Chronic Exposure to Interleukin-6 Causes Hepatic Insulin Resistance in Mice

Anti-IRβ (C-19) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-insulin receptor substrate (IRS)-1, anti-IRS-2, and anti-phosphotyrosine (clone 4G10) antibodies were from Upstate Biotechnology (Lake Placid, NY). Anti-phosphoSTAT3 (Tyr705) was purchased from Cell Signaling Technology (Beverly, MA). Recombinant human IL-6 was purchased from RDI systems (Flanders, NJ). All other chemicals were from Sigma except where indicated differently.

Male C57BL/6 mice purchased from the Jackson Laboratory (Bar Harbor, ME) were used in all studies. The age of animals used in the study ranged from 8–14 weeks for the chronic infusion pump experiments. All animal procedures were in accordance with University of Rochester animal care guidelines and approved by the animal care and use committee.

For all experiments examining chronic IL-6 exposure, Alzet osmotic pumps (model no. 2001; Durect, Cupertino, CA) with a 7-day pumping capacity and an infusion rate of 1 μl/h were used. Pumps were filled to capacity with 16 μg/ml hIL-6 diluted in carrier (0.9% NaCl and 0.1% BSA). Following induction of halothane general anesthesia, pumps were implanted into the intrascapular subcutaneous space. Incisions were closed with interrupted absorbable sutures.

Two models were used for acute insulin delivery, and both were performed under halothane general anesthesia. For portal insulin treatment the abdomen was opened through a midline incision and the portal vein identified. Insulin (at 300 ng) or carrier (0.9% NaCl and 0.1% BSA) was injected into the portal vein using a 28-gauge needle. Following injection, direct pressure was applied to the portal vein to prevent bleeding. Animals were killed 45 s after portal vein injection. Samples were placed in liquid nitrogen within 15 s. For liver/muscle comparisons a systemic insulin injection was used. The left femoral vein was identified by sharp dissection in the left inguinal region, and the same quantity of insulin was injected intravenously using a 28-gauge needle. Hemorrhage was again prevented with direct pressure from a cotton swab. Tissues were harvested at 90 s after femoral vein injection and frozen in liquid N₂ within 15 s.

Frozen liver and muscle were homogenized in 16 volumes (weight/volume) (liver) or 10 volumes (muscle) of lysis buffer (100 mmol/l HEPES pH 7.5, 150 mmol/l NaCl, 1% triton-X-100, 100 mmol/l NaF, 2 mmol/l EDTA, 2 mmol/l EGTA, 10% glycerol, 1 mmol/l benzamidine, 1 mmol/l tetrasodium pyrophosphate, 1 mmol/l PMSF, 5 mmol/l pervanadate, and 1× protease inhibitor cocktail I [Calbiochem]). Frozen tissues were homogenized using the Brinkman PT 10/35 Polytron. Extracts were kept ice-cold at all times. Extracts were cleared by microcentrifugation at 15,000g for 10 min at 4°C.

Protein content of extracts was determined by the Bradford method (14). Total extract protein (4 mg) was used for all immunoprecipitations. Extracts and immunoprecipitates were resolved by SDS-PAGE and transferred to nitrocellulose. Proteins were detected by immunoblotting and visualized using enhanced chemiluminescence (Amersham-Pharmacia).

Mice were fasted for 3 h before test. Mice were anesthetized under halothane and insulin (Novagen, Madison, WI) diluted in carrier (0.9% NaCl, 0.1% BSA) was injected intraperitoneally at 0.7 units/kg. Blood glucose levels were determined at indicated intervals using an Accu-Chek (Roche Indianapolis, IN) glucose meter and test strips on tail bleeds of awake mice.

Animals were fasted overnight (16 h). In the morning, a subset of fasted mice were given standard diet and allowed to feed freely for 90 min. Mice were anesthetized, subjected to tail bleeds for glucose levels, and immediately killed for removal of the liver. Total RNA was extracted from the livers of both fasted and refed animals using Trizol (Invitrogen, Carlsbad, CA) extraction. Total RNA (20 μg) was run on a 1% agarose gel containing 1× MOPS buffer and 0.66 mol/l formaldehyde. RNA was transferred to a Zeta-Probe (BioRad, Hercules, CA) membrane overnight, then cross-linked using a Stratalinker (Stratagene, La Jolla, CA). The glucokinase cDNA was a generous gift from Jacob E. Freidman (University of Colorado Health Sciences Center). Hybridization was done at 68°C using Express-Hyb hybrization buffer (Clontech, Palo Alto, CA). For quantitation, the membranes were exposed to a phosphoimager cassette (Molecular Dynamics, Sunnyvale, CA) and scanned in a Storm 840 phosphoimager (Molecular Dynamics). Quantitation was performed using ImageQuant software (Molecular Dynamics). Blots were stripped in boiling 2× SSC buffer with 0.1% SDS until no counts remained. Normalization was done following similar quantitation using a probe to GAPDH.

