Hepatic de Novo Synthesis of Glucose 6-Phosphate Is Not Affected in Peroxisome Proliferator-activated Receptor α-Deficient Mice but Is Preferentially Directed toward Hepatic Glycogen Stores after a Short Term Fast*

Apart from impaired β-oxidation, Pparα-deficient (Pparα–/–) mice suffer from hypoglycemia during prolonged fasting, suggesting alterations in hepatic glucose metabolism. We compared hepatic glucose metabolism in vivo in wild type (WT) and Pparα–/– mice after a short term fast, applying novel isotopic methods. After a 9-h fast, mice were infused with [U-13C]glucose, [2-13C]glycerol, [1-2H]galactose, and paracetamol for 6 h, and blood and urine was collected in timed intervals. Plasma glucose concentrations remained constant and were not different between the groups. Hepatic glycogen content was 69 ± 11 and 90 ± 31 μmol/g liver after 15 h of fasting in WT and Pparα–/– mice, respectively. The gluconeogenic flux toward glucose 6-phosphate was not different between the groups (i.e. 157 ± 9 and 153 ± 9 μmol/kg/min in WT and Pparα–/– mice, respectively). The gluconeogenic flux toward plasma glucose, however, was decreased in PPARα–/– mice (i.e. 142 ± 9 versus 124 ± 13 μmol/kg/min) (p < 0.05), accounting for the observed decrease (–15%) in hepatic glucose production in Pparα–/– mice. Expression of the gene encoding glucose-6-phosphate hydrolase (G6ph) was lower in the PPARα–/– mice compared with WT mice. In conclusion, Pparα–/– mice were able to maintain a normal total gluconeogenic flux to glucose 6-phosphate during moderate fasting, despite their inability to up-regulate β-oxidation. However, this gluconeogenic flux was directed more toward glycogen, leading to a decreased hepatic glucose output. This was associated with a down-regulation of the expression of G6ph in PPARα-deficient mice.

Fuel selection to meet the body's energy demand is of crucial importance during feeding-fasting transitions. The liver plays a central role in this switch. It changes from glucose uptake and glycogen synthesis during feeding to glucose production by gluconeogenesis (GNG) 1 and glycogenolysis during fasting. The origin of hepatic glucose production (HGP) shifts from mainly glycogenolysis to GNG as fasting prolongs. These changes in hepatic glucose metabolism are accompanied by adaptation of hepatic fatty acid metabolism (i.e. from fatty acid synthesis to fatty acid oxidation). This adaptation allows the optimization of fuel substrate utilization. These metabolic changes are, at least in part, effected by the reciprocal action of insulin and glucagon. However, the discovery of nuclear hormone receptors and the (partial) elucidation of their mode of action as ligandactivated regulators of gene expression has considerably complicated the picture. One member of the nuclear hormone receptors is of particular importance in mediating the adaptive response to fasting (i.e. peroxisome proliferator-activated receptor ␣ (PPAR␣)). PPAR␣ is a fatty acid-activated transcription factor that up-regulates the expression of a variety of genes that encode proteins involved in ␤-oxidation and lipoprotein metabolism (1). Lack of this receptor in Ppar␣ Ϫ/Ϫ mice results in the inability to up-regulate hepatic fatty acid oxidation and ketogenesis upon fasting in the face of increased concentrations of free fatty acids in the circulation (2). It also became apparent that mice lacking PPAR␣ develop hypoglycemia after a prolonged fast (2).
The etiology of hypoglycemia in fasting Ppar␣ Ϫ/Ϫ mice is still unclear. It has been hypothesized that it reflects decreased GNG secondary to impaired hepatic fatty acid ␤-oxidation (2). Surprisingly, a recent study suggested increased HGP in long term (24-h) fasted Ppar␣ Ϫ/Ϫ mice despite the development of hypoglycemia (3). Furthermore, glycogen content of the liver in Ppar␣ Ϫ/Ϫ mice after a prolonged fast was not reduced in comparison with wild type (WT) mice but, unexpectedly, tended to be higher in the Ppar␣ Ϫ/Ϫ mice. In contrast, glycogen content of the liver did not increase in Ppar␣ Ϫ/Ϫ mice upon refeeding, whereas in WT mice, glycogen content strongly increased. These data suggest that the balance between HGP, glycogen synthesis, and GNG at the level of glucose 6-phosphate (G6P) is perturbed in Ppar␣ Ϫ/Ϫ mice.
