Hydrogen Sulfide Improves Wound Healing via Restoration of Endothelial Progenitor Cell Functions and Activation of Angiopoietin-1 in Type 2 Diabetes

Diabetes mellitus and its chronic complications, such as foot problems, are prevalent worldwide. In fact, diabetic foot ulcer is the leading cause of nontraumatic amputation (1). That angiogenesis is crucial for wound healing is well known (2,3). Endothelial progenitor cells (EPCs) play an important role in angiogenesis and neovascularization (4). However, much evidence shows that some aspects of EPC angiogenic ability, as measured by tube formation, adhesion, and migration, decrease in diabetes (5,6). EPC-related angiogenic deficiency induced by diabetes causes chronic vascular complications, refractory wounds, and foot ulcers (3,7,8). Therefore, amendment of EPC function is a critical way to improve wound healing.

Mammalian tissues synthesize hydrogen sulfide (H₂S) by two pyridoxal-5′-phosphate–dependent enzymes catalyzing the metabolism of l-cysteine: cystathionine β-synthase (CBS) and cystathionine-γ-lyase (CSE), as well as by a recently identified third pathway by the combined action of 3-mercapto-pyruvate sulfurtransferase and cysteine aminotransferase (9,10). H₂S has been considered a poisonous gas for a long time. However, evidence accumulating over the last decade shows that H₂S could act as a novel gasotransmitter and take part in many physiological and pathological processes such as angiogenesis (11), vasodilation (12–14), and inhibition of apoptosis in vascular endothelial cells (15) and cardiomyocytes (16). In the cardiovascular system, the principal enzyme involved in the formation of H₂S is CSE, which converts cystathionine to l-cysteine, yielding pyruvate, ammonia, and H₂S (17). CSE is expressed in vascular endothelial cells, smooth muscle cells, and myocardial cells (18), but to date, whether CSE is expressed in the EPCs is unclear.

Regarding the relationship between H₂S and diabetes, current evidence shows that the synthesis and circulating level of H₂S decreases in NOD mice (19), in streptozotocin-induced type 1 diabetic mice (20), and in type 2 diabetic patients (20). H₂S has been reported to accelerate gastric ulcer healing (21) and healing of burn wounds in the skin (22,23). Papapetropoulos et al. (22) found that topical administration of H₂S improved the recovery from burn wounds in wild-type rats and that genetic ablation of CSE delayed healing in mice. However, little is known about the role of H₂S in chronic skin wound healing in diabetes, especially in type 2 diabetes. Whether H₂S stimulates angiogenesis by influencing the functions of EPCs has also not been elucidated.

Angiopoietin (Ang) plays a major role in endothelial survival and vascular maturation. As a paracrine agonist, Ang-1 can induce phosphorylation of its receptor Tie 2 and promote endothelial survival and vessel stabilization (24). The Ang-1/Tie 2 signal has been reported to contribute to diabetes-induced angiogenesis impairment (25,26), and reduced Ang-1 expression in high-glucose concentration downregulated the phosphatidyl inositol 3-kinase/Akt signaling pathway and phosphorylated endothelial nitric oxide (NO) synthase, which results in the impairment of EPC vessel-forming capacity (25,27). Vascular endothelial growth factor (VEGF) has been reported as a proverbial factor to stimulate angiogenesis and mediate the improving course of wound healing (28,29).

The current study aimed to test our hypothesis that H₂S might improve angiogenesis and restore EPC function in an in vivo wound-healing db/db diabetic mouse model. We also investigated the expression of Ang-1 in wound skin tissue and EPCs because it is a possible important factor that mediates the angiogenic effects of H₂S donors. Showing that controlling H₂S can improve wound healing effectively and become a potential strategy for chronic lower extremity ulcerations in diabetic populations would be of great clinical significance.

All animal procedures were approved by and conducted in accordance with the Central South University Institutional Animal Care and Use Committee established guidelines. The B6.Cg- m^+/+Lepr^db/J (leptin receptor-deficient diabetes, db/db, 8–12 weeks, purchased from The Jackson Laboratory) mouse is a well-established type 2 diabetic animal model with continuous hyperinsulinemia and high plasma glucose levels (30). Their matched nondiabetic littermates (db/+) were used as controls.

