Increased Antiangiogenic Protein Expression in the Skeletal Muscle of Diabetic Swine and Patients FREE

More than 20 million individuals in the United States have diabetes mellitus (DM).¹ In addition to their increase in risk for stroke and myocardial infarction,² diabetic patients account for 20% to 30% of patients with peripheral arterial disease (PAD). In fact, diabetes is one of the single strongest risk factors for the development of PAD,³ with every 1% increase in hemoglobin A_1c increasing the risk for developing PAD by 28%.⁴ Despite lifestyle modification and aggressive strategies aimed at improving glucose control, many of these patients have advanced PAD leading to claudication, rest pain, and nonhealing lower extremity ulcers necessitating revascularization. Unfortunately, patients with DM and PAD may have poorer clinical responses to pharmacologic and revascularization therapy than nondiabetic (ND) patients and may be poor candidates for surgical revascularization secondary to limited distal bypass targets and the presence of multiple comorbidities.^3,5 The summary results of these findings account for the 15-fold increase in risk for lower extremity amputation seen in diabetic patients.⁶

Therapeutic angiogenesis, the induction of neovascularization through administration of cell-, gene-, or protein-based proangiogenic factors, may offer a new modality for treating patients with severe PAD not amenable to surgical revascularization. Although the efficacy of proangiogenic therapy for chronic extremity ischemia has been shown in preclinical animal models⁷ and small patient series^8- 9 while phase 1 clinical trials of therapeutic angiogenesis for PAD have demonstrated safety,^10- 15 phase 2 clinical trials have thus far failed to demonstrate significant long-term objective benefit.^16- 17 These findings, combined with evidence that diabetic patients exhibit impaired endogenous collateral formation in response to chronic ischemia,¹⁸ highlight the need for an improved understanding of the effects of diabetes on molecular angiogenic signaling pathways influencing collateral formation. To address this issue, our group and others have investigated angiogenic signaling in the setting of diabetes in preclinical large animal models of hyperglycemia¹⁹ and diabetes.²⁰ A key finding of these studies, which investigated the myocardium and coronary circulation, indicated that hyperglycemia is associated with increased antiangiogenic signaling. This specifically implicated 2 well-studied proteins in the oncology literature—angiostatin and endostatin, which are derived from plasminogen and collagen XVIII, respectively, through a series of cleavage steps and ultimately by matrix metalloproteinases (MMPs). These proteins, discovered by Folkman and colleagues in the mid 1990s, have been shown to be potent antiangiogenic agents^21- 22 and are associated with impaired collateral formation.^23- 24 We hypothesized that angiostatin and endostatin expression would be increased in skeletal muscle in the setting of diabetes.

Sixteen male Yucatan miniswine (Sinclair Research Center, Inc, Columbia, Missouri) were used for the studies. Diabetes was induced at age 8 months using a single intravenous injection of alloxan (150 mg/kg). Only animals that achieved and maintained blood glucose levels greater than 200 mg/dL were included in the diabetes group (to convert glucose to millimoles per liter, multiply by 0.0555). Animals were fed ad libitum and no dietary modification was used. Two groups were created. In the DM group (n = 8), fasting blood glucose levels were maintained between 250 and 400 mg/dL with intramuscular insulin (70% amorphous, 30% crystalline) administered for blood glucose values exceeding 400 mg/dL. Fasting blood glucose levels were monitored every 2 days early after diabetes induction and 1 to 2 times per week thereafter by obtaining a small blood sample (50 μL) from a capillary bed in the ear. Age-matched ND miniswine (n = 8) served as controls. Animals were maintained in their diabetic state for 15 weeks prior to harvest of skeletal muscle. For tissue harvest, anesthesia was performed as described previously.²⁵ Briefly, anesthesia was induced with ketamine hydrochloride (10 mg/kg, intramuscular) and thiopental sodium (5-10 mg/kg, intravenous) and maintained with a gas mixture of oxygen at 1.5 to 2.0 L/min and isoflurane at 0.75% to 3.00%. The animals were intubated and mechanically ventilated at 12 to 20 breaths/min. A femoral groin dissection was performed and skeletal muscle was harvested and immediately placed in liquid nitrogen followed by storage at −80°C.

