- Email: [email protected]
Keratinocyte expression of transgenic hepatocyte growth factor affects melanocyte development, leading to dermal melanocytosis
Mechanisms of Development 94 (2000) 67±78
www.elsevier.com/locate/modo
Keratinocyte expression of transgenic hepatocyte growth factor affects melanocyte development, leading to dermal melanocytosis Takahiro Kunisada a,*, Hidetoshi Yamazaki a, Tomohisa Hirobe b, Shuichi Kamei a, Mitsuaki Omoteno a, Hisashi Tagaya a, Hiroaki Hemmi a, Uichi Koshimizu c, Toshikazu Nakamura c, Shin-Ichi Hayashi a a
Department of Immunology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori 683-8503, Japan b Division of Biology and Oncology, National Institute of Radiological Sciences, Anagawa, Chiba 263-8555, Japan c Division of Biochemistry, Department of Oncology, Biological Research Center, Osaka University Medical School, Suita, Osaka 565-0871, Japan Received 24 December 1999; received in revised form 8 February 2000; accepted 15 March 2000
Abstract Using the epidermis-speci®c cytokeratin 14 promoter to deliver HGF exclusively from epidermal keratinocytes, we have examined the potential of hepatocyte growth factor (HGF) secreted from the normal environment to control morphogenesis. The transgenic mice displayed a signi®cant increase of the number of melanocytes and their precursors in embryos starting not later than 16.5 dpc, and then after birth an explosive increase of dermal melanocytes started within 1 week, and these melanocytes were maintained throughout the entire life of the mice. Thus, HGF acts as a paracrine agent to promote survival, proliferation and differentiation of melanocyte precursors in vivo, and eventually causes melanocytosis. Loss of E-cadherin expression in dermal melanocyte precursors suggests that HGF caused dermal localization of melanocytes and their precursors by down-regulation of E-cadherin molecules. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Hepatocyte growth factor; Cytokeratin 14 promoter; Transgenic mouse; Melanocyte; Melanocytosis; E-cadherin; c-met; TRP-2
1. Introduction Epidermal melanocytes are derived from the neural crest, a highly migratory pluripotent embryonic cell population. Neural crest cells migrating along a lateral pathway at the outer periphery between the epidermis and the dermatomyotome of the embryo are speci®ed to differentiate into melanocytes. These lineage-restricted progenitors, called melanoblasts, invade the epidermis to ®nally colonize the skin and hair follicles. Apparently, signals from developing skin are important for proper melanocyte development (Wehrle-Haller and Weston, 1997; Molleman and Halaban, 1998; Reedy et al., 1998; Hirobe and Abe, 1999). Among over 90 gene loci that affect coat colors of mice, at least 25 loci affect melanocyte development (Bennett, 1993). Of these 25 loci, four have been identi®ed as receptors and their ligands. Kit and Mgf encode c-Kit receptor tyrosine kinase expressed on melanoblasts, and steel factor (SLF), the ligand for c-Kit produced by skin ®broblasts or kerati* Corresponding author. Tel.: 181-859-34-8270; fax: 181-859-34-8272. E-mail address: [email protected] (T. Kunisada).
nocytes, respectively (Geissler et al., 1988; Copeland et al., 1990). Ednrb encodes endothelin receptor B expressed on melanoblasts and its ligand, endothelin 3 (ET3) is encoded by the Edn3 locus (Baynash et al., 1994; Hosoda et al., 1994). The indispensable function of these genes in melanocyte development is expressed in the mutant mice of each gene as entirely white coats or white spotting of the coats due to the loss of neural crest±derived melanocytes. Although hepatocyte growth factor (HGF) and its receptor c-Met are not included in these 25 loci, promotion of proliferation, motility and melanin synthesis of human melanocytes in vitro by HGF was previously reported (Matsumoto et al., 1991; Halaban et al., 1992; Imokawa et al., 1998). Using mouse neural crest cell cultures, promotion of proliferation, survival and differentiation of melanoblasts by HGF was also detected (Kos et al., 1999). These in vitro activities of HGF on the melanocyte lineage partly overlap with those of SLF and ET3 (e.g. see discussion in Kos et al., 1999). In the developing skin, expression of HGF was reported (Defrances et al., 1992) although no evidence is available for the relevance of HGF to the normal development of melanocytes. In transgenic mice that ubiquitously
0925-4773/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0925-477 3(00)00308-7
68
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
expressed HGF due to use of a metallothionein (MT) I promoter, hyperpigmentation in skin was observed in conjunction with tumors of mesenchymal and epithelial origins. In addition, ectopic localization of melanocytes, including around the central nervous system, was found frequently (Takayama et al., 1996) and many neoplasms, including melanoma, arose spontaneously (Takayama et al., 1997). HGF was originally identi®ed as a mitogen for mature hepatocytes in primary culture (Nakamura et al., 1984; 1989), and now is well characterized as a cytokine that affects a variety of tissues in the body. It was also revealed that biological responses of target cells to HGF include not just stimulation of proliferation and/or motility, but also induction of morphogenesis, a more sophisticated mode of cell response to the secreted factor (reviewed by Birchmeier et al., 1995; Zarnegar and Michalopoulos, 1995; Matsumoto and Nakamura, 1996; Matsumoto and Nakamura, 1997). Thus, diverse phenotypes observed in different cell lineages caused by ubiquitous HGF expression are the natural consequences of the wide range of biological activities of HGF. This, in turn, makes it dif®cult to assess the effect of ubiquitously expressed HGF on particular cell lineages. Some of the phenotypes might be a direct effect of the overexpressed HGF itself, or some might be a less direct effect of ubiquitously expressed HGF overexpression (e.g. up-regulation of other factors by HGF that secondarily affect properties of other cell lineages). Autocrine activation of cells expressing c-met by transgenic HGF may induce profound physiological changes in these cells, such as neoplastic transformation (Otsuka et al., 1998). To avoid the dif®culties described above and to more precisely elucidate the function of HGF in melanocyte development in vivo, we developed transgenic mice using the human keratin 14 (k14) gene promoter to express HGF in the epidermis. In these transgenic mice, elevated expression of HGF was expected in the basal layer of the epidermis, including skin, tongue and esophagus from 13.5 days of gestation (Vassar et al., 1989; Byrne et al., 1994) and pleiotropic effects of HGF should be minimized when compared to the effects of ubiquitous expression. At least, the autocrine effect of HGF on melanoblasts or melanocytes could thus be eliminated and only the effect of normally secreted HGF from keratinocytes could be assessed. In the present investigation, we observed an increase of melanoblasts in the whole skin of embryos by at least 16.5 dpc, followed by an explosive increase of mature melanocytes after birth. In spite of the targeted expression of HGF in the epidermis, the primary site of melanoblast or melanocyte development was restricted to the dermis of interadnexal skin and the growth of these cells did not require SLF as a co-stimulating factor. Inability to express E-cadherin in melanoblasts of HGF transgenic mice was proposed as a cause of the preferential distribution of these melanoblasts in the dermis, and the roles of HGF in the distribution of dermal melanocytes in a restricted area of the normal skin are discussed. The potential use of the HGF transgenic mice as a model for human
dermal melanocytosis and the possible relationship between HGF and progression of melanomas is also discussed. 2. Results 2.1. Generation of HGF transgenic mice Three independent lines, HGFTg1, HGFTg43 and HGFTg75, which stably express HGF driven by the skinspeci®c cytokeratin 14 promoter, were generated. The extent of skin pigmentation varied from light in HGFTg1 to moderate in HGFTg43 and the heaviest pigmentation in HGFTg75 (Fig. 1B). The overall characteristics of skin pigmentation during postnatal development were similar in these three mouse lines, as described in a later section. Most of the analyses were performed using the HGFTg43 transgenic line, unless otherwise mentioned. Initially, we expected some effects of HGF on the development of skin and its associated tissues, because HGF is one of the key factors in epithelial-mesenchymal interactions during cell differentiation and organogenesis, but no clear alterations or abnormalities other than excessive skin pigmentation were recognized. Microscopic analysis also revealed no recognizable changes of the morphology of the skin or its associated tissues. Expression of the transgene was veri®ed by using rabbit anti-human HGF antibody. Skin-speci®c expression in the epidermis was observed in transgenic mice (Fig. 1C), and no signi®cant cross-reactive staining was observed in control wild-type littermates (Fig. 1D). 2.2. Pigmentation of HGF transgenic mouse skin during postnatal development Newborn HGF transgenic mice are hard to identify macroscopically, except by the pigmentation in the nasal skin (Fig. 2A,B). Then, on day 2 after birth, the external genitalia and ears become pigmented. Pigmentation of the dorsal skin becomes detectable on day 5 after birth. After 21 days, the pigmentation of the entire skin becomes visible and increases progressively until 1 month after birth, and is thereafter maintained throughout life. Hairless skin areas of the face, including the nose, mouth (Fig. 2C,D) and ears (Fig. 2G,H), are heavily pigmented. Although the tail skin is pigmented in normal mice (Fig. 2F), more extensive pigmentation was observed in HGF transgenic mice (Fig. 2E). In some cases, pigmentation in the palms, not seen in normal mice, was observed (Fig. 2E, arrowhead). The most striking phenotype of HGF transgenic mice is the distribution of melanocytes in the skin. In all three transgenic lines (HGFTg43 in Fig. 2I, HGFTg75 in Fig. 2J, HGFTg1 in Fig. 2K and control normal littermates in Fig. 2L), mature, pigmented melanocytes were observed mostly in the dermis. To discriminate the dermo-epidermal junction more clearly, skin sections were stained with anti-E-cadherin antibody (Fig. 2M±O). Although mature melanocytes have a tendency to reside close to E-cadherin-positive epidermal
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
69
Fig. 1. Transgene design, demonstration of transgenic skin phenotype, and expression of transgene product in mice. (A) Structure of k14-HGF transgene. The human cytokeratin 14 promoter is located upstream of the full-length human HGF cDNA. Speci®c coding regions of HGF are shown as `a peptide' and `b peptide' and the inserted rabbit b-globin intron and k14 polyadenylation site are indicated as `intron' and `poly a signal', respectively. (B) Shaved back skins of three k14-HGF transgenic mouse lines (HGFTg43/1, HGFTg75/1, HGFTg1/1) and wild type (1/1). Note that HGFTg1/1 is very lightly pigmented compared to the other two transgenic lines. Immunohistological identi®cation of HGF in the epidermis of HGFTg43/1 transgenic mice (C). No signi®cant staining was observed in 1/1 mice (D). Arrowheads indicate the junction of dermis and epidermis. All transgenic mouse lines were heterozygotes of the transgenes.
