High-Density Lipoproteins Rescue Diabetes-Impaired Angiogenesis via Scavenger Receptor Class B Type I

The vascular complications of diabetes are characterized by disordered angiogenesis and impairment of ischemia-induced neovascularization. Sufferers of diabetes have reduced coronary collateral formation following vascular occlusion (1), impaired wound healing, and increased rates of amputation (2). Despite advances in the treatment of athero-occlusive disease and the demonstration that intensive blood glucose control attenuates some vascular complications, many patients with diabetes and vasculopathy remain refractory to current treatment approaches.

Diabetes-impaired ischemia-induced neovascularization is associated with decreased hypoxia inducible factor-1α (HIF-1α) stability (3,4), reduced vascular endothelial growth factor (VEGF; specifically VEGFA) production and signaling via VEGF receptor 2 (VEGFR2) (5), and inhibition of endothelial nitric oxide synthase (eNOS) activity (6). However, the key triggers for these events are incompletely understood, as are potential therapies to minimize these abnormalities.

To date, preclinical studies have consistently demonstrated that high-density lipoprotein (HDL) and its main protein constituent, apolipoprotein (apo)A-I, exert antiatherogenic effects (7). Despite this, randomized trials of HDL-raising therapies have not demonstrated clinical benefit. However, long-term mortality follow-up of the Helsinki Heart study found that the incidence of coronary heart disease was significantly reduced when HDL levels were raised by only 5–10% (8), indicating that the biology of HDL is yet to be fully elucidated, and perhaps aggressive attempts to raise HDL have other effects at the cellular and molecular level. Increasing evidence demonstrates that HDL exerts endothelial protective and antidiabetic effects. Infusions of reconstituted HDL (rHDL) reduce plasma glucose levels, restore impaired endothelial function (9), and promote endothelial progenitor cell mobilization (10) in patients with type 2 diabetes. HDL interacts with the cholesterol transporters ABCA1 and ABCG1 and scavenger receptor class B type I (SR-BI). Although ABCA1 and ABCG1 are predominantly involved in HDL-mediated cholesterol efflux, SR-BI is known to mediate the vasculoprotective effects of HDL, including increasing re-endothelialization and endothelial cell migration (11,12). Despite a significant amount of work, the vascular biological effects of HDL remain incompletely understood and further efforts to translate HDL into a potential therapeutic agent require a fuller understanding of its properties.

We therefore sought to investigate the effect of rHDL on diabetes-impaired angiogenesis in two murine models of diabetic vascular complications and to elucidate the mechanisms of action. We report that rHDL rescues diabetes-related impairment of ischemia-driven angiogenesis and wound healing. This occurs via the receptor SR-BI and by HIF-1α stabilization, enhanced VEGFA/VEGFR2 production and signaling, and increased eNOS activity. These findings may have implications for therapeutic modulation of diabetic vascular complications.

apoA-I was isolated from plasma obtained from a pool of multiple healthy donors (>5) by ultracentrifugation and anion-exchange chromatography, as described previously (13). Discoidal rHDL was prepared by complexing apoA-I with 1-palmitoyl-2-linoleoyl-phosphatidylcholine.

All experimental procedures were conducted with approval from the Sydney Local Health District Animal Welfare Committee. Male 8-week-old C57Bl/6J, SR-BI^−/−, and wild-type (WT) littermates were rendered diabetic 2 weeks prior to surgery by a bolus intraperitoneal injection of streptozotocin (165 µg/g).

The hindlimb ischemia model was conducted as described previously (14). The left femoral artery and vein were ligated and excised from the hindlimb of mice (n = 8–12/group). A sham procedure was performed on the opposite hindlimb. Mice received intravenous injections of PBS (vehicle control) or rHDL (200 μg/mouse) via the tail vein every second day following surgery. Hindlimb blood reperfusion was determined by laser Doppler perfusion imaging prior to and immediately following surgery and then at days 2 to 3, 7, and 10 postsurgery.

The wound healing model was conducted as previously described (15). Two full-thickness excisions were created on the dorsum and a silicone splint secured around the wound. For each mouse, one wound received rHDL (50 µg/wound/day) and the other PBS topically applied directly on the wound. A transparent occlusive dressing (Opsite) was applied. Digital images and wound area were measured daily. Wound blood perfusion was determined using laser Doppler.

