Fenofibrate Rescues Diabetes-Related Impairment of Ischemia-Mediated Angiogenesis by PPARα-Independent Modulation of Thioredoxin-Interacting Protein

Diabetes-related lower limb amputation substantially impairs quality of life, imposing a major burden on the patient and health care systems. Despite improvements in the management of known risk factors, there has been no significant improvement in amputation rates among patients with diabetes in the last 30 years (1). The pathogenesis of diabetic vascular complications, including poor wound healing, impaired collateral development following vascular occlusion (2), and increased rates of lower limb amputations (3), is characterized by impaired angiogenesis in response to ischemia (4). Diabetes is associated with reduced vascular endothelial growth factor (VEGF) production following lower extremity ischemia (5) and reduced sensitivity to VEGF action (6). However, the molecular mechanisms for diabetes-related impairment in angiogenesis remain incompletely understood, and current treatments for diabetic complications are not fully effective.

Fenofibrate, a peroxisome proliferator–activated receptor α (PPARα) agonist, has traditionally been used for improving dyslipidemia (lowering triglyceride levels, increasing HDL levels, and reducing LDL levels) in patients with diabetes at risk for cardiovascular disease (7). The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) placebo-controlled randomized trial (8) demonstrated that fenofibrate therapy in patients with type 2 diabetes substantially reduced the risk of nontraumatic lower limb amputations (9). These effects of fenofibrate were independent of baseline lipid profiles and blood pressure (9); hence, fenofibrate may act via pleiotropic effects.

We hypothesize that fenofibrate may protect from diabetes-related amputations by rescuing diabetes-impaired angiogenesis. In this study, we report that fenofibrate restores diabetes-related impairment of hindlimb ischemia (HLI)–mediated neovascularization to nondiabetic control levels and attenuates high glucose–mediated impairment of endothelial cell (EC) function in vitro. Furthermore, we demonstrate that these proangiogenic effects of fenofibrate on diabetes-impaired angiogenesis are mediated by PPARα-independent regulation of thioredoxin-interacting protein (TXNIP), an exquisitely glucose-sensitive protein involved in the regulation of cellular homeostasis and angiogenesis (10).

Animal studies were performed under approval of the Sydney South West Area Health Service Animal Welfare Committee (#2013-095A). Diabetes was induced in male C57BL/6J or PPARα-knockout (KO) mice (B6) (129S4-Ppara^tm1Gonz/J; The Jackson Laboratory), as previously reported (n = 8/group/treatment) (11). Briefly, 6-week-old mice were fed with a 23% high-fat diet (HFD) for 6 weeks and injected intraperitoneally with a single dose of streptozotocin (STZ) (100 mg/kg). Diabetes was diagnosed as fasted blood glucose >15 mmol/L for 2 consecutive days. Mice received either fenofibrate (Abbott Laboratories) (30 mg/kg/day) in an HFD or HFD alone after developing diabetes at 2 weeks post–STZ injection and for the duration of the experiment. Age-matched wild-type (WT) and PPARα KO mice fed with normal chow diet served as controls.

After 2 weeks of fenofibrate pretreatment, HLI was induced in the left hindlimb by ligation and excision of femoral vein and superficial and deep femoral arteries as described previously (12), with a sham procedure conducted on the right hindlimb (nonischemic control). Blood perfusion recovery was measured using laser Doppler imaging (moorLDI2-IR; Moor Instruments, Devon, U.K.) as the laser Doppler perfusion index (LDPI) (ischemic/nonischemic blood flow). Mouse foot movement was scored at days 0, 2, 5, 7, and 14 post-HLI (3, normal; 2, plantar but not toe flexion; 1, no flexion; and 0, dragging of foot). Mice were euthanized 5 days post-HLI for protein expression by Western immunoblotting and 14 days to assess for neovascularization by immunohistochemistry (n = 8/group).

Gastrocnemius muscle samples were stained for capillary ECs by CD31 (ab28364; Abcam), myocytes by laminin (ab79057; Abcam), and vessel diameter by hematoxylin and eosin (H&E) staining. Capillary-to-myocyte ratio and vessel diameter were calculated using ImageJ software (National Institutes of Health).

