Acute Inhibition of Glucose-6-Phosphate Translocator Activity Leads to Increased De Novo Lipogenesis and Development of Hepatic Steatosis Without Affecting VLDL Production in Rats

Male Wistar rats (Harlan Laboratories, Zeist, the Netherlands), weighing between 290 and 350 g (means ± SD: 318 ± 25 g), were used to study the in vivo effects of S4048. To allow infusion and blood collection in freely moving, unrestrained animals, rats were equipped with two permanent heart catheters via the right jugular vein as described by Kuipers et al. (23). After surgery, animals recovered for a period of 7 days in individual cages in a temperature-controlled room (20°C) with food and water available ad libitum. At 8 h before the start of the experiment, the cages were cleaned and the food was removed.

Animals received an intravenous infusion of the G6P translocator inhibitor (S4048, 30 mg · kg^–1 · h^–1, infusion rate 3 ml/h) or of the solvent (phosphate-buffered saline [PBS] with DMSO) for 8 h. Rats were allowed to move freely throughout the experiment, and the animals did not show signs of stress. The S4048 compound was synthesized by the chemical department of Aventis Pharma (Frankfurt, Germany) (21). At the infusion rate used, S4048 is expected to partially inhibit G6P translocase activity (21). Simultaneously, all animals received an infusion of [1-¹³C]acetate (0.8 mmol · kg^–1 · h^–1). The infusates were freshly prepared before each experiment. S4048 was dissolved in 20% DMSO in PBS (vol/vol), and PBS was added to reach a final concentration of 6.1% DMSO (vol/vol). The solution was then immediately adjusted to pH 7.4. Before the infusion, a basal blood sample (∼300 μl) was taken to determine the baseline values of the metabolites studied. During the infusion, blood samples (∼200 μl) were taken after 30, 60, 120, 240, 360, 420, and 480 min. Samples were heparinized and immediately placed on ice and centrifuged at 5,000 rpm for 10 min at 4°C. The plasma was stored at −20°C until analysis. Despite the relatively high infusion rate, hematocrit levels did not fall significantly during the experiment. At the end of the infusion period, animals were anesthetized with sodium-pentobarbital, and a large blood sample was taken by heart puncture. The abdomen was opened, and the liver was rapidly exposed, excised, and stored in parts at −80°C for measurement of G6P and glycogen content, lipid analysis, and RNA isolation or rapidly frozen in liquid isopentane for histological analysis (see below).

The effects of S4048 on hepatic VLDL production were studied in a separate experiment. After the surgical procedures and recovery as described above, the rats were infused with S4048 or its solvent and, after 3 h of infusion, received an intravenous Triton WR 1339 injection (Tyloxapol; Sigma Chemical, St. Louis, MO) as a (12% wt/wt) solution dissolved in saline in a dose of 5 ml/kg body wt. Triton WR 1339 blocks lipolysis of lipoproteins, which accumulate over time in plasma, allowing for the calculation of hepatic VLDL-triglyceride production rates (24). To exclude any interference of the solvent containing DMSO on VLDL-triglyceride secretion, a separate group of rats received a saline infusion. After Triton WR 1339 was injected, blood samples were taken after 30, 60, 120, and 180 min for measurement of triglyceride concentrations. VLDL production rates were calculated from the slope of the linear triglyceride accumulation curves in time. After 180 min, a large blood sample was taken for isolation of VLDL/LDL particles, and the animals were killed. Lipoproteins were isolated according to Pietzsch et al. (25) using a solution of 15.3% NaCl and 35.4% KBr in saline with a density <1.019 g/ml. Plasma (0.5 ml) was overlayed with 0.6 ml of the NaCl-KBr solution, centrifuged for 100 min at 120,000 rpm and 4°C in a Beckman Optima TLX Ultracentrifuge (Beckman Instruments, Palo Alto, CA), and the VLDL fraction was collected by tube slicing and was frozen until analysis. VLDL particle size was calculated as described by Beil et al. (26).

