Treatment with bone marrow‐derived stromal cells accelerates wound healing in diabetic rats

Abstract

Bone marrow stem cells participate in tissue repair processes and may have a role in wound healing. Diabetes is characterised by delayed and poor wound healing. We investigated the potential of bone marrow‐derived mesenchymal stromal cells (BMSCs) to promote healing of fascial wounds in diabetic rats. After manifestation of streptozotocin (STZ)‐induced diabetic state for 5 weeks in male adult Sprague–Dawley rats, healing of fascial wounds was severely compromised. Compromised wound healing in diabetic rats was characterised by excessive polymorphonuclear cell infiltration, lack of granulation tissue formation, deficit of collagen and growth factor [transforming growth factor (TGF‐β), epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), platelet‐derived growth factor PDGF‐BB and keratinocyte growth factor (KGF)] expression in the wound tissue and significant decrease in biomechanical strength of wounds. Treatment with BMSC systemically or locally at the wound site improved the wound‐breaking strength (WBS) of fascial wounds. The improvement in WBS was associated with an immediate and significant increase in collagen levels (types I–V) in the wound bed. In addition, treatment with BMSCs increased the expression of growth factors critical to proper repair and regeneration of the damaged tissue moderately (TGF‐β, KGF) to markedly (EGF, VEGF, PDGF‐BB). These data suggest that cell therapy with BMSCs has the potential to augment healing of the diabetic wounds.

Introduction

Wound healing is a complex process involving a highly regulated cascade of events, initiated by interactions between many cell types, soluble factors and matrix components 1, 2. Recruitment of inflammatory cells, formation of granulation tissue with angiogenesis, fibroblast proliferation and migration of keratinocytes contribute to the restoration of anatomical and functional integrity 3, 4. This dynamic process of wound healing is also dependent on systemic signals, such as growth factors, cytokines, chemokines and proteolytic enzymes (5). The sum of all of these interrelated events determines the speed of wound healing. Compromises in wound healing occur when this orderly and complex sequence of cellular and molecular events is disrupted. The aetiologies of compromised wound healing are complex and multifactorial; impaired inflammatory cell migration, reduced growth factor levels and reduced collagen synthesis, all significantly impair the ability of wounds to heal normally 6, 7, 8.

Despite medical progress made in wound‐care management, problematic or chronic wounds are still a challenging complication affecting 10–40% of the population. Subset populations, namely those that suffer from diabetes or receive steroid therapy or chemotherapy regimes are known to have significantly impaired wound healing. The healing process in diabetes can be particularly prolonged and incomplete, resulting in poor anatomical and functional outcomes 9, 10, 11.

Diabetic wounds exhibit reduced chemotactic ability to recruit inflammatory cells into the damaged tissue (5), creating an attenuated inflammatory response. Growth factor production is also reduced, often correlating with the extent of prolonged or insufficient healing (12). These deficiencies lead to impaired angiogenesis as a result of dysfunctional endothelial progenitor cells that are unable to differentiate and proliferate. Impaired neovascularisation prolongs wound healing because of the lack of oxygen substrates 13, 14. Collagen matrixes, which are severely compromised in diabetic wounds, also contribute to poor wound healing 14, 15, 16.

Novel therapies aimed at improving neovascularisation and growth factor production could improve outcomes in poorly healing wounds. Numerous attempts to promote the healing of difficult wounds have focused on increasing systemic or local tissue growth factor levels. Topical growth factor applications including epidermal growth factor (EGF), platelet‐derived growth factor (PDGF), transforming growth factor‐beta (TGF‐β), and fibroblast growth factor (FGF) for the treatment of complex or chronic wounds have produced mixed results 17, 18, 19, 20, 21, 22. Other delivery vehicles, such as fibroblast implants, have been used as biological wound dressings with only modest improvements in wound healing (23). While these therapies have shown some promise, difficulties in delivery, application and cost have limited the use of these treatment strategies.

