• Open Access

Integration of Scaffolds into Full-Thickness Skin Wounds: The Connexin Response

Authors


Abstract

Scaffolds have been reported to promote healing of hard-to-heal wounds such as burns and chronic ulcers. However, there has been little investigation into the cell biology of wound edge tissues in response to the scaffolds. Here, we assess the impact of collagen scaffolds on mouse full-thickness wound re-epithelialisation during the first 5 days of healing. We find that scaffolds impede wound re-epithelialisation, inducing a bulbous thickening of the wound edge epidermis as opposed to the thin tongue of migratory keratinocytes seen in normal wound healing. Scaffolds also increase the inflammatory response and the numbers of neutrophils in and around the wound. These effects were also produced by scaffolds made of alginate in the form of fibers and microspheres, but not as an alginate hydrogel. In addition, we find the gap junction protein connexin 43, which normally down-regulates at the wound edge during re-epithelialisation, to be up-regulated in the bulbous epidermal wound edge. Incorporation of connexin 43 antisense oligodeoxynucleotides into the scaffold can be performed to reduce inflammation whilst promoting scaffold biocompatibility.

1. Introduction

Scaffolds are designed to promote the healing of wounds when there are large tissue deficits, such as in some chronic ulcers or in deep burns.1 The protracted healing of chronic wounds such as diabetic foot ulcers and venous leg ulcers have a measurable morbidity that is detrimental to patients as well as being a major burden on healthcare services. About £400 million a year is spent treating 120,000 ulcer patients in the UK.2 Similarly, following extensive full-thickness burns, where tissue damage extends deep into the lower layers of skin with loss of most or all adenexal structures, wound closure is significantly impeded. Around 13,000 burns patients are admitted to UK hospitals annually3 and in severe cases tissue grafts are frequently required.4 A major drawback in the use of autografting is the limited availability of donor tissue in extreme cases where burns cover a large proportion of the body. Culturing donor grafts to generate additional tissue typically involves several weeks of culture and results in a variable patient graft-take rate of between 15-85%, with the average take in the order of only 50% for full-thickness burns.5

In order to overcome the need for lengthy cell culture and use of donor tissue for autografts in full-thickness wounds, allogenic cellular products have been developed with the intention of improving wound healing. These include products such as Apligraf6 and Orcel7 which comprise a layer of human keratinocytes overlying an animal derived extracellular matrix (ECM), and Dermagraft,8 which is similar but lacks an epithelial component. Use of Apligraf and Dermagraft is only recommended for persistent full-thickness ulcers that have not healed through standard treatment, while Orcel is FDA approved for use in burns injuries.6–8 However, these biological dressings are expensive and there is a risk of patient rejection of allogenic cells.9 More simple scaffolds have therefore emerged that utilise fabricated acellular structures, either as lyophilized extracellular matrix extracts (Oasis, Alloderm) or made using synthetic components such as nylon (Integra).10 Scaffolds have also been experimentally fabricated from ECM-derived polymers including collagen,11 fibrinogen12 and hyaluronic acid,13 as well as from synthetic polymers such as polycaprolactone (PCL)14 and polylactic-co-glycolic acid (PLGA).15 These polymers have also been developed as solid microspheres and aqueous hydrogels in addition to large sheet structures.16 While successes in healing wounds using both cellular and acellular scaffolds have been reported, many studies pay little microscopic attention to the early stages of the healing process and make most observations at the macroscopic level.17

A new approach of scaffolds combining polymers with gene silencing drugs offers the potential for bioactivation of scaffolds to deliver drugs that promote healing as well as provide mechanical support structures.18 A potential candidate antisense oligodeoxynucleotide (asODN) that may be useful for this purpose targets the gap junction protein connexin 43 (Cx43). The Cx43 protein is naturally down-regulated in wound edge keratinocytes and fibroblasts in the first 24-48 h after injury.19 This remains down-regulated whilst the cells are migrating and returns when the wound is closed. Application of Cx43 antisense oligonucleotides (Cx43asODN) to acute wounds has been shown to speed down-regulation and double the rate of healing,20 as well as dampen the inflammatory response.21 In the STZ diabetic rat, where healing is perturbed, Cx43 is turned on in the cells of the wound edge and migration fails.22 Application of the Cx43asODN to wounds on STZ diabetic rats was shown to prevent the abnormal turn on of Cx43 and rescue healing to normal rates or better.22

Here we examine the microscopic healing response in the presence of different forms of scaffold with particular focus on wound edge outgrowth and integration with the scaffold, as well as the surrounding connexin dynamics. We also evaluate bioactivation of scaffolds using Cx43asODN to detect possible improvements to the healing process.

