Endocrinology and Metabolism

Molecular evidence supporting the portal theory: a causative link between visceral adiposity and hepatic insulin resistance

Morvarid Kabir, Karyn J. Catalano, Suchitra Ananthnarayan, Stella P. Kim, Gregg W. Van Citters, Melvin K. Dea, Richard N. Bergman


The mechanism by which increased central adiposity causes hepatic insulin resistance is unclear. The “portal hypothesis” implicates increased lipolytic activity in the visceral fat and therefore increased delivery of free fatty acids (FFA) to the liver, ultimately leading to liver insulin resistance. To test the portal hypothesis at the transcriptional level, we studied expression of several genes involved in glucose and lipid metabolism in the fat-fed dog model with visceral adiposity vs. controls (n = 6). Tissue samples were obtained from dogs after 12 wk of either moderate fat (42% calories from fat; n = 6) or control diet (35% calories from fat). Northern blot analysis revealed an increase in the ratio of visceral to subcutaneous (v/s ratio) mRNA expression of both lipoprotein lipase (LPL) and peroxisome proliferator-activated receptor-γ (PPARγ). In addition, the ratio for sterol regulatory element-binding transcription factor-1 (SREBP-1) tended to be higher in fat-fed dogs, suggesting enhanced lipid accumulation in the visceral fat depot. The v/s ratio of hormone-sensitive lipase (HSL) increased significantly, implicating a higher rate of lipolysis in visceral adipose despite hyperinsulinemia in obese dogs. In fat-fed dogs, liver SREBP-1 expression was increased significantly, with a tendency for increased fatty acid-binding protein (FABP) expression. In addition, glucose-6-phosphatase (G-6-Pase) and phosphoenolpyruvate carboxykinase (PEPCK) increased significantly, consistent with enhanced gluconeogenesis. Liver triglyceride content was elevated 45% in fat-fed animals vs. controls. Moreover, insulin receptor binding was 50% lower in fat-fed dogs. Increased gene expression promoting lipid accumulation and lipolysis in visceral fat, as well as elevated rate-limiting gluconeogenic enzyme expression in the liver, is consistent with the portal theory. Further studies will need to be performed to determine whether FFA are involved directly in this pathway and whether other signals (either humoral and/or neural) may contribute to the development of hepatic insulin resistance observed with visceral obesity.

  • visceral fat
  • lipolysis
  • gluconeogenic enzymes
  • messenger ribonucleic acid
  • dogs

visceral adiposity has been associated with insulin resistance, glucose intolerance, dyslipidemia, and cardiovascular disease (15, 18, 20). The liver has specifically been implicated as a primary site of insulin resistance observed with visceral obesity. The mechanisms relating visceral fat accumulation and hepatic insulin resistance are not well known, but several possible factors might be implicated. One hypothesis states that visceral fat secretes several substances [tumor necrosis factor-α (TNF-α) (22) and/or resistin (37) or decreased adiponectin (2)] that may induce hepatic insulin resistance. The alternative “portal hypothesis” posits a high rate of lipolysis of visceral adipose tissue leading to increased delivery of free fatty acids (FFA) to the liver via the portal vein, thus contributing to increased fat accumulation and liver insulin resistance (3, 7, 8). The role of FFA and secretory factors released by visceral fat in the pathogenesis of liver insulin resistance requires clarification.

Consistent with the portal hypothesis, several studies have shown that FFA turnover and lipolysis are higher in visceral than in subcutaneous fat and that the visceral adipose depot is less sensitive to the antilipolytic effect of insulin (31). Mittelman et al. (29) demonstrated in vivo that local infusion of insulin into the abdominal visceral depots suppressed lipolysis, only at much higher insulin concentrations compared with controls. Thus accumulation of visceral fat and an increased rate of adipocyte lipolysis in the visceral depot could lead to increased FFA flux to the liver. Chronic exposure of the liver to elevated FFA can promote liver gluconeogenesis (44), deplete enzymes involved in FFA oxidation, and increase hepatic lipogenesis (10, 45). In addition, elevated FFA are known to increase liver triglyceride content (30) and decrease insulin clearance (28), both of which are associated with insulin resistance.