Animals received humane care in compliance with the Harvard Medical Area Institutional Animal Care and Use Committee and the National Research Council's Guide for the Care and Use of Laboratory Animals.²⁶

Skeletal muscle tissue samples were obtained from the internal thoracic artery harvest site of 21 patients (11 ND patients, 10 patients with type 2 DM) undergoing initial elective coronary artery bypass grafting at the start of the surgical procedure. Tissue was immediately placed in liquid nitrogen and stored at −80°C for use in molecular studies. Patient samples were not harvested if any history of malignant neoplasm was reported. The study was approved by the institutional review board of the Beth Israel Deaconess Medical Center, Boston, Massachusetts.

Whole-cell lysates were isolated from the homogenized skeletal muscle samples with a ristocetin-induced platelet agglutination buffer (Boston Bioproducts, Worcester, Massachusetts) and centrifuged at 12 000g for 10 minutes at 4°C to separate soluble fractions from insoluble fractions. Protein concentration was measured spectrophotometrically at a 595-nm wavelength with a DC protein assay kit (Bio-Rad Laboratories, Inc, Hercules, California). Forty to eighty micrograms of total protein was fractionated by a 4% to 20% gradient in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (Invitrogen, San Diego, California) and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, Massachusetts). Each membrane was incubated with specific antibodies as follows: antiplasminogen antibody (dilution, 1:500) (Calbiochem, San Diego, California), antiangiostatin antibody (dilution, 1:2500) (BD Biosciences, San Jose, California), anti–collagen XVIII antibody (dilution, 1:300) (Santa Cruz Biotech, Santa Cruz, California), antiendostatin antibody (dilution, 1:1000) (Upstate, Chicago, Illinois), anti–MMP-2 antibody (dilution, 1:500) (Calbiochem), anti–MMP-9 antibody (dilution, 1:500) (Millipore), and anti–tissue inhibitor of metalloproteinase 2 (TIMP-2) antibody (dilution, 1:300) (Calbiochem). The membranes were subsequently incubated for 1 hour in diluted appropriate secondary antibody (Jackson Immunolab, West Grove, Pennsylvania). Immune complexes were visualized with the enhanced chemiluminescence detection system (Amersham, Piscataway, New Jersey). Bands were quantified by densitometry of radioautograph films. Ponceau S staining was performed to confirm equivalent protein loading.

Fasting blood glucose measurements in swine were performed with the Ascensia Contour Blood Glucose Monitoring System (Bayer Healthcare, Tarrytown, New York). Mean fasting blood glucose levels were used to determine glycemic control instead of hemoglobin A_1c levels in swine as evidence indicates that rates of glycated hemoglobin formation are quite different between swine and humans, possibly reflecting differences in erythrocyte permeability to glucose among the species.²⁷ Patient hemoglobin A_1c measurements were performed at the Beth Israel Deaconess Medical Center Clinical Laboratory. Reference ranges for ND patients are 4.8% to 5.9% hemoglobin A_1c. On average, each 1% increase in the glycated hemoglobin level represents roughly an increase of 30 mg/dL in the mean blood glucose level. Thus, a glycated hemoglobin level of 10% corresponds to a mean blood glucose level of 250 mg/dL: a normal 5% glycated hemoglobin level corresponds to a normal mean blood glucose level of 100 mg/dL, an increment of 5% (to get to 10%) corresponds to 5 × 30 = 150 mg/dL glucose, and 100 + 150 = 250.

To determine the degree of MMP-2 and MMP-9 activity in skeletal muscle specimens from patients, gelatin zymography was performed. Forty micrograms of protein in 20 μL of zymogram buffer (Bio-Rad Laboratories, Inc) was loaded from tissue homogenates onto a 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel copolymerized with gelatin (1 mg/mL). Proteins were electrophoretically separated at a 120-V constant current, washed for 1 hour in renaturation buffer (2.5% Triton X-100; Bio-Rad Laboratories, Inc) at room temperature, and incubated overnight in 50-mmol/L Tris hydrochloride (pH 7.5), 200-mmol/L sodium chloride, 5-mmol/L calcium chloride, and development buffer (0.02% Brij-35; Bio-Rad Laboratories, Inc). Gels were stained with 0.5% Coomassie blue in 30% methanol and 10% glacial acetic acid and destained in the same solution lacking Coomassie blue. The gelatinolytic activity was identified as transparent bands against the background of Coomassie blue–stained gelatin. Band size (based on migration) and loading of control purified MMP-2 and MMP-9 were used to identify bands of interest.

Data are reported as mean (standard error of the mean). Immunoblot results are expressed as a ratio of protein to loading band density and were analyzed after digitization of and quantification from radiographic films with ImageJ 1.33 software (National Institutes of Health, Bethesda, Maryland). Blots were analyzed using analyses of variance. Bonferroni corrections were applied to multiple tests, and P < .05 was considered statistically significant.

