Major Improvement in Wound Healing Through Pharmacologic Mobilization of Stem Cells in Severely Diabetic Rats

Foot ulceration is the most frequently recognized complication of diabetes, affecting 1.0–3.5 million in the U.S. alone (1). More than one-half of diabetic ulcers become infected (2) and ∼20% of moderate or severe diabetic foot infections lead to some level of amputation. The management of diabetic foot ulcers (DFUs) poses an increasing socioeconomic burden, with an estimated cost of $60 billion annually in the U.S. (3,4). The direct costs of treating diabetic foot complications exceed the treatment costs for many common cancers (5,6). In a 15-year study, the rate of DFUs in type 1 diabetes (T1DM) was 6.23%, which is similar to the rate of DFUs in type 2 diabetes (T2DM) (7). Approximately 23% of patients with T2DM will eventually have a foot problem (8). With appropriate therapy, ∼77% of DFUs heal within 1 year (9), but almost 60% recur within 3 years.

DFUs and their recurrence result partly from peripheral neuropathy that causes callus formation in the foot and subcutaneous hemorrhage, which ultimately lead to skin breakdown (10). Peripheral artery disease, which coexists in approximately one-half of patients with DFU, also contributes to the development of foot ulcers and their recurrence (11). Topical therapies have been unsatisfactory. Currently, there is only one U.S. Food and Drug Administration–approved drug, becaplermin, a gel made of a recombinant human platelet–derived growth factor (Regranex), available for treating patients with DFU (12), and a postmarketing phase IV trial failed to demonstrate becaplermin gel treatment efficacy (13). Cell-based therapeutic strategies, such as mesenchymal stem cells isolated from the bone marrow, peripheral blood, or adipose tissue (14–17), have been cited as a promising approach, but none of these studies showed significant reduction of healing time. Concern about the quantity and quality of stem cells for topical or injection treatment is an important limitation to cell therapy (14–17). Despite therapeutic promise, the presumed mechanism of action of stem cell populations often remains insufficiently validated.

The most fundamental limitation in all remedies investigated to date is their reliance on topical application. We propose a new systemic treatment paradigm for diabetic wound healing that exploits our serendipitous discovery of an intravascular stem cell mobilizing therapy. Using AF, a combination of two drugs (AMD3100 and low-dose FK506), we demonstrated long-term liver (18) and kidney (19,20) allograft survival with limited short-term treatment and freedom from further immunosuppression in otherwise strongly rejecting rat or pig strain combinations. Tolerance from this systemic stem cell mechanism was associated with allograft chimerism (host repopulation) and local downregulation of the immune response (18–21). Unexpectedly, in addition, an important new finding in those experiments was that AF improved wound healing. Validation experiments confirmed that AF produced significantly faster skin wound healing with less scar formation and resulted in regeneration of hair follicles in a rodent model of surgical excisional wounds (22). These benefits occurred because of the power of the drug to recruit mobilized multipotential bone marrow–derived stem cells into the injured sites. We hypothesized that the same systemic treatment with the AF combination would improve diabetic wound healing through increasing the number of circulating bone marrow–derived stem cells and recruiting them into the wound site where they would accelerate healing through the same mechanism demonstrated in those nondiabetic settings.

Diabetic wound healing studies to date have usually tested only the impact of hyperglycemia in early diabetic animals. Another limitation of most topical remedies tested is a lack of impact on impaired vasculature that contributes to impaired healing. To better recapitulate the clinical hallmarks of DFUs and to more generally test the therapeutic potential, both T1DM (from streptozocin [STZ]) and T2DM Goto-Kakizaki (GK) rats with severe diabetic changes, evidenced by neuropathy and peripheral vascular disease (PVD), were used in this wound healing study. We found that pharmacological mobilization and recruitment of bone marrow–derived stem cells with the AF combination not only normalizes wound healing of the integument in diabetes but also improves diabetic angiopathy of the injured hind limbs.

Sprague-Dawley rats were purchased from Charles River (Kingston, NY), and Lewis rats were purchased from Hilltop Lab Animals, Inc. (Scottdale, PA). The GFP transgenic Lewis rat was obtained from the National Institutes of Health–funded Rat Resource and Research Center at the University of Missouri (Columbia, MO). These rats were used at the age of 8–10 weeks. GK Wistar rats were obtained from Taconic (Hudson, NY) and used at the age of 20 weeks. Both male and female rats were included in each group for this study.

