Delayed wound healing caused by inefficient re-epithelialization underlines chronic skin lesions such as those found in diabetes. The dioxin receptor (AhR) modulates cell plasticity and migration and its activation by occupational polycyclic aromatic hydrocarbons (PAHs) results in severe skin lesions such as contact hypersensitivity, dermatitis and chloracne. Using wild-type (Ahr+/+) and AhR-null (Ahr–/–) mouse primary keratinocyte cultures and tissue explants, we show that lack of AhR increases keratinocyte migration and accelerates skin re-epithelialization without affecting cell proliferation or recruitment of inflammatory cells. Wounds in Ahr–/– animals had elevated numbers of fibroblasts and increased collagen content in their granulation tissue. Importantly, Ahr–/– dermal fibroblasts secreted higher levels of active TGFβ that increased keratinocyte migration in culture and that could account for over-activation of the TGFβ pathway and for faster wound healing in the AhR-null neo-epithelium. Consistently, a TGFβ neutralizing antibody decreased keratinocyte migration in culture and halted re-epithelialization in Ahr–/– mice. Moreover, in vivo treatment with an antisense oligonucleotide for AhR increased TGFβ signaling and improved re-epithelialization in wounds of wild-type mice. These data indicate that AhR is relevant for wound repair and suggest that AhR downmodulation might be a potential new tool for the treatment of chronic, surgical or accidental wounds.
Inefficient wound healing represents an important health problem for individuals suffering accidental, surgical or chronic skin lesions and has special relevance for the elderly and for immunosuppressed, diabetic or cancer patients (Reed and Clark, 1985; Schafer and Werner, 2008). Tissue repair is a complex process that involves new blood vessel formation and the recruitment to the wound of inflammatory cells, such as macrophages and neutrophils, to eliminate contaminants and bacteria and fibroblasts that produce extracellular matrix (ECM) components (e.g. collagen and fibronectin). In parallel to the formation of granulation tissue, increased proliferation and migration of keratinocytes takes place to repair the damaged epidermis (Eckes et al., 1999; Martin, 1997; Schafer and Werner, 2008; Singer and Clark, 1999). The intermediate molecules involved in adult skin repair are only partially known. Nevertheless, analysis of overexpression and knockout animal models has revealed the important contribution of adhesion molecules such as β1 integrins (Grose et al., 2002; White et al., 2004), growth factors such as FGF and HGF (Werner and Grose, 2003), ECM constituents such as osteopontin (Mori et al., 2008) and cytokines such as TGFβ (Ashcroft et al., 1997; Singer and Clark, 1999).
TGFβ is one of the best characterized profibrogenic molecules (Bauer and Schuppan, 2001; Kanzler et al., 1999) and alterations in its synthesis, secretion and intracellular signaling are associated with many pathological states including cancer and tissue fibrosis (Blobe et al., 2000; Corchero et al., 2004; Massague, 2000; Massague and Chen, 2000). Despite the fact that exogenous TGFβ3 administration is currently under clinical trial as a novel anti-scarring agent (Shah et al., 1995), the effect of this cytokine on tissue repair remains controversial. Some studies suggest that TGFβ regulates wound healing because its expression is increased by platelets, inflammatory cells and fibroblasts located at the site of injury (Amendt et al., 2002; Leibovich and Ross, 1975; Schafer and Werner, 2007). Genetic manipulation of TGFβ or its receptors, however, can both promote and inhibit tissue regeneration (Amendt et al., 2002; Brown et al., 1995; Crowe et al., 2000; Shah et al., 1999). Furthermore, additional work has revealed that depending on the level of TGFβ activity and/or the experimental model used, this cytokine has been shown not to affect (Leask et al., 2008), to inhibit (Hosokawa et al., 2005; Yang et al., 2001) or to promote skin re-epithelialization (Gailit et al., 1994; Reynolds et al., 2005; Reynolds et al., 2008). TGFβ-dependent signaling is mediated by its binding and activation of plasma membrane serine-threonine kinase receptors that will phosphorylate and activate intracellular intermediates of the Smad family of proteins (Smad2, Smad3 and Smad4). Activated Smads will heterodimerize and enter the cell nucleus where they activate target gene expression (Massague, 2000; Siegel and Massague, 2003).
The aryl hydrocarbon (dioxin) receptor (AhR) is a member of the class VII of basic-helix-loop-helix-PAS (bHLH-PAS) family of transcription factors. AhRs regulate gene expression through heterodimerization with the nuclear protein aryl hydrocarbon receptor nuclear translocator ARNT (Furness et al., 2007). In addition to its relevant role in xenobiotic-induced toxicity and carcinogenesis (Fernandez-Salguero et al., 1996; Mimura et al., 1997; Nebert et al., 2004; Shimizu et al., 2000), AhR is gaining considerable interest because of its contribution to the control of cell proliferation, differentiation and tissue homeostasis (Barouki et al., 2007; Gomez-Duran et al., 2008b; Puga et al., 2002). Among the different cell functions requiring AhR, the control of TGFβ activation appears particularly relevant. In cell culture systems, AhR activity has been functionally related to increased secretion and activation of TGFβ in primary hepatocytes (Zaher et al., 1998) and mouse embryo fibroblasts (Elizondo et al., 2000; Gomez-Duran et al., 2008a; Gomez-Duran et al., 2006; Santiago-Josefat et al., 2004) and, consistently, with diminished cell proliferation and increased apoptosis. In mice, knockdown of AhR expression results in fibrotic lesions in the liver (Corchero et al., 2004; Fernandez-Salguero et al., 1995; Peterson et al., 2000), heart and skin (Fernandez-Salguero et al., 1997), which, at least for the hepatic tissue, colocalized to the portal areas with increased levels of TGFβ (Corchero et al., 2004). Collectively, these studies offer the possibility that changes in AhR expression could affect tissue repair by controlling TGFβ activity.
