Unacylated Ghrelin Induces Oxidative Stress Resistance in a Glucose Intolerance and Peripheral Artery Disease Mouse Model by Restoring Endothelial Cell miR-126 Expression

Peripheral arterial disease (PAD) is a widespread condition caused by the atherosclerosis of peripheral arteries, most commonly in the lower extremities (1). Patients with both type 1 and type 2 diabetes display an increased risk of developing PAD (2,3). Moreover, individuals with diabetes develop PAD at an early age, and greater PAD severity is commonly caused by impaired vascular remodeling (2,3). The persistence of an oxidative environment is crucial for symptom occurrence/recurrence in the diabetic setting (3). A defective mitochondrial electron transport chain (4), along with an increase in reactive oxygen species (ROS) levels, are crucial determinants of tissue damage in PAD patients. This is particularly true in diabetes, in which endogenous antioxidant defense mechanisms are also impaired (3,5).

The role of glucose-mediated oxidative stress in vascular damage has been extensively documented and is considered to be the most relevant mechanism in what is known as metabolic memory (6–8). Metabolic memory refers to the long-term detrimental effects of hyperglycemia, maintained even after glucose normalization. It mainly relies on changes in the epigenome (6–8) and involves histone-modifying enzymes, DNA methylation, chromatin remodeling proteins, and micro-RNAs (miRs) (6–8). Proof of concept for the role of impaired ROS clearance in hyperglycemia-induced changes in the epigenome (9) can be found in the molecular approach-driven overexpression of superoxide dismutase-2 (SOD-2), the primary enzymatic antioxidant defense against ROS production, which prevents the sugar-induced histone methylation of the NF-κB subunit p65 (9). The transcriptional activation of NF-κB and the expression of its target genes, such as the vascular cell adhesion molecule 1 (VCAM-1), are some of the most relevant markers of epigenome modification in diabetes (9). In addition to the role of transcription factor NF-κB, posttranscriptional regulation, involving small noncoding miRs, is known to control VCAM-1 expression and inflammatory cell recruitment in developing atherosclerotic lesions and subsequent ischemia (10). Interestingly, the plasma miR profile of diabetic patients has revealed that miR-126, one of miRs involved in the regulation of VCAM-1 expression (11), is reduced (12). This suggests that aberrant miR-126 regulation might also contribute to the long-lasting detrimental vascular effects of hyperglycemia. Persistent downregulation of class III histone deacetylase sirtuin 1 (SIRT1) (13), in individuals suffering from diabetes has also recently been reported to contribute to mitochondrial superoxide accumulation, reduced scavenging enzyme activity, NF-κB activation, p53 and histone 3 lysate 56 (H3K56) acetylation, as well as vascular dysfunction (8,13,14).

Surgical and endovascular treatment are currently considered to be the front line in avoiding limb loss (15). However, restenosis is commonly found in up to 50% of patients with successful angioplasty (16). Moreover, alternative treatment options are still limited for PAD patients who are not eligible for surgery. Therapeutic angiogenesis has recently been proposed for these patients, but has failed to improve long-term patient outcomes (2). The failure to obtain a clinical response to current therapeutic options has spurred the development of alternative strategies, which are mainly focused on improving antioxidant machinery. We have previously shown that the more abundant circulating form of ghrelin, unacylated grelin (UnAG), exerts antioxidant effects on endothelial progenitor cells (EPCs) (17,18). A proof of concept is provided by UnAG administration, which induces SOD-2 expression and improves mitochondrial dysfunction in muscles subjected to ischemia (19). UnAG is one of the ghrelin forms that is mainly produced in the stomach (20). UnAG differs from acylated ghrelin (AG) in term of its biological activity (17–19,21). Moreover, while both AG and UnAG are released into circulation (22), their ratio (AG/UnAG) varies and ranges from 1:2 to 1:9 (23), while a relative excess of AG has interestingly been found in individuals suffering from insulin resistance–connoted metabolic disorders (24). These observations, coupled with the role of glucose-mediated oxidative stress in vascular damage, have led us to investigate the clinical impact of UnAG administration on a mouse model of massive obesity, glucose intolerance, and PAD.

A detailed list of the reagents and antibodies used in this study is reported in Supplementary Table 1.

