Control of Blood Glucose in the Absence of Hepatic Glucose Production During Prolonged Fasting in Mice

Abstract

OBJECTIVE

Since the pioneering work of Claude Bernard, the scientific community has considered the liver to be the major source of endogenous glucose production in all postabsorptive situations. Nevertheless, the kidneys and intestine can also produce glucose in blood, particularly during fasting and under protein feeding. The aim of this study was to better define the importance of the three gluconeogenic organs in glucose homeostasis.

RESEARCH DESIGN AND METHODS

We investigated blood glucose regulation during fasting in a mouse model of inducible liver-specific deletion of the glucose-6-phosphatase gene (L-G6pc^−/− mice), encoding a mandatory enzyme for glucose production. Furthermore, we characterized molecular mechanisms underlying expression changes of gluconeogenic genes (G6pc, Pck1, and glutaminase) in both the kidneys and intestine.

RESULTS

We show that the absence of hepatic glucose release had no major effect on the control of fasting plasma glucose concentration. Instead, compensatory induction of gluconeogenesis occurred in the kidneys and intestine, driven by glucagon, glucocorticoids, and acidosis. Moreover, the extrahepatic action of glucagon took place in wild-type mice.

CONCLUSIONS

Our study provides a definitive quantitative estimate of the capacity of extrahepatic gluconeogenesis to sustain fasting endogenous glucose production under the control of glucagon, regardless of the contribution of the liver. Thus, the current dogma relating to the respective role of the liver and of extrahepatic gluconeogenic organs in glucose homeostasis requires re-examination.

Since the pioneering work of Claude Bernard in the 19th century, the scientific community has long considered the liver to be the major source of endogenous glucose production (EGP) in any postabsorptive situation (1). Nevertheless, EGP can also be performed by the kidneys and the intestine, particularly during fasting (2–6) or under protein feeding (7,8). In these two situations, quantifications of kidney and intestinal glucose release have suggested that the liver might contribute far less glucose than expected (<50%). These data questioned the predominance of the liver in EGP. The liver, kidney, and intestine are the only organs capable of producing glucose in blood because they are the only organs known to express the catalytic subunit (G6PC) of the glucose-6-phosphatase (G6Pase) system (9). Indeed, G6Pase is the crucial enzyme of EGP. It catalyzes the hydrolysis of glucose-6 phosphate, an obligate step, common to glycogenolysis and gluconeogenesis, occurring immediately before the release of glucose in blood. Moreover, in humans, mutations in the G6PC gene result in glycogen storage disease type 1a (GSD1a), an autosomal recessive metabolic disorder. Patients are unable to maintain blood glucose concentration and suffer from severe hypoglycemia episodes if not treated by continuous feeding (10–12). A total knock-out model for the G6pc gene was developed, but these mice are unable to survive weaning despite glucose injections every 8 h (13).

Recently, we developed a novel mouse model of GSD1a in which G6pc deletion exclusively targeted the liver (L-G6pc^−/− mice) (14). L-G6pc^−/− mice are viable and show all hepatic hallmarks of GSD1a, i.e., accumulation of glucose-6 phosphate, glycogen, and triglycerides (14). This results in marked hepatomegaly and steatosis. It is noteworthy that L-G6pc^−/− mice exhibit normoglycemia in the fed state (14). Thus, this animal model can be used to address the question of the predominance of the liver and of the capacity of extrahepatic gluconeogenesis in the control of glucose homeostasis in the absence of food glucose. We hypothesized that L-G6pc^−/− would regulate their blood glucose during long-term fasting as a result of the induction of extrahepatic glucose production. We also identified novel mechanisms involved in the induction of gluconeogenesis in the kidneys and intestine, in the absence of liver glucose production. These novel findings were also extended to the wild-type (WT) physiology.

RESEARCH DESIGN AND METHODS

Generation of liver-specific G6pc-null mice.

The lack of hepatic G6Pase was obtained by a specific deletion of G6pc exon 3 in the liver using a CRE-lox strategy in mice, which was recently described by Mutel et al. (14). Male adult (6–8 weeks old) B6.G6pc^lox/lox.SA^creERT2/w and C57Bl/6J mice (Charles Rivers Laboratories) were injected intraperitoneally once daily with 100 µL of tamoxifen (10 mg/mL) on 5 consecutive days, to obtain L-G6pc^−/− and L-G6pc^+/+ (WT) mice, respectively (14). Mice were housed in the animal house of Lyon 1 University, in controlled temperature (22°C) conditions, with a 12-h light–12-h-dark cycle. Mice had free access to water and standard chow (unless fasted). Mice were studied 5 weeks after tamoxifen treatment. The specific hepatic deletion of G6pc exon 3 in the L-G6pc^−/− mice was always verified by PCR on purified liver genomic DNA after euthanasia, as described by Mutel et al. (14). All procedures were performed in accordance with the guidelines established by the European Convention for the Protection of Laboratory Animals. The animal care committee of the University approved all experiments.