STZ powder (Sigma-Aldrich, St. Louis, MO) was dissolved in a cold citrate buffer to reach a pH of 4.2 just before injection. Sprague-Dawley rats were fasted overnight and then received intraperitoneal STZ injection (45 mg/kg body weight). Rats with blood glucose levels >400 mg/dL for 6 weeks were considered to have had T1DM successfully induced and were included in the study.

Four full-thickness wounds were created on the dorsal skin of rats with a sterile disposable 5-mm (T1DM) or 8-mm (T2DM) diameter circular biopsy punch (Integra LifeSciences, Plainsboro Township, NJ) according to a method previously described (22). Wounded animals were randomly divided into two groups and received a combination of AMD3100 (1 mg/kg) and FK506 (0.1 mg/kg) subcutaneous injection (AF) or the same volume of saline for controls immediately after surgery and every other day until the wounds were completely healed. AMD3100 (plerixafor injection) and FK506 (tacrolimus) were diluted in 0.9% saline just before injection. Drugs (0.1 mL/100 g body weight) were injected into the loose skin of the neck, where it is far away from the wound area, since this is not a local treatment of the wound.

A modified size of 4.0 × 5.0 mm excisional wound was created on the dorsal surface of both feet in T1DM rats according to a method described before (23). Animals were randomly assigned into AF treatment or saline control group as described above.

Each wound site was photographed digitally (Sony DSC-RX100 and Canon EOS Rebel T5 with Canon EF-S 60mm f/2.8 AF USM Lens) at the indicated time intervals. Wound areas were calculated using Adobe Photoshop software as we previously described (22). All treatments and wound evaluations were double-blinded.

Skin wounds were surgically excised and fixed, and histologic sections were cut and stained as usual.

Cut sections of 4 μm were prepared from paraffin-fixed paraffin-embedded tissues for immunohistochemistry or frozen tissue for immunofluorescent staining. Immunohistochemistry stains were performed with the avidin-biotin peroxidase complex method, using VECTASTAIN ABC Kits (Vector, Burlingame, CA). Sections were counterstained with hematoxylin for 20 s. Double staining was performed by immunofluorescent stains using frozen sections. Cell nuclei were stained blue with DAPI (Life Technologies). The following antibodies were used: anti-CD34 (1:100; R&D Systems, Minneapolis, MN), anti-CD133 (1:100; Abcam, Cambridge, MA), anti-RECA-1 (1:500; Abcam), anti-SDF-1α (1:100; Abcam), anti-CXCR4 (1:100; Abcam), anti-CD31 (1:100; Abcam), anti-Ym1/2 (1:1,000; Abcam), biotin-conjugated bovine anti-goat (Jackson ImmunoResearch, West Grove, PA), biotin-conjugated goat anti-mouse (Cell Signaling Technology [CST], Danvers, MA), biotin-conjugated goat anti-rabbit (CST), Cy3-conjugated donkey anti-rabbit IgG (1:200; Jackson ImmunoResearch), Cy3-conjugated donkey anti-mouse (1:200; Jackson ImmunoResearch), Alexa Fluor 594–conjugated anti-rabbit/mouse (1:600; CST), and phycoerythrin (PE)-conjugated Ym1/2 antibody (1:1,000; Abcam).

To evaluate the nociceptive intraepidermal nerve fibers in diabetic rats, skin samples were collected by using a biopsy punch (3 mm in diameter). Sections of 50-μm thickness were cut and kept in an antifreeze buffer in −20°C before staining. Immunohistochemistry staining for PGP9.5 (1:2,000; DAKO) was performed using ABC kits (Vector) according to the method described before (24). Intraepidermal nerve fiber density was counted in a blinded fashion.

