Reduced Skeletal Muscle Inhibitor of κBβ Content Is Associated With Insulin Resistance in Subjects With Type 2 Diabetes: Reversal by Exercise Training

Six subjects with type 2 diabetes and eight healthy control subjects participated in the study. The metabolic data from these subjects have been previously reported (12). Each subject underwent a complete history, physical examination, screening laboratory tests, and a 75-g oral glucose tolerance test to determine the presence or absence of diabetes using established American Diabetes Association criteria. Three of six type 2 diabetic subjects were taking glyburide, which was withdrawn 3 days before the oral glucose tolerance test and insulin clamp studies. The remaining three type 2 diabetic subjects were treated with diet. The control subjects did not have a family history of type 2 diabetes and had normal glucose tolerance. Other than glyburide, no subject was taking any medication known to affect glucose metabolism. The study was approved by the institutional review board of the University of Texas Health Science Center at San Antonio, and all subjects gave written consent.

Following a 10- to 12-h overnight fast, subjects came to the General Clinical Research Center (GCRC) at the South Texas Veterans Health Care Hospital for a screening visit and glucose tolerance test. Within 3–7 days, subjects returned to the GCRC for determination of their peak aerobic capacity (Vo_2peak) (12). Within 3–7 days after the baseline Vo_2peak measurement, subjects returned to the GCRC at 8 a.m. after an overnight fast for a vastus lateralis muscle biopsy (13), followed by a 120-min euglycemic-hyperinsulinemic (40 mU · m⁻² · min⁻¹) insulin clamp study, which was performed with infusion of 3-³H-glucose (prime = 25 μCi, continuous infusion = 0.25 μCi/min) to determine rates of glucose appearance and disposal, as previously described (12). In control subjects, the 3-³H-glucose glucose was given as a prime (25 μCi)-continuous (0.25 μCi/min) infusion starting 120 min before insulin. In type 2 diabetic subjects, the prime was increased (25 μCi × fasting plasma glucose/100) and began 180 min before insulin. In diabetic subjects, following the start of insulin, the plasma glucose concentration was allowed to decrease to 100 mg/dl, at which level it was maintained. One week after the insulin clamp study, the subjects began an 8-week aerobic exercise training program on a stationary bicycle. The exercise intensity, duration, and frequency were progressively increased to 70% of Vo_2peak for 45 min four times per week, during the 8-week course of the exercise program. Once the training program was completed, the subjects returned to the GCRC for a repeat determination of Vo_2peak. A posttraining muscle biopsy and insulin clamp study was performed 24–36 h after the last exercise bout.

Plasma insulin concentrations were measured by radioimmunoassay (Diagnostic Products, Los Angeles, CA), glucose was measured using the oxidase method and a Beckman analyzer, and HbA_1c (A1C) was measured using a DCA 2000 analyzer (Bayer, Tarrytown, NY). FFA concentrations were determined by an enzymatic colorimetric method (Wako Chemicals, Nuess, Germany), and plasma IL-6 and TNFα concentrations were measured using an enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN).

Muscle samples were homogenized as previously described (12). Briefly, muscle samples were weighted while still frozen and were homogenized in ice-cold lysis buffer (1:10, wt/vol) containing 50 mmol/l HEPES (pH 7.6), 150 mmol/l NaCl, 20 mmol/l sodium pyrophosphate, 20 mmol/l β-glycerophosphate, 10 mmol/l sodium fluoride, 2 mmol/l sodium orthovanadate (Na₃VO₄), 2 mmol/l EDTA (pH 8.0), 1% Nonidet P-40, 10% glycerol, 1 mmol/l phenylmethylsulfonylfluoride, 1 mmol/l MgCl₂, 1 mmol/l CaCl₂, 10 μg/ml leupeptin, and 10 μg/ml aprotinin. Homogenates were incubated on ice for 20 min and then centrifuged at 15,000g for 20 min at 4°C. The supernatants were collected, and protein concentrations were measured by the Lowry method. Supernatants were stored at −80°C until used.