areas, only limited numbers of melanocytes were found in the epidermis. In previously reported SCF transgenic mice, in which targeted expression in epidermal cells was induced by the k14 promoter (Kunisada et al., 1998a), melanocytes were localized mostly in the epidermis (Fig. 2N). Normal littermate mice contained very few mature melanocytes (Fig. 2O). To quantitate the numbers of melanocytes and their precursors during development, the numbers of DOPApositive and DOPA-premelanin-positive cells were counted in HGF transgenic mice. DOPA-positive cells are known to represent differentiated melanocytes and DOPA-premelanin-positive cells consist of melanocyte precursors plus differentiated melanocytes (Hirobe, 1982). DOPA-positive and DOPA-premelanin-positive cells in the dorsal skin of 11-day-old transgenic and littermate mice are shown in Fig. 3B±E. The results presented in Fig. 3A do not include melanocytes inside hair follicles. In the dermis, an increase of differentiated melanocytes continued until day 23 after birth, at which time 300-fold more melanocytes were observed in the dorsal skin of transgenic mice than in that of normal littermates. Number of DOPA-premelanin-positive cells was nearly the same as the number of DOPApositive differentiated melanocytes, indicating that HGF does not selectively increase the number of melanoblasts, but rather promotes growth and/or differentiation of mature melanocytes from melanoblasts. The epidermal layer of HGF transgenic mice also contains more melanocytes than
that of to normal littermate mice; however, the total number per 0.1 mm 2 is at most several fold greater than that of normal littermates. It should also be noted that the number of epidermal melanocytes decrease with age both in HGF transgenic mice and control littermates, indicating that the primary location of proliferative melanocytes is the dermis of the skin. 2.3. Melanocyte precursors in HGF transgenic mice As shown in the previous section, most of the cells increased in the dermis of postnatal HGF transgenic mice are DOPA-positive differentiated melanocytes. We next investigated melanocyte lineage cells in the skin of embryonic-stage mice. Assuming that the number of DOPA-premelanin-positive cells minus the number of DOPA-positive cells corresponds to that of melanocyte precursors, these precursors increased signi®cantly in 16.5 or 17.5 dpc HGF transgenic mice in both the dermis and epidermis (Fig. 4). Thus, as observed in postnatal skin, overexpression of HGF increased the number of differentiated melanocytes in the embryonic dermal skin, although the absolute number of differentiated melanocytes was not as remarkably increased as in postnatal skin. This may highlight the relative increase of melanocyte precursors in both dermal and epidermal skin. To further investigate the melanocyte precursors or melanoblasts induced by the transgenic HGF, we used tyrosi-
70
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 2. Phenotype of k14-HGF transgenic mice. Newborn (A,B), and 21-day-old (C±F) transgenic mice, designated as Tg/1, were photographed with their normal littermates, designated as 1/1. The arrowhead in B shows pigmentation of the nose characteristic of HGF transgenic mice. The arrowhead in E shows pigmentation in the palm not observed in normal mice. The transgenic mice used were heterozygous HGFTg43 transgenic mouse lines. Hematoxylin and eosinstained sections of 2-month-old mice show melanocytes in the dermis of HGF transgenic mice (I; HGFTg43, J; HGFTg75, K; HGFTg1) in which virtually no melanocytes were found in the normal littermate mice (L). Anti-E cadherin immuno¯uorescent staining of 12-day-old skin clearly showed the dermal localization of melanocytes in HGFTg43 (M) mice, while melanocytes were abundant mostly in the epidermis of SLF transgenic mice (N) as originally reported in a previous paper (Kunisada et al., 1998a) and no melanocytes were detected in 1/1 littermates (O). Asterisks show hair follicles.
nase-related protein 2 (TRP2) and c-Kit receptor tyrosine kinase as markers for melanocyte lineages. It has been established that melanoblasts expressing both TRP2 and cKit appear as early as 10.5 dpc embryo (Steel et al., 1992). We performed immunohistological analysis of 17.5 dpc skin by using anti-TRP2 and anti-c-Kit antibodies, and the results are summarized in Fig. 5A. In the skin of control littermates, TRP2-positive cells were mostly found in hair follicles and epidermis. Ninety-eight percent of TRP2-positive cells in hair follicles are also c-Kit-positive, and 78% of TRP2-positive cells in the epidermis are c-Kit-positive. In HGF transgenic skin, TRP2-positive cells are mostly found in the dermis and hair follicles, although some TRP2-posi-
tive cells exist in the epidermis. The number of dermal TRP2-positive cells in HGF transgenic mice is ®ve times greater than that observed in the epidermis, and more than twice greater than that observed in normal epidermis. Unexpectedly, 95% of TRP2-positive cells observed in HGF transgenic dermis were c-Kit-negative (also see Fig. 5B,C). As expected, c-Met-positive cells were detected at a frequency almost equal with that of TRP-2-positive cells in the dermis of HGF transgenic mice (Fig. 5D). In the presence of the peptide corresponding to carboxyl-terminal sequence of c-Met used to raise anti-c-Met antibody, no positive staining was observed (Fig. 5E). Although melanoblasts with a c-Kit-negative have until now not been
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
71
Fig. 3. Changes in the number of DOPA positive cells (melanocytes) and DOPA-premelanin-positive cells (melanocytes plus melanoblasts) in the dorsal skin of HGF transgenic mice after birth. (A) At each time point, samples were prepared from two transgenic or normal littermate mice, and the numbers of melanocytes in the dermis and epidermis were counted in three arbitrarily chosen 0.1 mm 2 areas in a specimen prepared from each mouse. In the graph showing the number of dermal melanocytes (left), the scale for the vertical axis of HGF transgenic mice is indicated on the left and that of littermate mice is indicated on the right. For example, the number of dermal melanocytes of day 0 HGF transgenic mice is 162.8 ^ 55.8, which exceeds that of age-matched littermate mice, 8.9 ^ 3.0. Representative images of sections of the dorsal skin from 11-day-old HGF transgenic mouse stained with DOPA (B) or DOPA-premelanin (C), are shown, along with those of DOPA (D) and DOPA-premelanin (E)-stained sections from littermate mice. Large arrows indicate dermal cells positive to DOPA (B,D) or DOPA-premelanin reaction (C,E). Small arrows indicate epidermal cells positive to DOPA (B,D) or DOPA-premelanin reaction (C,E). Numerous cells positive to the reactions are observed in the dermis of transgenic mice.
reported, TRP2 positive, c-Kit negative, c-Met-positive cells of the dermis might be the most plausible candidates for dermal melanocyte precursors of HGF transgenic mice. 2.4. Hyperpigmentation of the skin of HGF transgenic mice is not in¯uenced by steel factor SLF is known to be indispensable for the development of melanocytes from their early precursors. In addition, synergistic effects of SLF with other growth factors or cytokines have been reported in precursors for hematopoietic cells (Galli et al., 1993; Broudy, 1997). Although the endogenous supply of SCF is not suf®cient to support melanocyte devel-
opment in the epidermis after birth, a suboptimal amount of SLF might be suf®cient for the melanocyte development observed in HGF transgenic skin. To test the possibility that HGF-induced hyperpigmentation is supported by the endogenous SLF, we blocked SLF in vivo by injecting anti-c-Kit antibody, ACK2 (Nishikawa et al., 1991). As shown in Fig. 6, while administration of ACK2 after birth completely prevented hair pigmentation at day 9 after birth both in HGF transgenic and control littermate mice, the pigmentation of the nose, ears, tail and the external genitalia region was not affected in HGF transgenic mice (Fig. 6A). On day 20 after birth, the entire skin of ACK2-treated HGF transgenic mice was pigmented, while their hairs were still
72
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 4. Changes in the number of DOPA-positive cells and DOPA-premelanin-positive cells in the dorsal skin of HGF transgenic and control embryos. The number of cells per 0.1 mm 2 was calculated as described in the legend for Fig. 3.
unpigmented (Fig. 6B). Histological analysis showed an increase of dermal melanocytes on day 9 (Fig. 6C) and further progression on day 20 (Fig. 6D) in ACK2-treated HGF transgenic skin. In the normal littermate mice, no melanocytes were found on day 20 (Fig. 6E). Thus, melanocyte proliferation in HGF transgenic skin does not require SLF as an essential cofactor. In the previous section, we identi®ed dermal TRP2-positive c-Kit-negative cells as possible melanocyte precursors of HGF transgenic mice. It is highly likely that these cells not expressing c-Kit are not affected by ACK2, indicating that these TRP2-positive cKit-negative cells are direct precursors for the melanocytes developed after ACK2 treatment. 2.5. Possible mechanisms for the dermal localization of melanocytes in HGF transgenic mice Recently, Nishimura et al. reported that the migration of melanoblasts from dermis to epidermis might be controlled by the cadherin molecule (Nishimura et al., 1999). They observed that melanocytes found in the dermis of developing skin did not express E-cadherin and those that had entered the epidermis all expressed E-cadherin, suggesting the importance of E-cadherin on the surface of melanoblasts for their entry into the epidermis. When 17.5 dpc HGF transgenic skin was double-stained with anti-TRP2 antibody and anti-E-cadherin antibody, TRP2-positive melanoblasts observed in the dermis of HGF transgenic mice did not express E-cadherin (indicated by arrowheads in Fig. 7A± C). Some melanoblasts in the epidermis, if not all, expressed E-cadherin (shown by arrows in Fig. 7A±C, and at higher magni®cation in Fig. 7D±F). As previously noted, transgenic mouse induced to express epidermal SLF showed preferential distribution of melanoblasts or melanocytes in the epidermis (also refer to Kunisada et al., 1998b). If HGF
somehow suppressed E-cadherin in developing melanoblasts, their epidermal entry might well be perturbed. 2.6. Expression of HGF in pinna of the ear and trunk skin of normal mice In the areas bearing reduced amount of hair, such as tail, nose, ear and external genitalia, dermal melanocytes were observed far more frequently than in skin regions with more hair (Silvers, 1979). If HGF is responsible for the dermal distribution of melanocytes, as discussed above, elevated levels of HGF production in these regions would be expected. Semi-quantitative RT-PCR analysis for the expression of HGF in normal mice showed that a signi®cantly larger amount of Hgf message was expressed in the ear than in trunk skin (Fig. 8). The existence of dermal melanocytes in the ears of normal mice was shown in Fig. 4J of our previous report (Kunisada et al., 1998b). Therefore, the dermal melanocytes induced by exogenous HGF might not be the outcome of non-physiological conditions generated in transgenic skin, and similar conditions may exist in some part of the wild-type mice. 3. Discussion Based on the fact that HGF participates in many types of epithelial-mesenchymal interactions during development, we at ®rst expected multiple effects on skin development in the HGF transgenic mice. In fact, it was reported that HGF affects the growth of keratinocytes, including hair (Shimosaka et al., 1995; Lindner et al., 2000). However, at least in the three independently isolated transgenic lines described here, targeted expression of HGF in the skin did not affect the development of the skin or its appendixes, as judged by close morphological inspection. As shown by our results presented in Fig. 8, Hgf mRNA was detected in the
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
73
Fig. 5. The number of melanocyte lineage cells expressing TRP2 or c-Kit markers in 17.5 dpc embryos. Double-label immuno¯uorescence were performed for sections prepared from Hgf/1 or 1/1 embryos, and the numbers of TRP2-positive cells (red) and c-Kit-positive cells (green) were counted. The numbers in each column in (A) represent the sum of 20 arbitrarily chosen ®elds of microscope (20£ objective lens with 3.3£ internal lens and 10£ eyepiece). In (B), representative images of TRP2-positive cells in the dermis (arrowheads) and the hair follicles (arrows) of HGF transgenic mouse are shown. The identical ®eld observed for c-Kit expression revealed that TRP2-positive cells in the dermis did not express c-Kit, while those in hair follicles did express c-Kit (C). Note that brightly stained round cells in (C) stained only with c-Kit are the mast cells of the hematopoietic cell lineage. In (D), 17.5 dpc HGF transgenic skin was stained with anti-c-Met antibody. In the presence of the peptide used to raise the antibody, positive signals of the dermis disappeared, as shown in (E).