H₂S was administered in the form of sodium hydrosulfide (NaHS; Sigma-Aldrich) and 4-hydroxylthio-benzamide (4-HTB; Alfa Aesar), which is well established as a reliable H₂S donor (22,31). NaHS is a rapid-releasing H₂S donor widely used in recent in vivo and in vitro H₂S studies (22,23,32), and 4-HTB is a slow-releasing H₂S donor that is effective in improving gastric ulcer healing in rats (21). dl-propargyl-glycine (PAG; Sigma-Aldrich), an irreversible competitive CSE inhibitor, was used to confirm the role of H₂S in db/+ mice subjected to skin wounding (33,34). Blood glucose from tail veins was measured with a rapid glucose meter (LifeScan OneTouch Ultra 2), and results were expressed as milligrams per deciliter. Diabetes was diagnosed when blood glucose was higher than 300 mg/dL.

Full-thickness punch biopsy wounds were created on male db/db mice and their age- and sex-matched db/+ mice, as we previously described (35,36). Briefly, 6-mm-diameter full-thickness skin wounds were made on the dorsal skin with a biopsy punch (Acuderm Inc., Fort Lauderdale, FL). Wounds were dressed with Tegaderm (Nexcare, 3M) and changed every other day, and wound-closure rates were measured by tracing the wound area every other day onto an acetate paper. Gross wound closure was quantified by Image-Pro Plus 5.1 software (Media Cybernetics), and wound healing was expressed as the percentage of the original wound area that had healed, calculated as [1 − (wound area day x/wound area day 0)] × 100.

Mice were randomly assigned to five groups: db/db plus NaHS (50 μmol ⋅ kg⁻¹ ⋅ day⁻¹, n = 9), db/db plus HTB (30 μmol ⋅ kg⁻¹ ⋅ day⁻¹, n = 9), db/db control (saline, n = 9), db/+ control (saline, n = 10), and db/+ plus DL-PAG (50 mg ⋅ kg⁻¹ ⋅ day⁻¹, n = 5). These drugs were injected intraperitoneally once daily from the day of wounding until 16 days later (22,23,32). At the end of 16 days, or at day 8 after wound creation, mice were killed under isoflurane anesthesia after the blood draw. Blood samples were drawn from the right ventricles by using syringes containing EDTA and then were centrifuged (10,000g at 4°C) for 10 min. Thereafter, plasma was aspirated and stored at −80°C for H₂S concentration measurement. The skin wound tissues were carefully dissected, and granulation tissues within the wound as well as a 10-mm margin of wound were collected.

To determine whether H₂S improves EPC-mediated wound healing in diabetes, 1 × 10⁶ diabetic EPCs were transplanted on db/db mouse wounds immediately after the skin punch biopsies. Described briefly, one group of db/db mice (n = 8) received an intraperitoneal injection of NaHS (50 μmol ⋅ kg⁻¹ ⋅ day⁻¹) every other day until 16 days later, and one group of db/db mice (n = 8) mice received an intraperitoneal injection of 0.9% saline every other day until 16 days later. These two groups of mice received topical transplants of 1 ×10⁶ diabetic EPCs on wounds, as we described previously (35). Another group of db/db (n = 8) mice used as a control only underwent skin punch biopsies.

H₂S in the plasma was determined using previously described methods (37–39). Plasma (120 μL) was mixed with 100 μL water and 120 μL trichloroacetic acid (10% volume for volume), reacted for 10 min at room temperature, and then centrifuged (14,000g at 4°C) for 10 min. The clear supernatant was transferred to an Eppendorf tube, then Zn acetate (1%) 60 μL, N,N-dimethyl-p-phenylenediamine sulfate, 40 μL (20 mmol/L in 7.2 mol/L HCl), and FeCl₃, 40 μL (30 mmol/L in 1.2 mol/L HCl), were added in proper sequence and reacted for 20 min at room temperature.

For CSE activity assay, the skin tissue was homogenized in an ice-cold 100 mmol/L potassium phosphate buffer (pH 7.4) with an electric homogenizer (PowerGen 125, Fisher Scientific). The homogenate was centrifuged (10,000g at 4°C) for 10 min, and the supernatant (430 μL in PBS) was mixed with 20 μL l-cysteine (20 mmol/L), 20 μL pyridoxal-5′-phosphate (2 mmol/L), and 30 μL 0.9% sodium chloride, and reacted for 30 min in a 37°C water bath. Then, 250 μL Zn acetate (1%) and 250 μL trichloroacetic acid (10%) were added to terminate the reaction. After that, N,N-dimethyl-p-phenylenediamine sulfate, 133 μL (20 mmol/L in 7.2 mol/L HCl), and FeCl₃, 133 μL (30 mmol/L in 1.2 mol/L HCl), were added. The absorbance of the standards and samples at 670 nm was measured on a spectrophotometer (SpectraMax 190, Molecular Devices) after 20 min. H₂S concentration was calculated against a calibration curve of NaHS (3.125–200 μmol/L) (39). All samples were assayed in duplicate. Results were expressed as plasma H₂S concentration in micromole per liter. The optimal density value of skin samples obtained at 670 nm was adjusted by protein concentration and extrapolated from the standard curve obtained from the same plate. Results of CSE activity were expressed as H₂S concentration in micromoles per microgram of protein.

