Increased Levels of Nuclear SREBP-1c Associated with Fatty Livers in Two Mouse Models of Diabetes Mellitus*

Hepatic steatosis is common in non-insulin-dependent diabetes and can be associated with fibrosis and cirrhosis in a subset of individuals. Increased rates of fatty acid synthesis have been reported in livers from rodent models of diabetes and may contribute to the development of steatosis. Sterol regulatory element-binding proteins (SREBPs) are a family of regulated transcription factors that stimulate lipid synthesis in liver. In the current studies, we measured the content of SREBPs in livers from two mouse models of diabetes, obese ob/ob mice and transgenic aP2-SREBP-1c436 (aP2-SREBP-1c) mice that overexpress nuclear SREBP-1c only in adipose tissue. The aP2-SREBP-1c mice exhibit a syndrome that resembles congenital generalized lipodystrophy in humans. Both lines of mice develop hyperinsulinemia, hyperglycemia, and hepatic steatosis. Nuclear SREBP-1c protein levels were significantly elevated in livers from ob/ob and aP2-SREBP-1c mice compared with wild-type mice. Increased nuclear SREBP-1c protein was associated with elevated mRNA levels for known SREBP target genes involved in fatty acid biosynthesis, which led to significantly higher rates of hepatic fatty acid synthesis in vivo. These studies suggest that increased levels of nuclear SREBP-1c contribute to the elevated rates of hepatic fatty acid synthesis that leads to steatosis in diabetic mice.

Non-insulin-dependent diabetes mellitus is a common disorder that affects approximately 5% of the population. Affected patients are usually obese, and they manifest insulin resistance, hyperinsulinemia, and hyperglycemia. As many as 40% of noninsulin-dependent diabetics develop evidence of hepatic steatosis or "fatty liver," a condition that leads to hepatic fibrosis and cirrhosis in a subset of individuals (1). Hepatic fatty acid synthesis is increased in rodent models of hyperinsulinemia and likely contributes to the development of fatty livers (2,3).
Sterol regulatory element-binding proteins (SREBPs) 1 are a family of transcription factors that activate the entire program of cholesterol and fatty acid synthesis in liver (4,5). SREBPs belong to the basic helix-loop-helix-leucine zipper family of transcription factors (4). Unlike other members of the basic helix-loop-helix-leucine zipper family, SREBPs are synthesized as Ϸ1150-amino acid precursors bound to the endoplasmic reticulum and nuclear envelope (4). To be active, the NH 2terminal segment must be released from the membrane by a sequential two-step cleavage process (6,7). Following the second cleavage (site-2 cleavage), the Ϸ500-amino acid NH 2 -terminal segment of SREBP is released from the membrane and translocates to the nucleus, where it binds to enhancer regions of target genes to activate transcription.
To date, three SREBP isoforms have been identified and characterized (4,8). SREBP-1a and -1c are derived from a single gene through the use of alternative transcription start sites that produce alternate forms of exon 1 (9). The third SREBP isoform, SREBP-2, is derived from a separate gene and is Ϸ45% identical to SREBP-1a (10). In most cultured cell lines, the predominant SREBP-1 isoform is SREBP-1a (11). In contrast, most animal tissues, including liver, express SREBP-1c as the predominant SREBP-1 isoform (11). Multiple lines of evidence suggest that SREBP-1 and SREBP-2 have different relative effects on target genes. SREBP-1 is relatively selective in activating genes involved in fatty acid synthesis, while SREBP-2 preferentially activates genes involved in cholesterol biosynthesis (12)(13)(14)(15).
In contrast, mice that overexpressed nuclear SREBP-2 (nSREBP-2) had 28-fold higher rates of cholesterol synthesis in vivo, and corresponding 10 -12-fold increases in mRNAs for several enzymes in the cholesterol biosynthetic pathway (15). The SREBP-2 transgenic mice also had 4-fold higher rates of fatty acid biosynthesis in vivo. This demonstrated that at high levels of expression, nSREBP-2 is capable of activating the enzyme cascade required for fatty acid biosynthesis, albeit much less efficiently than for that of cholesterol biosynthesis.
In the current studies, we demonstrate that the amount of nSREBP-1 is increased in fatty livers from two distinct animal models of non-insulin-dependent diabetes. The first model is the "obese" (ob/ob) mouse that develops severe obesity and hyperinsulinemia (18). The second model utilizes transgenic mice that overexpress nuclear SREBP-1c exclusively in adipose tissue (aP2-SREBP-1c). These transgenic animals are one of three recently described mouse models that resemble generalized lipodystrophy in humans (19 -21). The aP2-SREBP-1c transgenic mice have very little adipose tissue, apparently as a result of disturbed adipocyte differentiation. They also develop severe hyperglycemia, hyperinsulinemia, and lipid accumulation in liver. Despite the striking differences in the amounts of body fat, both diabetic mouse models manifest increased nSREBP-1c protein content in liver, corresponding elevations in mRNAs for multiple lipogenic genes, and increased rates of fatty acid biosynthesis. In contrast, nSREBP-2 levels and mRNAs for genes involved in cholesterol homeostasis were unchanged in the livers from ob/ob and aP2-SREBP-1c mice. We propose that increased nSREBP-1c protein results in the transcriptional activation of genes responsible for lipogenesis and therefore contributes to the fatty liver phenotype observed in diabetic mice.