skin of normal mice, although the level was less than that in liver or lung. Therefore, although HGF expressed in the skin may have important functions for skin development, an exogenous elevated supply of HGF from keratinocytes of the basal cell layer of the epidermis does not affect normal development of the skin, including that of keratinocytes themselves, except for the development of melanocytes. Fibroblasts are thought to be the major source of HGF expressed in the skin, and melanoblast cell lines or mature melanocytes are known to express c-met (Zarnegar and Michalopoulos, 1995; Kunisada et al., unpublished observations); however, in normal mice, the numbers of epidermal melanoblasts or melanocytes decrease starting at day 4 after birth and reach extremely low levels after 1 month of age
(Hirobe, 1984). Thus, although HGF might possibly regulate early precursors of melanocytes at embryonic stages, it might not participate in the development of melanoblasts or melanocytes after birth. Alternatively, the amount of HGF expressed in the dermis of the skin might be too low to induce a response in melanoblasts or melanocytes, or the site of expression might be inappropriate. It is also possible that normal melanoblasts or melanocytes lose responsiveness to HGF gradually after birth, or there is even the extreme possibility that melanocyte development in vivo never requires HGF. A null mutation of HGF could not provide insight into these questions because the HGF2/2 embryos die at around 13.5 dpc, probably due to the impaired formation of the placenta. Therefore, we tested
74
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 6. Effect of anti-c-Kit antibody on the postnatal pigmentation of HGF transgenic mice. The function-blocking anti-c-Kit monoclonal antibody ACK2 was injected on days 0, 1, 2, 4, 6, and 8 after birth. On day 9, while the hairs of the ACK2-treated mice were white in both HGF transgenic and 1/1 mice, pigmentation of the skin, nose, tail and ears progressed as in 1/1 control mice (A). On day 20, while the hair was still white, other parts of the skin were heavily pigmented (B). Dermal melanocytes were found in the histological sections of ACK2-treated HGF transgenic mice day 9 (C) and day 20 (D). No melanocytes were found in ACK2-treated normal littermates on day 20 (E).
Fig. 7. Expression of E-cadherin in the melanocyte lineage cells of HGF transgenic mice. Sections prepared from 17.5 dpc HGF transgenic skin were stained with TRP2 (red) and E-cadherin (green) double-label immuno¯uorescence. Images were taken for TRP2 (A), E-cadherin (B) and TRP2 and E-cadherin (C). TRP2-positive cells in the dermis (shown by arrowheads in (A)) did not express E-cadherin, as shown in (B). On the other hand, epidermal TRP2-positive cells (shown by arrows in (A,D)) expressed E-cadherin, as shown by arrows in (B,C,E,F).
the potential of HGF as a regulator for melanocyte development in vivo by exogenously supplying HGF molecules in the skin, in which melanocytes develop in normal mice. The effect of the HGF transgene on melanocyte development was ®rst detected in 14.5 dpc embryos as a slight increase of the number of dermal DOPA-premelanin-positive melanoblasts and/or melanocytes (Fig. 4). In contrast, no apparent differences in melanoblast number or location were reported between c-met 2/2 and 1/1 embryos examined from 10.5 to 13.5 dpc (Kos et al., 1999). In any case, HGF±c-Met signaling appears not to very signi®cantly affect melanocyte development up to 14.5 dpc. In the case of k14-SLF transgenic mice, no signi®cant difference of number or distribution of TRP2-positive melanoblasts between transgenic and normal mice was reported at 14.5 dpc (Kunisada et al., 1998b). Thus, assuming that the k14 promoter is active at 9.5 dpc, and its activity greatly increases by 13.5±14.5 dpc (Byrne et al., 1994), exogenous HGF and also SLF will exert minimal effects on the ongoing development of melanoblasts around this stage. SLF was shown to be a crucial factor for maintaining melanocyte development at this stage (Nishikawa et al., 1991), and a suf®cient amount of SLF might be supplied endogenously from the developing skin. This situation might also be true for endogenous HGF. In 16.5 and 17.5 dpc embryos, a signi®cant increase of melanoblasts represented by
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 8. Expression of endogenous HGF in normal mice. RNA was prepared from the tissues indicated and cDNAs synthesized from 1, 1/8, and 1/64 mg of each RNA preparation were subjected to RT-PCR.
DOPA-premelanin-positive cells was observed. Concomitantly, an increase of TRP2-positive melanoblasts was observed in 17.5 dpc HGF transgenic mice. At this stage, most of the TRP2-positive melanoblasts in HGF transgenic mice were observed to be localized in the dermis, whereas those of control mice were mostly found in the epidermis. In addition, these TRP2-positive cells do not express c-Kit on their surface. Therefore, even if a supply of endogenous SLF from the local environment is expected, these precursors do not respond to SLF, as clearly shown by the resistance of melanocyte development to ACK2 administration in HGF transgenic mice. Then, after birth, the number of DOPApositive differentiated melanocytes increased dramatically, and on day 23 after birth, nearly 1500 melanocytes were counted per 0.1 mm 2, which is an increase of about 300-fold compared to the number found in normal skin. At this stage, the number of DOPA-premelanin-positive cells, which include undifferentiated melanoblasts plus differentiated melanocytes, is almost equal to the number of cells positive only for DOPA, i.e. differentiated melanocytes only. Taking all these observations into account, what is occurring in the melanocyte development of HGF transgenic mice appears to be as follows: ®rst, TRP2-positive c-Kit negative precursors increase in the dermis around 16.5±17.5 dpc. These precursors may be the committed stem cells of melanocytes with self-renewing activity. After birth, these precursors are driven to differentiate into melanocytes, possibly through proliferation of intermediate-stage cells between the above described TRP2-positive c-Kit-negative precursors and mature DOPA-positive differentiated melanocytes. This process taking place after birth depends solely on transgenic HGF, and not on endogenous SLF. Transgenic HGF thus supports the survival, proliferation and differentiation of melanocytes in vivo, and the ability to induce melanocyte
75
development is attained solely by the expression of the HGF transgene in the epidermis. One of the striking features of k14-SLF transgenic mice was that extended distribution of melanocytes was observed in a number of sites, including oral epithelium and footpads, where neither melanocytes nor melanoblasts are normally detected (Kunisada et al., 1998b). Close inspection of a number of HGF transgenic mice did not reveal any melanocytes migrated out to the oral epithelium, gums for the incisor teeth, or pawpads, where no melanoblasts were observed during embryogenesis (Kunisada et al., 1998b; Kunisada, unpublished observation). In the nose, ears, lips, and external genitalia, where melanocytes were present in the wild-type mice, hyperpigmentation was common to every HGF transgenic mice (Fig. 2). HGF could thus expand the number of melanocytes in regions to which melanoblasts had already migrated; however, it could not extend their migration to regions where melanoblasts are not normally distributed. Rarely, a cluster of melanocytes was found in the palm of HGF transgenic mice (Fig. 2E). These might have arisen from an exceptional melanoblast that migrated out from the normal pathway and was expanded by transgenic HGF. Most of the melanocytes were localized in the dermis of HGF transgenic mice, in spite of the fact that HGF was expressed in the epidermis. Although melanocytes were not evenly spread in dermis, but instead tended to be located more closely to the epidermis, as seen in Fig. 2J,M, a major portion of the melanocytes did not invade the epidermis. As suggested in the cases of adhesion of epidermal Langerhans cells (Tang et al., 1993), human melanocytes (Tang et al., 1994) and T lymphocytes (Cepek et al., 1994; Lee et al., 1994) to keratinocytes in vitro, E-cadherin expression both in the adherent cells and target keratinocytes is likely to mediate epidermal localization of these cells. Recently, elevated expression of E-cadherin in the melanoblasts of mouse embryos during the entry of the melanoblasts into the epidermis was reported (Nishimura et al., 1999). Modulation of cadherin-mediated cell-cell interaction by HGF was reported in keratinocyte cell lines, in which the concentration of adhesion molecules, including E-cadherin, was reduced upon HGF treatment (Watabe et al., 1993) and co-localization of c-met with E-cadherin and beta-catenin was observed in a tumor cell line (Hiscox and Jiang, 1999). Considering these observations, suppression of the expression of adhesion molecules, including E-cadherin, by the transgenic HGF may explain the inability of melanoblasts to invade the epidermis in HGF transgenic mice. Clinically, abnormal growth of melanocytes in the dermis is well documented, and is termed dermal melanocytosis (Ortonne and Nordlund, 1998). The pathogenesis of dermal melanocytosis is thought to be heterogeneous, re¯ecting multiple factors or genes that regulate the development of melanocytes (Hori and Takayama, 1988). Apparently, the HGF transgenic mice described here show the symptom of dermal melanocytosis observed in human skin. Thus, these
76
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
HGF transgenic mice may provide an animal model for dermal melanocytosis. More than 200 HGF transgenic mice have been raised for over 3 months, and 20 of them were selected to be raised more than 1 year. Although not all of them underwent close inspection by autopsy, we found no apparent melanomas in HGF transgenic mice, even in those that lived more than a year. No apparent shortening of the life span was observed, either. These observations suggest that HGF expressed in keratinocytes does not markedly contribute to malignant tumor induction. On the other hand, systemic expression of HGF under the control of a MT promoter induced development of tumors including melanomas (Otsuka et al., 1998). The number of melanocytes found in the skin of mice that have systemic expression of HGF is comparable to that of the k14-HGF transgenic mice used in the present study; therefore, different modes of HGF expression affect proliferation and differentiation of melanocytes similarly, but affect tumorigenesis differently. Although expression of cytokines such as HGF may induce a broad range of physiological changes in transgenic mice irrespective of promoters used for transgene expression, strictly keratinocyte-speci®c expression of HGF and the lack of apparent incidence of melanomas in k14-HGF transgenic mice may suggest the role of HGF supplied by the autocrine mechanism for the neoplastic change of melanocytes to melanoma. It should be noted that keratinocytes are expressing c-met as shown in Fig. 5D and autocrine activation of HGF signaling is expected in the keratinocytes of k14-HGF transgenic mice. Lower expression level of k14 promoter driven HGF might not be adequate to promote tumor development of keratinocytes even under the possible autocrine activation of HGF signaling. While systemic expression of HGF was reported to cause inappropriate localization of melanocyte in skin of extremities including footpads (Takayama et al., 1996), k14-HGF transgenic mice were mostly lacking melanocytes in the footpads, as described in the previous section. This suggest that paracrine HGF signaling alone do not stimulate migration of melanocytes in vivo. It is likely that not only the complicated biological process such as neoplastic transformation but the normal biological process such as cellular migration are affected differently by the different mode of HGF expression in vivo.
