Discussion

The results generated from this research project provide novel insight into Met function during skin regeneration. Here it is shown that Met and HGF/SF expression is induced in the hyperproliferative epithelium, which is the major structure responsible for wound closure of the skin. The expression of both, Met and HGF/SF in the hyperproliferative epithelium suggests an autocrine system, which could play an important function during the process of skin regeneration. Through the use of genetic analysis, it was indeed shown that Met is important for reepithelialization of wounds, since Met mutant keratinocytes could not contribute to the generation of the hyperproliferative epithelium. Furthermore, wound epithelium was formed only of cells that expressed a functional Met receptor. Although Met and HGF/SF are expressed in unwounded skin neither alteration in development and maintenance of epidermis, nor in the hair cycling is observed following conditional mutagenesis of the Met receptor. Analysis of cultured keratinocytes during the closure of scratchwounds, in the presence of HGF/SF, indicates that the primary deficit in the mutant cells is caused by the inability of the cells to proliferate, to reorient themselves, and to migrate into the wounded area. Thus, it is shown here the importance of Met signaling in the skin regenerative process in vivo. Met is thus the first gene, which is absolutely required for re-epithelialization of wounds. This work therefore, contributes significantly to our understanding of wound healing regulation.

The function of the Met signaling system in skin wound healing in mice was previously investigated by blocking endogenous HGF/SF with a neutralizing antibody (Yoshida et al., 2003). However, such results have to be carefully interpreted, since cross-reactivity of the antibodies with related proteins is possible. Furthermore, it is unclear, whether neutralizing antibodies gain sufficient access to the tissue. Since this model has clear limitations, genetic models are preferred to accurately study skin development and repair. Therefore, the aim of this project was to analyze the Met function in the skin using conditional knockout mice.

A conditional knockout approach was employed utilizing mice with a null mutation of the tyrosine kinase receptor Met in the skin. Keratin 14cre was used allowing for the examination of Met function specifically in the epidermis. It should be pointed out that conditional mutagenesis is essential since the inactivation of Met by conventional knockout leads to an early lethal phenotype. In addition, previous knockout studies have shown that complete deletions of the Met receptor and/or its ligand, HGF/SF, in mice resulted in an absence of the muscle groups in the limbs, reduced liver size, a defective development of the placenta, and hence early embryonic lethality. Conditional mutagenesis permits the examination of genetic analysis in adult stages, circumventing early lethal phenotypes and moreover, it allows for the detailed study of gene function in particular cell lineages.

The ability to targeting specific genes in mice is based on the combination of pluripotent embryonic stem (ES) cells and the introduction of mutations by homologous recombination (Koller et al., 1990). Moreover, site-specific recombination can be induced by the CreloxP technology. The Cre recombinase of the P1 bacteriphage recognizes specific short consensus DNA sites and catalyses recombination between them (Gu et al., 1993; Dymecki, 1996). The efficiency of recombination depends on at least two parameters: the first is the distance between the two loxP sites along a chromosome; the further the two loxP sites are apart, the less often they are likely to collide, leading to lower rates of recombination. The second parameter is position variability and the local chromatin structure; as a result, recombination can be locus dependent. A comparison of recombination frequencies of different alleles in vivo showed marked differences. In this study, recombination of the Met^flox allele in the epidermis, using Keratin14 cre, was 95%. Only 5% of keratinocytes did not undergo recombination and retained the functional Met receptor. Usually, such a recombination is acceptable and sufficient to analyse loss of function of a particular gene. The distance between loxP sites in the Met^flox construct is 1.2 kb, which is within the optimal range for Cre recombinase. Another technical problem using conditional mutagenesis is Cre expression, which can provoke mutagenesis through strand breakage or recombination at cryptic lox sites in the genome (Schmidt et al., 2000). It is essential to be aware of the potential for undesired Cre-mediated mutagenesis to influence the experimental outcome. For experiments using conditional mutagenesis, a comparison of the Cre transgene in the heterozygous allele background with the Cre transgene in the homozygous allele background is an essential control.

