Early Hepatic Insulin Resistance Precedes the Onset of Diabetes in Obese C57BLKS-db/db Mice

C57BL/6J-db/db (B6-db) and C57BLKS/J-db/db (BKS-db) mice were purchased from The Jackson Laboratory. Mice were maintained in the animal care facilities of the University of California and VA Greater Los Angeles Healthcare System on a 12-h light/dark cycle. The mice were maintained on a standard diet (Harlan Teklad LM-485). Before all experiments, mice were fasted for 4 h (9:00 a.m. to 1:00 p.m.) unless indicated otherwise. Protocols were approved by the Institutional Animal Care and Use Committees of the respective institutions.

Plasma lipids, glucose, and insulin were determined as described previously (14). The Luminex 100 system was used to measure plasma leptin, resistin, tumor necrosis factor-α, and interleukin-6 levels. Adiponectin, corticosterone (R&D Systems, Minneapolis, MN), and glucagon (ALPCO Diagnostics, Salem, NH) were determined by enzyme-linked immunosorbent assays. Hepatic glycogen content was assayed at the Mouse Metabolic Phenotyping Center at Yale University. Hepatic triglyceride (TG) content was assayed using the L-type TG H kit (Wako Diagnostics, Richmond, VA) with modifications. Briefly, liver tissue was homogenized in ice-cold PBS with a Polytron homogenizer. After Folch extraction, samples were centrifuged and the lower (organic) phase was transferred to new tubes. The organic volume was dried in vacuum and re-dissolved in 90 μl of 10% Triton X-100 in isopropanol by vortexing. After the addition of 1.5 ml Wako Reagent A, samples were vortexed and incubated at 37° C for 5 min. Samples were vortexed until clear and 0.75 ml Wako Reagent B was added followed by incubation at 37°C for 30 min. Samples were vortexed again until clear and then put on ice until measurement of optical density at 600 nm. A calibration curve was established using the Wako Lipid Calibrator.

Fatty acid (FA) synthesis and oxidation were measured in liver slice cultures (15).

After an overnight fast, mice were anesthetized with isoflurane and killed by cervical dislocation. The left caudal lobe of the liver was rapidly excised, and fresh liver sections of uniform thickness (1 mm, averaging ∼60 mg) were obtained using a Stadie-Riggs microtome. The liver sections were weighed and immediately incubated at 37°C for 40 min in Krebs-Henseleit buffer under 95% O₂:5% CO₂ that contained 5.5 mmol/l glucose, 3% BSA, 1 mmol/l oleate, and [¹⁴C-U]-palmitic acid (2.5 μCi/ml). The average time from euthanizing the animals to getting the liver section into the buffer was <2 min. Rates of FA oxidation were assessed by determining ¹⁴CO₂ trapped in hyamine hydroxide saturated filter paper as described previously (16).

Liver sections were obtained and incubated essentially as described above, except that ³H₂O (0.5 mCi/vessel) was used as the radioactive tracer and incubations were carried out for 1 h (17). Media and tissue were then collected and total lipids extracted as described above. Total lipid isolates were then hydrolyzed in an acid/acetonitrile solution as described previously (18), and the lipids were re-extracted and dried under nitrogen and radioactivity in the fatty acid fraction determined by liquid scintillation spectrometry.

TG secretion was determined by comparing plasma TG levels at 0 and 30 min after intravenous injection of Triton WR1339 (Tyloxapol, Sigma Chemical) as described (19). Calculations assumed plasma volume of 3.5% of body weight.

Clearance of TG (10 μl peanut oil/g body weight delivered by gavage) was determined by following plasma TG levels at hourly intervals after the administration of lipid.

Ten units of heparin was administered by tail-vein injection followed by retro-orbital blood collection 10 min later as described (20). Plasma lipase activities were determined as described below.

Total RNA was isolated with Trizol (Invitrogen) and reverse transcribed using the Omniscript RT kit (Qiagen). Quantitative real-time RT-PCR was performed as described previously (21) using primers shown in supplementary Table 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0878/DC1. Each sample was measured in triplicates. Expression of the TATA-box binding protein (Tbp) and hypoxanthine guanine phosphorybosyl transferase (Hprt) genes were used for normalization.

Euglycemic clamps were performed in conscious 1-month-old mice using a 10 mU/kg/min insulin dose as described (22).