Clearly, the portal hypothesis is a strong candidate for the explanation of hepatic resistance in the face of increased visceral adiposity. However, the molecular evidence for the portal hypothesis is minimal, and studies addressing this hypothesis in the rodent are technically challenging. The dog genome exhibits a relatively high degree of conservation with that of the human (25); therefore, the dog model may serve well to understand the etiology of human disease while still allowing for detailed molecular study. Recently, our laboratory has demonstrated that the fat-fed dog is a powerful model to examine the role of obesity in insulin resistance longitudinally (24). The purpose of the present study was to determine whether transcriptional changes that occur in fat and the liver during fat feeding support the portal hypothesis. For this we measured several genes involved in lipid and glucose metabolism in adipose and liver tissue from moderate-fat-fed dogs vs. those fed a standard chow diet. In visceral and subcutaneous fat depots, we measured gene expression of the lipid-storing enzyme lipoprotein lipase (LPL), two transcription factors involved in lipid accumulation, peroxisome proliferator-activated receptor-γ (PPARγ) and sterol regulatory element-binding transcription factor 1 (SREBP-1), and a lipid mobilizing enzyme, hormone-sensitive lipase (HSL).

In the liver, we evaluated the expression of two gluconeogenic enzymes, glucose-6-phosphatase (G-6-Pase) and phosphoenolpyruvate carboxykinase (PEPCK). In addition, we examined the transcription factor SREBP-1, involved in liver lipogenesis, and fatty acid-binding protein (FABP) and carnitine palmitoyltransferase I (CPT I), involved in FFA oxidation. Triglyceride content in the liver was also assessed. Finally, to examine the efficacy of cell surface insulin receptors, we examined insulin receptor binding in liver tissue of fat-fed and control dogs.



Tissue biopsies were obtained from six male mongrel dogs fed a moderate-fat diet for 12 wk. Data depicting metabolic changes throughout the 12-wk diet are published elsewhere (28, 40). Six control dogs fed a normal chow diet were used as a reference group (14). Animals were housed under controlled kennel conditions (12:12-h light-dark cycle) at the University of Southern California (USC) Keck School of Medicine vivarium. All procedures were approved by the USC Institutional Animal Care and Use Committee.


Upon arrival, dogs underwent a period of acclimation for 3 wk, during which time they were fed a standard chow diet consisting of one-half of a can (∼200 g) of Hill’s Prescription Diet (9% protein, 8% fat, 10% carbohydrate, 0.3% fiber; Hills Pet Nutrition, Topeka, KS) and dry chow ad libitum (≤900 g/day: 26% protein, 15% fat, 40% carbohydrate, 3% fiber; Wayne Dog Food; Alfred Mills, Chicago, IL). The control diet consisted of ∼3,858 kcal, 39% from carbohydrates, 26.5% from protein, and 34.5% from fat. After 3 wk, dogs either continued to receive the control diet (n = 6) or received one with a modest fat supplementation (n = 6) provided via addition of 2 g cooked bacon grease/kg initial body wt to the control diet (obtained from the USC Keck School of Medicine cafeteria). Addition of fat to the diet increased total energy +13% to ∼4,360 kcal/day and calories from fat to ∼42%.

Body weight and metabolic parameters.

Body weight was measured on a weekly basis. All analyses were performed on blood samples drawn after an overnight fast (∼12 h) and stored at −70°C until analysis. Glucose, FFA, glycerol, and insulin were measured as previously described (28). Study of gene expression was performed in a cross-sectional manner, utilizing dogs from separate metabolic studies. Therefore, insulin sensitivity was assessed via the frequently sampled intravenous glucose tolerance test in moderate-fat-fed dogs, as published elsewhere (28, 40), and the euglycemic hyperinsulinemic clamp in control dogs (14). Values were converted to similar units for comparison (4).

RNA isolation.

At the end of the study, dogs were killed and tissue samples obtained from subcutaneous and visceral fat depots as well as from liver for assessment of gene expression. Tissues were immediately frozen in liquid nitrogen upon removal and stored at −80°C. RNA was extracted from frozen tissues using the Tri-Reagent Kit (Molecular Research Center, Cincinnati, OH). Total RNA concentration was quantified by spectrophotometric absorbance measurement at 260 nm. The 260-to-280 nm absorption ratio of all preparations ranged between 1.8 and 2.0. RNA integrity was assessed by gel electrophoresis using agarose-ethidium bromide gel.