Hepatic lipids were extracted using the method of Bligh and Dyer (27). To determine plasma and hepatic triglyceride and cholesterol concentrations, assay kits were obtained from Hoffmann-La Roche (Basel). To determine plasma and hepatic phospholipid and plasma free fatty acid (FFA) concentrations, assay kits were obtained from Waco Chemical (Marburg, Germany). Plasma β-hydroxybutyrate was measured using a commercially available kit from Sigma. Total protein content of tissue homogenates was determined according to Lowry et al. (28). Plasma insulin was determined by radioimmunoassay (RI-13K; Linco Research, St. Charles, MO). Plasma glucose concentration was determined enzymatically by use of the Beckman glucose analyzer II. After extraction with a 1-mol/l KOH solution, hepatic glycogen was determined by sonication. The extract was incubated for 30 min at 90°C, cooled, and then brought to pH 4.5 by adding 3 mol/l acetic acid. Precipitated protein was removed by rapid centrifugation (10,000 rpm for 1 min). Glycogen was converted to glucose by treating the samples with amyloglucosidase, followed by assay of glucose at pH 7.4 with ATP, NADP⁺, hexokinase, and G6P dehydrogenase (29). For the determination of G6P, liver samples were treated by sonification in a 5% (wt/vol) HClO₄ solution. Precipitated protein was removed by rapid centrifugation at 10,000 rpm for 1 min in a cold microcentrifuge, and the supernatant was neutralized to pH 7.0 by adding small amounts of a mixture of 2 mol/l KOH and 0.3 mol/l MOPS. G6P was determined fluorimetrically with NADP⁺ and G6P dehydrogenase (30). Activites of liver enzymes, i.e., alanine aminotransferase (ALAT) and aspartate aminotransferase (ASAT) were determined by routine chemical chemistry procedures.

To visualize fat deposition in the liver, staining with Oil-red-O was performed on 4-μm frozen liver slices, and counterstaining was performed with hematoxylin, according to standard procedures.

For gas chromatography/mass spectrometry analysis, plasma cholesterol was extracted and derivatized as described elsewhere (31). Palmitate from isolated VLDL fractions was trans-methylated according to Lepage et al. (32).

Cholesterol and fatty acid derivatives were analyzed on a magnetic sector mass spectrometer (70-250S; Micromass, Manchester, U.K.) using a CP-Sil 19 column (Chrompack, Middelburg, the Netherlands) for assessment of isotopomer distribution patterns. For cholesterol samples, the oven temperature increased from 120 to 260°C at a rate of 20°C/min, from 260 to 280°C at 2.5°C/min and finally from 280 to 300°C at 20°C/min. The ions at mass/charge ratio (m/z) 368–371 were measured under selected ion recording. For fatty acid samples, the oven temperature increased from 100 to 300 ^oC at a rate of 12.5°C/min. The ions of the palmitate derivative were measured at m/z 270–272 under selected ion recording.

The principle of the mass isotopomer distribution analysis (MIDA) technique is described in detail elsewhere (31,33). Briefly, MIDA allows the enrichment of the pool of acetyl-CoA precursor units that has entered newly synthesized cholesterol or palmitate molecules during the course of a [1-¹³C]acetate infusion to be determined by analyzing the isotopomer pattern of the molecules of interest. This isotopomer pattern is compared with a theoretical table generated using binomial expansion and known isotope frequencies of the atomic isotopes. When the enrichment of the acetyl-CoA pool is known, it becomes possible to calculate the fraction of newly synthesized cholesterol and palmitate molecules in plasma, VLDLs, and liver homogenates. For determination of the absolute amount of newly synthesized hepatic palmitate, we multiplied the fraction by the total amount of hepatic palmitate at the end of the experiment.