Many signaling pathways that involve tyrosine kinase receptors and ligands like FGF, EGF and TGFα are important for development and maintenance of the skin. In previous studies, mice deficient in the FGF receptor 2-IIIb, which is expressed in the epithelia of ectodermal and endodermal organs, show an extremely thin suprabasal layer, however with epidermal differentiation and establishment of unaffected barrier function. Furthermore, mice deficient for FGF10 (the main ligand of the FGF receptor 2-IIIb) display a similar, but less severe epidermal phenotype (Petiot et al., 2003). These data suggest that stem cell division in the basal layer is FGF/FGFR2-IIIb independent; however, receptorligand interaction is required for epidermal stratification. The EGF receptor on the other hand, is important for the regulation of the development of the epidermis and its appendages (Luetteke et al., 1994). In mammals and birds, overexpression or systemic administration of the ligand, EGF, can arrest epidermal appendage development and promote epidermal thickening (Moore et al., 1985; Kashiwagi et al., 1997). Furthermore, the EGF receptor can contribute to epithelial carcinogenesis, with elevated EGF receptor or its ligands, of which expression is reported in many types of epithelial cancers. TGFα involvement in the skin, shown in transgenic mice, reveal that overexpression leads to the development of skin tumors after treatment with the carcinogens 7,12-dimethylbenz(a)anthracene (DMBA) and/or 12-O-tetradecanoylphorbol-13 acetate (TPA) (Kiguchi et al., 1998). To date, the existence of an in vivo model to study the role of the Met receptor in the skin has not been established. However, in the present study Met and HGF/SF have been shown to be expressed in the hair follicles, and the function of the receptor in hair cycling has been analysed in vivo.

Hair follicles and other epithelial appendages develop as a result of interactions involving the embryonic ectoderm and specialized fibroblasts in the dermal papilla, which is located in the proximal mesenchyme underneath. It was shown that Met is expressed in the epithelium of hair follicles, whereas HGF/SF is produced in the dermal papilla (Lindner et al., 2000). Epithelial-mesenchymal interactions have been speculated to play an important role in hair growth. To date, the key signaling pathways involved in these epithelial-mesenchymal interactions are the Wnt/β-catenin, Sonic hedgehog, Notch, TGF-β superfamily and the FGF and EGF pathways. The secreted signaling molecules Sonic hedgehog (Shh) and bone morphogenetic proteins (Bmps) are of central importance in the regulation between proliferation and differentiation in postnatal hair follicles. Shh promotes proliferation, and Bmps promote differentiation. In the adult epidermis, Shh expression is restricted to cells at the distal portion of the growing follicle. Inhibition of Bmp signaling by the Bmp antagonist Noggin, is required for new hair growth in postnatal skin, and the growth-inducing effect of Noggin is mediated, at least in part, by Shh. Other factors that are believed to regulate the balance between epidermal proliferation and differentiation include the transcription factors Forkhead-box n1 (Foxn1) and nuclear factor κB (NF-κB) (Niemann and Watt, 2002). EGFR null mice do not survive longer than P20, and the skin and hair phenotypes could not be properly analysed in these mutant mice. Therefore, mice in which the endogenous EGFR is replaced by a human EGFR cDNA (hEGFR^KI/KI) were useful to analyse the role of the EGFR in hair follicle differentiation and cycling, because the early lethality has been overcome. After the first hair cycle, hair follicles of hEGFR^KI/KI  mice fail to enter into catagen and remain in aberrant anagen, indicating that EGFR signaling is necessary to regulate hair cycle progression (Sibilia et al., 2003). Overexpression of a dominant negative mutant of the EGFR in the skin induces striking alterations in hair cycling. These changes progressively lead to hair degeneration (Murillas et al., 1995).

The overexpression of HGF/SF in the skin is a topical debate since previous groups were able to demonstrate the importance of this growth factor in the hair cycle (Lindner et al., 2000). However, data generated from the present study cannot support such a function. Ablation of Met in the skin does not affect skin development and maintenance under normal conditions. Hair follicle cycling is unchanged in conditional Met mutants, which has been believed to be dependent on HGF/SF and Met (Lindner et al., 2000). Histological analyses of the skin sections of P1, P5, P8, P18, P20, P30, which represent all phases of first and second hair cycle, as well as 3-months old animals did not reveal any alteration. In addition, in conditional Met mutant mice, which were kept for nearly 2 years, no unusual hair loss or no other changes in the appearance of the skin were observed.

In contrast to normal hair cycling and skin development, wound healing was severely perturbed in conditional Met mutant mice. The events involved during the closure of a wound represent a classic example of a physiological process that has characteristics of both development and organ regeneration. For instance, coordinated proliferation coupled with migration and induction of cell polarity can be executed by epithelial cells during wound healing, when cells at the wound edge start dividing and moving over the provisional matrix to reconstitute tissue integrity. This invasive growth is observed during development, but in fact, is also a requirement for organ regeneration and in carcinoma progression. Therefore, an interesting aspect to investigate would be the function of Met during skin cancer, especially in metastasis stages, where migration plays important functions.