The involvement of AhR in the control of skin disease and tissue remodeling is only just beginning to emerge. A recent report showed that constitutive activation of AhR and increased expression of Cyp1a1, Gsta1, Fos and TGFA are underlying factors in the development of chloracne in human subjects occupationally or accidentally exposed to significant doses of polycyclic aromatic hydrocarbons (Imamura et al., 2007; Tang et al., 2008). Regeneration studies in adult zebra fish have revealed that AhR2 activation by acute exposure to dioxin (TCDD, 2,3,7,8-tetrachlorodibenzo-[p]-dioxin) impairs caudal (tail) fin regeneration (Andreasen et al., 2007), which suggests that AhR has an inhibitory role in tissue remodeling. Finally, transgenic mice expressing a constitutively activated form of AhR in their keratinocytes develop severe skin lesions with itching and inflammation that resembled atypical atopic dermatitis (Tauchi et al., 2005). These results indicate that AhR activation impairs skin wound healing and suggest that maintained receptor activity by chronic exposure to occupational and environmental xenobiotics could exacerbate an inflammatory response eventually affecting tissue repair.
In this work, we have analyzed whether the lack of AhR expression can accelerate wound healing in vivo and whether TGFβ has a role in such putative AhR-dependent process. Using wild-type and AhR-null mice, primary keratinocyte cultures and tissue explants, and by the modulation of TGFβ activity and AhR expression, we report that loss of AhR increases keratinocyte migration and enhances the efficiency of wound healing in vivo and that such mechanisms require TGFβ activity. Since downregulation of AhR at the wound site by antisense oligonucleotides significantly accelerated wound healing in Ahr+/+ mice, this study provides a potential new tool that could be used to improve re-epithelialization in diseased skin wounds.
Loss of AhR expression improves wound healing by accelerating re-epithelialization
Considering previous studies showing that AhR activation in mouse causes inflammatory skin lesions (Tauchi et al., 2005) and impairs fin regeneration in zebra fish (Andreasen et al., 2007), and since humans exposed to dioxin develop the skin disease chloracne associated with an increase in AhR-dependent transcription (Tang et al., 2008), we first analyzed whether AhR expression modulates wound healing in vivo. Histological analyses of wounds performed in the dorsal skin of Ahr+/+ and Ahr–/– mice revealed that AhR-null mice closed their wounds significantly faster than wild-type mice, this effect being more pronounced between days 3 and 5 (wound diameter was measured as the distance between both flanks of the regenerating neo-epithelium or granulation tissue) (Fig. 1, arrows). At 7 days after wounding, Ahr+/+ and Ahr–/– mice had completed closure of the skin. Further analysis of the healing process allowed us to determine that the neo-epithelium progressed faster in Ahr–/– than in Ahr+/+ mice. By contrast, the diameter of the granulation tissue did not significantly vary during wound healing, suggesting that the epidermal layer was a main target for AhR-dependent re-epithelialization (Fig. 1). Additionally, macroscopic analysis of the wounds confirmed that Ahr–/– mice were more efficient in the healing reaction than wild-type mice (data not shown). The largest differences in wound healing were observed at day 5; however, we performed the in vivo experiments at day 3 because the almost complete healing in Ahr–/– mice at day 5 could distort comparison with wild-type mice regarding relevant parameters such as keratinocyte proliferation and migration and TGFβ response (over 90% of the Ahr–/– wounds were totally closed at day 5). In addition, AhR expression did not significantly vary during the wound-healing process, or with respect to basal skin, suggesting that delayed re-epithelialization in Ahr+/+ mice was not due to alterations in AhR protein levels (supplementary material Fig. S1).
To further analyze the role of the epithelium in the AhR-dependent wound-healing phenotype, we measured the length of the neo-epithelial layer covering the area between the wound site and the margin of the regenerating tissue (Fig. 2A, arrows and red dotted line). It can be seen that the length of the neo-epithelium (e) was significantly larger in Ahr–/– mice than in Ahr+/+ mice (Fig. 2A, right). Since re-epithelialization requires an increase in keratinocyte proliferation (Schafer and Werner, 2007; Schafer and Werner, 2008), we next analyzed whether differences in keratinocyte proliferation could help explain the more efficient wound healing in Ahr–/– mice. PCNA immunostaining showed that proliferation rates were similar in Ahr+/+ and Ahr–/– wounds (Fig. 2B), which suggested the involvement of additional mechanisms.