Liver samples of ∼30 mg were used for total RNA isolation with the Trizol method (Gibco, Paisley, U.K.) and the SV Total RNA Isolation System (Promega, Madison, WI). Single-stranded cDNA was synthesized using materials from Boehringer Mannheim (Mannheim, Germany), according to the manufacturer’s instructions. Polymerase chain reaction (PCR) was performed in 50-μl preparations using 3 μl cDNA, 0.25 Taq polymerase, 5 μl 10-fold buffer, 0.75 μl dNTP-mix (10 mmol/l) (all from Hoffmann-La Roche), 2 μl DMSO, and 1 μl of each primer (25 pmol) (Gibco). The following primer sets were used: for acetyl-CoA carboxylase (ACC) (GenBank accession no. J03808), the sense primer was GGG ACT TCA TGA ATT TGC TGA TTC TCA GTT, and the anti-sense primer was GCT ATT ACC CAT TTC ATT ACC TCA ATC TC (34); for fatty acid synthase (FAS) (GenBank X13415), the sense primer was GGC TTT GGC CTG GAA CTG GCC CGG TGG CT, and the anti-sense primer was TCG AAG GCT ACA CAA GCT CCA AAA GAA TA (34); for sterol regulatory element binding protein (SREBP)-1 (SREBP-1a and -1c, GenBank L16995), the sense primer was CCT GTG TGT ACT GGT CTT CCT G, and the anti-sense primer was ACA AGA TGG CCT CCT GTG TAC T; for SREBP-2 (GenBank U02031), the sense primer was CAA TGG CAC GCT GCA GAC CCT TG, and the anti-sense primer was ATG GCC TTC CTC AGA ACG CCA G; for 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase (GenBank M29249), the sense primer was GAC ACT TAC AAT CTG TAT GAT G, and the anti-sense primer was CTT GGA GAG GTA AAA CTG CCA; for HMG-CoA synthase (GenBank X52625), the sense primer was TAC GAT GGT GTA GAT GCT GG, and the anti-sense primer was AGT TCT TCT GTG CTT TTC ATC CAC; for apolipoprotein B (apoB) (GenBank M14952), the sense primer was GAC ATG GTG AAT GGA ATC ATG, and the anti-sense primer was TGA AGA CTC CAG ATG AGG AC (34); for β-actin (GenBank M12481), the sense primer was AAC ACC CCA GCC ATG TAC G, and the anti-sense primer was ATG TCA CGC ACG ATT TCC C; for microsomal triglyceride transfer protein (MTP) (GenBank LA7970), the sense primer was ATC TGA TGT GGA CGT TGT GT, and the anti-sense primer was CCT CTA TCT TGT AGG TAG TG; for carnitine palmitoyltransferase I (CPT-I) (GenBank L07736), the sense primer was GCA TCA TCA CTG GTG TGT TC, and the anti-sense primer was TCT CCA TGG CGT AGT AGT TG. For each primer set, an increasing number of PCR cycles was performed while other conditions were fixed in order to determine the optimal number of cycles, which was chosen as the number halfway through the exponential phase. The PCR products were separated on 2.5% agarose gels. Images were made using a CCD video camera (Image Master VDS system; Pharmacia, Upsalla, Sweden). Intensities of the bands were quantified using the program Image Master 1D Elite 3.0.

All values reported are means ± SD. Significance was determined using the nonparametric Mann-Whitney U test for unpaired data. Differences were considered significant at P < 0.05.

Figure 1 shows the effects of S4048 infusion on plasma glucose, insulin, cholesterol, triglyceride, and FFA concentrations. Infusion of S4048 modestly decreased plasma glucose concentrations (P < 0.05), especially during the first 2 h of the experiment, with a subsequent significant decrease (P < 0.05) in plasma insulin concentrations. Both plasma glucose and insulin concentrations reached values approaching those in control rats at the end of the experiment. Triglyceride concentrations increased significantly during the course of the experiment from 0.4 ± 0.1 to 1.9 ± 0.6 mmol/l in S4048-treated rats. Cholesterol levels did not change during the course of the experiment, whereas FFA levels displayed a modest increase during S4048 infusion. The ketone body β-hydroxybutyrate concentration was 0.54 ± 0.23 mmol/l in control and 0.75 ± 0.71 mmol/l in S4048-treated animals. No effects of DMSO, S4048, or Triton WR1339 administration on ASAT or ALAT activities in plasma were found, indicating an absence of direct hepatotoxic actions of these compounds. A modest increase was found in ASAT and ALAT activities in the animals receiving all three compounds simultaneously.

