Diet-Induced Muscle Insulin Resistance Is Associated With Extracellular Matrix Remodeling and Interaction With Integrin α₂β₁ in Mice

Mice were housed under temperature- and humidity-controlled environment with a 12-h light/dark cycle. Distinctive murine models were used to address specific goals of the study: 1) the association between insulin resistance and ECM adaptations was studied in sex-matched C57BL/6J mice fed chow or HF diet (F3282, BioServ) containing 60% calories as fat for 20 weeks; 2) novel genetic (muscle-specific mitochondrial-targeted catalase transgenic, mcat^Tg) and pharmacological (sildenafil) models were used to alleviate insulin resistance in the presence of HF feeding, and the rescue of the ECM adaptations was assessed; and 3) integrin α₁β₁-null (itga1^−/−) and α₂β₁-null (itga2^−/−) mice were used to study the role of ECM-integrin interaction in insulin resistance.

We previously showed that mcat^Tg mice maintained on a HF diet for 16 weeks were protected from insulin resistance (21). HF-fed mice treated with subcutaneous injection of sildenafil, a selective phosphodiesterase (PDE)5a inhibitor, for 13 weeks are also protected from insulin resistance (22). The mcat^Tg and sildenafil-treated mice were selected as models that prevent diet-induced muscle insulin resistance based on two considerations. First, these approaches protect from diet-induced insulin resistance through specific actions that target distinct mechanisms. The mcat^Tg mice selectively overexpress catalase in muscle mitochondria. It is a genetically modified animal model in which the rescue of diet-induced muscle insulin resistance is related to a reduction of mitochondrial release of ROS. This contrasts with the PDE5a inhibition model, which causes smooth muscle relaxation and vasodilation pharmacologically. The second consideration was that ROS have been implicated in the regulation of collagen synthesis (13,23,24), and they do so by increasing the inflammatory response and activation of mitogen-activated protein kinase (MAPK) (25,26). Therefore, mcat^Tg mice provide a model for studying the hypothesis that the inflammatory response associated with insulin resistance increases collagen in muscle.

To further study the role of ECM-integrin interaction in insulin resistance, mice lacking the two major collagen binding receptor integrins α₁β₁ or α₂β₁ were studied. Male itga2^−/− mice and wild-type (WT) littermates (itga2^+/+) were chow-fed until 9 weeks of age at which time mice were continued on chow or switched to HF diet until 23 weeks of age. Male itga1^−/− mice and WT littermates (itga1^+/+) were chow-fed until 12 weeks of age at which time mice were continued on chow or switched to HF diet until 28 weeks of age. Murine body composition was determined by nuclear magnetic resonance. The Vanderbilt Animal Care and Use Committee approved animal procedures.

Catheters were implanted in a carotid artery and a jugular vein of mice for sampling and infusions 5 days before study (27). Insulin clamps were performed on 5-h fasted mice (28). [3-³H]glucose was primed (2.4 μCi) and continuously infused for a 90-min equilibration period (0.04 μCi/min) and a 2-h clamp period (0.12 μCi/min). Baseline blood or plasma parameters were determined in blood samples collected at −15 and −5 min. At t = 0, insulin infusion (4 mU/kg/min) was started and continued for 165 min. Blood glucose was clamped at 150 ∼160 mg/dL using a variable rate of glucose infusion (GIR). Mice received heparinized saline-washed erythrocytes from donors at 5 μL/min to prevent a fall of hematocrit. Insulin clamps were validated by assessment of blood glucose over time. Blood glucose was monitored every 10 min, and the GIR was adjusted as needed. Blood was taken at 80–120 min for the determination of [3-³H]glucose. Clamp insulin was determined at t = 100 and 120 min. At 120 min, 13 μCi of 2[¹⁴C]deoxyglucose ([¹⁴C]2DG) was administered as an intravenous bolus. Blood was taken at 2–35 min for the determination of [¹⁴C]2DG. After the last sample, mice were anesthetized and tissues were collected.