Keratinocytes lacking AhR have increased migration
Cell migration is an important parameter that markedly affects the quality of tissue repair (Schafer and Werner, 2007; Schafer and Werner, 2008). Our results prompted us to study whether an increase in keratinocyte migration could be important for the Ahr–/– phenotype. An ex vivo approach useful for analysis of changes in cell migration consists of culturing skin explants under conditions that favor keratinocyte emigration but inhibit cell proliferation in absence of serum (Guasch et al., 2007). In preliminary experiments, immunofluorescence staining for the keratinocyte-specific marker cytokeratin 14 confirmed the epithelial phenotype of the emigrating cells (supplementary material Fig. S2). To avoid interference due to cell proliferation, keratinocyte migration experiments (from tissue explants or primary cells) were performed in absence of serum. As shown in Fig. 3A, keratinocytes from Ahr–/– explants had increased migration rates over time (slope 145.96) than keratinocytes from Ahr+/+ explants (slope 68.55). Statistical analysis revealed that the difference in slopes between Ahr+/+ and Ahr–/– explants was highly significant at P=0.00214. Thus, migration appears to be relevant for the increased wound healing observed in absence of AhR in vivo. This hypothesis was further confirmed by establishing primary keratinocyte cultures from Ahr+/+ and Ahr–/– newborn mice. Primary keratinocytes were placed in collagen- or fibronectin-coated culture plates and grown to confluence to halt cell proliferation. Wounds were performed and cell migration measured 15 hours later as the distance between margins (Fig. 3B, left). Quantification of the data revealed that Ahr–/– primary keratinocytes migrated significantly faster than Ahr+/+ keratinocytes in both ECM components collagen and fibronectin. Altogether, these results indicate that keratinocytes lacking AhR have increased migration rates and suggest that this cellular characteristic could underline the higher efficiency of Ahr–/– mice in re-epithelialization and wound healing.
Ahr–/– wounds have increased fibroblasts and collagen but similar recruitment of inflammatory cells
At the early stages of wound healing (e.g. 3 days), an inflammatory response takes place that recruits inflammatory cells such as macrophages and neutrophils to the granulation tissue (Schafer and Werner, 2008). To determine whether differences in inflammation could contribute to increased wound healing in Ahr–/– mice, we analyzed by immunohistochemistry the content of macrophages (staining for F4/80) and neutrophils (Ly-6G antigen) in the granulation tissue. AhR expression did not significantly alter the ability of the wounded tissue to recruit those inflammatory cells, because their numbers were very similar between Ahr+/+ and Ahr–/– mice (Fig. 4A,B). Additionally, under basal conditions, the skin of Ahr+/+ and Ahr–/– mice had a very low content of macrophages and neutrophils (data not shown). Fibroblast cells also accumulate in the wounded tissue and promote re-epithelialization by the production of ECM proteins such as collagen and by the secretion of growth factors (Schafer and Werner, 2008). The marker protein vimentin, although expressed by endothelial cells, can be used to estimate the presence of fibroblasts. Immunohistochemistry for vimentin revealed that Ahr–/– wounds had a moderate, although consistent, increase in fibroblast numbers with respect to Ahr+/+ wounds (Fig. 4C). Western blot analyses also showed elevated vimentin levels in AhR-null wounds (Fig. 4C). Stromal fibroblasts can differentiate into α-smooth muscle actin (α-SMA)-expressing myofibroblasts in certain pathologies such as cancer (Elenbaas and Weinberg, 2001; Ronnov-Jessen et al., 1996). We analyzed such a possibility by measuring the expression of the myofibroblast-specific marker α-SMA in Ahr–/– mice wounds. As shown in Fig. 5A, α-SMA was found at similar levels in Ahr–/– and Ahr+/+ wounds, indicating that lack of AhR did not significantly affect myofibroblast differentiation during re-epithelialization. A moderate increase in fibroblast numbers could enhance matrix deposition in the granulation tissue. To address this issue, we performed Sirius red and fast green staining, and found that wounds from Ahr–/– mice had a significant increase in collagen content in their granulation tissue compared with Ahr+/+ wounds (Fig. 5B). To further support this observation, hydroxyproline levels were also measured to more accurately quantify collagen deposition. Consistently, Ahr–/– wounds showed a significant elevation in hydroxyproline content that was indicative of elevated collagen deposition (Fig. 5C). Thus, accelerated wound healing in untreated Ahr–/– mice does not seem to involve an exacerbated inflammatory response, although it requires increased fibroblast recruitment and enhanced collagen deposition in the ECM.