Infusion of S4048 clearly affected hepatic carbohydrate and lipid contents (Table 1). Liver weight, expressed as percentage of body weight, was 3.0 ± 0.4 and 3.3 ± 0.2% (NS) in the control and S4048 groups, respectively. Hepatic G6P and glycogen contents both increased almost threefold after S4048 infusion. Total cholesterol content was not affected, although a higher relative cholesteryl ester content was found. Triglyceride content was markedly increased in the S4048 group, i.e., S4048 induced massive steatosis within the 8-h time frame of the experiment.

Figure 2 shows representative sections of livers from solvent- and S4048-treated rats, stained for neutral fat by Oil-red-O. Massive amounts of neutral fat were present in livers of S4048-treated rats, mainly in periportal areas of the hepatic lobuli. In contrast, the relatively small amounts of neutral fat present in the control liver were concentrated around the central veins, i.e., in perivenous areas of the lobuli.

In Table 2, the effects of S4048 on palmitate and cholesterol synthesis rates are summarized. Plasma fractional cholesterol synthesis rates were similar in both groups of rats. Fractional de novo lipogenesis in plasma VLDL and liver were increased almost threefold in the treated animals. When the increased hepatic palmitate content was taken into account, the absolute amount of newly synthesized palmitate was increased >10-fold in the S4048-treated group compared with the control group. Calculated enrichments of the acetyl-CoA pools showed significantly decreased values in the S4048-treated rats.

Figure 3 shows the mRNA levels of ACC, FAS, HMG-CoA reductase, HMG-CoA synthase, and SREBP-1 and -2 in control and S4048-treated rats, as determined by a semiquantitative reverse transcriptase–PCR approach. Intensities of bands were normalized to those of β-actin (Fig. 3B). The results clearly show upregulation of the mRNA levels of ACC (∼4-fold) and FAS (∼14-fold) in S4048-treated rats but no effects on the mRNA levels of HMG-CoA reductase and HMG-CoA synthase. Furthermore, steady-state mRNA levels of transcription factors controlling de novo lipogenesis and cholesterol synthesis, i.e., SREBP-1 and -2, were not affected by S4048 infusion. As expected, based on unaltered plasma β-hydroxybutyrate concentrations, S4048 did not induce changes in CPT-I mRNA levels (data not shown).

Despite the significant upregulation of hepatic lipogenesis and hepatic lipid content, S4048 infusion did not affect hepatic VLDL-triglyceride production rate (Fig. 4); values of 151 ± 33 and 137 ± 14 μmol · kg^–1 · h^–1 were calculated for control and S4048-treated rats, respectively. The values obtained in the animals infused with DMSO solution alone were similar to those found in rats infused with saline for the same period of time (154 ± 29 μmol · kg^–1 · h^–1), indicating no effects of the solvent at this dose on VLDL-triglyceride production. The size of the VLDL particles produced by rats treated with the solvent or with S4048 was also similar, with average diameters of 40.7 ± 6.7 and 45.9 ± 9.6 nm in control and S4048-treated animals, respectively. As expected, no differences were found in hepatic expression of apoB or MTP genes (data not shown).