2. Results

2.1. Fabrication of Collagen Scaffolds

To clarify whether electrospun collagen scaffolds had a porous structure, they were imaged using SEM (Figure 1). Scaffolds displayed a fibrous structure when imaged from above (Figure 1A) and the side (Figure 1B), and were found to be permeable to the drug delivery vector, Pluronic gel, containing FITC dye (Figure 1C). The scaffold collagen fiber diameters ranged from 0.03 μm – 4.12 μm.

Figure 1.

Electrospun collagen scaffolds SEM imaged from (A): above and (B): side-on. (C): Scaffold permeability was tested by infusion with FITC-labelled pluronic gel (green). (D): Scaffolds were applied as 6 mm circular discs to a series of 4 full-thickness wounds to the mouse dorsum. (E): Diagram to illustrate scaffold placement at the wound. (A-B): Scale bar = 10 μm; (C): Scale bar = 200 μm. Images shown are representative of triplicate observations of a minimum of n = 3 scaffolds. Diagram legend: E = epidermis, D = dermis, PC = panniculus carnosus, WE = wound edge.

2.2. Effect of Collagen Scaffolds on Wound Closure

Collagen scaffolds were applied directly to individual full-thickness wounds on the backs of mice (Figure 1D-E). Following qualitative macroscopic assessment of wounds at D3 and D5 (Figure 2), wounds treated with scaffolds were visibly larger than untreated which were contracting down and re-epithelialising. In the case of non-treated wounds, re-epithelialisation was almost complete by D5 in the majority of wounds (Figure 2). The presence of a scaffold appeared macroscopically to prevent wound contraction and re-epithelialisation.

Figure 2.

Collagen scaffolds were applied to full-thickness wounds and compared to wounds with no treatment (NT) or collagen scaffolds bioactivated with a Cx43asODN containing pluronic gel. Macroscopic images of wounds were taken at D3 and D5 post wounding. The dotted line outlines a visible leading edge of re-epithelialisation. Images are typical representations across all mice (n = 7), Scale bar = 1.5 mm.

Sister sections from the wound site were H&E stained or immunostained for connexins for microscopic analysis (Figure 3 & Figure 4). At D3 and D5 keratinocytes at the edge of wounds and even up to several hundred microns away formed thickened, non-migratory bulbs compared to the thin tongues of cells formed in the rapidly healing untreated wounds (Figure 3A). At D3 the average thickness of the nascent epidermis of scaffold-treated wounds was 106 ± 24 μm; (mean +/- SEM) a significant 73% increase from 33 ± 5 μm for untreated wounds (P < 0.05, Figure 3D). This effect was also observed at D5 by the 57% increase in epidermal thickness in scaffold treated wounds (P < 0.05, Figure 4D). Microscopically, there was no evidence of keratinocyte entry into or over the scaffolds. Keratinocyte and fibroblast migration instead appeared by D5 to traverse down the side and underneath scaffolds, in a manner similar to re-epithelialisation underneath a scab yet severely more retarded (Figure 4A). Scaffolds were also surrounded by large numbers of neutrophils at both D3 and D5, and the scaffold had been considerably degraded by D5.

Figure 3.

H&E and Cx43 staining of D3 mouse full-thickness wounds following treatment with uncoated of Cx43asODN infused collagen scaffolds. (A): Boxed areas of representative H&E stained wound edge sare magnified in high power confocal images of sister sections. Expression of Cx26 (B), Cx43 (C) and epithelial thickness (D) was quantified in keratinocytes at the wound edge. Dotted lines mark the outline of the epidermis. Scale bar= 100 μm, n = 7 samples per group, data is plotted as mean +SEM. Diagram legend: S = scaffold, WE = wound edge. For wound treatments CS = collagen scaffold, as-CS = Cx43asODN treated collagen scaffold and NT = no treatment (* = P < 0.005, ** = P < 0.001).

Figure 4.

H&E and Cx43 staining of D5 mouse full-thickness wounds following treatment with uncoated of Cx43asODN infused collagen scaffolds. (A): Boxed areas of representative H&E stained wound edge sare magnified in high power confocal images of sister sections. Expression of Cx26 (B), Cx43 (C) and epithelial thickness (D) was quantified in keratinocytes at the wound edge. Dotted lines mark the outline of the epidermis. Scale bar= 100 μm, n = 7 samples per group, data is plotted as mean +SEM. Diagram legend: S = scaffold, WE = wound edge. For wound treatments CS = collagen scaffold, as-CS = Cx43asODN treated collagen scaffold and NT = no treatment (* = P < 0.005).