Muscle proteins (40 μg) were separated by 10% SDS-PAGE and transferred to nitrocellulose membranes, except for TNFα, for which 150 μg protein were used. After blocking in Tris-buffered saline with 5% nonfat dry milk, the membranes were incubated overnight at 4°C with antibodies against IκBα (Cell Signaling, Beverly, MA), IκBβ (Santa Cruz Biotechnology, Santa Cruz, CA), NFκB p50 (Cell Signaling), NFκB p65 (Santa Cruz Biotechnology), IKKβ (Cell Signaling), phospho-IKKβ (Ser^177/181) (Cell Signaling), and TNFα (Santa Cruz Biotechnology) (14,15). For all these proteins, a 0.45-μm nitrocellulose membrane was used, except for TNFα, for which a 0.2-μm membrane was used. Bound antibodies were detected with anti-rabbit immunoglobulin horseradish peroxidase–linked whole antibody and by using enhanced chemiluminescent reagents (PerkinElmer Life Science, Boston, MA). The membranes were exposed to film, and band intensity was quantified using Image Tool Software (University of Texas Health Science Center at San Antonio, San Antonio, TX).

During the postabsorptive state, steady-state conditions prevail and the rate of endogenous glucose appearance, which equals the rate of total body glucose uptake, is calculated as the 3-³H-glucose infusion rate (disintegrations per minute) divided by the steady-state plasma 3-³H-glucose specific activity (disintegrations per minute per milligram). During the insulin clamp, non–steady-state conditions prevail, and the rate of glucose production (R_a) is calculated using Steele’s equation (16). The rate of endogenous glucose production during the insulin clamp equals the tracer-derived R_a minus the exogenous rate of glucose infusion. The rate of whole-body glucose disposal (R_d) equals the rate of residual endogenous glucose production plus the exogenous glucose infusion rate adjusting to maintain euglycemia (90- to 120-min time period).

All data are expressed as means ± SE. Differences between the control and diabetic groups were determined using the unpaired Student’s t test. Differences within a group, pre- versus posttraining, were determined using the paired t test. Correlation analysis was performed by the Pearson product-moment method. For all analyses, a P < 0.05 was considered to be statistically significant.

Table 1 summarizes the subject’s clinical and laboratory characteristics. There was no significant difference in age, body weight, and BMI between the type 2 diabetic and control subjects. Subjects with type 2 diabetes had higher fasting plasma glucose and insulin concentrations (P < 0.05), higher A1C (P < 0.05), and lower aerobic capacity (Vo_2peak) (P < 0.05). There were no differences in FFA (690 ± 100 vs. 660 ± 100 μmol/l), IL-6 (2.3 ± 0.5 vs. 2.3 ± 0.6 pg/ml), and TNFα (1.0 ± 0.1 vs. 1.1 ± 0.2 pg/ml) plasma concentrations between the control and type 2 diabetic groups.

Eight weeks of exercise training had no effect on body weight, BMI, or fasting plasma glucose and insulin concentrations in either group (Table 1). Plasma FFA, IL-6, and TNFα concentrations did not change with training. However, physical training significantly increased the Vo_2peakby 17% (P < 0.05) and 31% (P < 0.05) in the control and diabetic subjects, respectively. In the diabetic subjects, the increase in Vo_2peak after training also was accompanied by a tendency for a 19% decrease in A1C (P = 0.05).

Basal endogenous glucose production was similar between groups, whereas insulin-mediated glucose disposal (R_d) was 95% higher in the control group (P < 0.05) (Table 1). Exercise training significantly increased R_d in the type 2 diabetic subjects by 37% (P < 0.05). In the control subjects, R_d tended to increase after training (P = 0.1). The smaller effect of training in control subjects is in part explained by a higher baseline R_d, resulting in a modest effect of training. Differences in sex distribution between groups did not explain the lesser effect of training on R_d observed in the control subjects (not shown).