Partitioning of G6P can be studied in vivo using the glycoconjugate probe technique and mass isotopomer distribution analysis (MIDA) as described by Hellerstein and co-workers (4). Recently, we applied these stable isotope techniques to study partitioning of newly synthesized G6P while glucose-6phosphatase activity was partially inhibited in 24-h fasted rats (5). We were able to show that when HGP was diminished by partial inhibition of glucose-6-phosphatase activity, quite surprisingly, de novo synthesis of G6P was unaffected. It appeared that newly synthesized G6P was repartitioned away from plasma glucose to glycogen synthesis, and, as a consequence, the glycogen content of livers of treated rats increased severalfold. The importance of these observations, substantiating data from other groups (6), is that partitioning of G6P independent of its de novo synthesis represents an additional mechanism of regulation of hepatic glucose production. We now have miniaturized these methods for application in mice (7). In the present study, we addressed the following questions. 1) Is there a role of PPAR␣ in the control of de novo synthesis of G6P? 2) What are the effects of PPAR␣ deficiency on partitioning of newly synthesized G6P during fasting? We approached these questions experimentally in short term fasted Ppar␣ Ϫ/Ϫ mice by infusion of [U-13 C]glucose, [2-13 C]glycerol, [1-2 H]-galactose, and paracetamol, collecting serial blood and urine spots on filter paper, and measuring the mass isotopomer distribution in glucose and paracetamol-glucuronide. The fluxes were subsequently compared with expression of genes encoding enzymes involved in hepatic glucose metabolism and fatty acid oxidation, enabling us to delineate functional consequences of PPAR␣ deficiency-induced changes in gene expression.

EXPERIMENTAL PROCEDURES
Animals-Male Ppar␣ Ϫ/Ϫ mice and WT mice on a SV129 background were housed in a temperature-controlled (21°C) room on a 10-h dark, 14-h light cycle. Experimental procedures were approved by the Ethics Committee for Animal Experiments of the State University Groningen. Mice were equipped with a permanent heart catheter that was attached to the skull with acrylic glue (8). Mice were allowed to recover from surgery for at least 4 days.
Fasting Experiments-The adaptive response to fasting in Ppar␣ Ϫ/Ϫ and WT mice was compared first. Mice were fasted up to 24 h, and blood samples were taken at t ϭ 0 (fed), 15, and 24 h by tail bleeding, and livers were removed at t ϭ 0, 15, and 24 h for lipid analysis and RNA isolation. Hepatic in vivo carbohydrate metabolism was measured according to the protocol described by Van Dijk et al. (7). On the day of the experiment, mice were placed in individual metabolic cages. Filter paper was placed under the wired floor of the cage to collect urine samples. Food was removed 9 h before the start of the experiments. Body weight was 25.0 Ϯ 1.9 g for WT mice and 24.1 Ϯ 1.1 g for Ppar␣ Ϫ/Ϫ mice at the time of the experiments. Mice received an infusion at a rate of 0.6 ml h Ϫ1 during 6 h of a solution consisting of [U-13 C]glucose (13 mol ml Ϫ1 ), [2-13 C]glycerol (160 mol ml Ϫ1 ), [1-2 H]galactose (33 mol ml Ϫ1 ), and paracetamol (1 mg ml Ϫ1 ). Blood glucose during the experiment was measured using EuroFlash™ test strips (LifeScan Benelux, Beerse, Belgium), and bloodspots obtained by tail bleeding for gas chromatography-mass spectrometry (GC-MS) measurements were collected before the start of the infusion and hourly afterward until 6 h after the start of the infusion. The filter paper placed under the wired floor of the cage was replaced at hourly intervals. A large blood sample was taken by heart puncture under halothane anesthesia at the end of the experiment, and the liver was quickly excised and frozen immediately in liquid N 2 for lipid analysis and RNA isolation.