2.3. Connexin Expression in the Epidermal Wound Edge

The scaffold induced a bulb like appearance in the epidermal wound edge keratinocytes which is reminiscent of that seen in STZ diabetic rat wounds which also exhibit elevated levels of Cx43 and Cx26.22 Here we found that, similarly, Cx43 and Cx26 expression in these wound edge bulbs was significantly increased when compared to the normal epidermis distal to the wound and also to the wound edges of non-treated wounds. Wound edge connexin levels were quantified as a percentage change from expression in the epidermis at a site distal to the wound and compared to that of non-scaffold treated wounds. Cx43 quantification revealed a 469% increase from distal levels at D3 (P < 0.01, Figure 3C) in scaffold treated wound edges and this was significantly higher than the 38% down-regulation in wound edge expression for non-treated wounds (P < 0.01, Figure 3C). Cx43 levels were also elevated at D5 as evidenced by a 457% increase at the scaffold-treated wound edge relative to a 10% up-regulation in non-treated wounds, which were restoring Cx43 levels to normal gradually as wounds closed (P < 0.05, Figure 4C). In acute wounds Cx26 expression naturally increases from very low baseline levels in the wound edge keratinocytes as they become migratory.20 However, we found Cx26 was up-regulated in wound edge keratinocytes almost 7,000% at D3 (P < 0.05, Figure 3B) relative to distal levels, significantly higher than the 2,326% up-regulation of non-treated wounds (P < 0.05, Figure 3B). This was also the case at D5; scaffold treated wounds had a 4,466% increase in wound edge keratinocyte Cx26 levels, significantly higher than the 167% detected in non-treated wounds which were down-regulating Cx26 to normal levels as re-epithelialisation completed (P < 0.05, Figure 4B).

2.4. Collagen Scaffold Bioactivation Using Cx43 asODN

We considered the possibility that scaffolds may be impairing the wound healing process by causing up-regulation of Cx43. We attempted to overcome this through the bioactivation of collagen scaffolds using Cx43asODN dissolved in Pluronic gel to prevent or reduce the elevation of Cx43 expression in wound edge keratinocytes induced by the scaffolds. Macroscopic evaluation of wounds treated with Cx43asODN bioactivated scaffolds at D3 and D5 showed that the scaffolds appeared to be integrating with the wound and were noticeably flush with the wound edge, compared with regular scaffolds which were generally not integrating well with the wound edge (Figure 2).

Tissue sections through these wounds were H&E stained or immunostained to reveal Cx43 or Cx26 expression. Bioactivation of the scaffold significantly reduced epidermal thickening at the wound edge by 60% at D3 (P < 0.01, Figure 3D) and by 52% at D5 (P < 0.05, Figure 4D) to a level that was not significantly different from untreated wounds. Although macroscopic analysis indicated improved integration of scaffolds with the wound edge, keratinocytes still migrated down the side of scaffolds rather than over them. Wounds that received bioactivated scaffolds had a 91% reduction in wound edge keratinocyte Cx43 protein levels relative to control scaffolds at D3 (P < 0.05, Figure 3C) and this level was not significantly different from untreated wounds at D3. However, by D5 there was no difference from regular scaffolds (Figure 4C). Also evident was a significant reduction in Cx26 levels at the wound edge of bioactive scaffold treated wounds relative to control scaffolds. Bioactivation with Cx43asODN resulted in a 68% decrease in wound edge keratinocyte Cx26 at D3 (P < 0.01, Figure 3B) and an 87% knockdown at D5 (P < 0.05, Figure 4B), to levels insignificantly different from non-treated wounds.

2.5. Altering Scaffold Properties

Because of differences in reported scaffold fabrication and cross-linking methods we investigated whether epithelial wound edge thickening and increased Cx43 expression relates to the thickness of the collagen scaffolds or the concentration of the cross-linker used. Collagen scaffolds were fabricated at a reduced thickness (0.4 mm) and compared to 0.8 mm thick scaffolds, and cross-linked using the original concentration or a 100-fold diluted concentration of EDC cross-linker (15% or 0.15% wt./v., respectively). Wounds received one type of scaffold and were harvested at D1 to observe any early effect on the initial healing process (Figure 5). With all scaffold thickness and cross-linker conditions, marked thickening of the wound edge epidermis was observed and was accompanied by a visible Cx43 up-regulation relative to distal levels.