Male C57BL/6J ob/ob mice (8 weeks old) (Charles River Laboratories International Inc., Wilmington, MA) were anesthetized on day 0, and unilateral hind limb ischemia was induced as described in Hellingman et al. (25). The normoperfused contralateral limb of each mouse was used as an internal control. After hind limb ischemia, animals were treated daily with intraperitoneal injection from day 0 to day 21 using either saline or UnAG (pharmacological dose: 100 μg/kg) (19,26,27). Mice were treated according to European guidelines and policies as approved by the University of Turin Ethical Committee.

Mice were anesthetized as above and subjected to a noninvasive assessment of ischemic limb microvascular perfusion using the laser Doppler perfusion imaging system (PIM3; Perimed) (performed at WIL Research Europe, Saint Germain-Nuelles, France). Details are reported in the Supplementary Data.

A semiquantitative estimation of foot damage (two-way ANOVA followed by the post hoc test using the Bonferroni correction for multiple comparison) was performed serially using the following classification: 3, dragging of foot (foot necrosis); 2, no dragging but no plantar flexion (foot damage); 1, plantar flexion but no toe flexion (toe damage); and 0, flexing the toes to resist gentle traction on the tail (no damage) (28).

Gastrocnemius muscles were recovered from the ischemic and normoperfused limbs of treated animals, fixed in 10% formalin, and embedded in paraffin. Tissue sections were stained with hematoxylin and eosin for histological analysis. The proportion of fibers with central nuclei (regenerating fibers) in the injured area was counted as previously described (19). Muscle sections were processed for immunofluorescence assays and immunohistochemistry analyses as previously described (29). A quantification of cells that expressed the indicated markers was obtained as previously described (30). Images were acquired using a Zeiss LSM 5 Pascal confocal laser-scanning microscope (Carl Zeiss, Jena, Germany) (19). Details are reported in the Supplementary Data.

For the ex vivo experiments, endothelial cells (ECs) were isolated (31) from gastrocnemius muscles that had been subjected to ischemia. Details are reported in the Supplementary Data. For the in vitro experiments, primary ECs were either cultured for 2 days in 10% (volume for volume) bovine calf serum and basic fibroblast growth factor alone or in combination with 400 μg/mL advanced glycosylation end product (AGE) or 25 mmol/L d-glucose and either treated with UnAG (1 µmol/L) or not. At day 3, in vitro ischemia was induced as previously described (19). Satellite cells (SCs) were also isolated as described in Musarò and Barberi (32).

Senescence was performed as previously described (33). Details are reported in the Supplementary Data.

Cells were lysed and processed as previously described (34). H3K56 acetylation was evaluated on nuclear extracts from ECs. Details are reported in the Supplementary Data.

Cells were lysed in a cold DIM buffer supplemented with 10 μmol/L trichostatin A. Equal amounts of protein (500 μg) were immunoprecipitated using the p53 monoclonal antibody (see Supplementary Data). Bound proteins were eluted and processed as previously described (35). Immunoprecipitates were subjected to SDS-PAGE and blotted with the antiacetylated Lys³⁸² p53 antibody.

Total RNA was isolated using the TRIzol reagent (Invitrogen) from ECs recovered from the muscles of treated animals or from ECs subjected to in vitro ischemia. Samples were processed as previously described (31). Loss-of-function experiments were performed in ECs transfected with an anti-miR negative control or an anti–miR-126 antagonist according to the manufacturer’s instructions.

All data are presented as mean ± SEMs. The D'Agostino-Pearson test was used to test normality. Data on blood perfusion, damage score, number of vessels, percentage of regenerating fibers, inflammatory cells, and miR-126 and VCAM-1 expression from the ischemic and nonischemic limbs of treated ob/ob mice at days 7 and 21 were analyzed using two-way ANOVA, followed by the post hoc test with the Bonferroni correction for multiple comparisons. Data on blood glucose, body weight, HbA_1c measurements from adhesion assays, senescence assays, ROS generation, nitric oxide (NO) production, CD31/SOD-2 colocalization, CD31/SIRT1 colocalization, proliferating cell nuclear antigen (PCNA) expression, on glucose and insulin tolerance test assays, from luciferase assays, on miR-126 expression in in vitro or loss-of-function experiments, and SIRT1 and SOD-2 inactivation were analyzed using one-way ANOVA followed by Tukey multicomparison post hoc test. Densitometric analysis data for the Western blots were analyzed using Student t tests for two-group comparison and one-way ANOVA, followed by Tukey's multiple comparison test, for ≥3 groups. The cutoff for statistical significance was set up at P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001). All statistical analyses were carried out using GraphPad Prism version 5.04 (Graph Pad Software).