Metabolite Concentrations and Enzyme Activities-Plasma was isolated by centrifugation, liver tissue was homogenized, and lipids were extracted using a modified Bligh & Dyer method (9). Plasma 3-hydroxybutyrate, lactate, free fatty acid, and liver triglyceride content were determined using commercially available kits (Roche Applied Science and Wako Chemicals GmbH, Neuss, Germany). Alanine concentrations were analyzed by ion exchange column chromatography followed by postcolumn ninhydrine derivatization (10, 11) on a Biochrome 20® automated amino acid analyzer (Amersham Biosciences). Glycerol concentrations were measured by GC-MS using [1,1,2,3,3-d 5 ]glycerol as an internal standard. Samples were derivatized to their triacetate by adding 100 l of pyridine and 200 l of acetic acid-anhydride to 50 l of plasma and incubating the solution at 80°C for 30 min. Samples were measured on a Trace-GC-MS (Finnigan Matt, San Jose, CA). Plasma insulin levels were determined by radioimmunoassay (RI-13K; Linco Research, St. Charles, MO). Total hepatic protein content was determined according to Lowry et al. (12). Hepatic glycogen content was determined in freeze clamped liver tissue after extraction in 1 M KOH solution by sonication. The extract was incubated for 30 min at 90°C, cooled, and brought to pH 4.5 by the addition of 3 M acetic acid. Precipitated protein was removed by centrifugation. Glycogen was converted to glucose by treating the samples with amyloglucosidase, followed by assay of glucose at pH 7.4 with ATP, NADP ϩ , hexokinase, and G6P dehydrogenase (13). Liver samples for the determination of G6P were treated by sonification in a 5% (w/v) HClO 4 solution and centrifuged, and supernatant was neutralized to pH 7 by the addition of small amounts of a mixture of 2 M KOH and 0.3 M MOPS. G6P was determined fluorimetrically with NADP ϩ and G6P dehydrogenase (13). Hepatic ATP levels were measured using a bioluminescence assay kit (Roche Applied Science).
Hepatic mRNA Levels-Total RNA was isolated from liver tissue using the Trizol method (Invitrogen). RNA was converted to cDNA with M-Mulv-RT (Roche Applied Science) according to the manufacturer's protocol using random primers. cDNA levels of the genes of interest were measured by real time PCR using the ABI Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). An amount of cDNA corresponding to 20 ng of total RNA was amplified using the qPCR core kit (Eurogentec, Seraing, Belgium) according to the manufacturer's protocol using the appropriate forward and reverse primers (Invitrogen) and a template-specific 3Ј-carboxy-N,N,NЈ,NЈ-tetramethylrhodamine-, 5Ј-carboxyfluorescein-labeled Double Dye Oligonucleotide probe (Eurogentec). Calibration curves were run on serial dilutions of pooled cDNA solutions as used in the assay. The data were processed using the ABI Sequence Detector 6.3 system (Applied Biosystems). Quantified expression levels were within the linear part of the calibration curves. PCR results were normalized to ␤-actin mRNA levels. The sequences of the primers and probes used in this study are listed in Table I.