In general, HGF/SF and Met are expressed in different cell types, although they may be closely apposed to allow an exchange of signals. For instance, HGF/SF is expressed primarily in mesenchymal cells, while Met is generally expressed in nearby epithelia (Sonnenberg et al., 1993; Yang and Park, 1995; Birchmeier and Gherardi, 1998). Moreover, Met is expressed in the epithelial dermomyotome and in migrating muscle progenitor cells that derive thereof, whereas HGF/SF is expressed in mesenchymal cells close to the somites and along the route of the migrating cells (Bladt et al., 1995; Birchmeier and Gherardi, 1998). In the cerebellum, granule cells express HGF/SF, while surrounding Bergmann glia cells express the Met receptor (Jung et al., 1994). In tumours, however, autocrine HGF/SF and Met signaling is frequently observed, e.g. in epithelial cells in human prostate cancer (Kurimoto et al., 1998; Birchmeier et al., 2003). Furthermore, the data presented in this study point to an autocrine signaling function of HGF/SF/Met in normal physiological process such as the healing of skin.

Several mouse models have recently been developed in which the function of particular molecules implicated in wound healing have been altered to study their genetic involvement (Ashcroft et al., 1999; Munz et al., 1999; Krampert et al., 2004; Reynolds et al., 2005; Munz et al., 1999; Ashcroft et al., 1999). Results from these studies demonstrate that the majority of factors involve in wound healing act in a paracrine manner and mediate cross-talk between mesenchymal and epithelial cells. All mutant skin cells described in these models were able to execute wound closure, although the wound healing process was either accelerated or delayed. For example, wound closure is delayed in mice that carry a targeted mutation in the gene encoding the fibroblast growth factor 2 (FGF2), which is produced by macrophages and endothelial cells and has major effects on fibroblast proliferation and angiogenesis in the skin (Ortega et al., 1998). In contrast, wound healing is accelerated in mice that are mutant for the gene encoding transforming growth factor β1 (TGFβ1), which is released from platelets and serves as a chemoattractant for macrophages and fibroblasts (Sellheyer et al., 1993; Koch et al., 2000; Amendt et al., 2002; Reynolds et al., 2005). TGFβ has been shown to induce antiproliferative actions in processes such as liver regeneration. Smad3null mice (Smad3, a downstream component of TGFβ signaling) displayed an increased rate of reepithelialization and reduced monocyte infiltration during wound healing (Ashcroft et al., 1999). On the other hand, conditional mutation of cJun and STAT3 in the epidermis, which participate in the signaling of several growth factors and interleukins, of integrins or leptin, exhibited delayed wound closure (Li et al., 2003; Sano et al., 1999). Lastly, ablation of the placental growth factor (PlGF) gene retarded wound angiogenesis, and delayed wound healing (Carmeliet et al., 2001). In contrast, mice that lack the Met receptor in the epidermis are capable of reepithelialization under wounded conditions; this however was demonstrated to be related to the overproliferation and migration of a small proportion of residual Met positive cells. Interestingly, no other paracrine or autocrine systems can compensate for a loss of Met function in skin regeneration. Met is thus the first gene identified, which is essential for the formation of the hyperproliferative epithelium and the closure of skin wounds.

In development, HGF/SF and Met control placentation, liver growth and muscle precursor cell migration, which are processes that appear late in evolution (Birchmeier et al., 2003). Genetic analyses demonstrate that Met is also important in regeneration of adult tissues, which is shown here for skin repair after wounding, and which was previously demonstrated for liver regeneration after damage (Borowiak et al., 2004; Huh et al., 2004). The conditional mutation of Met in the liver was established using the inducible Mxcre transgene, and from these mice a portion of the liver was removed by partial hepatectomy. Liver regeneration in these mice was severely impaired, with the livertobodyweight ratio particularly affected. The observed defect included downregulation of hepatocyte proliferation and altered cell cycle progression (Borowiak et al., 2004). Impaired regeneration is also characteristic of Met mutant mice in the skin. These data suggest that Met signaling could be part of a general physiological response to tissue injury.