Granulation tissue collected on day 7 was homogenized on ice. Total protein was extracted with radioimmunoprecipitation assay buffer (R0278; Sigma) with protease inhibitor cocktail (#5871; CST). Protein samples were adjusted to the same amount (30 μg) using BCA Protein Assay Kit (#7780; CST). Proteins were separated by NuPAGE Novex 4–12% Bis-Tris Midi Protein Gels (Thermo Fisher Scientific). Proteins were then transferred to immunoblot polyvinylidene fluoride membranes (Bio-Rad). Five percent nonfat milk was used to block unspecific binding at room temperature for 1 h, followed by primary antibodies at 4°C overnight, including anti–vascular endothelial growth factor (VEGF)-A (1:200; Abcam) and anti–hepatocyte growth factor (HGF)-α (1:500; Santa Cruz). Fluorescent secondary antibodies from IRDye (1:10,000; Li-Cor, Lincoln, NE) were used as secondary antibodies. Membranes were scanned and analyzed by Odyssey CLx imaging system (Li-Cor).

White blood cells (WBCs) were isolated from peripheral blood using red blood cell lysis buffer (Sigma-Aldrich). A single-cell suspension of WBCs (1 × 10⁶ cells) was blocked with donkey and goat serum (Sigma-Aldrich) for 30 min, followed by incubation with PE-conjugated anti-CD34 (Novus Biologicals, Littleton, CO), rabbit anti-CD133 (Novus Biologicals), or mouse anti-CD31 (Abcam) at 4°C for 60 min. Cells were then incubated with FITC-conjugated donkey anti-rabbit for CD133 or PE-conjugated goat anti-mouse antibodies for CD31 at room temperature for 30 min. CD34⁺, CD133⁺, and CD31⁺CD133⁺ cells were analyzed by flow cytometry using CELLQuest software (Becton Dickinson, Franklin Lakes, NJ). Thresholds for identifying positively staining cells were set with relevant isotype control antibodies. After acquisition, data were analyzed by using FlowJo 10.0.7 software (Tree Star Inc., Ashland, OR).

Eight-week-old wild-type Lewis rats were lethally irradiated with a single dose of 12 Gy using Gammacell 40 (Nordion, Ottawa, Ontario, Canada). Bone marrow cells (3–4 × 10⁷) recovered from GFP Lewis rats were injected intravenously through the femoral vein into wild-type Lewis rats at 24 h after total body irradiation as we described before (25). Recipient rats were used for induction of diabetes at least 10 weeks following bone marrow transplantation (BMT).

Data obtained from multiple experiments were reported as the mean ± SEM. Independent Student t test with two tails was performed to analyze the scientific difference between two groups. A P < 0.05 was considered a significant difference. Sex was considered a factor in the statistical analysis of the data; no difference was found between males and females in each group.

All animal studies were reviewed and approved by the Johns Hopkins University institutional animal care and use committee. Rats were housed and cared for according to the Johns Hopkins animal care and use committee animal welfare assurance standards and guidelines.

All data generated during the current study are available from the corresponding author upon reasonable request. No applicable resources were generated during the current study.

The T1DM rat model was induced by intraperitoneal injection of STZ (45 mg/kg). All animals developed hyperglycemia by 1 week after STZ injection, and blood glucose remained at higher levels during the study (Supplementary Fig. 1A). At 6 weeks following development of diabetes, heat hypersensitivity was measured to detect peripheral neuropathy. Increased duration time was observed for the hot plate test (50°C) in STZ-induced diabetic rats when compared with age-matched control rats (66.2 ± 14.3 vs. 41.0 ± 5.5 s, P = 0.006, n = 5 per group), indicating the development of neuropathy (Supplementary Fig. 1C). These diabetic rats were then used for wound healing studies.

Four full-thickness wounds were generated by 5-mm diameter circular excisions on the shaved backs of T1DM rats (Fig. 1A) or age-matched nondiabetic rats as control. They were measured and followed closely. The wounded T1DM rats were divided randomly into two experimental groups and received subcutaneous injections of control saline or AF combination (AMD3100 1 mg/kg, FK506 0.1 mg/kg) immediately after wounding and every other day until complete healing. Complete healing was defined as the entire wound area being reepithelialized and becoming a pale pink color (Fig. 1A).