Total, HDL, and LDL cholesterol concentrations on mouse plasma were determined enzymatically (Roche Diagnostics). HDL cholesterol concentrations were determined following polyethylene glycol precipitation of apoB-containing lipoproteins. Glucose concentrations were measured using a glucometer (Accu-Chek Performa; Roche).

Fresh frozen 5-µm sections of gastrocnemius muscle from ischemic and nonischemic hindlimbs were stained to detect the number of new capillaries (CD31⁺; DakoCytomation) per myocyte (laminin; Abcam). The 5-µm sections were taken from the midpoint of paraffin-embedded wound tissues and assessed for CD31⁺ neovessels (Abcam).

Human coronary artery endothelial cells (HCAECs; Cell Applications) were cultured in MesoEndo media and used at passages 4 to 5. Cells were seeded at 8 × 10⁴ cells/well, cultured for 8 h, and then treated for 18 h with rHDL (20 μmol/L; final apoA-I concentration) or PBS. Cells were replaced with fresh DMEM media in glucose conditions for 48 h. For high glucose conditions, media was supplemented with d-glucose to a final concentration of 25 mmol/L. For the measurement of phosphorylated proteins, cells were stimulated with 10 ng/mL recombinant human VEGF protein (R&D Systems). Each experiment was performed three times independently and in triplicate.

HCAECs were transfected for 6 h with 60 nmol/L small interfering RNA (siRNA) for Siah1, Siah2, or control scrambled (Santa Cruz Biotechnology), then treated with rHDL (20 μmol/L) or PBS for 18 h, and exposed to high glucose (25 mmol/L, 48 h).

Lentiviruses containing either short hairpin RNA (shRNA) for SR-BI (shSR-BI) or the empty vector (shControl) were generated in 293T17 cells as described previously (16,17). Viral titers were quantified using Lenti-X qRT-PCR Titration Kit (Clontech).

HCAECs were seeded and grown overnight to 50% confluency and then exposed for 24 h to 1 × 10⁷ lentiviral particles/mL containing either shSR-BI or shControl. Transduced cells were seeded at 8 × 10⁴ cells/well. Cells were treated with rHDL (20 μmol/L) or PBS for 18 h then exposed to high glucose (25 mmol/L, 48 h).

Quantitative real-time PCR was performed for: 1) Vegfa (F: 5′-GGCTGCTGTAACGATGAAG-3′; R: 5′-CTCTCTATGTGCTGGCTTTG-3′), murine Glut1 (forward [F]: 5′-TCAACACGGCCTTCACTG-3′; reverse [R]: 5′-CACGATGCTCAGATAGGACATC-3′), Ppargc1a (F: 5′-TGGAGTGACATAGAGTGTGCTG-3′; R: 5′-TGTTCGCAGGCTCATTGTTG-3′), Hif-2α (F: 5′-AGGTCTGCAAAGGACTTCGG-3′; R: 5′-CAAGTGTGAACTGCTGGTGC-3′), prolyl hydroxylase (Phd) 1, (F: 5′-TAAGGTGCATGGCGGCCTGC-3′; R: 5′-TGGCTGCTGCCCGTTCCTTG-3′), pyruvate dehydrogenase lipoamide kinase isozyme 4 (Pdk4) (F: 5′-CACGTACTCCACTGCTCCAA-3′; R: 5′-AGCGTCTGTCCCATAACCTG-3′), Scarb1 (F: 5′-CTGAGCACGTTCTACACGCA-3′; R: 5′-GGCCTGAATGGCCTCCTTAT-3′), Siah1a (F: 5′-GACTGCTACAGCATTACCCACT-3′; R: 5′-GTTGGATGCAGTTGTGCCG-3′), Siah2 (F: 5′-CTAACGCCCAGCATCAGGAA-3′; R: 5′-GAACAGCCCGTGGTAGCATA-3′), Hif-1α (F: 5′-TCCCTTGCTCTTTGTGGTTGGGT-3′; R: 5′-AACGTAAGCGCTGACCCAGG-3′), Vegfr2 (F: 5′-GCCCAGACTGTGTCCCGCAG-3′; R: 5′-AGCGCAAGACCGGGGAGAGC-3′), and 36B4 (F: 5′-CAACGGCAGCATTTATAACCC-3′; R: 5′-CCCATTGATGATGGAGTGTGG-3′) in murine hindlimbs and wound tissue, and 2) human SIAH1, SIAH2, and β₂-microglobulin (B2M) in cultured HCAECs using primers designed previously (16). Relative changes in gene expression were normalized using the ^ΔΔ threshold cycle method to murine 36B4 or human B2M.