To determine whether skeletal muscle from subjects with type 2 diabetes has excessive activity of IκB/NFκB pathway, we measured IκBα and IκBβ protein abundance. Because IκB sequesters NFκB in the cytoplasm and increases in IκB abundance inversely correlate with NFκB DNA binding in human skeletal muscle (17), a reduction in IκB abundance is considered to indicate activation of the IκB/NFκB pathway. In skeletal muscle from type 2 diabetic subjects, IκBβ was reduced by 60% compared with control subjects (P < 0.05) (Fig. 1A). IκBα was not significantly different in type 2 diabetic subjects (Fig. 1B). This reduction in IκB suggests that there is excessive activity of the IκB/NFκB axis in skeletal muscle from type 2 diabetic subjects. Importantly, there was a positive correlation between IκBβ and the rate of insulin-mediated glucose disposal (r = 0.63, P = 0.01) (Fig. 2A) and with the Vo_2peak (r = 0.6, P = 0.02) (Fig. 2B). Consistent with these findings, there was a negative correlation between IκBβ and the A1C concentration (r = −0.5, P = 0.05) (Fig. 2C). There was also a positive correlation between insulin-mediated glucose disposal and the Vo_2peak (r = 0.54, P = 0.04) (Fig. 2D). The predominant activating NFκB dimer in muscle is p50-p65 (9,10). We measured NFκB p50 and p65 protein in skeletal muscle and found that NFκB p50 protein was decreased by 30% in type 2 diabetic versus control subjects (P < 0.05) (Fig. 3A). There were no differences in p65 protein between groups (Fig. 3B). In addition, there was a positive correlation between NFκB p50 protein content and insulin-mediated glucose disposal (r = 0.58, P = 0.04) (Fig. 3C).

Because the IκB/NFκB pathway has been implicated in the cellular mechanisms responsible for skeletal muscle insulin resistance, we hypothesized that exercise training in patients with type 2 diabetes would reverse abnormalities in IκB/NFκB signaling and that this is a mechanism by which training improves insulin sensitivity. In control subjects, training caused a 50% increase in both IκBα (P < 0.05) and IκBβ (P < 0.05) protein content (Fig. 4A and B). After training, NFκB p50 protein tended to increase by 100% (P = 0.05) (Fig. 4C) in the control subjects but had no effect on NFκB p65 (Fig. 4D). Moreover, in the muscle from type 2 diabetic subjects, physical training increased IκBα and IκBβ protein by 98% (P < 0.05) and 185% (P < 0.05), respectively (Fig. 5A and B). Exercise training also increased NFκB p50 in subjects with type 2 diabetes by 140% (P < 0.05) (Fig. 5C), but similar to the control subjects, exercise had no effect on NFκB p65 (Fig. 5D). Furthermore, these increments in IκB and NFκB p50 protein were accompanied by a 37% increase in insulin-mediated glucose disposal (Table 1). These findings indicate that training can restore abnormalities in IκB and NFκB p50 content in muscle from subjects with type 2 diabetes.

We measured IKKβ phosphorylation to further examine the mechanism responsible for the reduction in IκB in the type 2 diabetic group. There was no difference in IKKβ phosphorylation between groups, and training had no effect on IKKβ phosphorylation (Fig. 6) or protein content (not shown).

It has been shown that type 2 diabetes is associated with higher muscle expression of TNFα, an NFκB-regulated gene (18). Accordingly, in the type 2 diabetic subjects the reduction in IκBβ was accompanied by a tendency for higher basal TNFα protein content by 25% compared with the control subjects (P = NS) (Fig. 7), although we could only measure TNFα muscle content in five control and four diabetic subjects because a large amount of muscle protein is required to measure TNFα muscle content by Western blotting and muscle tissue was no longer available for mRNA expression analysis. It is possible that this trend would have reached statistical significance with a higher number of samples. Consistent with previous reports (19), the increases in IκB and NFκB p50 caused by training were associated with a reduction in TNFα muscle content in both groups (P < 0.05) (Fig. 7).

In this study, we found that type 2 diabetic subjects have decreased IκBβ muscle content, suggesting enhanced IκB/NFκB signaling. Importantly, there was a positive correlation between IκBβ protein and insulin-mediated glucose disposal in type 2 diabetic and control subjects. These findings support the hypothesis that enhanced activation of the IκB/NFκB pathway is associated with insulin resistance in human skeletal muscle.

