A Role for iNOS in Fasting Hyperglycemia and Impaired Insulin Signaling in the Liver of Obese Diabetic Mice

Male ob/ob mice and lean wild-type C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The ob/ob mice were backcrossed onto wild-type C57BL/6J mice for at least 30 generations. The institutional animal care committee approved the study protocol. The mice were housed in mesh cages in a room maintained at 25°C and illuminated in 12:12-h light-dark cycles; they were provided with standard rodent diet and water ad libitum.

Male ob/ob mice at 10 weeks of age were treated with a specific inhibitor of iNOS, l-NIL (60 mg/kg body wt, i.p., b.i.d.; Cayman Chemical, Ann Arbor, MI), or PBS for 10 days. Following 10 days of treatment, the animals treated with l-NIL (n = 17) or PBS (n = 18) were fasted overnight and then received an injection of insulin (Humulin R; Eli Lilly, Indianapolis, IN; 15 units/kg body wt in PBS with 0.1% BSA) or vehicle (PBS with 0.1% BSA) via the portal vein under anesthesia with pentobarbital sodium (70 mg/kg body wt, i.p.). At 5 min after insulin injection, the liver, gastrocnemius muscle, and epididymal fat were taken for biochemical analyses. Male ob/ob mice at 10 weeks of age were also treated with l-NIL or PBS for 10 days under a pair-fed condition. After 10 days of treatment, the liver was taken from pair-fed mice under fed condition or after overnight fasting. Twenty-four mice were randomly assigned into four groups (l-NIL fasted, PBS fasted, l-NIL fed, and PBS fed) of six animals. The average food intake of both l-NIL and PBS-treated animals under pair-fed condition was 5.6 g/day.

After 10 days of treatment, the animals treated with l-NIL (n = 10) or PBS (n = 10) were fasted for 4 h and then received an injection of insulin (2.5 units/kg body wt in PBS with 0.1% BSA, i.p.). At 0, 15, 30, 60, 90, and 120 min after insulin injection, blood samples were collected to measure glucose level.

Liver samples were homogenized as previously described (28). Immunoprecipitation and immunoblotting were performed (29) using anti-IRβ, IRS-1 and -2, p85 phosphatidylinositol 3(PI3)-kinase, iNOS, endothelial NOS, neuronal NOS (Upstate, Lake Placid, NY), Akt, phosphorylated Akt (Ser473) (Cell Signaling, Beverly, MA), and phosphotyrosine antibodies (Santa Cruz, Santa Cruz, CA) (online appendix available at http://diabetes.diabetesjournals.org).

Partial purification of iNOS was performed using immobilized 1400W beads (Calbiochem, San Diego, CA), which selectively bind to iNOS, according to the manufacturer’s instruction (online appendix).

PI3-kinase activities in the immunoprecipitates with anti–IRS-1 or -2 antibody were measured in vitro (30).

Immunostaining with anti-nitrotyrosine antibody (Upstate) was performed according to the manufacturer’s instruction (online appendix).

RNase protection assay was performed as previously described (11) using the RPA III kit (Ambion, Austin, TX). mRNA levels of IRS-1 and -2 and sterol regulatory element–binding protein (SREBP)-1c were normalized to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (online appendix).

Mouse Hepa1c1c7 and human HepG2 cells (American Type Culture Collection, Manassas, VA) were maintained in α-minimum essential medium and minimum essential medium supplemented with 10% fetal bovine serum, respectively. After 8 h serum starvation, the cells were treated with the indicated concentrations of S-nitrosoglutathione (GSNO) or 3-morpholino-sydnonimine (Cayman Chemical, Ann Arbor, MI) in the presence or absence of ODQ (1 μmol/l; Sigma) or 8-bromoguanosine-3′,5′-cyclic monophosphorothioate (Br-cGMP; 100 μmol/l; Axxora, San Diego, CA) for 24 h, unless otherwise indicated. Hepa1c1c7 cells were transfected with pCDNA3/iNOS or pCDNA3 using Lipofectamin 2000 (Invitrogen). pCDNA3/iNOS was kindly provided by Dr. Kone. At 48 h after transfection, the cells were harvested.

Blood glucose level and plasma insulin concentration were determined by the glucose oxidase method (Ascensia Elite Glucometer; Bayer, Elkhart, IN) and enzyme-linked immunosorbent assay (Crystal Chem, Chicago, IL), respectively. Glycogen content was measured using a glucose hexokinase assay kit (Sigma) (31). A Glycerol phosphate oxidase trinder reagent (Sigma) was used to measure triglyceride (32). The concentrations of adiponectin and free fatty acids in serum from fasted animals were determined using an enzyme-linked immunosorbent assay kit (Linco Research, St. Charles, MO) and an enzymatic colorimetric assay kit (Wako Chemicals, Richmond, VA), respectively.