Digital images showed that treatment with AF therapy had significant effects accelerating granulation formation and reepithelialization (Fig. 1B). Wounds reached complete healing on day 18 after surgery in age-matched nondiabetic animals, which is consistent with the known healing kinetics in this established model (22). The diabetic animals exhibited significantly slower healing compared with nondiabetic control animals, and wounds reached complete healing on day 27 after surgery. In contrast, the complete healing time was reduced to nearly normal at 19 days in the animals with AF combination therapy (Fig. 1C and D). Scars in the wound site were significantly smaller at 3 months after surgery in diabetic animals treated with AF combination compared with saline controls (Fig. 1E and F). Remarkably, hair follicles appeared histologically in tissue sections of healed wounds from animals with AF combination therapy but not in controls (Fig. 1G and H).

On the basis of these findings, we further tested AF combination therapy in a more clinically relevant model of full-thickness excision wounds on the footpad of T1DM rats. Using a scalpel, a 4 × 5 mm rectangular full-thickness wound was created in the skin of the footpad on both hind legs of each rat (Fig. 2A). Digital images showed that treatment with AF therapy significantly reduced the size of the skin defect as soon as day 6 and accelerated reepithelialization and wound healing (Fig. 2B and Supplementary Fig. 2). For example, at 20 days post-injury, no wound was healed in diabetic rats treated with saline, while 71% of footpad wounds were healed in T1DM rats treated with AF combination. Notably, only 33% of footpad wounds were healed in age-matched nondiabetic rats at the same time (Fig. 2C). Wounds on the footpad of nondiabetic rats reached complete healing on day 21 after surgery, while healing time was delayed by 4 days in T1DM rats. AF combination therapy restored the healing time in T1DM rats, with complete healing times reduced from 25 to 20 days (Fig. 2D). Histologically, there were few hair follicles, and collagen was abundant and disorganized in tissue sections of healed wound sites from saline-treated animals at 2 months after surgery (Fig. 2E and F). In contrast, the number of hair follicles in tissue sections of healed wound sites was significantly higher in the AF-treated animals compared with the saline control group. Thus, the AF combination therapy improved T1DM wound healing by promoting both reepithelialization and differentiation of skin components.

In addition to hyperglycemia, 5-month-old GK rats developed diabetic neuropathy and peripheral artery disease. PGP9.5 staining for intraepidermal nociceptive nerve fibers showed a reduction of intraepidermal nerve fiber density in skin biopsies from T2DM GK rats (1.32 ± 1.51 vs. 7.26 ± 3.74/mm, n = 15 per group) (Supplementary Fig. 3A and B). Periodic acid Schiff (PAS) staining demonstrated the diffuse intimal thickening of skin small arteries in T2DM GK rats (Supplementary Fig. 3C). Further, as in T1DM rats, increased duration time was observed for the hot plate test (50°C) in diabetic GK rats compared with age-matched control rats (100 ± 25.3 s [n = 10] vs. 41.0 ± 5.5 s [n = 5], P = 0.00021) (Supplementary Fig. 3D). These results indicate that 5-month-old GK rats developed diabetic complications similar to patients with T2DM.

Four full-thickness wounds were created by 8-mm diameter circular excisions on the shaved back of 5-month-old GK rats (Fig. 3A) or age- and strain-matched nondiabetic rats as control. Wounded GK rats received subcutaneous injections of saline or AF combination (AMD3100 1 mg/kg, FK506 0.1 mg/kg) immediately after wounding and every other day until complete healing. All wound evaluations were double-blinded.

Similar to T1DM rats, digital images showed that treatment with AF therapy had produced a significant reduction in the size of the skin defect as early as day 12 and accelerated reepithelialization (Fig. 3B). Wounds reached complete healing on day 19 after surgery in aged-matched nondiabetic Wistar rats. Healing time was delayed by 7 days in 5-month-old T2DM GK rats. Treatment with AF combination therapy minimized the delay of wound healing in these T2DM GK rats so that wounds healed on day 21 after surgery (Fig. 3C and D) with no significant difference in healing time between nondiabetic rats and the diabetic GK rats treated with AF combination. Again, a reduction in scar tissue was also observed in GK rats treated with AF combination (Fig. 3E and F).