Acute inhibition of the G6Pase system by S4048 stimulated hepatic de novo lipogenesis >10-fold in rats, despite decreased circulating levels of insulin, a well-known stimulator of lipogenesis (36). Increased de novo lipogenesis was associated with an increased flux through the acetyl-CoA pool because lower ¹³C-acetyl-CoA enrichments were calculated for rats treated with S4048. Thus, more substrate for fatty acid synthesis was produced in the S4048-treated rats than in the control rats, presumably attributable to an increased glycolytic flux. Therefore, fatty acid synthesis was not only stimulated by an increased precursor supply. Concomitantly, hepatic mRNA levels of two key enzymes in de novo lipogenesis, ACC and FAS, were markedly induced within the 8-h time frame of the experiment.

Pathways of de novo lipogenesis are under transcriptional control of SREBPs (37,38), a group of transcription factors (SREBP1a, -1c, and -2) that regulate the expression of genes involved in cholesterol, fatty acid, and glucose metabolism. SREBP-1 gene knockout mice show a very low basal expression of ACC and FAS and hardly possess the ability to upregulate de novo lipogenesis (37). In contrast, overexpression of the nuclear form of SREBP-1a leads to massive steatosis and increased de novo lipogenesis, albeit in the absence of hypertriglyceridemia (39). In S4048-treated rats, SREBP-1 mRNA expression was not significantly induced, but this obviously does not exclude the possibility of direct SREBP-1–mediated activation of gene expression, particularly in view of the recent results of Foretz et al. (40). These authors have shown that enhancement of mRNA expression of ACC and FAS by SREBP-1c in isolated hepatocytes critically depends on the presence of glucose in the medium. Furthermore, it is well-established that glucose exerts stimulatory effects of lipogenic gene expression only after being metabolized to either G6P (41) or xylose-5-phosphate (42). In a recent study (18), it was shown that overexpression of glucokinase in fed rats, leading to increased G6P concentrations, resulted in a marked increase in plasma triglyceride levels. In light of the threefold increase in G6P concentration in the liver of rats infused with S4048, potentiation of transcriptional activity of SREBP-1 by G6P is highly likely to occur in our model.

Despite increased production of the obligatory precursor, i.e., acetyl-CoA, our stable isotope study showed unaffected cholesterol synthesis rates during infusion of S4048 and unaffected hepatic mRNA levels of HMG-CoA synthase and HMG-CoA reductase. In accordance, overexpression of glucokinase resulted in an increased hepatic G6P content and did not increase plasma cholesterol levels (18). SREBP-2 is a strong regulator of cholesterol synthesis, it has the ability to upregulate various genes involved in the cholesterogenic pathway, such as HMG-CoA synthase and HMG-CoA reductase (43,44). SREBP-2 mRNA levels were not affected in the S4048-treated animals. These combined results clearly demonstrate that in spite of the common regulatory mechanisms involved, i.e., SREBP-modulated activation of gene expression, de novo lipogenesis and cholesterogenesis are differentially regulated under conditions associated with increased glycolytic flux in rat liver.

Partial inhibition of G6Pase for an 8-h period was associated with massive steatosis. Fat accumulation was much more pronounced in hepatocytes located in the zone surrounding the portal vein and hepatic artery (periportal area) than in those surrounding the hepatic vein (perivenous area). This is probably related to the predominant periportal localization of G6Pase (45). Increased de novo lipogenesis contributed to the development of steatosis, but the quantitative contribution of newly synthesized fatty acids to the steatosis appeared rather limited, i.e., was <10%. Furthermore, VLDL-triglyceride secretion was similar in both groups (see below) and thus did not differentially influence hepatic lipid stores. Increased fat uptake, therefore, must have contributed, either in the form of VLDL/LDL-triglycerides or as FFAs, although plasma FFA concentrations were only moderately increased on treatment. The latter, however, does not exclude an enhanced FFA flux to the liver. Furthermore, fatty acid oxidation might have been impaired in the S4048-treated rats, although similar mRNA levels of CPT-I in livers and unaffected plasma β-hydroxybutyrate concentrations were found. Malonyl-CoA is produced during the course of fatty acid synthesis and is expected to accumulate in the livers of the S4048-treated rats. Malonyl-CoA is a strong, allosterically acting inhibitor of fatty acid oxidation (46). Furthermore, the increase in hepatic triglyceride content with similar β-hydroxybutyrate concentrations strongly suggests a shift in the balance between fatty acid oxidation and esterification in the liver.