Hepatocytes were isolated by a two-step perfusion protocol that is based on minor modifications of the method described by Seglen (23). Briefly, mice were anesthetized by intraperitoneal injection of pentobarbital (50 mg/kg). The abdominal cavity was opened and the liver perfused via the portal vein first with a Ca²⁺ Mg⁺⁺ free EDTA chelating solution, followed by perfusion with a Ca²⁺ Mg⁺⁺ replete buffer containing type I collagenase (Worthington Biochemical, Lakewood, NJ). The hepatocytes were then dissociated by cutting the liver capsule and gently shaking in the collagenase solution. A series of five low-speed (50g) centrifugations were performed to wash and differentially sediment hepatocytes from other cell types, particularly Kupffer and endothelial cells. The resulting cell pellet was re-suspended, and an aliquot was taken to determine cell number and viability by trypan blue exclusion. The cells were plated at a density of 2.0 × 10⁶ cells on 60-mm collagen-coated plates (BD Biosciences, San Jose, CA) in hepto-stim media (BD Biosciences, San Jose, CA) containing 10% FBS (ATCC, Manassas, VA) and supplemented with pen/strep (ATCC, Manassas, VA) and fungizone (Invitrogen, Carlsbad, CA) and allowed to adhere for 4 h. The media were then removed, the cells washed once with PBS, and fresh media added. The following day, the media were changed and replaced with the same media except without the fungizone. After 24 h in culture, cells were serum-starved overnight in 199 medium supplemented with glutamine, NaHCO₃, and pen/strep. After incubation in fresh serum-free media for 2 h, cells were stimulated with 100 nmol/l of porcine insulin for 10 min in triplicate wells. Cells were solubilized in lysis buffer (50 mmol/l Tris-HCl, pH 7.5, 150 mmol/l NaCl, 1% NP-40) containing inhibitors of proteases (Complete Mini; Roche, Indianapolis, IN) and phosphatases (PhosStop; Roche). Protein concentration was determined by the BCA assay (Pierce, Rockford, IL).

Tissue and cell lysates were analyzed with Western blotting. Primary antibodies against insulin receptor substrate (IRS)-1 (1:500 dilution, Upstate Biotechnology, Lake Placid, NY), IRS-2 (1:2,000 dilution, Upstate), Akt (1:5,000 dilution, Cell Signaling Technologies, Danvers, MA), and phospho-Akt (1:1,000 dilution, CST) were used followed by goat anti-rabbit–horseradish peroxidase (1:10,000 dilution; Jackson ImmunoResearch, West Grove, PA) secondary antibody. Blots were visualized using the ECL+ chemiluminescent detection system (Roche).

The activity of lipoprotein lipase (LPL) and hepatic lipase was measured using the respective triolein substrates prepared by sonication (24). To discriminate between LPL and hepatic lipase activities further, we took advantage of the difference in salt sensitivity between these lipases. Lipase activities were measured using the LPL substrate in the presence and absence of 1 mol/l NaCl; LPL activity was calculated as the salt-inhibited fraction. Tissues were homogenized in 25 mmol/l Tris-HCl, pH 7.5, containing 10% glycerol, 0.2% deoxycholic acid (sodium salt), and 10 μ/ml heparin. Homogenization (100 mg/ml) was carried out for 30 s on ice, using a Tekmar tissue homogenizer, followed by centrifugation for 10 min at 10,000g to remove insoluble debris and, in the case of adipose tissue, to separate the floating lipid layer from the lysate.

Data are presented as mean ± SE. Differences between two groups were considered statistically significant at P < 0.05 using Student's t test. ANOVA was used for multiple group comparisons followed by Bonferroni's post hoc test, as implemented in the SigmaPlot software package (Systat Software, San Jose, CA). Non-normally distributed trait values were transformed before analysis as indicated.

To characterize the development of diabetes in the BKS-db strain, we performed a metabolic time-course study (Fig. 1, line graphs, and supplementary Table 3). B6-db mice were used as controls, because they are diabetes resistant despite their genetic similarity to BKS-db. Consistent with previous reports (2,4), BKS-db and B6-db mice exhibited similar body weights and adiposity, with the exception of male BKS-db, which started losing weight at 3 months because of severe diabetes (Fig. 1,A and B). Despite similar degrees of obesity, BKS-db and B6-db mice displayed dramatic differences in glucose homeostasis. B6-db mice compensated for insulin resistance associated with obesity by sharply increasing plasma insulin levels (Fig. 1,D). In contrast, insulin levels in BKS-db gradually declined and were associated with progressive hyperglycemia (Fig. 1,C). These data are consistent with previous studies implicating impaired β-cell viability in BKS-db as the primary determinant of diabetes susceptibility (10,11,13). However, our time-course study also revealed metabolic differences between the strains at 1 month of age, before the onset of diabetes in BKS-db (Fig. 1, right panels). At this age, BKS-db mice maintained similar glucose levels, but over twofold higher insulin concentrations in comparison to B6-db mice (Fig. 1,C and D). Moreover, these mice exhibited elevated plasma TG (Fig. 1,E), but lower free fatty acid Fig. 1 F and total and HDL cholesterol levels (data not shown) relative to their B6-db counterparts. To assess the effects of strain background and genetic interaction with the db mutation, we also phenotyped lean B6 and BKS mice (supplementary Table 2). Among the traits analyzed, only HDL cholesterol and insulin levels were different between wild-type B6 and BKS mice, whereas total cholesterol, free fatty acid, and glucose traits revealed significant strain × db interactions. In conclusion, our results indicate β-cell–independent metabolic derangements in BKS-db mice. To investigate the role of different tissues in the early metabolic differences observed between B6-db and BKS-db, we further characterized 1-month-old mice in subsequent studies.