Whole-cell and nuclear protein extracts were subjected to Western blot analysis and probed with antibodies for phosphoinositide 3-kinase (PI3K; p85) (Abcam), phosphorylated Akt (Ser⁴⁷³), total Akt (Cell Signaling Technology), PHD2, PHD3, HIF-1α (Novus Biologicals), VEGFA (Abcam), VEGFB, VEGFC, VEGFD (R&D Systems), phosphorylated VEGFR2 (Tyr¹¹⁷⁵), total VEGFR2, phosphorylated eNOS (Ser¹¹⁷⁷), total eNOS (Cell Signaling Technology), and SR-BI (Novus Biologicals). Even protein loading was confirmed by α-tubulin (Abcam) for whole-cell lysates or lamin B1 (Abcam) for nuclear fractions. Secreted and cytoplasmic VEGFR1 expression were measured in the media and cytoplasmic fractions by ELISA (R&D Systems).

Pretreated HCAECs were seeded at 8 × 10³ cells/well on polymerized growth factor–reduced Matrigel and incubated for 4 h. Tubules were photographed at ×40 magnification under light microscopy and total number tubules formed determined using ImageJ (National Institutes of Health).

Data are expressed as mean ± SEM. Differences between treatment groups were calculated using a one-way ANOVA (Bonferroni comparison test post hoc) or Student t test. A two-way ANOVA (Bonferroni comparison test post hoc) was performed when comparing data at multiple time points. Significance was set at a two-sided P < 0.05.

We first studied the effects of rHDL in two models of diabetes-impaired vascular complications including: 1) ischemia-mediated neovascularization, and 2) wound healing and angiogenesis. In the hindlimb ischemia model, femoral artery ligation reduced blood flow equally in all mice at day 0 (Fig. 1A). In the nondiabetic mice, rHDL infusions promoted blood flow recovery compared with PBS-infused mice, reaching significance at day 7 (P < 0.05). Diabetes severely impaired blood flow recovery in PBS-infused mice. However, this was rescued by rHDL infusions (P < 0.01). Consistent with this, rHDL increased capillary density in the gastrocnemius muscle of ischemic hindlimbs in both the nondiabetic (P < 0.01) and diabetic mice (P < 0.001) (Fig. 1B). The impact of diabetes on wound healing and angiogenesis was more striking. In nondiabetic mice, topical rHDL application increased wound closure compared with PBS-treated wounds at day 10 (Fig. 1C). As expected, diabetic mice exhibited delayed wound closure. Topical rHDL rescued diabetes-related impairment in wound closure to the level of nondiabetic controls. Furthermore, in nondiabetic and diabetic mice, blood perfusion was elevated in rHDL-treated wounds in the important early stages of wound recovery (days 2–6) (Fig. 1D). Diabetic mice had reduced wound capillary density in PBS-treated wounds (Fig. 1E). However, topical rHDL enhanced capillary density in diabetic wounds (P < 0.05). In both models, we found that diabetes suppressed Vegfa expression, which was rescued by rHDL treatment (Supplementary Fig. 1A and B). Taken together, we show that rHDL rescues diabetes-impaired angiogenesis. These effects were independent of changes in glucose and lipid levels (Supplementary Tables 1 and 2). Furthermore, rHDL had no effect on markers of glucose metabolism, including the glucose transporter Glut1, the metabolic regulator Ppargc1a and three genes involved in metabolic cellular programming: Hif-2α, Phd1, and Pdk4 (Supplementary Fig. 2). This indicates that rHDL does not mediate its effects on angiogenesis via changes in glucose metabolism. Finally, we found there were no differences in Scarb1 (SR-BI) levels between diabetic and nondiabetic animals in both studies (Supplementary Fig. 3A). However, rHDL significantly augmented hindlimb Scarb1 expression in diabetic mice.