Metabolic studies.

Blood was withdrawn from the retro-orbital vein for plasma metabolite determinations. Glucose, triglyceride, and total cholesterol concentrations were determined with Biomérieux colorimetric kits. Free fatty acid and glycerol concentrations were determined with a Diasys colorimetric kit. Insulin, glucagon, and corticosterone concentrations were determined with mouse ELISA kits from Chrystal Chem, Biovendor, and Neogen, respectively. Catecholamines and amino acids levels were determined by high-performance liquid chromatography (Centre Hospitalier Universitaire, Lyon). β-Hydroxybutyrate was measured using Optium β−ketone test strips with Optium Xceed sensors (Abbott Diabetes Care, U.K.). Blood glucose was determined with an Accu-Chek Go glucometer (Roche Diagnostics) during fasting experiments. Hepatic glycogen determinations were carried out as described by Keppler and Decker (15).

In vivo analyses.

EGP was determined in conscious, unrestrained, catheterized mice. A catheter was inserted into the right jugular vein under anesthesia, and mice were allowed to recover for 4–6 days. After 6 h or 24 h of fasting, mice were infused with a bolus (92.5 kBq) of [3-³H] glucose. We then administered [3-³H] glucose at a rate of 6.3 kBq/min. Blood glucose was monitored every 15 min. In preliminary experiments, we checked that a steady state of plasma [3-³H] glucose-specific activity reached from 30 min of infusion (data not shown). Plasma glucose concentration did not vary during the infusion time. At a steady state of glycemia and [3-³H] glucose-specific activity (between 60 to 90 min), blood was collected to determine glucose specific activity and glucose concentration (16). EGP was calculated as Ra using the simplified equation of Steele, by dividing the [3-³H] glucose infusion rate by the [3-³H] glucose-specific activity (16).

l-Alanine and l-glutamine tolerance tests were performed after 6 h or 24 h of fasting. Mice were injected intraperitoneally with l-alanine or l-glutamine (2 g/kg BW). Blood glucose levels were determined from the tail vein at 0, 15, 30, 45, 60, and 120 min after injection.

For glucagon challenge, WT mice were injected intraperitoneally with 0.9% NaCl or glucagon (Glucagen, Novo Nordisk) at a dose of 200 ng/g of weight. Mice were killed by cervical dislocation after 30 min. Kidneys and intestine were rapidly removed and fixed in formaldehyde for chromatin immunoprecipitation (ChIP) experiments, as described previously (17).

For GcgR antagonist experiments, L-G6pc^−/− mice were dosed by gavage with 50 mg/kg of GcgR antagonist L168,049 (Glucagon receptor antagonist II, Calbiochem) in 5% PEG-400/5% Tween-80/90% H2O or vehicle, as described previously (18). Immediately after gavage, mice were fasted and killed 6 h later.

Gene expression analysis.

Mice were killed by cervical dislocation. The liver and kidneys were rapidly removed, with tongs chilled previously in liquid N₂, and frozen. The intestine (proximal jejunum) was rinsed and immediately frozen in liquid N₂. G6Pase and PEPCK-c activities were assayed at maximal velocity, as described previously (19). Immunoblotting was carried out using purified anti-rat G6PC (20), anti-PEPCK (Santa Cruz Biotechnology), and antiglutaminase (kindly provided by Dr. N. Curthoys) antibodies. Membranes were reprobed with anti-β-actin monoclonal antibody for standardization. Total RNAs were isolated from tissues with TRIzol reagent. Reverse transcription and real-time PCR were performed using sequence-specific primers (Supplementary Table 1), with ribosomal protein mL19 transcript (Rpl19) as a reference.

ChIP assays.

Nuclear isolation and ChIP assays were performed as described (17). Chromatin obtained was then sheared with the Enzymatic shearing kit (Active Motif), yielding chromatin fragments of 500–200 bp in size. Each immunoprecipitation was performed with about 10 µg of chromatin using the ChIP-It Express kit (Active Motif). Chromatin complexes were immunoprecipitated for 16–18 h at 4°C while rotating with either phospho-CREB (pS133) antibody (Epitomics), glucocorticoid receptor antibody (sc-1004; Santa Cruz Biotechnology), or GFP antibody (Santa Cruz) as a negative control. After DNA purification, quantitative PCR amplification was performed using primers specific for the cAMP or glucocorticoid response units of Pck1 and G6pc promoters (Supplementary Table 1).