Recently, stem cell therapy has been successfully used for restorative repair and tissue regeneration 24, 25. The versatile nature of bone marrow‐derived mesenchymal stem cells has been widely published 26, 27, 28. These pluripotent stem cells, also referred to as bone marrow stromal cells, are capable of differentiation into numerous cell types, including myoblasts, cartilage, bone, muscle and brain cells, and are also able to produce growth factors and cytokines and engraft to the sites of injury 27, 28, 29, 30. They have been shown to be successful in adjunctive restorative therapy in stroke, myocardial infarctions and osteogenesis imperfecta 31, 32, 33. Others and we have shown the ability BMSCs to promote healing of normal wounds 34, 35, 36. However, efficacy of BMSCs or other adult tissue‐derived stem cells for promoting healing of diabetic wounds has not been adequately investigated.

In the present study, we show that systemic or local administration of syngeneic BMSCs augments healing of the fascial wounds in diabetic rats. Treatment with BMSCs improved wound healing histologically, increased wound‐breaking strength (WBS) and collagen production and upregulated the expression of growth factors critical to wound healing.

Materials and methods

Animals: Adult male, Sprague–Dawley rats (Charles River, Protage, MI) of an average weight of 250–300 g were housed in the animal care facility of the Henry Ford Health System. Rats consumed standard rat chow and water ad libitum. Age‐matched animals were used within any given experiment. The experimental protocol for these studies was approved by the Institutional Animal Care and Use Committee and all animals received humane care.

Diabetic rat model: A diabetic model was established in male Sprague–Dawley rats using streptozotocin (STZ) as described by Junod et al. with some modifications (37). Briefly, rats were fasted for 16 hours before one time intraperitoneal (IP) injection of STZ dissolved in sodium citrate (0·1 mM, pH 4·5) at a dose of 64 mg/kg. Five days post‐STZ injection and three times per week thereafter, blood samples were obtained via a fine needle prick at the tail vein and the blood glucose levels were analysed by glucometer. Glucose levels were maintained between 300 and 500 mg/dl with subcutaneous protamine zinc insulin (PZI) U40 insulin (0·5–2·0 units) given as needed to maintain these levels. Rats were allowed to manifest the diabetic state for 5 weeks prior to surgical wounding. Age‐matched normal rats served as controls.

Wound‐healing model: Control and diabetic rats were fasted overnight prior to surgical intervention and were anaesthetised with IP ketamine (60 mg/kg) and xylazine (8 mg/kg). The abdomen was shaved and the skin was cleansed with alcohol and Betadine (The Purdue Frederick Co. Norwalk, CT) solutions. A 5‐cm‐long midline incision through the skin was made; the skin was retracted, followed by another 5‐cm‐long midline abdominal fascial incision into the peritoneum. The fascial incision was immediately closed with a running 3‐0 prolene suture. The skin was then closed with an interrupted 4‐0 nylon suture. The animals were allowed to recover and they were returned to their housing and allowed chow and water ad libitum.

Determination of WBS: Animals were euthanised with an overdose of ketamine and xylazine on day 7 post‐wounding and abdominal walls were removed for determination of WBS. The incision line was examined macroscopically; integrity of incision line, existence of abscess or abnormal adhesion formation was noted. The running prolene suture was carefully removed to preserve the surgically manipulated abdominal fascia. Two strips of abdominal muscle and fascia, measuring 1 cm in width and 4 cm in length, were removed from the middle portion of the abdominal wall with a specially constructed tissue harvester. The strips were analysed for WBS in a TCD 200 computer‐driven tensiometer (Chatillon, NY) at a distraction rate of 20 mm/minute. Data are presented as means of WBS ± SD.

BMSCs: BMSCs, generated from marrow aspirates from the long bones (femur and tibia) of normal adult Sprague–Dawley rats were provided by Cognate BioServices Inc, Baltimore, MD. Cryopreserved BMSCs were thawed at 37°C and washed twice in cold phosphate‐buffered saline (PBS) before use. The viability of thawed cells typically ranged from 60% to 65%. Flow cytometry showed the presence of stromal surface markers CD29, CD73 and CD90 on more than 90% of rat BMSCs and absence of CD11b/c, CD45 and major histocompatibility complex class II (<1% of cells). BMSCs harvested between passages 2–6 were used in these experiments.

