Human Obesity and Type 2 Diabetes Are Associated With Alterations in SREBP1 Isoform Expression That Are Reproduced Ex Vivo by Tumor Necrosis Factor-α

The characteristics of the populations studied are summarized in Table 1. Group A comprised 17 lean, 22 obese, and 7 type 2 diabetic patients from Indiana University (IU) who donated subcutaneous adipose tissue biopsies. Group B comprised 11 lean, 7 obese, and 8 morbidly obese (BMI >40 kg/m²) patients undergoing elective open-abdominal surgery at Addenbrooke’s Hospital (AH) who donated subcutaneous adipose tissue biopsies. Group C comprised 12 lean, 15 obese, and 6 weight-reduced formerly obese patients undergoing elective abdominal surgery at East Carolina University (ECU) who donated rectus abdominus muscle samples. Three subjects from the latter group were studied in both the obese and weight-reduced state. Weight reduction in the weight-reduced group was induced by gastric bypass surgery. Five of these patients had stable weight for at least 4 months before the second biopsy. Group D comprised subjects studied at the VA San Diego Healthcare System and University of California San Diego (UCSD). Muscle biopsy specimens were obtained from vastus lateralis of lean, obese, and type 2 diabetic subjects before and after a 3-h hyperinsulinemic (300 mU · m⁻² · min⁻¹) euglycemic (5.0–5.5 mmol/l) clamp as described in detail previously (18). Protocols involving muscle biopsies were approved by ECU policy and the UCSD Review on Human Research. The Institutional Review Board at IU and the Cambridge Local Research Ethics Committee (at AH) approved protocols involving adipose biopsies. Informed consent was obtained from all patients. For all subjects, obesity was defined as BMI >27.3 kg/m² for men and BMI >27 kg/m² for women (19).

All chemicals and reagents were purchased from Sigma (Dorset, U.K.) unless otherwise stated. Tumor necrosis factor-α (TNF-α) was purchased from R&D Systems Europe (Abingdon, U.K.), and the anti-SREBP1 antibody was obtained from Santa Cruz Biotechnology (Calne, U.K.). Secondary antibodies were from DAKO (Goldstrup, Denmark), and enhanced chemiluminescence (ECL) reagents and Hyperfilm were from Amersham Pharmacia Biotech (Little Chalfont, U.K.). RNeasy extraction kit was obtained from Qiagen (West Sussex, U.K.). Polyethyleneglycol methacrylate-T easy cloning vector and RT reagents were from Promega (Madison, WI), and Cyclophilin RPA internal control was from Ambion (Austin, TX).

Muscle tissue biopsies at UCSD and abdominal subcutaneous adipose tissue biopsies at IU, ECU, and AH were obtained as previously described (20). Total RNA was obtained by guanidinium thiocyanate-phenol chloroform extraction (21), except samples for AH, which were extracted using the RNeasy mini extraction kit according to the manufacturer’s recommendations. RNA samples were quantified by spectrophotometry, and integrity was assessed by agarose gel electrophoresis and ethidium bromide staining. The RNA samples were then diluted as appropriate in RNase-free water and stored at −80° C.

Adipocyte isolation for group B was as follows. Adipose tissue biopsies were placed in normal saline and immediately processed (transport time to the laboratory was <5 min). The adipose tissue was finely diced and digested in a collagenase solution (Hank’s balanced salt solution containing 3 mg/ml type II collagenase and 1.5% BSA) for 1 h in a shaking water bath at 37°C. After digestion, the mature adipocytes were separated from the stromo-vascular cells by centrifugation (10 min, 1,500g) of the digestion mixture. The mature adipocytes were removed from the top layer, and then RNA was immediately extracted or adipocytes were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F12 plus 10% fetal bovine serum (FBS) at 37°C/5% CO₂ in the absence or presence of 10 or 100 ng/ml TNF-α, 100 nmol/l insulin, or 10 ng/ml TNF-α plus 100 nmol/l insulin. After 24 h, total RNA or protein was extracted. Protein extraction was carried out using RIPA buffer (1% NP40, 1% sodium deoxycolate, 0.1% SDS, 0.9% NaCl, 250 mmol/l Tris HCl [pH 6.8], 1 mmol/l EDTA, and 1 × protease cocktail inhibitor; Boehringer Mannheim, Mannheim, Germany).

