming glucose through the hexosamine biosynthesis pathway (HBP) culminates in the induction of insulin resistance (1,2). Desensitization is likely mediated by altered gene expression, because desensitization occurs relatively slowly over several hours and because transcriptional inhibitors completely block glucose-induced desensitization (3). Numerous studies, both in vitro and in vivo, have now confirmed the role of the HBP in glucose-induced desensitization and extended the scope of this regulatory system to muscle tissue and other cell types (4 -9).
In the current study, we have used GlcN and isolated adipocytes to examine the hypothesis that GlcN-6-P can allosterically enhance glycogen biosynthesis (GBS) through in vivo stimulation of glycogen synthase (GS). This idea originated from earlier studies showing that GlcN-6-P can allosterically activate GS in vitro (10,11). The choice of adipocytes as our cellular model system derives from the fact that adipocytes represent a classical insulin-responsive target tissue. Moreover, adipocytes have proven useful in elucidating the mechanism(s) underlying glucose-induced insulin resistance (1,4,5,12) and in identifying various genes regulated through the HBP (13). Our experimental approach entailed treating adipocytes with GlcN and then measuring the subsequent changes in glycogen metabolism (GBS and UDP-glucose levels) and the intracellular levels of two key hexosamine products (GlcN-6-P and UDP-GlcNAc). The rationale for this pharmacological approach is based on the finding that GlcN is rapidly transported into adipocytes through the insulin-responsive GTS, where it undergoes phosphorylation to GlcN-6-P, the first product of the HBP (1,14). Thus, GlcN is relatively specific, because it directly enters the HBP and bypasses the first and rate-limiting enzyme of this pathway, which is glutamine: fructose-6-phospahte amidotransferase. GlcN action was evaluated as a function of the treatment time (5 min to 4 h) and the GlcN dose (20 M to 10 mM) by using insulin-pretreated cells, insulin-and glucose-pretreated cells, and control cells (no insulin or glucose pretreatment).
The obtained results support the hypothesis that relatively small increases in GlcN-6-P can rapidly stimulate GBS (within minutes) through allosteric activation of GS. Moreover, enhanced formation of glycogen appears to result in a decrease in the steady-state levels of UDP-glucose through precursor utilization. In general, our findings are conceptually important because they add a new regulatory dimension to the HBP that encompasses very rapid control of glucose metabolism through an allosteric mechanism. Thus, we postulate that enhanced flux within the HBP orchestrates both short term (allosteric) and long term (transcriptional) regulatory actions in response * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ¶ Present address: School of Medicine, University of Tsukuba, Tsukuba, Ibaraki 305, Japan. 1 The abbreviations used are: GTS, glucose transport system; G-6-P, glucose-6-phosphate; GBS, glycogen biosynthesis; GlcN-6-P, glucosamine-6-phosphate; GS, glycogen synthase; HBP, hexosamine biosynthesis pathway.
to nutritional information derived from the internal and external environments.

EXPERIMENTAL PROCEDURES
Materials-Sources of materials were as follows. Porcine insulin was from Sigma, collagenase was provided by Worthington Biochemicals (Freehold, NJ), bovine serum albumin came from the Armour Company (Kankakee, IL), and penicillin-streptomycin, Dulbecco's modified Eagle's medium (DMEM), and DMEM formulated without D-glucose were purchased from Invitrogen. All other reagents were from Sigma or Fisher unless otherwise specified.