TGFβ signaling is increased in Ahr–/– wounds
The higher collagen content in the granulation tissue of Ahr–/– mice suggested the possibility that an increase in growth factor secretion could modulate wound healing. In addition, AhR has a role in TGFβ activation because AhR-null mice produce increased levels of active cytokine (Corchero et al., 2004; Elizondo et al., 2000; Gomez-Duran et al., 2008a; Gomez-Duran et al., 2008b; Gomez-Duran et al., 2006; Santiago-Josefat et al., 2004). Based on this information, we performed experiments aimed to analyze the contribution of TGFβ signaling in the AhR-dependent wound-healing phenotype. Protein expression in whole cell extracts from basal skin showed that Ahr–/– mice expressed higher levels of total TGFβ than did Ahr+/+ mice. Wounding induced an increase in total TGFβ protein in both mouse genotypes that normalized the differences present in basal skin (Fig. 6A). Since active, rather than latent TGFβ accounts for TGFβ-dependent signaling (Massague, 2000; Massague and Gomis, 2006), we determined by ELISA whether the lack of AhR expression affected the secretion of active cytokine by primary keratinocytes (as target cells for re-epithelialization) and dermal fibroblasts (as a source for active TGFβ). Ahr–/– dermal fibroblasts secreted higher levels of active TGFβ (measured as the active:total cytokine ratio) than did wild-type cells (Fig. 6B, CM-DF). By contrast, similar amounts of active cytokine were secreted by primary keratinocytes in both experimental groups (Fig. 6B, CM-Ker). Increased production of active TGFβ by Ahr–/– dermal fibroblasts might be relevant for progression of the neo-epithelium during wound healing. Analysis of TGFβ-dependent signaling revealed that the number of keratinocytes activating the TGFβ pathway (determined as the ratio of Smad2-P-positive cells/area) was significantly higher in Ahr–/– than in Ahr+/+ wounds (Fig. 6C). Thus, although wounding increases total TGFβ to similar levels in Ahr+/+ and Ahr–/– mice, the increased secretion of active cytokine by AhR-null dermal fibroblasts could enhance TGFβ signaling and re-epithelialization in vivo.
Modulation of TGFβ levels alters Ahr+/+ and Ahr–/– keratinocyte migration and rescues the Ahr–/– wound-healing phenotype
Altogether, the data obtained led to the hypothesis that accelerated wound healing in Ahr–/– mice could be due, at least in part, to an increase in keratinocyte migration induced by overproduction of active TGFβ by dermal fibroblasts. Nevertheless, the fact that AhR-null primary keratinocytes had intrinsic differences in migration (Fig. 3B), suggests that these cells could secrete molecules modulating motility in a cell-type-autonomous fashion. To test this possibility, migration of Ahr+/+ and Ahr–/– keratinocytes was determined in the presence of their own conditioned medium or medium conditioned by the opposite phenotype (Fig. 7A). Ahr–/– keratinocytes in their conditioned medium migrated faster than Ahr+/+ keratinocytes growing in their own medium. Addition of conditioned medium from Ahr–/– keratinocytes (CM-Ker Ahr–/–) marginally increased migration of Ahr+/+ cells (P=0.125) whereas medium from Ahr+/+ cells (CM-Ker Ahr+/+) slightly decreased migration of Ahr–/– keratinocytes (P=0.069). Thus, although keratinocytes could modulate their own migration autonomously, and these observations are of potential interest, we focused our study on dermal fibroblasts based on the functional relationship between AhR and TGFβ in fibroblast cells. We first analyzed the effect of conditioned medium from dermal fibroblasts (CM-DF) on keratinocyte migration in culture. Growth of wild-type keratinocytes with medium from AhR-null dermal fibroblasts increased their migration rates above levels obtained by the addition of wild-type conditioned medium (Fig. 7B). Consistently, culture of Ahr–/– keratinocytes with conditioned medium from wild-type dermal fibroblasts inhibited their migration rates to levels below those produced by AhR-null CM-DF (Fig. 7B). Since these results suggested that dermal fibroblasts secreted a molecule(s) affecting keratinocyte migration, and considering the AhR-TGFβ relationship, as well as the relevance of TGFβ in wound healing, we performed further experiments designed to modulate keratinocyte migration by adjusting TGFβ activity. Ahr–/– keratinocytes, growing in CM-DF Ahr+/+ medium and treated with 10 ng/ml recombinant TGFβ, exhibited a significant increase in migration compared with the same experimental conditions without TGFβ addition (Fig. 7C, compare bars 2 and 4). Accordingly, Ahr–/– keratinocytes, cultured in CM-DF Ahr–/– medium and treated with 1 μg/ml of neutralizing anti-TGFβ antibody, decreased their migration to levels close to those induced by CM-DF Ahr+/+ medium (Fig. 7C, compare bars 6 and 8 with 2). Ahr+/+ keratinocytes had a similar response, and their culture in CM-DF Ahr+/+ medium supplemented with 10 ng/ml recombinant TGFβ also increased migration compared with results obtained with CM-DF Ahr+/+ alone (Fig. 7C, compare bars 3 and 1). Moreover, Ahr+/+ keratinocytes cultured in CM-DF Ahr–/– migrated faster than the same cells in presence of CM-DF Ahr+/+ (Fig. 7C, compare bars 5 and 1), whereas addition of 1 μg/ml of neutralizing anti-TGFβ antibody to CM-DF Ahr–/– significantly decreased migration of Ahr+/+ keratinocytes (Fig. 7C, compare bars 7 and 5). Therefore, increasing TGFβ levels in medium conditioned by Ahr+/+ or Ahr–/– dermal fibroblasts increased Ahr–/– and Ahr+/+ keratinocytes migration, whereas lowering TGFβ activity in medium conditioned by Ahr–/– or Ahr+/+ dermal fibroblasts decreased migration of wild-type and AhR-null keratinocytes. The neutralizing activity of the anti-TGFβ antibody has been previously determined (IC50=0.06 μg/ml to inhibit the proliferation of Mv1Lu cells by 50%) (Santiago-Josefat et al., 2004).
To determine whether an increase in TGFβ activity is relevant to stimulate more efficient keratinocyte migration and improved wound healing in Ahr–/– mice, we first cultured skin explants from Ahr–/– mice in the presence of 1 μg/ml of neutralizing anti-TGFβ antibody. We found that neutralizing TGFβ activity reduced migration of Ahr–/– keratinocytes to a level similar to that observed in explants from Ahr+/+ mice (Fig. 8A). In agreement with our hypothesis, neutralization of TGFβ activity in vivo also inhibited wound healing in Ahr–/– mice to values that were similar to those observed in untreated Ahr+/+ wounds (Fig. 8B). In agreement with the data presented in Fig. 1, wound-healing inhibition by the anti-TGFβ antibody blocked migration of the epithelial layer without a significant effect on the progression of the granulation tissue (Fig. 8B). It is interesting to note that increased TGFβ response in Ahr–/– keratinocytes might not only involve higher levels of active cytokine but also changes in TGFBR1 and TGFBR2 receptors. Real-time RT-PCR analyses of Tgfbr1 and Tgfbr2 mRNA expression showed that although mRNA levels for the type 1 receptor did not significantly vary between Ahr+/+ and Ahr–/– keratinocytes, the expression of the cytokine-binding TGFBR2 receptor was moderately increased in AhR-null cells (supplementary material Fig. S3). Thus, increased TGFβ signaling in Ahr–/– wounds could involve a complex mechanism of cytokine overactivation and TGFBR2 overexpression.
AhR downmodulation by antisense oligonucleotide mimics the Ahr–/– wound-healing phenotype and increases TGFβ-dependent signaling in the neo-epithelium
Collectively, our data indicate that AhR has a causal role in modulating the efficiency of wound healing and that such a process involves the regulation of TGFβ activity. To further demonstrate the role of AhR in wound healing, we applied antisense oligonucleotide to wounds in Ahr+/+ mice and quantified differences in re-epithelialization and in the activation of the TGFβ pathway. Addition of a gel containing antisense oligonucleotides decreased AhR levels in the granulation tissue of Ahr+/+ wounds whereas a control sense oligonucleotide did not show a significant effect, as determined by immunofluorescence (Fig. 9A, area below dotted line) or western immunoblotting (Fig. 9A). Remarkably, antisense oligonucleotides significantly increased re-epithelialization and accelerated wound healing in Ahr+/+ mice (Fig. 9B), which supported a causal role for AhR in epithelial regeneration. Furthermore, in agreement to the regulatory role of TGFβ activity in wound healing in Ahr–/– mice, treatment with antisense oligonucleotides for AhR also increased the number of keratinocytes activating TGFβ-dependent signaling in the neo-epithelial layer (quantified as the ratio of Smad2-P-positive cells/area in Fig. 9C). Thus, downregulation of AhR expression accelerates wound healing in vivo through a mechanism involving increased TGFβ activity.
The cellular functions of the dioxin receptor AhR appear far more complex than those related to the regulation of xenobiotic metabolism (Barouki et al., 2007; Gomez-Duran et al., 2008b; Puga et al., 2005), and different studies have demonstrated its role in cell proliferation, differentiation and apoptosis (Barouki et al., 2007; Furness et al., 2007; Nebert and Dalton, 2006). Interestingly, early reports already indicated that AhR had a role in epithelial cell adhesion because suspension of human keratinocytes activated this receptor in the absence of xenobiotics (Sadek and Allen-Hoffmann, 1994). Later work on the keratinocyte cell line HaCaT showed that AhR regulates the expression of the epithelial-to-mesenchymal marker Slug and that AhR becomes activated in cells located at the leading edge in wound healing in vitro (Ikuta and Kawajiri, 2006). Despite these studies in cultured cells, the involvement of AhR in the control of epithelial cell migration in vivo is mostly unknown. In this study, we used AhR-null mice, skin explants and primary keratinocytes cultures to demonstrate that lack of AhR expression accelerates wound healing and re-epithelialization by increasing keratinocyte migration, and that such a phenotype is dependent on elevated levels of active TGFβ.
Although AhR activation in cultured epithelial cells has been associated with increased cell migration (Ikuta and Kawajiri, 2006; Sadek and Allen-Hoffmann, 1994), we found that genetic knockout of AhR expression accelerates wound healing in vivo, therefore suggesting that skin wound repair could be hampered under conditions of maintained AhR activation. This initial result has in fact been confirmed in a different AhR-null mouse line in which a preliminary macroscopic examination of skin wounds also revealed that AhR-null mice had a faster wound-healing response (Ikuta et al., 2008). Interestingly, the accelerated wound healing that we found in Ahr–/– mice involved a larger progression of the epithelial layer rather than changes in the progression of the granulation tissue. However, the increased length of the neo-epithelium in Ahr–/– wounds did not result from higher proliferation rates of their keratinocytes, which suggested that additional parameters, such as increased cell migration and/or augmented inflammatory reaction, are involved. Interestingly, although TGFβ inhibits cell proliferation, Ahr–/– wounds had only a marginal decrease in keratinocyte proliferation with respect to Ahr+/+ wounds, suggesting that TGFβ could affect proliferation and migration of keratinocytes with differing sensitivity. Future experiments will be required to determine whether TGFβ secreted by dermal fibroblasts differentially affects cell proliferation and migration of Ahr+/+ and Ahr–/– keratinocytes. This hypothesis is particularly interesting considering that keratinocytes lacking AhR overexpress the cytokine-interacting TGFβR-2 receptor, which could cooperate with TGFβ to exert different effects on epithelial cell proliferation and migration. Recruitment of inflammatory cells important for wound healing, such as macrophages and neutrophils, was similar in Ahr+/+ and Ahr–/– wounds, indicating that a difference in the inflammatory response is not a critical factor in the AhR-null phenotype. In agreement with a delaying activity of AhR in wound healing, a previous study has shown that transgenic mice expressing a constitutively active form of the receptor in their keratinocytes developed skin lesions that resembled atypical atopic dermatitis and that involved a significant inflammatory reaction (Tauchi et al., 2005). Thus, although Ahr–/– wounds develop a normal inflammatory response, inflammation becomes aggravated in the skin lesions of mice overexpressing a constitutively activated AhR. Taken together, these studies suggest that a reduction in the physiological levels of AhR can increase the efficiency of wound healing in vivo without a major local inflammatory reaction.