We then studied the effects of rHDL on key angiogenic pathways in vitro in high-glucose conditions. Decreased HIF-1α stability, impaired VEGFA production, and eNOS inhibition are strongly implicated in the pathogenesis of diabetes-impaired angiogenesis (3–6). HIF-1α is posttranslationally modulated, beginning with the PI3K/Akt signaling pathway, which induces the E3 ubiquitin ligases Siah1 and Siah2 (18). Under high-glucose conditions, rHDL increased PI3K protein expression (Fig. 2A) (P < 0.05) and induced an increase in phosphorylated Akt (pAkt) (Fig. 2B) (P < 0.001). rHDL increased SIAH1 and SIAH2 mRNA levels, irrespective of glucose conditions (Fig. 2C and D) (P < 0.001). The Siahs suppress PHD2 and PHD3 that ubiquitinate and target HIF-1α for degradation. Exposure to high glucose increased PHD2 and PHD3 protein levels (Fig. 2E and F). However, rHDL suppressed both PHD2 (P < 0.01) and PHD3 (P < 0.05) levels. Consistent with the decreases in PHDs, rHDL rescued high glucose–induced reductions in both total and nuclear HIF-1α protein levels (Figs. 2G and H) (P < 0.05). These results show that rHDL rescues high glucose–impaired HIF-1α stabilization in vitro.

Although the VEGFA/VEGFR2 signaling axis has been strongly implicated in diabetic vascular complications (5), other members of the VEGF ligand-receptor family have also been implicated in diabetes (19–23). We determined the effects of rHDL on the VEGF ligand-receptor family. rHDL prevented high glucose–induced inhibition of VEGFA and VEGFB protein (Fig. 3A and B) (P < 0.05), but did not change VEGFC or VEGFD levels (Fig. 3C and D). Although high glucose did not impact on VEGFR1 expression, rHDL significantly decreased both soluble and cytoplasmic VEGFR1 protein irrespective of glucose conditions (Fig. 3E and F). rHDL promoted VEGFR2 phosphorylation (pVEGFR2) (Fig. 3G) and augmented eNOS phosphorylation in high glucose (Fig. 3H). Finally, consistent with increases in HIF-1α stabilization, VEGFA/VEGFR2 signaling, and eNOS activation, rHDL augmented endothelial cell tubule formation (Fig. 3I) (P < 0.05).

To elucidate the importance of Siah1 and Siah2 in rHDL-induced HIF-1α stabilization and VEGFA augmentation, an siRNA knockdown approach was used. In the scrambled siRNA (siScr) cells, rHDL augmented HIF-1α, VEGFA, pVEGFR2, and tubulogenesis (Fig. 4A–D) in high glucose. However, knockdown of Siah1 and Siah2 abrogated several steps in the angiogenic pathway, including rHDL-induced increases in HIF-1α, VEGFA, pVEGFR2, and tubulogenesis, highlighting an important role for Siahs in mediating the proangiogenic effects of rHDL in high glucose.

In HCAECs, high glucose induced a 50% decrease in SR-BI protein (Supplementary Fig. 3B) (P < 0.05), consistent with previous studies (24–26). Incubation with rHDL prevented high glucose–induced SR-BI inhibition (P < 0.05). The role of SR-BI in mediating the proangiogenic effects of rHDL in high glucose conditions was next assessed in vitro using a lentiviral shSR-BI. In the lentiviral control shRNA (shControl) cells, rHDL augmented pAkt; however, this was abrogated in shSR-BI cells (Fig. 5A). rHDL also failed to augment HIF-1α expression in shSR-BI cells (Fig. 5B). Although rHDL increased VEGFA protein in shControl cells, this induction was abrogated by shSR-BI (Fig. 5C). Finally, rHDL-induced tubule formation was attenuated in shSR-BI cells (Fig. 5D).

We then investigated the role of SR-BI in mediating the effects of rHDL on ischemia-mediated neovascularization in vivo using SR-BI^−/− mice. In nondiabetic WT mice, rHDL augmented hindlimb blood perfusion, which was attenuated in nondiabetic SR-BI^−/− mice (Supplementary Fig. 4A). In diabetic WT littermates, rHDL infusions promoted blood flow recovery, capillary density, and augmented mRNA levels of key angiogenic mediators including Siah1a, Siah2, Hif-1α, and Vegfa (Fig. 6A–F) (P < 0.05). However, the proangiogenic ability of rHDL was completely abrogated in SR-BI^−/− diabetic mice, suggesting that rHDL augments ischemia-induced angiogenesis in diabetes via SR-BI.