EXPERIMENTAL PROCEDURES
Materials and General Methods-All chemicals used were from Sigma. [␣-32 P]dCTP and [␣-32 P]CTP were obtained from Amersham Pharmacia Biotech. The content of cholesterol and triglyceride in plasma and liver was measured as described previously (22,23). Plasma insulin was measured with a monoclonal anti-rat insulin radioimmunoassay using the Rat Insulin RIA kit (Linco Research, Inc., St. Charles, MO). Plasma glucose and free fatty acids were measured using the Glucose (Trinder) 100 kit (Sigma Diagnostics, St. Louis, MO), and the Wako NEFA C test kit (Wako Chemicals, Richmond, VA), respectively.
Animals-"Obese" (C57BL/6J-ob/ob), and C57BL/6J wild-type mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Transgenic mice that overexpress the truncated form of human SREBP-1c (amino acids 1-436) in adipose tissue under the control of the adipocyte-specific aP2 enhancer/promoter were described previously (21). The aP2-SREBP-1c mice are on a mixed genetic background consisting of C57BL/6J and SJL strains. All mice were housed in colony cages in a 14-h light/10-h dark cycle and were maintained on Teklad 6% (w/w) Mouse/Rat Diet 7002 from Harlan Teklad Premier Laboratory Diets (Madison, WI). In experiment A, 5 male wild-type C57BL/6J and 5 male ob/ob mice were studied at 13 weeks of age. For experiment B, 4 male wild-type and 4 male transgenic aP2-SREBP-1c littermate mice were studied at 17-weeks of age. All mice were sacrificed during the mid-dark cycle following a 1-2-h fast.
Immunoblots-Pooled liver membranes and nuclear extracts from livers of mice in experiments A and B were prepared and immunoblot analysis for endogenous SREBP-1 and SREBP-2 was performed as described previously (14,16).