Isolation and cloning of probes.

Genes of interest were isolated from total RNA of adipose tissue and liver RNA. Each cDNA was synthesized from 1 μg of total RNA in 50 μl of final incubation volume by means of reverse transcriptase-polymerase chain reaction (RT-PCR; Advantage One-Step RT-PCR Kit; Clontech Laboratories, Palo Alto, CA). The reverse transcription was carried out under the following conditions: 50°C for 1 h and 94°C for 15 min, and amplification of cDNA was performed in 35 temperature cycles (94°C for 30 s, 57°C for 30 s, 68°C for 1 min) and one cycle at 72°C for 2 min. The PCR products were analyzed by gel electrophoresis using 2% agarose and purified by a gel purification kit (Qiagen, Valencia, CA). PCR products were subcloned into pT-Adv using the AdvanTAge PCR Cloning Kit (Clontech Laboratories) and linearized using EcoRI. Clone products of interest (three clones for each sequence) were sequenced in duplicate using an automated ABI Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA). Characteristics of the dog gene fragments are presented in Table 1. Primer pairs were designed by aligning sequences for rat, mouse, and human (when possible) and looking for conserved regions for each gene of interest.

View this table:
Table 1.

Characteristics of dog gene sequences

Northern blot analysis.

Equal amounts (20 μg for liver tissue and 15 μg for fat tissue) of total RNA were fractionated on 1% agarose gel under denaturing conditions, using the NorthernMax Kit (Ambion, Austin, TX). Gels were blotted to a nylon membrane (Hybond-XL Nylon Membrane; Amersham Pharmacia Biotech). For all genes (LPL, PPARγ, SREBP-1, HSL, G-6-Pase, PEPCK, and β-actin) cDNA was labeled with [α-32P]dCTP using a multiprime DNA labeling system kit (DECAprime II; Ambion). Membranes were prehybridized (2 h at 42°C) and then hybridized overnight at 42°C with Ultrahyb Solution (Ambion). Hybridized membranes were washed in 1× SSC-0.1% SDS twice (20 min at 42°C) and exposed to BioMax MS Double Emulsion Photographic Film (Eastman Kodak, Rochester, NY) for 7 h at −80°C using intensifying screens. Radioactivity in each band was quantified by Scion Image software, and the fold change for each mRNA was calculated after correction for loading variation by use of the 18S rRNA and β-actin probe.

Semiquantitative RT-PCR.

In addition to gluconeogenic enzyme gene expression, we investigated the expressions of both CPT I, a gene involved in FFA oxidation, and FABP. First-strand complementary DNA was generated from 0.2 μg of RNA in a 50-μl volume with the Advantage One-Step RT-PCR Kit (Clontech Laboratories). All primer sequences and PCR conditions are shown in Table 1. Synthesis of cDNA by reverse transcription was performed under the following conditions: 50°C for 1 h and 94°C for 15 min. Cycle number for cDNA amplification was dependent on the gene being studied (32 for CPT I and FABP; 27 for β-actin). Each cycle was performed under the following conditions: 94°C for 30 s, 57°C for 30 s, 68°C for 1 min and one cycle at 72°C for 2 min. All PCR reactions were optimized and duplicate aliquots amplified in the linear phase. PCR products were electrophoresed on 1.5% agarose gel and visualized using ethidium bromide, and relative intensities of transcript signals were compared quantitatively using Scion Image software. Target gene mRNA was expressed as the ratio of signal intensity relative to that of β-actin.

Liver triglyceride content.

Triglyceride content was assessed in liver samples obtained at the time of biopsy and determined as described by Szczepaniak et al. (39).

Insulin receptor binding.

Liver insulin receptor binding was measured as described by Kadama et al. (26), with some modifications. Crude liver membranes (∼1 mg protein/tube), 20 nCi/tube of 125I-labeled human insulin (Novo Nordisk, Bagsvaerd, Denmark) and graded amounts of unlabeled bovine pancreatic insulin were incubated in 0.4 ml of 50 mmol/l Tris·HCl buffer solution (pH 7.4) containing 2% (wt/vol) BSA and 2 mg/ml bacitracin (Sigma, St. Louis, MO). After a 16-h incubation at 25°C, samples were centrifuged at 1000 g for 30 min. The sample pellets were washed with 1 ml of the same buffer and measured for membrane-bound radioactivity. Nonspecific binding was determined in the presence of an excess of unlabeled insulin (2 μg/tube). Insulin binding and affinity were evaluated by Scatchard analysis of the dose-response data.