Many factors are known to influence hepatic VLDL-triglyceride production and secretion. De novo lipogenesis has been suggested to be of regulatory importance for VLDL production (47). Furthermore, many studies have shown that increases in fatty acid delivery to the liver lead to increased triglyceride synthesis and are accompanied by increases in VLDL secretion (48,49,50). The balance between apoB synthesis and degradation is an important factor in controlling hepatic triglyceride secretion, and inhibition of protein synthesis has been shown to reduce VLDL-triglyceride secretion (51). Additionally, insulin is a well-known acute inhibitor of VLDL secretion (52,53,54), and insulin resistance is associated with increased VLDL-triglyceride and apoB secretion (9,10,11). Furthermore, recent data has shown that phosphorylation of glucose is also of importance in the regulation of assembly and secretion of triglyceride-containing VLDL (19). In our study, surprisingly, we did not find increased VLDL-triglyceride secretion in the S4048-treated rats. Moreover, neither the number nor the size of the VLDL particles were affected by S4048 infusion. In our model, the induction of de novo lipogenesis, which strongly increased hepatic triglyceride content in combination with increased hepatic G6P levels and decreased insulin concentrations, was also expected to increase the secretion of VLDL-triglycerides. However, a number of factors have to be taken into account. First of all, it is not known whether apoprotein synthesis was impaired in the S4048-treated rats. Furthermore, if glucose itself is also important in this process, lowered plasma glucose concentrations in the S4048-treated rats might cause an inability to adequately upregulate VLDL secretion. It should be stressed that in our model, G6P content is increased in the cytoplasm but probably decreased in the ER. Compartmentalization of G6P could potentially play a role in its capacity to influence VLDL secretion, but further studies are needed to clarify this phenomenon.

The observation that VLDL-triglyceride secretion was not increased in rats after S4048 treatment indicates that the hyperlipidemia observed after G6Pase inhibition must have originated from decreased triglyceride clearance. Insulin is a well-known stimulator of adipocyte lipoprotein lipase activity. In the S4048 model with a low concentration of insulin, lipoprotein lipase activity was probably decreased, leading to decreased lipolysis of VLDL-triglycerides. Indeed, studies in GSD patients have shown low lipoprotein lipase activity in GSD patients (55).

Data in the literature on the relation between G6Pase activity and lipid metabolism is confusing. Based on our results, we postulate that G6P concentrations in the liver, specifically in certain compartments, play a pivotal role in determining triglyceride concentration in liver and plasma. Altered activation of SREBP-1 and/or changes in the intrahepatic concentration of G6P in itself are most likely a better explanation than the plasma concentration of glucose or insulin for the apparently conflicting effects of G6Pase activity on hepatic lipid metabolism in diabetes and GSD.

In conclusion, acute inhibition of G6Pase activity in rats leads to increased de novo lipogenesis and massive steatosis within a relatively short time frame. Cholesterogenesis was not affected in our study, implying a dissociated regulation of cholesterol and fatty acid synthesis under the conditions used. Increased de novo lipogenesis and hepatic lipid accumulation alone is not sufficient to stimulate VLDL-triglyceride secretion. This study underlines the important function of G6P in the control of hepatic lipid metabolism.

This work was supported by a grant from the Dutch Foundation for Scientific Research (NWO 920-03-123) and a grant from the Netherlands Diabetes Research Foundation (DFN 96.604).

We thank Gijs Nagel and Theo Boer for their help with the gas chromatography/mass spectrometry measurements and Vincent Bloks and Torsten Ploesch for the development of primers.