4. Experimental procedures 4.1. Generation and analysis of transgenic animals A full-length human HgfcDNA cloned into pBluescript (pHHGF, Honda et al., 1995) was cut by HpaI at its 3 0 end, ligated with BamHI linker, and then digested with BamHI. Gel-puri®ed Hgf cDNA with BamHI sites on both ends was cloned into the BamHI site of the K14 expression vector, provided by Dr. E. Fuchs (Vassar et al., 1989). To excise
transgenes from the plasmid, in which the Hgf sequence contains a KpnI site, DNA was ®rst cut with ClaI, and then cut partially with KpnI to obtain the fragment shown in Fig. 1A. After separation by agarose gel electrophoresis, the fragment was puri®ed by CsCl/Ethidium Bromide centrifugation, and then DNA was dissolved in 10 mM Tris (pH 7.5), 0.1 mM EDTA at the concentration of 2 mg/ml, and injected into fertilized oocytes collected from pregnant (C57BL/6xDBA/2) F1 mice as described (Kimura et al., 1995). Integration of transgenes was screened by PCR of genomic DNA with K14-speci®c primers, and then veri®ed by Southern hybridization with Hgf cDNA used for transgene construction. 4.2. Histology and immunohistochemistry For DOPA and DOPA-premelanin staining, pieces of skin were excised from the dorsal side of transgenic and littermate mice and ®xed with formalin in phosphate buffer (pH 7.0) for 16 h at 48C, and subjected to the DOPA reaction. The samples were embedded in paraf®n, and then subjected to the DOPA-premelanin reaction as described (Mishima, 1960; Hirobe, 1982). For immunohistochemistry, tissues were embedded in polyester wax, sectioned (7 mm) and stained as described (Yoshida et al., 1993; Kunisada et al., 1998b). The primary antibodies used were; rabbit antihuman HGF polyclonal antibody (Matsumoto et al., 1996); rabbit anti-mouse TRP2 polyclonal antibody (provided by V. Hearing; Tsukamoto et al., 1992), diluted 300£ in 5% normal goat serum in PBS for use as blocking solution; a monoclonal rat anti-mouse c-Kit antibody (ACK2 hybridoma cells were provided by S.-I. Nishikawa; Nishikawa et al., 1991), used at the concentration of 20 mg/ ml; monoclonal rat anti-mouse E-cadherin (ECCD2, conjugated with biotin, provided by S.-I. Nishikawa; Shirayoshi et al., 1986), used at 10 mg/ml. For double-labeled immuno¯uorescent staining, sections were incubated with biotinlabeled or non-labeled primary antibody overnight at room temperature, washed three times, and then incubated with ¯uorescence-labeled secondary antibodies for 2 h. Then, after they were washed three times, sections were incubated with streptoavidin-rhodamine or rhodamine-labeled antibodies bind to the ®rst antibody (purchased from Vector and diluted as speci®ed by the manufacturer). Because this protocol did not work well for rabbit anti-mouse c-met polyclonal antibody (R&D systems), cryostat sections were prepared from 17.5 dpc embryos. After they were immersed successively in 12, 15, and 18% sucrose dissolved in PBS for 30 min each, the skins were embedded in Tissue Tek (Miles), and frozen in liquid nitrogen. Cryostat sections of 10-mm thickness were rehydrated in PBS, and then processed as described for the sections from polyester wax-embedded specimens. 4.3. RT-PCR analysis Total RNAs were isolated from tissues from B6 mice
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
using Isogen (Nippon Gene). Random hexamer-primed cDNA was synthesized from 4 mg of RNA from each tissue using Super Script reverse transcriptase (GIBCO-BRL) in a 25-ml reaction mixture as speci®ed by the manufacturer. Two and one-half microliter aliquots or serial dilutions of the reaction mixture were then subjected to PCR in 50 ml of reaction mixture containing Taq polymerase and the buffer supplied by the manufacturer (TOYOBO). The primer sequences and the condition of the PCR cycle for mouse HGF were described previously (Ohmichi et al., 1998). For the control, a G3PDH sequence was ampli®ed using the previously described conditions and primers (Yamane et al., 1999). After PCR, 15-ml aliquots of the reaction mixture were separated by agarose gel electrophoresis, transferred to nylon membranes, and hybridized with mouse HGF and human G3PDH probes to obtain sequence-speci®c signals. Acknowledgements We thank Drs. E. Fuchs, V. Hearing, S.-I. Nishikawa for providing materials. We also thank Drs. H. Yoshida, E. Nishimura for discussion. This work was supported in part by grants from the Ministry of Education, Science, Sports and Culture of Japan, Special Coordination Funds for Promoting Science and Technology from the Science and Technology Agency of Japan and Yamanouchi Foundation for Research on Metabolic Disorders. References Baynash, A.G., Hosoda, K., Giaid, A., Richardson, J.A., Emoto, N., Hammer, R.E., Yanagisawa, M., 1994. Interaction of endothelin-3 with endothelin-B receptor is essential for development of epidermal melanocytes and enteric neurons. Cell 79, 1277±1285. Bennett, D.C., 1993. Genetics, development, and malignancy of melanocytes. Int. Rev. Cytol. 146, 191±260. Birchmeier, C., Meyer, D., Riethmecher, D., 1995. Factors controlling growth, motility, and morphogenesis of normal and malignant epithelial cells. Int. Rev. Cytol. 160, 221±266. Broudy, V.C., 1997. Stem cell factor and hematopoiesis. Blood 90, 1345± 1364. Byrne, C., Tainsky, M., Fuchs, E., 1994. Programming gene expression in developing epidermis. Development 120, 2369±2383. Cepek, K.L., Shaw, S.K., Parker, C.M., Russell, G.J., Morrow, J.S., Rimm, D.L., Brenner, M.B., 1994. Adhesion between epithelial cells and T lymphocytes mediated by E-cadherin and the aEb7 integrin. Nature 372, 190±193. Copeland, N.G., Gilbert, D.J., Cho, B., Donovan, P.J., Jenkins, N.A., Cosman, D., Anderson, D., Lyman, S.D., Williams, D.E., 1990. Mast cell growth factor maps near the Steel locus on mouse chromosome 10 and is deleted in a number of Steel alleles. Cell 63, 175±183. Defrances, M.C., Wolf, H.K., Michalopoulos, G.K., Zarnegar, R., 1992. The presence of hepatocyte growth factor in the developing rat. Development 116, 387±395. Galli, S.J., Zsebo, K.M., Geissler, E.N., 1993. The kit ligand. Stem cell factor. Adv. Immunol. 55, 1±96. Geissler, E.N., Ryan, M.A., Houseman, D.E., 1988. The dominant-white spotting (W) locus of the mouse encodes the c-kit proto-oncogene. Cell 55, 185±192. Halaban, R., Rubin, J.S., Funasaka, Y., Cobb, M., Boulton, T., Faletto, D.,
77
Rosen, E., Chan, A., Yoko, K., White, W., Cook, C., Moellman, G., 1992. Met and hepatocyte growth factor/scatter factor signal transduction in normal melanocytes and melanoma cells. Oncogene 7, 2195± 2206. Hirobe, T., 1982. Genes involved in regulating the melanocyte and melanoblast-melanocyte populations in the epidermis of newborn mouse skin. J. Exp. Zool. 223, 257±264. Hirobe, T., 1984. Histochemical survey of the distribution of the epidermal melanoblasts and melanocytes in the mouse during fetal and postnatal periods. Anat. Rec. 208, 589±594. Hirobe, T., Abe, H., 1999. Genetic and epigenetic control of the proliferation and differentiation of mouse epidermal melanocytes in culture. Pigment Cell Res. 12, 147±163. Hiscox, S., Jiang, W.G., 1999. Association of the HGF/SF receptor, c-met, with the cell-surface adhesion molecule, E-cadherin, and catenins in human tumor cells. Biochem. Biophys. Res. Commum. 261, 406±411. Honda, S., Kagoshima, M., Wakana, A., Tohyama, M., Matsumoto, K., Nakamura, T., 1995. Localization and functional coupling of HGF and c-Met/HGF receptor in rat brain: implication as neurotrophic factor. Mol. Brain Res. 32, 197±210. Hori, Y., Takayama, O., 1988. Circumscribed dermal melanoses: classi®cation and histologic features. Dermatol. Clin. 6, 315±326. Hosoda, K., Hammer, R.E., Richardson, J.A., Baynash, A.G., Cheung, J., Giaid, A., Yanagisawa, M., 1994. Targeted and natural (Piebald-lethal) mutations of endothelin B receptor gene produce megacolon associated with spotted coat color in mice. Cell 79, 1276±1297. Imokawa, G., Yada, Y., Morisaki, N., Kimura, M., 1998. Biological characterization of human ®broblast-derived mitogenic factors for human melanocytes. Biochem. J. 330, 1235±1239. Kimura, S., Niwa, H., Moriyama, M., Araki, K., Abe, K., Miike, T., Yamamura, K., 1995. Improvement of germ cell line transmission by targeting b-garactosidase to nuclei in transgenic mice. Dev. Growth Differ. 38, 87±97. Kos, L., Aronzon, A., Takayama, H., Maina, F., Ponzetto, C., Merlino, G., Pavan, W., 1999. Hepatocyte growth factor/Scatter factor-MET signaling in neural crest-derived melanocyte development. Pigment Cell Res. 12, 13±21. Kunisada, T., Lu, S-Z., Yoshida, H., Nishikawa, S., Nishikawa, S.-I., Mizoguchi, M., Hayashi, S.I., Tyrrell, L., Williams, D.A., Wang, X., Longley, B.J., 1998a. Murine cutaneous mastocytosis and epidermal melanocytosis induced by keratinocyte expression of transgenic stem cells factor. J. Exp. Med. 187, 1565±1573. Kunisada, T., Yoshida, H., Yamazaki, H., Miyamoto, A., Hemmi, H., Nishimura, E., Shultz, L.D., Nishikawa, S.-I., Hayashi, S.I., 1998b. Transgene expression of steel factor in the basal layer of epidermis promotes survival, proliferation, differentiation and migration of melanocyte precursors. Development 125, 2915±2923. Lee, M.G., Sharrow, S.O., Farr, A.G., Singer, A., Udey, M.C., 1994. Expression of the homotypic adhesion molecule E-cadherin by immature murine thymocytes and thymic epithelial cells. J. Immunol. 152, 5653±5659. Lindner, G., Menrad, A., Gherardi, E., Merlino, G., Eelker, P., Handjiski, B., Roloff, B., Paus, R., 2000. Involvement of hepatocyte growth factor/ scatter factor and Met receptor signaling in hair follicle morphogenesis and cycling. FASEB J. 14, 319±332. Matsumoto, K., Nakamura, T., 1996. Emerging multipotent aspects of hepatocyte growth factor. J. Biochem. 119, 591±600. Matsumoto, K., Nakamura, T., 1997. Hepatocyte growth factor (HGF) as a tissue organizer for organogenesis and regeneration. Biochem. Biophys. Res. Commun. 239, 639±644. Matsumoto, K., Tajima, H., Nakamura, T., 1991. Hepatocyte growth factor is a potent stimulator of human melanocyte DNA synthesis and growth. Biochem. Biophys. Res. Commun. 176, 45±51. Matsumoto, K., Date, K., Shimura, H., Nakamura, T., 1996. Acquisition of invasive phenotype in gallbladder cancer cells via mutual interaction of stromal ®broblasts and cancer cells as mediated by hepatocyte growth factor. Jpn. J. Cancer Res. 87, 702±710.