Treatment with BMSCs: For systemic administration, 1·5 × 10⁶ BMSCs viable cells were resuspended in 1 ml PBS and injected once daily over a 2‐minute period for 4 days into each animal via the tail vein starting 24 hours after wounding. For local treatment with BMSCs at the wound site, 6 × 10⁶ cells were resuspended in 0·5 ml of PBS and 50 μl of cell suspension was injected at 10 different sites 1 mm lateral to the wound edge along the entire length of the incision immediately after closing. Control animals were treated with PBS without BMSCs.

Collagen measurement: Total soluble collagen (types I–V) in wound tissue was measured colorimetrically using Sircol Collagen Assay kit (Newtonabbey, Northern Ireland) as per manufacturer‘s protocol. Briefly, wound tissue samples were homogenised in lysis buffer (100 mM potassium phosphate, 0·1% Triton X‐100, 2 mM dithiothreitol (DTT), 100 μg/ml phenylmethylsulphonylflouride (PMSF), pH 7·8) in which the ratio of tissue weight to lysis buffer was 0·5 g per 2·5 ml of buffer. After homogenisation, tissue debris was removed by centrifugation. One millilitre of Sircol Dye reagent was added to 100 μL of tissue extract and stirred for 30 minutes at room temperature. The collagen–dye complex was separated by centrifugation at 10 000 × g for 15 minutes and the bound dye was recovered in 1 ml of alkali reagent. Absorbance was measured at 540 nm in a spectrophotometer. The quantity of collagen in each sample was determined from the standard curve generated by using collagen standards.

Detection of growth factors by reverse transcriptase polymerase chain reaction (RT–PCR): Total cellular RNA was extracted from wound tissue with TRI‐zol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s recommendation. 2·5 μg of RNA was then reverse transcribed by using random primers (Boehringer Mannheim, Germany) and reverse transcriptase to generate cDNAs. The growth factor primer sequences for PCR amplification were as follows: TGF‐β, upper, 5′‐CGA GGT GAC CTG GGC ACC ATC CAT GAC‐3′, and lower, 5′‐CTG CTC CAC CTT GGG CTT GCG ACC CAC‐3′; PDGF, upper, 5′‐TCC CTC TAC CCC AAG AAC CT‐3′, and lower, 5′‐GAT CTG GGT GCC ATG AGA GT‐3′; EGF, upper, 5′‐ATG TCT GCC AAT GCT CAG AAG G‐3′, and lower, 5′‐TAG GAC CAC AAA CCA AGG TTG GG‐3′; VEGF, upper, 5′‐GAG TAT ATC TTC AAG CCG TCC TGT‐3′, and lower 5′‐ATC TGC ATA GTG ACG TTG CTC TC‐3′; KGF, upper, 5′‐GTA GCG ATC AAC TCA AGG TC‐3′, and lower, 5′‐ATT TAA GGC CAC GAA CAT TT‐3′; FGF, upper, 5′‐GCA GTA TAA ACT CGG ATC CAA AAC‐3′, and lower, 5′‐GCC TGA GAG TGA CAG TGT CTA AAG‐3′; hepatocyte growth factor (HGF), upper, 5′‐CCC AAA TGT GAC GTG TCA AG‐3′, and lower, 5′‐ATC CCA AGG AAC GAG AGG AT‐3′; glyceraldehyde‐3‐phosphate dehydrogenase (GADPH), upper, 5′‐TCC CTC AAG, ATT, GTC AGC AA‐3′, and lower, 5′‐AGA TCC ACA ACG GAT ACA TT‐3′. 0·5–1·0 μg of cDNA was amplified by PCR for 35 cycles of denaturation (94°C for 1 minute), annealing (61°C for 1 minute), and polymerisation (72°C for 2 minutes). The PCR products were separated by 1·5% agarose gel electrophoresis and visualised by ethidium bromide staining. Gels were photographed with Direct Screen Instant Polaroid Camera System from Hoefer Scientific Instruments, San Francisco, CA. Photographs were digitised and band intensity of PCR products was determined using Scion image analysis program version 4·0·3·2 (Scion Corporation, Frederick, MD). Values are presented as arbitrary units that have been normalised against GADPH signal determined in each experiment. Various primers used in these experiments amplified DNA fragments of 404 base pairs (bp) (TGF‐β), 215 bp (PDGF), 617 bp (EGF), 230bp (VEGF), 570 bp (KGF), 180 (FGF), 246 bp (HGF), and 300 bp (GADPH).