Figure 5.

Cx43 staining of D1 full-thickness wound edges following treatment with varying physical properties of collagen scaffold. (A): Diagram illustrating site of confocal image capture. (B): High power single optical section confocal images of the boxed area in (A) of the leading edge epidermis. Scaffolds properties were varied (either 15 mM or 0.15 mM EDC used to cross-link, and scaffolds spun to a thickness of either 0.4 mm–termed ‘thin’ or 0.9 mm–termed ‘thick’). Dotted lines mark the outline of the epidermis. Scale bar = 100 μm, images are typical observations of n = 5 samples per group. Diagram legend: E = epidermis, D = dermis, PC = panniculus carnosus, WE = wound edge.

2.6. The Effect of Alginate Fabrications on Healing

In order to understand which properties of scaffolds may be causing adverse effects on re-epithelialisation and Cx43 expression, we applied three different alginate fabrications (woven scaffolds, microspheres and fluid hydrogels) to full-thickness wounds to assess the healing response to a different reagent and other scaffold structures (Figure 6A). Wounds were harvested at D1 to assess the effect of the different forms of alginate on the wound edge keratinocytes during the early healing response. Solid alginate microspheres and alginate dressing treated wounds consistently exhibited thickened bulbs of keratinocytes at the wound edge containing up-regulated Cx43 (Figure 6B). However, these effects were not observed with the alginate hydrogel where keratinocytes behaved as they do in normal acute wounds, elongating as a thin tongue and down-regulating Cx43.

Figure 6.

Alginate application to D1 full-thickness wounds. (A): Diagram to illustrate the application method of one of three alginate fabrications to wounds. One type of scaffold was applied to each individual wound. (B): Cx43 fluorescence staining of full-thickness wound edges. High-power confocal single section images. Images focus on the boxed area of the wound edge shown in (A). Dotted lines mark the outline of the epidermis. Scale bar = 100 μm, n = 5 samples per group. Diagram legend: E = epidermis, D = dermis, PC = panniculus carnosus, WE = wound edge.

2.7. Cellular Response to Dressings

In order to better understand the impact of scaffolds and dressings on wounds, H&E stained sections were examined at high power to investigate the early inflammatory response at D1 (Figure 7). This revealed numerous leukocytes with a polymorphonuclear leukocyte neutrophil-like morphology surrounding all solid structures including the collagen scaffold, alginate microspheres and the alginate dressing (Figure 7A). The effects were reduced in alginate hydrogel treated wounds and neutrophils were mostly restricted to beneath the dermis in the panniculus carnosus. Numbers of neutrophils in the wound edge dermis were also significantly higher following application of scaffolds or dressings (Figure 7B). At D1 in the dermis at 3 fields of view distal from the wound edge there was a significant 3.5-fold increase in numbers of invading neutrophils for wounds treated with collagen scaffolds relative to untreated wounds (P < 0.01, Figure 7B). Similarly the alginate dressing induced a 3.8-fold increase (P < 0.01) and the alginate microspheres a 5.2-fold increase (P < 0.001) in the number of invading neutrophils. The number of neutrophils present in this region of the dermis of alginate hydrogel treated wounds was not significantly different from control wounds. These findings overall suggest that the presence of solid structures in wounds increases the inflammatory response. Collagen scaffolds induced a protracted inflammatory response (Figure 7C) as evidenced by a strong 10-fold increase in dermal neutrophils at D3 (P < 0.001) and an 11.1-fold increase at D5 (P < 0.001). However, bioactivation of collagen scaffolds with Cx43asODN was found to significantly reduce the number of neutrophils in the dermis at D3 (P < 0.001) and D5 (P < 0.001) after wounding to a level statistically insignificant from control wounds. Cx43 may be involved in the pro-inflammatory process induced by these solid scaffolds and dressings.

Figure 7.