Plasma insulin was determined by ELISA (Millipore). Radioactivity of [3-³H]glucose, [¹⁴C]2DG, and [¹⁴C]2DG-6-phosphate were determined by liquid scintillation counting (22). Glucose appearance (R_a) and disappearance (R_d) rates were determined using non-steady-state equations (29). Endogenous glucose production (endoR_a) was determined by subtracting the GIR from total R_a. The glucose metabolic index (R_g) was calculated as previously described (30).

ColI, ColIII, ColIV, and Von Willebrand factor (VWF) were assessed by immunohistochemistry in paraffin-embedded tissue sections. Sections (5 μm) were incubated with the following primary antibodies for 60 min: anti-ColI (Abcam), anti-ColIII (CosmoBio), anti-ColIV (Abcam), or anti-VWF (DakoCytomation). Slides were lightly counterstained with Mayer hematoxylin. The EnVision+HRP/DAB System (DakoCytomation) was used to produce localized, visible staining. Immunostaining of CD31 was performed in either frozen sections or paraffin-embedded sections using anti-CD31 (BD Biosciences). Images were captured using a Q-Imaging Micropublisher camera mounted on an Olympus upright microscope. Immunostaining was quantified by ImageJ or BIOQUANT Life Science 2009. ColI, ColIII, and ColIV protein were measured by the integrated intensity of staining. Muscle vascularity was determined by counting CD31-positive structures and by measuring areas of VWF-positive structures.

Total RNA was isolated from gastrocnemius using the RNeasy Fibrous Tissue Kit (Qiagen). Total RNA (1 μg) was reverse transcribed using the iScript cDNA Synthesis Kit (Bio-Rad). Real-time PCR was then performed using the SYBR Green JumpStart Taq Readymix (Sigma) on an IQ5 Multicolor Real-Time PCR Detection System (Bio-Rad). The primers used are listed in Supplementary Table 1. Data were analyzed by the 2^−ΔΔCt method (31).

Gelatin zymography was performed to measure the activities of type IV collagenase-matrix metalloproteinase (MMP)2 and MMP9 (32). Gastrocnemius was homogenized in buffer containing 0.5% Triton X-100, 100 mM Tris-HCl, 10 mM EDTA, and 10 μL/mL protease inhibitor (pH 7.5). Homogenates were centrifuged at 13,000 rpm for 20 min at 4°C. The supernatant (500 μg) was incubated at 4°C for 2 h with 40 μL of gelatin-Sepharose (Pharmacia). After the incubation, the gelatin-Sepharose beads were resuspended in nonreducing SDS-sample buffer and loaded on 10% zymogram gels (Invitrogen). The gel was developed according to the manufacturer’s instructions. Conditioned medium from HT-1080 cells was used as positive control.

Gastrocnemius was homogenized in buffer containing 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 10% glycerol, 1% Triton X-100, 1 mM DTT, 1 mM PMSF, 5 μg/mL protease inhibitor, 50 mM NaF, and 5 mM sodium pyrophosphate and centrifuged at 13,000 rpm for 20 min at 4°C. The supernatant (40 μg) was applied to 4–12% SDS-PAGE gel. Phosphorylated and total Akt/protein kinase B (PKB) were probed using phospho-Akt(Ser473) and Akt antibodies (Cell Signaling). Phosphorylated and total insulin receptor substrate-1 (IRS-1) were probed using phospho–IRS-1 (Tyr612) and IRS-1 antibodies (Upstate).

Data are expressed as means ± SE. Statistical analyses were performed using Student ttest or two-way ANOVA followed by Turkey post hoc tests as appropriate. The significance level was P < 0.05.