78
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Mishima, Y., 1960. New technique for comprehensive demonstration of melanin, premelanin, and tyrosinase sites; combined DOPA-premelanin reaction. J. Invest. Dermatol. 34, 355±360. Molleman, G., Halaban, R., 1998. Growth-factor receptors and signal transduction regulating the proliferation and differentiation of melanocytes. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 135±149. Nakamura, T., Nawa, K., Ichihara, A., 1984. Partial puri®cation and characterization of hepatocyte growth factor from serum of hepatectomozed rats. Biochem. Biophys. Res. Commun. 122, 1450±1459. Nakamura, T., Nishizawa, T., Hagiya, M., Seki, T., Shimonishi, M., Sugimura, A., Tashiro, K., Shimizu, S., 1989. Molecular cloning and expression of human hepatocyte growth factor. Nature 342, 440±443. Nishikawa, S., Kusakabe, M., Yoshinaga, K., Ogawa, M., Hayashi, S.I., Kunisada, T., Era, T., Sakakura, T., Nishikawa, S.-I., 1991. In utero manipulation of coat color formation by a monoclonal anti-c-kit antibody: two distinct waves of c-kit-dependency during melanocyte development. EMBO J. 10, 2111±2118. Nishimura, E., Yoshida, H., Kunisada, T., Nishikawa, S.-I., 1999. Cadherin expression associated with melanocyte migration and diversi®cation re¯ects the microenvironmental setting. Dev. Biol. 215, 155±166. Ohmichi, H., Koshimizu, U., Matsumoto, K., Nakamura, T., 1998. Hepatocyte growth factor (HGF) acts as a mesenchyme-derived morphogenic factor during fetal lung development. Development 125, 1315± 1324. Ortonne, J.-P., Nordlund, J.J., 1998. Mechanisms that cause abnormal skin color. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 489±502. Otsuka, T., Takayama, H., Sharp, R., Celli, G., Larochelle, W.J., Bottaro, D.P., Ellmore, N., Vieira, W., Owens, J.W., Anver, M., Merlino, G., 1998. c-Met autocrine activation induces development of malignant melanoma and acquisition of the metastatic phenotype. Cancer Res. 15, 5157±5167. Reedy, M.V., Parichy, D.M., Erickson, C.A., Mason, K.A., Frost-Mason, S.K., 1998. The regulation of melanoblast migration and differentiation. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 75±95. Shimosaka, S., Tsuboi, R., Jindo, T., Imai, R., Takamori, K., Rubin, J.S., Ogawa, H., 1995. Hepatocyte growth factor/scatter factor expressed in follicular papilla cells stimulates human hair growth in vitro. J. Cell. Physiol. 165, 333±338.
Shirayoshi, Y., Nose, A., Iwasaki, K., Takeichi, M., 1986. N-linked oligosaccharides are not involved in the function of a cell-cell binding glycoprotein E-cadherin. Cell Struct. Funct. 11, 245±252. Silvers, W.K., 1979. The Coat Colors of Mice: A Model for Mammalian Gene Action and Interaction, Springer±Verlag, New York. Steel, K.P., Davidson, D.R., Jackson, I.J., 1992. TRP-2/DT, a new early melanoblast marker, shows that steel growth factor (c-kit ligand) is a survival factor. Development 115, 1111±1119. Takayama, H., LaRochelle, W.J., Anver, M., Bockman, D.E., Merlino, G., 1996. Scatter factor/hepatocyte growth factor as a regulator of skeletal muscle and neural crest development. Proc. Natl. Acad. Sci. USA 93, 5866±5871. Takayama, H., LaRochelle, W.J., Sharp, R., Otsuka, T., Kriebel, P., Anver, M., Aaronson, S.A., Merlino, G., 1997. Diverse tumorigenesis associated with aberrant development in mice overexpressing hepatocyte growth factor/scatter factor. Proc. Natl. Acad. Sci. USA 94, 701±706. Tang, A., Amagai, M., Granger, L.G., Stanley, J.R., Udey, M.C., 1993. Adhesion of epidermal Langerhans cells to keratinocytes mediated by E-cadherin. Nature 361, 82±85. Tang, A., Eller, M.S., Hara, M., Yaar, M., Hirohashi, S., Gilchrest, B.A., 1994. E-cadherin is the major mediator of human melanocyte adhesion to keratinocytes in vitro. J. Cell. Sci. 107, 983±992. Tsukamoto, K., Jackson, I.J., Urabe, K., Montague, P.M., Hearing, V.J., 1992. A second tyrosinase-related protein. TRP-2, is a melanogenic enzyme termed DOPAchrome tautomerase. EMBO J. 11, 519±526. Vassar, R., Rosenberg, M., Ross, S., Tyner, A., Fuchs, E., 1989. Tissuespeci®c and differentiation-speci®c expression of a human K14 keratin gene in transgenic mice. Proc. Natl. Acad. Sci. USA 86, 1563±1567. Watabe, M., Matsumoto, K., Nakamura, T., Takeichi, M., 1993. Effect of hepatocyte growth factor on cadherin-mediated cell-cell interaction. Cell Struct. Funct. 18, 117±124. Wehrle-Haller, B., Weston, J.A., 1997. Receptor tyrosine kinase-dependent neural crest migration in response to differentially localized growth factors. BioEssays 19, 337±345. Zarnegar, R., Michalopoulos, G.K., 1995. The many faces of hepatocyte growth factor: from hepatopoiesis to hematopoiesis. J. Cell. Biol. 129, 1177±1180. Yamane, T., Hayashi, S.I., Mizoguchi, M., Yamazaki, H., Kunisada, T., 1999. Derivation of melanocytes from embryonic stem cells in culture. Dev. Dyn. 216, 450±458. Yoshida, H., Nishikawa, S.-I., Okamura, H., Sakakura, T., Kusakabe, M., 1993. The role of c-kit proto-oncogene during melanocyte development in mouse. In vivo approach by the in utero microinjection of anti-c-kit antibody. Dev. Growth Differ. 35, 209±220.
www.elsevier.com/locate/modo
Keratinocyte expression of transgenic hepatocyte growth factor affects melanocyte development, leading to dermal melanocytosis Takahiro Kunisada a,*, Hidetoshi Yamazaki a, Tomohisa Hirobe b, Shuichi Kamei a, Mitsuaki Omoteno a, Hisashi Tagaya a, Hiroaki Hemmi a, Uichi Koshimizu c, Toshikazu Nakamura c, Shin-Ichi Hayashi a a
Department of Immunology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori 683-8503, Japan b Division of Biology and Oncology, National Institute of Radiological Sciences, Anagawa, Chiba 263-8555, Japan c Division of Biochemistry, Department of Oncology, Biological Research Center, Osaka University Medical School, Suita, Osaka 565-0871, Japan Received 24 December 1999; received in revised form 8 February 2000; accepted 15 March 2000
Abstract Using the epidermis-speci®c cytokeratin 14 promoter to deliver HGF exclusively from epidermal keratinocytes, we have examined the potential of hepatocyte growth factor (HGF) secreted from the normal environment to control morphogenesis. The transgenic mice displayed a signi®cant increase of the number of melanocytes and their precursors in embryos starting not later than 16.5 dpc, and then after birth an explosive increase of dermal melanocytes started within 1 week, and these melanocytes were maintained throughout the entire life of the mice. Thus, HGF acts as a paracrine agent to promote survival, proliferation and differentiation of melanocyte precursors in vivo, and eventually causes melanocytosis. Loss of E-cadherin expression in dermal melanocyte precursors suggests that HGF caused dermal localization of melanocytes and their precursors by down-regulation of E-cadherin molecules. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Hepatocyte growth factor; Cytokeratin 14 promoter; Transgenic mouse; Melanocyte; Melanocytosis; E-cadherin; c-met; TRP-2
1. Introduction Epidermal melanocytes are derived from the neural crest, a highly migratory pluripotent embryonic cell population. Neural crest cells migrating along a lateral pathway at the outer periphery between the epidermis and the dermatomyotome of the embryo are speci®ed to differentiate into melanocytes. These lineage-restricted progenitors, called melanoblasts, invade the epidermis to ®nally colonize the skin and hair follicles. Apparently, signals from developing skin are important for proper melanocyte development (Wehrle-Haller and Weston, 1997; Molleman and Halaban, 1998; Reedy et al., 1998; Hirobe and Abe, 1999). Among over 90 gene loci that affect coat colors of mice, at least 25 loci affect melanocyte development (Bennett, 1993). Of these 25 loci, four have been identi®ed as receptors and their ligands. Kit and Mgf encode c-Kit receptor tyrosine kinase expressed on melanoblasts, and steel factor (SLF), the ligand for c-Kit produced by skin ®broblasts or kerati* Corresponding author. Tel.: 181-859-34-8270; fax: 181-859-34-8272. E-mail address: [email protected] (T. Kunisada).