Measurement of Mass Isotope Distribution by GC-MS-Analytical procedures for extraction of glucose and paracetamol-glucuronide from bloodspot and urine filter paper strips, respectively, derivatization of the extracted compounds, and GC-MS measurements of derivatives were essentially according to Van Dijk et al. (5,7). In short, glucose was extracted by incubating a disk (6.5 mm) punched out of a bloodspot with ethanol/water (10/1, v/v) mixture. After drying the sample under a stream of N 2 , glucose was derivatized to its pentaacetate-ester and aldonitrile pentaacetate-ester. The final derivatives were dissolved in 100 l of ethyl acetate for injection. Paracetamol-glucorionide (Par-GlcUA) was extracted from filter paper with methanol/water (3/1, v/v) and subsequently isolated by a Milton Roy high pressure liquid chromatography system (Interscience, Breda, The Netherlands) on a Nucleosil 7C18 SP250/10 column (Bester, Amstelveen, The Netherlands) eluted with a gradient of 0.2% (v/v) ammonium formate in water (pH 4.8) and 40% (v/v) acetonitrile in water. The fraction containing the isolated compound was dried under a stream of N 2 and subsequently derivatized to its trimethylsilyl-ethyl-ester or oxidized to saccharic acid by nitrite in nitric acid and derivatized to its tetraacetate-diethyl-ester. After drying of the samples under a stream of N 2 , the dry residues were dissolved in 200 l of ethyl acetate for injection. All samples were analyzed by GC-MS (SSQ7000; ThermoFinnigan, San Jose, CA) on an AT-5MS 30 m ϫ 0.25-mm inner diameter (0.25-m film thickness) capillary column (Alltech, Breda, The Netherlands). For all calculations of mass isotopomer distribution, Excaliber software (Ther-moFinnigan, San Jose, CA) was used. Mass spectrometric analyses of glucose pentaacetate, glucose aldonitrile pentaacetate, and saccharic acid diethyl-, tetraacetate-ester were performed by positive ion chemical ionization with methane. Ions monitored for glucose pentaacetate were m/z 331-337, ions for glucose aldonitrile pentaacetate were m/z 328 -334, and ions for saccharic acid diethyl-ester tetraacetate were m/z 375-381, all corresponding to the m 0 -m 6 mass isotopomers. Mass spectrometric analyses for Par-GlcUA ethyl-, tetra-(trimethylsilyl)ester were performed by electron impact ionization. The ions monitored were m/z 331-337, corresponding to the m 0 -m 6 mass isotopomers. Series of measurements were composed of experimental samples, control samples, and a dilution series obtained from a mixture of the last, most enriched, samples taken at the end of an experiment.
A series of measurements was accepted for further calculations when two conditions were met as described by Van Dijk et al. (7). First, for each derivative, the coefficient of variance of the fractional contribution of m 0 , m 1 , and m 2 to total ion abundance in control samples must be smaller than 1% for m 0 and 2% for m 1 and m 2 . Second, for each derivative, the fractional contribution of m 1 , m 2 , and m 6 to total ion abundance measured in experimental samples must be within the range of constant response of the GC-MS as estimated from the values of the fractional contribution of m 1 , m 2 , and m 6 to total ion abundance of the inserted dilution series.
MIDA-The measured fractional isotopomer distribution by GC-MS (m 0 -m 6 ) was corrected for the fractional distribution due to natural abundance of 13 C. This was done by multiple linear regression as described by Lee et al. (14) to obtain the excess fractional distribution of mass isotopomers (M 0 -M 6 ) due to incorporation of infused labeled compounds (i.e. [   Ra(UDPglc;whole body) was calculated with the assumption of a constant and complete entry of infused galactose into the hepatic UDPglucose pool. Furthermore, it was assumed that the fractional isotopomer distribution observed in urinary Par-GlcUA reflects the fractional isotopomer distribution in hepatic UDPglc.