The stromo-vascular pellet containing the preadipocytes was treated with erythrocyte lysis solution (154 mmol/l NH₄Cl, 10 mmol/l KHCO₃, and 0.1 mmol/l EDTA) for 5 min at room temperature and centrifuged (5 min, 1,500g). The preadipocyte pellet was cultured in DMEM/Ham’s F12 supplemented with 10% FBS, 2 mmol/l glutamine, 100 units penicillin, and 0.1 mg/ml streptomycin at 37°C in a humidified 95% air and 5% CO₂ incubator. Cultures were passaged four times and grown to confluence (day 0). At confluence, the medium was changed to a serum-free hormonally modified differentiating medium (DMEM/Ham’s F12 supplemented with 2 mmol/l glutamine, 100 units penicillin, 0.1 mg/ml streptomycin, 33 μmol/l Biotin, 17 μmol/l pantothenic acid, 10 μ g/ml human apo-transferrin, 0.2 nmol/l tri-iodothyronine, 100 nmol/l cortisol, and 500 nmol/l insulin) with the addition of the following compounds: 10⁻⁷ mol/l BRL 49653 and 10⁻⁷ mol/l LG 100268. For the first 3 days only, 250 μmol/l iso-butylmethylxanthine was added to the differentiating medium. Medium was replaced every 2–3 days. After 15 days postdifferentiation, TNF-α was added to the medium (10 ng/ml). Control samples were maintained in medium without TNF-α after 24 h RNA was extracted.

Expression of mRNA was analyzed using two different methods, depending on RNA availability.

The studies described were performed using a partial human ADD1/ SREBP1 cDNA probe common to both isoforms. This probe was generated as previously described (22, 23) by RT-PCR using total RNA from human adipose tissue using primers (5′TCT ACC ATA AGC TGC ACC AGC TG3′ and 5′CAG TCC CCA TCC ACG AAG AAA CG3′) designed to amplify 319 bp of the hSREBP1 sequence. A 103-bp cDNA corresponding to human cyclophilin (Ambion, Austin, TX) was used as internal control. SREBP1 and cyclophilin were quantitated using previously described solution hybridization RNase protection assay methods and phosphorimager analysis (22,23).

Real-time quantitative PCR was used to analyze the RNA samples from group B. This PCR-based method was developed to overcome the problem of limiting amounts of human RNA. Total RNA (100 ng) was reverse-transcribed for 1 h at 37°C in a 20-μl reaction containing 1× RT buffer (50 mmol/l Tris-HCl, 75 mmol/l KCl, 3 mmol/l MgCl₂, and 10 mmol/l dithiothreitol), 100 ng random hexamers, 1 mmol/l dNTPs, and 100 units Moloney murine leukemia virus RT (Promega). Reactions in which RNA was omitted served as negative controls. A reaction containing 500 ng of adipocyte total RNA was also included as a standard. Following first-strand cDNA synthesis, this standard was serially diluted 1:2 in DNase-free water to generate a standard curve for the PCR analysis.