Histological assessment of cells infiltrating scaffolds and dressings. (A): D1 H&E images illustrating the clustering of round cells within and surrounding polymer applications. (B): High-power images focusing on a region of dermis 3 fields of view from the wound edge at the position indicated by the diagram at D1 following polymer application (top-left). Subsequent quantification of the number of leukocytes is also displayed (top-right). Cells were observed at high-power to identify polymorphonuclear leukocytes prior to counting (example shown at top-left). (C): High-power images of the dermis of D3 and D5 treated wounds captured identically to those in (B). Leukocyte counts at both time points were also quantified (bottom-right). (A): Scale bar = 100 μm. (B): Polymorphonuclear leukocyte image scale bar = 10 μm, all other images = 50 μm. n = 5 (D1) or n = 6 (D3 & D5) samples were used, data is plotted as mean ± SEM. Diagram legend: E = epidermis, D = dermis, PC = panniculus carnosus, WE = wound edge, PA = polymer application. For wound treatments NT = no treatment, CS = collagen scaffold, AD = alginate dressing, AM = alginate microspheres, AH = alginate hydrogel and as-CS = Cx43asODN treated collagen scaffold (** = P < 0.01, *** = P < 0.001).

3. Discussion

In this study we set out to examine the in vivo cellular response to the presence of a variety of scaffolds and dressings in the wound bed and to assess whether bioactivation could improve aspects of scaffold incorporation. In the rodent skin wound healing takes place through a combination of wound contraction and the active migration of keratinocytes and fibroblasts into the wound bed. Application of collagen scaffolds to a full thickness acute wound inhibited healing and induced the formation of a bulb of non-migratory keratinocytes in which the gap junction proteins Cx43 and Cx26 were significantly upregulated. This response was reminiscent of the wound healing response in STZ diabetic rats which also form a thickened bulb of non-migratory keratinocytes with high levels of Cx43.22

Cx43 is normally downregulated in the wound edge keratinocytes and fibroblasts in the first 24 h of acute healing.19, 23, 24 The abnormal expression of Cx43 has previously been reported in diabetic rat wounds and was found to underlie a failure of these cells to migrate forward and heal the wound. When this abnormal Cx43 expression was selectively targeted with Cx43asODN it restored healing to a normal rate.22, 25 It is likely that the scaffold induced upregulation of Cx43 in wound edge keratinocytes and inhibited their migration and may therefore induce many of the features of a chronic or impaired-healing wound. It has been shown that elevation of Cx43 expression in fibroblasts attenuates their migration by inhibiting their cytoskeletal dynamics.24 Conversely, downregulation of Cx43 in fibroblasts reduces cell-cell adhesion and activates the small GTPases Rac1 and RhoA, enhancing cytoskeletal dynamics and promoting migration.25 This may be in part due to the interaction of the Cx43 cytoplasmic tail with the adhesion molecules zonular occludens 1 & 2, and components of the cytoskeleton.26

While Cx43 is downregulated in wound edge keratinocytes, Cx26 is naturally upregulated in those same cells and is maintained at high levels during the process of cell migration.19 However, we found that in the presence of scaffolds Cx26 is expressed at far greater levels than achieved during normal wound healing. As Cx26 is normally expressed during keratinocyte migration we would not expect it to interfere with the migration process, especially as it has a short cytoplasmic tail that is unable to interact with other cell structural proteins in the way that Cx43 does. Conversely, enhanced levels of Cx26 have been associated with a hyper-proliferative epidermis as well as barrier dysfunction and inflamed conditions.27–29 It has been suggested that high levels of Cx26 result in leakage of ATP, which in turn promotes inflammation,27 a feature we have observed in scaffold containing wounds which have very high levels of neutrophils in the wound bed and surrounding dermis.

It was not immediately apparent as to why collagen spun scaffolds induced this perturbed healing state. We explored the possibility that it might relate to the thickness of the scaffold or the concentration of the cross-linker used in its production. However, altering the scaffold thickness and reducing cross-linker concentration by 100-fold had no effect on the wound edge reaction and all permutations produced very similar epidermal wound edge thickening and inflammation. We also investigated the effects on healing of different materials in diverse forms inserted into the wound bed. We chose to investigate alginates, which are commercially available as a fibrous wound dressing or hydrogel and can be manufactured in the form of microspheres. We found that the alginates in the form of either a fibrous dressing or microspheres induced an apparent foreign body reaction of enhanced inflammation and formation of a bulb of non-migratory keratinocytes at the wound edge. However, this was not the case for the alginate hydrogel, which did not induce an elevated inflammatory response or inhibit re-epithelialisation. Thus, it appears that the negative response observed in the presence of scaffolds is due to the presence of a physical foreign body in the wound bed rather than a specific substance inducing the reaction. Indeed these physical structures appear to induce a concerted invasion of neutrophils, a feature previously found by others in response to scaffolds placed subcutaneously.30 These neutrophils quite rapidly destroy the scaffolds and also spread beyond the wound bed into the surrounding dermis where there is a danger that they may cause more damage than benefit.31 Despite considerable degradation of scaffolds by D5 there was still a substantial presence of inflammatory cells and continued repression of keratinocyte migration and overexpression of Cx43 and Cx26 within the bulb. It is possible that the enhanced connexin expression relates to the elevated inflammatory response present in these wounds. Compared to normal acute wounds, there was very little outgrowth of keratinocytes and fibroblasts or integration of these cells with any of the scaffolds. Macroscopically there appeared to be very little integration of the collagen scaffolds with the surrounding tissues at D3 with this just beginning by D5 when most acute control wounds had already re-epithelialised. Chronic ulcers can also contain high numbers of inflammatory cells prior to treatment, so it may be questioned whether placement of scaffolds that can promote further neutrophil invasion is most beneficial to healing.32 Regardless of the effect on neutrophils, no fibroblast-like cells could be seen to associate with solid collagen or alginate scaffolds in this study which calls into question the purported role of scaffolds as a supportive matrix for fibroblast attachment and proliferation.1