The mechanism responsible for the apparent increase in IκB/NFκB signaling in muscle from the type 2 diabetic subjects is not clear. One hypothesis involves lipid-induced activation of IκB/NFκB signaling. Intramyocellular lipid metabolites, particularly long-chain fatty acyl CoAs, diacylglycerol, and ceramides, have been shown to induce insulin resistance (20–25). Moreover, skeletal muscle of insulin-resistant subjects is characterized by increases in fatty acyl CoA (26) and ceramides (27). Recently, it was shown that lipid-induced insulin resistance in L6 myotubes (28) and muscle from rodents (3,4,29) and humans (5) is associated with activation of the IκB/NFκB pathway. A role for this pathway in mediating insulin resistance is further strengthened by the findings that inhibition of IκB/NFκB signaling improves insulin sensitivity (4,11). In support of our findings, Bhatt et al. (29) recently reported that diet-induced obesity in rats, an intervention that causes insulin resistance, lead to a decrease in muscle IκB content. Interestingly, this group found that differences in IκB muscle content between muscle fiber types were not explained by muscle triglyceride content (29), suggesting that intramyocellular lipids were not responsible for the reduction in IκB. However, metabolites of triglycerides and fatty acids (i.e., fatty acyl CoAs, diacylglycerol, and ceramides), and not triglycerides per se, are believed to be responsible for the insulin resistance. We were unable to measure fatty acyl CoA, ceramide, or diacylglycerol muscle content because of the limited amount of tissue that can be obtained in humans using the percutaneous biopsy technique. Infusion of FFAs to humans (5) and rats (29) has been associated with activation of the IkB/NFkB pathway. In the present study, fasting plasma FFA concentrations were similar between groups. It should be noted, however, that fasting FFA levels do not reflect FFA turnover and day-long plasma FFA concentrations. Reaven et al. (30) studied type 2 diabetic subjects with similar blood glucose concentrations compared with the subjects included in the present study and found that the type 2 diabetic subjects had similar fasting FFA plasma levels compared with the nondiabetic control subjects, yet mean 24-h plasma FFA concentrations were significantly elevated.

Type 2 diabetes is characterized by low-grade chronic inflammation (1,31,32), and cytokines, including TNFα and IL-6, have been shown to activate the IκB/NFκB pathway. TNFα and IL-6 plasma concentrations were similar between groups. Therefore, it seems unlikely that these cytokines were responsible for the decrease in muscle IκB. Nonetheless, there are many other exogenous inflammatory stimuli that were not measured in this study and could have caused IκB/NFκB axis activation. Reactive oxygen species are also known to simulate the IκB/NFκB pathway (33–35), and hyperglycemia-induced reactive oxygen species generation could have contributed to the lower IκB seen in the type 2 diabetic group. Consistent with our findings, Bandyopadhyay et al. (36) did not find increased IKKβ phosphorylation in muscle from insulin-resistant subjects. Because IκB phosphorylation by IKK leads to IκB ubiquination and subsequent degradation, the decrease in IκB content would then have to be explained by IKK-independent phosphorylation. Protein kinase C and eIF2 kinase can also associate with and phosphorylate IκB (37).

How excessive activity of the IκB/NFκB pathway leads to muscle insulin resistance has not been fully elucidated. Insulin infusion increased insulin receptor substrate-1–associated phosphoinositide 3-kinase activity by 33% (P < 0.05) in control subjects but not in type 2 diabetic subjects (12). One potential mechanism for these decreases in insulin-stimulated glucose disposal and phosphoinositide 3-kinase activity in the diabetic group involves insulin receptor substrate-1 serine phosphorylation by IKKβ (38–40), but, as mentioned above, IKKβ phosphorylation was unchanged. Moreover, training-induced improvements in insulin sensitivity were not explained by changes in IKKβ phosphorylation or phosphoinositide 3-kinase activity. In insulin-resistant (lipid-treated) L6 muscle cells, inhibition of NFκB using a selective inhibitory peptide improved insulin-stimulated glucose transport (28). This indicates that NFκB, and genes that this transcription factor controls, may directly influence insulin sensitivity. Yet, overactivation of NFκB in transgenic mice was not associated with insulin resistance (41). Because the phenotype of these animals was characterized by severe muscle wasting, it is unclear how the muscle wasting influenced insulin sensitivity measurements. It is also possible that lifelong overactivation of NFκB could lead to compensatory mechanisms aimed at normalizing insulin sensitivity. Further studies will help clarify how the IκB/NFκB pathway modulates insulin sensitivity in humans.