RNA Analysis-For
Northern gel analysis, equal aliquots of total RNA made from each mouse liver were pooled (total, 10 g), denatured with formaldehyde and formamide, subjected to electrophoresis in a 1% agarose gel, and transferred to Hybond Nϩ membranes (Amersham Pharmacia Biotech) for hybridization. Hybridization conditions and cDNA probe preparation were carried out as described (15)(16)(17). As a loading control, a cDNA probe for mouse ␤-actin was prepared using reverse transcriptase-PCR and mouse liver poly(A) ϩ RNA as a template as described previously (16). The PCR primers used were as follows: 5Ј primer, 5Ј-CATTGAACATGGCATTGTTACCAACTGGGA-3Ј, and 3Ј primer, 5Ј-GCCATCTCCTGCTCAGAGTCTAGAGCAACA-3Ј (24). Northern blot filters were exposed to a Fuji PhosphorImager, and the resulting bands were quantified using a Bio-Imaging Analyzer with BAS1000 Mac-BAS software (Fuji Medical Systems, Stamford, CT). The -fold change for each mRNA was calculated after normalization by the signal generated by ␤-actin. The RNase protection assay for SCD1 and SCD2 mRNA transcripts was carried out as described previously (17).
Cholesterol and Fatty Acid Synthesis in Vivo-The rates of liver cholesterol and fatty acid synthesis were measured in nonfasted mice using [ 3 H]water as described previously (16). The data presented are pooled results from two independent experiments, each of which included 5 12-16-week-old male mice of the indicated genotype.
Tissue Cholesterol and Fatty Acid Composition-The relative hepatic fatty acid compositions in the indicated lipid fractions were measured as described previously (17). Table I shows the phenotypic characteristics of the mice used in the current studies. In experiment A, we studied male ob/ob mice on the C57BL/6J background and age-matched wild-type C57BL/6J male mice as controls. In experiment B, we studied aP2-SREBP-1c mice that overexpress human nSREBP-1c exclusively in adipose tissue and sex-matched littermate wildtype controls. The aP2-SREBP-1c mice were derived from the C57BL/6J and SJL strains. The ob/ob and aP2-SREBP-1c mice had similar elevations in plasma glucose and insulin, indicating they had similar levels of insulin resistance and diabetes. Liver cholesterol was increased 2-3-fold in both diabetic models. Liver triglyceride content was markedly elevated in livers from ob/ob (10-fold) and aP2-SREBP-1c (16-fold) mice compared with wild-type levels. The two models differ in that the ob/ob mice had massive peripheral fat stores and higher plasma free fatty acid concentrations compared with their wildtype controls. Conversely, the aP2-SREBP-1c mice had markedly reduced peripheral fat stores with a slight reduction in plasma free fatty acid levels, compared with their littermate controls. Fig. 1 shows immunoblots of SREBP-1 and -2 in nuclear extracts (N) and membranes (P) from livers of ob/ob mice, aP2-SREBP-1c mice, and their respective wild-type controls after a brief 1-2-h fast. The amount of transcriptionally active nSREBP-1 was 3-4-fold increased in livers from ob/ob (lane 2) and aP2-SREBP-1c (lane 6) mice compared with their respective wild-type controls (lanes 1 and 5). Consistent with previous experiments (11,17,25), the predominant SREBP-1 mRNA transcript present was SREBP-1c as determined by an RNase protection assay (data not shown). This suggests that the SREBP-1 isoform measured by immunoblot is predominantly the SREBP-1c isoform. In contrast to the increase in nSREBP-1c, the amount of nSREBP-2 showed very little change in livers from either diabetic model (lanes 4 and 8 versus lanes 3 and 7,  respectively). Similar results were obtained in one other independent experiment.

RESULTS