Measurement of growth factors by enzyme‐linked immunosorbent assays (ELISA): As there is greater than 95% homology between human and rat TGF‐β and EGF and rat‐specific ELISA kits for these growth factors are not commercially available, we used human ELISA kits to measure TGF‐β and EGF in wound tissue lysates. All reagents and working standards were prepared as directed by the manufacturer. Fifty microlitres of assay diluent and 50 μl cell culture supernatant were added to each well of antibody‐coated 96‐well plates in duplicates. The contents in the wells were gently mixed and plates incubated for 2 hours at room temperature. After incubation, wells were washed five times with cold PBS and then 200 μl of secondary antibody‐horseradish peroxidase conjugate was added to each well. Following incubation for 2 hours at room temperature, wells were washed and 200 μl of substrate solution was added to each well and incubated for 30 min at room temperature. Reaction was stopped by adding 50 μl of stop solution to each well and optical density of each well was determined using a microplate reader set to 590 nm.

Statistical methods: Statistical analysis was performed by t‐test and two‐way analysis of variance. Data are presented as mean values ± SD. Statistical probability were assigned as *P < 0·05 and **P < 0·01·

Results

Within 1 week of STZ injection, all rats had elevated blood glucose levels in the range of 350–500 mg/dl (normal range 87–122 mg/dl, 95 ± 14). Over the course of 5 weeks, the rats were allowed to manifest diabetic state and glucose levels were maintained between 300 and 500 mg/dl with subcutaneous PZI U40 insulin (0·5–2·0 units) given as needed. Average body weights (g) at time of surgery were normal control, 449 ± 23; diabetic, 343 ± 29, that is, 23–24% reduction in body weight following STZ injection.

Diabetic wounds exhibit poor wound healing and cell therapy with BMSCs partially restores wound healing

Normal control, diabetic control and diabetic rats treated with BMSCs were euthanised on day 7 post‐wounding to determine the biomechanical strength of the wounds. Diabetic control rats exhibited significant impairment in wound healing compared with normal control rats, as determined by WBS. In normal rats, WBS on day 7 was 8·9 ± 2·6 N compared with diabetic rats whose WBS was significantly reduced, for example, 2·6 ± 1·6 N (Figure 1A, P < 0·01). WBS of diabetic wounds treated with BMSCs systemically increased modestly (3·9 ± 2·1 N, P = 0·06) but did not achieve statistical significance compared with untreated diabetic wounds (P = 0·06). However, single treatment of diabetic wounds with BMSCs via local injection of cells at multiple sites along the entire length of the wounds significantly improved WBS on day 7. The breaking strength of diabetic wounds treated locally with BMSCs was 4·3 ± 1·3 N compared with 2·6 ± 1·6 N WBS of untreated diabetic wounds (P < 0·05). These data show that induction of diabetic state with STZ in rats compromises wound healing and systemic and local cell therapy with BMSCs partially restores healing of these wounds.

Figure 1.

Treatment of diabetic wounds with BMSCs increases wound‐breaking strength (WBS). A 5‐cm long midline abdominal fascial incision was made with a scalpel in normal and diabetic rats. One day after wounding, diabetic rats were treated with syngeneic bone marrow‐derived mesenchymal stromal cells (BMSCs) systemically or locally. For systemic administration, 1·5 × 10⁶ BMSCs cells in 1 ml of phosphate‐buffered saline (PBS) were injected once daily over a 2‐minute period for four consecutive days into each animal via the tail vein (n = 7). For local treatment with BMSCs at the wound site, 6 × 10⁶ viable cells were suspended in 0·5 ml PBS and 50 μl cell suspension was injected at 10 different sites 1 mm lateral to the wound edge along the entire length of the incision immediately after closing (n = 6). Normal control (n = 8) and diabetic control (n = 8) animals were treated with PBS without BMSCs. Wounds were harvested on day 7 and WBS measured using a tensiometer as described in Materials and Methods. (A) Data shown are means ± SD. *P < 0·05, **P < 0·01. (B) Photomicrographs of histological sections of untreated normal control (a), untreated diabetic control (b) and diabetic wounds treated with BMSCs intravenously (c). Wounds were harvested on day 7 and processed for histological examination of haematoxylin and eosin‐stained tissue sections. Original magnification ×40 (a,b,c).