Differences in plasma insulin and lipid levels are frequently associated with different insulin response in different tissues. To directly investigate insulin sensitivity in young BKS-db and B6-db mice, hyperinsulinemic-euglycemic clamps were performed. Upon infusion with 2.5 mU/kg/min insulin, the two strains exhibited similar metabolic parameters (data not shown). However, administration of increased doses of insulin (10 mU/kg/min) unexpectedly revealed elevated glucose infusion rate and insulin-stimulated whole-body glucose turnover in BKS-db, indicating better systemic insulin sensitivity in this strain (Table 1). Consistent with these results, insulin-stimulated glucose uptake in gastrocnemius muscle was also higher in BKS-db mice, suggesting that muscle tissue is likely responsible for their relative insulin sensitivity. In contrast, the clamp showed severe hepatic insulin resistance in both strains. In fact, even high-dose insulin infusion failed to achieve any suppression of hepatic glucose production in either strain. In conclusion, these data indicate relative muscle insulin sensitivity in BKS-db, but severe hepatic insulin resistance in both strains.

As insulin sensitivity in muscle is expected to protect against, rather than contribute to, the diabetic phenotype of BKS-db, we focused on liver metabolism in subsequent experiments. To investigate hepatic insulin resistance, we compared insulin signaling in the two strains. Immunoblot analysis revealed lower levels of both IRS-1 and IRS-2 proteins in BKS-db liver, suggesting reduced proximal insulin signaling in this strain (Fig. 2,A). To confirm this prediction, we assessed downstream signaling in primary hepatocytes treated with insulin. Consistent with reduced IRS expression, insulin-stimulated Akt phosphorylation was reduced by ∼50% in BKS-db hepatocytes (Fig. 2 B). These results indicate a higher degree of hepatic insulin resistance in BKS-db mice.

Next, we investigated the metabolic consequences of elevated hepatic insulin resistance in BKS-db. Upon visual examination of internal organs, BKS-db livers appeared darker than their B6-db counterparts (Fig. 3,A), suggesting reduced steatosis in the former. Indeed, despite higher plasma insulin levels in BKS-db mice, hepatic TG content was over threefold lower in this strain (Fig. 3,B). Similarly, hepatic glycogen content was also significantly reduced in BKS-db mice (Fig. 3 E).

Because plasma free fatty acid levels are lower in BKS-db mice, reduced hepatic TG content may reflect reduced substrate delivery for TG biosynthesis. Alternatively, metabolic differences intrinsic to liver may also contribute to reduced steatosis in BKS-db mice. To discriminate between these possibilities, we used short-term in vitro liver slice cultures, which allow examination of metabolic pathways under conditions of controlled substrate delivery (15). We first examined hepatic lipogenesis by measuring the incorporation of radioactivity into de novo synthesized FAs from ³H₂O. Despite elevated plasma insulin levels in vivo, the rate of hepatic lipogenesis was significantly lower in BKS-db liver slices (Fig. 3,C), suggesting intrinsic metabolic differences in hepatic lipid synthesis between the strains. Hepatic FA oxidation was not significantly different between the two strains, although there was a trend toward increased oxidation in BKS-db (Fig. 3 D). We also tested the potential involvement of circulating factors in the observed metabolic differences by measuring plasma levels of various adipokines and hormones implicated in hepatic metabolism (25). Plasma concentrations of adiponectin, leptin, resistin, corticosterone, and glucagon were similar in B6-db and BKS-db mice (supplementary Table 3). Taken together, our results indicate altered lipid and carbohydrate metabolism consistent with elevated hepatic insulin resistance in BKS-db mice.