Oligonucleotide primers and taqman probe for SREBP1a and -1c were designed using Primer Express, Version 1.0 (Applied Biosystems, Warrington, U.K.) and sequences from the Genebank database (accession nos. U00968 and S66167). For quantitation of SREBP1a and -1c isoforms, the same reverse primer and fluorogenic probe but different forward primers were used. The sequences were as follows: SREBP1c forward–5′ CCATGGATTGCACTTTCGAA 3′, SREBP1a forward–5′ TGCTGACCGACATCGAAGAC 3′, reverse–5′CCAGCATAGGG TGGGTCAAA 3′, and probe–5′ TATCAACA ACCAAGACAG TGACTTCCCTGGC3′. The taqman probe was labeled at the 5′ end with the reporter dye FAM (6-carboxy-fluorescein) and at the 3′ end with the quencher TAMRA (6-carboxy-tetramethyl-rhodamine). Oligonucleotide primers and taqman probes for the glyceraldehyde-3-phosphate dehydrogenase and β-Actin internal controls were purchased from Perkin-Elmer. PCR was carried out in duplicate for each sample on an ABI 7,700-sequence detection system (PE Applied Biosystems). Each 25-μl reaction contained 2 μl first-strand cDNA, 1× PCR master mix, forward (900 or 300 nmol/l for SREBP1a/1c, respectively) and reverse (900 or 300 nmol/l for SREBP1a/1c, respectively) primers, and Taqman probe (125 or 50 nmol/l for SREBP1a/1c, respectively). All reactions were carried out using the following cycling parameters: 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. After PCR, standard curves were constructed from the standard reactions for each target gene and internal control by plotting Ct values, i.e., the cycle number at which the fluorescence signal exceeds background, versus log cDNA input (ng). The Ct readings for each of the unknown samples were then used to calculate the amount of either target or internal control relative to the standard. Because the amplification efficiency of SREBP1a and -1c were equal, the 1c-to-1a ratio in each sample could be calculated using the formula 2^−ΔΔCt where ΔΔCt = 1c Ct − 1a Ct. For each sample, results were normalized by dividing by the amount of target by the amount of internal control. Reproducibility of the PCR was examined by measuring the Ct readings of five samples in two different assays. Intra- and interassay coefficients of variation ranged from 0.19 to 4.2% and from 0.54 to 3.0%, respectively.

In brief, protein concentrations were measured using the Coomassie Plus Protein Assay reagent (Pierce, Rockford, IL). Equal amounts of total protein were resolved through an 8% SDS-polyacrylamide gel and transferred to Immobilon-P polyvinylidene fluoride membranes (Millipore, Bedford, MA). After a 2.5-h blocking time in blotto A (Tris-buffered saline, 0.05% Tween, and 5% milk), membranes were probed overnight with anti-SREBP1 antibodies detecting both forms of SREBP1, uncleaved and mature (cleaved) form. Identity of these bands was confirmed previously by analyzing nuclear and cytoplasmic fractions separately. The membrane was probed with an appropriate horseradish peroxidase– coupled secondary antibody. Antibody binding was visualized by ECL (Amersham Pharmacia Biotech, Little Chalfont, U.K.).

Total SREBP1 and cyclophilin were expressed as phosphoimager arbitrary units. Levels of ADD1/SREBP1c were expressed as arbitrary units of mRNA/cyclophilin. All results are presented as the means ± SE (unless specifically noted). Statistical significance was assessed by ANOVA and specific differences among groups using Bonferroni/Dunn post hoc test. Protein data were standardized to the amount seen in the control experiment. All analyses were performed with Statview statistical package. Comparisons between subjects were analyzed using nonpaired Wilcoxon’s nonparametric test or Student’s t test. The level of significance was set at P < 0.05.

In isolated human adipocytes, SREBP1 mRNA levels increased by approximately threefold in response to 24 h of exposure to 100 nmol/l insulin (Fig. 1A). Under the same experimental conditions, there was a modest increase in the expression of full-length SREBP1 protein (130 ± 12%), whereas the abundance of the mature, cleaved form of SREBP1 was greatly increased (338 ± 51%, P = 0.0015; n = 6) (Fig. 1B). The in vivo effect of 3 h of hyperinsulinemia on SREBP1 mRNA expression in skeletal muscle (vastus lateralis) from healthy subjects was also examined. Hyperinsulinemia significantly increased levels of SREBP1 mRNA in skeletal muscle (1,031 ± 82 vs. 1,961 ± 197 AU, P = 0.016; n = 4 for previous postclamp) (Fig. 1C).