Preparation and Treatment of Isolated Adipocytes-Isolated adipocytes were obtained from the epididymal fat pads of male Sprague-Dawley rats (180 -225 g) by collagenase digestion as described previously (15). After digestion, cells were washed three times in Hepesbuffered balanced saline solution consisting of 25 mM Hepes, 120 mM NaCl, 0.8 mM MgSO 4 , 2 mM CaCl 2 , 5.4 mM KCl, 1 mM NaH 2 PO 4 , 1 mM sodium pyruvate, 100 units/ml penicillin, 100 g/ml streptomycin, and 1% bovine serum albumin, pH 7.6. Cells were then diluted to a final concentration of 5 ϫ 105 cells/ml (12 ml of Hepes-buffered saline solution per gram of initial fat weight). From a common pool of cells, 400 l of adipocytes were aliquoted into 17 ϫ 100-mm sterile polystyrene tubes containing 1.6 ml of Hepes-buffered saline solution. Unless otherwise indicated, insulin was added (25 ng/ml, final concentration), and the cells were incubated at 37°C to stimulate the GTS and facilitate the subsequent uptake of glucose and GlcN. After 30 min GlcN was added, and the adipocytes were further incubated at 37°C for either 4 h (dose-response experiments) or for various time periods (time course experiments).
Measurement of UDP-N-acetylglucosamine and UDP-glucose Levels-After treatment the adipocytes were transferred to 1.5-ml microfuge tubes, and the cells were washed three times with ice-cold, bovine serum albumin-free Hepes-buffered saline solution. During the final wash the cell volume was reduced to 150 l, and 0.4 ml of cold perchloric acid (0.3 mol/liter) was added to adipocytes for 10 min. The mixture was then centrifuged at 20,800 ϫ g (10 min at 4°C), and the deproteinated metabolite extract (420 l) was transferred to a new microfuge tube. After the extract was neutralized by adding a small amount of K 2 CO 3 (5 M), extracts were frozen and stored at Ϫ20°C. The concentration of UDP-GlcNAc was measured by high performance liquid chromatography as described previously (16) by first passing the cell extract through a strong anion-exchange Supelclean LC-SAX column (Supelco, Bellefornte, PA). Analysis was performed using a high performance liquid chromatography system (Waters Associates, Milford, MA) and two reverse phase LC18T columns in series (Supelco). Metabolite concentration was quantitated by UV light absorption (260 nm), and levels were determined from a standard calibration curve. Recovery of standards added to adipocyte extract was Ͼ95%.
Measurement of GlcN-6-P Levels-We have developed a fluorometric assay to measure GlcN-6-P levels based on the same enzymatic principles used to measure G-6-P levels. At high concentrations we found that G-6-P dehydrogenase can act on GlcN-6-P and NADP, resulting in equimolar production of NADPH, whereas at low concentrations (used to measure G-6-P) NADPH was not generated. Thus, the low affinity of G-6-P dehydrogenase for GlcN-6-P allows for the resolution of G-6-P from GlcN-6-P in the same sample by first carrying out the assay for G-6-P with a low amount of G-6-P dehydrogenase (1.2 units/ml) followed by the addition of a much greater concentration of G-6-P dehydrogenase (43 units/ml). Optimization of assay methods revealed that GlcN-6-P measurements are not affected by glucose, GlcN, fructose-6phosphate, or N-acetylglucosamine-6-phosphate and that standard calibration curves (using GlcN-6-P) are linear (data not shown). To further validate the assay results, we incubated adipocytes for 4 h with various concentrations of GlcN, prepared metabolite extracts as described above, and then measured GlcN-6-P levels using two methods. The first method was the fluorometric assay. The second assay entailed derivatization of GlcN-6-P with o-pthaldialdehyde so that it could be detected fluorometrically (following separation by reverse phase high performance liquid chromatography). Both assay methods yielded similar results, but the fluorometric enzymatic assay was faster, less expensive, and allowed for the measurement of numerous samples (using a 96-well assay format).
Assessment of the Glycogen Biosynthesis Rate-GBS was measured using a modified method derived from Smith and Lawrence (17). Briefly, isolated adipocytes were added to 17 ϫ 100 polypropylene tubes and preincubated for 30 min at 37°C with 25 ng/ml insulin (total volume of 200 l). GlcN was then added for various time periods. After treatment, the rate of GBS was determined by adding 10 l of [U-14 C]glucose for 5 min at 37°C. At the end of 5 min, the reaction was terminated by transferring 180 l of adipocytes to a 400-l microfuge tube and centrifuging the cells through silicone oil to separate cells from extracellular buffer and free the [ 14 C]glucose. Adipocytes were then lysed, and glycogen was extracted by boiling cells for 10 min in 250 l of 30% KOH. The mixture was centrifuged for 2 min at 8,200 ϫ g, after which time 200 l of the solubilized cell extract was transferred to a 1.5-ml microfuge. To precipitate glycogen, 95% ethanol was added (600 l). The samples were cooled for 10 min at Ϫ20°C and centrifuged at 4°C for 8 min at 8,200 ϫ g. The glycogen pellet was then washed three times with 70% ethanol, resuspended, and transferred to a liquid scintillation vial. The amount of [ 14 C]glucose incorporated into glycogen was determined by scintillation counting.