Statistical analysis.

Data are expressed as means ± SE. Significance was established using the unpaired Student’s t-test for between-group comparisons and the paired Student’s t-test for comparisons between week 0 and week 12 of the moderate-fat diet, with significance set at P < 0.05.


Body weight and metabolic parameters.

Body weight and metabolic parameter data from control and fat-fed dogs are summarized in Table 2. No differences were observed in body weight, fasting plasma glucose, or systemic FFA levels between the two groups. However, moderate-fat feeding increased fasting insulin by nearly 20% compared with controls. As calculated in separate studies (14, 28, 40), insulin sensitivity (SI) was markedly impaired in fat-fed compared with control dogs. It should be noted that, despite an apparent difference in mean values, SI was relatively low and not significantly different between groups before the nutritional period (e.g., week 0: 2.07 ± 0.40 vs. 1.29 ± 0.12 dl·min−1·pM−1, P = 0.08).

View this table:
Table 2.

Body weight and metabolic parameters after 12 wk of control or moderate-fat diet

Adipose tissue and liver gene expression.

Seven genes were partially cloned and sequenced for the first time during this study, and results were reported in GenBank (Table 1). We observed high homology between dog and human sequences (range 85–91%). In all gene expression measurements, fat feeding did not alter expression of either of the control genes 18S or β-actin.

To determine whether alterations in adipose tissue gene expression might explain some of the changes in adiposity and sensitivity observed previously in moderate-fat-fed dogs, we measured the v/s mRNA ratio for HSL, LPL, and two transcription factors involved in lipid accumulation, PPARγ and SREBP-1 (Fig. 1). Under control conditions, the v/s expression ratios for HSL and LPL approached unity. In contrast, the v/s ratio for LPL and HSL was twofold greater in dogs fed fat, suggesting augmented lipolytic activity in the visceral fat depot relative to subcutaneous fat (P < 0.05). Animals on the control diet exhibited slightly higher PPARγ and SREBP-1 expression in visceral vs. subcutaneous adipose tissue. The v/s ratio for PPARγ was markedly increased approximately fourfold with fat feeding, demonstrating molecular evidence for increased lipid accumulation in the visceral vs. subcutaneous depot with moderate fat (P < 0.05). In addition, the SREBP-1 expression ratio tended to increase, although this change did not reach significance (P = 0.09).

Fig. 1.

Visceral (Vis)/subcutaneous (Sub) adipose tissue mRNA ratio of hormone-sensitive lipase (HSL), lipoprotein lipase (LPL), peroxisome proliferator-activated receptor-γ (PPARγ), and sterol regulatory element-binding transcription factor-1 (SREBP-1) in moderate-fat-fed and control dogs (n = 6 for both). Total RNA was isolated from the visceral and subcutaneous adipose tissue of fat-fed and control dogs under overnight-fasting conditions. Levels of mRNA were detected by Northern blot analysis. To account for loading differences, all values were normalized to ribosomal mRNA. Data are means ± SE. *P < 0.05 vs. control.

To determine whether moderate-fat feeding and liver insulin resistance was associated with an increased potential for lipid deposition in the liver, we measured the expression of genes involved in fatty acid sequestration. FABP tended to be elevated in livers of fat-fed animals (P = 0.06). In addition, the liver transcription factor SREBP-1 was 38% greater in fat-fed compared with control livers (P < 0.01; Fig. 2A).

Fig. 2.

A: total RNA from livers of fat-fed (n = 6) and control dogs (n = 6) under overnight-fasting conditions. Levels of SREBP-1 mRNA were detected by Northern blot analysis, and fatty acid-binding protein (FABP) gene expression was quantified by semiquantitative RT-PCR. All values were normalized to ribosomal mRNA and β-actin. Data are means ± SE. *P < 0.05 vs. control. B: expression of phosphoenolpyruvate carboxykinase (PEPCK), glucose-6-phosphate (G-6-Pase), SREBP-1 and carnitine palmitoyltransferase I (CPT I) in fat-fed and control dogs after an overnight fast (n = 6 for both). Levels of PEPCK, G-6-Pase, and SREBP-1 mRNA were detected by Northern blot analysis, whereas CPT I gene expression was quantified by semiquantitative RT-PCR. All values were normalized to ribosomal mRNA and β-actin. Data are means ± SE. **P < 0.05 vs. control.

