Report

Regulation of Blood Glucose by Hypothalamic Pyruvate Metabolism

See allHide authors and affiliations

Science  05 Aug 2005:
Vol. 309, Issue 5736, pp. 943-947
DOI: 10.1126/science.1112085

Abstract

The brain keenly depends on glucose for energy, and mammalians have redundant systems to control glucose production. An increase in circulating glucose inhibits glucose production in the liver, but this negative feedback is impaired in type 2 diabetes. Here we report that a primary increase in hypothalamic glucose levels lowers blood glucose through inhibition of glucose production in rats. The effect of glucose requires its conversion to lactate followed by stimulation of pyruvate metabolism, which leads to activation of adenosine triphosphate (ATP)–sensitive potassium channels. Thus, interventions designed to enhance the hypothalamic sensing of glucose may improve glucose homeostasis in diabetes.

Diabetic hyperglycemia is due in part to an inappropriately elevated rate of liver glucose production (1). This is paradoxical, since hyperglycemia should restrain glucose production (24). Hypothalamic glucose sensing plays an important role in preserving nutrient homeostasis (510). Here we test the hypothesis that activation of neuronal pyruvate flux is required for hypothalamic glucose sensing and for control of blood glucose levels and liver glucose metabolism. Blood glucose gains access to the CNS by means of a facilitated transport system (11), and indeed, the extracellular concentration of glucose in the brain is significantly lower (∼2 mM) than its concentration in blood (∼5 mM). However, it remains controversial whether neurons directly metabolize glucose to pyruvate or use lactate derived from the anaerobic glycolysis of glucose within astrocytes (1214) (Fig. 1A). The transfer of glucose-derived lactate from astrocytes to neurons is referred to as the “astrocyte-neuron lactate shuttle” (12, 15).

Fig. 1.

Central administration of glucose or lactate lowers blood glucose by means of suppression of endogenous glucose production. (A) Schematic representation of hypothesis. Glucose enters astrocytes where it is metabolized to pyruvate by means of glycolysis. Pyruvate is preferentially converted to l-lactate (l-LACT) by lactate dehydrogenase (LDH-A in astrocytes) and can be taken up by neurons to generate pyruvate by means of LDH (LDH-B in neurons). Finally, pyruvate is converted to acetyl-CoA by pyruvate dehydrogenase (PDH). The increased flux through PDH ultimately leads to the activation of KATP channels. Oxamate (OXA) is a competitive inhibitor of LDH. Dichloroacetate (DCA) is an activator of PDH. (B) Effect of the central administration of glucose (black squares and bars; n = 6) or l-lactate (gray triangles and bars; n = 6), mannitol (open squares, white bars; n = 5), or d-lactate (open triangles, white bars; n = 5) on glucose levels. Glucose and l-lactate lowered blood glucose, whereas mannitol or d-lactate did not. During clamp, (C) glucose and l-lactate increased glucose infusion, (D) decreased glucose production, and (E) decreased liver G6pc mRNA relative to mannitol or d-lactate. *P < 0.01 versus mannitol or d-lactate. Values are means ± SEM.

To examine the central effects of glucose on systemic glucose homeostasis, we infused d-glucose (2 mM) into the third cerebral ventricle of conscious rats. Intracerebro-ventricular (ICV) glucose resulted in a 69% increase in the hypothalamic glucose concentration (from 0.88 to 1.48 μmol/g) and in decreased blood glucose and insulin levels (Fig. 1B) (table S1). In the presence of basal insulin (table S1), the rate of glucose infusion required to maintain euglycemia was marginal in rats receiving ICV mannitol (2 mM). However, when rats received ICV glucose, glucose had to be infused systemically to prevent hypoglycemia (Fig. 1C). The increased glucose infusion was due to suppression of liver glucose production (Fig. 1D).

Glucose production represents the net contribution of gluconeogenesis and glycogenolysis. However, a portion of glucose entering the liver by means of phosphorylation of glucose is also a substrate for dephosphorylation by means of glucose-6-phosphatase (G6Pase, encoded by G6pc), creating a futile cycle. To delineate the mechanisms by which central activation of glucose metabolism modulates blood glucose and liver glucose homeostasis, we estimated the in vivo flux through G6Pase and the relative contribution of gluconeogenesis and glycogenolysis to glucose output. ICV glucose markedly decreased the flux through G6Pase and the hepatic expression of G6pc (Fig. 1E; fig. S1A). Inhibition in both gluconeogenesis and glycogenolysis accounted for the decrease in glucose production (fig. S1A). Conversely, the levels of glucoregulatory hormones, the rate of glucose utilization, and the hepatic expression of the gluconeogenic gene phosphoenolpyruvate carboxykinase, encoded by Pck1, were unchanged (fig. S1A, table S1). These results indicate that the potent effects of central glucose on liver glucose fluxes are largely due to marked inhibition of hepatic G6Pase.