EPCs were isolated from the bone marrow of db/+ and db/db mice and cultured afterward according to our established methods with minor modifications (35,40). Described briefly, bone marrow mononuclear cells were isolated from the femurs and tibias of mice by density gradient centrifugation. After two washing steps, mononuclear cells were plated onto vitronectin-coated (Sigma-Aldrich) six-well plates and maintained in endothelial basal medium-2 (Lonza) supplemented with endothelial growth medium 2 (EGM-2, Lonza) with 5% FBS. Cells were cultured at 37°C with 5% CO₂ in a humidified atmosphere. EPCs after 7 days of culture were used for further analysis. To characterize EPCs isolated from bone marrow, stem cell markers (CD34, Sca-1), endothelial cell markers (Flk-1, CD144), and monocyte marker (CD11b) were examined on cultured EPCs by flow cytometry according to our established methods (35,41).

The CellTiter 96 AQ_ueous One Solution Cell Proliferation Assay kit (Promega) was used to determine the cytotoxicity of NaHS and PAG in the in vitro studies. Described briefly, 2 × 10⁴ EPCs in 100 μL media were dispensed into each well of 96-well culture plates and cultured overnight. Plates were incubated for 30 min with NaHS (100 μmol/L) or PAG (100 nmol/L). A total of 20 μL of the CellTiter 96 AQ_ueous One Solution Reagent containing tetrazolium compound (3-[4,5-dimethylthiazol-2-yl]-5-[3-carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium, MTS) and an electron coupling reagent (phenazine ethosulfate; PES) was added to each well. Two hours later, the absorbance was recorded at 490 nm using a 96-well plate reader (Molecular Devices, Sunnyvale, CA).

EPCs isolated from db/db mice were reseeded into six-well plates at 5 × 10⁵ cells/well, incubated at 37°C for 24 h, and serum starved for 8 h. Then they were treated with NaHS at concentrations of 10, 50, and 100 μmol/L in EGM-2 medium for 30 min in a 37°C incubator before function evaluation.

After 7 days in culture, EPCs from db/+ mice were reseeded into six-well plates at 2 × 10⁵ cells/well and incubated at 37°C until the cells were 60%–80% confluent. After that, EPCs were serum starved for 6 h before small interfering (si)RNA transfection. The EPC-conditioned media on CSE-siRNA (sc-142618, Santa Cruz Biotechnology) was diluted to a 10 μmol/L working solution and delivered to cells at a 5 nmol/L final concentration through a siRNA transfection reagent (sc-29528, Santa Cruz Biotechnology). After 5–7 h of transfection, the transfection mixture was removed and replaced with fresh EGM-2 and incubated for an additional 48 h. A nonrelated scrambled sequence siRNA (Control siRNA-A, sc-37007; Santa Cruz Biotechnology) was used as a transfection control.

EPCs isolated from db/+ mice and cultured for 7 days were reseeded into six-well plates at 5 × 10⁵ cells/well and incubated at 37°C for 24 h, then serum starved for 6 to 8 h and treated with DL-PAG at concentrations of 3, 10, and 30 mmol/L in medium for 30 min; the medium containing PAG was then removed and washed twice with PBS for function assay.

EPCs isolated from db/db mice were reseeded into six-well plates at 5 × 10⁵ cells/well, incubated at 37°C for 24 h, and serum starved for 8 h. Then, the EPC-conditioned media on Ang-1 siRNA (Dharmacon) was diluted and delivered to cells at a 5 nmol/L concentration through a siRNA transfection reagent (Dharmacon). After 6 to 7 h, the transfection mixture was removed and replaced with fresh EGM-2 and incubated for an additional 48 h before 30-min NaHS treatment at concentration of 50 μmol/L. Of the nonrelated scrambled sequence siRNA–treated db/db EPCs, only NaHS 50 μmol/L–treated db/db EPCs and the untreated db/+ EPCs were used as controls. EPC tube formation was assessed after this treatment.