Fat feeding profoundly elevated the expression of the gluconeogenic enzymes G-6-Pase and PEPCK in the liver. Both enzyme expressions were increased to nearly the same extent; PEPCK increased by 161% and G-6-Pase by 166% (P < 0.001; Fig. 2B). Thus rate-limiting enzymes for gluconeogenesis were increased in coordination with visceral lipolyitc enzyme expression. Interestingly, there was no difference observed between control and fat-fed animals for CPT I expression, a known modulator of fatty acid oxidation in the liver (Fig. 2B).

Liver triglyceride content.

To determine whether the moderate-fat diet increased lipid deposition in the liver, triglyceride levels were measured. Liver triglyceride content increased by 45% in the fat-fed compared with the control dogs (P < 0.05; Fig. 3).

Fig. 3.

Liver triglyceride contents in fat-fed and control dogs (n = 6 for both). Data are means ± SE. *P < 0.05 vs. control.

Insulin receptor binding.

All fat-fed dogs demonstrated significantly impaired insulin sensitivity (28). In addition, hepatic insulin clearance was reduced (28). Consistent with these results, we observed a 50% decrement in the number of insulin receptors in the fat-fed dog compared with controls. Receptor affinity, however, remained unchanged (Table 3).

View this table:
Table 3.

Changes in hepatic insulin-binding capacity and affinity in control and moderate-fat-fed dogs


The association between visceral fat and insulin resistance has been widely reported (5, 14); however, the mechanism of this association is not understood. Although it is possible that insulin resistance causes deposition of visceral fat, it appears more likely that accumulation of fat in the abdomen induces insulin resistance (33). It has been suggested that accumulated visceral fat elicits a signal that in turn results in liver insulin resistance. Although it is possible that the signal is a protein product of visceral adipose cells, so-called adipokines, it is also possible that FFA themselves are responsible for hepatic insulin resistance. The present study examines this latter possibility at a molecular level, looking at the expression levels of proteins involved in FFA metabolism in the liver and adipose tissue of the moderate-fat-fed dog.

Our results demonstrate differential expression of several genes involved in FFA metabolism of both the visceral adipose depot and the liver in control vs. fat-fed dogs. Although there were no differences between visceral and subcutaneous gene expression in control dogs, Northern blot analysis revealed that the ratio of visceral to subcutaneous expression of LPL, HSL, and PPARγ increased twofold in fat-fed dogs, suggesting increased relative lipid turnover of the visceral fat depot. Furthermore, genes in the liver that are influenced by FFA availability were also elevated (PEPCK 161% increase; G-6-Pase 166% increase) compared with control dogs. Triglyceride deposition was augmented (45%) in fat-fed livers, providing further support for potentially increased lipid delivery to the liver. Moreover, we previously demonstrated that the fat-fed dog model not only exhibited reduced insulin sensitivity but also reduced hepatic insulin clearance (23, 28). Consistent with these data, we find here that liver insulin receptor binding is decreased by 50% in fat-fed dogs. Thus our studies provide circumstantial evidence supporting the so-called portal theory.

In this study, we partially cloned and sequenced seven canine genes that exhibited high homology with their human equivalents. Consistent with our findings, recent sequencing of the dog genome revealed high homology between the dog and human genome (25). These data further support usage of the dog model to study human metabolic disorders. Moreover, the dog model offers a unique advantage, in that one is able to study longitudinal metabolic derangements and associate these changes with molecular studies.

Previously, our laboratory has shown (24) that moderate-fat feeding causes a substantial increase (69%) in visceral fat, as measured by magnetic resonance imaging after 12 wk of diet, without any significant changes in body weight. To elucidate the molecular mechanism of fat storage and accumulation in visceral and subcutaneous fat depots, we studied several genes involved in lipid accumulation in this model. In the moderate-fat-fed dog, we demonstrated that fatty acid storage and lipid accumulation are higher in visceral compared with subcutaneous fat depots. In adipocytes, lipid storage is almost entirely dependent on uptake of fatty acids released from hydrolysis of circulating triglyceride-rich lipoproteins by LPL (17). Although not significant, LPL expression was slightly elevated in visceral compared with subcutaneous fat in control dogs. However, this difference was enhanced two times by fat feeding, suggesting preferential fat accumulation in the visceral fat depot in the moderate-fat-fed dog model.