Peripheral blood monocytes were elevated fourfold in the diabetic rats at 3 h after subcutaneous injection of the AF combination (Fig. 4A). Flow cytometry analysis of blood samples showed that CD34⁺ stem cells and CD31⁺CD133⁺ endothelial progenitor cells (EPCs) were significantly increased in the diabetic rats receiving AF combination treatment (Fig. 4B and Supplementary Fig. 4). Immunohistochemistry staining of sections of newly formed granulation tissues of the wounds at 7 days post-injury showed that the number of SDF-1⁺ cells was dramatically increased by AF combination treatment, as were CD133⁺ and CD34⁺ cells (Fig. 4C and D). Immunofluorescence double staining showed that a large number of ED1⁺ (CD68, a macrophage marker) macrophages had appeared in newly formed granulation tissues of the wounds, but few of these ED1⁺ macrophages stained with Ym1/2 (a marker of M2 macrophages) in saline-treated diabetic rats, while abundant Ym1/2 and ED1 double-positive macrophages were present in newly formed granulation tissues of the wounds in the diabetic rats receiving AF combination at 7 days post-injury (Fig. 5A and B and Supplementary Fig. 5A). Interestingly, at this 7-day time point, a majority of Ym1/2⁺ cells costained with SDF-1, an attractant of stem cells (Fig. 5C and Supplementary Fig. 5B).

Western blot analysis of the granulation tissues showed that protein expression of the angiogenic cytokine VEGF-A and an important stem cell homing factor and wound healing mediator, HGF-α, were both significantly increased in the diabetic rats treated with AF combination compared with saline controls on day 7 post-injury (Fig. 6A–C). Immunohistochemistry staining demonstrated that the number of RECA (an endothelial marker)–positive endothelial cells and α-smooth muscle actin (α-SMA)–positive myofibroblast cells dramatically increased in the granulation tissue recovered from the AF combination–treated rats compared with the saline-treated controls (Fig. 6D and E). These RECA-positive endothelial cells or α-SMA–positive cells formed vasculature-like structures in the granulation tissues at 7 days, indicating angiogenesis.

BMT studies using syngeneic GFP transgenic Lewis rats were performed to confirm the critical role of mobilized bone marrow–derived stem cells in improving diabetic wound healing by AF combination therapy. In this model, flow cytometry analysis demonstrated that all bone marrow–derived cells were GFP positive at 2 months following BMT (25). Three months later, recipients were injected with STZ to induce T1DM. At 6 weeks after STZ injection, four full-thickness wounds were created by 5-mm diameter circular excisions on the shaved back of each BMT rat (Fig. 7A) and evaluated for stem cells during healing.

Both CD133⁺ and GFP-positive cells were increased in the granulation tissues at 7 days after wounding (Fig. 7B). The numbers of CD133⁺ and GFP-positive cells were significantly higher in the granulation tissues recovered from AF combination–treated rats compared with the saline controls (Supplementary Fig. 6). Several clusters of GFP-positive cells appeared in the granulation tissues recovered from rats treated with AF combination, and most CD133⁺ cells localized in the clusters of GFP-positive cells and costained with GFP (Fig. 7B bottom panels). These results imply that pharmacologically mobilized bone marrow–derived stem cells and their progeny are principal contributors to skin wound healing in diabetic rats.

Diabetic PVD increases the risk of nonhealing foot ulcers, amputation, and mortality (8,9). Decreased limb skin temperature as a result of the decrease of blood flow is one of the symptoms of diabetic PVD. To determine whether AF combination therapy improves peripheral microcirculation, we measured hind limb skin temperature using a FLIR E8 Thermal Camera at 30 days after skin wounding (Fig. 8A). The average skin temperature from the ankle joint to the toes was determined as foot skin temperature. Foot skin temperature in age-matched nondiabetic Wister rats was ∼34°C, while the foot skin temperature was reduced to 32°C in diabetic GK rats. Decreased limb skin temperature indicates the development of diabetic PVD. Interestingly, AF combination therapy restored the foot skin temperature to the normal 34°C in diabetic GK rats (Fig. 8B). To further confirm whether AF combination therapy improves diabetic angiopathy, PAS staining of the foot ulcers was performed (Fig. 8C). No PAS-positive interphalangeal blood vessels were observed in tissue sections recovered from age-matched nondiabetic Wistar rats. In contrast, most interphalangeal blood vessels stained with purple-magenta color (PAS positive) in diabetic GK rats treated with saline, and some blood vessels were thicker compared with the age-matched nondiabetic rats. In AF combination–treated diabetic GK rats, few PAS-positive interphalangeal blood vessels appeared in tissue sections. In addition to normalization of wound healing, these results suggest that AF combination therapy ameliorates diabetic angiopathy of the hind limbs.