Statistical analyses were performed using Student’s t test or one-way ANOVA followed by Fisher’s protected least significant differences test. Values of P < 0.05 were considered statistically significant. Data are expressed as means ± SE.

We found a 2.5-fold increase in iNOS protein expression in the liver of ob/ob mice compared with wild-type mice (Fig. 1A). No difference was found in the protein expression of endothelial or neuronal NOS between ob/ob and wild-type mice (online appendix Fig. 1). iNOS expression was also increased in skeletal muscle and adipose tissue of ob/ob mice compared with wild-type mice to the extent comparable to that in the liver (online appendix Fig. 2).

The immunoreactivity for nitrotyrosine, a marker of nitrosative stress, was also elevated in the liver of ob/ob mice compared with wild-type mice (Fig. 1B). Treatment with iNOS inhibitor (60 mg/kg body wt, i.p., b.i.d., for 10 days) reversed the elevated nitrotyrosine immunoreactivity in the liver of ob/ob mice compared with PBS. l-NIL treatment, however, did not alter the expression of iNOS and endothelial and neuronal NOS (data not shown). These results suggest that l-NIL, a competitive inhibitor for iNOS, effectively reduced the activity of iNOS in the liver of ob/ob mice without altering iNOS expression level, as expected.

At the inception of the treatment, there was no difference in fed blood glucose level (l-NIL: 292.8 ± 21.8; PBS: 290.3 ± 18.7 mg/dl), plasma insulin concentration (l-NIL: 57.0 ± 0.5; PBS: 57.0 ± 0.5 ng/ml), or body weight (l-NIL: 57.0 ± 0.5; PBS: 56.8 ± 0.6 g) between l-NIL –and PBS-administered animals. Treatment with l-NIL did not affect body weight at the time of its termination (l-NIL: 57.2 ± 0.5; PBS: 57.0 ± 0.6 g). There was no statistical difference in food intake between l-NIL –and PBS-treated animals (l-NIL: 5.8 ± 0.2; PBS: 6.4 ± 0.2 g/day), although food intake in l-NIL–treated animals appeared to be lower than in PBS-treated animals. However, l-NIL treatment reversed fasting hyperglycemia and significantly ameliorated fasting hyperinsulinemia in ob/ob mice (Fig. 2A and B). The insulin sensitivity index (fasting blood glucose × plasma insulin concentration) was also significantly improved by l-NIL (l-NIL: 16.9 ± 4.0; PBS: 51.0 ± 12.6 mg²/l², P < 0.05). Improved insulin sensitivity by l-NIL treatment was confirmed by insulin tolerance test. ob/ob mice treated with l-NIL responded to insulin with a more profound decrease in blood glucose level than in PBS-administered ob/ob mice (Fig. 2C). Area under the curve analysis in response to the insulin tolerance test also revealed the beneficial effects of l-NIL (Fig. 2D).

To exclude a possible influence of food intake, the effects of l-NIL treatment were also examined in pair-fed ob/ob mice. The beneficial effects of l-NIL were corroborated in pair-fed mice. In pair-fed ob/ob mice, l-NIL treatment also led to the reversal of fasting hyperglycemia (l-NIL: 95.3 ± 13.4; PBS: 210.8 ± 18.1 mg/dl, P < 0.05) and significant amelioration of fasting hyperinsulinemia (l-NIL: 11.4 ± 3.2; PBS: 19.4 ± 1.9 ng/ml, P < 0.05) and insulin sensitivity index (l-NIL: 10.3 ± 2.8; PBS: 40.8 ± 5.9 mg²/l², P < 0.05). Under fed condition, l-NIL treatment appeared to mitigate hyperglycemia in ob/ob mice (l-NIL: 195.7 ± 29.1; PBS: 272.2 ± 24.0, P < 0.10), although no statistical significance was found. Plasma insulin concentration under fed condition tended to be higher in l-NIL–treated animals than PBS-treated animals (l-NIL: 86.4 ± 7.0; PBS: 57.5 ± 12.4 ng/ml, P < 0.10), although there was no statistical difference.