For our in vitro studies, primary mouse lung ECs were isolated as reported elsewhere (13). Briefly, lung tissues from male C57BL/6J or PPARα KO mice were digested in collagenase (Worthington Biochemical Corporation). ECs were selected with anti-mouse CD31 antibody-coated magnetic Dynabeads (Life Technologies) and tested for purity (>90%) by flow cytometry. Murine ECs, human umbilical vein ECs (HUVECs), and human artery ECs (Lonza) were grown in MesoEndo medium (Cell Applications). Cells were cultured in low (5 mmol/L) or high (25 mmol/L) glucose conditions for 48 h with fenofibric acid (FFA) (Abbott Laboratories) or DMSO (vehicle control) and then subjected to functional assays or Western blotting. To determine the importance of PPARα, cells were treated with PPARα antagonist MK886 (Tocris Bioscience) and agonist WY14643 (Tocris Bioscience). Parallel cells were pretreated with the broad-spectrum caspase inhibitor Z-VAD-FMK (50 μmol/L) (Selleckchem) 1 h prior to and during glucose and FFA treatments.

Cell migration, tubulogenesis, and apoptosis assays were conducted as previously described (12,14). Cell migration was performed in 0.8-µm pore size Transwell chambers (Corning) with ECs plated in RPMI medium in the top chamber and complete growth medium with or without recombinant VEGF165 (20 ng/mL) (R&D Systems) in the lower chamber. After 16 h, membranes were fixed in ethanol and stained with DAPI to count the cell number. After 4 h in RPMI medium with or without VEGF165 (20 ng/mL) on growth factor–reduced Matrigel (BD Biosciences) tubulogenesis was imaged by photomicrography and assessed as tubule area (ImageJ software). Apoptosis was assessed by costaining ECs with FITC-labeled Annexin V and propidium iodide (BD Biosciences) and measuring Annexin V⁺/propidium iodide⁻ apoptotic cells by flow cytometry. Apoptotic cells were also detected using ApopTag Fluorescein In Situ Apoptosis Detection Kit (TUNEL assay; Millipore). Caspase-3/7 activation was measured by Apolert Caspase Assay Plate (BD Biosciences).

Quantitative RT-PCR (qRT-PCR) of TXNIP mRNA was performed using iQ SYBR Green Supermix (Bio-Rad Laboratories) with β-actin or 36B4 as a reference gene. Primer sequences are as follows: human TXNIP (forward: 5′-AGCCTTCGGGTTCAGAAGAT-3′; reverse: 5′-TTGGATCCAGGAACGCTAAC-3′), human β-actin (forward: 5′-GCCGGGACCTGACTGACTAC-3′; reverse: 5′-CGGATGTCCACGTCACACTT-3′), mouse TXNIP (forward: 5′-TGTCAATACCCCTGACCTAATG-3′; reverse: 5′-TGTCATCACCTTCACAGAATCC-3′), and mouse 36B4 (forward: 5′-CAACGGCAGCATTTATAACCC-3′; reverse: 5′-CCCATTGATGATGGAGTGTGG-3′).

Protein extracts were subjected to Western blot analysis and probed with antibodies for TXNIP (NBP1-54578; Novus Biologicals), VEGF (ab46154; Abcam), phosphorylated Akt (4058), total Akt (9272), KDR (2479), phosphorylated endothelial nitric oxide synthase (eNOS) (9571), and total eNOS (9572; Cell Signaling Technology). Even protein loading was confirmed using β-actin (A1978; Sigma-Aldrich).

To determine whether the effects of FFA are KDR dependent, cells were transduced with either a KDR-overexpressing adenovirus (Adv-KDR; Vector Biolabs) or GFP-overexpressing control (Adv-GFP). To elucidate the role of TXNIP in the effects of FFA, TXNIP levels were overexpressed by TXNIP adenoviral (Adv-TXNIP) transduction or suppressed by siRNA-mediated knockdown as previously reported (14).

Data are expressed as mean ± SEM. Differences between treatment groups were calculated using an unpaired t test or a one-way ANOVA with post hoc analyses for pairwise comparisons (Tukey multiple comparison). Significance was set at P < 0.05.