Twenty weeks of HF diet in C57BL/6J mice, a regimen that induces insulin resistance (33), caused inflammation in muscle, as gene expression of F4/80, a macrophage marker, and TNF-α, a cytokine involved in systemic inflammation, was increased in muscle of HF-fed mice compared with chow-fed mice (Fig. 2A and B). To test the hypothesis that the inflammatory response associated with diet-induced insulin resistance in mice is accompanied by increased muscle collagen, we measured collagen expression in HF-fed murine muscle. Muscle ColIII and ColIV protein were increased in HF-fed mice compared with chow-fed mice (Fig. 1A), but protein expression of ColI was very low and differences were undetectable.

Increases in collagen could be because of either increased synthesis or decreased degradation via MMPs. mRNA of specific collagen chains were used to assess collagen synthesis. HF feeding increased muscle mRNA of ColIα1, ColIα2, and ColIIIα1 chains, whereas mRNA of ColIV was unchanged (Fig. 1B). To test whether increased ColIV protein was due to reduced degradation, the activities of type IV collagenases, MMP2 and MMP9, were measured. HF feeding decreased MMP9 activity, without affecting pro-MMP2 and MMP2 activities in muscle (Fig. 1C).

Muscle insulin resistance was absent in HF-fed mcat^Tg (21) and sildenafil-treated mice (22). In line with decreased muscle insulin resistance, both mcat^Tg and sildenafil-treated mice had reduced F4/80 expression in muscle (Fig. 2A and C). HF diet-induced increase in muscle TNF-α mRNA was also eliminated in mcat^Tg mice (Fig. 2B); although, the decrease in TNF-α mRNA in muscle of sildenafil-treated mice was nearly 50%, the measurements in WT were variable and differences were not significant (Fig. 2D). ColIII and ColIV proteins were decreased in mcat^Tg and sildenafil-treated mice (Fig. 2E). The lower expression of ColIII in HF-fed mcat^Tg mice was because of reduced synthesis, as the HF-induced increases in mRNA of ColIIIα1 were abolished (Fig. 3B). HF-induced increase in mRNA of ColIα2 was also abolished in mcat^Tg mice (Fig. 3A). Decreased ColIV protein was due at least in part to increased degradation via MMP9 as MMP9 activity was increased in muscle of HF-fed mcat^Tg and sildenafil-treated mice (Fig. 3C).

Decreased collagen deposition in HF-fed mcat^Tg and sildenafil-treated mice was associated with improved muscle vascularization, since blood vessel staining by CD31 and VWF was increased (Fig. 3D). Taken together, preventing diet-induced muscle insulin resistance using different approaches that focus on distinct targets results in common ECM adaptations. These common adaptations include reduced collagen, increased MMP9 activity, and increased vascularization.

Integrins α₁β₁ and α₂β₁ are collagen receptors expressed on the endothelial cell surface (19,20), as well as on several other cell types (8). To test the role of ECM-integrin signaling in the pathogenesis of insulin resistance, we assessed insulin action in chow and HF-fed itga1^−/− and itga2^−/− mice using insulin clamps. HF feeding increased body weight, body fat, and basal 5-h fasting glucose and insulin levels in all genotypes (Table 1). Body weights and body fat were not different between null mice and their respective WT littermates either on chow or HF diet. Basal 5-h fasting blood glucose did not differ between itga1^+/+ and itga1^−/− mice regardless of diet; however, the basal 5-h fasting blood glucose was lower in the HF-fed itga2^−/− mice compared with HF-fed itga2^+/+ mice. Blood glucose was maintained at 150 ∼160 mg/dL during the insulin clamp in all groups (Table 1 and Fig. 4A, D, G, and J). Plasma insulin was not different between null mice and their respective WT littermates either in basal or insulin-clamped state within a specific diet (Table 1).