According to the astrocyte-neuron lactate shuttle hypothesis (12), central administration of glucose increases the formation of lactate in astrocytes (with the conversion of each glucose molecule into two molecules of lactate) to provide extracellular lactate for neurons, which in turn rapidly convert it to pyruvate (Fig. 1A). Therefore, we investigated whether ICV lactate could recapitulate the effects of ICV glucose on blood glucose and liver glucose metabolism. Indeed, the ICV infusion of l-lactate (5 mM) decreased blood glucose and insulin levels (Fig. 1B; table S2) compared with equimolar infusions of its isomer d-lactate, which cannot be metabolized. In the presence of basal levels of circulating insulin (table S2), ICV l-lactate increased the rate of glucose infusion required to maintain euglycemia (Fig. 1C). This increase was due to suppression of liver glucose production (Fig. 1D). The flux through G6Pase, the hepatic expression of G6pc, gluconeogenesis, and glycogenolysis were all decreased in rats receiving ICV l-lactate (Fig. 1E; fig. S1B). However, the levels of glucoregulatory hormones, the rate of glucose utilization, and the hepatic expression of Pck1 were unchanged (fig. S1B, table S2). Thus, a moderate increase in the central availability of either glucose or lactate is sufficient to lower blood glucose by means of rapid changes in liver glucose metabolism and gene expression.

The use of extracellular lactate by neurons requires its conversion to pyruvate by the enzyme lactic dehydrogenase (mainly LDH-B in neurons) (16). If the effects of ICV lactate on glucose homeostasis are mediated by neuronal lactate use, the inhibition of LDH should negate these effects. Therefore, we examined whether impeding the conversion of lactate to pyruvate negates the metabolic effects of the central administration of l-lactate (Fig. 1A). Oxamate is a competitive inhibitor of LDH (17). The coinfusion of oxamate (50 mM) abolished the effects of ICV l-lactate on blood glucose and insulin levels (Fig. 2A; table S2). During clamp, ICV oxamate also negated the effects of ICV l-lactate on glucose infusion (Fig. 2B), glucose production (Fig. 2C), G6Pase flux (fig. S2A), G6pc mRNA (Fig. 2D), gluconeogenesis (Fig. 2E), and glycogenolyis (Fig. 2F). These data indicate that the metabolism of lactate to pyruvate is an obligatory biochemical step for the regulation of liver glucose homeostasis by central lactate.

Fig. 2.

Central inhibition of LDH negates the effects of l-lactate and glucose on glucose homeostasis. (A) The ICV coinfusion of the LDH inhibitor oxamate with glucose (black squares, dotted lines; n = 6) or with l-lactate (gray triangles, dotted lines; n = 6) abolished the blood glucose–lowering effect of glucose (black squares; n = 6) or l-lactate (gray triangles; n = 6). Oxamate alone (white squares) did not modify blood glucose. *P < 0.01 versus oxamate and versus oxamate with glucose or l-lactate. (B) During clamp, ICV coinfusion of oxamate with l-lactate (right gray bars) or glucose (right black bars) negated the effects of ICV l-lactate (left gray bars) or glucose (left black bars) on the rate of glucose infusion, (C) on the suppression of glucose production, (D) onliver G6pc mRNA, (E) on gluconeogenesis, and (F) on glycogenolysis. *P < 0.05 versus oxamate and versus oxamate with glucose or l-lactate. Values are means ± SEM.

Glucose could be directly metabolized to pyruvate within neurons or it may first be converted to lactate, the metabolism of which then generates pyruvate. If the conversion of glucose to lactate is required for the effects of ICV glucose on liver glucose homeostasis, the coinfusion of oxamate should also prevent the metabolic effects of ICV glucose. Indeed, the central administration of oxamate abolished the effects of ICV glucose on blood glucose and insulin levels (Fig. 2A; table S1). During clamp, ICV oxamate also negated the effects of ICV d-glucose on glucose infusion (Fig. 2B), glucose production (Fig. 2C), G6Pase flux (fig. S2B), G6pc mRNA (Fig. 2D), gluconeogenesis (Fig. 2E), and glycogenolysis (Fig. 2F). Together these findings reveal a negative-feedback loop between the central availability of glucose or lactate and the regulation of liver glucose homeostasis.