We first studied the effect of fenofibrate on diabetes-related impairment in ischemia-mediated neovascularization. WT C57BL/6J mice developed diabetes after 6 weeks of an HFD and a single dose of STZ (Supplementary Fig. 1A). Following HLI, diabetic mice exhibited impaired blood flow recovery (Fig. 1A and B) and reduced foot movement (Fig. 1C) compared with nondiabetic mice. Diabetic mice that received fenofibrate had significantly increased blood flow recovery compared with untreated controls (P < 0.05) (Fig. 1A and B), with blood flow in the ischemic hindlimb restored to levels of nondiabetic controls. This finding is corroborated by accelerated functional recovery of the ischemic hindlimb in fenofibrate-treated diabetic mice (P < 0.005) (Fig. 1C). Angiogenesis and arteriogenesis were assessed as capillary density and vessel diameter, respectively, after CD31 and H&E staining. Diabetes was associated with a reduction in both ischemia-induced angiogenesis (Fig. 1D) and arteriogenesis (Fig. 1E) compared with nondiabetic mice. Congruent with the blood flow recovery data, fenofibrate treatment of diabetic mice significantly increased capillary density (P < 0.001) and vessel diameter (P < 0.05) to levels similar to nondiabetic controls. Taken together, these data demonstrate that fenofibrate rescues diabetes-related impairment in ischemia-mediated neovascularization.

Impaired VEGF production and signaling are strongly implicated in the pathogenesis of diabetes-impaired angiogenesis (5,15). VEGF signaling activates the Akt/eNOS pathway (16). We assessed expression of VEGF and VEGF receptor 2 (KDR) and the activation (phosphorylation) of downstream signaling molecules (phosphorylated eNOS [peNOS]/eNOS and pAkt/Akt) in the gastrocnemius muscle of ischemic hindlimbs. Diabetes reduced the expression of KDR (Fig. 1F) and the activation of both Akt (Fig. 1G) and eNOS (Fig. 1H) relative to nondiabetic controls. Fenofibrate increased KDR in diabetic mice (P < 0.05), pAkt/Akt (P < 0.01), and peNOS/eNOS (P < 0.001), restoring expression back to nondiabetic levels. Compared with untreated diabetic mice, fenofibrate also enhanced VEGF production in ischemic tissues of diabetic mice (P < 0.01) (Fig. 1I) to levels higher than in nondiabetic WT mice (Supplementary Fig. 2). These results suggest that fenofibrate restores diabetes-related impairment in VEGF signaling in ischemia.

To investigate the role of PPARα in the restorative effects of fenofibrate, the HLI study was repeated in PPARα KO mice. PPARα KO mice developed diabetes after 6 weeks of HFD and a single dose of STZ (Supplementary Fig. 1B). Compared with WT C57BL/6J mice, nondiabetic PPARα KO mice had significantly reduced blood flow recovery after HLI (Fig. 2A), indicating that PPARα plays a role in ischemia-mediated angiogenesis. Similar to WT mice, diabetes impaired blood flow recovery, foot movement score, capillary density, vessel diameter, and VEGF signaling molecules in PPARα KO mice (Fig. 2A–H). Fenofibrate significantly increased blood flow recovery in diabetic PPARα KO mice to levels equal to or better than nondiabetic mice and was associated with the restoration of foot movement score, capillary density, vessel diameter, and VEGF signaling molecules to nondiabetic levels and enhanced VEGF expression (Fig. 2I). The effects of fenofibrate in diabetic PPARα KO mice are strikingly similar to those in diabetic WT mice (Fig. 1), indicating that these protective effects of fenofibrate are largely PPARα independent.