Blood glucose measurements, cell proliferation, FACS analyses, adhesion assays, SIRT1 and SOD-2 silencing, oxidative stress, NO detection, glucose and insulin tolerance tests, and the luciferase gene reporter are reported in the Supplementary Data.

PAD is one of the most challenging medical-surgical diseases worldwide (1–3). The beneficial vascular effects of UnAG (17,19) have led us to investigate UnAG's therapeutic potential in ob/ob mice with PAD. ob/ob Mice that had been subjected to unilateral hind limb ischemia were treated daily (starting at day 0) with either saline or UnAG. No significant differences in large vessel reperfusion (Fig. 1A) were detected in the treatment groups as shown by laser Doppler perfusion imaging performed at the indicated number of days postsurgery. However, the functional scores reported in Fig. 1B demonstrate a significantly higher damage score in the saline- than in the UnAG-treated group. UnAG-treated ob/ob mice contained an increased number of regenerating myofibers in ischemia-subjected muscles (Fig. 1C and D), which is consistent with data obtained in wild-type mice (19). Also consistent with our previous data (19) is the finding that UnAG was able to induce an expansion in SCs (Pax-7⁺/MyoD⁺ cells) as well as their terminal differentiation (myogenin⁺ cells) (Fig. 1E). In order to investigate the possibility that such differences depended on UnAG-mediated vascular protection, the number of vessels in ischemic and nonischemic muscles was counted in the two groups of mice. As shown in Fig. 2A, saline-treated mice showed significantly lower capillary density in this study than in the contralateral muscles, unlike UnAG-treated animals. The expression of PCNA and cyclin D1 was then evaluated in ECs recovered from the ischemic muscles of saline and UnAG mice so as to rule out the possibility that proliferation signals may be able to mediate UnAG effects. We found that both PCNA and cyclin D1 expression were similar in the two animal groups at days 7 and 21 after surgery (PCNA: saline, day 7 = 24.09 ± 5.2; UnAG, day 7 = 20.3 ± 3.3; saline, day 21 = 20.73 ± 4.6; UnAG, day 21 = 24.5 ± 3.8; n = 3) (Fig. 2B). No changes in plasma glucose concentration (saline, 326.3 ± 28.9; UnAG, 310.1 ± 31.2 mg/dL), body weight (saline, 45.83 ± 0.76; UnAG, 45.91 ± 0.9 g), HbA_1c (saline, 12.38 ± 1.09%; UnAG, 11.89 ± 1.19), or insulin sensitivity (Supplementary Fig. 1A and B) were detected in UnAG mice over the entire treatment period. Thus, it can be said that UnAG protects vessels against ischemia-induced damage in ob/ob mice independently of metabolic control.

Impaired antioxidant machinery and increased ROS production are crucial mediators of vascular damage in diabetes (36). This is particularly true in PAD patients (37), and so ECs recovered from ischemic muscles were first evaluated for intracellular ROS content. Data reported in Fig. 2C demonstrate a significant decrease in ROS generation in UnAG over saline-treated mice. In order to validate the possibility that UnAG promotes its biological effects by inducing an efficient antioxidant response, the mitochondria-specific antioxidant enzyme SOD-2 was evaluated in vessels. We demonstrate that the intramuscle vessels of UnAG-treated mice express high levels of SOD-2 (Fig. 2D). Similar results were obtained in ex vivo experiments (Fig. 2E).

It has been reported that the impaired SIRT1 level contributes to superoxide mitochondrial accumulation in both diabetes- and insulin resistance–connoted metabolic disorders (38–40). We therefore evaluated SIRT1 expression in both saline and UnAG animals. Data reported in Fig. 3A demonstrate that UnAG treatment was associated with increased SIRT1 expression in vessels. ECs were recovered from ischemic muscles and analyzed for SIRT1 expression as a means to validate these observations. As expected, ex vivo experiments demonstrated that ECs that had been recovered from UnAG animals displayed increased SIRT1 content (Fig. 3B) both in ischemic and nonischemic muscles. Moreover, a reduced level of H3K56 acetylation (Fig. 3C) (14) and a reduced number of senescent ECs were found in ob/ob mice challenged with UnAG (Fig. 3D), which is consistent with the role of SIRT1 in regulating EC fate.