Statistical analyses.

Results are reported as means ± SEM. The various groups were compared by Mann-Whitney tests. Differences were considered statistically significant at P < 0.05.

RESULTS AND DISCUSSION

Control of blood glucose in L-G6pc^−/− mice during fasting.

G6Pase activity was disrupted specifically in the liver by temporal and tissue-specific deletion of the G6pc gene based on a CRE-lox strategy (14). In brief, transgenic B6.G6pc^lox/lox mice were crossed with transgenic B6.SA^creERT2/w mice (21) to generate B6.G6pc^lox/lox.SA^creERT2/w mice, expressing inducible CRE^ERT2 specifically in the liver. The treatment of adult B6.G6pc^lox/lox.SA^creERT2/w mice with tamoxifen induced the excision of G6pc exon 3, specifically in the liver (14), leading to a complete lack of hepatic G6Pase activity (Fig. 1A). This finding is consistent with these mice being unable to mobilize their glycogen stocks during fasting (Fig. 1C), leading to a hepatomegaly (Table 1). In WT mice, indeed, glycogen stock was about 40% of the fed level at 6 h of fasting and was completely exhausted after 12 h of fasting (0.6 mg/g of liver, n = 4). Although mice with the total knock-out of G6pc die after weaning, these L-G6pc^−/− mice are viable. Moreover, all L-G6pc^−/− mice survived prolonged fasting (45 h), although hepatic G6Pase activity was still undetectable throughout the fasting period (Fig. 1A). Because hepatic glucose production is thought to be critical for glucose homeostasis during fasting periods, plasma glucose concentration and EGP were determined in L-G6pc^−/− and WT mice during prolonged fasting. L-G6pc^−/− mice displayed the same blood glucose level as control mice in the fed state (Fig. 1C). Consistent with the role of hepatic glycogenolysis as an important pathway of glucose production in the early postabsorptive state, EGP levels were found to be lower in 6 h–fasted L-G6pc^−/− animals than in control animals (Fig. 1D). This was correlated with a lower blood glucose in 6 h–fasted L-G6pc^−/− mice (Fig. 1C).

FIG. 1.

Hepatic glucose production is not crucial for the control of fasting glycemia. G6Pase activities (A) and glycogen content (B) were assayed in the liver of L-G6pc^−/− (black bar) and WT (white bar) mice in the fed state (0 h) or in fasted states (6 h, 24 h, and 40 h; n = 5 to 6 mice per group). C: Blood glucose concentrations were determined afterward for 0, 6, 16, 24, 30, 40, and 45 h fasting in L-g6pc^−/− and WT mice. D: EGP was determined in conscious L-G6pc^−/− (black bar) and WT (white bar) mice fasted for 6 h or 24 h (n = 6 mice per group). Mice had free access to water during fasting. Data were obtained 5 weeks after gene deletion and are expressed as means ± SEM. Values significantly different from WT (**P < 0.01). ††, significantly different with respect to the value in fed state in each group (P < 0.01). Φ and ΦΦ, significantly different with respect to the value at 6 h of fasting in each group (P < 0.05 and P < 0.01, respectively).

TABLE 1.

Body and liver weights and plasmatic parameters of 6 h–fasted L-g6pc^−/− mice

	L-g6pc−/−	L-g6pc+/+
Mice per group	6	6
Body wt (g)	27.8 ± 0.8	26.5 ± 0.8
Liver wt (g)	1.8 ± 0.04**	1.1 ± 0.1
Glucose (mmol/L)	3.4 ± 0.2**	7.2 ± 0.3
Insulin (ng/mL)	0.35 ± 0.02**	0.67 ± 0.07
Glucagon (pg/mL)	552 ± 50**	302 ± 28
Corticosterone (ng/mL)	191 ± 19**	71.0 ± 9.7
Norepinephrine (ng/mL)	0.43 ± 0.06	0.43 ± 0.03
Epinephrine (ng/mL)	0.36 ± 0.10	0.36 ± 0.04
Total cholesterol(g/L)	1.33 ± 0.09**	0.89 ± 0.04
Triglycerides (g/L)	1.42 ± 0.1**	0.78 ± 0.03
Free fatty acids (mmol/L)	1.07 ± 0.09**	0.69 ± 0.06
Glycerol (mg/dL)	4.3 ± 0.2	4.2 ± 0.3
Lactate (mmol/L)	5.5 ± 0.5**	3.5 ± 0.3
Alanine (µmol/L)	336 ± 34**	224 ± 18
Glutamine (µmol/L)	538 ± 16*	618 ± 26

Values are expressed as means ± SEM. Data were determined 5 weeks after gene invalidation from L-g6pc^−/− and WT L-g6pc^+/+ mice upon 6 h of fasting.

Values significantly different from WT (*P < 0.05 and **P < 0.01) are indicated (Mann-Whitney test).

Concomitant with this drop in blood glucose observed at 6 h fasting, ketogenesis was rapidly induced in L-G6pc^−/− animals. At 6 h fasting, the amounts of ketone bodies (β-hydroxybutyrate) in L-G6pc^−/− mice were fivefold greater compared with the control mice that were able to produce high levels of glucose by inducing glycogenolysis (Supplementary Fig. 1A). These results were accounted for by fast induction of the mitochondrial 3-hydroxy-3-methylglutaryl-CoA synthase (Hmgcs2) expression in 6 h–fasted L-G6pc^−/− mice (Supplementary Fig. 1B). It is important to recall here that ketone bodies provide an alternative form of cellular energy fuel in postabsorptive situations characterized by low glucose availability. This is true notably for the brain, which may derive up to 70% of its energy from ketone bodies (6), and for the kidney and intestine, the two alternative gluconeogenic organs, which derive 50% of their energy from ketone bodies in postabsorptive state (22,23). It therefore seems unlikely that L-G6pc^−/− mice lacked energy fuels for these essential organs.

It is noteworthy that differences in EGP between L-G6pc^−/− and control animals were no longer observed after longer fasting periods, and that L-G6pc^−/− mice controlled blood glucose effectively during long fasting periods (Fig. 1D). We even observed a significant increase in blood glucose between 6 h and 16 h fasting, such that the blood glucose concentration of L-G6pc^−/− mice eventually reached that of control mice after 30 h fasting (Fig. 1C). By contrast, blood glucose concentration decreased steadily during fasting in control mice, eventually reaching a plateau at 100 mg/dL after 24 h of fasting (Fig. 1C).

Induction of extrahepatic gluconeogenesis in L-G6pc^−/− mice.

It was hypothesized that L-G6pc^−/− mice maintained their glycemia through rapid induction of extrahepatic gluconeogenesis, compensating for lack of glucose production by the liver. It is noteworthy that blood glucose was the same in L-G6pc^−/− and control mice in the fed state, although glucose injection was necessary in the total G6pc knockout mice to maintain blood glucose (24). To explain this difference, we assessed the expression of gluconeogenic genes in the kidneys and intestine of fed animals. Renal G6pc mRNA levels of fed L-G6pc^−/− mice were about 1.7-fold higher than those in control mice (Fig. 2A). This was associated with an increase in renal G6PC protein (Fig. 2B). Finally, specific G6Pase activity in the kidney was 40% higher in L-G6pc^−/− than in control mice (Fig. 2C). On the contrary, renal Pck1 expression and PEPCK activity were not modified (Fig. 2A–C). In the fed state, no modification of G6PC and PEPCK-c expression was observed in the intestine of L-G6pc^−/− compared with control mice (Fig. 2D–F). Thus, only G6Pase was upregulated in the kidneys of L-G6pc^−/− mice in the fed state, reflecting constitutive induction of the gluconeogenic pathway in the kidneys of L-G6pc^−/− compared with WT mice.

FIG. 2.

Expression of main gluconeogenic enzymes in the kidneys (A–C) and the intestine (D–F) of fed L-G6pc^−/− mice. A and D: Level of G6pc and Pck1 mRNA expressed as a ratio relative to the Rpl19 mRNA level. B and E: Western blot and quantification analysis for G6PC and PEPCK-C proteins. Actin is shown as a loading control. C and F: Specific G6Pase and PEPCK-c activity of homozygous L-G6pc^−/− (black bar) and WT (white bar) mice. Data were obtained 5 weeks after gene deletion in fed mice (n = 6 mice per group) and are expressed as the mean ± SEM. Values significantly different from WT (*P < 0.05, **P < 0.01) are indicated. AU, arbitrary units.

After 6 h fasting, the expression of the genes encoding major regulatory enzymes of extrahepatic gluconeogenesis [G6Pase, PEPCK-c, and glutaminase] was dramatically induced in both the kidneys and intestine of L-G6pc^−/− mice. In the kidneys, amounts of G6pc mRNA and protein were about twice those in the control (Fig. 3A and B), and specific G6Pase activity was 36% higher than in control mice (Fig. 3C). Renal PEPCK-c expression was increased concomitantly, resulting in a threefold increase of Pck1 mRNA and protein and specific PEPCK-c activity more than twice that of L-G6pc^−/− mice (Fig. 3A–C). There was also an induction of glutaminase in the kidneys of L-G6pc^−/− mice (Fig. 3B). A similar pattern of induction of gluconeogenic gene expression was observed for the three enzymes in the intestine of L-G6pc^−/− mice (Fig. 3D–F). Thus, the expression of gluconeogenic genes was rapidly induced in both the kidneys and intestine of L-G6pc^−/− mice in postabsorptive state.

FIG. 3.

Expression of main gluconeogenic enzymes in the kidneys (A–C) and the intestine (D–F) of 6 h–fasted L-G6pc^−/− mice. A and D: Level of G6pc and Pck1 mRNA expressed as a ratio relative to Rpl19 mRNA level. B and E: Western blot analysis and quantification for G6PC, PEPCK-C, and glutaminase proteins. Actin is shown as a loading control. Results are expressed as fold induction versus WT. C and D: Specific G6Pase and PEPCK-c activity of homozygous L-G6pc^−/− (black bar) and WT (white bar) mice. Data were obtained 5 weeks after gene deletion, in mice fasted for 6 h (n = 6 mice per group), and are expressed as means ± SEM. Values significantly different from WT (*P < 0.05, **P < 0.01) are indicated.

These results strongly suggested that the L-G6pc^−/− mouse rapidly upregulated gluconeogenesis in the kidneys and intestine once food glucose was lacking. It is noteworthy that there is clear-cut organ substrate selectivity for the liver, kidney, and intestine gluconeogenesis (9). Whereas gluconeogenesis from glutamine only takes place in the kidneys and intestine, gluconeogenesis from alanine and lactate is the key pathway in the liver (4,9,25). To substantiate the view that postabsorptive or fasting L-G6pc^−/− mice were able to regulate plasma glucose from extrahepatic gluconeogenesis in the absence of hepatic gluconeogenesis, we performed tolerance tests from alanine (exclusive hepatic substrate) and glutamine (exclusive extrahepatic substrate). After 6 h of fasting, there was a marked increase in plasma glucose upon glutamine injection in L-G6pc^−/− mice, whereas only a marginal increase was observed in WT mice (Fig. 4A). On the contrary, L-G6pc^−/− mice were unable to increase their plasma glucose in response to alanine injection (Fig. 4B). Surprisingly, there was no increase in plasma glucose in WT mice injected with alanine; this might be a result of the preponderant role of hepatic glycogenolysis in EGP and the nonactivation of gluconeogenesis at 6 h fasting in normal mice (see Fig. 1). In agreement with this rationale, WT mice dramatically increased plasma glucose in response to alanine injection after 24 h of fasting, a time where glycogen stores were exhausted and liver gluconeogenesis was enhanced (Fig. 1A and B and Fig. 4C). On the contrary, L-G6pc^−/− mice were still unable to convert the injection of alanine in increased plasma glucose (Fig. 4C). Also in line with the absence of liver gluconeogenesis in L-G6pc^−/− mice, plasma alanine and lactate concentrations were 50–60% higher than in WT mice (Table 1). On the other hand, the plasma concentration of glutamine was slightly lower in L-G6pc^−/− mice than in WT mice, which was consistent with an increased utilization of glutamine (Table 1). Taken together, these data strongly suggested that L-G6pc^−/− mice regulated plasma glucose from extrahepatic gluconeogenesis exclusively.

FIG. 4.

Glutamine and alanine tolerance tests in L-G6pc^−/− and WT mice . After 6 h (A and B) or 24 h (C) of fasting, L-G6pc^−/− (black circles) and WT (open squares) mice were injected with glutamine (A) or alanine (B and C). Blood glucose levels were measured every 15 min for 2 h. Data were obtained 5 weeks after tamoxifen treatment. Percent values relative to time 0 were expressed as means ± SEM (n = 6 mice/group). Values significantly different from WT (*P < 0.01); †significantly different with respect to the value before alanine or glutamine injection (P < 0.01).

Role of glucagon in physiological adaptation of the L-G6pc^−/− mouse to fasting.

We then investigated the mechanisms underlying expression changes of gluconeogenic genes in both the kidneys and intestine in the absence of G6pc in the liver.

First, we analyzed the hormonal status of L-G6pc^−/− mice in both fed and 6 h fasting states. In mammals, sophisticated hormonal control by insulin, glucagon, glucocorticoids, and cathecholamines maintains blood glucose within narrow limits, orchestrating the uptake and production of glucose. In both the fed state and after 6 h fasting, glucagon levels were higher in L-G6pc^−/− mice than in control mice (Fig. 5A). On the contrary, L-G6pc^−/− mice exhibited a lower insulin level than that of control mice after 6 h fasting (Fig. 5B), but were not hypoinsulinemic in fed state (Fig. 5B). This resulted in a twofold (fed state) to threefold (6 h fasted state) increase in the glucagon-to-insulin ratio in L-G6pc^−/− mice in comparison with that of the control mice (Fig. 5C). Moreover, corticosterone levels were markedly increased in L-G6pc^−/− mice compared with the control mice after 6 h fasting (Table 1), but not in the fed state (63.0 ± 10.8 ng/mL in L-G6pc^−/− mice vs. 65.0 ± 14.0 ng/mL in WT mice). However, the epinephrine and norepinephrine plasma concentrations were similar in both L-G6pc^−/− and control mice (Table 1). This is in agreement with the observation that a hypoglycemic state around 60 mg/dL is mild and does not represent a stressed condition from the metabolic viewpoint, especially in the light of the concomitant increased availability of ketone bodies (see above).

FIG. 5.

Glucagon regulates gluconeogenic gene expression in the kidneys and intestine of L-G6pc^−/− mice. Plasma glucagon (A), plasma insulin (B), and glucagon-to-insulin ratio (C) of fed or 6 h–fasted L-G6pc^−/− (black bars) and control (white bars) mice are shown. D: RT-PCR analysis of the expression of glucagon receptor in the liver, kidney, and intestine of control mice fasted for 6 h. E-G: P-CREB binding to the G6pc or Pck1 promoter was analyzed by ChiP assay from kidneys or intestine of fed mice (E) or 6 h–fasted mice (F and G). H: Level of Pck1 mRNA expressed as a ratio relative to Rpl19 mRNA level in the liver of L-G6pc^−/− mice treated or not with GcgR antagonist L-168,049 (black bars) and WT mice (white bars) fasted for 6 h. I and J: Level of G6pc mRNA expressed as a ratio relative to Rpl19 mRNA level in the kidneys (I) and intestine (J) of L-G6pc^−/− mice treated or not with GcgR antagonist L-168,049 (black bars) and WT mice (white bars) fasted for 6 h. Data were obtained 5 weeks after gene deletion (n = 6 mice per group) and are expressed as means ± SEM. Values significantly different from WT (*P < 0.05, **P < 0.01) and untreated L-G6pc^−/− mice ($P < 0.05, $$P < 0.01) are indicated. †, significantly different fed state (P < 0.01).

In the light of these findings, glucagon could be a major candidate responsible for the induced expression of gluconeogenic genes in the kidneys and intestine of L-G6pc^−/− mice. Glucagon receptors (GcgR) are known to be present in the liver, but GcgR are also expressed in the kidneys, notably in the proximal tubules and, to a lesser extent, in the intestine (26–28). We confirmed these data in L-G6pc^−/− mice where GcgR mRNA was substantially expressed in the kidneys (10 times less than in the liver, however) and more weakly in the intestine (about 500 times less than in the liver) of both control and L-G6pc^−/− mice (Fig. 5D). To date, the specific role of GcgR in these organs is still unexplained. However, two previous articles suggested a possible role of glucagon in the kidney cortex and intestine: 1) glucagon stimulated cAMP production in a kidney cortical cell line (29) and 2) glucagon infusion caused a net release of glucose into portal blood in dog (30). The GcgR is coupled to Gαs proteins, stimulating adenylate cyclase and increasing intracellular cAMP levels (26). This activates protein kinase A (PKA), which phosphorylates cAMP-response element binding protein (CREB). The phosphorylated form of CREB (P-CREB) induces the transcription of G6pc in the kidneys and intestine, via binding to the gene promoter (17). We then examined whether P-CREB could bind to the G6pc and Pck1 promoters by performing a ChIP assay. On the fed state, P-CREB was bound to the G6pc promoter in the kidneys of the fed L-G6pc^−/− mice, but not to that of the WT mice (Fig. 5E). P-CREB was bound neither to the G6pc promoter nor to the Pck1 promoter in the intestine of fed L-G6pc^−/− or WT mice (not shown). After 6 h fasting, P-CREB was bound to the renal G6pc promoter, but still not to the Pck1 promoter of L-G6pc^−/− mice (Fig. 5F). It is noteworthy that in the intestine of 6 h–fasted L-G6pc^−/− mice, P-CREB was bound to both G6pc and Pck1 promoters (Fig. 5G). To demonstrate unequivocally that the binding of P-CREB on G6pc promoter in the intestine and kidney of L-G6pc^−/− mice was dependent on the increase of plasmatic glucagon, we administered orally a glucagon receptor antagonist to L-G6pc^−/− mice (L-168,049, 50 mg/kg body) (18). L-168,049 inhibits the fixation of glucagon on GcgR and glucagon-stimulated cAMP synthesis in CHO cells expressing human GcgR (31). In the liver of L-G6pc^−/− mice, Pck1 mRNA expression was decreased by half 6 h after the administration of GcgR antagonist (Fig. 5H), demonstrating the efficiency of the suppression of glucagon signaling. In agreement with the role of glucagon in the induction of extrahepatic gluconeogenesis, the administration of the antagonist prevented the increase of the G6pc expression in both the kidneys and intestine of 6 h–fasted L-G6pc^−/− mice (Fig. 5I and J). These results strongly suggested that the increase of glucagon levels could account for the induction of G6pc expression in the kidneys and intestine of L-G6pc^−/− mice.