Fenofibrate is a prodrug that is converted into an active metabolite, FFA (17). To further elucidate the proangiogenic mechanisms of fenofibrate in diabetes, we investigated the effects of FFA on key angiogenic events in HUVECs. FFA attenuated the high glucose–mediated impairment of HUVEC migration (Fig. 3A) and tubulogenesis (Fig. 3B) in a dose-dependent manner with maximal effect at a therapeutically relevant concentration of 50 μmol/L (17). FFA exhibited similar effects in human artery ECs, where it also attenuated high glucose–mediated impairment of migration and tubulogenesis (Supplementary Fig. 3). High glucose concentrations induced HUVEC apoptosis, and this was attenuated by FFA at 50 μmol/L as assessed by Annexin V (Fig. 3C), TUNEL assays (Fig. 3D) and caspase-3/7 activity (Fig. 3E), suggesting that FFA protects ECs from the deleterious effects of high glucose on angiogenic function and cell survival. When apoptosis was inhibited, the suppression of VEGF-induced EC migration by high glucose was abrogated, and FFA had no incremental effect on EC migration (Fig. 3F). In contrast to the striking effects of caspase inhibition on cell migration in high glucose, the broad-spectrum caspase inhibitor, Z-VAD-FMK had no effects on high glucose–mediated inhibition of tubulogenesis or the effects of FFA on tubule formation (Fig. 3G).

Impaired sensitivity to VEGF action has been reported in patients with diabetes (18) and in ECs cultured in high glucose conditions (14). Upon addition of exogenous VEGF (20 ng/mL), HUVECs cultured in 5 mmol/L glucose media showed significant increases in both migration (Fig. 4A) and tubulogenesis (Fig. 4B) compared with untreated control subjects. HUVEC sensitivity to exogenous VEGF was abolished in high glucose (25 mmol/L). FFA restored HUVEC sensitivity to VEGF under high glucose conditions, with increases in migration (P < 0.001) and tubulogenesis (P < 0.001). Blocking VEGF action using a monoclonal antibody prevented all FFA-induced improvements in migration and tubulogenesis, confirming that the observed effects were VEGF dependent. Furthermore, overexpression of KDR promotes VEGF-induced EC migration and rescues high glucose–mediated impairment in EC migration (Supplementary Fig. 4). When KDR is overexpressed, FFA has no significant effect on EC migration. These results demonstrate that high glucose–mediated impairment of EC migration is dependent on VEGF action via KDR and are consistent with fenofibrate rescuing impaired EC migration in high glucose by restoring VEGF action through KDR.

The proportion of phosphorylated Akt and eNOS in HUVECs was not significantly depressed by high glucose. Although FFA did not affect Akt and eNOS phosphorylation in 5 mmol/L glucose, treatment with FFA did increase the phosphorylation of Akt (P < 0.05) (Fig. 4C) and eNOS (P < 0.01) (Fig. 4D) under high glucose conditions. The restoration of Akt and eNOS phosphorylation by fenofibrate in high glucose conditions coincided with its effects on rescuing high glucose–impaired, VEGF-induced EC function.

To evaluate the involvement of PPARα in the protective effects of fenofibrate under high glucose conditions in vitro, we subjected HUVECs to a PPARα antagonist (MK886) (20 μmol/L) and a PPARα agonist (WY14643) (50 μmol/L). Consistent with our in vivo observation that PPARα plays a role in angiogenesis (Fig. 2A), the PPARα antagonist MK886 alone reduced EC migration (Fig. 5A) and markedly inhibited tubulogenesis (Fig. 5B) in both 5 mmol/L and 25 mmol/L glucose conditions. Following preadministration of MK886, the PPARα agonist WY14643 had no effect on EC migration or tubulogenesis in either 5 mmol/L or 25 mmol/L glucose. In contrast, FFA attenuated the MK886-induced impairment in migration and tubulogenesis in both glucose conditions. Consistent with previous observations (Fig. 4C and D), VEGF failed to enhance EC migration (Fig. 5C) and tubulogenesis in 25 mmol/L glucose (Fig. 5D). However, FFA rescued high glucose–mediated insensitivity to exogenous VEGF to a proportionately similar degree in both the absence and presence of MK886 (absence vs. presence of MK886: migration, 2.2-fold vs. 1.8-fold increase; tubulogenesis, 1.6-fold vs. 1.7-fold increase upon treatment with FFA). These results indicate that, although PPARα is important for EC function, the protective effects of FFA in attenuating high glucose–induced EC dysfunction are largely PPARα independent.