iNOS inhibitor, l-NIL, did not alter the protein expression of insulin receptor (IR). However, the protein expression of IRS-1 and -2 under a fasted condition was increased by 1.5- and 2-fold, respectively, compared with PBS-treated animals (Fig. 3A). Increased protein expression of IRS-1 and -2 by l-NIL was further confirmed in pair-fed mice. In pair-fed ob/ob mice, the protein expression of IRS-1 and -2 in l-NIL–treated animals was greater than in PBS-treated animals under both fasted and fed conditions (Fig. 3B and online appendix Fig. 3). l-NIL treatment resulted in upregulated mRNA level of IRS-2 under both fasted and fed conditions (Fig. 3C and D). However, the mRNA level of IRS-1 was not altered by l-NIL treatment. l-NIL did not affect GAPDH mRNA level (data not shown). No difference was found in the protein expression of p85, p55α, and p50α PI3-kinase in the liver between ob/ob mice treated with l-NIL and PBS (online appendix Fig. 4).

Insulin-stimulated tyrosine phosphorylation of IRS-1 and -2 in the liver was upregulated ∼1.5- and 3-fold by iNOS inhibitor, respectively, compared with PBS (Fig. 4). However, iNOS inhibitor did not alter insulin-stimulated tyrosine phosphorylation of IR. l-NIL treatment did not significantly alter basal (insulin-unstimulated) tyrosine phosphorylation of IR and IRS-1 and -2. The ratio of insulin-stimulated tyrosine phosphorylation to protein abundance for IR and IRS-1 and -2 was not significantly altered by l-NIL treatment (online appendix Fig. 5). These results suggest that the improvement by iNOS inhibitor in insulin-stimulated tyrosine phosphorylation of IRS-1 and -2 may be mainly attributable to the increased expression of IRS-1 and -2. Insulin-stimulated IRS-1–and IRS-2–associated PI3-kinase activity was also enhanced by l-NIL up to 1.5- and 2.5-fold, respectively (Fig. 4D and E). l-NIL treatment also increased insulin-stimulated phosphorylation of Akt/protein kinase B, a downstream signaling molecule, while the expression of Akt/protein kinase B was not altered (Fig. 4F and G).

Glycogen content in the liver of fasted animals was significantly increased by l-NIL treatment (l-NIL: 14.9 ± 1.8; PBS: 7.3 ± 0.6 mg/g liver wt, P < 0.01). Triglyceride content in the liver of fasted animals appeared to be lower in l-NIL–treated animals than in PBS-treated animals (l-NIL: 52.9 ± 5.6; PBS: 65.7 ± 4.0 μg/g protein, P < 0.10), although there was no statistical difference. In skeletal muscle of ob/ob mice, l-NIL treatment increased IRS-1 protein expression, while IRS-2 abundance was unaltered. In adipose tissue, neither IRS-1 nor -2 expression was affected by l-NIL treatment (online appendix Fig. 6). We did not find a significant difference in circulating adiponectin (l-NIL: 6.32 ± 0.6; PBS: 6.00 ± 0.45 ng/ml) and free fatty acid (l-NIL: 1.83 ± 0.1; PBS: 1.71 ± 0.12 mEq/l) concentration between l-NIL–and PBS-administered ob/ob mice.

Next, we asked whether iNOS can downregulate IRS-1 and -2 expression in cultured hepatocytes. Exposure to NO donor, GSNO (1 mmol/l), for 24 h decreased the protein expression of IRS-1 and -2 in Hepalc1c7 and HepG2 cells (Fig. 5 and online appendix Fig. 7), whereas the expression of IR and p85 PI3-kinase was not affected by NO donor. Treatment with another NO donor, 3-morpholino-sydnonimine-1 (0.5 mmol/l), for 24 h also decreased the expression of IRS-1 and -2 (online appendix Fig. 7). Ectopically expressed iNOS reduced the expression of IRS-1 and -2 without altering p85 expression. The inhibitory effects of GSNO on IRS-1 and -2 expression were dose and time dependent (Fig. 6). In accordance with the greater increase in IRS-2 expression than in IRS-1 by l-NIL treatment in the liver of ob/ob mice (Fig. 3), the inhibitory effects of NO donor or iNOS transfection appeared to be more prominent on IRS-2 than -1 (Figs. 5 and 6). Treatment with 1 mmol/l GSNO for 24 h did not affect viability of the cells, as judged by the trypan blue exclusion test. Consistent with the effects of iNOS inhibitor on IRS-1 and -2 mRNA in the liver of ob/ob mice, 1 mmol/l GSNO decreased IRS-2 mRNA level but did not affect IRS-1 mRNA level in cultured Hepa1c1c7 cells (Fig. 7).