These experiments confirm our serendipitous finding that pharmacologic stem cell mobilization and homing with AF has multiple healing capacities that include improved healing in normal mouse skin (22) and extend that now to show the power of AF treatment to promote skin healing in diabetes. Moreover, we report that beyond its power to heal the integument, AF treatment produces resolution of peripheral angiopathy. These studies were in animals with advanced diabetes that allowed us to detect those benefits.

AF combination therapy consists of AMD3100 (plerixafor) and low-dose FK506 (tacrolimus) subcutaneously injected every other day. AMD3100 is an antagonist of CXCR4 and has been used clinically to mobilize hematopoietic stem cells for BMT. FK506 is a calcineurin inhibitor and a potent immunosuppressive drug when used in standard clinical doses. The dosage of FK506 we used in the AF combination was quite low (0.1 mg/kg), which is about one-tenth of the effective dosage for immunosuppression in rats, thus unlikely to increase the risk of infections. Previous studies demonstrated the synergistic effects of AMD3100 and low-dose FK506 (not in conventional dose) in mobilizing and recruiting bone marrow–derived stem cells into injured tissues (18–22). AMD3100 blocks the interaction of SDF-1/CXCR4, which releases stem cells from the bone marrow into the circulation, while low-dose FK506 increases SDF-1–producing macrophages in the wound sites (22) through activation of bone morphogenetic protein (BMP) signaling (26), which attracts mobilized stem cells into the injured sites. AF combination therapy induces liver and kidney allograft chimerism (18–20), accelerates wound healing (22), and promotes liver regeneration (27). On the basis of these findings, we focused our attention on improvement of diabetic wound healing by mobilizing bone marrow–derived stem cells with AF combination. In this study, we show that the healing of dorsal skin wounds was delayed by 7–8 days in both T1DM and T2DM rats with neuropathy and angiopathy compared with age-matched nondiabetic rats. Systemic therapy with the two-drug combination normalizes the impaired back and foot pad wound healing so that the complete healing time was the same as in age-matched nondiabetic rats. The restored healing is associated with recruitment of bone marrow–derived stem cells into the wound sites, providing improvement of angiogenesis and amelioration of diabetic PVD. While a large number of published approaches in rodent models describe somewhat improved diabetic wound healing or normalized healing in 8-week-old genetically engineered diabetic mice (28), a therapeutic strategy that normalizes the impaired wound healing and improves diabetic PVD is previously unreported.

The major factors affecting wound healing in diabetes include hyperglycemia, PVD, and neuropathy. Hyperglycemia can induce oxidative stress when the production of reactive oxygen species exceeds the antioxidant capacity (29). Oxidative stress and overproduction of reactive oxygen species cause damage to vascular endothelial cells, impair EPC mobilization and homing, cause dysfunction in fibroblasts and epidermal cells, decrease host immune resistance, and increase the levels of metalloproteases. Hypoxia caused by insufficient perfusion as a result of PVD can amplify the early inflammatory response, thereby prolonging injury by increasing the levels of oxygen radicals. The neuropathy that occurs in DFUs probably also contributes to impaired wound healing. Neuropeptides, such as nerve growth factor, promote cell chemotaxis, induce growth factor production, and stimulate the proliferation of cells. A decrease in neuropeptides has been associated with DFU formation, and denervated skin exhibits reduced leukocyte infiltration (30,31). An animal model closely approximating these hallmarks of DFUs in humans is necessary to evaluate therapeutic strategy. Published studies done at an early stage of diabetes in rats or mice without PVD and/or neuropathy are not fully clinically relevant. Experiments in animals do not guarantee successful clinical trials, so it is important to use a model that approximates the disease in human. For these reasons, we tested the AF combination therapy in the more rigorous severely diabetic rat model with neuropathy and vasculopathy, representing the hallmark of DFUs in human.

In this study, T1DM rats were used for wound healing studies when animals developed signs of neuropathy (increased duration time tolerated for the hot plate test) at 6 weeks after STZ injection, while T2DM GK rats were used at 5 months old when animals developed neuropathy and peripheral artery disease. Impaired wound healing was observed in these diabetic animals compared with age-matched controls, and the healing time of dorsal wounds was prolonged by 8 days in T1DM and 7 days in T2DM rats because of the severe diabetes that provided a clinically relevant model to test.