For the tube formation assay, 48-well plates were coated with growth factor–reduced Matrigel (150 μL/well, Becton, Dickinson and Co.). EPCs (5 × 10⁴), prepared as described above, were plated in 200 μL EGM-2 medium and incubated at 37°C with 5% CO₂ for 12 h overnight to form the tube. Images of tube morphology in a random five fields per well were taken by inverted microscope (Nikon Eclipse TE2000-U) at original magnification ×4 (23). Tube lengths were drawn with Adobe Photoshop CS4 software and analyzed with Image-Pro Plus 5.1 software. For adhesion assay, EPCs were plated at 2.5 × 10⁴/well in triplicate to a vitronectin precoated 96-well plate. After 2 h of incubation at 37°C with 5% CO₂ in EGM-2, nonadherent cells were removed by washing twice with PBS. Cells were fixed with 2% paraformaldehyde (PFA) for 20 min, and the nuclei were stained with Hoechst (5 μg/mL, Sigma-Aldrich) for 20 min. The adherent cells were counted in five random fields in each well with an inverted fluorescence microscope (Nikon) at original magnification ×10, and the mean value of three wells was determined for each sample.

Capillary density was quantified by histological analysis as previously described (35). At day 8 after wounding, round-shaped 6-mm skin samples, including the healed or unhealed wound area, were fixed with 2% PFA (Sigma-Aldrich), embedded in a cross-section position in paraffin, and sectioned at 4-μm intervals. Slides were deparaffinized and hydrated, blocked with normal rabbit serum (Vector Laboratories) for 30 min, incubated for 60 min at room temperature with an anti-CD31 antibody (1:50; Santa Cruz Biotechnology), and further incubated with Vectastain Elite ABC Reagent (Vector Laboratories) for 30 min and Nova Red (Vector Laboratories) for 15 min. Slides were counterstained with Gill (Lerner) 2 Hematoxylin (VWR Scientific) for 10 s, differentiated in 1% glacial acetic acid, and rinsed in running tap water. Photographs were taken under a Nikon Eclipse TE2000-U microscope using Metamorph software. Five random microscopic fields (original magnification ×20) were counted to determine the number of capillaries per wound, which was expressed as capillaries per high-power field.

EPCs were isolated as described above and cultured in EGM-2. BrdU staining was performed on EPCs as we previously described (35). Cultured EPCs were labeled with BrdU and 5-fluro-2′-deoxyuridine (BrdU labeling reagent, Invitrogen). Then, 1 × 10⁶ EPCs were transplanted to 6-mm punch biopsy wounds in db/db mice, as described above. On day 8 of wound healing, the mouse was killed, and the wound and surrounding skin was recovered and fixed in 2% PFA for 18 h. Samples were then switched to 70% ethanol. The samples were embedded in paraffin, and 4-mm sections were used for analysis. For staining with rabbit CD31 (1:50; Santa Cruz Biotechnology) and chicken BrdU (1:500; Abcam), primary antibodies were incubated overnight at 4°C. Slides were incubated with Alexa-488 (1:500; Invitrogen Molecular Probes) and Cy3 (1:1,000; Jackson ImmunoResearch Laboratory), followed by 30 s of incubation with Hoechst nuclear stain. Positively stained cells in five random fields were imaged at original magnification ×10 on a Nikon fluorescence microscope (Olympus, Melville, NY). Imaging conditions were maintained at identical settings with original gating performed within each antibody-labeling experiment with the negative control (no primary antibody).

Eight nondiabetic and seven diabetic subjects underwent a full-skin excisional biopsy in their lower extremities. The nondiabetic subjects were recruited from patients admitted to the Division of Vascular Surgery of Shanghai Clinical Center because of traumatic wounds and did not have cardiovascular, renal, or hepatic disease. The diabetic subjects were recruited from Shanghai Clinical Center of Diabetes, and those who had foot or leg ulceration and underwent the debridement operation in our Division of Vascular Surgery were included in the study. The clinical characteristics of participating subjects are outlined in Table 1. All lower limb skin biopsies were performed during the limb operation by surgeons, and the skin beside the wound (margin within 1 cm) was taken. Protocols were approved by the institutional review boards of Shanghai Jiao Tong University and the Shanghai Clinical Center of Diabetes. Written informed consent was obtained from every subject. The skin tissues were used to examine the expressions of Ang-1, CSE, and VEGF by Western blot.

Total protein was extracted from cultured EPCs after in vitro culture and treatment, as described above, or from harvested wound skin tissue, including a 6-mm region of skin located in the wound and 10 mm of skin tissue around the wound bed, by homogenization of skin wounds in tissue lysis buffer (Sigma-Aldrich). Protein concentration was determined by a bicinchoninic acid protein assay kit (Pierce). Equal amounts of proteins (30–50 μg) were boiled and separated by 6–12% SDS-PAGE and transferred onto nitrocellulose membrane (Amersham Hybond-ECL, GE Healthcare). CSE (mouse monoclonal anti-CSE, Santa Cruz Biotechnology), CBS (mouse monoclonal anti-CBS, Santa Cruz Biotechnology), and Ang-1 (rabbit polyclonal anti–Ang-1, Abcam) were determined with its specific first antibodies. The dilutions of primary antibody were 1:500 for Ang-1, 1:10000 for β-actin (Santa Cruz Biotechnology), and 1:250 for all other antibodies. Luciferin-conjugated secondary antibody was used at 1:5000. The bands were scanned and analyzed by the Odyssey system.