In addition to LPL, we also examined two transcription factors involved in adipocyte lipid accumulation and adipocyte differentiation, PPARγ and SREBP-1. In addition to stimulating differentiation, thus augmenting adipocyte number, PPARγ regulates the expression of LPL and a number of genes involved in lipid metabolism (27). PPARγ gene expression increased significantly in the visceral fat depot in the fat-fed dog. Consistent with our results, expression of PPARγ has been shown to increase in adipose tissue of insulin-resistant rodents and humans (41, 42). Thus we suggest that an increased visceral PPARγ gene expression in the fat-fed dog enhanced adipocyte differentiation and lipid accumulation. SREBP-1 also plays a role in adipocyte differentiation via stimulation of PPARγ activity (36). In addition, enzymes of fatty acid synthesis, such as fatty acid synthase, appear to be regulated by the products of the SREBP-1 gene in liver and adipose tissue (19, 35). We found similar results for SREBP-1 gene expression; SREBP-1 tended to increase in visceral/subcutaneous fat (P = 0.09). Taken together, these data are consistent with higher fat accumulation in the visceral fat depot.

If the portal theory is operative, it is anticipated that the rate-limiting enzyme for lipolysis (HSL) should be elevated. Indeed, we found that HSL expression was increased twofold in visceral fat with fat feeding. Although we did not determine the protein level or activity of HSL in the present study, we can infer that the overwhelming increase in transcription of HSL corresponds to greater lipolysis in visceral fat tissue of fat-fed animals. In accord with this notion, Berraondo et al. (6) demonstrated a positive correlation between HSL mRNA levels and lipolytic capacity in high-fat-fed rats. In addition, HSL protein expression is at least partly determined by HSL mRNA expression in human fat cells (9).

It would be reasonable to conclude that overall visceral depot lipolysis is increased in fat-fed dogs due to increased SREBP-1 and PPARγ expression in this fat depot as well as increased gene expression of the key regulators of FFA release (HSL and LPL). Therefore, even if the lipolytic rate of individual cells remained unchanged, the augmented cell number would lead to an increase in absolute lipolysis from the viscera, thus leading to increased delivery of FFA to the liver. Although direct measures of lipolysis could not be made in this study, these data do provided molecular evidence for elevated portal FFA. Further investigation will be necessary to test this hypothesis further.

Exposure of the liver to elevated FFA can alter the expression of genes involved in FFA metabolism. Here, we demonstrate that moderate-fat feeding increased the expression of genes important for FFA sequestration in the liver. FABP binds to fatty acids and facilitates trafficking between cells. The moderate-fat-fed dog tended to exhibit elevated FABP expression. Similarly, others have demonstrated that liver FABP protein levels are increased by high-fat diet (1), and in vitro evidence suggests that FFA upregulate liver FABP gene expression (11). Moreover, it is clear from studies in the transgenic SREBP-1 mouse and studies with fasting and refeeding protocols that SREBP-1 regulates fatty acid synthesis in the liver (21); in fact, fatty acid synthase appears to be regulated by SREBP-1 in liver (19, 35). Consistent with these data, we observed increased liver SREBP-1 gene expression in fat-fed dogs. Additionally, it has been demonstrated that overexpression of liver PPARγ in the mouse induces adipogenic transformation of hepatocytes, leading to adipose tissue-specific gene expression and increased lipogenesis and lipid accumulation (41). Interestingly, we observed only a nonsignificant (∼20%) increase in liver PPARγ expression (data not shown). Nevertheless, it is possible that this difference would have reached significance with a greater change in visceral fat and/or longer time on the diet. Thus our data demonstrating increased expression of proteins involved in FFA metabolism suggest potentially elevated FFA present at the liver.