It has been reported that hyperglycemia impairs EPC mobilization and homing and, therefore, reduced angiogenesis in diabetic wounds (32–34). Few CD34⁺ and CD133⁺ stem cells in tissue sections of wounds, absence of an increase in VEGF-A and HGF-α expression in the granulation tissues, and less vasculature formation (RECA and α-SMA staining) were observed in control saline-treated diabetic rats at day 7 post-injury. Fewer stem cells present in granulation tissues may result in impaired healing in diabetic rats, and increasing the number of stem cells present may improve diabetic wound healing. Indeed, nonintravascular delivery by local injection or topical administration of bone marrow–derived mesenchymal stem cells showed beneficial effects in diabetic wound healing in rodent and human trials (14–17). Our AF therapy greatly increased the number of CD133⁺ stem cells and CD31⁺CD133⁺ EPCs in peripheral blood, indicating that pharmacological mobilization of bone marrow stem cells was not impaired by hyperglycemia. Further, SDF-1–expressing cells, as well as CD133⁺ and CD34⁺ stem cells, were dramatically increased in the granulation tissues recovered from diabetic rats with AF combination therapy at 7 days post-wounding, in spite of hyperglycemia. Increased number of CD133⁺ stem cells in the granulation tissues was accompanied by higher levels of VEGF-A and HGF-α expression and increased vasculature formation. These results suggest that AF combination not only mobilizes stem cells into the circulation, increasing the availability of stem cells, but also recruits these stem cells into the wound sites in diabetic rats. Importantly, accumulation of stem cells in the granulation tissues promotes angiogenesis and enables normalization of wound healing in diabetes.

Macrophages play an important role in the entire progress of wound healing, and transition of macrophages from predominantly proinflammatory (M1-like phenotypes) at the first stage of wound healing (hemostasis phase) to anti-inflammatory (M2-like phenotypes) at the proliferative phase is critical for healing processes, including angiogenesis and vasculogenesis (35,36). Nonhealing wounds, such as diabetic ulcers, remain indefinitely inflamed at the first stage of wound healing (37,38). Restoration of macrophage function is considered to be a therapeutic strategy for chronic nonhealing wounds (39). In our study, the recruitment of macrophages was not impaired by hyperglycemia, and a large number of macrophages (ED1⁺) were detected in the granulation tissues recovered from saline-treated diabetic rats at 7 days post‐injury, but few of these macrophages costained with Ym1/2, a marker of M2-like phenotypes. Interestingly, a dramatically increased number of Ym1/2⁺ macrophages (M2) appeared in the granulation tissues recovered from diabetic rats treated with AF combination therapy (Fig. 5), and the increased number of M2-like phenotypes was associated with improved angiogenesis (Fig. 6) and wound healing. We have reported that low-dose FK506 increased Ym1⁺ M2 macrophages in the injured peritoneum, and low-dose FK506 had a synergistic effect with AMD3100 in increasing M2 macrophages in the injured sites (40). FK506 at a low dose activates the BMP signaling pathway (41) through FKBP12 ligands (26), and BMP4 expression induces macrophage polarization toward M2 (42). Notably, most Ym1/2⁺ M2-like macrophages costained with SDF-1 (Fig. 5C), and the accumulation of SDF-1–producing macrophages led to the recruitment of more stem cells (22). In addition, stem cells can induce macrophage polarization toward M2-like phenotypes (43). Thus, the concentrated CD133⁺ stem cells in the granulation tissues may not only promote angiogenesis but also reduce the inflammatory response through an M1-to-M2 transition, allowing the wound to enter the proliferative phase.