An intriguing observation, however, is that the G6pc gene was induced in the kidneys but not in the intestine in fed L-G6pc^−/− mice (Figs. 2 and 5). This might be dependent on differential sensitivity to glucagon in both organs, possibly related to the higher expression level of GcgR in the kidneys (see above). Although this reasoning may appear speculative, this phenomenon might to some extent be related to the fact that the kidney rapidly become predominant in participating in EGP during fasting (8), whereas the participation of the intestine is weaker and is induced later in the same situation (5). In agreement with this rationale, the binding of P-CREB to the renal G6pc promoter occurred from 6 h of fasting in WT mice (Fig. 5F). To increment our findings and confirm that the extrahepatic glucagon effects observed herein were not an adaptation specific to L-G6pc^−/− mice but could also take place in WT, we carried out a glucagon challenge in fed WT mice. After 30 min of glucagon administration, the plasma glucagon rose to the level of 6 h–fasted L-G6pc^−/− mice (Figs. 5A and 6A). Consistent with the results obtained in L-G6pc^−/− mice, P-CREB was firmly bound to the G6pc promoter in both tissues (Fig. 6B). Moreover, in response to glucagon injection, P-CREB bound to the Pck1 promoter only in the intestine (Fig. 6B), as already noted in L-G6pc^−/− mice (Fig. 5). In conclusion, these results show definitively for the first time that G6pc gene expression is regulated by glucagon not only in the liver, but also in the kidneys and intestine, under physiological conditions. Moreover, glucagon can also regulate Pck1 in the intestine, but not in the kidneys. It must be noted that the increased concentration of corticosterone might also play a role in the induction of extrahepatic gluconeogenesis. Glucocorticoids activate gluconeogenesis via the binding of glucocorticoid receptors (GRs) to glucocorticoid response element on both G6pc and Pck1 promoters. These regulations have been essentially documented in the liver up to now (for review see [32]). Using ChIP experiments, we showed that GRs were bound to G6pc and Pck1 promoters in both the kidney and intestine of 6 h–fasted L-G6pc^−/− and WT mice (Supplementary Fig. 2). The recruitment of GRs to Pck1 promoter was markedly increased in both the kidney (about threefold compared with WT) and intestine (sixfold compared with WT) of L-G6pc^−/− mice (Supplementary Fig. 2A and B). There was also a robust recruitment of GRs to the G6pc promoter in the intestine of L-G6pc^−/− mice compared with WT mice (about eightfold compared with WT; Supplementary Fig. 2B). These data suggested that glucocorticoids concurred in the induction of G6pc and Pck1 gene expression in the kidney and intestine of L-G6pc^−/− mice. This does not exclude that they could act in synergy with P-CREB to stimulate both G6pc and Pck1 gene transcription in the kidney and intestine, as it was observed previously in the liver (33,34).

FIG. 6.

Glucagon induces P-CREB binding to G6pc and Pck1 promoters in WT mice. A: Plasma glucagon of WT mice injected with saline solution (white bar) or glucagon (black bar). Values significantly different from saline solution (*P < 0.05) are indicated. B: P-CREB binding to G6pc or Pck1 promoter was analyzed by ChIP assay from the kidneys and intestine, 30 min after the injection of saline solution or glucagon in WT mice.

Role of metabolic acidosis in the adaptation of renal gluconeogenesis of L-G6pc^−/− mice to fasting.