Our results prompted us to analyze whether accelerated wound healing in Ahr–/– mice was the result of increased keratinocyte migration. Experiments performed in skin explants ex vivo and in primary keratinocyte cultures clearly indicated that lack of AhR expression significantly increased epithelial cell migration. It is well known that many cytokines and growth factors regulate the efficiency of wound healing (Grose and Werner, 2003; Scheid et al., 2000; Singer and Clark, 1999). We focused our attention on TGFβ for several reasons: (1) mouse models with targeted inactivation of the genes encoding β3 integrin (Reynolds et al., 2005) and Dpr2 (Meng et al., 2008) showed accelerated re-epithelialization and an enhanced response to TGFβ, whereas decreased TGFβ activity correlated with impaired wound healing in PKCϵ-null mice (Leask et al., 2008) and (2) we have extensively shown that absence of AhR expression results in increased TGFβ activity in certain cell types such as fibroblasts and hepatocytes (Corchero et al., 2004; Elizondo et al., 2000; Gomez-Duran et al., 2008a; Gomez-Duran et al., 2006; Santiago-Josefat et al., 2004). Fibroblasts produce many components of the ECM and are also a relevant source of TGFβ. After synthesis and secretion, TGFβ is first linked to the ECM via LTBP, from which the cytokine will be released and activated by extracellular proteases (Annes et al., 2003; Gomez-Duran et al., 2006). Ahr–/– wounds had a moderate increase in fibroblast numbers in their granulation tissue that correlated with a significant accumulation of collagen in the ECM. Since increased collagen content colocalized with elevated TGFβ levels in Ahr–/– liver (Corchero et al., 2004), we analyzed whether AhR-null wounds produced higher levels of TGFβ. Wounding increased total TGFβ content to a similar extent in Ahr+/+ and Ahr–/– mouse skin, normalizing the differences present in non-wounded skin. Regardless the amount of total TGFβ secreted, this cytokine can only initiate signaling through its receptors when released in its active form (Massague, 2000; Massague and Gomis, 2006; Massague and Wotton, 2000). Therefore, we determined the amount of active TGFβ secreted by dermal fibroblasts and keratinocytes. Interestingly, Ahr–/– dermal fibroblasts, but not keratinocytes, produced significantly more active TGFβ than did wild-type cells, indicating that the granulation tissue of AhR-null wounds could contain higher amounts of active TGFβ. In agreement with such a possibility, TGFβ-dependent signaling, measured as the number of Smad2-P-positive cells, was increased in Ahr–/– neo-epithelium. We hypothesize that, in the absence of AhR expression, increased TGFβ activity by dermal fibroblasts could overactivate the TGFβ pathway in keratinocytes, promoting their migration and accelerating re-epithelialization.
Different sets of experimental results support our hypothesis. First, although conditioned medium from Ahr–/– dermal fibroblasts increased migration of Ahr+/+ keratinocytes, the opposite was also true and medium conditioned by Ahr+/+ dermal fibroblasts decreased migration of Ahr–/– keratinocytes. Second, exogenous TGFβ increased migration of Ahr+/+ keratinocytes, whereas a neutralizing anti-TGFβ antibody inhibited Ahr–/– keratinocyte migration. Finally, in vivo treatment of Ahr–/– wounds with the neutralizing anti-TGFβ antibody inhibited re-epithelialization and wound healing to a degree of closure that was similar to that found in Ahr+/+ wounds. Therefore, active TGFβ secreted by dermal fibroblasts can modulate keratinocyte migration and wound healing by a mechanism involving AhR. Despite this experimentally supported mechanism, the fact that media conditioned by wild-type or AhR-null keratinocytes seems to modulate, to some extent, keratinocyte migration in culture, led us to suggest that a cell-autonomous mechanism could cooperate with fibroblast-secreted TGFβ-dependent signaling. Importantly, AhR was required for TGFβ-dependent wound healing, because in vivo downregulation of AhR expression by antisense oligonucleotides significantly improved wound healing in Ahr+/+ mice as well as increasing TGFβ-responsiveness in their neo-epithelium.