Mean (SEM) fasting blood glucose levels in swine were 73.5 (5.9) mg/dL for the ND group vs 262.1 (23.4) mg/dL for the DM group (P = .001). Patient groups ND and DM were similar in terms of age, sex, and comorbidities. Based on the hemoglobin A_1c levels, the calculated mean blood glucose levels of patients were 118.0 mg/dL in the ND group and 184.0 mg/dL in the DM group (P = .01). One patient in the DM group had type 2 DM and was using insulin therapy. For patient characteristics, see the Table.

Skeletal muscle expression of the precursors of angiostatin and endostatin were similar between both groups of swine. Mean (SEM) plasminogen expression was 63.7 (17.3) densitometry units (DU) in the ND group vs 70.5 (12.9) DU in the DM group (P = .77). Mean (SEM) collagen XVIII expression was 33.7 (17.3) DU in the ND group vs 40.5 (12.9) DU in the DM group (P = .71).

Similar results were found for patient skeletal muscle levels of plasminogen (mean [SEM], 65.7 [18.2] DU in the ND group vs 53.2 [13.9] DU in the DM group; P = .64) and collagen XVIII (mean [SEM], 31.0 [2.0] DU in the ND group vs 31.8 [3.7] DU in the DM group; P = .85).

Porcine skeletal muscle expression of angiostatin was increased 1.70-fold in DM swine relative to ND swine (mean [SEM], 50.0 [12.4] DU in the ND group vs 86.0 [7.5] DU in the DM group; P = .04) (Figure 1A). Expression of endostatin in porcine skeletal muscle was increased 1.84-fold in DM swine relative to ND swine (mean [SEM], 29.9 [2.9] DU in the ND group vs 55.0 [8.6] DU in the DM group; P = .04) (Figure 1B).

Skeletal muscle expression of angiostatin in patients demonstrated no significant difference between groups (mean [SEM], 45.8 [16.9] DU in the ND group vs 71.0 [15.0] DU in the DM group; P = .35) (Figure 2A). Endostatin expression in the skeletal muscle of patients was 1.69-fold higher in patients with DM in comparison with ND patients (mean [SEM], 36.4 [7.3] DU in the ND group vs 61.5 [9.4] DU in the DM group; P = .04) (Figure 2B).

No significant differences were seen between swine groups for expression of the latent form of MMP-2 (mean [SEM], 32.5 [4.2] DU in the ND group vs 35.3 [8.6] DU in the DM group; P = .78), whereas expression of the active form of MMP-2 was significantly higher in DM swine (mean [SEM], 4.4 [1.6] DU in the ND group vs 17.0 [4.8] DU in the DM group; P = .04) (Figure 3A). Expression of the latent form of MMP-9 (mean [SEM], 51.8 [7.0] in the ND group vs 64.3 [5.3] DU in the DM group; P = .21) and active form of MMP-9 (mean [SEM], 63.5 [12.0] DU in the ND group vs 83.5 [6.6] DU in the DM group; P = .20) remained similar between DM and ND swine (Figure 3B).

In the patients, no significant differences were observed between expression of the latent form of MMP-2 (mean [SEM], 92.7 [3.6] DU in the ND group vs 87.0 [4.1] DU in the DM group; P = .28) and active form of MMP-2 (mean [SEM], 60.0 [11.4] DU in the ND group vs 35.3 [8.8] DU in the DM group; P = .12). Expression of the latent form of MMP-9 (mean [SEM], 83.5 [4.7] DU in the ND group vs 74.0 [5.0] DU in the DM group; P = .20) and active form of MMP-9 (mean [SEM], 81.5 [3.7] DU in the ND group vs 66.0 [6.2] in the DM group; P = .06) remained similar between the DM and ND groups as well.

Human skeletal muscle MMP-2 activity (mean [SEM], 20.6 [5.5] DU in the ND group vs 39.6 [2.7] in the DM group; P = .01) was significantly increased by 1.92-fold in patients with DM relative to ND patients (Figure 4A). There were no significant changes in skeletal muscle MMP-9 activity (mean [SEM], 25.0 [10.0] DU in the ND group vs 62.2 [13.4] DU in the DM group; P = .12) (Figure 4B).

Skeletal muscle expression of TIMP-2, an inhibitor of MMP-2 and MMP-9, was significantly higher in DM swine (mean [SEM], 24.3 [4.8] DU in the ND group vs 56.3 [9.7] DU in the DM group; P = .02).