Biphasic Actions of GlcN on Glycogen Biosynthesis-As
shown in the time course study depicted in Fig. 1A, the addition of 2 mM GlcN to insulin-pretreated adipocytes rapidly enhanced the rate at which [ 14 C]glucose was incorporated into glycogen. By 10 min, maximal stimulatory effects were seen (an ϳ2-fold increase; t1 ⁄2 of Ͻ5 min). The rapidity of GlcN action is further illustrated by the fact that stimulatory effects were observed when GlcN was added immediately before the short 5-min assay (at time 0). As can also be seen in Fig. 1A, GlcN treatment longer than 20 min resulted in a progressive decrease in GBS, culminating in a GBS rate that was 50% below control levels at 2 h. In Fig. 1B, insulin-pretreated adipocytes FIG. 1. Biphasic actions of glucosamine on glycogen biosynthesis. A, isolated adipocytes were pretreated with 25 ng/ml insulin for 30 min to stimulate the insulin-response GTS and facilitate GlcN uptake. GlcN (2 mM) was then added to cells, and, after various times, the rate of GBS was assessed by measuring [ 14 C]glucose incorporation into glycogen during a 5-min assay. Each point represents the mean of triplicate determinations. B, insulin-pretreated adipocytes were incubated with the indicated concentrations of GlcN for 30 min, after which time the rate of GBS was assessed by measuring the incorporation of [ 14 C]glucose into glycogen. Each point represents the mean of duplicate determinations.
were exposed to various concentrations of GlcN for 30 min. As depicted, a similar biphasic curve was obtained in which GlcN concentrations up to 1 mM stimulated GBS (ED 50 of ϳ300 mM), whereas greater GlcN concentrations inhibited GBS. Fig. 2, we used insulin-pretreated cells to assess GlcN action on intracellular levels of GlcN-6-P, UDP-GlcNAc, and UDP-glucose as a function of GlcN treatment time (5 min to 4 h) or GlcN dose (20 M to 10 mM GlcN). In Fig. 2A, it can be seen that 2 mM GlcN rapidly increased GlcN-6-P levels at 5 min from undetectable levels to ϳ300 nmol/g. Because GlcN-6-P has been shown to be an allosteric activator of GS (10, 11), we believe that elevated levels of this key hexosamine metabolite explain the rapid stimulation of GBS (Fig. 1). The data presented in Fig. 2A also show that prolonged treatment with 2 mM GlcN results in a massive accumulation of GlcN-6-P (Ͼ1,400 nmol/g) that reaches a plateau at 2-3 h (t1 ⁄2 of ϳ32 min). In Fig. 2B, it can be seen that 2 mM GlcN had little or no effect on UDP-GlcNAc levels during the first 5 min of treatment but did double UDP-GlcNAc levels during the 1st h (t1 ⁄2 of ϳ30 min). In Fig. 2C, it can be seen that GlcN treatment rapidly (t1 ⁄2 of ϳ1.4 min) and profoundly (Ͼ95% decrease) reduced intracellular levels of UDP-glucose. Because the rapid decrease in UDP-glucose levels correlates with increased rates of GBS ( Fig. 1), this result indicates that the increased demands of enhanced GBS quickly reduce the precursor pool of substrate (UDP-glucose).