ECs were treated with AGE and subjected to in vitro ischemia to further deepen the understanding of these data. Fig. 4A–C demonstrates that a UnAG challenge interferes with ROS generation and was able to induce both SOD-2 and SIRT1 expression. Moreover, in a finding that is consistent with our previous data in EPCs (17), we demonstrate that UnAG was able to increase NO levels in an ischemic setting (Supplementary Fig. 1C). Similar results were obtained using high glucose concentrations (data not shown). AGE was thus used throughout the study. We found that pS36p66Shc (38) was reduced in these experimental conditions, as expected from SIRT1 expression (Fig. 4D). In addition, p53 acetylation was also reduced upon UnAG treatment, and, in accordance with ex vivo experiments, this effect was associated with a reduced number of senescent ECs (Fig. 4E and F). These data were further confirmed by silencing SIRT1 (Supplementary Fig. 2A and Fig. 5A–D). Finally, UnAG treatment is only effective in reducing H3K56 acetylation in cells that express SIRT1, in vitro as well (Figs. 4G and 5E).

Moreover, UnAG-induced SIRT1 activity is not enhanced by the activation of AMP-activated protein kinase (AMPK) (41), unlike calorie restriction (Fig. 4H).

It has been shown that SIRT1 activates, via its deacetylation ability, peroxisome proliferator activated receptor-γ coactivator 1α, which is involved in SOD-2 expression (42). ECs were therefore silenced for SIRT1 and subjected to the same experimental conditions. SIRT1 silencing led to reduced SOD-2 expression and increased ROS generation, even in the presence of UnAG, as shown in Fig. 5F and G. These data indicate that SIRT1 is crucial for UnAG-mediated SOD-2 expression and p53 and H3K56 deacetylation.

Oxidative stress–mediated EC activation strictly controls inflammatory cell recruitment in ischemic conditions (2,3). We found that the ischemic muscles of UnAG mice had a reduced number of CD68-positive inflammatory cells (Fig. 6A). In order to investigate the possibility that UnAG treatment protects ECs against oxidative stress–induced activation, the expression level of VCAM-1 was evaluated. Indeed, we found that vessels from UnAG animals express low levels of VCAM-1 at days 7 and 21 after ischemia when compared with control animals (Fig. 6B). Similar results were obtained both ex vivo (Fig. 6C) and in vitro when diabetes and ischemia were recapitulated (Fig. 6D and Supplementary Fig. 2B).

It has been reported that molecular approaches that force SOD-2 expression led to reduced VCAM-1 expression (9,10). Thus, ECs that had been silenced for SOD-2 were evaluated (Supplementary Fig. 2C). As expected, high levels of VCAM-1 and an increased number of mononuclear cells that were adherent to ECs were detected, even in presence of UnAG (Fig. 6E and Supplementary Fig. 2D). Similar results were obtained when ECs were silenced for SIRT1 (Fig. 5H).

Besides precise transcriptional regulation, posttranscriptional regulation involving miR-126 is known to play a role in VCAM-1 expression (11). Indeed, reduced miR-126 plasma levels have been observed in diabetic patients' miR profile (12). Therefore, miR-126 levels were evaluated first on ECs recovered from ischemic muscles of UnAG- and saline-treated mice. The results shown in Fig. 7A demonstrate that the miR-126 level decreased even in ECs from saline animals. By contrast, the miR-126 level was high in ECs recovered from UnAG mice. Again, the effects of UnAG were evaluated in vitro. In accordance with the ex vivo experiments, a UnAG challenge led to miR-126 upregulation in ECs treated with AGE and subjected to ischemia (Fig. 7B, left panel). In order to gain further insight into the possibility that the posttranscriptional mechanism controls VCAM-1 expression upon UnAG treatment, the direct effect of miR-126 on VCAM-1–3′-untranslated region (UTR) was evaluated by transfecting ECs with the luciferase reporter vector containing wild-type full-length VCAM-1–3′-UTR. Increased luciferase activity was only detected in VCAM-1–3′-UTR that expressed ECs that had been exposed to AGE and in vitro ischemia, but not when the same experiments were performed in the presence of UnAG (Fig. 7C).