To further characterize the effects of NO donor on IRS-1 and -2, we examined the effects of ODQ, a soluble guanylate cyclase inhibitor, and Br-cGMP, a cell-permeable cGMP analog in Hepa1c1c7 cells. ODQ (1 μmol/l) failed to inhibit GSNO-induced reduction in IRS-1 and -2 expression. cGMP analog, Br-cGMP (100 μmol/l), did not affect the expression of IRS-1 and -2, either. These observations suggest that the inhibitory effects of GSNO are cGMP independent.

Since a recent study (33) revealed that SREBPs suppress the transcription of IRS-2 in the liver, we examined the effects of l-NIL and GSNO on SREBP-1c expression. SREBP-1c mRNA level was increased by GSNO (1 mmol/l) in cultured Hepa1c1c7 cells (Fig. 7C) and was reduced by l-NIL treatment in the liver of ob/ob mice under both fasted and fed conditions (Fig. 7D and E).

We found that iNOS expression was significantly elevated in the liver as well as skeletal muscle and adipose tissue in ob/ob mice compared with wild-type mice (Fig. 1) and that iNOS inhibitor reversed fasting hyperglycemia and ameliorated whole-body insulin resistance in ob/ob mice (Fig. 2). The improved glycemic control was accompanied by increased protein expression of IRS-1 and -2 and enhanced IRS-1–and IRS-2–mediated insulin signaling in the liver of ob/ob mice (Figs. 3 and 4). Moreover, we found that exposure to an NO donor and ectopically expressed iNOS reduced the protein expression of IRS-1 and -2 in cultured hepatocytes (Figs. 5 and 6). These results indicate that increased expression of iNOS may play an important role in obesity-induced fasting hyperglycemia and suggest that iNOS may contribute to the defects in hepatic insulin signaling, in particular, depressed expression of IRS-1 and -2 in ob/ob mice.

Fasting blood glucose levels are determined primarily by glucose output from the liver, and hepatic glucose output is suppressed by insulin in insulin-sensitive animals (21,22,25). Therefore, fasting hyperglycemia is accounted for by the failure of insulin to inhibit the production and release of glucose by the liver. A predominant role for defects in hepatic insulin signaling in fasting hyperglycemia is also supported by previous studies in mice with tissue-specific gene targeting of IR. Tissue-specific gene disruption of IR in skeletal muscle (34) or adipose tissue (35) was not associated with hyperglycemia, whereas liver-specific knockout of IR did exhibit hyperglycemia (36,37). Taken together, our findings suggest that increased iNOS expression may contribute to fasting hyperglycemia, in part, by exacerbating deranged insulin signaling in the liver of ob/ob mice.

In the liver, IRS-2 rather than IRS-1 plays a prominent role in metabolic actions of insulin, including the inhibition of hepatic glucose output, whereas IRS-1 has a major role in skeletal muscle (38–41). Previous studies (42–44) showed a marked reduction in IRS-2 expression in the liver of ob/ob mice, whereas IRS-1 expression was unaltered or modestly decreased. These findings suggest that defective IRS-2–mediated insulin signaling is a major component of obesity-related hepatic insulin resistance.

iNOS inhibitor improved the protein expression of both IRS-1 and -2 in the liver of ob/ob mice. However, the beneficial effect of iNOS inhibitor on IRS-2 seemed more prominent compared with that on IRS-1 in the liver. Similarly, the reduction in protein expression induced by NO donor or iNOS transfection was more profound in IRS-2 compared with IRS-1 in Hepa1c1c7 and HepG2 cells. These findings suggest that iNOS may downregulate preferentially the expression of IRS-2 and, to a lesser extent, that of IRS-1 in the liver.

To date, limited knowledge is available about the regulatory mechanisms of IRS-2 expression in normal and disease states such as insulin resistance and diabetes. Previous studies (25,45,46) showed that insulin downregulates IRS-2 expression. Therefore, one can hypothesize that the improved expression of IRS-2 might be due to amelioration of hyperinsulinemia by l-NIL via as yet determined mechanisms rather than direct effects of iNOS inhibitor in the liver. However, our data argue against this hypothesis. NO donor and iNOS transfection reduced IRS-2 expression in cultured hepatocytes, suggesting that increased iNOS expression without hyperinsulinemia can reduce IRS-2 expression in the liver. Moreover, l-NIL increased IRS-2 expression at both protein and mRNA levels under fed conditions as well as in fasted animals, although the effects of l-NIL on plasma insulin level appeared to differ between fasted and fed ob/ob mice. Fasting plasma insulin concentration was significantly lower in l-NIL–treated ob/ob mice than in PBS-treated animals, while plasma insulin concentration under fed condition tended to be higher in l-NIL–treated animals, albeit with no statistical significance.