GlcN Treatment Rapidly Lowers UPD-glucose Levels-In
From the dose-response curve depicted in Fig. 2D, it can be seen that GlcN-6-P levels were minimally affected when cells were incubated for 4 h with GlcN concentrations Ͻ250 M. At 500 M GlcN the levels were only slightly elevated, whereas GlcN-6-P levels were maximally elevated at 4 mM GlcN (ED 50 of 620 M). Measuring UDP-GlcNAc levels as a function of GlcN concentration (Fig. 2E) revealed a very different pattern. Low doses of GlcN progressively increased UDP-GlcNAc levels (4fold increase, ED 50 of 60 M), whereas higher GlcN doses (750 M to 4 mM) diminished the maximal levels of UDP-GlcNAc. When UDP-glucose levels were measured under these conditions (Fig. 2F) we found that GlcN concentrations Ͻ100 M had little or no effect on UDP-glucose levels, but as the concentration was increased from 100 M to 1 mM a progressive decrease was observed (ED 50 of 300 M). Several conclusions can be drawn by comparing the dose-response curves in panels D, E, and F of Fig. 2. First, by comparing panels D and E in Fig. 2, it is apparent that low concentrations of GlcN can markedly elevate UDP-GlcNAc levels without measurably affecting GlcN-6-P levels. From this finding, we conclude that low concentrations of GlcN are rapidly taken up by adipocytes, phosphorylated to GlcN-6-P, and then routed through the HBP resulting in a 4 -5-fold increase in UDP-GlcNAc levels. At higher levels of extracellular GlcN (Ͼ250 M), we believe that the uptake and the intracellular phosphorylation of GlcN exceed the capacity of the HBP, which leads to a progressive accumulation of GlcN-6-P. The second conclusion is derived by comparing panels D and F in Fig. 2, which show that loss of UDP-glucose is closely correlated with the rise in GlcN-6-P levels. Thus, we believe that depletion of UDP-glucose is mediated by the accumulation of GlcN-6-P and the resulting allosteric activation of GS.
Removal of Extracellular GlcN Decreases Intracellular GlcN-6-P Levels and Increases UDP-glucose Levels-To gain additional insights into the kinetics of hexosamine flux, we pretreated adipocytes for 30 min with insulin and then added 2 mM GlcN. After 1 h, adipocytes were washed three times to remove extracellular GlcN. As shown in Fig. 3A, we found that intracellular GlcN-6-P levels were reduced from 1500 nmol/g to Ͻ200 nmol/g after 45 min (t1 ⁄2 of 31 min). In contrast, UDP-GlcNAc levels actually increased during this time (Fig. 3B). From this observation we conclude that the high intracellular levels of GlcN-6-P were reduced by the continued flux of GlcN-6-P through the HBP. In Fig. 3C we measured UDP-glucose levels as a function of time following GlcN removal. As can be seen, the increase in UDP-glucose levels correlates well with the time-dependent decrease in GlcN-6-P levels (but not with UDP-GlcNAc levels). This finding adds further support for the supposition that GlcN enhances GBS through allosteric regulation of GS rather than O-linked glycosylation.

GlcN Rapidly Lowers UPD-glucose Levels Under Euglycemic Conditions (Presence of 5 m M Glucose)-Several in vivo studies
have shown that prolonged GlcN infusion in rats can increase skeletal muscle hexosamine levels and induce whole body insulin resistance (10, 19 -23). To reproduce euglycemic-hyperinsulinemic clamp conditions under defined in vitro conditions, we modified the protocol used in Fig. 2 such that all treatment groups contained 5 mM glucose (in addition to GlcN). As can be seen in the time course experiment depicted in Fig. 4A, GlcN-6-P levels were again markedly increased by 5 min (undetectable to about 160 nmol/g) and reached maximal levels by 90 min (t1 ⁄2 of 12 min). It should be noted that compared with cells treated with GlcN alone (Fig. 2A), total intracellular accumulation was reduced by 50%. This can be explained by the fact that glucose competitively inhibits GlcN uptake through the insulin-responsive GTS. In other words, when a fixed concen- tration of GlcN is used (2 mM), less GlcN enters cells when glucose is present. As can be seen in Fig. 4C, although the rate and the extent of GlcN uptake was diminished, UDP-glucose levels were still rapidly and extensively reduced (Ͼ60% decrease by 5 min; t1 ⁄2 of 3 min). It is also apparent that the initial level of UDP-glucose in control cells treated with 5 mM glucose was higher than that in cells incubated under glucose-free conditions (ϳ7 nmol/g versus ϳ2.5 nmol/g). This observation indicates that extracellular glucose was effectively internalized and that a portion of the incoming glucose was routed to the GBS pathway where it increased the precursor product used for glycogen formation (UDP-glucose). Thus, despite a continuing supply of substrate for GBS, steady-state levels of UDP-glucose were still reduced by Ͼ80%. Because both glucose and GlcN enter the HBP in adipocytes and raise the intracellular levels of UDP-GlcNAc, it is not unexpected that UDP-GlcNAc levels were slightly elevated in the presence of 5 mM glucose and continued to increase as a function of GlcN treatment time (Fig. 4B).