In nondiabetic WT mice, rHDL promoted the rate of wound closure, which did not occur in nondiabetic SR-BI^−/− mice (Supplementary Fig. 4B). In diabetic WT littermates, rHDL promoted wound closure and wound angiogenesis and induced Siah1a, Siah2, Hif-1α, Vegfa, and Vegfr2 mRNA levels (Fig. 7A–H) (P < 0.05). The ability of rHDL to promote wound healing and angiogenesis was attenuated in diabetic SR-BI^−/− mice. Taken together with the in vitro studies reported in Fig. 5 and the in vivo hindlimb ischemia studies described in Fig. 6, this highlights an important role for SR-BI in the induction of angiogenesis by rHDL in diabetes.

We report for the first time that rHDL rescues diabetes-impaired angiogenesis by enhancement of ischemia-mediated neovascularization and acceleration of wound closure and wound angiogenesis. In vitro studies indicate that these effects may, at least in part, be via enhanced posttranslational HIF-1α modulation and nuclear translocation, increased VEGFA/VEGFR2 production and signaling, and augmented eNOS activity. Siah siRNA knockdown in vitro confirmed the importance of posttranslational HIF-1α modulation in mediating the proangiogenic effects of rHDL in high glucose. Furthermore, in vitro studies using lentiviral shSR-BI knockdown and in vivo studies with SR-BI^−/− mice indicate that these effects of rHDL are mediated by the receptor SR-BI. In summary, we have demonstrated a key role for rHDL in the attenuation of diabetes-related impairment of angiogenesis with implications for the therapeutic modulation of diabetic vascular complications.

Collateral vessel network development is an important response to tissue ischemia following vascular occlusion (27). Similarly, the extent of neovascularization in the early stages following wounding is a key determinant of wound closure rate (28). Diabetes is associated with poor outcomes following vascular occlusion and impaired coronary collateral development (29), and patients with peripheral vascular disease manifest increased peripheral limb ulceration and amputation, with cutaneous wounds more prone to amputation (2,30). Consistent with this, our in vivo studies found that diabetes caused impairment of ischemia-induced neovascularization in hindlimbs and in wound closure/angiogenesis, with a more striking impact on wound healing. These are two distinctly different models; therefore, the timing for capturing changes may vary. Compared with the hindlimb ischemia model, which is primarily driven by ischemia-mediated angiogenesis, wound healing is more complex and not solely dependent on angiogenesis but involves other cellular processes, including epithelialization and cellular proliferation. Additionally, a chronic inflammatory state as is commonly seen in patients with diabetes, significantly impairs wound healing and can lead to severe unfavorable outcomes such as amputation. HDL also exhibits anti-inflammatory effects (31,32) and may assist in the more robust impact of HDL. Regardless, rHDL promoted neovascularization in both angiogenic models, rescuing both ischemia-mediated neovascularization and wound healing and angiogenesis, potentially highlighting a new role for HDL in attenuating vascular complications associated with diabetes-impaired neovascularization.

Our in vitro studies found that high glucose suppressed posttranslational HIF-1α modulation (via augmentation of PHDs), VEGFA/VEGFR2 production and signaling, and tubulogenesis. HIF-1α is the pivotal transcription factor involved in ischemia-mediated neovascularization and governed by a complex orchestration of posttranslational regulation. Previous studies have reported that hyperglycemia inhibits hypoxia-induced HIF-1α stabilization and suggest that mechanisms involving proline hydroxylation are important (33). The current study, however, is the first to directly show that high glucose decreases HIF-1α stability via posttranslational effects including an increase in PHD (PHD2 and PHD3) expression and a decrease in both total and nuclear HIF-1α protein. More importantly, rHDL rescued HIF-1α stabilization at each step in its posttranslational regulation by: 1) activating the PI3K/Akt signaling pathway, which 2) triggered an increase in the expression of the E3 ubiquitin ligases Siahs that 3) suppressed PHD expression and ultimately 4) rescued hyperglycemia-induced reductions in HIF-1α. These observations complement a recent study that found rHDL increased posttranslational HIF-1α stabilization, but this study was not conducted in high-glucose conditions (17). Nuclear localization studies found rHDL augmented nuclear HIF-1α protein levels, demonstrating strong evidence of the impact of rHDL and the Siah/PHD axis on promoting nuclear HIF-1α translocation. Furthermore, siRNA knockdown of Siahs confirmed the importance of posttranslational HIF-1α modulation in the effects of rHDL in high glucose, as silencing of the Siahs abrogated key steps in the angiogenic pathway in vitro.