Rates of endogenous plasma glucose (Ra(glc;endo)) and UDP-glucose (Ra(UDPglc;endo)) appearance were calculated as follows. Fractional gluconeogenic contribution to both plasma glucose (f(glc) and hepatic UDP-glucose (f(UDPglc)) were calculated using MIDA of glucose and Par-GlcUA derivatives, respectively, as described in detail elsewhere (see Refs. 16 and 17). The absolute gluconeogenic flux into both plasma glucose (GNG(glc) and hepatic UDP-glucose (GNG(UDPglc)) were calculated as follows. Statistical Analysis-All values reported are mean Ϯ S.D. Significance for metabolite and activity levels was determined using the nonparametric Mann Whitney test for unpaired data. Significance for fluxes calculated over time was determined using multiple measurement analysis of variance. Differences were considered significant at p Ͻ 0.05. Table II the effects of fasting are shown in Ppar␣ Ϫ/Ϫ and WT mice of various blood-borne compounds. Glucose concentration tended to be lower at 15 h of fasting but became significantly lower after 24 h of fasting in Ppar␣ Ϫ/Ϫ mice compared with WT mice. No differences were observed between WT and Ppar␣ Ϫ/Ϫ mice with respect to plasma lactate concentration; lactate concentrations decreased to a similar extent in both groups of mice. Lactate/pyruvate ratios after 24 h fasting were 21.4 Ϯ 4.3 and 22.3 Ϯ 5.0 in WT and Ppar␣ Ϫ/Ϫ mice, respectively. Free fatty acid concentrations were significantly higher at 15 h of fasting in Ppar␣ Ϫ/Ϫ mice and remained elevated up to 24 h of fasting. In contrast, the ketotic fasting response was diminished in Ppar␣ Ϫ/Ϫ mice. After an initial rise in the concentration of 3␤-hydroxybutyrate in both groups of mice, the 3␤hydroxybutyrate concentration increased further in WT but not in Ppar␣ Ϫ/Ϫ mice. The changes in plasma glycerol concentration did not differ between Ppar␣ Ϫ/Ϫ or WT mice. After 24 h of fasting, a decrease in glycerol concentration was observed in both groups of mice. Alanine concentrations were 393 Ϯ 61 mol/liter in WT mice and 212 Ϯ 27 mol/liter (p Ͻ 0.05) in the Ppar␣ Ϫ/Ϫ mice at the end of the fasting period.

The Adaptive Response of Ppar␣ Ϫ/Ϫ Mice to Fasting-In
In Fig. 1 shows the changes in hepatic glycogen and triglyceride contents in Ppar␣ Ϫ/Ϫ and WT mice upon fasting. Liver weight was not different between groups (i.e. 0.9 Ϯ 0.1 and 1.0 Ϯ 0.1 g in WT and Ppar␣ Ϫ/Ϫ mice, respectively). When fed, hepatic glycogen content was significantly lower in Ppar␣ Ϫ/Ϫ mice than in WT mice. Upon fasting, glycogen decreased more strongly in WT than in knock-out mice. After 15 h of fasting, hepatic glycogen content in Ppar␣ Ϫ/Ϫ mice was 90 Ϯ 28 mol/g liver, compared with 69 Ϯ 10 mol/g liver in WT mice. This difference in response of the hepatic glycogen content was apparent also after 24 h of fasting (i.e. 9 Ϯ 9 mol/g liver in WT mice and 36 Ϯ 10 mol/g liver in Ppar␣ Ϫ/Ϫ mice). Hepatic G6P levels, determined after 15 h of fasting, were not significantly different (i.e. 322 Ϯ 57 and 397 Ϯ 51 mol/g liver in the WT and Ppar␣ Ϫ/Ϫ mice, respectively). During fasting, hepatic triglyceride content increased in both group of mice. However, in Ppar␣ Ϫ/Ϫ mice, the increase in hepatic triglyceride content was significantly more pronounced than in WT mice .