No significant differences were found in the skeletal muscle expression of TIMP-2 between patients with DM and ND patients (mean [SEM], 50.5 [8.5] DU in the ND group vs 47.8 [7.5] DU in the DM group; P = .82).

The development of collateral vessels represents an endogenous adaptive response to chronic ischemia and is impaired in diabetic patients.¹⁸ From a therapeutic perspective, these patients may have limited distal bypass targets and more extensive PAD limiting surgical options for revascularization,²⁸ highlighting the need for alternative therapeutic options in this setting. Proangiogenic therapy may be a promising modality for these patients, yet the disappointing results of phase 2 clinical trials^16- 17 highlight the need for an improved understanding of angiogenesis at the molecular level. Multiple factors have been implicated in the diminished endogenous angiogenic response to chronic skeletal muscle ischemia, including impairments in vascular endothelial growth factor signaling²⁹ and endothelial progenitor cell function.³⁰ In this study, we identify additional components in the angiogenic cascade that may contribute to impaired collateral formation. Primarily, this study demonstrates that antiangiogenic protein expression is increased in the skeletal muscle in the setting of diabetes. Specifically, angiostatin and endostatin levels are increased in the skeletal musculature of diabetic swine, and endostatin expression is increased in the skeletal musculature of diabetic patients. The expression levels of the 2 proteins are not regulated at the level of production of their precursors, plasminogen for angiostatin and collagen XVIII for endostatin, but may be regulated at the level of MMP-2, an enzyme responsible for their production through cleavage.

Angiostatin has been the focus of intensive study for the treatment of cancer, where it has been shown to be an effective antiangiogenic agent.^31- 32 As a cleavage product of plasminogen, angiostatin is produced after a series of cleavage steps involving plasmin, disulfide reductase, and, terminally, MMP-2 and MMP-9.³³ The molecular mechanisms by which it exerts its antiangiogenic effects have yet to be fully elucidated but presently center around endothelial cells. Angiostatin binds to endothelial adenosine triphosphate synthase, resulting in endothelial cell death secondary to energy dysregulation.^34- 35 It has also been shown to impair endothelial cell proliferation,³⁶ migration, and tube formation by binding to angiomotin.³⁷ Angiostatin appears to have multiple proapoptotic effects through binding of integrin α_Vβ₃³⁸ and activating the Fas-L extrinsic apoptotic pathway.³⁹ Despite these findings, numerous angiostatin-binding partners exist for which functionality has yet to be determined.³³

Two prior articles have demonstrated that angiostatin expression is increased in the setting of hyperglycemia¹⁹ and diabetes.⁴⁰ The first of these studies used a canine model of hyperglycemia (via continuous glucose infusions) for 21 days and examined myocardial interstitial fluid, finding increases in angiostatin levels in myocardial interstitial fluid at day 21. Chung et al⁴⁰ examined internal mammary artery samples of diabetic and nondiabetic patients and found a 1.62-fold increase in angiostatin expression in mammary arteries of diabetic patients. These findings are interesting in that they indicate increases in angiostatin levels may be systemic in the setting of diabetes and not localized to specific organs or tissues.

Endostatin, the 20-kDa cleavage product of collagen XVIII, was discovered in 1997 and has been shown to be a potent inhibitor of angiogenesis.²² The mechanisms of endostatin's action remain under investigation. Endostatin has multiple actions leading to endothelial cell apoptosis. It can inhibit antiapoptotic signaling induced by the phosphoinositol-3-kinase pathway⁴¹ and can activate proapoptotic pathways by increasing levels of caspase 9 and decreasing levels of antiapoptotic proteins (Bad, Bcl-2, and Bcl-Xl).⁴² Cell migration, adhesion, and proliferation are diminished through endostatin impairments in integrin binding,⁴³ generation of mitogen-activated protein kinases 1 and 2 and c-myc,⁴⁴ and vascular endothelial growth factor R2 function.⁴⁵ These findings are likely only a few of those relevant to endostatin function as microarray studies of endothelial cells have shown that endostatin can regulate 12% of the genome, including a significant number of antiangiogenic genes (which are upregulated) and proangiogenic genes (which are downregulated).^46- 47