To determine if altered gene expression underlies the metabolic differences observed between BKS-db and B6-db mice, we measured mRNA levels of genes involved in hepatic lipid and carbohydrate metabolism. Consistent with reduced lipogenesis in BKS-db mice, expression levels of all lipogenic genes tested (liver X receptor-α [LXRα], sterol regulatory element binding protein-1c [Srebp1c], peroxisome proliferator-activated receptor-γ [Pparγ], acetyl-CoA carboxylase-1 [AccI], fatty acid synthase [Fas], ATP citrate lyase [Acl], stearoyl-CoA desaturase-1 [Scd1], and cytosolic malic enzyme [Me1]) were significantly lower in this strain (Fig. 4 A). The expression of enzymes involved in TG synthesis, including glycerol-3-phosphate acyltransferase (Gpat) and lipin (Lpin1), was also lower in BKS-db. As Srebp1c is a direct transcriptional activator of lipogenic genes and Gpat (26), suppressed Srebp1c expression is likely to be responsible for diminished hepatic lipid synthesis in BKS-db mice. Other targets of Srebp1c, such as glucokinase (Gk) and the FA transporter Cd36, also exhibited reduced expression in this strain. In contrast, mRNA levels of acyl-CoA:diacylglycerol acyltransferase-2 (Dgat2), a TG biosynthetic enzyme not known to be regulated by Srebp1c, were similar in the two strains.

We also analyzed gluconeogenic gene expression (Fig. 4,B). After a 4-h fast, which results in more than twofold higher insulin levels in BKS-db than in B6-db mice (Fig. 4,B, left panel), the mRNA levels of phosphoenopyruvate carboxykinase (Pepck), fructose-1,6-bisphosphatase (Fbp), and glucose-6-phosphatase (G6p) were similar in the two strains (Fig. 4,B, right panel). Conversely, similar plasma insulin levels after an overnight fast were associated with elevated expression of Pepck and Fbp, but not G6p, in BKS-db mice (Fig. 4 B). In conclusion, our gene expression results indicate diminished hepatic insulin action in BKS-db mice. Consistent with a proximal defect in insulin signaling, hepatic insulin resistance affects both the lipogenic and anti-gluconeogenic actions of insulin in this strain.

BKS-db mice exhibit twofold higher plasma TG levels than B6-db mice at 1 month of age (Fig. 1,E, right panel). Considering the lower hepatic TG content in BKS-db mice (Fig. 3,B), this result was initially surprising. To investigate the mechanisms underlying these differences, we measured hepatic TG secretion in 1-month-old mice after inhibiting clearance with Triton WR1339. This analysis indicated similar rates of hepatic TG secretion in B6-db and BKS-db mice (Fig. 5,A). As plasma TG levels are determined by the relative rates of secretion and clearance, we also measured plasma TG clearance after an oral lipid load. In B6-db mice, lipid ingestion resulted in a modest (less than twofold) rise in plasma TG levels peaking at 2 h after gavage and returning to baseline levels after 4 h (Fig. 5 B). In contrast, BKS-db mice exhibited an over fourfold increase in TG at 2 h, and TG levels remained elevated threefold after 4 h. These data indicate significantly reduced clearance of circulating TG in BKS-db mice.

To test the potential role of lipases in reduced TG clearance in BKS-db mice, we measured the activities of these enzymes in postheparin plasma. Consistent with reduced TG clearance, total lipase, LPL, and hepatic lipase activities are significantly lower in BKS-db plasma (Fig. 5,C). To identify the tissue(s) responsible for differences in LPL activity, we assayed lipase activity in isolated tissues (Fig. 5 D). Adipose and, to a smaller extent, heart from BKS-db mice exhibit lower LPL activities, whereas skeletal muscle shows the opposite pattern. In conclusion, these data suggest that elevated plasma TG levels in BKS-db mice result from reduced clearance primarily due to low LPL activity in adipose. Further studies will be required to explore the underlying mechanisms in this tissue.

Mutations in the leptin receptor gene cause severe obesity and insulin resistance in rodents and humans (27,–29). In leptin receptor–deficient (db mutation) mouse strains, obesity is variably associated with hyperglycemia depending on genetic background (30). Diabetes-resistant strains, such as B6-db, exhibit islet hypertrophy and insulin hypersecretion and are able to maintain mildly elevated plasma glucose levels. In contrast, islets in the diabetes-susceptible BKS-db strain undergo atrophy leading to insulin deficiency and hyperglycemia. The genes underlying these phenotypic differences are largely unknown. Based on reduced viability and proliferative capacity of BKS β-cells both in vivo and in vitro, diabetes susceptibility of this strain has been ascribed to genetic determinants acting in islets (10,,–13). However, it is conceivable that genetic differences manifested in other tissues may also contribute to β-cell failure through mechanisms that increase metabolic stress on these cells (e.g., insulin resistance). Indeed, in the current study, we demonstrate increased hepatic insulin resistance in young BKS-db mice compared with B6-db mice and propose that this metabolic abnormality may lead to elevated β-cell stress and the diabetic phenotype in BKS-db.