To quantify the mRNA levels of known SREBP-1c target genes, a series of Northern blot analyses were performed on pooled samples of total RNA from the livers of the mice described in Table I. The mRNA levels for several key genes involved in cholesterol synthesis and lipogenesis are shown in Fig. 2 for wild-type (WT) and ob/ob (ob) mice (panel A) and for wild-type (WT) and aP2-SREBP-1c (Tg) mice (panel B). Both hyperinsulinemic models had similar 2.5-5-fold elevations in the mRNAs for the lipogenic enzymes ATP citrate lyase, ACC, FAS, GPAT, and the NADPH-producing enzymes malic enzyme and Glu-6-PD. Less significant changes were measured in the mRNAs for the genes involved in cholesterol metabolism (i.e. low density lipoprotein receptor, HMG-CoA synthase, and HMG-CoA reductase). Similar results were obtained in one additional independent experiment (data not shown).
To confirm that the changes measured in the mRNAs for the lipogenic enzymes resulted in increased rates of fatty acid synthesis in vivo, we used [ 3 H]water to directly measure newly synthesized cholesterol and fatty acids in livers of wild-type and diabetic mice (Table II). The fatty acid synthetic rates were 6-fold higher per gram of liver and 23-fold higher per organ in ob/ob mice. Livers from aP2-SREBP-1c mice had fatty acid synthetic rates that were 2.5-and 5.4-fold increased per gram and organ, respectively. In contrast, livers from ob/ob and aP2-SREBP-1c mice had 6-and 3-fold reductions in the rate of incorporation of [ 3 H]water into digitonin-precipitable sterols per gram of liver, respectively. The lower cholesterol synthetic rates presumably reflect post-transcriptional suppression of HMG-CoA reductase activity owing to the increased level of hepatic cholesterol (26).
To determine whether SCD1 and/or SCD2 mRNA was elevated in livers of the two diabetic mouse models, we employed an RNase protection assay to detect the mRNA for each gene (Fig. 3). Lanes 1 and 2 show the protected bands for the SCD1 and SCD2 mRNA in 5 g of total RNA from epididymal fat pads from wild-type mice. Inasmuch as both SCD1 and SCD2 mRNA transcripts are present in white fat, this RNA was used as a positive control (28). Compared with their respective wild-type levels, the SCD1 mRNA was 3.5-fold higher in ob/ob mouse livers (lanes 3 and 4) and 2-fold higher in aP2-SREBP-1c mouse Table I (experiments A and B) were pooled, and aliquots (30 g of protein) of the membranes and nuclear extracts were subjected to 8% SDS-PAGE. Immunoblot analysis was carried out using 5 g/ml rabbit anti-mouse SREBP-1 IgG (lanes 1, 2, 5, and 6) or SREBP-2 IgG (lanes 3, 4, 7, and 8) as the primary antibody and 0.25 g/ml horseradish peroxidase-coupled donkey anti-rabbit IgG as the secondary antibody. Filters were exposed to Reflection TM NEF496 film for 15 s (lanes 1, 2, 5, and 6) and 30 s (lanes 3, 4, 7, and 8). P and N denote the precursor and cleaved nuclear forms of SREBP, respectively. -1c (Tg) mice (B). The mice used in this experiment are described in Table  I. Total RNA isolated from livers of individual mice was pooled, and 10-g aliquots were subjected to electrophoresis and blot hybridization with the indicated 32 P-labeled cDNA probe. The dried filters were exposed to Reflection TM NEF 496 film with intensifying screens at Ϫ80°C for 3-36 h. The amount of radioactivity in each band was quantified by exposing the filters to a Fuji PhosphorImager and using a Bio-Imaging Analyzer with BAS1000 MacBAS software (Fuji Medical Systems, Stamford, CT). The -fold change in each mRNA from ob/ob mice (A) and aP2-SREBP-1c mice (B), relative to that of the respective wild-type mice, was calculated after correction for loading differences with ␤-actin. These values are shown below each blot. livers (lanes 7 and 8) after correction for levels of the control mRNA encoding ␤-actin. The mRNA for SCD2 was not detected in livers from wild-type, ob/ob, or aP2-SREBP-1c mice at exposures up to 36 h at Ϫ80°C (lanes 5 and 6 and lanes 9 and 10). This result is consistent with previous observations that nSREBP-1c overexpression in liver and cultured cells results in increased mRNA levels for SCD1 but not SCD2 (12,13).