Hepatic Glucose Metabolism in PPAR␣ Ϫ/Ϫ Mice-In Fig. 2, the time courses are shown of plasma glucose concentration and endogenous glucose production rates during the last 3 h of the label infusion experiment. The end of the experiment corresponds to 15 h of fasting. Mice were fasted for 9 h before the label infusion started. Plasma insulin concentrations showed no difference between both groups at the end of the experiment (i.e. 0.23 Ϯ 0.04 ng/ml in WT mice and 0.24 Ϯ 0.02 ng/ml in Ppar␣ Ϫ/Ϫ mice). These values were not different from those measured after 24 h of fasting (i.e. 0.25 Ϯ 0.05 and 0.25 Ϯ 0.03 ng/ml in WT and Ppar␣ Ϫ/Ϫ mice, respectively). As is clear from Fig. 2A, plasma glucose concentrations remained constant and the values were very similar in both groups of animals. Glucose concentration was 8.0 Ϯ 0.4 and 7.7 Ϯ 1.5 mM in WT and Ppar␣ Ϫ/Ϫ mice, respectively. During the last 3 h of the experiment, a constant endogenous glucose production could be documented. The data clearly show a significantly decreased endogenous glucose production in Ppar␣ Ϫ/Ϫ mice (129 Ϯ 13 mol kg Ϫ1 min Ϫ1 ) in comparison with WT mice (149 Ϯ 11 mol kg Ϫ1 min Ϫ1 , p Ͻ 0.05).  Fig. 3, the rates of de novo synthesis of G6P into plasma glucose (Fig. 3A), into UDP-glucose (Fig. 3B) and the total de novo synthesis of G6P (Fig. 3C) are shown during the final 3 h of the label infusion experiment. De novo synthesis of G6P into plasma glucose was constant during the experiment. In Ppar␣ Ϫ/Ϫ mice, the absolute rate of GNG toward plasma glu-cose was significantly diminished compared with WT mice (i.e. 124 Ϯ 13 versus 142 Ϯ 9 mol kg Ϫ1 min Ϫ1 (p Ͻ 0.05), respectively). A different observation was made with regard to the absolute rate of appearance of newly synthesized G6P into the UDP-glucose pool. In the course of the experiment, a slow but significant decline was observed in the rate of appearance of newly formed UDP-glucose. In Ppar␣ Ϫ/Ϫ mice, the decline, from 70 Ϯ 4 mol kg Ϫ1 min Ϫ1 at 3 h to 60 Ϯ 5 mol kg Ϫ1 min Ϫ1 at 6 h after the start of the label infusion (p Ͻ 0.05), was less pronounced than in WT mice, in which it declined from 65 Ϯ 11 mol kg Ϫ1 min Ϫ1 at 3 h to 42 Ϯ 5 mol kg Ϫ1 min Ϫ1 at 6 h after the start of the label infusion (p Ͻ 0.05). At the end of the experiment, the rate of appearance of newly formed UDPglucose in Ppar␣ Ϫ/Ϫ mice was therefore significantly higher than in WT mice. Interestingly, the total rate of de novo synthesis of G6P remained constant during the second half of the experiment and was not different between the two groups (i.e. 153 Ϯ 9 and 157 Ϯ 9 mol kg Ϫ1 min Ϫ1 in Ppar␣ Ϫ/Ϫ and WT mice, respectively).
In Fig. 4, the partitioning of newly synthesized G6P into plasma glucose and UDP-glucose is given at 6 h after the start of the label infusion. As is clear from this figure, there was a significantly larger part of newly synthesized G6P diverted to UDP-glucose in Ppar␣ Ϫ/Ϫ mice than in WT mice. At the end of the experiment, the fractional contribution of newly formed G6P to UDP-glucose synthesis was 0.40 Ϯ 0.04 in Ppar␣ Ϫ/Ϫ mice and 0.29 Ϯ 0.03 in WT mice, a significant increase of 32% in Ppar␣ Ϫ/Ϫ mice as compared with WT mice.
Hepatic Gene Expression Profiles-Hepatic expression of genes involved in glucose and fat metabolism in fed and 15-and 24-h fasted mice are shown in Fig. 5. The expected response was observed in expression of genes involved in hepatic glucose and fatty acid oxidation during fasting in WT mice. Expression of Pepck, G6ph, G6pt, Gp, Cpt1a, Mcad, and Hmgs all increased upon fasting. Pepck expression was similarly increased after the 15-h fast compared with the fed situation in both groups but significantly less in the Ppar␣ Ϫ/Ϫ than in WT mice after a 24-h fast. Only the expressions of G6ph and Gs were significantly affected at 15 h of fasting in Ppar␣ Ϫ/Ϫ mice compared with WT mice. Hepatic expression of the other genes involved in glucose metabolism (i.e. G6pt, Gp, Gk, and the transcription factor Chrebp) were not different in Ppar␣ Ϫ/Ϫ mice when compared with WT mice after 15 h of fasting. Expression of genes involved in fatty acid oxidation were differently affected. Expression of Cpt1a did not differ between Ppar␣ Ϫ/Ϫ and WT mice at 15 h of fasting, but expression of Mcad and particularly of Hmgs was significantly decreased in Ppar␣ Ϫ/Ϫ mice when compared with WT mice. The hepatic expression of the Ppar␥ gene was increased under all circumstances tested in Ppar␣ Ϫ/Ϫ mice in comparison with WT mice. Thus, after a 24-h fast, expression of most genes studied was decreased compared with the fed situation and more severely so in Ppar Ϫ/Ϫ mice. This raised the question of whether this effect of 24-h fasting could be aspecific and simply due to a severe energy shortage in livers of these animals. To try to clarify this, we measured hepatic ATP levels. Hepatic ATP levels showed a trend, albeit not significant, to a decrease after 24 h of fasting in the Ppar␣ Ϫ/Ϫ mice compared with WT mice (i.e. 0.56 Ϯ 0.13 versus 0.93 Ϯ 0.39 nmol of ATP/mg of liver).