In chow-fed mice, GIR during the insulin clamp were not different between itga1^−/− and itga1^+/+ mice or itga2^−/− and itga2^+/+ mice (Fig. 4B and E). R_g in muscle was also not affected by genotype (Fig. 4C and F). HF feeding induced insulin resistance in all groups as shown that GIR and R_g were dramatically decreased by HF feeding (Fig. 4B, C, E, and F vs. 4H, I, K, and L). In contrast with chow-fed mice, GIR was lower in the HF-fed itga1^−/− mice compared with HF-fed itga1^+/+ mice (Fig. 4H) but higher in the HF-fed itga2^−/− mice when compared with HF-fed itga2^+/+ mice (Fig. 4K). Muscle R_g was not different between HF-fed itga1^−/− and itga1^+/+ mice (Fig. 4I). However, muscle R_g was three- to fourfold higher in HF-fed itga2^−/− compared with HF-fed itga2^+/+ (Fig. 4L).

Because integrins α₁β₁ and α₂β₁ are expressed in a cell type that is widely distributed (i.e., endothelial cells), the metabolic phenotype of HF-fed null mice is not necessarily restricted to muscle. Therefore, we measured insulin action in liver and adipose tissue, two other insulin-sensitive glucoregulatory organs, in HF-fed integrin null mice and WT littermates. Basal endoR_a was not different between HF-fed itga1^−/− and itga1^+/+ mice (Fig. 5A). Insulin infusion decreased endoR_a by 50% in the HF-fed itga1^+/+ mice. In contrast, endoR_a was not suppressed in the itga1^−/− mice. These data are consistent with the lower GIR and reflects greater hepatic insulin resistance in HF-fed itga1^−/− mice. R_d was not different between the HF-fed itga1^−/− and itga1^+/+ mice either in basal or insulin-clamped state (Fig. 5B). Insulin-stimulated adipose tissue R_g was decreased in HF-fed itga1^−/− mice relative to itga1^+/+ mice (Fig. 5C). R_d was not reduced since white adipose tissue is a small contributor to total glucose disposal. EndoR_a was not different between the itga2^−/− mice and the itga2^+/+ mice either in basal or insulin-clamped state (Fig. 5D); however, R_d during the insulin clamp was higher in itga2^−/− compared with itga2^+/+ (Fig. 5E). This is consistent with increased R_g in muscles. Adipose tissue R_g did not differ between HF-fed itga2^−/− and itga2^+/+ mice (Fig. 5F). These data suggest that the HF-fed itga1^−/− mice have a further impairment in hepatic and adipose insulin resistance with no impairment in muscle insulin resistance. HF-fed itga2^−/− mice have improved insulin sensitivity because of decreased muscle insulin resistance.

Changes in muscle insulin signaling after the clamp were consistent with flux measurements observed during the clamp in HF-fed itga2^−/− mice. Phosphorylation of IRS-1 and Akt were increased in muscle of HF-fed itga2^−/− mice compared with those of the itga2^+/+ littermates, indicative of improved insulin signaling (Fig. 6A and B). There was no effect on total Akt and IRS-1 in HF-fed itga2^−/− mice.

In contrast with our findings in mcat^Tg and sildenafil-treated mice, the HF-fed itga2^−/− mice were protected against HF-induced muscle insulin resistance despite no reduction in muscle ColIII and ColIV proteins (Fig. 6C). Nevertheless, immunostaining of the blood vessels by CD31 and VWF was increased in muscle of HF-fed itga2^−/− compared with HF-fed itga2^+/+ mice (Fig. 6C). This is consistent with results obtained in mcat^Tg and sildenafil-treated mice. These data provide further evidence for the association between increased muscle insulin sensitivity and vascularization.

Our studies show that HF-fed insulin-resistant mice are characterized by an increase in muscle collagen. It is important to recognize that this is not a response that is isolated to this murine model, but is an established characteristic of human insulin resistant muscle (4,5). Thus the study of collagen and ECM in HF-fed mouse has great relevance to the human condition. By the use of genetically modified and sildenafil-treated mice, the current study demonstrates a close association between collagen dynamics and the pathogenesis of muscle insulin resistance. Furthermore, studies in itga2^−/− mice provide mechanistic insight for this association by showing that the link between muscle insulin resistance and increased collagen is uncoupled in the absence of collagen-integrin α₂β₁ interaction.