As expected, high glucose suppressed VEGFA protein levels in vitro. rHDL treatment prevented this decrease, most likely via the stabilization of HIF-1α, the critical transcription factor mediating VEGFA expression. Our in vitro analysis of the VEGF ligand-receptor family found that rHDL augmented VEGFB but had no effect on VEGFC or VEGFD. Both VEGFA and VEGFB bind to VEGFR1, which is also implicated in diabetes-impaired neovascularization (19,34). However, we found that rHDL decreased VEGFR1 expression, suggesting that VEGFR1 is not mediating rHDL-induced rescue of angiogenesis in high glucose. VEGFR2 is the receptor that regulates the proangiogenic effects of VEGFA. Following binding of VEGFA to VEGFR2, the receptor dimerizes and causes the activation of receptor-kinase activity, leading to the phosphorylation of the receptor. pVEGFR2 induces the activation of an array of angiogenic signaling pathways (35). We found that rHDL increased the phosphorylation (activation) of VEGFR2 in high glucose. This is particularly important, as hyperglycemia reduces VEGFA sensitivity via suppression of VEGFR2 activation (5). Previous studies have found rHDL augments hypoxia-induced VEGFR2 total protein levels (14), but this is the first study to show that in high glucose, rHDL promotes both VEGFA protein expression and signaling via VEGFR2 phosphorylation/activation. The current study also found that rHDL promoted eNOS phosphorylation in high glucose, which is likely to be due to the increase in VEGFA/VEGFR2 and the subsequent increase in Akt phosphorylation (downstream of VEGFR2). Elevated eNOS phosphorylation leads to nitric oxide release, promoting angiogenic functions including endothelial cell migration, proliferation, and vessel growth (36,37). Consistent with this, other studies have found that rHDL increases eNOS activity in vitro and in vivo (38,39). Finally, using a functional Matrigel assay for endothelial tubule formation, we found that rHDL augmented high glucose–impaired tubule formation, suggesting that the effects of rHDL on the key angiogenic proteins is translated into critical cellular processes involved in angiogenic functions.

It is becoming increasingly recognized that HDL exhibits antidiabetic effects (9,40). We show in this study that rHDL rescues diabetes-mediated impairment of hindlimb and wound angiogenesis. This is supported by previous work showing that apoA-I/rHDL augment ischemia-driven angiogenesis and promote angiogenesis-related functions including migration and re-endothelialization (14,17,41). However, these studies were not done in the clinically relevant setting of hyperglycemia, one of the key contributors to diabetes-impaired vascular complications. In the current study, the effects of rHDL were independent of changes in glucose and lipid levels. Our in vivo gene analysis of both hindlimb and wound tissues found that rHDL had no effect on the expression of the glucose transporter GLUT1 and the transcriptional coactivator peroxisome proliferator-activated receptor γ coactivator 1-α, a powerful regulator in metabolism that is involved in HIF-independent regulation of VEGF and angiogenesis in diabetes (42). Hypoxia tolerance and HIF-1α stabilization are central to a hypometabolic state characterized by reduced oxygen consumption, such as that seen in diabetes-impaired neovascularization. At a cellular level, metabolic cellular reprogramming involves increased efficiency of energy-producing pathways, via increased anaerobic glycolysis activity, and decreased energy-consuming processes and is mediated via HIF-2α, PHD1, and PDK4 (43). rHDL did not have any effect on gene expression of any of these markers. Taken together, this confirms that the action of rHDL on diabetes-impaired angiogenesis is not due to glucose stress dependent effects but via a number of mechanisms previously found to contribute to diabetes-impaired neovascularization including: 1) promotion of posttranslational HIF-1α stabilization, 2) VEGFA/VEGFR2 production and signaling, and 3) eNOS activation.