Side-by-side phenotypic characterization of BKS-db and B6-db mice revealed differences in plasma lipid and insulin levels as early as 1 month of age, even before the onset of β-cell dysfunction in BKS-db mice. These observations indicated metabolic differences in tissues other than β-cells and prompted us to investigate the early BKS-db phenotype further. Unexpectedly, hyperinsulinemic-euglycemic clamp studies demonstrated higher whole-body and muscle insulin sensitivity in BKS-db, indicating that this tissue does not contribute to, but rather counteracts, the diabetes-sensitive phenotype of this strain. However, the clamp also revealed profound hepatic insulin resistance, which completely abolished the effect of high doses of insulin on hepatic glucose production in both strains. However, in contrast to muscle, hepatic insulin resistance is more severe in BKS-db than B6-db mice. Despite elevated plasma insulin levels, BKS-db mice exhibit signs of reduced hepatic insulin action including diminished lipogenesis, glycogen accumulation, insulin-stimulated gene expression and TG secretion, and suppression of gluconeogenic gene expression. Furthermore, direct analysis of hepatic insulin signaling demonstrated lower IRS-1 and IRS-2 expression as well as insulin-stimulated Akt phosphorylation in BKS-db hepatocytes. Diminished IRS-1/2 expression has been linked to reduced hepatic insulin signaling in various obese animal models in comparison to their lean counterparts (31,,–34). Thus, our data suggest that lower IRS-1/2 expression in BKS-db mice results in relative hepatic insulin resistance in this strain.

Hepatic insulin resistance is a principal component of type 2 diabetes (35). Decreased insulin sensitivity in liver leads to elevated hepatic glucose production, hyperinsulinemia, increased β-cell stress, and hyperglycemia (36). Animal models indicate that hepatic insulin resistance can play a primary role in the development of diabetes (37). For example, diminished insulin signaling due to hepatic insulin receptor (38) or IRS-1/2 deficiency (34,39) causes hyperglycemia. In line with these studies, we propose that early hepatic insulin resistance in BKS-db mice contributes to the diabetes-susceptible phenotype of this strain.

What are the metabolic consequences of hepatic insulin resistance in BKS-db mice? Although we did not directly demonstrate it in our study, hepatic glucose output is expected to be higher in BKS-db mice than in B6-db mice. Diminished insulin signaling in BKS-db liver results in impaired suppression of gluconeogenic gene expression and likely elevated glycogenolysis, consistent with lower hepatic glycogen content observed in this strain. Another relevant finding is markedly reduced lipogenic capacity in the BKS-db liver. It has been proposed that active lipogenesis reduces fuel partitioning into the gluconeogenic pathway, thereby reducing the rate of glucose production at the substrate level (40). Thus, reduced lipogenesis resulting from proximal defects in insulin signaling in BKS-db mice may contribute to elevated gluconeogenesis, increased β-cell stress, and the development of diabetes in this strain. Consistent with this hypothesis, decreased hepatic lipogenesis and steatosis are also associated with diabetes susceptibility in the obese BTBR-ob (40), lipoatrophic FVB-AZIP (41), and liver-specific Pparγ knockout mouse models (42). Moreover, increased hepatic lipogenesis and steatosis achieved by overexpressing Srebp1c (43) or the administration of Pparγ (44,45) or LXR agonists (46) markedly improve glycemia in diabetic mice.

In addition to β-cells and liver, metabolic abnormalities in other tissues may also contribute to the development of diabetes in BKS-db mice. For example, our results suggest that low LPL activity in BKS-db adipose tissue is associated with reduced clearance and elevated levels of plasma TG, which may inflict lipotoxic stress in β-cells (47). Further studies will be needed to address the potential contribution of adipose tissue to the diabetic phenotype.

This work was supported by National Institutes of Health Grants DK071673 and DK80756, the Cedars-Sinai Medical Center, and American Diabetes Association Grants 1-05-RA-96 and 7-07-RA-80.

No potential conflicts of interest relevant to this article were reported.

The authors thank Dr. Nicole Ehrhardt for expert statistical analysis.