DISCUSSION
In the present study, we addressed the role of PPAR␣ in the control of hepatic glucose metabolism (i.e. de novo synthesis of G6P and its partitioning). The results show that after a short term fast of 15 h, de novo synthesis of G6P was not affected by PPAR␣ deficiency. However, newly synthesized G6P was partitioned away from plasma glucose to glycogen synthesis. Furthermore, deficiency of PPAR␣ resulted in a reduced hepatic expression of G6ph, Gs, and, to a lesser extent, Gp.
The validity of the isotope model, with the application of glycoconjugates and the MIDA approach, has been substantiated in various studies, although some controversy still re-mains (18). We have validated the application of MIDA in mice in a separate study (7). In that study, we observed major adaptations in whole body and hepatic glucose metabolism in 24-h fasted mice but not in 9-h fasted mice during the course of a 6-h infusion of stable isotopically labeled compounds.
The alterations in the adaptive response to fasting of Ppar␣ Ϫ/Ϫ mice that we observed were in line with those reported by other investigators (2,3). The hypoglycemia in Ppar␣ Ϫ/Ϫ mice did not appear to be due to an enhanced glucose consumption by peripheral tissues (i.e. metabolic clearance of glucose was similar in WT and Ppar␣ Ϫ/Ϫ mice). Differences exist with respect to the reported time course of development of hypoglycemia in fasted Ppar Ϫ/Ϫ mice. Kersten et al. (2) and Xu et al. (3) reported significantly lower plasma glucose concentrations after, respectively, 15 and 17 h of fasting in Ppar␣ Ϫ/Ϫ mice compared with WT mice. In our hands, blood glucose concentrations decreased to a larger extent in WT mice than reported by Kersten et al. (2). As yet, no explanation can be given for this discrepancy.
During infusion of stable isotopically labeled compounds, no decrease was observed in plasma glucose concentration of either Ppar␣ Ϫ/Ϫ or WT mice. This was different from the observations made in noninfused mice during the adaptive response. Differences in plasma insulin concentrations do not offer an explanation; they were low as would have been expected for fasting mice and did not differ from the values observed in 24-h fasted, noninfused WT and Ppar␣ Ϫ/Ϫ mice. In our opinion, the absence of a decrease in plasma glucose concentration upon prolonged fasting in infused mice might have been due to the delivery of gluconeogenic substrates by the infusion of stable isotopically labeled compounds. The rates of infusion were considerable, particularly that of [2-13 C]glycerol at ϳ60 mol kg Ϫ1 min Ϫ1 . Whereas all of the infused glycerol would have been utilized for GNG to plasma glucose, however, only 15-20% of the amount of glucose produced through GNG can be accounted for by the glycerol infusion. Furthermore, Previs et al. (18) showed that infusion of this amount of glycerol did not increase endogenous glucose production in 30-h fasted BALBc mice. Their experiments, however, consisted of short term infusion (3 h) of labeled glycerol instead of 6 h as in our experiments and were performed under conditions in which blood glucose concentrations were normal. On the other hand, Xu et al. (3) concluded from their experiments that in Ppar␣ Ϫ/Ϫ mice glycerol appeared to be the preferred gluconeogenic substrate at the expense of lactate. Our results imply that de novo synthesis of G6P exhibits a high elasticity toward the supply of gluconeogenic substrates.