Hence, FFA resulting from the increased lipolysis and lipid turnover in visceral fat may play a role in the mechanism of hepatic insulin resistance. Increased FFA may exert dysfunction in the liver via several mechanisms, including stimulation of gluconeogenesis and/or fatty acid oxidation (44). In addition, FFA are known to increase liver triglyceride content (30) and decrease insulin clearance (28, 43).

A recent study in our laboratory (24) showed that dogs fed a moderate-fat diet for 12 wk exhibit a complete inability of insulin to suppress glucose production during physiologically hyperinsulinemic, but euglycemic, conditions. Although they were not measured in that study, we proposed that elevated portal FFA levels could induce hepatic insulin resistance by promoting gluconeogenesis. In the present study, we demonstrate that the expression of key gluconeogenic enzymes PEPCK and G-6-Pase increased by 161 and 166%, respectively, in fat-fed dogs. These results are in accord with other studies demonstrating increased basal gluconeogenesis in fat-fed rats (32). It has also been shown that gluconeogenesis can be stimulated by FFA oxidation (12). Therefore, we evaluated the expression of CPT I, an enzyme involved in FFA β-oxidation. However, semiquantitative RT-PCR revealed that the expression of CPT I was not altered by the moderate-fat diet. Consistent with this, Oakes et al. (30) have shown that treatment with etomoxir, a known fatty acid oxidation blocker, does not improve hepatic insulin resistance in fat-fed rats. Furthermore, it is important to note that CPT I is regulated posttranslationally by malonyl-CoA, as well as by its own concentration. Therefore, gene expression can be considered only an estimate of the state of this enzyme. Further studies will be necessary to determine whether CPT I protein levels and activity are altered in the present model. Alternatively, these data might also suggest that a likely source of hepatic insulin resistance in this model is simply an alteration in the gluconeogenic pathway.

One might also expect that elevated FFA delivery would induce an elevation of triglyceride deposition in liver; excess FFA may accumulate as triglyceride if they are not oxidized by mitochondria or peroxisomes. It has been shown that increased liver fat content is associated with hepatic insulin resistance (34). Consistent with these data, we found that fat feeding resulted in a 45% elevation of liver triglyceride in fat-fed, insulin-resistant dogs. although these data are consistent with increased FFA delivery to the liver via the portal vein, we cannot negate the possible effects of triglycerides in the fat diet per se to triglyceride accumulation in the liver.

Excess FFA are associated not only with liver insulin resistance but also with decreased hepatic insulin clearance (30). Previously, our laboratory demonstrated a significant decrease in first-pass hepatic insulin extraction after 12 wk of fat feeding (23). Insulin receptor binding is necessary for degradation in the hepatocyte (16). In the current study, insulin receptor binding was decreased by 50% in fat-fed dogs, as has been demonstrated elsewhere (13), effectively reducing the capacity of the liver to clear insulin. Furthermore, Svedberg et al. (38) have shown that FFA can interfere directly with insulin binding and degradation in rat hepatocytes in vitro, possibly implicating the portal FFA in the reduction in clearance.

In conclusion, a moderate-fat diet nearly doubled transcription of genes involved in lipid turnover in the visceral vs. subcutaneous adipose tissue, indicating a possible increase in FFA flux to the liver. The depot-specific differences in the expression of LPL, PPARγ, and HSL genes in the fat-fed dog arise through a mechanism that requires further analysis. We also demonstrated enhanced gluconeogenic enzyme gene expression, liver triglyceride content and decreased liver insulin receptor binding. Future studies will attempt to examine the effect of this diet on protein levels of genes studied here. Furthermore, although this study posits that elevated portal FFA are the basis of hepatic insulin resistance observed with fat feeding and obesity, we acknowledge that, because it is experimentally challenging to measure portal FFA, we have yet to make longitudinal measurements in this model. Future studies are being performed to examine portal FFA concentrations as well as the expression of other important factors, such as TNF-α, leptin, resistin, and adiponectin, to clarify the roles of adipose tissue secretory factors in insulin resistance.


This work was supported by the research grants of R. N. Bergman from the National Institute of Diabetes and Digestive and Kidney Diseases (DK-27619 and DK-29867). M. Kabir performed this work during a postdoctoral fellowship supported by a mentor-based grant from the American Diabetes Association.


We thank Linda Kirkman and Doug A. Davis for assistance with surgeries and animal handling.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract