Neutralising antibody to TGF-β_1,2 reduces cutaneous scarring in adult rodents

Scarring is a major cause of many clinical problems. Post-burn contractures, post-operative adhesions and strictures causing intestinal obstruction, mid-facial contractures following cleft palate surgery and painful neuromas are a few examples of the problems caused by scarring. Scar tissue interferes with growth, causes deformities, impairs function and is aesthetically unsightly. Failure to regenerate tissue following iatrogenic or accidental trauma results in the formation of scar tissue. However scarring does not appear to be a problem with embryonic and early fetal wounds, which heal without scar tissue formation (Whitby and Ferguson, 1991a).

Fetal wound healing differs from that in the adult by a number of parameters such as rapid re-epithelialisation, lower and different inflammatory response, different growth factor response, less angiogenesis, different rates of extracellular matrix deposition and restoration of the architecture of the involved tissue (Whitby and Ferguson, 1991a). The mechanism of tissue regeneration and scarless healing is, however, not yet understood. The warm and sterile fetal environment, excess and prolonged hyaluronan deposition and intrinsic tissue differences have all been put forward as possible explanations of scarless healing in the fetus (Adzick and Longaker, 1992).

Growth factors are known to play a prominent role in wound healing (McGrath, 1990). The growth factor profile of fetal wounds differs from that of the adult. Using immunocyto-chemical techniques, Whitby and Ferguson (1991b)failed to immunolocalise transforming growth factor-β (TGF-β) or basic fibroblast growth factor (bFGF) at fetal wound sites whilst they were present in adult wounds. Exogenous addition of TGF-β to fetal wounds induces scarring (Krummel et al., 1988).

TGF-βs are members of a superfamily. Three isoforms of TGF-β (1, 2 and 3) have been identified in mammals. The three isoforms have distinct 5′ flanking sequences that regulate their expression; however, the activated mature peptides are 60-80% homologous in their amino acid sequences. Many cultured cells respond equally to the different isoforms but several differences in their biological potencies and actions have been reported. TGF-β₃is 10 times more active than TGF-β₂in inducing mesoderm formation in Xenopus laeviswhile TGF-β₁is inactive (Roberts et al., 1990); TGF-β₁and TGF-β₃readily inhibit the growth of endothelial cells in monolayers whilst substantially greater concentrations of TGF-β₂are required for the same effect (Merwin et al., 1991a,b).

Transforming growth factor-β₁is chemotactic to macrophages and fibroblasts, induces angiogenesis and granulation tissue (Roberts et al., 1986), stimulates extracellular matrix synthesis and decreases extracellular matrix degradation. Thus TGF-β plays a central role in wound healing.

TGF-β has also been implicated in various fibrotic diseases such as scleroderma (Gruschwitz et al., 1990), glomerulonephritis (Border et al., 1990), pulmonary fibrosis (Broekelmann et al., 1991), liver cirrhosis (Castilla et al., 1991), proliferative vitreoretinopathy (Connor et al., 1989) and post-operative peritoneal adhesions (Williams et al., 1992).

We have used neutralising antibody to TGF-β_1,2to alter the growth factor profile of dermal wounds in adult rodents and studied the effect on scar tissue formation. This paper extends our preliminary report that exogenous addition of neutralising antibody to TGF-β reduces scarring and restores the dermal architecture of healing incisional wounds (Shah et al., 1992). Here, we describe in detail, the effects of exogenous addition of neutralising antibody to TGF-β on healing incisional wounds in adult rodents.

Adult, male Sprague-Dawley rats (Charles River UK Ltd, Kent, UK) weighing 225-250 g were anaesthetised by halothane, nitrous oxide and oxygen inhalation. Four full-thickness, linear incisions, 1 cm in length, down to and including the panniculus carnosus were made on the dorsal skin of the animal. The incisions were placed equidistant from the midline and adjacent to the four limbs (Fig. 1). One of the wounds (control) was unmanipulated; the second wound (sham-control) was injected with an irrelevant antibody (rabbit immunoglobulin IgG, Serotec Ltd Oxford, UK); the third wound (positive control) was injected with transforming growth factor-β₁purified from porcine platelets (BDA 3; British Biotechnology Ltd Oxford, UK) and reconstituted in phosphate buffered saline (PBS) containing 1 mg/ml bovine serum albumin (Serotec Ltd Oxford, UK) and 4 mM HCl (hydrochloric acid) and the fourth wound (experimental) was injected with the neutralising antibody to TGF-β (IgG raised in rabbit against porcine platelet TGF-β₁): 10 μg antibody neutralises 0.25 ng TGF-β_1,2(BDA 1; British Biotechnology Ltd, Oxford, UK). Each injection was made up to 100 μl with PBS. All treatments were administered by infiltrating along both wound margins using an insulin syringe with the needle inserted through a single entry point, 0.5 cm distal to the caudal end of each wound (inset in Fig. 1). For experiments where treatments were administered on day 0, the wound sites were infiltrated intra-dermally just prior to wounding. The actual sites of the experimental treatments were rotated between the four wounds to correct for the known anterior-posterior differences in the healing of rodent wounds (Auerbach and Auerbach, 1982). The wounds were left unsutured to heal by secondary intention. Animals were allowed to recover and housed in individual cages and fed normal rat chow and water ad libitum. At various times after wounding, the animals were killed by chloroform overdose and the wounds harvested and processed as described below. All procedures were carried out according to Home Office regulations and under appropriate licences.

To determine how long the neutralising antibody to TGF-β remains in and around the wound site, wounds were treated with a single injection of 50 μg of the neutralising antibody at the time of wounding. Sham control wounds were treated with 50 μg of irrelevant rabbit IgG. The wounds were harvested 15 minutes, 30 minutes, 6 hours, 12 hours, 18 hours, 24 hours, 36 hours and 48 hours post-wounding. All wounds were frozen and processed for immunohisto-chemistry as described later. A total of 12 animals were studied in this experiment.

To determine the optimum dose of neutralising antibody to TGF-β, wounds were injected daily for 7 days starting on the day of wounding and harvested 10 days post-wounding. Experimental wounds were injected with 10 μg/injection, 50 μg/injection or 250 μg/injection of the neutralising antibody to TGF-β. Sham control wounds were injected with an equivalent amount of irrelevant rabbit IgG and the positive control wounds were injected with 2 ng/injection, 10 ng/injection or 50 ng/injection of TGF-β₁. After harvesting, the wounds were bissected and processed for immunohistochemistry and routine histology as described later; 24 wounds from 6 animals were analysed in this experiment.

To determine the optimum time and duration of treatment, animals were divided into several groups. In order to reduce the levels of TGF-β during the early phases of wound healing, we treated the wounds for 3 or 7 days. The first group of animals (group I) was treated daily for three days, starting on the day of wounding (day 0). The second group of animals (group II) was treated daily for seven days starting on day 0. The third group of animals (group III) was treated daily for seven days starting on the day after wounding (day 1). Previous studies by Cromack et al. (1987)and Grotendorst et al. (1988)demon-strated that TGF-β levels in wound fluid from wound chambers implanted subdermally into rats, peaked around days 7 and 20 post-implantation. They reported that the peak levels of TGF-β occurred during the fibroblast proliferation, collagen synthesis and remodelling phases of wound healing. Hence we attempted to reduce these levels by treating the wounds on days 7, 8 and 9 (group IV) and days 19, 20 and 21 (group V) post-wounding (Table 1). Wounds from all groups were harvested on various days post-wounding as shown in Table 1. The wounds were bissected; one half was processed for immunohistochemistry and routine histology and the other half was processed for biochemical analyses as described later; 20 animals were studied in this experiment.

After completing and analysing the kinetics, dose response and duration of treatment experiments, for detailed investigation of the effects of neutralising antibody to TGF-β, the experimental wounds were injected with 50 μg/injection of the antibody for three days, starting on day 0. Sham control wounds were injected with 50 μg/injection of rabbit IgG and the positive control wounds were injected with 10 ng/injection of TGF-β₁for three days starting on day 0. Wounds were harvested on days 7, 14, 28, 42, 70 and 168 post-wounding. After harvesting, the wounds were bissected; one half was processed for immunohistochemistry and routine histology and the other half was either subjected to tensiometry or biochemical analyses; 144 wounds from 36 animals were analysed.

The wounds were embedded in OCT compound (Miles Inc., Elkhart, IN) and frozen in liquid nitrogen. Serial sections, 7 μm in thickness, were cut onto poly-L-lysine (Sigma Chemical Co., Dorset, UK)-coated slides and stored at −20°C until they were used for staining. For indirect immunofluorescence without amplification, sections of wounds were incubated with the primary antibody for an hour at room temperature followed by three, 5 minute washes with PBS. The sections were then incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody for an hour at room temperature. After a further three, 5 minute washes with PBS, the sections were mounted in Gelvatol (Rodriguez and Deinhardt, 1960).

The biotin-streptavidin amplification technique involved incubating the sections with the primary antibody for an hour at room temperature followed by three, 5 minute washes with PBS. The sections were next incubated with biotinylated secondary antibody for an hour at room temperature; washed thrice with PBS and then incubated with FITC-conjugated streptavidin for 20 minutes at room temperature. This was followed by three, 5 minute washes with PBS and the sections were then mounted in Gelvatol (Tables 2and 3). The stained sections were visualised by epifluorescent microscopy using a Leitz Dialux microscope. Photography used identical settings and either T-Max 100 film for black and white photos or Kodak 200 ASA film for colour photography.

The macrophage profile was determined by the monoclonal antibody ED2 (Serotec, Oxford, UK) which recognises a surface antigen on mature resident macrophages in the rat (Dijkstra et al., 1985). Resident macrophages provide a first line of defence (Gordon et al., 1988) and this is taken over by blood-borne macrophages and monocytes. The monocyte and macrophage profile was determined by the monoclonal antibody ED1 (Serotec, Oxford, UK), which recognises mainly cytoplasmic and some surface antigens on most monocytes and macrophages in the rat (Dijkstra et al., 1985) and appears to be related to phagocytosis.

At least 3 sections (140 μm from each other in the wound), were incubated with ED2 antibody and at least 3 additional sections were incubated with ED1 antibody, from each wound harvested on days 7 and 14 post-wounding. With the aid of an epifluorescent microscope and a Magiscan image analyser (Joyce Loeble, Gateshead, UK), the number of macrophages (ED2) and macrophages and monocytes (ED1) present within a defined area in the subepidermal and deep dermal regions of the wounds and adjacent normal skin were counted to give the macrophage and monocyte profiles of healing wounds. The defined areas in the wounds were 0.162 mm²in area and selected such that they did not transgress the wound margins. The selected area in the subepidermal region of the wound was immediately subjacent to the epidermis and that in the deep dermal region was immediately above the level of the panniculus carnosus muscle; all areas were in the midline of the wound. In the normal dermis, similar areas were selected for measurements (Fig. 2).

To study the angiogenic response of healing wounds, at least 3 sections (140 μm from each other in the wound), from each wound harvested on days 7 and 14 post-wounding, were stained for laminin and a further 3 sections for von Willebrand factor to mark the blood vessels. Using epifluorescent microscopy, the sections were analysed and scored on a scale of −4 to +4 (vascularity: 0= same as normal dermis; 1= 1-25% of normal dermis; 2= 26-50% of normal dermis; 3= 51-75% of normal dermis; 4= >75% of normal dermis; prefix +, more than and prefix −, less than).

At least 3 sections (140 μm from each other in the wound) from each wound harvested on days 7 and 14 post-wounding were stained for fibronectin. Using epifluorescent microscopy, the sections were scored for the staining intensity on a scale of −4 to +4 (staining intensity: 0= same as normal dermis; 1= 1-25% of normal; 2= 26-50% of normal; 3= 51-75% of normal; 4= >75% of normal; prefix +, more than and prefix −, less than).

At least 3 sections from each wound were stained for collagen I and collagen III after pre-incubation with sheep testicular hyaluronidase type III (Sigma Chemical Co., Dorset, UK). Using epifluorescent microscopy, the sections were scored for the staining intensity using a scale of −4 to +4 as described for fibronectin.

Wounds were fixed in 4% paraformaldehyde overnight and processed for paraffin embedding, sectioned at 7 μm thickness and stained with Masson’s trichrome to highlight the connective tissue and Picrosirius red to enhance polarisation of collagen fibres (Junquiera et al., 1979).

The extent to which the dermal architecture of healing wounds was different from that of normal unwounded dermis, was assessed from sections stained with Masson’s trichrome or Picrosirius red stain (examined using a polarising microscope (Zeiss, Germany)). The following criteria were considered for scoring sections of wounds:

1. differences in the collagen bundle size;

2. compactness/spacing between the collagen bundles/fibers;

3. orientation of the collagen bundles/fibers compared to those of the normal dermis; and

4. organisation of collagen into a normal basket-weave pattern. At least 6 sections (140 μm apart from each other in the wound) from each wound were assessed and scored on a scale of 0-3 (0, normal dermal architecture; 1, better than scar in control wound; 2, equivalent to scar in control wound; 3, worse than scar in control wound).

All materials were of analytical grade where appropriate and purchased from BDH Chemicals Ltd, Poole, Dorset or from Sigma Chemical Co. Poole, Dorset unless otherwise stated.

With the aid of an operating microscope (Leitz), the individual wounds were microdissected free from the underlying muscle, fat and fascia and excised flush with the wound margins. A sample of normal skin from each animal was also similarly dissected. The tissue samples were freeze dried and stored at −70°C till further use. At the time of analyses, the tissue samples were weighed (dry weights), digested with papain for 24 hours at room temperature and centrifuged (10,000 g). Hydroxyproline content of the protein pellet was determined using the method of Stegmann and Stalder (1967)and the total collagen content was calculated by assuming that it contains 13.6% hydroxyproline. The glycosaminoglycans were precipitated from the supernatant with 2% cetyl pyridinium chloride. Total glycosaminoglycan content was determined using the method of Bartold and Page (1985).

The quantitative data of monocyte and macrophage profiles were subjected to multivariate analysis of variance using the SPSS programme. The multivariate analysis of variance was used to compare differences between treatments and to study the interactions between time, treatment and depth (subepidermal/deep dermal). As no differences were found between the profiles of control and sham control wounds, these were pooled together and compared with the profiles of antibody-treated wounds or TGF-β-treated wounds. As all comparisons were not orthogonal, the significance levels were adjusted by a factor of 0.75 using the procedure outlined by Girden (1992). Hence a nominal 0.0375 significance level was used in place of a 0.05 level.

Biochemical data were analysed using the repeated measurements analysis of variance.

The neutralising antibody to TGF-β was detected in the wound margins, fibrin clot and in the overlying scab up to 36 hours after a single injection at the time of wounding. At 48 hours post-injection, only faint staining was detected in the overlying scab. Interestingly, 15 minutes and thereafter, the antibody was also detected within and on the cell-surface of macrophages and monocytes (a rich source of TGF-β) present at the wound site. By contrast, 12 hours after injection, the irrelevant rabbit IgG in the sham control wounds, was only faintly detected in the overlying scab, in the inflammatory cells infiltrating the wound and in the wound margins. By 24 hours the irrelevant IgG could not be detected at the wound site. Neutralising antibody to TGF-β was not detected in sections from control wounds (Fig. 3).

There were no differences between any of the control and sham-control wounds. The macrophage and monocyte profile, fibronectin and collagen deposition were effectively reduced by treating the wounds with 50 μg/injection of the neutralising antibody to TGF-β. The wounds treated with 10 μg/injection of the antibody also showed slight reduction in the intensity of immunostaining for extracellular matrix compared to the control wounds. Surprisingly, wounds treated with 250 μg/injection of the antibody resembled the control wounds.

Wounds treated with 2 ng/injection of TGF-β₁did not differ markedly from the control wounds whilst the wounds treated with 10 ng/injection and 50 ng/injection had more blood vessels and more extracellular matrix compared to the control wounds (Tables 4, 5, 6; Figs 4, 5). Therefore, for all subsequent experiments, the experimental wounds were treated with 50 μg/injection of the neutralising antibody to TGF-β, positive control wounds with 10 ng/injection of TGF-β₁and the sham-control wounds with 50 μg/injection of irrelevant rabbit IgG.

At 28 days post-wounding, there were no differences between the experimental and any of the control wounds in groups IV and V. Wounds treated with neutralising antibody to TGF-β in groups I, II and III showed an improvement in the dermal architecture as demonstrated by Picrosirius red and Masson’s trichrome staining whilst there was scarring in the control, sham control and TGF-β₁-treated wounds in all groups (Fig. 6and Table 7). The collagen content of these wounds did not show any significant differences (Table 8).

Sections of wounds harvested 14 days post-wounding showed less fibronectin and collagen deposition in the neu-tralising antibody-treated wounds compared to the controls (Tables 9and 10). These wounds also appeared less vascular than the control wounds (Table 11). These effects were best seen in groups I and II (Tables 9, 10; Fig. 7), indicating the necessity for early application of the neutralising antibody to TGF-β. These effects were similar in wounds treated on days 0-2 and 0-6. Hence for all subsequent experiments, the wounds were treated on days 0-2 post-wounding.

Treatment of wounds with neutralising antibody to TGF-β markedly reduced both the macrophage profile and the monocyte and macrophage profile of healing wounds during the first 14 days post-injury compared to control wounds. The difference in the degree of inflammatory infiltrate was appreciable as early as 15 minutes. The reduction in the macrophage profile was more obvious on day 7 post-wounding (P=0.01; Fig. 8). At both 7 and 14 days post-wounding, the monocyte and macrophage profile of the antibody-treated wounds was significantly lower than that of the controls (P<0.01; Fig. 9).

At both time points, the macrophage profile and the monocyte and macrophage profile of normal skin surrounding all wounds were similar suggesting a localised effect of the treatments. This also suggests that there are no anatomical differences in the density of resident macrophages.

The vascularity of the wounds was assessed using laminin and Factor VIII-related antigen to mark the blood vessels. At all time points, laminin was detected in the basement membranes of the epidermis, hair follicles, blood vessels and in the per-imysium of the underlying muscle. By 7 days post-wounding, the basement membrane of the epidermis was completely formed in all wounds. The wounds treated with neutralising antibody to TGF-β were less vascular than the control, sham control and TGF-β₁-treated wounds but more vascular than the surrounding normal skin. Wounds treated with TGF-β₁were the most vascular of the four groups. By 10 weeks post-wounding the vascularity of the healed wounds appeared similar to that of normal skin.

Immunostaining with antibody to von Willebrand factor showed a similar staining pattern for blood vessels except for the smaller blood vessels surrounding the hair follicles, which were not detected by this antibody. Table 11shows vascularity of time and duration experiments.

In the normal skin, fibronectin staining was present through-out the dermis with more intense staining at the dermo-epidermal junction, in the subepidermal dermis and around the hair follicles and blood vessels. At 7 days post-wounding, fibronectin staining was more intense in the healing wounds than in the surrounding skin. The staining intensity decreased at 14 days post-wounding and by 28 days post-wounding, fibronectin staining in all wounds resembled that of the sur-rounding normal skin (Fig. 10).

At 7 and 14 days post-wounding, both the intensity of fibronectin staining and the density of the fibre network in the wounds treated with neutralising antibody to TGF-β were markedly less than that in the control wounds. There was no difference between the sham control and control wounds whilst the wounds treated with TGF-β₁had the most intense staining for fibronectin (Fig. 10, Table 12).

The staining patterns for both these collagens were similar. However, the intensity of staining for collagen III at the wound site was more than that for collagen I on days 7 and 14 post-wounding (Table 13).

Collagen staining was present throughout the normal dermis in a reticular pattern. Collagen deposition in the wounds increased with time. At all time points examined, the intensity of staining in the wounds treated with neutralising antibody to TGF-β was less than that in the control wounds. There were no differences between control and sham control wounds. At 7 and 14 days post-wounding, the intensity of staining in wounds treated with TGF-β₁was more than that in the control wounds; but thereafter, this difference was not observed. Interestingly, the collagen fibres were deposited compactly in an abnormal pattern in the control, sham control and TGF-β₁-treated wounds. By contrast, the collagen fibres in the antibody-treated wounds had more spacing between them and were laid down in a reticular pattern resembling that of the normal dermis (Fig. 11).

This difference in the orientation of collagen fibres was better visualised using Picrosirius red staining and polarised light microscopy or by Masson’s trichrome staining and bright field microscopy (Fig. 12).

The microscopic differences in the dermal architecture were directly related to the macroscopic appearance of the healed wounds. At 28 days post-wounding and thereafter, wounds treated with neutralising antibody to TGF-β had no obvious scar, were difficult to visualise and were only detected as a fine line due to the absence of hair. In shaved animals, neutralising antibody-treated wound sites were indistinguishable from the surrounding skin. By contrast, all other wounds healed with obvious scar formation (Fig. 13).