Application of collagen scaffolds has been previously associated with a decrease in rodent wound contraction but little is understood of the mechanisms underlying the perturbed wound re-epithelialisation.33–36 It is known that over-expression of Cx43 can inhibit migration of keratinocytes and fibroblasts in wound healing models in vivo and in vitro and that down-regulating Cx43 protein expression in vivo can accelerate wound healing whilst reducing the inflammatory response.20, 22, 24, 25, 37 We therefore attempted to overcome the negative effects of the collagen scaffolds by bioactivating them with Cx43asODN. We found that Cx43asODN bioactivated collagen scaffolds integrated considerably better with the surrounding wound edge tissues at both D3 and D5. Microscopic evaluation of H&E stained wounds with bioactivated scaffolds showed reduced levels of inflammation and neutrophils in wound edge dermis on both D3 and D5, similar to the effect induced by a SDF-1 bioactivated collagen-glycosaminoglycan scaffold in a separate study.38 This is consistent with the anti-inflammatory effects of Cx43asODN on acute wound healing37 but it also implies that the treatment can repress the inflammatory response in a pro-inflammatory situation. Due to the similarities in features of the scaffold wounds to impaired-healing wounds, this treatment may be of therapeutic benefit in the treatment of chronic and other slow-to-heal wounds, including burns, with their sustained inflammatory background or indeed other conditions with chronic inflammation.

Bioactivation of the collagen scaffold repressed the abnormal upregulation of Cx43 in the wound edge keratinocytes which consequently lost much of their bulbous appearance and could also be seen actively migrating between the wound edge dermis and the scaffold at D3 and D5. In the vast majority of cases keratinocytes migrated between the wound edge and scaffold and underneath the scaffold as if it were a scab. At D3 in the wound edge epidermis Cx43 was found to be significantly reduced in the presence of bioactivated scaffolds compared to control scaffolds. Interestingly Cx26 levels were also found to be significantly reduced with the bioactivated scaffolds on D3. The reduction in levels of Cx26, in addition to Cx43, is not likely to be from a direct effect of the antisense as this is specific for Cx43 and does not directly affect Cx26. It is also very unlikely that in the current formulation there would be any active antisense remaining after the first 24 h as this is an unmodified ODN which is rapidly degraded within minutes in the presence of sera and has a short intracellular half-life.39 Cx43asODN appears to reduce inflammation and also reduce any subsequent scaffold-induced elevations in the levels of both Cx43 and Cx26. This is supported by the fact that Cx43 and Cx26 have been reported as elevated in a number of different conditions associated with increased inflammation with injury or disease.21, 27, 29 By D5 the effects of the antisense from the bioactivated scaffold had diminished and in the continued presence of the scaffolds re-epithelialisation was still attenuated compared to non-scaffold control wounds. It was however still significantly ahead of the control scaffold wounds that had not begun the re-epithelialisation process. Clearly there is scope to improve the bioactivation in the future with either longer lasting asODNs or a more long term sustained release of the ODN from the scaffolds.