Recently, SREBPs were shown to negatively regulate the transcription of IRS-2 in the liver (33). We found that SREBP-1c expression was reduced by l-NIL in the liver in vivo and was elevated by NO donor in cultured hepatocytes (Fig. 7). In aggregate, therefore, it is possible that increased expression of IRS-2 may be associated with reduced SREBP-1 expression by iNOS inhibitor in the liver of ob/ob mice.

Protein expression of IRS-1 was increased by iNOS inhibitor in the liver of ob/ob mice, while NO donor and ectopically expressed iNOS decreased IRS-1 in cultured hepatocytes. However, mRNA of IRS-1 was not altered by iNOS inhibitor or NO donor. This is consistent with our preliminary observation that iNOS and NO donor decreased IRS-1 expression via proteasome-dependent proteolysis in skeletal muscle cells (H. Sugita, M.K., unpublished observations).

The effects of l-NIL on the expression of IRS-1 and -2 in skeletal muscle and adipose tissue were distinct from those in the liver. In contrast to the preferential improvement in IRS-2 expression in the liver, l-NIL treatment was associated with increased expression of IRS-1, but not of IRS-2, in skeletal muscle. In adipose tissue, l-NIL failed to modulate the expression of either IRS-1 or -2. These results in skeletal muscle are in agreement with our aforementioned unpublished observation that NO donor and iNOS transfection reduced IRS-1 expression but did not affect IRS-2 expression in skeletal muscle cells. Based on the predominant role of hepatic insulin signaling in fasting hyperglycemia, one can reasonably speculate that ameliorated expression of IRS-2 in the liver may be the more important contributor to the reversal of fasting hyperglycemia by l-NIL rather than improved IRS-1 expression in skeletal muscle. It is conceivable, however, that the protective effects of iNOS inhibitor in the skeletal muscle as well as in the liver may contribute in concert to improve whole-body insulin sensitivity in ob/ob mice.

The study of Perreault and Marette (12) showed that high-fat diet–induced elevation in iNOS expression was restricted to skeletal muscle and adipose tissue, but not observed in the liver in mice, in contrast to our results in ob/ob mice (Fig. 1). This discrepancy might be attributable to the methodological difference to induce obesity, namely leptin mutation in ob/ob mice versus high-fat diet. It is also possible that the difference in mouse strains could influence the results. We used ob/ob mice that were backcrossed onto C57BL/6J background, while C57BL/6X129SvEv mice were used by Perreault and Marette. Moreover, in contrast to the glucose-lowering effects of iNOS inhibitor (Fig. 2), iNOS disruption did not significantly alter high-fat diet–induced fasting hyperglycemia (12). The lack of iNOS induction in the liver might account for the ineffectiveness of iNOS disruption to mitigate fasting hyperglycemia in high-fat diet–fed mice, although iNOS inhibitor effectively reversed fasting hyperglycemia in ob/ob mice in which iNOS expression was elevated in the liver.

Notably, there is a marked difference in the severity of fasting hyperglycemia between ob/ob mice and high-fat diet–fed mice (12). In ob/ob mice, overt fasting hyperglycemia was associated with increased iNOS expression in the liver. But in mice fed a high-fat diet, fasting hyperglycemia was mild, and iNOS expression did not increase in the liver (12). Thus, iNOS induction in the liver seems to correlate with the magnitude of fasting hyperglycemia. Collectively, one can speculate that increased expression of iNOS in the liver might be a culprit for overt fasting hyperglycemia via suppressing IRS-1 and -2 expression. Further studies will be required to clarify this point.

In summary, the present data show that iNOS inhibitor reversed fasting hyperglycemia in parallel with the concomitant improvement in hepatic IRS-1–and IRS-2–mediated insulin signaling in ob/ob mice. The present study sheds a new light on a therapeutic potential of iNOS inhibitor to improve glycemic control in obesity-related diabetes.

This work was supported by grants from the National Institutes of Health (R01DK058127 to M.K. and R01GM031569, R01GM055082, and R01GM061411 to J.A.J.M.) and grants from the Shriners Hospital for Children to J.A.J.M.