Histologically there were discernible differences between normal, diabetic and diabetic wounds treated with BMSCs with respect to overall tissue organisation, cellular infiltration, neovascularisation and collagen matrix. On day 7, wounds of healthy rats exhibited increased neovascularisation set in an oedematous stroma, predominately mononuclear cell infiltration and collagen fibrils arranged haphazardly (Fig 1B). Control diabetic wounds (untreated) showed scant neovascularisation, abundant inflammatory infiltrate mostly comprising polymorphonuclear cells and foci of necrotic tissue with neutrophils. In contrast, diabetic wounds treated with BMSCs intravenously showed evidence of neovascularisation set in an oedematous stroma and predominantly mononuclear inflammatory cells with few polymorphonuclear cells without tissue necrosis. Thus, histopathological analysis of wound tissue show impaired resolution of early inflammation and lack of granulation tissue formation in diabetic wounds compared with wound of normal control rats and improved healing of diabetic wounds following BMSC therapy.

BMSC cell therapy increases wound collagen

To determine whether treatment with BMSCs increases collagen levels in the diabetic wounds, total soluble collagen types I–IV in normal control and diabetic untreated and treated wounds was measured colorimetrically using Sircol Collagen Assay kit. As shown in Figure 2, the collagen content of diabetic wounds on day 7 was significantly reduced compared with collagen levels of control wounds (day 7, normal control = 1400 ± 35 μg/g, diabetic control = 701 ± 38 μg/g). Systemic treatment of diabetic wounds with BMSCs significantly increased collagen levels compared with diabetic control rats (901 ± 50 μg/g versus 700 ± 38 μg/g, P < 0·05), suggesting that increase in tensile strength of diabetic wounds treated with BMSCs is attributable, at least in part, to an increase in the synthesis of collagen, a major component of the extracellular matrix (ECM).

Figure 2.

Treatment with bone marrow‐derived mesenchymal stromal cells (BMSCs) increases collagen content of diabetic wounds. Diabetic animals were wounded and treated with BMSCs intravenously as described in Figure 1. Wound tissue was harvested on days 7, homogenised in lysis buffer and clarified by centrifugation. Total soluble collagen was measured colorimetrically at 540 nm using the Sircol Collagen Assay kit. Bar graphs represent mean collagen concentration (μg/g) ± SD.

Effect of BMSC therapy on growth factor expression in diabetic wounds

We first compared the expression of various growth factors (e.g. TGF‐β, EGF, VEGF, PDGF‐BB and KGF) in wound tissue of normal and diabetic rats by RT–PCR. Figure 3A compares the expression levels of these growth factors in 7 day wound tissue of normal and diabetic rats (five rats per group). There was 30–35% reduction in the expression of TGF‐β, EGF and KGF in diabetic wounds, but expression of PDGF‐BB or VEGF expression was not affected.

Figure 3.

Effect of bone marrow‐derived mesenchymal stromal cells (BMSC) treatment on growth factor expression in diabetic wounds. (A) Reverse transcriptase polymerase chain reaction (RT–PCR) analysis of growth factor expression in wound tissue of normal and diabetic rats. Total cellular RNA was isolated and reverse transcribed to generate cDNAs from wound tissue harvested 7 days after wounding as described in Materials and Methods. cDNA (0·5–1·0 μg) was amplified using growth factor‐specific upper and lower primers and amplified products were separated on 1·5% DNA agarose gel. Gels were stained with ethidium bromide and amplified DNA fragments were identified by base pair (bp) sizes. Top panel shows pattern of growth factor expression in wounds of normal and diabetic rats. Bar graph compares band densities of various growth factors in normal and diabetic wounds (n = 5). (B) RT–PCR analysis of growth factor expression in diabetic wounds treated with BMSCs. Untreated diabetic wounds and diabetic wounds treated with BMSCs intravenously (i.v.) or locally as described in Figure 1 were analysed for growth factor expression by RT–PCR as described above. Data presented compare pattern (top panel) and relative band densities (bar graph) of growth factors in treated and untreated diabetic rats (n = 3). Similar results were obtained in two independent experiments. (C) Measurement of growth factor secretion by enzyme‐linked immunosorbent assays (ELISA). Levels of TGF‐β and EGF in tissue homogenates prepared from normal control, diabetic control and diabetic wounds treated with BMSCs systemically or locally were measured by ELISA following the instructions provided in the kit. Data shown are mean ± SD of three replicate measurements with three rats in each group. EGF, epidermal growth factor; GADPH, glyceraldehyde‐3‐phosphate dehydrogenase; KGF, keratinocyte growth factor; PDGF‐BB, platelet‐derived growth factor; TGF‐β, transforming growth factor; VEGF, vascular endothelial growth factor.