4. Conclusions

Whilst acellular scaffolds are intended to encourage tissue repair and provide a surface for wound edge cells to crawl into or over and form new tissue it appears that their physical presence can also provoke a ‘foreign body’ response. In the case of collagen, this induces an enhanced inflammatory response with neutrophils destroying the scaffolds and invading the surrounding dermis. The inflamed condition appears to elevate Cx43 and Cx26 levels in the wound edge epidermis that in turn inhibits outgrowth and healing in the presence of the scaffold. Bioactivating the scaffolds with Cx43asODNs can repress the inflammatory response and reduces the Cx43 and Cx26 elevation in wound edge keratinocytes thus improving the healing process and macroscopic scaffold integration. However, wound edge cells still seem to be reluctant to grow into any of the scaffolds we tested and instead grew underneath them as if they were scabs. Future improvements in bioactivation of the scaffolds with Cx43asODNs have scope to improve the therapeutic action of the scaffolds in chronic wounds.

5. Experimental Details

Collagen Scaffold Fabrication: Collagen scaffolds were fabricated by electrospinning acid-soluble bovine collagen (>99% purity, Kensey Nash, PA, USA) blended with poly-ϵ-caprolactone (>99% purity, Sigma, Poole, UK) at a 10:1 weight ratio, respectively. The collagen used had been previously processed by the manufacturer to remove antigenic components.40 Blended polymers were dissolved in Hexafluoropropan-2-ol (‘HFP’, 99% purity, Apollo Scientific, Stockport, UK) at 10% (wt./v.). The electrospinning process has been described previously.41 Briefly, scaffolds were electrospun at a flow rate of 5 ml/h with a needle to collector distance of 130 mm and using an output voltage of 13 kV. 14 ml of polymer solution was spun onto a 9 cm2 foil sheet to produce scaffolds with an average thickness of 0.4 mm. To generate scaffolds with an average thickness of 0.8 mm, 30 ml of polymer solution was electrospun. Scaffolds have previously been cross-linked using different concentrations of N-(3-Dimethylaminopropyl)-N’-ethylcarbodiimide (EDC) hydrochloride.7, 21 Scaffolds in this study were cross-linked by immersion in either a high (15% wt./v.) or a low (0.15% wt./v.) concentration of EDC hydrochloride (99.5% purity, Apollo) in a 1:10 water to acetone solution. Scaffolds were washed for 20 min with sterile Phosphate Buffered Saline (PBS), sterilised for 1 h using 70% ethanol, and washed 3 times with sterile PBS. Scaffolds were punched out using a 6 mm biopsy punch to match the size and shape of the mouse excision punch biopsy wounds.

Alginate Microsphere Fabrication: Alginate microspheres were produced by electrospraying medium 2% (wt./v.) viscosity alginic acid powder PBS solution (Sigma, Poole, UK). The electrospraying process has been described previously.42 Briefly, the polymer was electrosprayed at a distance of 10 cm from the collector, using a voltage of 8 kV and a flow rate of 10 ml/h. Microspheres were sprayed into a gelling bath containing a 200 μM concentration of calcium chloride dehydrate solution (≥98% purity, Sigma, Poole, UK) to become crosslinked. The average sphere diameter generated using this method was measured as 685 μm.

Scaffold Bioactivation: Murine Cx43 antisense oligodeoxynucleotides (Cx43asODN), sequence 5′-GTA ATT GCG GCA GGA GGA ATT GTT TCT GTC-3′ (Sigma, Poole, UK) were dissolved in molecular grade water to make a 1 mM stock. They were then diluted into a 30% (wt./v.) Pluronic gel solution (Pluronic F-127 powder, Sigma, Poole, UK) to a 30 μM final concentration. Pluronic gel is liquid below 4 °C and collagen scaffolds were immersed in 20 μl of either Cx43asODN Pluronic gel or PBS in the fridge overnight. Alginate microspheres were soaked in a volume of Cx43asODN Pluronic gel equal to the volume of polymer used in electrospraying. When applied to wounds the gel warmed and set in place facilitating a sustained release of asODN to the wound bed over a number of h.

Animals: 6-week old male ICR mice (Harlan Laboratories, UK) were used with UK Home Office approval and according to their regulations. Anaesthesia was induced using 4% isoflurane, 2 L/min oxygen and 1 L/min nitrous oxide and maintained using 1.5% isoflurane. Animals were injected subcutaneously with 0.03 mg/kg buprenorphine (Vetergesic, Recknitt Benckiser Healthcare, Hull, UK) before operation. Mouse backs were shaved then covered in a thin layer of Nair hair removal cream. Nair and hair was then removed from the animal using a warm moist gauze. Mice were placed on a heated mat during the operation. Four full-thickness excisional biopsy punch wounds were made using a 6 mm medical biopsy punch (Kai Industries, UK) to mouse backs, two on each side of the dorsal midline. Collagen scaffolds, commercial products (either an ActivHeal brand alginate dressing or hydrogel; Advanced Medical Solutions, Winsford, UK) or 20 μl Pluronic gel containing microspheres were applied directly to wounds, and covered with Tegaderm film (3M, UK). One wound on each mouse was always left untreated as a control. Post-operation, mice were kept warm in a heated incubator and monitored whilst recovering before returning animals to cages.