nocytes, respectively (Geissler et al., 1988; Copeland et al., 1990). Ednrb encodes endothelin receptor B expressed on melanoblasts and its ligand, endothelin 3 (ET3) is encoded by the Edn3 locus (Baynash et al., 1994; Hosoda et al., 1994). The indispensable function of these genes in melanocyte development is expressed in the mutant mice of each gene as entirely white coats or white spotting of the coats due to the loss of neural crest±derived melanocytes. Although hepatocyte growth factor (HGF) and its receptor c-Met are not included in these 25 loci, promotion of proliferation, motility and melanin synthesis of human melanocytes in vitro by HGF was previously reported (Matsumoto et al., 1991; Halaban et al., 1992; Imokawa et al., 1998). Using mouse neural crest cell cultures, promotion of proliferation, survival and differentiation of melanoblasts by HGF was also detected (Kos et al., 1999). These in vitro activities of HGF on the melanocyte lineage partly overlap with those of SLF and ET3 (e.g. see discussion in Kos et al., 1999). In the developing skin, expression of HGF was reported (Defrances et al., 1992) although no evidence is available for the relevance of HGF to the normal development of melanocytes. In transgenic mice that ubiquitously
0925-4773/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0925-477 3(00)00308-7
68
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
expressed HGF due to use of a metallothionein (MT) I promoter, hyperpigmentation in skin was observed in conjunction with tumors of mesenchymal and epithelial origins. In addition, ectopic localization of melanocytes, including around the central nervous system, was found frequently (Takayama et al., 1996) and many neoplasms, including melanoma, arose spontaneously (Takayama et al., 1997). HGF was originally identi®ed as a mitogen for mature hepatocytes in primary culture (Nakamura et al., 1984; 1989), and now is well characterized as a cytokine that affects a variety of tissues in the body. It was also revealed that biological responses of target cells to HGF include not just stimulation of proliferation and/or motility, but also induction of morphogenesis, a more sophisticated mode of cell response to the secreted factor (reviewed by Birchmeier et al., 1995; Zarnegar and Michalopoulos, 1995; Matsumoto and Nakamura, 1996; Matsumoto and Nakamura, 1997). Thus, diverse phenotypes observed in different cell lineages caused by ubiquitous HGF expression are the natural consequences of the wide range of biological activities of HGF. This, in turn, makes it dif®cult to assess the effect of ubiquitously expressed HGF on particular cell lineages. Some of the phenotypes might be a direct effect of the overexpressed HGF itself, or some might be a less direct effect of ubiquitously expressed HGF overexpression (e.g. up-regulation of other factors by HGF that secondarily affect properties of other cell lineages). Autocrine activation of cells expressing c-met by transgenic HGF may induce profound physiological changes in these cells, such as neoplastic transformation (Otsuka et al., 1998). To avoid the dif®culties described above and to more precisely elucidate the function of HGF in melanocyte development in vivo, we developed transgenic mice using the human keratin 14 (k14) gene promoter to express HGF in the epidermis. In these transgenic mice, elevated expression of HGF was expected in the basal layer of the epidermis, including skin, tongue and esophagus from 13.5 days of gestation (Vassar et al., 1989; Byrne et al., 1994) and pleiotropic effects of HGF should be minimized when compared to the effects of ubiquitous expression. At least, the autocrine effect of HGF on melanoblasts or melanocytes could thus be eliminated and only the effect of normally secreted HGF from keratinocytes could be assessed. In the present investigation, we observed an increase of melanoblasts in the whole skin of embryos by at least 16.5 dpc, followed by an explosive increase of mature melanocytes after birth. In spite of the targeted expression of HGF in the epidermis, the primary site of melanoblast or melanocyte development was restricted to the dermis of interadnexal skin and the growth of these cells did not require SLF as a co-stimulating factor. Inability to express E-cadherin in melanoblasts of HGF transgenic mice was proposed as a cause of the preferential distribution of these melanoblasts in the dermis, and the roles of HGF in the distribution of dermal melanocytes in a restricted area of the normal skin are discussed. The potential use of the HGF transgenic mice as a model for human
dermal melanocytosis and the possible relationship between HGF and progression of melanomas is also discussed. 2. Results 2.1. Generation of HGF transgenic mice Three independent lines, HGFTg1, HGFTg43 and HGFTg75, which stably express HGF driven by the skinspeci®c cytokeratin 14 promoter, were generated. The extent of skin pigmentation varied from light in HGFTg1 to moderate in HGFTg43 and the heaviest pigmentation in HGFTg75 (Fig. 1B). The overall characteristics of skin pigmentation during postnatal development were similar in these three mouse lines, as described in a later section. Most of the analyses were performed using the HGFTg43 transgenic line, unless otherwise mentioned. Initially, we expected some effects of HGF on the development of skin and its associated tissues, because HGF is one of the key factors in epithelial-mesenchymal interactions during cell differentiation and organogenesis, but no clear alterations or abnormalities other than excessive skin pigmentation were recognized. Microscopic analysis also revealed no recognizable changes of the morphology of the skin or its associated tissues. Expression of the transgene was veri®ed by using rabbit anti-human HGF antibody. Skin-speci®c expression in the epidermis was observed in transgenic mice (Fig. 1C), and no signi®cant cross-reactive staining was observed in control wild-type littermates (Fig. 1D). 2.2. Pigmentation of HGF transgenic mouse skin during postnatal development Newborn HGF transgenic mice are hard to identify macroscopically, except by the pigmentation in the nasal skin (Fig. 2A,B). Then, on day 2 after birth, the external genitalia and ears become pigmented. Pigmentation of the dorsal skin becomes detectable on day 5 after birth. After 21 days, the pigmentation of the entire skin becomes visible and increases progressively until 1 month after birth, and is thereafter maintained throughout life. Hairless skin areas of the face, including the nose, mouth (Fig. 2C,D) and ears (Fig. 2G,H), are heavily pigmented. Although the tail skin is pigmented in normal mice (Fig. 2F), more extensive pigmentation was observed in HGF transgenic mice (Fig. 2E). In some cases, pigmentation in the palms, not seen in normal mice, was observed (Fig. 2E, arrowhead). The most striking phenotype of HGF transgenic mice is the distribution of melanocytes in the skin. In all three transgenic lines (HGFTg43 in Fig. 2I, HGFTg75 in Fig. 2J, HGFTg1 in Fig. 2K and control normal littermates in Fig. 2L), mature, pigmented melanocytes were observed mostly in the dermis. To discriminate the dermo-epidermal junction more clearly, skin sections were stained with anti-E-cadherin antibody (Fig. 2M±O). Although mature melanocytes have a tendency to reside close to E-cadherin-positive epidermal
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
69
Fig. 1. Transgene design, demonstration of transgenic skin phenotype, and expression of transgene product in mice. (A) Structure of k14-HGF transgene. The human cytokeratin 14 promoter is located upstream of the full-length human HGF cDNA. Speci®c coding regions of HGF are shown as `a peptide' and `b peptide' and the inserted rabbit b-globin intron and k14 polyadenylation site are indicated as `intron' and `poly a signal', respectively. (B) Shaved back skins of three k14-HGF transgenic mouse lines (HGFTg43/1, HGFTg75/1, HGFTg1/1) and wild type (1/1). Note that HGFTg1/1 is very lightly pigmented compared to the other two transgenic lines. Immunohistological identi®cation of HGF in the epidermis of HGFTg43/1 transgenic mice (C). No signi®cant staining was observed in 1/1 mice (D). Arrowheads indicate the junction of dermis and epidermis. All transgenic mouse lines were heterozygotes of the transgenes.
areas, only limited numbers of melanocytes were found in the epidermis. In previously reported SCF transgenic mice, in which targeted expression in epidermal cells was induced by the k14 promoter (Kunisada et al., 1998a), melanocytes were localized mostly in the epidermis (Fig. 2N). Normal littermate mice contained very few mature melanocytes (Fig. 2O). To quantitate the numbers of melanocytes and their precursors during development, the numbers of DOPApositive and DOPA-premelanin-positive cells were counted in HGF transgenic mice. DOPA-positive cells are known to represent differentiated melanocytes and DOPA-premelanin-positive cells consist of melanocyte precursors plus differentiated melanocytes (Hirobe, 1982). DOPA-positive and DOPA-premelanin-positive cells in the dorsal skin of 11-day-old transgenic and littermate mice are shown in Fig. 3B±E. The results presented in Fig. 3A do not include melanocytes inside hair follicles. In the dermis, an increase of differentiated melanocytes continued until day 23 after birth, at which time 300-fold more melanocytes were observed in the dorsal skin of transgenic mice than in that of normal littermates. Number of DOPA-premelanin-positive cells was nearly the same as the number of DOPApositive differentiated melanocytes, indicating that HGF does not selectively increase the number of melanoblasts, but rather promotes growth and/or differentiation of mature melanocytes from melanoblasts. The epidermal layer of HGF transgenic mice also contains more melanocytes than
that of to normal littermate mice; however, the total number per 0.1 mm 2 is at most several fold greater than that of normal littermates. It should also be noted that the number of epidermal melanocytes decrease with age both in HGF transgenic mice and control littermates, indicating that the primary location of proliferative melanocytes is the dermis of the skin. 2.3. Melanocyte precursors in HGF transgenic mice As shown in the previous section, most of the cells increased in the dermis of postnatal HGF transgenic mice are DOPA-positive differentiated melanocytes. We next investigated melanocyte lineage cells in the skin of embryonic-stage mice. Assuming that the number of DOPA-premelanin-positive cells minus the number of DOPA-positive cells corresponds to that of melanocyte precursors, these precursors increased signi®cantly in 16.5 or 17.5 dpc HGF transgenic mice in both the dermis and epidermis (Fig. 4). Thus, as observed in postnatal skin, overexpression of HGF increased the number of differentiated melanocytes in the embryonic dermal skin, although the absolute number of differentiated melanocytes was not as remarkably increased as in postnatal skin. This may highlight the relative increase of melanocyte precursors in both dermal and epidermal skin. To further investigate the melanocyte precursors or melanoblasts induced by the transgenic HGF, we used tyrosi-
70
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 2. Phenotype of k14-HGF transgenic mice. Newborn (A,B), and 21-day-old (C±F) transgenic mice, designated as Tg/1, were photographed with their normal littermates, designated as 1/1. The arrowhead in B shows pigmentation of the nose characteristic of HGF transgenic mice. The arrowhead in E shows pigmentation in the palm not observed in normal mice. The transgenic mice used were heterozygous HGFTg43 transgenic mouse lines. Hematoxylin and eosinstained sections of 2-month-old mice show melanocytes in the dermis of HGF transgenic mice (I; HGFTg43, J; HGFTg75, K; HGFTg1) in which virtually no melanocytes were found in the normal littermate mice (L). Anti-E cadherin immuno¯uorescent staining of 12-day-old skin clearly showed the dermal localization of melanocytes in HGFTg43 (M) mice, while melanocytes were abundant mostly in the epidermis of SLF transgenic mice (N) as originally reported in a previous paper (Kunisada et al., 1998a) and no melanocytes were detected in 1/1 littermates (O). Asterisks show hair follicles.
nase-related protein 2 (TRP2) and c-Kit receptor tyrosine kinase as markers for melanocyte lineages. It has been established that melanoblasts expressing both TRP2 and cKit appear as early as 10.5 dpc embryo (Steel et al., 1992). We performed immunohistological analysis of 17.5 dpc skin by using anti-TRP2 and anti-c-Kit antibodies, and the results are summarized in Fig. 5A. In the skin of control littermates, TRP2-positive cells were mostly found in hair follicles and epidermis. Ninety-eight percent of TRP2-positive cells in hair follicles are also c-Kit-positive, and 78% of TRP2-positive cells in the epidermis are c-Kit-positive. In HGF transgenic skin, TRP2-positive cells are mostly found in the dermis and hair follicles, although some TRP2-posi-
tive cells exist in the epidermis. The number of dermal TRP2-positive cells in HGF transgenic mice is ®ve times greater than that observed in the epidermis, and more than twice greater than that observed in normal epidermis. Unexpectedly, 95% of TRP2-positive cells observed in HGF transgenic dermis were c-Kit-negative (also see Fig. 5B,C). As expected, c-Met-positive cells were detected at a frequency almost equal with that of TRP-2-positive cells in the dermis of HGF transgenic mice (Fig. 5D). In the presence of the peptide corresponding to carboxyl-terminal sequence of c-Met used to raise anti-c-Met antibody, no positive staining was observed (Fig. 5E). Although melanoblasts with a c-Kit-negative have until now not been
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
71
Fig. 3. Changes in the number of DOPA positive cells (melanocytes) and DOPA-premelanin-positive cells (melanocytes plus melanoblasts) in the dorsal skin of HGF transgenic mice after birth. (A) At each time point, samples were prepared from two transgenic or normal littermate mice, and the numbers of melanocytes in the dermis and epidermis were counted in three arbitrarily chosen 0.1 mm 2 areas in a specimen prepared from each mouse. In the graph showing the number of dermal melanocytes (left), the scale for the vertical axis of HGF transgenic mice is indicated on the left and that of littermate mice is indicated on the right. For example, the number of dermal melanocytes of day 0 HGF transgenic mice is 162.8 ^ 55.8, which exceeds that of age-matched littermate mice, 8.9 ^ 3.0. Representative images of sections of the dorsal skin from 11-day-old HGF transgenic mouse stained with DOPA (B) or DOPA-premelanin (C), are shown, along with those of DOPA (D) and DOPA-premelanin (E)-stained sections from littermate mice. Large arrows indicate dermal cells positive to DOPA (B,D) or DOPA-premelanin reaction (C,E). Small arrows indicate epidermal cells positive to DOPA (B,D) or DOPA-premelanin reaction (C,E). Numerous cells positive to the reactions are observed in the dermis of transgenic mice.