Next we analysed the expression of these growth factors in diabetic wounds treated with BMSCs systemically and locally. Figure 3B shows moderate to marked increase in expression of VEGF (28%), EGF (68%) and PDGF‐BB (48%) in diabetic wounds treated with BMSCs intravenously. There was only a slight increase in expression of TGF‐β or KGF (<10%). Similar increases in growth factor expressions were observed in wounds treated with BMSCs locally with the exception that increase in EGF expression was less pronounced in locally treated wounds compared with systemically treated wounds. Together, these data show that in general the expression of growth factors involved in the repair of injured tissue is decreased in the diabetic wounds and cell therapy with BMSCs increases their expression moderately (TGF‐β, KGF) to significantly (EGF, PDGF‐BB and VEGF).

We also measured the levels of TGF‐β and EGF protein levels in wound tissue from normal, diabetic and diabetic rats treated with BMSCs systemically or locally by ELISA. Both growth factors were decreased in diabetic wounds and were significantly increased following treatment with BMSCs systemically or locally (Figure 3C, P < 0.05).

Discussion

Optimal wound healing depends on a multitude of factors including cell migration, growth factor production and highly regulated collagen synthesis and matrix deposition. Diabetes, however, disrupts this highly coordinated series of events (38); there is abnormal inflammatory response, decreased vascularisation, compromised production and functionality of growth factors and defective collagen metabolism (14).

BMSC therapy has emerged as a highly promising form of regenerative cell therapy, from cardiac injury to neuronal regeneration 31, 32, 39. The cells can enter into a trans‐differentiation pathway to be programmed into cells required for regenerating damaged tissues. They are also capable of secreting essential cytokines critical to wound repair 29, 30. The extraordinary plasticity and the regenerative potential of BMSCs to promote tissue repair prompted our investigation into the beneficial effects on wound healing in a diabetic model. Systemic and local administration of BMSCs resulted in improvement in healing of fascial wounds in diabetic rats as determined from improvement in histology of the wounds and an increase in WBS. Diabetic wounds showed excessive polymorphonuclear leucocyte infiltration that failed to resolve and wounds were found to have lost the ideal synchrony of cellular events that leads to rapid healing of wounds, that is, early mononuclear cell infiltration, fibroblast proliferation, formation of granulation tissue and synthesis of ECM. Consequently, the biomechanical strength of diabetic wounds was significantly reduced. Although diabetic wounds treated with BMSCs still showed dyssynchrony of the healing phases, wound healing had definitely improved with respect to inflammatory cell infiltration, evidence of neovascularisation, amount and arrangement of collagen matrix and WBS, all indicative of more rapid maturation of BMSC‐treated wounds.

The improvement in WBS was associated with increased collagen and growth factor expression in wounds treated with BMSCs, corroborating the results of our previous studies showing improvement in healing of fascial and cutaneous wounds with BMSC cell therapy in healthy rats (34). These results are also in agreement with studies of Badiavas et al. 35, 40 showing participation of bone marrow‐derived cells in cutaneous wound healing in rats and chronic venous ulcers in patients and Yamaguchi et al. (36) showing that topically applied bone marrow cells to cutaneous excision wounds accelerate wound healing when combined with occlusive dressings. However, whether wound‐healing effects of whole bone marrow cells were derived from mesenchymal or haematopoietic stem cells was not clear from these reports. In contrast, our present studies show the wound‐healing capacity of in vitro expanded BMSCs.