The activation of hypothalamic adenosine triphosphate (ATP)–sensitive potassium (KATP) channels is critical for modulation of blood glucose levels and liver glucose fluxes (18), as well as for the metabolic effects of other nutrient-dependent signals (1820). To examine the role of central KATP channels in carbohydrate sensing, we coinfused ICV the KATP blocker glibenclamide (100 μM) with lactate or glucose (Fig. 1A). This inhibitor negated the effects of both ICV lactate and glucose on blood glucose (Fig. 3A). Furthermore, during clamp, the central infusion of glibenclamide prevented the increase in the rate of glucose infusion (Fig. 3B), as well as the decrease in glucose production (Fig. 3C) induced by either ICV lactate or glucose. Similarly, the KATP blocker abolished the effects of central lactate and glucose on G6Pase flux (fig. S3, A and B), G6pc mRNA (Fig. 3D), gluconeogenesis (Fig. 3E), and glycogenolysis (Fig. 3F). Glucoregulatory hormone levels, glucose use, and liver Pck1 mRNA were unchanged (fig. S3, A and B; tables S1 and S2). These findings indicate that the central activation of KATP channels bridges the hypothalamic sensing of glucose and lactate with the regulation of liver glucose homeostasis.

Fig. 3.

Central administration of a KATP channels blocker negates the effects of l-lactate and glucose on glucose homeostasis. (A) The ICV coinfusion of the KATP channels blocker glibenclamide with glucose (black squares, dotted lines; n = 6) or l-lactate (gray triangles, dotted lines; n = 6) abolished the blood glucose lowering effect of glucose (black squares; n = 6) or l-lactate (gray triangles; n = 6). Glibenclamide alone (white squares) did not modify blood glucose; *P < 0.01 versus glibenclamide and versus glibenclamide with glucose or l-lactate. (B) During clamp, ICV coinfusion of glibenclamide with l-lactate (right gray bars) or glucose (right black bars) negated the effects of ICV l-lactate (left gray bars) or glucose (left black bars) on the rate of glucose infusion, (C) on the suppression of glucose production, (D) on liver G6pc mRNA, (E) on gluconeogenesis, and (F) on glycogenolysis; *P < 0.01 versus glibenclamide or versus glibenclamide with glucose or l-lactate. Values are means ± SEM.

To investigate the anatomical localization of these effects, we infused a dose of glucose that was lower (by a factor of 15) than that used in the ICV studies bilaterally within the parenchyma of the mediobasal hypothalamus (intrahypothalamic, IH) (fig. S4, A and B). IH glucose (2 mM) lowered circulating glucose levels (Fig. 4A). During clamp, glucose had to be infused systemically to prevent hypoglycemia when glucose was administered IH (Fig. 4B). The increased glucose infusion was due to a reduction in liver glucose production (Fig. 4C); however, the rate of glucose use was not affected (fig. S4C). Importantly, the IH coinfusion of glibenclamide negated the effects of IH glucose on blood glucose, glucose infusion, and glucose production (Fig. 4, A to C). Thus, the increased availability of glucose leads to the activation of a KATP-dependent pathway within the mediobasal hypothalamus that is sufficient to lower blood glucose by means of the inhibition of glucose production. Because the relevant (Sur1-containing) KATP channels do not appear to be expressed in glial cells (18, 19), it is likely that this step in hypothalamic glucose sensing is occurring in neurons.

Fig. 4.

Intrahypothalamic infusions of glucose, PDH activator, or LDH inhibitor regulate glucose production. (A) Bilateral cannulae were placed within the parenchyma of the mediobasal hypothalamus. The IH coinfusion of the KATP channels blocker glibenclamide with glucose (gray squares; n = 5) abolished the blood glucose–lowering effect of glucose (black squares; n = 5). Glibenclamide alone (white squares) did not modify blood glucose. *P < 0.01 versus glibenclamide or glucose glibenclamide. (B) During clamp, IH coinfusion of glibenclamide with glucose + dichloroacetate (DCA, black bars; n = 5) lowered blood glucose, (E) increased glucose infusion, and (F) decreased glucose production; *P < 0.01 versus vehicle (white bars). (G) During pancreatic-basal insulin clamps, circulating blood glucose was doubled and then maintained at these levels in the presence (gray bars, n = 4) or absence (black bars, n = 4) of IH oxamate. IH oxamate resulted in (H) decreased rates of glucose infusion during pancreatic-hyperglycemic clamps, (I) did not alter glucose production in the presence of euglycemia (white bar), but increased glucose production during the hyperglycemia so that the suppressive effect of hyperglycemia on glucose production was diminished. *P < 0.001, IH OXA versus IH vehicle. Values are means ± SEM.