Furthermore, UnAG no longer had any effect on ROS production, VCAM-1 expression, or the adhesion of mononuclear cells to ECs (Fig. 7D and Supplementary Fig. 3) in loss-of-function experiments involving the transfection of ECs with anti–miR-126 antago-miRs (Fig. 7B, right panel). Interestingly, we also noticed that SOD-2 and SIRT1 content, as well as p53 and H3K56 deacetylation, were reduced by the knocking down of miR-126 (Fig. 7E–G). Furthermore, UnAG was no longer effective in protecting ECs from ROS production.

In this study, we have shown that UnAG improves hind limb functional score in a mouse model of insulin resistance and PAD by protecting vessels against ROS-mediated damage. In particular, we have noticed that UnAG rescues ROS imbalance by restoring miR-126 expression, as well as SIRT1 and SOD-2 levels in vessels from ischemic muscles. These results indicate that restoring the AG/UnAG ratio in an insulin resistance setting can offer clinical benefits in the treatment of PAD and provide novel mechanistic insight into UnAG action.

The imbalance of two different forces, ROS and antioxidants, is crucial in long-term diabetic complications, including PAD (3,5,37). Indeed, the persistence of an oxidative milieu and the inefficient antioxidant machinery in diabetic patients with PAD also contribute to mitochondrial damage, leading to the activation of a vicious circle that clinically results in the appearance of symptoms (3,5,37). In the current study, we demonstrate that UnAG administration to mice with insulin resistance and PAD induces skeletal muscle regeneration. A resident population of adult stem cells, named SCs, is present in skeletal muscles and contributes to muscle regeneration after ischemia (43). In accordance with our previous data (19), we found that UnAG promotes SC expansion in response to ischemia. This effect results in a low damage score that, however, appears to be independent of neovascularization (PCNA and cyclin D1 content in ECs recovered from both saline and UnAG mice was similar), but arguably dependent on its antioxidant effects on ECs. As proof of concept in UnAG mice, the number of vessels in ischemic muscles was comparable to the number in nonischemic ones, and ECs recovered from ischemic muscles exhibited low levels of intracellular ROS. This suggests that UnAG, besides inducing muscle regeneration, exerts its beneficial effects in our PAD model of insulin resistance by improving the antioxidant machinery in ECs as well.

Physiologically, superoxide radicals from mitochondria are dismutated to hydrogen peroxide and oxygen by a class of enzymes called superoxide dismutases (37). Such a first line of defense against ROS mainly depends on SOD-2 (37). Human and rodent studies provide evidence for impaired SOD-2 expression in PAD muscles, particularly in patients with diabetes (37). We demonstrate in this study that SOD-2 expression is also reduced in the vessels of saline-treated mice, while it was highly expressed in vessels from UnAG-treated animals, in vivo and ex vivo. That UnAG can induce EC resistance against oxidative stress via SOD-2 is further sustained by data obtained in SOD-2–silenced ECs.

It has been reported that glucose-mediated oxidative stress and NAD⁺ depletion is associated with the impaired expression of SIRT1 in diabetic patient peripheral cells (40). We found that, in ob/ob mice, the SIRT1 level was lower in ECs recovered from both ischemic and nonischemic muscles of saline mice, while it increased upon a UnAG challenge. A number of SIRT1 targets have been described in mammals (44), and tumor suppressor p53 is included (38). Transcription factor p53 is involved in DNA damage mechanisms (45). However, it also acts as a negative regulator of cell proliferation in human atherosclerotic lesions (46) and contributes to accelerated vascular cell senescence in diabetes (47). We have previously reported that the silencing of p53 in EPCs from diabetic patients prevents cell senescence (47). We demonstrate in this study that UnAG-induced SIRT1 expression results in p53 deacetylation and EC protection against senescence in the ischemic condition. The pS36p66Shc level further sustains this (48).