Both MMP-2 and MMP-9 are able to cleave plasminogen and collagen XVIII to generate angiostatin and endostatin, respectively. Interestingly, these proteases act as proangiogenic mediators initially—allowing for degradation of the extracellular matrix to facilitate collateralization—but can ultimately lead to inhibition of angiogenesis through generation of angiostatin and endostatin, revealing the fine dynamic balance of the angiogenic process. Controversy exists as to whether diabetes induces an upregulation or downregulation of MMP activity. The 2 prior studies investigating MMP activity in human samples used mammary artery and reached opposite conclusions: according to Chung et al,⁴⁰ MMP-2 and MMP-9 activity is increased in mammary arteries of diabetic patients, but according to Portik-Dobos et al,⁴⁸ MMP-2 and MMP-9 activity is decreased in mammary arteries of diabetic patients. Our results are consistent with the former, although comparison is limited owing to the samples examined—skeletal muscle vs mammary artery. It is unknown whether these differences are owing to patient variables or methods.

To investigate molecular signaling governing MMP activity, we examined expression of TIMP-2, a protein capable of inhibiting MMP-2 and MMP-9 that has been shown to be downregulated in the setting of diabetes.⁴⁰ Our study demonstrated an increase in TIMP-2 expression in diabetic swine, with no significant difference observed in patients. This may reflect differences in the tissue studied by Chung et al.⁴⁰ At present, no exhaustive approach investigating which TIMPs most effectively inhibit which MMPs has been undertaken,⁴⁹ leaving the possibility that alternative TIMPs play a role in modulating MMP-2 and MMP-9 in skeletal muscle and are differentially expressed in diabetes.

While providing data from a preclinical large animal model and patients, this study has several limitations. The use of insulin-dependent diabetic swine may limit generalizability of some findings from the animal model to patients, as most diabetic patients have type 2 DM. Currently, there are no suitable, validated, type 2 DM, large animal models available, necessitating use of the alloxan model studied here. Notably, this model does induce endothelial dysfunction and impaired collateral formation common between type 1 and type 2 diabetes, providing some foundation for its use.²⁰ The differences between the chronicity and type of diabetes between the animal model and the patient studies also add to the difficulty of drawing comparisons between the two. The patient portions of this study are limited by our sample size, which may limit discernment of certain differences between the patients with DM and the ND patients. The patient studies used ND patients with coronary artery disease as control subjects, whereas ideally, a group of healthy, age-matched control subjects would have been studied. Future studies examining chronically ischemic skeletal muscle will likely be of interest. Additionally, this study does not give data providing a direct causal linkage between antiangiogenic protein levels and collateral formation. Future in vitro studies examining the effects of angiostatin and endostatin on angiogenesis will be of value.

Diabetes mellitus is associated with an increase in skeletal muscle levels of the potent antiangiogenic proteins angiostatin and endostatin in a preclinical large animal model, with endostatin expression increased in skeletal muscle of patients with DM. These increases may be secondary to increased activity of MMP-2, an enzyme capable of cleaving plasminogen and collagen XVIII into angiostatin and endostatin, respectively, in the skeletal muscle of patients with DM. Targeting these 2 proteins may have significant therapeutic utility in enhancing both endogenous and exogenous growth factor–induced skeletal muscle collateral development.

Correspondence: Frank W. Sellke, MD, Division of Cardiothoracic Surgery, Beth Israel Deaconess Medical Center, 110 Francis St, Lowry Medical Office Bldg, Ste 2A, Boston, MA 02215 (fsellke@caregroup.harvard.edu).

Accepted for Publication: December 21, 2007.

Author Contributions:Study concept and design: Sodha and Sellke. Acquisition of data: Sodha, Boodhwani, Clements, Xu, and Khabbaz. Analysis and interpretation of data: Sodha, Clements, and Sellke. Drafting of the manuscript: Sodha. Critical revision of the manuscript for important intellectual content: Sodha, Boodhwani,Clements, Xu, Khabbaz, and Sellke. Statistical analysis: Sodha, Boodhwani, Clements, and Sellke. Obtained funding: Sellke. Administrative, technical, and material support: Xu, Khabbaz, and Sellke. Study supervision: Sellke.

Financial Disclosure: None reported.

Funding/Support: This work was supported by grants R01 HL46716 (Dr Sellke) and R01 HL69024 (Dr Sellke) from the National Heart, Lung, and Blood Institute. Dr Sodha is supported in part by grant T-32HL076130-02 from the National Institutes of Health and by the Irving Bard Memorial Fellowship.

Previous Presentation: This paper was presented at the 88th Annual Meeting of the New England Surgical Society; September 28, 2007; Burlington, Vermont; and is published after peer review and revision.