In view of these results, another mechanism should be involved in the induction of PEPCK-c expression in the kidneys of 6 h–fasted L-G6pc^−/− mice. Despite a high glucagon-to-insulin ratio, Pck1 was not activated in the kidneys of fed L-G6pc^−/− mice. The transcription of Pck1 gene is known to be controlled by metabolic acidosis in the kidneys, whereas it does not respond to changes in pH in the liver (35,36). It is noteworthy that the plasmatic levels of free fatty acids, ketone bodies, and lactate were higher in L-G6pc^−/− mice than in control mice, after 6 h fasting (Table 1, Supplementary Fig. 1B). This could lead to a change in the acid-base status. Consistently, urinary pH of 6 h–fasted L-G6pc^−/− mice was more acidic compared with that of 6 h–fasted control mice (Fig. 7A). In the fed state, both L-G6pc^−/− and WT mice presented the same neutral urinary pH, ranging from 6.5 to 7.0. To test whether acidosis is the mechanism by which the Pck1 gene was induced in L-G6pc^−/− mice, we counteracted acidosis by the addition of 0.28M NaHCO₃ in the drinking water of L-G6pc^−/− mice for 3 days (37). The urinary pH of 6 h–fasted L-G6pc^−/− mice increased after treatment with NaHCO₃ and was not different from that of control mice (Fig. 7A). This led to a specific normalization of Pck1 expression at the mRNA and protein levels in the kidneys, but not in the intestine, of 6 h–fasted L-G6pc^−/− mice treated with NaHCO₃ (Fig. 7B–G). On the contrary, G6Pase expression was not affected in the kidneys and in the intestine after NaHCO₃ treatment (Fig. 7B–G), confirming that G6pc expression is not regulated by acidosis (35). These data also strongly suggested that Pck1 expression is not regulated by acidosis in the intestine.

FIG. 7.

Renal gluconeogenesis of L-G6pc^−/− mice is regulated by acidosis. A: Follow-up of urinary pH of L-G6pc^−/− mice (black bars), L-G6pc^−/− mice treated with 0.28 mol/L NaHCO₃ in drinking water (gray bars), and WT (L-G6pc^+/+ mice, white bars) on the fed or postabsorptive state. Values of pH were determined using strips with ΔpH = 0.2. B and C: Expression levels of mRNA encoding G6pc or Pck1 gene in the kidneys (B) or in the intestine (C) of 6 h–fasted mice. Results are expressed as a ratio relative to Rpl19 expression levels. D–G: Western blot quantification and enzyme activity assays of G6Pase and PEPCK determined in the kidneys (D and F) or in the intestine of 6 h–fasted mice (E and G). Data were obtained 5 weeks after gene deletion and are expressed as mean ± SEM. Values significantly different from WT (*P < 0.05; **P < 0.01), from the fed state ($$P < 0.01), and from L-G6pc^−/− without NaHCO₃ treatment (££P < 0.01).

Conclusion.

The liver has always been considered the major source of EGP until now. In contradiction with this dogma, we here report that, after a transient drop in plasma glucose as a result of incapacity to mobilize glycogen stores, the absence of hepatic glucose production has no major effect on the control of fasting plasma glucose. Instead, the early induction of gluconeogenesis in the kidneys and intestine occurs, permitting sustentation of EGP and blood glucose right from the start of fasting periods. This perfectly matches what has been observed during the anhepatic phase of liver transplantation in humans (38,39). Our data also emphasize that an essential function of the liver is the rapid tuning of blood glucose during nutritional transitions, via the handling of glycogen stores. However, the first major finding of this study is that the liver is not an irreplaceable source of endogenous glucose in the absence of food glucose. Similarly, glucagon, which is well known to stimulate hepatic glucose release via activation of glycogenolysis and gluconeogenic gene expression, has up to now been considered as only targeting the liver. In opposition to this other dogma, the second major finding here is that glucagon also plays a key role in the transcriptional regulation of renal and intestinal gluconeogenic genes. This may account for the basal induction of renal G6pc in the absence of hepatic glucose production and for the rapid induction of intestinal G6pc and Pck1 once food is lacking. Either fasting or type 2 diabetes is characterized by increased glucagon secretion in humans, which is believed to exert a key role in the augmented EGP of diabetes (40). In these two situations, the renal glucose production is increased (4,41–43). Moreover, the renal glucose production could play a crucial role in the counterregulation of insulin-induced hypoglycemia in humans, a situation of increased glucagon and cortisol secretions (41,42). At least, the important role of the kidney evidenced here might also explain why patients with renal failure are prone to hypoglycemia (44).

In conclusion, our study provides a definitive quantitative estimate of the capacity of extrahepatic gluconeogenesis to sustain fasting EGP, regardless of the contribution of the liver. It also extends the regulatory role of glucagon to the control of gluconeogenesis in the kidneys and intestine. This leads us to conclude that the current dogma relating to the relative role of the liver vis-à-vis extrahepatic gluconeogenic organs in glucose homeostasis should be reconsidered.