7 days post-wounding the glycosaminoglycan content of TGF-β₁-treated wounds was higher than that of the other wounds though this difference was not statistically significant. For all wounds, the glycosaminoglycan content of wounds decreased with time. Surprisingly, on day 42 post-wounding, TGF-β₁-treated wounds had significantly more (P=0.05) glycosamino-glycans than control wounds, which in turn had more glycosaminoglycans than the neutralising antibody-treated wounds (Table 14).

The temporal changes in the collagen content of wounds have previously been reported (Shah et al., 1992). At 7 days post-wounding wounds treated with neutralising antibody to TGF-β_1,2had significantly less collagen, whilst wounds treated with TGF-β₁had significantly more collagen, than did the control wounds. Thereafter there were no differences in the collagen content of any wounds.

The temporal changes in the tensile strength of wounds have previously been reported (Shah et al., 1992). Despite the lower collagen content of wounds treated with neutralising antibody to TGF-β_1,2on day 7 post-wounding, the tensile strength of these wounds was similar (i.e. not less) to that of the control wounds.

This study expands on our preliminary report (Shah et al., 1992) that exogenous addition of neutralising antibody to TGF-β_1,2reduces scar formation after cutaneous injury in adult rodents. Early application of neutralising antibody to TGF-β_1,2reduces the magnitude of monocyte and macrophage infiltration at the wound site, reduces neovascularisation, decreases the amount of fibronectin and collagen at day 7 post-wounding without reducing the tensile strength of the wounds and, most importantly, improves the architecture of the neodermis to resemble that of the normal dermis and thereby reduces scarring.

After a single injection, the neutralising antibody persisted in and around the wound site for up to 36 hours. Interestingly, the antibody was also detected on some of the macrophages and monocytes in the wound. Monocytes and macrophages are a rich source of TGF-β (Assoian et al., 1987), however, binding of the neutralising antibody to the Fc receptors cannot be ruled out. Dose response experiments showed that the high dose of neutralising antibody to TGF-β (250 μg/injection) did not produce the beneficial effects seen with lower dose (50 μg/injection) of this antibody. This may be due to neutralisation of more TGF-β₃in the healing wound at the higher dose compared to the lower doses. The polyclonal antibody used in this study has been reported to block 36% of the activity of recombinant TGF-β₃(Roberts et al., 1990). There are isoform-specific differences in the antiscarring effects of TGF-β (manuscript submitted). Reduced efficacy of the antibody at higher concentrations due to steric hinderance may be an additional explanation for lack of effect. It is also possible that a high concentration of foreign protein in the wound would increase the inflammatory response and thereby nullify the beneficial effects of the neutralising antibody. However, we did not observe a marked inflammatory response in these wounds.

This study shows that a reduction in scarring was achieved when the neutralising antibody to TGF-β was administered early (days 0-2, 0-6, 1-7) but the best effects were seen when the treatment regime included the day of wounding i.e. day 0 (Fig. 6). This suggests that crucial events occur immediately after injury. TGF-β₁protein has been detected as early as 5 minutes after injury to human skin (Kane et al., 1991). Platelets, which are a rich source of TGF-β (Van den Eijndenvan Raaij et al., 1988) release TGF-β on degranulation of the α-granules at the site of injury. TGF-β₁upregulates its own expression as well as that of basic fibroblast growth factor (bFGF), platelet derived growth factor (PDGF) and interleukin 1 (IL-1) in monocytes (McCartney-Francis et al., 1990). Early application of the neutralising antibody to TGF-β may cause a reduction in the level of TGF-β at the wound site, which in turn may alter the levels of PDGF, bFGF and other cytokines thereby altering the overall cytokine profile of the adult wound to one that is conducive to reduced scar formation. Application of the neutralising antibody to TGF-β to healing wounds on days 7 to 9 or days 19 to 21 post injury, did not reduce scarring. Whilst inadequate dosage cannot be ruled out, the timing of the treatment seems to be more important. Cellular response to a cytokine is affected by the presence of other cytokines as well as the extracellular matrix surrounding the cells (Nathan and Sporn, 1991). As the milieu of a healing wound changes continuously, it seems crucial to alter the cytokine profile at the right time in order to influence the final outcome. In this context, the antiscarring effect of neutralising antibody to TGF-β appears to be analogous to the therapeutic benefits of monoclonal antibody to tumour necrosis factor-α (TNF-α) in septic shock where the ‘therapeutic window’ is narrow (Exley et al., 1990).