Results are expressed as mean ± SEM. Differences between the two groups were compared with paired or unpaired Student t tests. One-way ANOVA with Tukey multiple comparison tests were used to compare each parameter when there were two or more independent groups. Two-way ANOVA of grouped analyses was used to compare the wound recovery rate at different time points, and then Bonferroni post hoc t tests were performed to compare replicate means by row when ANOVA was statistically significant. Values of P < 0.05 were considered statistically significant.

At 8–10 weeks of age, the body weights of the db/db mice were significantly greater than those of the db/+ mice (40.06 ± 3.37 g vs. 25.99 ± 1.22 g; P < 0.01), and the blood glucose levels of the db/db mice were significantly higher than those of the db/+ mice (475.89 ± 42.27 mg/dL vs. 139.9 ± 33.49 mg/dL; P < 0.01). The baseline plasma H₂S concentration of the db/+ mice was significantly higher than that of the db/db mice (Fig. 1A). The plasma H₂S concentrations of the db/db mice were further decreased than those of db/+ mice after wound creation and were reversed by NaHS and HTB at day 8 (Fig. 1B) and day 18 (Fig. 1C). In contrast, PAG downregulated the H₂S plasma concentrations in the db/+ mice (Fig. 1B and C). Although the baseline level of H₂S synthase CSE activity in the skin tissue of db/db mice was higher than in that of the db/+ mice (Fig. 1D), the CSE activity of the db/db mice was significantly lower compared with that of the db/+ mice after wound creation at day 8 (Fig. 1E).

The baselines of wound healing in db/db and db/+ mice were determined. The data demonstrated that the wound areas of the db/db mice closed significantly slower than those of the db/+ mice from day 2 to day 16 (Fig. 2A and B). The administration of the rapid-releasing NaHS or slow-releasing HTB H₂S donor improved wound recovery rates compared with saline-treated db/db controls (Fig. 2B and C) starting from day 2 to day 16. The overall wound-healing rate in the two H₂S donor groups was still slower than that of the db/+ group (Fig. 2B and C). In contrast, the administration of H₂S synthase CSE inhibitor, DL-PAG, significantly delayed the wound closure rate compared with the db/+ group (Fig. 2B and C). These data suggest that H₂S treatment improved wound healing in type 2 diabetic mice, whereas the inhibition of H₂S synthesis in vivo delayed the wound-healing course.

At day 8 after treatments, capillary numbers in the diabetic group were significantly lower than those in the normal control group (Fig. 3A and B). The slow-releasing H₂S donor HTB significantly increased the newly formed capillary densities compared with that of diabetic controls (Fig. 3A and B), whereas the vessel densities in skin tissue treated by NaHS augmented the newly formed capillary densities to some extent but without reaching statistical significance. On the contrary, new vessel densities were lower in wound skin taken from mice treated with PAG than from wound skin taken from normal db/+ controls (Fig. 3A and B). The difference in the baselines of capillary numbers of unwounded skin between db/+ mice and db/db mice was not significant (Fig. 3C).

Wound healing improved significantly in the diabetic mice that received EPC transplantation. Furthermore, NaHS significantly augmented the efficiency of EPC cell therapy in diabetic mice (P < 0.05 vs. db/db mice; Fig. 4A and B). These data suggest that H₂S significantly accelerated EPC-mediated wound healing in diabetic mice. On day 8 of wound beds, some BrdU-labeled EPCs integrated into vascular-like structures (CD31-positive) and the other BrdU-labeled EPCs integrated in the dermis. On day 16 of wound beds, more BrdU-labeled cells integrated into the epithelial layers (Fig. 4C).