SREBP1 mRNA levels were measured by RNase protection assay in subcutaneous adipose tissue from lean, obese, and obese diabetic individuals (study group A). Adipose tissue expression of SREBP1 mRNA was decreased in the obese versus lean subjects (0.11 ± 0.02 vs. 0.2 ± 0.02 AU, P < 0.0001) (Fig. 2A) and further reduced in obese diabetic patients (0.07 ± 0.01 AU, P < 0.0001 vs. control subjects) (Fig. 2B).

Having established that obesity was associated with reduced expression of total SREBP1 in whole adipose tissue, we examined whether this effect of obesity was seen in isolated mature adipocytes. Real-time, quantitative PCR assays were developed to allow RNA measurement in small amounts of human tissue and to permit specific quantitation of mRNAs encoding the human SREBP1a and -1c splice variants.

Consistent with previous studies, SREBP1c (ADD1) was the predominant isoform expressed in human subcutaneous adiopocytes (ratio of SREBP1c to -1a 20.0 ± 7.3) and skeletal muscle tissue (ratio of SREBP1c to -1a 10.3 ± 2.7). RNA from patient-group B (see above) was used to examine the specific expression of the two SREBP1 isoforms in isolated human adipocytes (Fig. 3). An inverse correlation between SREBP1c and BMI in both depots (r = −0.63, P = 0.0008) was observed. SREBP1a was expressed at much lower level than SREBP1c in isolated adipocytes, and its expression was not significantly correlated with BMI.

In contrast to adipose tissue, SREBP1 mRNA levels were similar in biopsy samples from rectus abdominus of lean (0.23 ± 0.02 AU, n = 12), obese (0.24 ± 0.02 AU, n = 15), and stable weight-reduced patients (0.23 ± 0.02 AU, n = 6) (study group C). Similarly, in an independent set of patients (study group D), SREBP1 mRNA expression levels were similar in vastus lateralis of lean (0.24 ± 0.02 AU, n = 5) and obese individuals (0.22 ± 0.03 AU, n = 6) (Fig. 4)

In contrast to the findings in obese subjects, SREBP1 mRNA levels in skeletal muscle (vastus lateralis) of diabetic patients (n = 6) were significantly reduced compared with those of control subjects (0.17 ± 0.04 vs. 0.24 ± 0.02, P < 0.05) (Figs. 4A and B). Notably, the three patients with poor metabolic control, as indicated by markedly elevated HbA_1c levels (> 10%), had the greatest reduction is SREBP1, and a negative correlation between SREBP1 and HbA_1c levels (−0.339, P = 0.036) was established (Fig. 4C). Interestingly, exposure of diabetic subjects to 3 h of supraphysiological hyperinsulinemia restored the expression of SREBP1 mRNA to levels similar to control subjects (Figs. 4A and B). Although a similar trend was observed in obese patients, this difference did not reach statistical significance.

As SREBP1 expression was decreased in adipose tissue of morbidly obese and diabetic patients, we examined whether TNF-α, a cytokine whose production is increased in the adipose tissue of obese patients and which has been implicated in the induction of obesity-associated insulin resistance, could affect SREBP1 expression. Isolated mature subcutaneous adipocytes were cultured in the absence or presence of 10 or 100 ng/ml TNF-α for 24 h, and SREBP1 isoform expression was measured by real-time RT-PCR. SREBP1c mRNA expression was reduced by 63.7 ± 18.8 and 47.2 ± 11.1% by treatment with 10 or 100 ng/ml TNF-α, respectively (P < 0.01) (Fig. 5A). In contrast, TNF-α had no effect on SREBP1a mRNA expression (Fig. 5B). The effect of TNF-α on SREBP1a and -1c was also determined in human preadipocytes that had been differentiated in vitro. As in mature adipocytes, TNF-α decreased SREBP1c mRNA expression by 83.1 ± 6.4% (P = 0.03) but had no effect on SREBP1a mRNA expression.