TXNIP is an exquisitely glucose-inducible gene that is overexpressed in the tissues of animals and humans with diabetes (19,20). TXNIP is a multifunctional protein that is increasingly recognized as a key modulator of angiogenesis signaling by cytokines such as VEGF (10). We have recently shown that targeted TXNIP knockdown rescues diabetes-related impairment in angiogenesis (14). We therefore investigated the potential role of TXNIP in mediating the protective effects of fenofibrate in diabetes. We first assessed the effects of FFA on high glucose–mediated overexpression of TXNIP in ECs. In HUVECs, FFA reversed high glucose–mediated upregulation of TXNIP mRNA and protein (Fig. 6A and C). To investigate whether the modulation of TXNIP by FFA in high glucose is mediated by PPARα, primary lung ECs were cultured from WT and PPARα KO mice and treated with FFA in combination with 5 or 25 mmol/L glucose. FFA attenuated the high glucose–induced upregulation of TXNIP mRNA (Fig. 6B) and protein (Fig. 6D) in ECs from both WT and PPARα KO mice, indicating that fenofibrate normalizes high glucose–induced overexpression of TXNIP and that this is a PPARα-independent effect.

The effects of fenofibrate on TXNIP expression were assessed in the gastrocnemius muscles of diabetic and nondiabetic WT and PPARα KO mice after HLI. In both WT (Fig. 6E and G) and PPARα KO mice (Fig. 6F and H), the diabetes-induced overexpression of TXNIP mRNA and protein was reversed by fenofibrate in both ischemic and nonischemic hindlimbs. The results are strikingly similar between WT and PPARα KO mice, thus supporting a PPARα-independent mechanism of TXNIP modulation by fenofibrate in diabetes.

To elucidate the role of TXNIP in the protective effects of FFA, HUVECs were transduced with a TXNIP-expressing adenovirus (Adv-TXNIP) or an Adv-GFP adenoviral control. Consistent with Fig. 3A and B, Adv-GFP–transduced HUVECs cultured in high glucose had impaired migration and tubulogenesis that was restored upon FFA treatment (Fig. 7A and B). TXNIP overexpression impaired EC function, as we have previously shown (14), and additionally abrogated the protective effects of FFA on high glucose–mediated impairment in EC migration and tubulogenesis. In contrast to TXNIP overexpression, which abrogated the effects of FFA, we find that FFA still augments tubule formation in high glucose conditions (Supplementary Fig. 5) (P < 0.05) in the context of partial siRNA-mediated TXNIP knockdown (50%). This finding is also consistent with a TXNIP-dependent effect but may also indicate that the protubulogenic effects of FFA potentially extend beyond ability to inhibit TXNIP.

In VEGF-mediated migration and tubulogenesis assays, the VEGF sensitivity of the control Adv-GFP–transduced HUVECs was impaired in high glucose and restored by FFA (Fig. 7C and D). Adenoviral overexpression of TXNIP further reduced high glucose–impaired function and, moreover, abrogated the ability of FFA to rescue VEGF-mediated migration and tubulogenesis in high glucose (Fig. 7C and D). Finally, blocking VEGF using a monoclonal antibody prevented all VEGF-induced migration and tubulogenesis. These data demonstrate that the protective effects of FFA may be dependent, at least in part, on the ability of FFA to reduce TXNIP expression.

Fenofibrate is the first pharmacotherapy to significantly reduce the risk of lower limb amputations in type 2 diabetes (9). However, the mechanisms underlying the protective effects of fenofibrate are poorly understood. The salient findings of this study are that: 1) fenofibrate protects from high glucose–mediated EC dysfunction and apoptosis and rescues diabetes-related impairment of ischemia-mediated neovascularization, 2) the protective effects of fenofibrate are associated with rescue of impaired VEGF signaling and cell sensitivity to the action of VEGF, 3) the protective effects of fenofibrate are largely PPARα independent, and 4) exogenous overexpression of TXNIP blocks the protective effects of FFA. Our studies have revealed critical mechanisms for the protective effects of fenofibrate in diabetes-related impairment of angiogenesis, including PPARα-independent rescue of VEGF signaling pathways and suppression of TXNIP. We provide new insights for understanding the reduction in lower limb amputations seen with oral fenofibrate in patients with type 2 diabetes.