Following a 5-day constant infusion of hIL-6 at a rate of 16 ng/h, animals underwent general anesthesia and plasma was collected by cardiac puncture with a 25-gauge needle. Human IL-6 and mouse IL-6 Quantikine enzyme-linked immunoassays (R&D Systems, Minneapolis, MN) were used to measure the level of exogenous hIL-6 and endogenous mouse IL-6, respectively. As stated by the manufacturer, there is no cross-reactivity between human and mouse IL-6 in their respective assays. Circulating recombinant hIL-6 is reported as the mean ± SE from 20 animals in 5 experiments. Mouse endogenous IL-6 is reported as the mean ± SD from six mice in two experiments.

Densitometric scanning was performed on a Gel Doc gel documentation system (BioRad). Data were analyzed using Quantity One Software (BioRad). Statistical analysis was performed using the Student’s t test and GraphPad Intuitive Software (GraphPad, San Diego, CA).

We have previously reported that an acute injection of IL-6 in mice impairs hepatic insulin receptor signaling in response to portal vein injection of insulin. Chronic rather than acute cytokine elevations, however, are characteristic of insulin-resistant states of infection, cachexia, cancer, obesity, and type 2 diabetes (5). We therefore tested the effect of chronic exposure to IL-6 delivered continuously for 5 days using an osmotic pump. The concentration of human recombinant IL-6 in the pumps was such that IL-6 was delivered at 16 ng/h. This dose resulted in an approximate sixfold serum elevation of exogenous hIL-6 over endogenous levels (rhIL-6 112 ± 27 pg/ml and endogenous mouse IL-6 18 ± 5 pg/ml). This elevation approximates that seen in obese individuals, though it does not take into account any differences in species-dependent bioactivity between human and mouse IL-6.

On the 4th day of IL-6 infusion, an overnight fast was initiated. On the 5th day, insulin (300 ng) was administered by intraportal injection and livers were harvested 45 s later. Hepatic insulin receptor autophosphorylation and IRS-1 and -2 tyrosine phosphorylation were analyzed by immunoprecipitation and immunoblot analysis. As expected, insulin stimulated tyrosine phosphorylation of these proteins while IL-6 alone did not. In mice chronically treated with IL-6, however, insulin-dependent insulin receptor autophosphorylation was suppressed by ∼60% (Fig. 1). No consistent change in insulin receptor mass was observed. IL-6 also inhibited insulin-induced IRS-1 tyrosine phosphorylation by ∼60% (Fig. 2A) and IRS-2 tyrosine phosphorylation by ∼40% (Fig. 2B). Again, mass levels of IRS-1 and -2 were not affected by chronic IL-6 treatment. STAT3 phosphorylation at tyrosine 705 was modestly elevated on day 5 in the liver of mice treated with chronic IL-6 infusion (Fig. 3, compare lanes 1–4 with lanes 5–8). These results demonstrate that chronic IL-6 exposure can suppress early hepatic insulin signaling in mice.

Approximately 75% of insulin-dependent glucose uptake occurs in the skeletal muscle (15). Since our data indicated that hepatic insulin signaling was reduced in mice chronically treated with IL-6, we sought to understand the insulin sensitivity of muscle compared with that of the liver under these conditions. To this end, mice were exposed to IL-6 for 5 days as described above. Insulin (300 ng) was delivered by femoral vein injection and liver and muscle harvested 90 s later. Insulin receptor signaling in the liver and muscle was then assayed by immunoprecipitation and immunoblot analysis. Again, insulin-dependent tyrosine phosphorylation of insulin receptor, IRS-1, and IRS-2 (Fig. 4, left panels) was inhibited in livers of animals treated with chronic IL-6. Thus, for liver insulin receptor signaling, femoral vein injection of insulin recapitulates the results obtained via portal vein injection. However, chronic IL-6 treatment did not affect insulin-dependent insulin receptor signal transduction in the skeletal muscle of these animals (Fig. 4, right panels).