FIG. 2. Amounts of various mRNAs as measured by blot hybridization in livers of wild-type C57BL/6J (WT) and C57BL/6J (ob/ob) (ob) mice (A), and wild-type (WT) and transgenic aP2-SREBP
To determine whether the increased SCD1 mRNA resulted in increased hepatic monounsaturated fatty acid content, we measured the relative fatty acid compositions of various lipid fractions from livers of the mice described in Table I. Table III shows the relative fatty acid compositions in total lipid extracts as well as the three major lipid classes after fractionation. Livers from ob/ob and aP2-SREBP-1c mice had similar 2-3fold increases in the percentage of 16:1 and 18:1 in total lipid extracts. Therefore, approximately 50% of the total fatty acids in the livers from the diabetic mice are monounsaturated (primarily oleic acid). These differences are primarily a result of changes in fatty acid composition in the triglyceride and cholesteryl ester fractions.

DISCUSSION
The purpose of the current studies was severalfold: 1) to determine whether the amounts of nSREBPs were altered in fatty livers from obese and lipodystrophic mouse models of non-insulin-dependent diabetes, 2) to correlate the changes in nSREBP-1c content with the relative mRNA levels of known lipogenic SREBP-1c target genes, and 3) to determine whether increased nSREBP-1c expression contributes to the development of fatty liver in both mouse models of diabetes.
Previous studies in fasted and refed mice suggested that nSREBP-1c was regulated in parallel with the amounts of mRNA encoding lipogenic enzymes (29). Fasted mice had markedly reduced levels of hepatic nSREBP-1c and nSREBP-2. Refeeding a high carbohydrate/low fat diet led to a 4-fold "overshoot" in the amount of nSREBP-1c compared with pre-fasted levels, while nSREBP-2 returned only to pre-fasted levels (29). The pattern of regulation of nSREBP-1c closely paralleled the changes in mRNAs for lipogenic genes, whereas the changes in nSREBP-2 protein paralleled the changes in mRNA levels for genes encoding enzymes of cholesterol synthesis (29). These studies suggested that nSREBP-1c contributed to the fasting and refeeding response that has been reported for lipogenic enzymes in liver (30). This response has been previously attributed to the direct effects of ingested glucose or the secondary effect of the elevated insulin that occurs in response to ingested glucose (30).
Inasmuch as lipogenesis is increased in livers from ob/ob mice (3), we hypothesized that increased levels of nSREBP-1c could contribute to this process. Indeed, we found markedly elevated levels of nSREBP-1c protein in livers from ob/ob mice. To address whether this observation is specific to ob/ob mice or whether it represents a more general consequence of hyperinsulinemia/hyperglycemia, we studied a phenotypically different model of insulin-resistant diabetes, the aP2-SREBP-1c transgenic mouse. These mice exhibited similar 3-4-fold elevations in hepatic nSREBP-1c. The increased nSREBP-1c levels in both models were associated with increased mRNAs for multiple lipogenic enzymes and increased rates of fatty acid synthesis and triglyceride accumulation in liver. The predominant fatty acid synthesized was oleic acid, presumably as a consequence of the increase in SCD1 mRNA. Similar results were reported in mice in whom the phosphoenolpyruvate carboxykinase promoter was used to overexpress nSREBP-1c in liver (13). These transgenic mice exhibited increased mRNA levels for genes involved in fatty acid synthesis and elevated fatty acid synthetic rates in vivo in the absence of hyperinsulinemia or hyperglycemia. These observations suggest that the mRNA changes measured in the diabetic mice are a direct result of increased nSREBP-1c levels in liver.
Nuclear SREBP-2 levels were unchanged in livers from the two diabetic mouse models, and no significant changes were measured in the mRNAs for genes involved in cholesterol homeostasis. These studies support previous cell culture and in vivo studies showing that the isoforms of SREBP-1 preferen- FIG. 3. Changes in the amounts of mRNAs for SCD1 and SCD2 in livers from wild-type (WT), C57BL/6J (ob/ob) (ob), and transgenic aP2-SREBP-1c (Tg) mice, as measured by the RNase protection assay. Total RNA was isolated from epididymal fat of C57BL/6J mice, and 5-g aliquots were subjected to an RNase protection assay for SCD1 (lane 1) and SCD2 (lane 2) as described under "Experimental Procedures." White fat contains mRNA transcripts for both SCD1 and SCD2, and was thus used as a positive control. Total RNA isolated from livers of mice described in Table I was pooled, and 5-g aliquots were hybridized with 32 P-labeled cRNA probe for SCD1 (lanes 3 and 4 and lanes 7 and 8) or SCD2 (lanes 5 and 6 and lanes 9 and 10). After RNase digestion, the protected fragments were separated by gel electrophoresis and exposed to film at Ϫ80°C for 6 h.

TABLE II
In vivo synthesis of cholesterol and fatty acids in livers from wild-type, ob/ob, and aP2-SREBP-1c transgenic mice Each value represents the mean Ϯ S.E. from two independent studies each of which had 5 12-16-week-old male mice of the indicated genotype. The mice were maintained on a chow diet and studied nonfasted. All mice were injected with [ 3 H]water intraperitoneally, and 1 h later the liver was removed for measurement of its content of 3  tially activate enzymes involved in lipogenesis, whereas SREBP-2 is primarily responsible for the transcriptional regulation of genes involved in cholesterol homeostasis (12,15). The mechanism for the increase in nSREBP-1c in the diabetic mice is not established in these studies. It likely involves post-transcriptional as well as transcriptional control. Foretz et al. (31) have recently reported that insulin mediates increased SREBP-1 transcription in rat primary hepatocytes and that SREBP-1 protein is required for the transcriptional activation of several lipogenic genes by glucose. In the current in vivo studies, a 2.4-fold increase in the mRNA level for SREBP-1c was measured in livers from ob/ob mice but not in livers from aP2-SREBP-1c mice (Fig. 2). However, in livers from both models, we consistently measured a significant increase in the amount of nSREBP-1c protein. Therefore, increased SREBP-1c transcription is not solely responsible for the measured increase in nSREBP-1c protein. These findings suggest that enhanced cleavage of the membrane-bound SREBP-1c precursor and/or delayed post-transcriptional degradation of nSREBP-1c is altered in hyperinsulinemic/hyperglycemic states. Determining which of the components in the insulin-signaling cascade ultimately regulates nSREBP-1c protein levels will shed new light on the link between hyperinsulinemia and the molecular events that result in increased lipogenesis and steatosis in liver.

TABLE III
Fatty acid composition of livers from wild-type, ob/ob, and aP2-SREBP-1c mice Liver samples from individual mice of each genotype were extracted, and the major classes of lipids were separated on silica columns. The lipid fractions were methyl-esterified and quantified by gas-liquid chromatography as described under "Experimental Procedures." Each value represents the mean from the mice described in Table I. Standard errors of the means were all less than 15% of the mean and are omitted for clarity. Bold values denote a level of statistical significance of p Ͻ 0.05 between wild-type and ob/ob or aP2-SREBP-1c mice.