To investigate the functions of EPCs, EPC tube formation on Matrigel and EPC adhesion in vitronectin precoated plates were observed in EPCs isolated from mice after 16 days of treatment with H₂S or PAG. The cell population profiles of cultured EPCs (day 7) were as follows: Sca-1 (60.8 ± 7.52%), CD34 (47.6 ± 4.37%), Flk-1 (22.6 ± 3.17%), CD144 (8.7 ± 1.19%), and CD11b (37.5 ± 2.54%), indicating that 7-day cultured EPCs are heterogeneous populations containing both higher percentages of stem cells and endothelial cells. The db/db EPCs formed significantly fewer tube networks than the db/+ EPCs (Fig. 5A). In vivo H₂S therapy with NaHS and HTB significantly improved the ability of diabetic EPCs to form tube networks compared with db/db control EPCs (Fig. 5A). On the contrary, PAG treatment significantly reduced the tube length compared with normal controls (Fig. 5A). Similar to tube formation, the adhesive numbers of EPCs in the db/db group was markedly less than that in the normal group (Fig. 4B), and administration of NaHS and HTB significantly increased the adhesion function of EPCs compared with untreated db/db EPCs (Fig. 5B). Contrarily, EPC adhesion in the PAG group was significantly attenuated compared with that of the db/+ control group (Fig. 4B).

To observe the influence of H₂S on the functions of diabetic EPCs in vitro, EPCs isolated from the bone marrow of db/db mice were treated with NaHS at concentrations of 10, 50, and 100 μmol/L in EMG-2 for 30 min before tube formation and adhesion determination. NaHS significantly augmented the tube networks of diabetic EPCs at all three concentrations compared with nonmodified db/db EPCs (Fig. 5C). The deficient adhesion of diabetic EPCs was also reversed by NaHS with concentrations of above 10 μmol/L in vitro (Fig. 5D). These aforementioned data suggest that in vivo and in vitro H₂S donor treatment both restore deficient EPC tube formation and adhesive functions in db/db mice.

To verify whether H₂S inhibition could impair EPC function, H₂S synthesis of EPCs was inhibited with different concentrations of DL-PAG (3, 10, and 30 mmol/L), which was added to EGM-2 medium for 30 min. The results indicated that the tube length of EPCs treated with PAG 10 mmol/L and 30 mmol/L decreased significantly compared with that of normal controls (Fig. 5E). Similarly, the adhesive EPC numbers were reduced markedly in the PAG 10 mmol/L and 30 mmol/L intervention groups compared with control EPCs (Fig. 5F). These results suggest that the inhibition of H₂S synthesis in EPCs reduces their functions.

The tube formation results from the CSE silencing experiment in EPCs further demonstrated the critical effect of H₂S on EPC angiogenesis. The CSE expression of EPCs after siCSE treatment was reduced by nearly 70%, as shown by Western blotting (Fig. 5G), whereas the expression of CBS was not influenced by the silencing of CSE in EPCs (Fig. 5H). The average tube length of EPCs after CSE knockdown decreased to ∼40% compared with controls (Fig. 5I), and EPC adhesion also was markedly attenuated compared with controls (Fig. 5J). These data identified that H₂S is critical for EPC functions, including tube formation and adhesion abilities. In addition, the cytotoxicity assay showed that treatment with NaHS or PAG did not affect the viability of EPCs (Fig. 5K).

To understand the possible mechanism through which H₂S improves EPC function, the expressions of Ang-1 and VEGF in wound skin tissue and EPCs were measured by Western blotting. The results indicated that the in vivo H₂S donor treatment significantly increased the expressions of Ang-1 and VEGF not only in EPCs from these mice but also in wound skin tissues (Fig. 6A–C). On the contrary, the CSE inhibitor PAG inhibited the expression of Ang-1 in the wound skin tissue (Fig. 6A) and in EPCs (Fig. 6B). Similarly, the silencing of CSE in cultured normal EPCs with CSE siRNA in vitro treatment significantly decreased their Ang-1 expression (Fig. 6D).

To identify whether H₂S stimulates the angiogenic function of EPCs through Ang-1 signaling, EPCs were pretreated with Ang-1 siRNA before in vitro NaHS treatment. EPCs were isolated from the bone marrow of db/db and normal mice and cultured in EGM-2 for 7 days. Diabetic EPCs (5 × 10⁵) were then reseeded and treated in vitro with NaHS at 50 μmol/L only in EGM-2 for 30 min or with Ang-1 siRNA at 5 nmol/L in medium for 6–7 h before NaHS treatment. The results revealed that the improved tube length with NaHS was significantly inhibited by prior Ang-1 siRNA treatment and was significantly lower than that of normal EPCs (all P < 0.05; Fig. 7A and B). These results suggest that H₂S exerts its role of promoting EPC angiogenesis, at least partially, by restoration of Ang-1.

To investigate whether the dysregulations of Ang-1, CSE, and VEGF also applied to the skin tissue from foot ulcers of diabetic subjects, we compared their expression between nondiabetic and type 2 diabetic subjects. The data showed that the expressions levels of Ang-1, CSE, and VEGF were 36%, 41%, and 46% lower, respectively, in the diabetic subjects compared with the nondiabetic subjects (Fig. 8).