In summary, we report here that lack of AhR expression in mouse skin accelerates wound healing by enhancing re-epithelialization. Consistent with the already known role of AhR in modulating TGFβ activity, our mechanism proposes that increased production of active TGFβ by dermal fibroblasts can exert a paracrine effect on the keratinocytes of the neo-epithelium that will result in increased migration along the regenerating wound. Remarkably, the fact that in vivo administration of an antisense oligonucleotide against AhR increased wound healing in wild-type mice, offers the potential for AhR knockdown to be useful in the treatment of accidental, surgical or chronic skin wounds. Therefore, AhR, as earlier suggested for osteopontin (Mori et al., 2008), could be a novel therapeutic target to improve the quality of skin repair.
Materials and Methods
Antibodies and reagents
Proliferating cell nuclear antigen (PCNA) antibody was obtained from Neomarkers. The antibody to detect mouse AhR was from ABR. Antibodies against Smad2-P and vimentin were purchased from Cell Signaling and Anacrom Diagnostics, respectively. The macrophage marker F4/80 and the neutrophil antigen Ly-6G were detected using specific antibodies from Serotec and Transduction Laboratories, respectively. Anti-TGFβ neutralizing antibody (clone 1D11) was from R&D. Antibody against α-smooth muscle actin, recombinant TGFβ protein and β-actin antibody were obtained from Sigma. TRICT-labeled anti-cytokeratin 14 (AF 64) was purchased from Covance.
Ahr+/+ and Ahr–/– mice were produced by homologous recombination in embryonic stem cells as described (Fernandez-Salguero et al., 1995). All the experimentation involving animals were performed following the guidelines established by the Animal Care and Use Committee of the University of Extremadura. Adult male mice were used at 9-12 weeks of age and had free access to water and rodent chow. Before performing surgical procedures, animals were anesthetized by an 300 μl i.p. injection of 2.5% avertin (100% stock prepared by mixing 10 g tribromoethyl alcohol in 10 ml tertiary amyl alcohol).
Ahr+/+ and Ahr–/– mice were anesthetized using avertin and their dorsal skin shaved and sterilized by topical application of povidone. Typically, two wounds of 4 mm in diameter were performed in each flank of each mouse (total of four wounds in each animal) for experiments requiring treatments. Untreated mice received a single wound in each dorsal flank. Experiments were performed using 3-4 mice of each genotype. Skin biopsies containing both dermis and epidermis were taken from each wound at days 3, 5, and 7 after surgery. Tissues were fixed at 4°C in 4% paraformaldehyde and processed for immunohistochemistry as described below. In some experiments, wounds were treated in vivo with anti-TGFβ neutralizing antibody or with AhR-antisense (ODN-As) or AhR-sense (ODN-Se) oligonucleotides as detailed below. Sections were routinely obtained at 8 μm.
Whole-skin explant culture
Whole-skin biopsies of 4 mm in diameter were obtained from the dorsal area of shaved Ahr+/+ and Ahr–/– mice. These tissues were flattened with their dermis down on tissue culture plates previously treated with 5 μg/ml collagen or 15 μg/ml fibronectin. Experiments were performed in the absence of serum to prevent cell proliferation. To obtain a kinetic of keratinocytes migration from the explants, pictures were taken every 24 hours for 7 days using a NIKON TE2000U microscope. Migration at each time point was quantified using the ImageJ software as the distance from the edge of the skin explant to the border of the keratinocyte monolayer.
Primary keratinocytes and dermal fibroblast culture
Primary keratinocytes and dermal fibroblasts were obtained from Ahr+/+ and Ahr–/– newborn mice at 2-3 days of age. After sterilization in povidone solution, mice were washed in sterile water and rinsed in 70% ethanol in PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM PO4HNa2, 1.5 mM PO4H2K pH 7.2). All four legs and the tail were removed and the complete skin dissected using forceps. The resulting skins were floated dermis down in sterile culture dishes containing 0.25% trypsin for 16-18 hours at 4°C. Next, dermis and epidermis were separated and individually minced in 2-3 ml/mouse of plating medium (E-MEM containing 4% fetal bovine serum pre-treated with chelex and 0.2 mM Ca2+ and gentamycin as antibiotic). Tissues were further digested by incubation for 45 minutes at 4°C with gentle agitation. For the preparation of keratinocytes, the digested epidermis was filtered through a 140 μm mesh to remove aggregates and undigested tissue and the cell suspension was seeded at a density of 2×106 cells in 60 mm culture plates pre-treated with 5 μg/ml collagen or 15 μg/ml fibronectin. After 24 hours, keratinocytes were washed with PBS and grown in maintenance culture medium (plating medium supplemented with 0.05 mM Ca2+) to promote proliferation and to inhibit differentiation. Dermal fibroblasts were obtained from mouse skin following the protocol used in our laboratory to isolate mouse embryo fibroblasts (Santiago-Josefat et al., 2001). Conditioned medium was produced by culturing Ahr+/+ and Ahr–/– keratinocytes and Ahr+/+ and Ahr–/– dermal fibroblasts for 72 hours in E-MEM or OptiMEM, respectively. For keratinocyte wound-closure assays, cells were allowed to reach confluence in serum-containing medium and wounds were performed with the aid of a pipette tip. After incubation for 15 hours in serum-free medium, culture plates were photographed in a NIKON TE2000U microscope.