The entry of pyruvate in the tricarboxylic cycle (TCA) is governed by its conversion to acetyl-coenzyme A (acetyl-CoA) by pyruvate dehydrogenase (PDH). If the flux of pyruvate into the TCA cycle is the key signal generated by the central administration of glucose or lactate, then the stimulation of pyruvate flux should recapitulate their effects on glucose homeostasis (Fig. 1A; fig. S4A). Dichloroacetate (DCA) activates PDH by means of inhibition of PDH kinase and increases pyruvate flux in neurons and in astrocytes (21). The delivery of DCA (1 mM) within the mediobasal hypothalamus lowered circulating glucose levels (Fig. 4D). During clamp, the IH administration of DCA increased the rate of glucose infusion required to maintain euglycemia (Fig. 4E). This effect was due to suppression of liver glucose production (Fig. 4F), rather than to increased glucose utilization (fig. S4D). Because activation of PDH in astrocytes would actually decrease lactate formation, our results with DCA are consistent with neuronal activation of PDH in vivo. This in turn reproduced the metabolic effects of central glucose or lactate. Because the enhanced conversion of pyruvate to acetyl-CoA should lead to increased flux into the TCA cycle (Fig. 1A), it seems plausible that the neuronal TCA cycle serves as a biochemical sensor for carbohydrate availability in the hypothalamus, which in turn regulates liver glucose homeostasis.

To estimate the contribution of hypothalamic glucose sensing to the overall effects of circulating glucose on liver glucose homeostasis, we used the pancreatic-hyperglycemic clamp technique to double the circulating glucose levels in rats receiving IH oxamate or IH vehicle (Fig. 4G). This level of hyperglycemia led to a twofold increase in the hypothalamic glucose levels (to 2.1 μmol/g). The rate of glucose infusion required to maintain hyperglycemia was less with IH oxamate than with IH vehicle (Fig. 4H). Hyperglycemia inhibited glucose production by 88% in the presence of IH vehicle and by only53% in the presence of IH oxamate (Fig. 4I). Thus, blocking the metabolism of lactate in the hypothalamus results in a loss of 40% of the inhibitory action of circulating glucose on glucose production, which indicates that the neuronal circuit engaged in response to increased hypothalamic glucose metabolism plays an important role in restraining glucose production in response to an increase in the circulating glucose levels.

Our results are also relevant to the ongoing debate on the physiological role of the astrocyte-neuron lactate shuttle (12). According to this hypothesis, an increase in neuronal activity raises the extracellular levels of the excitatory neurotransmitter glutamate, which in turn activates glycolysis and lactate production in astrocytes. Consistent with this notion, the IH infusion of glutamate (1 mM) recapitulated the effects of IH glucose on blood glucose levels and on liver glucose production (22). Because the entry of pyruvate into the TCA cycle is a dominant pathway in neuronal energetics, the role of pyruvate flux in glucose sensing appears to link neuronal activity within this hypothalamic region to the regulation of liver glucose homeostasis. Alterations in the function of mitochondria have been suggested to play important roles in the etiology of metabolic changes associated with aging, diabetes, and obesity (2325). These alterations, if extended to hypothalamic neurons, could result in decreased flux through PDH and, therefore, could hamper the responses to nutritional cues converging on the increased availability of pyruvate.

The arcuate nucleus of the hypothalamus is emerging as a major site for the integration of multiple nutritional (19, 26) and hormonal (18, 20, 2731) signals, which are central to the modulation of liver glucose homeostasis. Type 2 diabetes and the metabolic syndrome are typical examples of diseases the prevalence of which is dependent on environmental and/or nutritional factors operating on genetic susceptibility. Impairment in the biochemical sensing of carbohydrates may represent a basic under-pinning for defects in the regulation of food intake (6), β-cell function (32), and liver glucose homeostasis (4).

In conclusion, neurons appear to have developed mechanisms for glucose sensing designed to protect them from hypoglycemic injury by triggering the rapid secretion of counterregulatory hormones in response to low extracellular glucose levels (810). Here, we have shown that moderate increases in extracellular glucose levels within a specific region of the hypothalamus are sufficient to lower blood glucose levels through a robust inhibition of liver glucose production. This discovery paves the way for biochemical interventions designed to restore central glucose sensing and glucose homeostasis.

Supporting Online Materials

www.sciencemag.org/cgi/content/full/309/5736/943/DC1

Materials and Methods

Fig. S1 to S4

Table S1 and S2

References and Notes

References and Notes

View Abstract

Navigate This Article