Inflammation can cause insulin resistance (34), and collagen deposition and ECM remodeling are hallmarks of inflammation (2). Here we tested the hypothesis that the inflammatory response associated with a HF diet would increase collagen, causing a formidable barrier to muscle glucose uptake and affecting muscle vascular reactivity (Fig. 7). In the current study, overexpression of catalase in the mitochondria, which reduces oxygen free radical production, prevented muscle inflammation, diminished collagen expansion, improved vascularity, and rescued insulin resistance. It is known that PDE5a inhibition increases vascular reactivity. PDE5a inhibition prevented collagen expansion by decreasing the inflammatory response. PDE5a inhibition, like catalase overexpression, rescued insulin resistance. Integrin α₂β₁ is expressed in endothelial cells (8), and loss of integrin α₂β₁ leads to increased endothelial cell proliferation and angiogeneis in vivo (20). Endothelial/vascular dysfunction is associated with insulin resistance (35). Here we found that enhanced muscle insulin action in mice lacking the integrin α₂β₁ receptor was also associated with increased angiogenesis. We postulate that the lack of an adaptive increase in vascularization and endothelial function contributes to diet-induced insulin resistance in C57BL/6J mice. Murine models that are able to adapt in this way have diminished insulin resistance.

TNF-α, a cytokine involved in systemic inflammation, has been shown to regulate collagen expression (36,37). In line with decreased muscle inflammation and collagen expression, mcat^Tg mice also had decreased muscle TNF-α gene expression. These data suggest that inflammation increases collagen expression possibly mediated via TNF-α. Although TNF-α mRNA tended to be reduced in HF-fed sildenafil-treated mice when compared with the HF-fed control mice, differences were not significant. This suggests that factors other than TNF-α could also play a role in inflammation-mediated collagen regulation.

We showed that the increases in muscle collagen were corrected when muscle insulin resistance was rescued either by genetic manipulation (mcat^Tg) or pharmacological treatment (sildenafil). It is important to note that these two models improve muscle insulin sensitivity and suppress the diet-induced increases in ColIII and ColIV, by distinct primary mechanisms. mcat^Tg reduces mitochondrial release of ROS (21), whereas chronic treatment with sildenafil inhibits PDE5a, causing an induction of smooth muscle relaxation and vasodilation (22). Thus, despite targeting different initial pathways within skeletal muscle, both models reverse excess collagen deposition in response to HF feeding. Bajaj et al. (38) showed that a sustained reduction in plasma free fatty acids in insulin-resistant relatives of patients with type 2 diabetes leads to an increase in insulin sensitivity, but paradoxically increases mRNA for ECM proteins. Differences in the experimental paradigm make it difficult to compare those results with findings from the current study and previous studies in humans from the same laboratory (4). Whether this paradoxical increase in collagen gene expression observed in the study of Bajaj et al. is paralleled by an increase in protein was not assessed. Acipimox, a nicotinic acid analog that activates G_i protein-coupled receptors (39), was used to lower free fatty acids in the study of Bajaj et al. It is possible that acipimox affects collagen mRNAs independent of effects on insulin action. It is clear that the association between ECM remodeling, free fatty acids, and insulin resistance is complex and remains to be fully defined.