Tissue sections were deparaffinized and gradually rehydrated to PBS. Endogenous peroxidase activity was blocked by treatment with H2O2 for 45 minutes at room temperature (1% H2O2 diluted in PBS-T: PBS containing 0.05% Triton X-100). After rinsing in PBS-T, non-specific epitopes were blocked by incubation for 1 hour in PBS-T containing 2 mg/ml gelatin and 0.1 M lysine. Sections were then incubated overnight at 4°C with the corresponding primary antibodies diluted in PBS-T-gelatin. Following extensive washing in PBS-T-gelatin, sections were incubated for 1 hour at room temperature with the appropriate biotinylated secondary antibody. After additional washing, tissues were incubated with peroxidase-conjugated streptavidin and color developed using a diaminobenzidine (DAB) solution (0.025% DAB w/v, 0.06% H2O2 v/v in PBS). Sections were dehydrated, mounted and visualized using a NIKON TE2000U microscope. PCNA, Smad2-P, Ly-6G and F4/80-reactive cells were quantified using ImageJ software. Immunofluorescence for AhR in whole skin sections was performed as described above using a mouse anti-AhR primary antibody and an Alexa Fluor 488-labeled secondary antibody.
Hematoxylin and eosin staining
Deparaffinized and rehydrated sections of skin wounds were incubated with Harris hematoxylin for 3 minutes at room temperature. After washing with tap water, eosin solution was added for 1 minute. A final washing step was performed and the tissues were dehydrated, mounted and observed in a NIKON TE2000U microscope.
Sirius red and fast green staining, and hydroxyproline enzymatic assay
The presence of total collagen in the granulation tissue was analyzed using Sirius red and fast green staining as previously described (Gascon-Barre et al., 1989; Lopez-De Leon and Rojkind, 1985; Peterson, 1993). Briefly, tissue sections were incubated in 0.04% fast green in saturated picric acid for 15 minutes at room temperature. Sections were then washed with distilled water and further incubated for 30 minutes at room temperature in 0.04% fast green containing 0.1% Sirius red in picric acid (Sigma). After washing in distilled water, sections were mounted and observed in a NIKON TE2000U microscope. Collagen content was also measured in the wounds by quantifying hydroxyproline levels as described previously (Sauzeau et al., 2007).
In vivo treatment with a TGFβ antibody and with AhR antisense oligonucleotide
To analyze how blockade of TGFβ activity affects wound healing in vivo, a neutralizing antibody for this cytokine was used. The two wounds on one flank of the dorsal skin of Ahr–/– mice were injected in their marginal area with three doses of 50 μl of neutralizing anti-TGFβ antibody at 50 μg/ml concentration. The two wounds on the opposite flank in each mouse were injected under the same conditions with sterile PBS. Wounds were dissected and processed for histology and immunohistochemistry as indicated above.
To determine how AhR downmodulation affects re-epithelialization in vivo, an AhR antisense oligonucleotide was applied to Ahr+/+ wounds. Both AhR antisense oligonucleotide and a negative control sense oligonucleotide were synthesized as described (Peters and Wiley, 1995). To prevent degradation, oligonucleotides were modified by the addition of phosphorothiolated linkages at their 5′ and 3′ ends. The sequence of the antisense oligonucleotide is fully complementary to that of the murine AhR mRNA between nucleotides 39 and 59. Sequences used were: antisense, 5′-GGGGATGGGCTTTACTGTTT-3′ and sense, 5′-AACCTTGGGTTTGGGTTTGG-3′. Before use, oligonucleotides were mixed with 30% pluronic F127 (Sigma) in sterile PBS to obtain a soft gel. The two wounds on one flank of each Ahr+/+ mice were treated with 50 μl antisense oligonucleotide whereas the two wounds on the opposite flank were treated with the same amount of control sense oligonucleotide, as described (Mori et al., 2008; Reynolds et al., 2008). Wound tissue was dissected and analyzed by histology and immunohistochemistry as indicated above.
SDS-PAGE and western immunoblotting
SDS-PAGE and western immunoblotting for TGFβ in whole-skin cell extracts were performed essentially as described (Mulero-Navarro et al., 2005).
Data are shown as mean ± s.e.m. Statistical comparison between experimental conditions was done using GraphPad Prism 4.0 software (GraphPad). Comparisons between conditions were made using unpaired Student's t-test.
We are very grateful to Francisco Javier Martin-Romero for assistance with microscopy. A detailed protocol for TGFβ treatment in vivo was kindly provided by Louise E. Reynolds and Kairbaan M. Hodivala-Dilke. This work was supported by Grants from the Spanish Ministry of Education and Sciences (SAF2005-00130 and SAF2008-00462), from the Junta de Extremadura (2PR04A060) and from the Red Temática de Investigación Cooperativa en Cáncer (RTICC) (RD06/0020/1016, Fondo de Investigaciones Sanitarias (FIS), Carlos III Institute, Spanish Ministry of Health) (to P.M.F.-S.). A.C.R. and J.M.C.-G. were supported by fellowships from the Spanish Ministry of Education and Sciences and Junta de Extremadura, respectively. All Spanish funding is co-sponsored by the European Union FEDER program.