CD133⁺ stem cells have a vital role in effective skin regeneration. CD133 is a marker for long-term repopulating murine epidermal stem cells, and CD133⁺ keratinocytes formed both hair follicles and epidermis after injection into immunodeficient mice (44). Lineage tracing studies demonstrated the presence of CD133⁺ cells in the hair follicles of intact skin (45), and CD133⁺ stem cells generate hair follicles and epithelium during wound healing (22). In the present study, AF combination therapy dramatically increased CD133⁺ cells in the granulation tissues, but the question remains whether these CD133⁺ stem cells were liberated from bone marrow by AF combination treatment or whether these CD133⁺ stem cells originated in adjacent normal skin. By using a GFP transgenic BMT model, we found that CD133⁺ stem cells were GFP positive, and these CD133⁺ cells localized in clusters of GFP-positive cells in the granulation tissues. This demonstrated that CD133⁺ cells in the wound were derived from the bone marrow. It is possible that bone marrow–derived CD133⁺ stem cells differentiated into other cell populations and formed GFP-positive tissues, while the CD133 marker waned as the cells differentiated. Further study is warranted to better understand the details of the mechanism responsible for the benefit from AF treatment.

The incidence of PVD is much higher in patients with diabetes than in those without diabetes. Impaired wound healing is frequently associated with PVD in patients with diabetes. Neuropathy in the presence of PVD leads to neuroischemic ulcers, which account for most nonhealing wounds and potential limb loss (46). Surgical revascularization significantly improves the rate of limb salvage, but there is often no large artery compromise to correct, and regardless, the problems of healing a DFU resulting from microangiopathy and impairment of angiogenesis remain unchanged (47). An unexpected finding in the present study is that AF combination therapy not only normalized the impaired healing but also restored the foot skin temperature in diabetic GK rats. Histological studies demonstrated the amelioration of angiopathy in interphalangeal blood vessels (Fig. 8). Hyperglycemia-induced endothelial cell damage through oxidative stress is considered the initial injury in the blood vessels (48). Endothelial injury/dysfunction in diabetes is closely linked to the occurrence of angiopathy. If damaged endothelial cells can be repaired/repopulated by EPCs, it is possible to prevent/ameliorate diabetic angiopathy. In patients, the number of EPCs in peripheral blood is negatively correlated with the severity of atherosclerosis (49,50). Indeed, vascular repair by EPCs was detected in mice (51), and chronic treatment with bone marrow–derived EPCs prevented the development of the atherosclerotic disease in apolipoprotein E knockout mice. EPCs also enhanced reendothelialization and reduced neointima formation after induction of endothelial cell damage using the carotid artery model (52). Circulating CD31⁺CD133⁺ EPCs were increased fourfold in diabetic rats treated with AF combination compared with saline controls (Fig. 4B). Mobilized EPCs by the AF combination treatment may repair and/or repopulate the endothelium damaged by hyperglycemia and therefore prevent or ameliorate the development of angiogenesis and atherosclerosis.

In conclusion, we have developed a simple and safe systemic therapy for DFUs in both T1DM and T2DM animals that have neuropathy and PVD. AF combination therapy mobilizes bone marrow–derived stem cells into the circulation, recruits the stem cells into the wound sites, promotes M1-to-M2 transition in the granulation tissues, and thereby accelerates healing in diabetic rats. Mobilized CD31⁺CD133⁺ EPCs may also ameliorate angiopathy. Improvement of angiogenesis and amelioration of angiopathy by pharmacological mobilization and recruitment of bone marrow–derived stem cells not only contributes to faster, more complete diabetic wound healing but also may help to protect against recurrence.

Acknowledgments. The authors thank Dr. John Harmon and Dr. G. Melville Williams (Johns Hopkins University) for comments and advice on the study.

Funding. This work was made possible by a Start-up fund from Surgery, Johns Hopkins Hospital, and a grant from MedRegen LLC. L.Q. was partially supported by the China Scholarship Council.

Duality of Interest. J.B. is currently chief medical officer of MedRegen LLC. Z.S. is a cofounder of and holds equity in MedRegen LLC. Additionally, Z.S. is an inventor of technology and participates in research funded by MedRegen LLC that intends to further develop the technology. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.Q. performed the surgical excisional wound model, BMT, and histological studies. A.R.A., J.H., M.C., H.K., K.I., W.W., and R.W. gave animal treatment, evaluated the wounds, and analyzed data. B.P. evaluated the nociceptive intraepidermal nerve fibers in diabetic rats. A.M.C. and J.B. analyzed data and edited the manuscript. S.C. and Z.S. designed the study. Z.S. supervised the study, analyzed data, and wrote the manuscript. Z.S. 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 79th Scientific Sessions of the American Diabetes Association, San Francisco, CA, June 7–11, 2019.