Exogenous addition of neutralising antibody to TGF-β resulted in decreased numbers of macrophages and monocytes at the wound site. TGF-β is the most potent chemoattractant for macrophages and monocytes in the femtomolar range (Wahl et al., 1987). These inflammatory cells themselves produce TGF-β, thereby perpetuating the presence of TGF-β in the wounds. Exogenous neutralising antibody to TGF-β, applied early, probably dampens this autocatalytic cascade, providing another explanation for optimum antiscarring effects with early application. TGF-β induces the expression of Fc_gRIII (CD16) on newly recruited monocytes, which promote phagocytic activity and the release of toxic oxygen radicals in vivo (Welch et al., 1990; Wahl et al., 1992). The antigen recognised by the monoclonal antibody ED1 used in this study to measure the monocyte and macrophage profile appears to be related to the phagocytic function of these cells (Dijkstra et al., 1985). Therefore exogenous treatment with neutralising antibody to TGF-β may also have reduced the local release of toxic oxygen radicals thereby reducing further tissue damage. Thus, neutralisation of TGF-β may also alter the macrophage differentiation pathway preventing large numbers from differentiating down the inflammatory lineage.

TGF-β is known to upregulate the expression and production of various cytokines such as basic FGF, PDGF and interleukin 1 (IL-1) by monocytes. Basic FGF is a potent angiogenic factor (Baird and Bohlen, 1990) and TGF-β itself appears to control the organisation of endothelial cells into tubules (Yang and Moses, 1990). Subcutaneous injection of TGF-β₁into the necks of new-born mice resulted in neovascularisation and granulation tissue formation (Roberts et al., 1986). Treatment with neutralising antibody to TGF-β resulted in healing wounds with less angiogenesis compared to controls. This is probably also due to the altered cytokine profile, probably lower basic FGF levels, in the antibody-treated wounds. Neutralising antibody treatment may have reduced the levels of basic FGF, which in turn lead to reduced plasminogen activator levels thereby resulting in less activation of latent TGF-β (Flaumenhaft et al., 1992).

The effects of TGF-β on matrix deposition have been studied extensively (Roberts and Sporn, 1990). TGF-β is a potent chemoattractant for fibroblasts and by inducing the expression and production of PDGF and its receptors, causes proliferation of these cells (Ishikawa et al., 1990). TGF-β induces the expression and incorporation into the extracellular matrix of fibronectin, collagen I and collagen III (Ignotz and Massague, 1986; Varga and Jimenez. 1986; Ignotz et al., 1987). In addition to increasing matrix synthesis, TGF-β down-regulates proteolytic enzyme production and enhances protease inhibitors (Edwards et al., 1987; Overall et al., 1989) so resulting in reduced matrix degradation. TGF-β also modulates the effects of other growth factors on extracellular matrix turnover (Laiho and Keski-Oja, 1989). Exogenous addition of neutralising antibody to TGF-β resulted in less extracellular matrix at day 7 post-wounding. Despite the lower collagen content, these wounds were functionally no different from the control wounds as assayed by tensile strength measurements (Shah et al., 1992). This, we believe is due to the better orientation of collagen in the antibody-treated wounds. Modulation of the cytokine profile early in the healing process, probably alters extracellular matrix turnover resulting in a less compact and better organised neodermis. Although the intensity of staining for the matrix macromolecules was persistently less than that in the control wounds, this difference was not seen in biochemical analyses of wounds at later time points. This discrepancy may be due to a technical error in sampling for bio-chemistry as the (scar-free) wounds treated with the antibody became increasingly difficult to differentiate and dissect from the surrounding skin with time.