The scavenger receptor SR-BI has been implicated in mediating a number of the endothelial protective effects of HDL such as migration, tubulogenesis, and re-endothelialization (11,12,17,44,45). This is the first study to date that has provided a direct link between SR-BI and the proangiogenic effects of HDL in diabetes. We found that diabetic animals had slightly lower levels of Scarb1 mRNA, although this did not reach significance, whereas SR-BI protein levels were significantly reduced under high-glucose conditions in vitro. Currently, there is no consensus on what happens to SR-BI expression in patients with diabetes, with one study reporting elevated SR-BI mRNA levels in patients with diabetes (46) whereas two studies reported no differences (47,48). Data from in vitro studies are more robust. Exposure to high glucose is shown to reduce SR-BI levels across several cell types, including intestinal Caco-2/15 cells (24), hepatic HepG2 cells (25), and human monocyte-derived macrophages (26). More importantly, rHDL increased SR-BI expression in vivo and in vitro. Our in vitro and in vivo data show that SR-BI is critical in mediating the ability of rHDL to rescue diabetes-impaired neovascularization. rHDL was unable to restore high glucose–impaired angiogenesis in shSR-BI cells in vitro or rescue diabetes-impaired hindlimb and wound angiogenesis/closure in SR-BI^−/− mice in vivo. Furthermore, compared with WT littermates of both the hindlimb ischemia and wound-healing models, we found that the ability of rHDL to augment Siah1a, Siah2, Hif-1α, Vegfa, and Vegfr2 mRNA levels were attenuated in SR-BI^−/− mice. Interestingly, SR-BI is the preferred cholesterol acceptor for spherical HDL particles rather than discoidal rHDL (49). However, it would be expected that following incubation in vitro, systemic injection in vivo, or topical application to wounds, rHDL would rapidly acquire lipid, forming a spherical particle that is able to interact with SR-BI (50). In support of our findings, a recent study found that SR-BI acts as a cholesterol sensor triggering intracellular signaling and is important for the actions of HDL on endothelial cells (45). SR-BI was also important in mediating ischemia-mediated neovascularization and wound healing/angiogenesis in nondiabetic mice. The role of SR-BI in angiogenesis is further supported by reports that show important signaling pathways including PI3K/Akt are downstream of SR-BI and are associated with angiogenesis-related functions such as migration and proliferation (12).

Despite the vast number of reports demonstrating the therapeutic benefits of HDL on the cardiovascular system, to date, there is no translated use of HDL-targeted treatments. There is, however, increasing evidence for the antidiabetic effects of HDL, which may present an alternative translation pathway. We found that rHDL rescues both high glucose–related impairment of tubulogenesis in vitro and diabetes-impaired neovascularization in vivo. The mechanisms for these effects are via increased posttranslational HIF-1α stabilization, VEGFA/VEGFR2 production/activation and signaling, and eNOS activation (Fig. 8). Furthermore, SR-BI is important in mediating rHDL-induced rescue of diabetes-impaired angiogenesis. The current study provides a greater understanding into the vascular biological effects of HDL in the context of diabetic vascular complications. This may ultimately facilitate the translation of HDL, not only for cardiovascular disease but also diseases associated with impaired angiogenesis and the vascular complications of diabetes.

Acknowledgments. The authors thank Pat Pisansarakit, from the Heart Research Institute, for the maintenance of HCAECs and Emily King, from the Heart Research Institute, for the use of the murine Hif-2α, Phd1, and Pdk4 primers.

Funding. This work was supported by the National Health and Medical Research Council of Australia Project Grant (632512 to M.K.C.N. and C.A.B.), Early Career Fellowship (537537 to L.L.D.), PhD Scholarship (APP1038394 to S.C.G.Y.), National Heart Foundation Career Development Fellowship (CR07S3331 to C.A.B.), and PhD Scholarship (PB12S6959 to L.Z.V.).

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

Author Contributions. J.T.M.T., H.C.G.P., and C.A.B. developed the study design, designed the experiments, interpreted the data, performed the experiments, and wrote the manuscript. L.L.D. and S.R. performed the experiments and reviewed and edited the manuscript. L.Z.V., A.R., T.T., L.L., Z.E.C., S.C.G.Y., and Y.T.L. performed the experiments. D.S.C. reviewed and edited the manuscript. M.K.C.N. developed the study design, designed the experiments, interpreted the data, and reviewed and edited the manuscript. C.A.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.

Prior Presentation. Parts of this study were presented in abstract form at the 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.