reported, TRP2 positive, c-Kit negative, c-Met-positive cells of the dermis might be the most plausible candidates for dermal melanocyte precursors of HGF transgenic mice. 2.4. Hyperpigmentation of the skin of HGF transgenic mice is not in¯uenced by steel factor SLF is known to be indispensable for the development of melanocytes from their early precursors. In addition, synergistic effects of SLF with other growth factors or cytokines have been reported in precursors for hematopoietic cells (Galli et al., 1993; Broudy, 1997). Although the endogenous supply of SCF is not suf®cient to support melanocyte devel-
opment in the epidermis after birth, a suboptimal amount of SLF might be suf®cient for the melanocyte development observed in HGF transgenic skin. To test the possibility that HGF-induced hyperpigmentation is supported by the endogenous SLF, we blocked SLF in vivo by injecting anti-c-Kit antibody, ACK2 (Nishikawa et al., 1991). As shown in Fig. 6, while administration of ACK2 after birth completely prevented hair pigmentation at day 9 after birth both in HGF transgenic and control littermate mice, the pigmentation of the nose, ears, tail and the external genitalia region was not affected in HGF transgenic mice (Fig. 6A). On day 20 after birth, the entire skin of ACK2-treated HGF transgenic mice was pigmented, while their hairs were still
72
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 4. Changes in the number of DOPA-positive cells and DOPA-premelanin-positive cells in the dorsal skin of HGF transgenic and control embryos. The number of cells per 0.1 mm 2 was calculated as described in the legend for Fig. 3.
unpigmented (Fig. 6B). Histological analysis showed an increase of dermal melanocytes on day 9 (Fig. 6C) and further progression on day 20 (Fig. 6D) in ACK2-treated HGF transgenic skin. In the normal littermate mice, no melanocytes were found on day 20 (Fig. 6E). Thus, melanocyte proliferation in HGF transgenic skin does not require SLF as an essential cofactor. In the previous section, we identi®ed dermal TRP2-positive c-Kit-negative cells as possible melanocyte precursors of HGF transgenic mice. It is highly likely that these cells not expressing c-Kit are not affected by ACK2, indicating that these TRP2-positive cKit-negative cells are direct precursors for the melanocytes developed after ACK2 treatment. 2.5. Possible mechanisms for the dermal localization of melanocytes in HGF transgenic mice Recently, Nishimura et al. reported that the migration of melanoblasts from dermis to epidermis might be controlled by the cadherin molecule (Nishimura et al., 1999). They observed that melanocytes found in the dermis of developing skin did not express E-cadherin and those that had entered the epidermis all expressed E-cadherin, suggesting the importance of E-cadherin on the surface of melanoblasts for their entry into the epidermis. When 17.5 dpc HGF transgenic skin was double-stained with anti-TRP2 antibody and anti-E-cadherin antibody, TRP2-positive melanoblasts observed in the dermis of HGF transgenic mice did not express E-cadherin (indicated by arrowheads in Fig. 7A± C). Some melanoblasts in the epidermis, if not all, expressed E-cadherin (shown by arrows in Fig. 7A±C, and at higher magni®cation in Fig. 7D±F). As previously noted, transgenic mouse induced to express epidermal SLF showed preferential distribution of melanoblasts or melanocytes in the epidermis (also refer to Kunisada et al., 1998b). If HGF
somehow suppressed E-cadherin in developing melanoblasts, their epidermal entry might well be perturbed. 2.6. Expression of HGF in pinna of the ear and trunk skin of normal mice In the areas bearing reduced amount of hair, such as tail, nose, ear and external genitalia, dermal melanocytes were observed far more frequently than in skin regions with more hair (Silvers, 1979). If HGF is responsible for the dermal distribution of melanocytes, as discussed above, elevated levels of HGF production in these regions would be expected. Semi-quantitative RT-PCR analysis for the expression of HGF in normal mice showed that a signi®cantly larger amount of Hgf message was expressed in the ear than in trunk skin (Fig. 8). The existence of dermal melanocytes in the ears of normal mice was shown in Fig. 4J of our previous report (Kunisada et al., 1998b). Therefore, the dermal melanocytes induced by exogenous HGF might not be the outcome of non-physiological conditions generated in transgenic skin, and similar conditions may exist in some part of the wild-type mice. 3. Discussion Based on the fact that HGF participates in many types of epithelial-mesenchymal interactions during development, we at ®rst expected multiple effects on skin development in the HGF transgenic mice. In fact, it was reported that HGF affects the growth of keratinocytes, including hair (Shimosaka et al., 1995; Lindner et al., 2000). However, at least in the three independently isolated transgenic lines described here, targeted expression of HGF in the skin did not affect the development of the skin or its appendixes, as judged by close morphological inspection. As shown by our results presented in Fig. 8, Hgf mRNA was detected in the
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
73
Fig. 5. The number of melanocyte lineage cells expressing TRP2 or c-Kit markers in 17.5 dpc embryos. Double-label immuno¯uorescence were performed for sections prepared from Hgf/1 or 1/1 embryos, and the numbers of TRP2-positive cells (red) and c-Kit-positive cells (green) were counted. The numbers in each column in (A) represent the sum of 20 arbitrarily chosen ®elds of microscope (20£ objective lens with 3.3£ internal lens and 10£ eyepiece). In (B), representative images of TRP2-positive cells in the dermis (arrowheads) and the hair follicles (arrows) of HGF transgenic mouse are shown. The identical ®eld observed for c-Kit expression revealed that TRP2-positive cells in the dermis did not express c-Kit, while those in hair follicles did express c-Kit (C). Note that brightly stained round cells in (C) stained only with c-Kit are the mast cells of the hematopoietic cell lineage. In (D), 17.5 dpc HGF transgenic skin was stained with anti-c-Met antibody. In the presence of the peptide used to raise the antibody, positive signals of the dermis disappeared, as shown in (E).
skin of normal mice, although the level was less than that in liver or lung. Therefore, although HGF expressed in the skin may have important functions for skin development, an exogenous elevated supply of HGF from keratinocytes of the basal cell layer of the epidermis does not affect normal development of the skin, including that of keratinocytes themselves, except for the development of melanocytes. Fibroblasts are thought to be the major source of HGF expressed in the skin, and melanoblast cell lines or mature melanocytes are known to express c-met (Zarnegar and Michalopoulos, 1995; Kunisada et al., unpublished observations); however, in normal mice, the numbers of epidermal melanoblasts or melanocytes decrease starting at day 4 after birth and reach extremely low levels after 1 month of age
(Hirobe, 1984). Thus, although HGF might possibly regulate early precursors of melanocytes at embryonic stages, it might not participate in the development of melanoblasts or melanocytes after birth. Alternatively, the amount of HGF expressed in the dermis of the skin might be too low to induce a response in melanoblasts or melanocytes, or the site of expression might be inappropriate. It is also possible that normal melanoblasts or melanocytes lose responsiveness to HGF gradually after birth, or there is even the extreme possibility that melanocyte development in vivo never requires HGF. A null mutation of HGF could not provide insight into these questions because the HGF2/2 embryos die at around 13.5 dpc, probably due to the impaired formation of the placenta. Therefore, we tested
74
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 6. Effect of anti-c-Kit antibody on the postnatal pigmentation of HGF transgenic mice. The function-blocking anti-c-Kit monoclonal antibody ACK2 was injected on days 0, 1, 2, 4, 6, and 8 after birth. On day 9, while the hairs of the ACK2-treated mice were white in both HGF transgenic and 1/1 mice, pigmentation of the skin, nose, tail and ears progressed as in 1/1 control mice (A). On day 20, while the hair was still white, other parts of the skin were heavily pigmented (B). Dermal melanocytes were found in the histological sections of ACK2-treated HGF transgenic mice day 9 (C) and day 20 (D). No melanocytes were found in ACK2-treated normal littermates on day 20 (E).
Fig. 7. Expression of E-cadherin in the melanocyte lineage cells of HGF transgenic mice. Sections prepared from 17.5 dpc HGF transgenic skin were stained with TRP2 (red) and E-cadherin (green) double-label immuno¯uorescence. Images were taken for TRP2 (A), E-cadherin (B) and TRP2 and E-cadherin (C). TRP2-positive cells in the dermis (shown by arrowheads in (A)) did not express E-cadherin, as shown in (B). On the other hand, epidermal TRP2-positive cells (shown by arrows in (A,D)) expressed E-cadherin, as shown by arrows in (B,C,E,F).
the potential of HGF as a regulator for melanocyte development in vivo by exogenously supplying HGF molecules in the skin, in which melanocytes develop in normal mice. The effect of the HGF transgene on melanocyte development was ®rst detected in 14.5 dpc embryos as a slight increase of the number of dermal DOPA-premelanin-positive melanoblasts and/or melanocytes (Fig. 4). In contrast, no apparent differences in melanoblast number or location were reported between c-met 2/2 and 1/1 embryos examined from 10.5 to 13.5 dpc (Kos et al., 1999). In any case, HGF±c-Met signaling appears not to very signi®cantly affect melanocyte development up to 14.5 dpc. In the case of k14-SLF transgenic mice, no signi®cant difference of number or distribution of TRP2-positive melanoblasts between transgenic and normal mice was reported at 14.5 dpc (Kunisada et al., 1998b). Thus, assuming that the k14 promoter is active at 9.5 dpc, and its activity greatly increases by 13.5±14.5 dpc (Byrne et al., 1994), exogenous HGF and also SLF will exert minimal effects on the ongoing development of melanoblasts around this stage. SLF was shown to be a crucial factor for maintaining melanocyte development at this stage (Nishikawa et al., 1991), and a suf®cient amount of SLF might be supplied endogenously from the developing skin. This situation might also be true for endogenous HGF. In 16.5 and 17.5 dpc embryos, a signi®cant increase of melanoblasts represented by
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Fig. 8. Expression of endogenous HGF in normal mice. RNA was prepared from the tissues indicated and cDNAs synthesized from 1, 1/8, and 1/64 mg of each RNA preparation were subjected to RT-PCR.