Few studies have examined the effect of exercise on the IκB/NFκB pathway. In rats, acute exercise activates IκB/NFκB signaling in muscle (42,43), while acute fatiguing exercise in humans reduces NFκB activity (44). In the present study, we found that training increased IκB and reduced TNFα muscle content, suggesting decreased IκB/NFκB signaling. This increase in IκB was not associated with changes in plasma TNFα, IL-6, adiponectin (not shown), or FFA concentrations. Stimuli intrinsic to muscle could also be responsible for the training-induced changes in IκB and TNFα. Τraining significantly increased AMP-activated protein kinase (AMPK) phosphorylation in both control and type 2 diabetic groups (N.M., L.J.M., unpublished observations). In endothelial cells, chemical AMPK activation with 5-aminoimidazole-4-carboxamide ribonucleoside has been shown to inhibit NFκB activity (45). These results suggest that exercise-stimulated AMPK activation may be responsible for the inhibition of the IκB/NFκB axis. In contrast, AMPK activation with 5-aminoimidazole-4-carboxamide ribonucleoside had no effect on NFκB activity in rat skeletal muscle (43), arguing against a role for AMPK as a regulator of IκB/NFκB signaling. A reduction in intramyocellular lipid content caused by training could help explain the changes in IκB and TNFα. However, several studies have shown that trained athletes have increased triglyceride muscle content (46,47). The greater triglyceride storage in the trained athlete represents an adaptive response to training and provides a readily available source of energy for the contracting muscle. In contrast, elevated lipid stores in type 2 diabetes result from an imbalance between plasma FFA availability and oxidation and have been implicated in the development of insulin resistance. Future studies will be needed to examine whether elevated levels of fatty acyl CoA, diacylglycerol, and ceramides are related with abnormalities in IκB/NFκB signaling and to assess the effect of training on the content of these metabolites.

Type 2 diabetic subjects had a more pronounced reduction in IκBβ content compared with IκBα. While the IκBα and IκBβ isoforms share many similarities, they also display differences. The IκBα gene has a κB recognition sequence in its promoter region (48). Thus, activation of NFκB can induce resynthesis of IκBα. In contrast, NFκΒ does not induce IκBβ synthesis (8,49). Since the muscle of type 2 diabetic subjects appears to have excessive NFκB activity, this could lead to selective resynthesis of IκBα but not of IκBβ. Although it is not known whether IκBβ remains suppressed upon long-term NFκB activation, in Jurkat cells cytokine-induce IκBα degradation is rapidly restored within 2 h, whereas IκBβ remains low for at least 24 h (50). Whether this differential regulation between IκBα and IκBβ occurs in other chronic diseases remains to be determined.

We found that NFκB p50 subunit content was reduced in muscle of type 2 diabetic subjects, and this decrease correlated with reduced insulin sensitivity. This might appear counterintuitive since NFκB stimulates transcription of inflammatory genes. While the p50-p65 heterodimer is the predominant activating NFκB dimer, the p50 subunit also can dimerize with another p50 subunit to form p50-p50 homodimers. While the p50-p65 heterodimer activates transcription of inflammatory genes, the p50-p50 homodimer inhibits gene transcription (9,51–53). In view of our findings of reduced p50, one might speculate that type 2 diabetic subjects could have decreased abundance of the p50-p50 repressor homodimers, resulting in excessive NFκB p50-p65 activity, and that training inhibited NFκB activity by increasing p50-p50 homodimers. Future investigations are needed to determine whether NFκB p50-p50 DNA binding is indeed reduced in muscle from type 2 diabetic subjects.

The present study shows an important association between insulin sensitivity and the content of IκBβ and NFκB p50 in human muscle. Yet, these results do not prove causality. Studies using genetic and pharmacologic approaches to manipulate IκB/NFκB signaling will help establish the role of this pathway in insulin sensitivity regulation.

This work was supported by grants DK47936 (to L.J.M.), DK66483 (to L.J.M.), and DK24092 (to R.A.D.) from the National Institutes of Health; Grant RR01346 (GCRC, Audie L. Murphy Veterans Hospital); an Executive Research Committee Research Award from the University of Texas Health Science Center at San Antonio (to N.M.); a VA Merit Award (to R.A.D.); an American Diabetes Association Junior Faculty Award (to N.M.); and a grant from the Faculty of Medicine Siriraj Hospital, Mahidol University, Thailand (to A.S.).

We also thank all the study participants.