H₂S is the “third” gasotransmitter next to NO and carbon monoxide (CO). Its biological effects have not yet been completely understood. The current study demonstrates that 1) plasma H₂S levels and CSE activities in skin wounds decreased in type 2 diabetic mice; 2) in vivo and in vitro H₂S treatment both amended the tube formation and adhesive functions of diabetic EPCs; 3) H₂S donor therapy improved EPC-mediated wound healing in diabetic mice via elevating angiogenesis; 4) H₂S treatment increased the expression of Ang-1 and VEGF in EPCs from diabetic mice and Ang-1 siRNA inhibited the ameliorative effect of NaHS on EPC angiogenic functions; and 5) the expressions of Ang-1, CSE, and VEGF were decreased in diabetic ulcers in diabetic patients.

The baselines of plasma H₂S in diabetic mice were significantly lower than those of the nondiabetic mice, which is consistent with the plasma H₂S measurements in NOD mice (19) and streptozotocin-induced type 1 diabetic mice (20). On day 8 or day 16 after skin wounds, the plasma H₂S levels were further decreased in the diabetic mice, whereas the plasma H₂S levels in the nondiabetic mice were maintained at baseline level. Interestingly, the baseline activity of CSE in skin tissues was ∼1.3-fold higher in the diabetic mice than in the nondiabetic mice, which might be a compensative response to decreased plasma H₂S levels. However, on day 8 after wounds, CSE activity in skin wounds of the diabetic mice was significantly attenuated and failed to produce enough H₂S for wound healing, whereas CSE activity in the nondiabetic skin wounds was higher than at baseline, which might be a mechanism for the unchanged H₂S levels. Because the expression of another H₂S synthase, CBS, was not changed in skin tissue and in the EPCs of diabetic mice, the delayed wound healing in diabetic mice is more likely due to the deficiency of H₂S synthesis in the vascular system, just as Brancaleone et al. (19) reported in NOD mice.

H₂S had been found to improve the burn skin wound and gastric ulcers induced by acetate acid in wild-type mice or rats (18–20) and to improve endothelial wound healing in an in vitro model (23). Wallace et al. (21) revealed an enhancing effect of H₂S in acetic acid–induced gastric ulcers in a Wistar rat model and in a skin burn wound model in CSE-knockout mice (22). This evidence suggests that H₂S contributes to wound healing. However, no evidence of its efficacy on diabetic wounds has been reported so far.

In the current study, two H₂S donors, NaHS and 4-HTB, were administrated in vivo; the first is the widely used rapid-releasing donor and the latter is the slow-releasing H₂S donor. The usual dosage NaHS range is 10–100 μmol/kg, as reported in recent in vivo studies (22,23,32). On the basis of this evidence and considering that the physiological H₂S levels in the plasma of mammals were 10–100 μmol/L (42), we chose 50 μmol/kg as the treatment dosage of NaHS in the current study. As for 4-HTB, only one experience, reported by Wallace et al. (21), could be referenced. They used HTB 10–30 μmol/kg twice daily by intragastric administration in male Wistar rats (21), so we administered the dose of 30 μmol/kg to diabetic mice and proved its efficacy on diabetic wound healing. The results of this study demonstrate that an intraperitoneal injection of the H₂S donors NaHS and 4-HTB both improved the recovery rate of wound healing in type 2 diabetic mice and had similar effects on wound healing in diabetic mice. Because NaHS and 4-HTB are easily available H₂S donors, they may be potential remedies to ameliorate foot ulcerations in diabetic patients.

Insufficient angiogenesis is one of most important reasons for delayed diabetic wound healing. The prominent functions of EPCs in angiogenesis have been widely demonstrated (35,43,44). In the current study, the tube formation and adhesive function of diabetic EPCs was impaired compared with normal mice with a matched genetic background, suggesting that their angiogenic effect was attenuated. The proangiogenic effect of H₂S has been recently recognized. Cai et al. (23) reported a novel Akt-mediated proangiogenic role of H₂S in C57BL/6 mice. Topical administration of H₂S enhanced wound healing in CSE knockout mice, and the researchers concluded that endogenous H₂S stimulates endothelial cell–related angiogenic properties through a ATP-sensitive K⁺ channel/mitogen-activated protein kinase pathway (22). H₂S also enhances tube formation and migration in cultured vascular endothelial cells (32). The current study demonstrated that H₂S donors facilitated diabetic wound healing by stimulating capillary formation at the site of injury. Furthermore, we demonstrated that the H₂S donor accelerated EPC-mediated wound healing in diabetic mice. To date, whether H₂S ameliorates diabetic wound healing by influencing EPC functions has not been elucidated.