As depicted in Fig. 4, D-F, GlcN dose-response curves for all three metabolites were shifted to the right in the presence of 5 mM glucose (compared with cells similarly treated in the absence of glucose as shown in Fig. 2). This rightward shift is most likely due to the reduced uptake of GlcN in the presence of glucose, because glucose has a higher affinity for cell surface glucose transporters and can competitively inhibit GlcN uptake. Although glucose was able to reduce the steady-state levels of GlcN-6-P by half (Fig. 4, A and D), no such reduction was observed when measuring UDP-GlcNAc levels (Fig. 4, B and E). In fact, the combination of glucose and GlcN increased UDP-GlcNAc levels by ϳ30% above values obtained with maximally effective concentrations of GlcN alone (compare Fig. 4E with Fig. 2E). When the dose-response curves in Fig. 4, D and E are compared, it can be seen that that increases in UDP-GlcNAc levels occur before any measurable changes in GlcN-6-P levels. Specifically, 750 M GlcN failed to elevate GlcN-6-P, whereas this dose almost maximally increased UDP-GlcNAc levels. This finding again illustrates that above a certain threshold concentration of GlcN, the ability of the HBP to process an incoming substrate becomes increasingly rate-limited, resulting in the progressive accumulation of GlcN-6-P. These studies, performed under euglycemic conditions, indicate that our in vitro results on metabolite levels are not due to glucose deprivation.
GlcN Lowers UDP-glucose Levels in the Absence of Insulin-In the aforementioned experiments we used insulin to stimulate the insulin-responsive GTS and facilitate the uptake of GlcN. However, insulin itself initiates a rapid cascade of phosphorylation/dephosphorylation reactions that culminate in various pleiotropic actions, including the stimulation of GS. Therefore, we performed time course and dose-response experiments to assess the extent to which GlcN alone elevates GlcN-6-P levels and to determine whether GlcN can decrease UDPglucose levels in the absence of insulin. As depicted in Fig. 5A, we found that 2 mM GlcN only modestly increased GlcN-6-P levels (to ϳ120 nmol/g versus Ͼ1400 nmol/g) and that the time required to attain steady-state levels was much shorter (t1 ⁄2 of 8 min versus 32 min for insulin-treated cells). These differences are likely explained by a markedly reduced rate of GlcN uptake in the absence of insulin (10-fold reduction in uptake) and by the fact that the rate of incoming GlcN no longer exceeds the capacity of the HBP. As depicted in Fig. 5C, the decrease in UDP-glucose proceeded at a much slower rate (t1 ⁄2 of 8 min), which we believe results from the lower intracellular levels of GlcN-6-P. From dose-response studies (Fig. 5, D-F), we obtained results very similar to those in Fig. 2, D-F, except that the dose-response curves were right-shifted by ϳ5-10-fold. This makes sense, because insulin stimulates GlcN uptake ϳ10-fold in isolated adipocytes. It is interesting to note that insulin treatment alone can lower UDP-glucose levels but that it affects neither GlcN-6-P nor UDP-GlcNAc levels. From these insulin-free experiments, we conclude that GlcN is the major variable responsible for the observed metabolite changes and that insulin serves simply to enhance the rate of uptake by its ability to stimulate the GTS. It is also readily apparent from the time course studies that relatively small increases in GlcN-6-P (that do not exceed the biosynthetic capacity of the HBP) can have a profound effect on lowering UDP-glucose levels.