The most interesting observation however, was the difference in the orientation of the collagen fibres. Whilst the control, sham control and TGF-β₁-treated wounds healed with obvious scar tissue and abnormally oriented collagen, the collagen in wounds treated with neutralising antibody to TGF-β was deposited in a reticular fashion resembling that of the normal dermis. Early application of neutralising antibody to TGF-β probably reduces the production of plasminogen activator inhibitor (PAI-1), increases the net amount of plasminogen activator and plasmin and thereby increases fibrinolysis. This makes the initial fibronectin-fibrin scaffold less compact enabling the cells to migrate into the wound site easily and lay down the extracellular matrix over the loose scaffold in a more reticular pattern. To this end, extracellular matrix deposition and organisation in a wound may be an iterative process and hence the importance of early application of the neutralising antibody to TGF-β. Additionally, increased cell surface plasminogen activator and plasmin on fibroblasts of the neutralising antibody-treated wound margins may facilitate egress of these dermal cells and their invasion of the fibrin clot to form a more dermal-like provisional matrix. However, it should be noted that the epidermal appendages and the pan-niculus carnosus showed no signs of regeneration. In this respect, ‘scarless healing’ caused by application of neutralising antibody to TGF-β in adult wounds differs from the regenerative response seen in embryonic wound healing.

Embryonic/fetal wound healing is characterised by a lower and different inflammatory and growth factor response, less angiogenesis, different extracellular matrix composition and regeneration of the tissue (Whitby and Ferguson, 1991a,b). By contrast, the wound healing process in the adult appears to be optimised for rapid restoration of tissue integrity and the prevention of infection by ensuring rapid clearance of debri, rapid closure of the defect and prevention of dehiscence. Thus adult wound healing is characterised by a massive (? excessive) inflammatory response, increased neovascularisation, excessive extracellular matrix deposition and scar formation. This is probably the outcome of an overdrive response to injury, resulting in an excess of certain cytokines in the adult healing wound. Treatment of the wounds with neutralising antibody to TGF-β did not result in delayed healing. Similarly exogenous addition of neutralising antibodies to basic FGF or PDGF to incisional dermal wounds did not delay healing (Shah et al., 1991) supporting the cytokine overdrive hypothesis. Exogenous addition of TGF-β₁to incisional and excisional wounds increases the mRNA for fibronectin, collagen I and collagen III and reduces the mRNA of stromelysin (Quaglino et al., 1990, 1991). The present study demonstrated an increase in collagen content of TGF-β₁-treated wounds, 7 days post-wounding (Shah et al., 1992). However, this effect was not sustained and in the long term the scars in the TGF-β₁-treated group were similar to those of the control group.

The importance of TGF-β in wound healing is becoming increasingly evident (Roberts and Sporn, 1990). TGF-β₁protein and mRNA have been detected in healing wounds (Quaglino et al., 1990, 1991; Kane et al., 1991; Majesky et al., 1991). Impaired wound healing caused by glucocorticoids or streptozotocin (Pierce et al., 1989; Broadley et al., 1989; Beck et al., 1991), adriamycin (Lawrence et al., 1986; Salomon et al., 1990) and irradiation (Bernstein et al., 1991) can be reversed by exogenous addition of TGF-β₁.

However, it is also becoming clear that effects of excessive or inappropriately expressed TGF-β are detrimental resulting in tissue damage by fibrosis (Border and Ruoslahti, 1992). We have shown that early application of neutralising antibody to TGF-β to adult dermal wounds reduces scarring (Shah et al., 1992). Neutralising antibody to TGF-β has also been reported to reduce the fibrotic process in glomerulonephritis (Border et al., 1990) and the synovitis and destructive phases of arthritis (Wahl et al., 1993).

Thus, alteration of the cytokine profile of incisional cutaneous wounds, by exogenous application of neutralising antibody to TGF-β, markedly modulates the early events of the healing process resulting in a structured deposition of the extra-cellular matrix and restoration of the dermal architecture. This study therefore confirms the view that acute wounds in the adult are optimised for speed of closure under dirty conditions but at the expense of final wound quality (Shah et al., 1992) so that manipulation of their early cytokine profile is a rational and novel therapeutic target to reduce adult scarring and its adverse cosmetic and functional sequelae.

This work was supported by reseach grants from the North Western Regional Health Authority and the British Technology Group. We thank Chris Roberts (Dental school, Manchester) for his help with the statistical analysis and Mr P. J. Davenport (Consultant Plastic Surgeon, University Hospital of South Manchester) for help and discussion.