Tissue Histology: Animals were culled at either 1, 3 or 5 days post wounding. Individual wounds were excised, imaged and fixed by immersion in a 4% paraformaldehyde-PBS solution for 24 h. Samples were washed 3 × 1 h in PBS then transferred to 20% sucrose-PBS solution for 24 h. Wounds were bisected and one half was snap frozen in TissueTek O.C.T. medium (Sakura Finetek, UK), then cryosectioned at both 5 μm and 10 μm thicknesses. The 5 μm thick tissue cryosections were stained with haematoxylin and eosin43 whilst 10 μm sections were used for immunostaining. Polymorphonuclear leukocytes were identified through high power imaging of nuclei and counted in a 300 μm2 region at 3 fields of view away from the wound edge in the same position of the dermis in each sample.

Immunostaining: 10 μm thick tissue cryosections were permeabilised in cold acetone for 5 min. Tissue was blocked against non-specific antigen binding for 30 min using a 0.1 M lysine (Sigma, Poole, UK) PBS solution containing 0.1% Triton-X-100 (Sigma). Tissue was stained using polyclonal rabbit antibodies to connexin 43 (Sigma C6219, 1:2000 dilution), or connexin 26 (1:200 dilution,44) for 1 h at room temperature, followed by 3 × 5 min washes using blocking solution. Goat anti-rabbit Alexa 488 secondary antibody (Invitrogen A11008, UK) was applied at a 1:400 dilution for 1 h at room temperature followed by a 5 min counterstain using Hoechst solution (Hoechst 33258 and Hoechst 33342 dyes each diluted to 1:50,000 in PBS, Sigma, Poole, UK). Tissue sections were washed 2 × 5 min in PBS and then mounted and cover-slipped using Citifluor AF1 mountant (Citifluor Ltd., UK).

SEM and Macroscopic Imaging: Samples of collagen scaffolds were air dried, snapped, and mounted onto aluminium stubs. Samples were tungsten sputter coated then imaged using a JEOL JSM 7401F electron microscope system set at 2.0kV. Images were acquired at 800X magnification. Each mouse wound was imaged macroscopically using a Leica MZ8 dissection microscope at 0.63X magnification. Scaffold and microsphere porosity was assayed by placing samples in either a 5% Fluorescein isothiocyanate (FITC) or Cy3 Pluronic gel overnight, respectively. Fluorescently labeled microspheres were imaged using the dissection microscope and fluorescent scaffolds using a Leica DMLB fluorescence microscope.

Confocal Imaging: Single optical section images of Cx43 or Cx26 immunostained tissue sections were acquired using a Leica SP2 confocal microscope using a 40X 1.2NA oil objective. Fluorophores were excited sequentially (Hoechst using both 351 and 364 nm wavelength lasers, Alexa 488 using a 488 nm wavelength laser). 8-bit greyscale images were acquired using a frame average of 4 at 1024 × 1024 pixels. Image acquisition settings remained identical between images in order to allow comparisons in image analysis.

Data Analysis: Expression of connexin protein was quantified by counting connexin positive pixels on identically thresholded binary images of wound edges using ImageJ software (for method see22). Only the number of connexin positive pixels contained within the end 100 μm of nascent epidermis were counted to ensure only the leading edge was quantified. Connexin pixels were counted per micron squared of epidermis in order to account for differences in epidermal thickness, and connexin levels were normalised to those of a distal site away from the wound edge to factor in differences in innate connexin levels between animals. Epidermal thickness was determined by measuring the thickest point from the basal layer to spinous layer within the end 100 μm of nascent epidermis using ImageJ. Both connexin pixel quantification and leukocyte count data was subjected to Kolmogorov-Smirnov tests to ascertain normality, followed by one-way ANOVA and Tukey's post-hoc tests to confer statistical significance.

Acknowledgements

We would like to thank Prashant Ghadjahur, an MSc student for his contribution to the microsphere part of this project. DJG is a BBSRC-CASE student supported by CoDa Therapeutics Inc. and DLB is a founding scientist of CoDa Theraputics Inc. and holds shares in the company. AP is an employee of CoDa Therapeutics Inc.