DISCUSSION
Hyperglycemia is the hallmark of type 2 diabetes mellitus and contributes to disease pathogenesis by impairing both insulin action and insulin secretion (24 -27). Thus, hyperglycemia is not only a consequence of diabetes, it is a pathophysiological factor that can perpetuate and sustain the diabetic state. This detrimental effect of hyperglycemia is generally referred to as "glucose toxicity" (26 -28), and for many years the question of how hyperglycemia mediates desensitization at the cellular level remained unanswered. In 1991 we provided an explanation by showing that an excessive flux of glucose through the HBP culminates in the induction of insulin resistance in isolated adipocytes (1). Evidence supporting the hexosamine hypothesis of glucose toxicity included the finding that the induction of insulin resistance requires three components, namely insulin, glucose, and glutamine. When any one of these components is omitted, little or no desensitization is observed (1,2,4). This was explained by the fact that the formation of hexosamine products requires a supply of both glucose (in the form of fructose-6-phosphate) and glutamine as a cofactor for glutamine:fructose-6-phosphate amidotransferase in the transfer of an amide group to fructose-6-phosphate. The primary role of insulin in this scheme is to facilitate the uptake of glucose (ϳ10-to 20-fold) by increasing the number of cell surface glucose transporters. Two independent approaches were used to substantiate this hypothesis (1). First, we demonstrated that glucose-induced desensitization could be prevented by glutamine analogs that irreversibly inactivate glutamine-requiring enzymes such as glutamine:fructose-6-phosphate amidotransferase, the first and rate-limiting enzyme of the HBP. Second, we showed that GlcN was Ͼ40 times more potent than glucose in inducing insulin resistance by virtue of its ability to undergo internalization though the GTS and directly enter the HBP at the level of GlcN-6-P (the first product of the HBP). Taken together, these studies led to the conclusion that the HBP serves as a glucose sensor (or nutrient sensor) coupled to a complex signal transduction system that plays an integral role in the development of glucose-induced insulin resistance.
With the realization of the glucose-sensing/transductional capabilities of the HBP, several investigators began exploring the potential role of the HBP in regulating glycogen metabolism (29 -32). In part, these studies were initiated because hyperglycemia and type 2 diabetes culminate in an impaired ability of insulin to enhance GS activity and stimulate GBS (33,34). Thus, it was logical to postulate that insulin resistance and defective glycogen metabolism were both causally related to hyperglycemia and enhanced flux through the HBP. Because avenues available to test this hypothesis were limited, several investigators used a direct pharmacological approach that entailed the use of GlcN. When these glycogen studies were collectively considered, a conceptual problem emerged in that GlcN treatment yielded results that differed from the defects in glycogen metabolism found under hyperglycemic conditions associated with type 2 diabetes. Moreover, findings among the various studies were not in good agreement. For example, when rats were infused with GlcN during a 6-h euglycemic-hyperinsulinemic clamp, glycogen content in heart and rectus abdominis tissue remained constant (or increased), whereas glycogen synthase activity was diminished in both tissues (18). In contrast, GlcN treatment in vitro was found to abolish insulin stimulation of glycogen synthesis in isolated rat diaphragm but to activate glycogen synthase (29). In fibroblasts, GlcN treatment similarly induced a defect in GBS and increased GS activity (31). From these studies, it is apparent that information derived from the use of GlcN is often conflicting and inconsistent and does not fit into a coherent understanding of GlcN action on glycogen metabolism.