In contrast to its effect on SREBP1c mRNA, 24 h exposure of isolated adipocytes to TNF-α treatment had no significant effects on the expression levels of either the full-length or cleaved form of SREBP1 protein (Fig. 5C). However, TNF-α completely blocked the ability of insulin to promote cleavage of SREBP1 protein to the mature form (Fig. 5D).

SREBP1 is a transcription factor, highly expressed in metabolically active tissues, with important functions in the regulation of processes such as lipid metabolism, adipogenesis, and insulin action. Despite the growing appreciation of its importance in metabolic regulation, little information is available regarding the regulation of SREBP1 expression in human tissues, either in normal or insulin-resistant states. In the studies of human adipose tissue and skeletal muscle reported here, we have 1) established that insulin positively modulates human SREBP1 expression and proteolytic cleavage; 2) demonstrated that altered SREBP1 isoform expression is found in human insulin-resistant states, such as obesity and type 2 diabetes; and 3) shown that TNF-α blocks the effects of insulin on SREBP1 in human adipocytes and recapitulates, at least in part, the altered SREBP1 isoform expression seen in human metabolic disease states.

The ability of insulin to increase SREBP1 mRNA expression in the liver has been repeatedly demonstrated in rodents (16,24). In isolated rat hepatocytes, insulin and glucose together have been shown to induce the expression of SREBP1c (15). Our studies provide new evidence that insulin also induces SREBP1 gene expression in human isolated adipocytes ex vivo and in skeletal muscle in vivo. Furthermore, we have demonstrated that incubation of isolated adipocytes with insulin increases the amount of cleaved, transcriptionally active SREBP1 protein.

Having established the effect of insulin on human SREBP1 gene expression, we explored whether common insulin-resistant states such as obesity and type 2 diabetes were associated with impaired expression of SREBP1. Our data show that both obese normoglycemic and obese type 2 diabetic patients have decreased expression of SREBP1 mRNA in subcutaneous adipose tissue. Using a PCR-based method, we showed that SREBP1c is the most abundant isoform in subcutaneous human adipocytes and that this isoform is markedly reduced in isolated adipocytes from obese patients. Indeed, we observed that SREBP1c is negatively correlated with BMI. In contrast, SREBP1a mRNA was also detectable in human adipocytes, but it was expressed at much lower levels, and its expression level was not correlated with BMI. Altogether, these data indicate that in states of insulin resistance or deficiency, there is a specific decrease in the expression of SREBP1c, suggesting a specific role of this isoform in the mediation of the effect of insulin on lipid metabolism. The mechanism whereby insulin promotes SREBP1c expression is unclear, although recent evidence suggests that, at least in hepatocytes, increased glucose flux in response to insulin may also have a role in SREBP1 regulation (26).

Consistent with our findings in adipose tissue, SREBP1 mRNA levels in skeletal muscle of type 2 diabetic patients were also reduced, particularly in those with markedly impaired metabolic control. Interestingly, 3 h of hyperinsulinemia was able to restore SREBP1 expression to levels comparable with those of nondiabetic individuals. In contrast to the subjects with type 2 diabetes, SREBP1 levels in skeletal muscle of obese subjects was not reduced. As insulin resistance is generally more severe in type 2 diabetes than obesity, it is possible that there may be quantitative differences in the sensitivity of particular tissues to the effects of insulin resistance on SREBP1 expression. Alternatively, the regulation of SREBP1 by insulin may be fundamentally different in different tissues.