In the present study, only keratiniocytes that express functional, non-recombined Met are capable of participating in the hypreproliferative epithelium. It is still puzzling as to how the non-recombined keratinocytes in conditional Met mutant skin are able to contribute to wound closure. In vivo, this process was delayed, but occurred and required only twice as much time, despite the fact that the vast majority (95% mutant cells) did not participate in the process. Moreover, the overall kinetics of wound closure was similar in both, control and mutant wounds, but the rate of closure was decrease in the mutants, which may indicate that the identity of cells contributing to the hyperproliferative epithelium was the same. Reepithelialization was delayed in the skin of the conditional mutant mice (3 days after injury), but recovered fast, since the proliferation of the keratinocytes in the hyperproliferative epithelium was increased during early wound healing. The proliferation index, for instance at day 3, was quite impressive. It shows 20% increase when compared to the controls. Although it should be noted that at day 3 the relative number of proliferating Metpositive cells only reached 50%. Thus, nearly 100% of the keratinocytes that escaped recombination in the mutant population proliferate at day 3, while only 35% proliferate in the case of control. Moreover, in the control skin, proliferation of keratinocytes comes to a halt 5 days after the injury, but was still increasing at that time in the conditional Met mutant mice. By this, a selection of Met positive over Met negative cells occurs within a few days in vivo, which results in an overproliferative compensation mechanism to complete wound healing. If only 5% cells escaped recombination, 4.125 cell cycles of these would be sufficient to generate the same amount of cells that can contribute to the wound epithelium (100%) in control mice. Compensation by overproliferation in mosaic animals has been previously observed and can provide an astoundingly efficient compensatory mechanism (Riethmacher et al., 1997). Aside from these findings, another interesting aspect is the issue of the critical mass, which relates to the amount of cells necessary and sufficient to heal the wound. It is known that for some organs that this number can be relatively small, for example 500 of MTS24+ epithelial cells are enough to restore a whole thymus in nude mice (Gill et al., 2002). In other regenerative organs, e.g. the liver, the opposing situation occurs, in that the regeneration process can be triggered by only the certain number of cells, which are beyond the critical number (approximately 30% of total number, see (Michalopoulos and DeFrances, 1997). As a consequence, the recombination over time in the skin of conditional Met mutant mice was examined and revealed that the percentage of recombined cells remained constant in normal, unwounded skin.

Regeneration of the epidermis after wounding involves activation, migration and proliferation of keratinocytes from the surrounding epidermis, but also keratinocytes derived from hair follicles and sweat glands may participate in the healing process. Epidermal stem cells may also contribute to the wound epithelium since they constitute an unlimited source of cells that contribute to tissue morphogenesis, homeostasis, and also injury repair (Cowan et al., 2004). It is generally thought that each of the epithelial compartments (the interfollicular epidermis, the hair follicle, and the sebaceous gland) has its own specialized stem cells capable of sustaining tissue growth independently. Furthermore, it has been demonstrated that at times such as rapid growth or injury; hair follicle stem cells can leave their niche and contribute to the hyperproliferative epithelium. Although it has been demonstrated that hair follicle stem cells are capable of such contributions, it is less apparent that stem cells from the sebaceous gland and the interfollicular epidermis exhibit similar abilities. Additionally, studies suggest that hair follicle stem cells contribute only transiently to the interfollicular epidermis compartment, and thus are not capable of replenishing this stem cell compartment (Miller et al., 1998; Ito et al., 2005; Levy et al., 2005). The bulge cells of the skin may therefore provide another potential source of cells that reconstitute the injured epidermis. The K14-cre-mediated recombination also occurs in the hair bulge leading to cre-mediated deletion in the bulge as in the epidermis. Of note is that Met expression is not excluded from the hair follicle stem cells. Moreover, dividing cells did not form any clusters that could correspond to cells that had escaped recombination close to hair follicle remnants. This indicates that repopulating cells that escaped recombination do not derive from a particular site. It was observed that keratin 6-positive cells form continuous layers in the wound epithelium and are not only preferentially located close to the hair follicle. Repopulating cells that escaped recombination could thus originate from both, cells of the epidermis and of the hair bulge. The rapid, but transient contributing nature of the bulge cells to repopulate wounded skin is also reminiscent of the behaviour of embryonic stem cells injected into myocardium. Early after injection, embryonic cells are plentiful, but they quickly disappear (Fraidenraich et al., 2004). However, the signals leading to the recruitment of bulge cells to the epidermis after wounding are not known. Identification of these signals could eventually lead to treatment for wound and other skin disorders, such as epidermal atrophy seen in aging, by identifying therapeutic targets for enhancing the movements of bulge cells into the epidermis. It is possible that the underlying dermis also contributes to this compensatory process, with increases of other growth factors or cytokines.