A potential pitfall in the aforementioned studies lies in the use of excessively high concentrations of GlcN for prolonged times. Thus, rather than examining the physiological consequences of increased hexosamine flux, it is possible these investigators were actually assessing the actions of abnormally high intracellular accumulation of GlcN-6-P. Among the first to address this consideration were Virkamaki and Yki-Jarvinen (10), who found that GlcN-6-P levels were increased 500 -700fold in muscle tissue after rats were infused with GlcN for 6 h. These investigators postulated that GlcN-6-P could allosterically regulate GS based on the structural similarities between glucose-6-P (a known allosteric activator of GS) and GlcN-6-P. When the ability of GlcN-6-P to stimulate glycogen synthase activity was directly assessed in vitro using enzyme derived from rectus abdominis and heart muscle, a 21 and 542% increase in enzyme activity was seen.
Although GlcN-6-P was shown to allosterically activate GS, several key questions remained unanswered. For example, could elevated levels of GlcN-6-P stimulate GBS in intact cells and, if so, what is the dose-response relationship between intracellular levels of GlcN-6-P and enhanced GBS? In the current study we addressed these questions and designed experimental protocols with three goals in mind. The first goal was to test the hypothesis that in vivo accumulation of GlcN-6-P (in adipocytes) has a direct effect on GBS. Second, we sought to determine the rapidity of GlcN action, because it should be manifested in minutes if allosteric regulation is involved. Our third goal was to test whether the actions of GlcN on glycogen metabolism were biphasic, with stimulatory effects seen at low concentrations of GlcN and inhibitory actions manifested at high doses. Such a finding could provide a tentative explanation for the conflicting data regarding GlcN action on glycogen metabolism. Our approach was direct in that we sought to elucidate the temporal and dose-dependent relationship between GlcN treatment and the resulting changes in glycogen metabolism (GBS and UDP-glucose levels) and intracellular hexosamine levels (GlcN-6-P and UDP-GlcNAc). More specifically, GlcN action was evaluated as a function of treatment time (5 min to 4 h) and GlcN dose (20 M to 10 mM) by using insulin-treated cells, glucose-treated cells, and control cells (no insulin or glucose treatment).
Based on obtained results, we conclude that GlcN-6-P accumulation in intact adipocytes enhances GBS within minutes, resulting in lower steady-state levels of UDP-glucose (due to rapid precursor utilization). Supporting this idea was the finding that the incorporation of [ 14 C]glucose into glycogen was markedly enhanced over 10 min (2-fold increase; t1 ⁄2 of Ͻ5 min) in insulin-pretreated adipocytes exposed to 2 mM GlcN. During this time, intracellular levels of UDP-glucose were concomitantly reduced (t1 ⁄2 of 1.4 min; Ͼ90% decrease). It is unlikely that the rapid and extensive decrease in UDP-glucose levels was due to the glucose-free conditions (with GlcN treatment), because similar results were obtained under euglycemic conditions (the presence of 5 mM glucose). For several reasons, we believe that the stimulation of GBS and the depletion of UDPglucose are the result of the intracellular accumulation of GlcN-6-P and the allosteric activation of GS. First, we found that GlcN-induced effects on glycogen metabolism correlated well with the early and rapid rise in GlcN-6-P levels (a marked increase by 5 min), which is consistent with the finding that GlcN-6-P is an in vitro allosteric activator of GS. Second, we observed a rapid reduction in GlcN-6-P levels after the removal of GlcN (by washing) and a concomitant restoration of UDP-glucose.
With prolonged GlcN treatment (at 2 mM), we observed a progressive and massive accumulation of GlcN-6-P (Ͼ1,400 nmol/g; t1 ⁄2 of ϳ32 min; GlcN ED 50 of 620 M) that correlated with a reduction in GBS. As we reported previously, high GlcN concentrations can overwhelm the biosynthetic capacity of the HBP, resulting in a massive accumulation of GlcN-6-P and the depletion of cellular ATP levels (14). ATP depletion likely results from phosphate sequestration (with the formation of GlcN-6-P) or the increased energy demands of phosphorylation. Thus, the observed reduction in GBS with excessive GlcN treatment may result from a cytotoxic effect rather than a physiological action mediated through the HBP. This finding is significant, because most studies have examined the functional role of the HBP by using high GlcN concentrations and treatment periods from hours to days. The use of such protocols could easily explain the confusion in the literature regarding the role of HBP in glycogen metabolism.