Impaired angiogenesis is a hallmark of the vascular complications of diabetes (21), and key to this is the diabetes-related impairment of VEGF signaling (4), characterized by reduced VEGF production (5), reduced KDR availability (6), and attenuated sensitivity to VEGF action (18). Our data demonstrate that fenofibrate is involved in restoring the VEGF signaling pathways that are impaired by diabetes or hyperglycemia, augmenting KDR expression, and increasing sensitivity to VEGF action. Furthermore, our VEGF-blocking and KDR-overexpressing experiments show that the proangiogenic action of fenofibrate is dependent on VEGF action via KDR, suggesting that this is a key mechanism by which fenofibrate rescues impaired angiogenesis. We previously showed that TXNIP knockdown restores high glucose–impaired VEGF expression and sensitivity to VEGF action (14). TXNIP is known to mediate nuclear export of von Hippel–Lindau protein and target degradation of hypoxia-inducible factor (HIF)-1α, a key promoter of VEGF (22). Furthermore, the COOH-terminal arrestin domain of TXNIP was shown to downregulate VEGF expression via HIF-1α (23). In this study, we show that fenofibrate inhibited hyperglycemic induction of TXNIP, and importantly, TXNIP overexpression abrogated the proangiogenic effects of fenofibrate on VEGF sensitivity. Based on these findings and in tandem with previous studies, we therefore postulate that fenofibrate may be modulating VEGF expression via TXNIP.

Fenofibrate has been reported to attenuate several mediators involved in inflammation, metabolic reprogramming, oxidative stress, and endothelial dysfunction (17) and may therefore modulate the intracellular signaling pathways implicated in diabetic vascular complications. We find that fenofibrate protects cells from high glucose–induced death via inhibition of caspase-3/7 activity. Interestingly, when apoptosis is inhibited using a caspase inhibitor, this abrogates high glucose–mediated inhibition of VEGF-mediated EC migration but not tubulogenesis. With caspase inhibition, FFA has no incremental effects on EC migration but still significantly enhances tubulogenesis. These findings indicate that FFA protection from high glucose–induced impairment in migration is dependent on its antiapoptotic effects, whereas its protection of tubulogenesis is independent of these effects. These divergent findings highlight that although EC migration and tubulogenesis are both functions of angiogenesis, their regulation can be subjected to differences.

Previous studies have shown that PPARα regulates ischemia-mediated angiogenesis. PPARα activation, using a selective PPARα agonist, increased neovascularization in a murine corneal angiogenic model; however, angiogenesis was absent in PPARα-KO mice (24). PPARα is also involved in hypoxia-driven angiogenesis in zebrafish embryonic development (25). Consistent with this, our current study showed that PPARα KO mice exhibited delayed blood flow recovery compared with WT mice in vivo, an effect also observed in nondiabetic mice. Nevertheless, fenofibrate rescues diabetes-related impairment in neovascularization in both WT and PPARα KO mice, indicating that the proangiogenic effects of fenofibrate occur largely via a PPARα-independent mechanism.

Our current study shows that the proangiogenic action of fenofibrate may be attributed, at least in part, to its ability to suppress the induction of TXNIP in high glucose. High glucose–induced TXNIP overexpression has been implicated as a key mediator of diabetes-impaired angiogenesis (Fig. 8A) (10). We previously found that TXNIP knockdown rescued diabetes-related impairment of ischemia-induced neovascularization (14). In this study, we now show that fenofibrate reverses high glucose–induced overexpression of TXNIP via a PPARα-independent mechanism. The mechanism by which fenofibrate suppresses TXNIP expression remains to be elucidated. TXNIP is both a redox- and glucose-sensitive protein; it is therefore highly likely that many signaling pathways/regulators that influence cellular oxidative stress and/or high glucose conditions may be involved. For example, fenofibrate was reported to enhance neovascularization via an endothelial nitric oxide (NO)–dependent mechanism, augmenting the production and downstream signaling of eNOS (26,27). Consistent with this, we found that fenofibrate rescues diabetes or high glucose–induced attenuation of eNOS activation. TXNIP expression is suppressed by NO via transcriptional regulation (28). Therefore, we postulate that fenofibrate induces eNOS activation and NO production that in turn suppresses TXNIP expression in high glucose. In support of this, we have previously shown that siRNA silencing of TXNIP increases NO production (14). Taken together, these findings suggest that fenofibrate may inhibit TXNIP in high glucose through induction of NO.