ECM-integrin signaling has been associated with insulin resistance (9,10). Because the two major collagen binding receptors integrin α₁β₁ and α₂β₁ have been implicated in inflammation (40) and angiogenesis (18,20), as well as collagen synthesis (14,16), we used mice selectively lacking integrin α₁β₁ or α₂β₁ to examine the role of collagen-integrin interaction in the development of insulin resistance. Consistent with our hypothesis, insulin resistance was exacerbated in HF-fed itga1^−/− mice because of further impairments in hepatic and adipose insulin resistance. Muscle insulin resistance in HF-fed itga1^−/− mice was not further impaired, possibly because of an already high resistance in muscle. In contrast with integrin α₁β₁, deletion of integrin α₂β₁ protected against muscle insulin resistance induced by HF feeding. Integrin α₂β₁ is shown to be a positive regulator of collagen expression when cultured cells expressing α₂β₁ grow in three-dimensional collagen gel (16,17). However, we found no changes in muscle collagen expression in HF-fed itga2^−/− mice. This disparity could be because of differences in models and/or cell types studied. Conversely muscle vascularization was increased, which we hypothesize contributes to decreased HF-induced muscle insulin resistance in itga2^−/− mice. This observation is consistent with the antiangiogenic nature of integrin α₂β₁ (20). The fact that deletion of integrin α₂β₁ improves diet-induced muscle insulin resistance without a reduction in muscle collagen suggests that ECM collagen expansion is associated with muscle insulin resistance by interaction with integrin α₂β₁.

Although deletion of integrin α₂β₁ improves HF-induced muscle insulin resistance, as shown by an approximate fourfold increase in R_g, muscle insulin sensitivity in the HF-fed itga2^−/− mice is still below that seen in chow-fed itga2^+/+. This suggests that although the collagen-integrin α₂β₁ interaction is an important determinant of insulin resistance it, not surprisingly, is just one piece of the insulin resistance syndrome. Deletion of the integrin β₁-subunit specifically in striated muscle results in insulin resistance in chow-fed mice (10). It is possible that α-subunits other than α₁ and α₂, which dimerize with the integrin β₁-subunit, may also contribute to insulin resistance. ColIII and ColIV protein was increased in muscle of HF-fed mice. It is also possible that the expanded collagens increase the spatial barrier and the nature of the physical barrier from blood to muscle, thus contributing to insulin resistance. In contrast with ColIII and ColIV, protein expression of ColI was very low in muscle, suggesting it does not create a spatial barrier. However, the mRNA levels of both ColIα1 and ColIα2 were increased in muscle of HF-fed mice, implying that ColI remodeling could also be a part of insulin resistance. The interaction between ColI and integrins even at a very low level could conceivably impact cell signaling. There is the potential for interaction between the ECM and many cell surface receptors besides integrins, such as discoidin domain and growth factor receptors (41). These interactions could produce chemical or mechanical signal transduction changes that have impact on intracellular pathways.

These data obtained in murine models extend studies conducted in human subjects, which have shown that insulin-resistant humans have increased muscle collagen (4,5). We show for the first time that distinct genetic and pharmacological murine models that correct insulin resistance prevent the accumulation of collagen in skeletal muscle and exhibit increased muscle vascularity. Through the use of itga2^−/−mice we provide mechanistic insight into this association by showing that the link between muscle insulin resistance and increased collagen is uncoupled in the absence of integrin α₂β₁. The deletion of integrin α₂β₁ also promotes increased vascularization in muscle. The results support the concept that increased collagen contributes to insulin resistance through interaction with endothelial integrin α₂β₁ protein. Therefore, our findings demonstrate novel roles of ECM expansion in the pathogenesis of insulin resistance. Manipulations that limit ECM expansion or target integrin signaling may provide new opportunities to explore therapeutics for insulin resistance and type 2 diabetes.

This work was supported by National Institutes of Health Grants DK-54902 and DK-59637 (Mouse Metabolic Phenotyping Center).

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

L.K. researched data and wrote the article. J.E.A. and R.S.L.-Y. contributed to discussion and reviewed and edited the article. Z.Z. contributed to discussion. F.D.J. researched data. P.D.N., A.P., M.M.Z., and D.H.W. contributed to discussion and reviewed and edited the article.

The authors thank Melissa B. Downing, Pamela S. Wirth, and Dr. Lillian B. Nanney of the Vanderbilt Immunohistochemistry Core Laboratory for performing the immunohistochemistry.