The storage of intracellular glucose as a branched glucose polymer of glycogen is crucial to glucose homeostasis and provides a means to regulate energy balance between meals (35). After eating, elevated blood glucose levels are reduced through a combination of glucose utilization and glucose storage. The liver plays a major role in storage because it can convert glucose to glycogen, which can later be released to maintain constant blood glucose levels. In skeletal muscle, the catabolism of stored glycogen to glucose provides much of the postprandial energy needed for muscle contraction. In both tissues, GS is the rate-limiting enzyme in GBS and is regulated by complex mechanisms, which include rapid allosteric activation by glucose-6-P, inhibition of GS activity by phosphorylation, and possibly the intracellular compartmentalization of GS (11,35). Interestingly, it has been recently postulated that GS activity may also be regulated by posttranslational modification through the addition of a single GlcNAc monosaccharide on serine/threonine resides (36,37). In other words, defect(s) in glycogen storage observed in hyperglycemic, type 2 diabetics may be mediated by increased flux of glucose into the HBP, enhanced formation of UDP-GlcNAc, and rapid O-linked glycosylation of GS (which inhibits enzyme activity). Despite the physiological appeal of this hypothesis, several aspects of the current study are inconsistent with this proposal. For example, we found that increased hexosamine flux resulted in marked stimulation of GBS, not inhibition. Moreover, we found that that rapid activation of GBS occurred prior to measurable increases in UDP-GlcNAc levels. Although additional studies will be required to resolve this apparent discrepancy, one potential explanation is that both mechanisms are operative, with rapid allosteric activation of glycogen synthase by GlcN-6-P followed by a slower inactivation due to O-linked glycosy-lation. This would be consistent with the biphasic actions of GlcN that we observed in the current studies.
A conceptually important idea derived from the current study is that enhanced flux within the HBP orchestrates both short term (allosteric) and long term (transcriptional) regulatory actions in response to nutritional information derived from the internal and external environment. Although we focus on rapid allosteric control of the GBS pathway, elevated levels of GlcN-6-P may exert allosteric actions that influence other metabolic pathways. For example, it has been shown that GlcN-6-P can allosterically inhibit glucose-6-P dehydrogenase (38,39), the first and rate-limiting enzyme of the pentose phosphate shunt. GlcN-6-P has also been shown to inhibit hexokinase, which is a key enzyme that phosphorylates incoming glucose to glucose-6-P (18). Thus, intracellular levels of GlcN-6-P could potentially influence the insulin-responsive GTS to mimic glucose-induced desensitization. Within the hexosamine pathway itself, it appears that GlcN-6-P can inhibit glutamine:fructose-6-phosphate amidotransferase (40), allosterically activate GlcN-6-P deaminase (41), and activate UDP-GlcNAc pyrophosphorylase at physiological levels of GlcN-6-P (18). From these studies it appears that elevated intracellular levels of GlcN-6-P may coordinately regulate several metabolic pathways and may also play a role in directly modulating the glucose-sensing/ transductional capabilities of the HBP.
Understanding the pathophysiology of type 2 diabetes and the etiology of hyperglycemia-mediated diabetic complications is crucial in developing treatment regimens and discovering novel therapeutic agents. However, a major problem in elucidating the mechanism(s) underlying the cellular abnormalities associated with type 2 diabetes is that various clinical parameters are assessed under chronic hyperglycemic conditions. Consequently, it is difficult to separate the initial triggering events of hyperglycemia from the complicated and long term adaptive actions of various cells and tissues. We know that the HBP plays a key role in sensing hyperglycemic conditions, and we postulate that this pathway initiates or triggers the adaptive responses that ultimately lead to the clinical abnormalities of diabetes and the development of diabetic complications. What is required is an understanding of both the short term and long term consequences of hyperglycemia and the finding of new insights into how these changes are interrelated.