In this regard, recent studies have shown that SREBP1c expression in the liver of the highly insulin-resistant ob/ob mouse is not altered, despite the presence of profound resistance of other processes to insulin in the same tissue (27). Furthermore, we have recently observed opposite regulation of SREBP1c mRNA in adipose and hepatic tissues of ob/ob mice in response to insulin sensitizing drugs (H. Roche and A.J.V.-P., unpublished observations). By preventing lipid accumulation in adipose tissue, the selective downregulation of adipocyte SREBP1 in obesity might promote the partitioning of free fatty acids toward other tissues, such as skeletal muscle, pancreas, or liver, where through ‘lipotoxicity’ they may impair insulin action and/or secretion (28,29). While this manuscript was in preparation, Ducluzeau et al. (25) reported that SREBP1c gene expression was decreased in adipose tissue of obese and type 2 diabetic patients. Similarly, this group showed that SREBP1c was not altered in skeletal muscle of obese individuals (5) and only slightly decreased in diabetic patients. Also, Ducluzeau et al. (25) reported insulin-induced SREBP1c expression in adipose tissue and skeletal muscle of lean and obese individuals but not in diabetic patients. Our study supports these results, although we also observed restoration of SREBP1 expression in skeletal muscle of diabetic patients in response to insulin. The reason for this difference is unclear but may be related to the different degree of previous metabolic control of these patients.

TNF-α production by adipocytes increases in obese states and, through a paracrine and autocrine effects, has been implicated in the causation of obesity-associated insulin resistance (30,31). We therefore examined whether TNF-α might be capable of mediating the alterations in SREBP1 isoform expression seen in adipose tissue of obese human subjects. Incubation of human isolated adipocytes and preadipocytes differentiated in vitro with TNF-α markedly decreased the expression of SREBP1c but not SREBP1a, a pattern that recapitulates the alterations in SREBP1 isoform expression seen in vivo in obese and type 2 diabetic individuals. TNF-α also completely blocked the effects of insulin to increase SREBP1 protein in human adipocytes and inhibited insulin’s effect on cleavage of the protein to its active form. TNF-α is thought to induce insulin resistance through the production of serine phosphorylated signaling molecules that act as inhibitors of the insulin receptor tyrosine kinase (32,33). Thus, it is possible that all of the effects of TNF-α on SREBP1 expression and cleavage are mediated through its inhibitory effects on insulin signaling. In other systems, however, TNF-α has been reported to have effects on SREBPs independent of insulin signaling. For example, Lawler et al. (34) reported that SREBP cleavage occurs as part of the TNF-α–induced apoptotic program in hepatocytes. Thus, the “discrepancies” between RNA and protein data in vitro are intriguing and suggest that the regulation of SREBP1 by TNF-α may involve several pathways. However, TNF-α–specific effects on SREBP1c mRNA expression and inhibition of the insulin-induced SREBP1 cleavage indicates that TNF-α interferes with insulin regulation of SREBP1.

In summary, we have shown for the first time that insulin induces SREBP1 gene expression in isolated human adipocytes and skeletal muscle and also promotes SREBP1 cleavage in human isolated adipocytes. We have also provided evidence that common insulin-resistant states, such as obesity and type 2 diabetes, are characterized by decreased expression of SREBP1c mRNA and indicated a potential mechanism whereby TNF-α could contribute to the dysregulation of SREBP1. As SREBP1c/ADD1 appears to play such a key role in the coordination of fuel metabolism, we suggest that the decreased expression of this molecule in obese and diabetic human subjects could play an important role in the induction and/or maintenance of the insulin resistance seen in these disorders.

A.J.V.P. and S.O.R. are funded by Wellcome Trust, G.M. is funded by a fellowship from the Spanish Ministry of Education, D.B. is funded by Deutsche Forschunsgemeinschaft (DFG), and T.C. and R.H. are supported by a grant from Medical Research Services, Department of Veterans Affairs and VA San Diego Healthcare System, a grant from Chiron, and Grant M01 RR-00827 in support of the General Clinical Research center from the General Clinical Research Branch Division of Research Sources.

We acknowledge Peter Murgatroyd and Chris Lelliott for thoughtful review of the manuscript and Helen Roche for helpful discussions.