A further mechanism that might interfere with regeneration in Met mutant mice is an increase in apoptosis. Met, like other receptor tyrosine kinases, provides antiapoptotic signals by activating the Akt kinase. Furthermore, previous studies reported that Met can directly interacts with the Fas receptor and can therefore prevent Fasinduced apoptosis (Wang et al., 2002). However, apoptosis rates in the regenerating skin of control and conditional Met mutant mice were comparable, indicating that the lack of the antiapoptotic function of Met is not a dominant mechanism that accounts for the impaired regeneration.

The effect of Met signaling on migration of primary keratinocytes was analysed utilizing an in vitro based system. Primary keratinocytes were isolated from the skin of control and conditional Met mutant mice and were examined for their ability to close scratchwounds. Indeed, only cultured Met positive cells initially migrated towards the scratch-wounds in the presence of HGF/SF. In addition, at later stages Met positive cells exclusively could be detected in the scratched area. It was also observed that control keratinocytes at the edges of the scratchwounds reoriented themselves, i.e. focal adhesions and stress fibres pointed rapidly towards the wound edges, and RhoA accumulated at the retracting ends of the cells, which has previously been reported for other cell types (Ridley et al., 1995; Nobes and Hall, 1999). In contrast, reorientation of the cells did not occur in scratchwounds of Met mutant keratinocytes that were exposed to HGF/SF. The major arrangement of microtubules was not centrosomal in keratinocytes, which is in agreement with a previous study (Yvon et al., 2002). Primary keratinocytes after HGF/SF stimulation extend numerous filopodia, which were packed with actin cytoskeleton. Keratinocytes can actively move forward and slide along each other into the scratch-wound. It is already established that exogenous HGF/SF is capable of accelerating wound closure in Madin-Darby canine kidney (MDCK) epithelial cell monolayers (Sponsel et al., 1994). Other ligands for receptor tyrosine kinases such as insulin and IGF however, have been reported to promote single cell keratinocyte migration while these ligands were unable to promote colony scattering (Ando and Jensen, 1993).

The mutation of Met did not disturb the reorientation of cells in response to other growth factors, and reorientation occurred in the presence of TGFα, which signals via the EGF receptor. The genetic data indicate that the signals provided by HGF/SF are the only ones capable of reorienting kerationocytes at the wound edges and contribute to the hyperproliferative epithelium in vivo. This might be reflected by a limited availability of other growth factors in vivo that can elicit similar responses as HGF/SF. Met signaling in keratinocytes activates Erk1/2, Akt, Gab1, and PAK1/2 for the motility response. The phosphorylation of PAK1/2 might be of particular importance, since PAK1/2 is a target of Rho signaling that causes remodeling of actin cytoskeleton and focal adhesion sites (Frost et al., 1998; Royal et al., 2000).

In the last decade, several growth factors have been implicated in wound healing, for instance FGFs, factors that signal via the EGF receptor, and members of the TGFβ superfamily (Werner and Grose, 2003). Ablation of these factors or their receptors in mice affected the kinetics of wound healing, but mutant cells contributed to the newly formed epithelium. As yet, Met is the only example of a receptor, in which the loss of its expression in skin cells make them unable to contribute to wound epithelium and is an essential factor required for efficient wound healing. The application of HGF/SF and/or HGF/SF variants in the therapy of wounds therefore is an attractive possibility (Bevan et al., 2004). Treatments of wounds with exogenous growth factors accelerate healing in normal animals. Topical administration of HGF/SF to wounds of diabetic mice enhanced neovascularization and formation of new tissue. It is also worth notice that HGF/SF was proven to prevent fibrotic disorders or facilitate resolution in liver cirrhosis, rental fibrosis or lung fibrosis (Matsuda et al., 1997; Mizuno et al., 1998; Ueki et al., 1999). The improvement of tissue repair processes after acute injury or chronic inflammatory disease, the reconstruction of damaged organs as well as the treatment of devastating diseases associated with tissue remodeling are major challenges in medical science. Until recently, progress in this area has been hampered by the fact that these repair and disease processes are based on complex interactions between different cell types and between cells and the extracellular matrix, which are still poorly understood. In the future, growth factors may be administrated sequentially, in combination, or at timed intervals to more closely mimic the normal healing process. The knowledge from this study will offer more information for clinical intervention and the design of new therapeutic targets for wound treatments, from molecular diagnostic of Met to therapeutics tailored to the genetic background of each patient.