We have previously shown that compared with BMSCs, treatment with skin‐derived rat fibroblasts is significantly less effective in augmenting healing of the fascial wounds, showing the specificity of BMSC therapy for wound healing (34).

There appears to be several reasons why diabetic wounds treated with BMSCs heal better than untreated diabetic wounds. The improved WBS and upregulation of growth factors suggest at least two possibilities by which BMSCs enhance wound healing in this diabetic model: (i) improved collagen metabolism and (ii) improved functionality of growth factors, such as TGF‐β, EGF and PDGF, which affect the healing of wounded tissue.

The improvement in WBS was associated with an increase in collagen concentration at the wound bed. Increase in collagen expression was immediate and high in normal control wounds. In the diabetic untreated wounds, there was diminished collagen formation and WBS, both of which improved following treatment with BMSCs. Collagen is an important ECM component that provides strength, integrity and structure to normal tissues and also is needed to repair the defect created by injuries, thereby restoring tissue structure and function 15, 41, 42. The improved collagen production correlated with improved WBS in the BMSC‐treated diabetic animals. However, it can be argued that Sircol collagen assay measures collagen rather non specifically; therefore, collagen measured by this assay may not accurately quantitate collagen types I and III that are critical to the formation of ECM in healing wounds. As stated earlier, Sircol kit measures collagen types I–IV, given that types I and III are the major collagen species in the wounds we make the assumption that the amount of collagen measured by Sircol is largely because of collagen types I and III.

Secondly, the local production of chemokines acts to enhance the wound‐healing process. In a normal healing wound, the early appearance of cytokines and growth factors presumably provide a receptive environment in which the other growth signals originating from inflammatory cells can exert their effects. However, there is significantly reduced expression of growth factors in chronic wounds and their activity levels are dysregulated in non healing wounds. In the untreated diabetic wounds, reduced expression of TGF‐β, EGF, KGF and PDGF‐BB was associated with impaired wound healing. Following BMSC therapy, an increase in expression of these growth factors correlated with increase in collagen production and improved WBS. The increased growth factor levels at the wound site presumably stimulated cell adhesion at the site of injury, induced cells to secrete more chemokines or forced stem cells to trans‐differentiate into cells required for repairing damaged tissue. Although improved healing of diabetic wounds treated with BMSCs correlated with increased growth factor expression, these results should be interpreted with caution, because expression levels were only marginally improved. This might be attributed to the low sensitivity of RT–PCR technique that we used to measure the expression of growth factors. Thus, changes seen in growth factor expression in treated wounds are more qualitative than quantitative in nature. Clearly, vigorous analysis of growth factors using more sensitive and quantitative technique such as real‐time PCR is warranted to establish their role in BMSC therapy of diabetic wounds.

Alternatively BMSCs may facilitate wound healing by trans‐differentiating into a multiple cell types that integrate into the wound. However, BMSCs do not necessarily need to trans‐differentiate following injury. Rather, through the local upregulation of cytokines and growth factors, they may enhance the wound‐healing programme either through the improved inflammatory response or the recruitment and proliferation of cells (e.g. fibroblasts) critical to wound healing. This investigation suggests that upregulation of growth factor expression in wound tissue following BMSC therapy plays a key role in promoting cellular processes such as chemotaxis, cell proliferation, cell signalling, ECM formation, and angiogenesis that are critical to proper wound healing.

Although systemic therapy of diabetic wounds with BMSCs did not achieve statistical significance in WBS, benefits of early BMSC cell therapy are still promising because there was an overall improvement in healing of wounds microscopically and an increase in WBS, especially of wounds treated with BMSCs locally. We attribute differences in efficacy of systemic versus local BMSC therapy of wounds to a decrease in the number of BMSCs migrating to the wound after systemic administration as opposed to wounds treated with BMSCs locally. Perhaps the presence of arterial‐venous shunts in diabetic skin contributed to the decreased migration of BMSCs to the wounds treated systemically. Thus, further understanding of the biology of BMSCs and their mechanisms of action, including homing of BMSCs to the wound and their cell signalling and differentiation pathways during wound‐induced activation could lead to novel therapeutic strategies to enhance efficacy of BMSC cell therapy of diabetic wounds.