Interestingly, we find that FFA retains its ability to promote tubule formation when TXNIP is inhibited, suggesting that there are additional mechanisms that contribute to the proangiogenic action of fenofibrate. Fenofibrate has also been shown to enhance ischemia-induced angiogenesis via other signaling pathways, including the adiponectin-dependent AMPK/eNOS regulatory signaling (29). As AMPK is an important regulator of TXNIP transcription in the presence of high glucose (30), we postulate that fenofibrate regulates EC survival via a novel PPARα-independent mechanism of TXNIP modulation, mediated by AMPK. The eNOS signaling pathway is a critical mediator of ischemia-driven angiogenesis and is one of the mechanisms that underpins diabetes-impaired responses (31). We find that FFA augments eNOS phosphorylation, suggesting that the proangiogenic effects of FFA may also be mediated through direct activation of eNOS signaling. Additionally, it has been reported that fenofibrate downregulates fatty acid–induced TXNIP transcription via SIRT1 upregulation and AMPK signaling (32). Conversely, fenofibrate has been shown to suppress angiogenesis in pathological inflammatory angiogenesis (33). Fenofibrate inhibits pathological ocular angiogenesis by suppressing cytochrome P450 epoxygenase (CYP)2C activity, which decreases CYP2C-derived proangiogenic metabolites involved in the development of retinopathy (34). Moreover, fenofibrate suppresses proliferation and migration of neuroblastoma cells by upregulating TXNIP, in which depletion of TXNIP, not PPARα, attenuates the effect of fenofibrate on tumor cells (35). These studies suggest that the mechanisms by which fenofibrate’s effects on angiogenesis differ may be related to tissue-specific factors and/or the physiologic/pathologic conditions. FFA may also have differential effects depending on the vasculature (micro- vs. macrovasculature). Similar differential regulatory effects of angiogenesis in physiological and pathological conditions have been reported for statins (36) and HDLs (37).

In summary, our investigations have shown that fenofibrate rescues diabetes-related impairment of ischemia-mediated neovascularization in vivo and high glucose–mediated EC dysfunction in vitro. We find that the proangiogenic action of fenofibrate in diabetes is independent of PPARα and 1) suppresses hyperglycemic induction of TXNIP, 2) increases VEGF sensitivity, and 3) activates downstream signaling pathways including eNOS to improve endothelial function, ultimately increasing angiogenesis (Fig. 8B). These findings enhance understanding of the mechanisms underlying the therapeutic effects of fenofibrate in reducing nontraumatic lower limb amputations in people with type 2 diabetes.

Acknowledgments. The authors thank David Celermajer of the Cardiology Department, Royal Prince Alfred Hospital, New South Wales, Australia, for critical review of the manuscript.

Funding. This work was supported by National Health and Medical Research Council of Australia (NHMRC) project grants to M.K.C.N. (512299 and 1066541). K.R. was supported by an NHMRC Post-Graduate Scholarship and the NHMRC Clinical Trials Centre Scholarship. A.J.J. is a University of Sydney, Sydney Medical School Foundation Fellow. A.C.K. is an NHMRC Fellow.

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

Author Contributions. J.Y. designed the study, executed the experiments, analyzed the data, and drafted the manuscript. J.T.M.T. designed the study, analyzed the data, and revised the manuscript. K.R. designed the study, executed the experiments, and analyzed the data. E.L.S., E.J.K., and L.L. executed the experiments and analyzed the data. P.J.L.S. interpreted the data and wrote the manuscript. Y.T.L. analyzed the data and revised the manuscript. A.J.J. conceived the project and designed the study. C.A.B. designed the study, analyzed the data, and revised the manuscript. A.C.K. and M.K.C.N. conceived the project, designed the study, and critically interpreted the data. All authors contributed to the critical review and final approval of the manuscript. M.K.C.N. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the American Heart Association Scientific Sessions, Chicago, IL, 15–19 November 2014, and the Cardiology Society of Australia and New Zealand Annual Scientific Meeting, Auckland, New Zealand, 4–6 June 2015.