To determine the effect of chronic IL-6 exposure on insulin-dependent glucose homeostasis, insulin tolerance tests were performed. Insulin (0.7 units/kg body wt) was administered by intraperitoneal injection, and glucose levels were taken by tail vein sampling at indicated time points (Fig. 5). At early times after insulin injection (t <60 min), no statistical difference in glucose levels between chronic IL-6-treated mice and controls was observed, though glucose levels in IL-6-treated animals trended toward a smaller decrease in response to insulin. After 60 min, however, mice that were chronically treated with IL-6 had significantly higher glucose levels consistent with a modest degree of insulin resistance. A more profound insulin resistance would not be expected given that ∼75% of the drop in blood glucose following insulin challenge is due to skeletal muscle glucose uptake, and the inhibitory effect of IL-6 appears to be restricted to the liver, as suggested by the absence of impairment in insulin receptor signaling in skeletal muscle after chronic exposure to IL-6.

The results of the insulin tolerance test suggested that chronic IL-6 exposure altered glucose homeostasis and produced a mild state of insulin resistance in mice. If the liver was the major contributor to this insulin resistance state, it would be anticipated that induction of insulin responsive genes that are central to glucose homeostasis would be compromised. Therefore, we tested whether feeding-dependent glucokinase induction was altered by chronic IL-6 treatment. Maximal glucokinase induction has been shown to occur in mice ∼90 min after feeding (16). Mice chronically treated for 5 days with IL-6 (16 ng/h) were fasted overnight and then freely fed for 90 min. Postprandial glucose levels were 24 ± 17% (n = 6, P < 0.05) higher in the mice infused with 16 ng/h IL-6 relative to controls. Liver glucokinase RNA levels were measured by Northern blot analysis of total RNA. As expected, a large postprandial glucokinase gene induction was observed. This induction was suppressed by ∼40% in mice chronically treated for 5 days with 16 ng/h (Fig. 6). These results indicate that chronic IL-6 exposure induces insulin resistance that is manifest on key physiological end points of glucose homeostasis in the liver.

The current results provide strong direct evidence that a chronic elevation of circulating IL-6, approximating that observed in obesity and type 2 diabetes, inhibits early hepatic insulin receptor signaling and downstream insulin action in vivo. This provides support for the hypothesis that this proinflammatory cytokine impairs insulin sensitivity and glucose regulation in humans and is linked to obesity-related insulin resistance. These data are particularly consistent with the strong clinical link that has been established between IL-6 and glucose levels (7), obesity, and insulin resistance (3,6,7). We have observed that chronic IL-6 exposure in vivo inhibits the ability of insulin to signal through its receptor and to phosphorylate two major metabolic substrates, IRS-1 and -2. This is consistent with data demonstrating impaired insulin receptor autophosphorylation and substrate phosphorylation in obese and diabetic subjects (17–21). Furthermore, we demonstrated defects in downstream actions of insulin in mice chronically treated with IL-6. These include an abnormal insulin tolerance test, increased postprandial serum glucose levels, and reduced postprandial induction of glucokinase. Together, these data demonstrate that chronic exposure to circulating levels of IL-6 that are within the range attainable in obesity and type 2 diabetes are able to recapitulate several of the characteristic metabolic defects of obesity-mediated insulin resistance and diabetes.