PPAR␣ deficiency resulted in a lower endogenous rate of appearance of glucose than observed in WT mice. Whereas GNG would have been calculated based on its fractional contribution to plasma glucose alone, our results would have led us to infer that total gluconeogenic flux was inhibited in parallel with inhibition of glucose production. By analyzing both plasma glucose and urinary paracetamol-glucoronic acid, however, we were able to show that the decrease in hepatic glucose production was not associated with a decrease in the rate of de novo synthesis of G6P but with a more predominant partitioning of newly synthesized G6P into glycogen. In line with this observation, glycogen content of liver of Ppar␣ Ϫ/Ϫ mice remained higher during fasting when compared with WT mice. The metabolic fate of newly synthesized G6P, therefore, seems to be without consequence for the rate of its de novo synthesis. Similar lack of feedback inhibition of de novo synthesis of G6P by its product has been documented by us (5) as well as by others (6). In contrast to our conclusions, Xu et al. mice. However, they did not study partitioning of newly synthesized G6P. It is therefore not clear whether the reported increase in glucose production resided in an increased rate of de novo synthesis of G6P or in an altered partitioning of newly synthesized G6P. Another potentially important aspect is that these authors performed their experiments in Ppar␣ Ϫ/Ϫ mice bred on to a C57Bl/6 background, whereas we studied Ppar␣ Ϫ/Ϫ mice bred on to a SV129 background.
Our study brings into focus another mechanism by means of which impairment of fatty acid oxidation influences GNG. It is widely accepted that inhibition of ␤-oxidation impairs GNG by diminished delivery of energy and/or reducing equivalents. However, early data already indicated that inhibiting ␤-oxidation elicited different effects on GNG, depending on the substrate available. When livers of fasted guinea pigs were perfused with glycerol or propionate as gluconeogenic substrates, inhibition of fatty acid oxidation by CPT I blockade did not perturb GNG from these substrates (19). Only when lactate and pyruvate were used as substrate, did inhibition of fatty acid oxidation at CPT I result in inhibition of GNG. A similar mechanism might have been active in the experiments of Xu et al.
(3) on substrate selection for gluconeogenesis in Ppar␣ Ϫ/Ϫ mice. The data presented in this study show that, in addition, impairment of ␤-oxidation of fatty acids can affect partitioning of newly synthesized G6P between plasma glucose and glycogen without affecting the rate of de novo synthesis of G6P. As discussed earlier by us (5) and others (6), the response of the hepatocyte to maintain homeostasis of intracellular G6P might be central in this partitioning.
Our experiments allow us only to speculate how PPAR␣ deficiency brings about the preferential partitioning of newly synthesized G6P to glycogen. The mRNA levels of the G6ph gene were decreased, whereas expression of the G6pt gene was unaffected in the knockout mice. Gene expression of Pepck was not affected in short term fasted Ppar␣ Ϫ/Ϫ mice. This would indicate that reduced G6ph gene expression might be a way by which PPAR␣ influences endogenous glucose production. In the promotors of either gene, no PPREs have been reported so far. The effects of PPAR␣ might therefore be indirect. Changes in intracellular concentration of long chain acyl-CoA thioesters, ligands for NHF4␣, might be of importance. In the G6ph promotor, a region has been identified that binds HNF4␣ (20). Furthermore, binding of fatty acyl-CoAs to HNF4␣ modulates its DNA binding activity and thereby the expression of the G6ph gene (21). We should realize, however, that expression of the genes involved in glycogen metabolism (i.e. Gs and Gp) were also depressed, whereas glycogen metabolism measured by isotopic means appeared to be enhanced.
In conclusion, PPAR␣ deficiency results in partitioning of newly synthesized G6P away from plasma glucose, resulting in a decreased endogenous glucose production during short term fasting. Decreased expression of G6ph might have been a mechanism by which this effect of PPAR␣ deficiency is mediated.