The occurrence of vascular complications in diabetes has recently been linked to the complex interactions between genes and environmental cues that are driven by the persistence of high glucose concentrations (6–8). The impaired clearance of ROS of mitochondria origin is the most relevant environmental cue in the long-lasting detrimental effect of hyperglycemia (6–8). Growing amounts of evidence indicate that mechanisms that account for so-called “metabolic memory” rely on the epigenetic regulation of gene expression, which can be irreversible over time (6–8). The important role that epigenetics play in long-term diabetes-associated vascular complications means that particular attention has been devoted to developing therapeutic strategies that erase epigenetics. We demonstrate in this study that, in diabetic mice subjected to ischemia, UnAG has no effect on blood glucose metabolism but interferes with some epigenetic, hyperglycemia-associated mechanisms, including histone modifying enzymes and miRs. In fact, UnAG increases SIRT1 expression by restoring miR-126 expression in ECs, which is, in turn, involved in p53 and H3K56 deacetylation and protection against DNA damage (8,13,14). Moreover, SIRT1 rescue translates into SOD-2 expression, which restores an efficient antioxidant response and further protects vessels against ROS-mediated damage. Overall, UnAG treatment recapitulates the beneficial vascular effects of SOD-2 overexpression (6,9,10). Finally, UnAG also prevents EC activation and inflammatory cell recruitment into ischemic tissues via miR-126–driven VCAM-1 posttranscriptional regulation. It has been extensively documented that the oxidant-sensitive expression of adhesion molecules is a crucial determinant of lesion progression during atherosclerosis evolution (3,5). In addition, ROS imbalance is believed to play a major role in accelerating vascular dysfunction and PAD in diabetes. Thus, UnAG might also protect vessels that are prone to atherosclerosis from undergoing lesion progression by rescuing miR-126 expression and interfering with mechanisms that control inflammatory cell homing. Finally, as NO bioavailability loss in ECs is a crucial feature of endothelial dysfunction, the increased endothelial NO detected in ECs treated with UnAG, besides controlling vascular tone, could further prevent atherogenesis and PAD progression.

We have previously shown that UnAG improves diabetic EPC mobilization by rescuing endothelial NO synthase activity in bone marrow (17). We have not analyzed EPC's contribution in the ischemic model; however, the impaired inflammatory cell recruitment observed in UnAG animals strongly suggests that UnAG protection, rather than relying on EPC homing, may rely on their functional improvement. In accordance with this possibility is the fact that renewing miR-126 levels in diabetes rescues EPC functional capabilities and protects them against apoptosis (49). Overall, these data provide further evidence to support the theory that restoring the AG/UnAG ratio in insulin resistance–connoted pathological settings can have an impact on patients' cardiovascular risk factor profile.

The incidence of PAD, in particular, in diabetic patients is high and considered the most challenging of medical-surgical diseases (16). As current therapeutic approaches have failed to improve long-term outcomes, new therapeutic options for PAD are urgently needed. Even though an optimal treatment has yet to be identified, our results indicate that UnAG administration to ob/ob animals counteracts oxidative stress-mediated vascular dysfunction and offers clinical benefits in the treatment of PAD (Fig. 8). Although we have no direct evidence to suggest that restoring UnAG plasma concentrations in the insulin resistance state can ameliorate overall clinical outcomes, data provided in this study suggest that UnAG or UnAG-related peptides may constitute an alternative therapeutic approach to the management of one of the most relevant vascular complications in insulin resistance–connoted pathological settings. Finally, the results of this study, together with the role of mitochondrial dysfunction in the development of several metabolic and age-related disorders, suggest that exploiting UnAG-related peptides or UnAG receptor-specific agonists may also be a useful therapeutic strategy with which to address a number of currently unmet medical needs that require ROS scavenging.

Funding. This work was supported by grants from the Ministero dell’Università e della Ricerca Scientifica Progetti di Ricerca di Interesse Nazionale and Unito-Compagnia S. Paolo (to M.F.B., R.G., and E.G.), Fondazione per la Ricerca Diabetologica (FO.Ri.SID) (to M.F.B.), and Ministero della Salute (New Idea Award) and AIRC 5x1000 (SPMCO 9979) (to R.F.).

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

Author Contributions. G.T. was involved in the in vivo experiments, immunoprecipitation, cell culture, and in vitro ischemia. A.T. was involved in SC isolation, histological analysis FACS, and statistical analysis. P.D. was involved in EC isolation, immunofluorescence, adhesion experiments, and luciferase assay. S.G. was involved in Western blotting, RNA isolation, and quantitative real-time PCR. A.R. was involved in construct generation and transfection. P.C. performed animal anesthesia and contributed to the implementation of the murine hind limb ischemia model. R.G. revised the manuscript. R.F. and T.D. made scientific contributions. E.G. revised the manuscript. M.F.B. was involved in study conception and design and in writing the manuscript. M.F.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.