While the current results support a role for IL-6 in mediating hepatic insulin resistance, several additional adipose tissue-derived factors and cytokines have been similarly linked to glucose dysregulation and diabetes. It is likely that multiple cytokines and other factors work in concert with IL-6 to induce obesity-related insulin resistance. This is suggested by the absence of an effect of IL-6 on skeletal muscle. There has been a particular interest in the role of TNF-α in obesity-related insulin resistance. TNF-α has been shown to impair insulin action in vivo and in vitro (22–24). In rodent studies TNF-α overproduction correlates with insulin resistance, and neutralization of TNF-α restored insulin sensitivity to obese rats (25,26). A role for TNF-α is more controversial in humans. TNF-α expression by adipose tissue, but not plasma levels, has been correlated with obesity and insulin resistance (3,27,28). Neutralization of TNF-α in humans did not fully restore insulin sensitivity (29,30). Additionally, a direct role for TNF-α in suppressing insulin receptor signal transduction in the liver has been questioned (25,31,32). Kern et al. (3) reported that IL-6 plasma levels correlate better to insulin resistance than TNF-α levels. However, IL-6 plasma levels in obese subjects are elevated three- to fivefold in these studies (3,6,33). While the role of circulating TNF-α is uncertain, it is possible that an autocrine or paracrine role for TNF-α may exist since TNF-α has been shown to induce IL-6 production (30,34,35). It is therefore possible that TNF-α has both a direct effect on insulin responsive tissues, particularly skeletal muscle and adipose tissue, and an indirect effect through induction of IL-6 secretion.

The current in vivo study confirms and extends our finding that IL-6 causes insulin resistance in hepatocytes. Treating HepG2 hepatoma cells and primary mouse hepatocytes with IL-6-inhibited insulin signaling at the level of insulin-dependent IRS-1 tyrosine phosphorylation, phosphatidylinositol 3-kinase association with IRS-1, and AKT/protein kinase B activation (10). Furthermore, insulin-dependent glycogen synthesis was markedly impaired in primary hepatocytes pretreated with IL-6 (10). In the current study, acute and chronic in vivo IL-6 exposure inhibited hepatic insulin sensitivity. Interestingly, chronic exposure to IL-6 for 5 days did not inhibit insulin receptor signaling in skeletal muscle. These results suggest that IL-6 may not play a role in mediating insulin resistance in skeletal muscle. This agrees with reports indicating exceedingly low levels of IL-6 receptors expressed in muscle cells (36,37). Alternatively, the concentrations of IL-6 used in our current study may be too low to elicit a response in skeletal muscle. Several reports have described responses to IL-6 in various models of muscle atrophy, but the IL-6 levels were considerably greater than those observed in obesity and type 2 diabetes, and its effects may not have been direct (38,39). Further investigations are required to clarify the physiological role of skeletal muscle as a target for IL-6 in obesity-mediated insulin resistance.

How do the proinflammatory cytokines mediate their effects on insulin signaling? TNF-α appears to exert its inhibitory effect by activating several serine kinases, resulting in serine phosphorylation of the IRS-1 and -2 in muscle and adipose tissue (25,40,41). The serine phosphorylated IRS is a poor substrate for the insulin receptor kinase and is also subject to increased degradation (42–44). Although the mechanism through which IL-6 exerts its effect has not been proven in vivo, it is likely that IL-6 acts through a mechanism that is different from that for TNF-α. We have not detected significant serine phosphorylation of IRS-1 in either hepatocytes or liver in response to IL-6 (Klover and Mooney, unpublished observations). Changes in IRS-1 and -2 degradation are also not observed in response to chronic IL-6 treatments (Fig. 2). Our group has strong evidence to suggest that induction of suppressor of cytokine signaling-3 (SOCS-3) by IL-6 in hepatocytes may play a role in mediating the inhibitory properties of IL-6 on the insulin receptor (13). Overexpressed SOCS-3 has been shown to impair insulin-dependent insulin receptor autophosphorylation, IRS-1 tyrosine phosphorylation, phosphatidylinositol 3-kinase association with IRS-1 and AKT/protein kinase B activation in hepatocytes. Furthermore, the timing of SOCS-3 induction by IL-6 correlates with IL-6-induced impairments of insulin receptor signaling in these cells. SOCS-3 is also induced in response to acute IL-6 treatment in the liver of mice (13). Whether SOCS-3 is responsible for the effects of acute or chronic IL-6 exposure in mice is the subject of current investigation.

Changes in the hormonal milieu may be critically important for the development of insulin resistance and type 2 diabetes. The proinflammatory cytokines that are implicated in this disorder may each utilize a distinct mechanism to contribute to the insulin resistance of obesity. The results of this study and our previous report demonstrate that acute as well as chronic exposure to IL-6 inhibits insulin action in vivo. This implicates IL-6 as a contributor to the insulin resistance of obesity, particularly in the liver. Future studies are required to elucidate the mechanism by which IL-6 acts.

This work was supported by U.S. Public Health Service Grants R01-DK38138 and R01-DK60732 to R.A.M.