In the current study, we first investigated the angiogenic effects of H₂S donors on EPCs isolated from diabetic mice in vitro. Our results revealed that the tube formation and adhesion of EPCs increased significantly by chronic treatments with both the rapid-releasing and slow-releasing H₂S donor in diabetic mice. Furthermore, the angiogenic function of diabetic EPCs also was restored by NaHS in vitro. These results support that H₂S improves the angiogenic function of diabetic EPCs. On the contrary, after the normal mice and their EPCs were treated with the H₂S synthetase (CSE) inhibitor DL-PAG in vivo and in vitro, or siCSE in vitro, the tube network formation and adhesive ability of normal EPCs declined markedly at the same time the wound closure rate was delayed. We have demonstrated that topical EPCs transplantation significantly improves wound healing in the diabetic mice (35,45). In the current study, we found that NaHS significantly augmented the efficiency of EPC cell therapy in diabetic mice. Consistent with our previous studies, some EPCs integrated into vascular-like structures and the other EPCs integrated in the dermis at the early stage of wound beds (on day 8), indicating that H₂S accelerates wound healing by EPC-mediated angiogenesis. Interestingly, we also found that more cells integrated into the epithelial layers at the later stage of wound beds (on day 16), which may be investigated in future studies. In vitro and in vivo evidence both suggests that H₂S is vital for preserving normal EPC functions and wound repair in type 2 diabetic mice.

Angiogenic signals are mainly activated by endothelium-specific receptor tyrosine kinase, of which VEGF, fibroblast growth factor, and Ang-1 are ligands. Ang-1 promotes endothelial differentiation from embryonic stem cells and induces pluripotent stem cells (46). Ang-1 gene expression was downregulated in EPCs cultured in a high glucose environment, and recombinant Ang-1 increased the vessel-forming capacity of EPCs in vitro (47). The current study showed that in vivo treatment with H₂S donors significantly increased the expressions of Ang-1 and VEGF in EPCs from diabetic mice. Most importantly, we also found attenuated expression of Ang-1 and VEGF in diabetic wounds in patients. Although the current study cannot exclude the effects of sex, medications, and comorbidities on the expressions of Ang-1, CSE, and VEGF, our results provided evidence that angiogenic factors and H₂S were significantly attenuated in local diabetic wounds, which may contribute to refractory wounds in diabetic patients. Ang-1 was reported to potentiate VEGF-induced angiogenesis (48). Ang-1 signaling was known to regulate vascular quiescence and angiogenesis through its cognate receptor Tie-2 (49). However, we did not observe altered expression of the Ang-1 receptor Tie-2 between diabetic EPC or diabetic wounds. In parallel, the tube formation and adhesion ability of H₂S-amended EPCs increased significantly, which was blocked by pretreatment by Ang-1 siRNA before H₂S. These results suggest that the restoration of EPC angiogenic functions by H₂S may occur, at least partly, through the mediation of Ang-1. H₂S activates phosphatidylinositol-3-kinase and mitogen-activated protein kinases to regulate angiogenic factors in endothelial cells (22,50,51). However, the mechanisms by which H₂S regulates Ang-1 in EPCs are still unknown. Further research should be done to clarify how and to what degree H₂S improves diabetic wound healing and EPC function via Ang-1.

In summary, the current study demonstrates for the first time that H₂S is vital in maintaining normal EPC angiogenic function and that an H₂S donor can accelerate wound healing in type 2 diabetic mice. These effects of H₂S are related to restoring the angiogenic functions of EPCs by upregulating Ang-1 expression. The findings that an H₂S donor can amend EPC function and improve skin wound recovery may lead to novel therapeutic strategies for diabetic vascular complications and diabetic skin ulcerations.

Acknowledgments. The authors thank Drs. Patricia Loughran, Jie-Mei Wang, and Mingwei Zhang for their technical support.

Funding. This work was supported in part by the National Basic Research Program (973 Program) 2014CB542400 and the National Science Foundation of China (81130004, 81370359, and 913392019 to A.F.C.; 81070650 and 81270397 to F.L.; 30900519 and 81370429 to D.-D.C.). Support was also received from Shanghai Jiao Tong University Affiliated Sixth People’s Hospital Fellowship to F.L.

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

Author Contributions. F.L. and D.-D.C. designed the experiments, conducted the in vitro cell functional tests and the in vivo cell therapy, and wrote the manuscript. X.S. and H.-H.X. performed the Western blot on patient skin samples. H.Y. conducted the in vitro cell biochemical tests and contributed to discussion. W.J. and A.F.C. designed the study, contributed to discussion, and reviewed and edited the manuscript. F.L., D.-D.C., and A.F.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.