DOPA-premelanin-positive cells was observed. Concomitantly, an increase of TRP2-positive melanoblasts was observed in 17.5 dpc HGF transgenic mice. At this stage, most of the TRP2-positive melanoblasts in HGF transgenic mice were observed to be localized in the dermis, whereas those of control mice were mostly found in the epidermis. In addition, these TRP2-positive cells do not express c-Kit on their surface. Therefore, even if a supply of endogenous SLF from the local environment is expected, these precursors do not respond to SLF, as clearly shown by the resistance of melanocyte development to ACK2 administration in HGF transgenic mice. Then, after birth, the number of DOPApositive differentiated melanocytes increased dramatically, and on day 23 after birth, nearly 1500 melanocytes were counted per 0.1 mm 2, which is an increase of about 300-fold compared to the number found in normal skin. At this stage, the number of DOPA-premelanin-positive cells, which include undifferentiated melanoblasts plus differentiated melanocytes, is almost equal to the number of cells positive only for DOPA, i.e. differentiated melanocytes only. Taking all these observations into account, what is occurring in the melanocyte development of HGF transgenic mice appears to be as follows: ®rst, TRP2-positive c-Kit negative precursors increase in the dermis around 16.5±17.5 dpc. These precursors may be the committed stem cells of melanocytes with self-renewing activity. After birth, these precursors are driven to differentiate into melanocytes, possibly through proliferation of intermediate-stage cells between the above described TRP2-positive c-Kit-negative precursors and mature DOPA-positive differentiated melanocytes. This process taking place after birth depends solely on transgenic HGF, and not on endogenous SLF. Transgenic HGF thus supports the survival, proliferation and differentiation of melanocytes in vivo, and the ability to induce melanocyte
75
development is attained solely by the expression of the HGF transgene in the epidermis. One of the striking features of k14-SLF transgenic mice was that extended distribution of melanocytes was observed in a number of sites, including oral epithelium and footpads, where neither melanocytes nor melanoblasts are normally detected (Kunisada et al., 1998b). Close inspection of a number of HGF transgenic mice did not reveal any melanocytes migrated out to the oral epithelium, gums for the incisor teeth, or pawpads, where no melanoblasts were observed during embryogenesis (Kunisada et al., 1998b; Kunisada, unpublished observation). In the nose, ears, lips, and external genitalia, where melanocytes were present in the wild-type mice, hyperpigmentation was common to every HGF transgenic mice (Fig. 2). HGF could thus expand the number of melanocytes in regions to which melanoblasts had already migrated; however, it could not extend their migration to regions where melanoblasts are not normally distributed. Rarely, a cluster of melanocytes was found in the palm of HGF transgenic mice (Fig. 2E). These might have arisen from an exceptional melanoblast that migrated out from the normal pathway and was expanded by transgenic HGF. Most of the melanocytes were localized in the dermis of HGF transgenic mice, in spite of the fact that HGF was expressed in the epidermis. Although melanocytes were not evenly spread in dermis, but instead tended to be located more closely to the epidermis, as seen in Fig. 2J,M, a major portion of the melanocytes did not invade the epidermis. As suggested in the cases of adhesion of epidermal Langerhans cells (Tang et al., 1993), human melanocytes (Tang et al., 1994) and T lymphocytes (Cepek et al., 1994; Lee et al., 1994) to keratinocytes in vitro, E-cadherin expression both in the adherent cells and target keratinocytes is likely to mediate epidermal localization of these cells. Recently, elevated expression of E-cadherin in the melanoblasts of mouse embryos during the entry of the melanoblasts into the epidermis was reported (Nishimura et al., 1999). Modulation of cadherin-mediated cell-cell interaction by HGF was reported in keratinocyte cell lines, in which the concentration of adhesion molecules, including E-cadherin, was reduced upon HGF treatment (Watabe et al., 1993) and co-localization of c-met with E-cadherin and beta-catenin was observed in a tumor cell line (Hiscox and Jiang, 1999). Considering these observations, suppression of the expression of adhesion molecules, including E-cadherin, by the transgenic HGF may explain the inability of melanoblasts to invade the epidermis in HGF transgenic mice. Clinically, abnormal growth of melanocytes in the dermis is well documented, and is termed dermal melanocytosis (Ortonne and Nordlund, 1998). The pathogenesis of dermal melanocytosis is thought to be heterogeneous, re¯ecting multiple factors or genes that regulate the development of melanocytes (Hori and Takayama, 1988). Apparently, the HGF transgenic mice described here show the symptom of dermal melanocytosis observed in human skin. Thus, these
76
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
HGF transgenic mice may provide an animal model for dermal melanocytosis. More than 200 HGF transgenic mice have been raised for over 3 months, and 20 of them were selected to be raised more than 1 year. Although not all of them underwent close inspection by autopsy, we found no apparent melanomas in HGF transgenic mice, even in those that lived more than a year. No apparent shortening of the life span was observed, either. These observations suggest that HGF expressed in keratinocytes does not markedly contribute to malignant tumor induction. On the other hand, systemic expression of HGF under the control of a MT promoter induced development of tumors including melanomas (Otsuka et al., 1998). The number of melanocytes found in the skin of mice that have systemic expression of HGF is comparable to that of the k14-HGF transgenic mice used in the present study; therefore, different modes of HGF expression affect proliferation and differentiation of melanocytes similarly, but affect tumorigenesis differently. Although expression of cytokines such as HGF may induce a broad range of physiological changes in transgenic mice irrespective of promoters used for transgene expression, strictly keratinocyte-speci®c expression of HGF and the lack of apparent incidence of melanomas in k14-HGF transgenic mice may suggest the role of HGF supplied by the autocrine mechanism for the neoplastic change of melanocytes to melanoma. It should be noted that keratinocytes are expressing c-met as shown in Fig. 5D and autocrine activation of HGF signaling is expected in the keratinocytes of k14-HGF transgenic mice. Lower expression level of k14 promoter driven HGF might not be adequate to promote tumor development of keratinocytes even under the possible autocrine activation of HGF signaling. While systemic expression of HGF was reported to cause inappropriate localization of melanocyte in skin of extremities including footpads (Takayama et al., 1996), k14-HGF transgenic mice were mostly lacking melanocytes in the footpads, as described in the previous section. This suggest that paracrine HGF signaling alone do not stimulate migration of melanocytes in vivo. It is likely that not only the complicated biological process such as neoplastic transformation but the normal biological process such as cellular migration are affected differently by the different mode of HGF expression in vivo.
4. Experimental procedures 4.1. Generation and analysis of transgenic animals A full-length human HgfcDNA cloned into pBluescript (pHHGF, Honda et al., 1995) was cut by HpaI at its 3 0 end, ligated with BamHI linker, and then digested with BamHI. Gel-puri®ed Hgf cDNA with BamHI sites on both ends was cloned into the BamHI site of the K14 expression vector, provided by Dr. E. Fuchs (Vassar et al., 1989). To excise
transgenes from the plasmid, in which the Hgf sequence contains a KpnI site, DNA was ®rst cut with ClaI, and then cut partially with KpnI to obtain the fragment shown in Fig. 1A. After separation by agarose gel electrophoresis, the fragment was puri®ed by CsCl/Ethidium Bromide centrifugation, and then DNA was dissolved in 10 mM Tris (pH 7.5), 0.1 mM EDTA at the concentration of 2 mg/ml, and injected into fertilized oocytes collected from pregnant (C57BL/6xDBA/2) F1 mice as described (Kimura et al., 1995). Integration of transgenes was screened by PCR of genomic DNA with K14-speci®c primers, and then veri®ed by Southern hybridization with Hgf cDNA used for transgene construction. 4.2. Histology and immunohistochemistry For DOPA and DOPA-premelanin staining, pieces of skin were excised from the dorsal side of transgenic and littermate mice and ®xed with formalin in phosphate buffer (pH 7.0) for 16 h at 48C, and subjected to the DOPA reaction. The samples were embedded in paraf®n, and then subjected to the DOPA-premelanin reaction as described (Mishima, 1960; Hirobe, 1982). For immunohistochemistry, tissues were embedded in polyester wax, sectioned (7 mm) and stained as described (Yoshida et al., 1993; Kunisada et al., 1998b). The primary antibodies used were; rabbit antihuman HGF polyclonal antibody (Matsumoto et al., 1996); rabbit anti-mouse TRP2 polyclonal antibody (provided by V. Hearing; Tsukamoto et al., 1992), diluted 300£ in 5% normal goat serum in PBS for use as blocking solution; a monoclonal rat anti-mouse c-Kit antibody (ACK2 hybridoma cells were provided by S.-I. Nishikawa; Nishikawa et al., 1991), used at the concentration of 20 mg/ ml; monoclonal rat anti-mouse E-cadherin (ECCD2, conjugated with biotin, provided by S.-I. Nishikawa; Shirayoshi et al., 1986), used at 10 mg/ml. For double-labeled immuno¯uorescent staining, sections were incubated with biotinlabeled or non-labeled primary antibody overnight at room temperature, washed three times, and then incubated with ¯uorescence-labeled secondary antibodies for 2 h. Then, after they were washed three times, sections were incubated with streptoavidin-rhodamine or rhodamine-labeled antibodies bind to the ®rst antibody (purchased from Vector and diluted as speci®ed by the manufacturer). Because this protocol did not work well for rabbit anti-mouse c-met polyclonal antibody (R&D systems), cryostat sections were prepared from 17.5 dpc embryos. After they were immersed successively in 12, 15, and 18% sucrose dissolved in PBS for 30 min each, the skins were embedded in Tissue Tek (Miles), and frozen in liquid nitrogen. Cryostat sections of 10-mm thickness were rehydrated in PBS, and then processed as described for the sections from polyester wax-embedded specimens. 4.3. RT-PCR analysis Total RNAs were isolated from tissues from B6 mice
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
using Isogen (Nippon Gene). Random hexamer-primed cDNA was synthesized from 4 mg of RNA from each tissue using Super Script reverse transcriptase (GIBCO-BRL) in a 25-ml reaction mixture as speci®ed by the manufacturer. Two and one-half microliter aliquots or serial dilutions of the reaction mixture were then subjected to PCR in 50 ml of reaction mixture containing Taq polymerase and the buffer supplied by the manufacturer (TOYOBO). The primer sequences and the condition of the PCR cycle for mouse HGF were described previously (Ohmichi et al., 1998). For the control, a G3PDH sequence was ampli®ed using the previously described conditions and primers (Yamane et al., 1999). After PCR, 15-ml aliquots of the reaction mixture were separated by agarose gel electrophoresis, transferred to nylon membranes, and hybridized with mouse HGF and human G3PDH probes to obtain sequence-speci®c signals. Acknowledgements We thank Drs. E. Fuchs, V. Hearing, S.-I. Nishikawa for providing materials. We also thank Drs. H. Yoshida, E. Nishimura for discussion. This work was supported in part by grants from the Ministry of Education, Science, Sports and Culture of Japan, Special Coordination Funds for Promoting Science and Technology from the Science and Technology Agency of Japan and Yamanouchi Foundation for Research on Metabolic Disorders. References Baynash, A.G., Hosoda, K., Giaid, A., Richardson, J.A., Emoto, N., Hammer, R.E., Yanagisawa, M., 1994. Interaction of endothelin-3 with endothelin-B receptor is essential for development of epidermal melanocytes and enteric neurons. Cell 79, 1277±1285. Bennett, D.C., 1993. Genetics, development, and malignancy of melanocytes. Int. Rev. Cytol. 146, 191±260. Birchmeier, C., Meyer, D., Riethmecher, D., 1995. Factors controlling growth, motility, and morphogenesis of normal and malignant epithelial cells. Int. Rev. Cytol. 160, 221±266. Broudy, V.C., 1997. Stem cell factor and hematopoiesis. Blood 90, 1345± 1364. Byrne, C., Tainsky, M., Fuchs, E., 1994. Programming gene expression in developing epidermis. Development 120, 2369±2383. Cepek, K.L., Shaw, S.K., Parker, C.M., Russell, G.J., Morrow, J.S., Rimm, D.L., Brenner, M.B., 1994. Adhesion between epithelial cells and T lymphocytes mediated by E-cadherin and the aEb7 integrin. Nature 372, 190±193. Copeland, N.G., Gilbert, D.J., Cho, B., Donovan, P.J., Jenkins, N.A., Cosman, D., Anderson, D., Lyman, S.D., Williams, D.E., 1990. Mast cell growth factor maps near the Steel locus on mouse chromosome 10 and is deleted in a number of Steel alleles. Cell 63, 175±183. Defrances, M.C., Wolf, H.K., Michalopoulos, G.K., Zarnegar, R., 1992. The presence of hepatocyte growth factor in the developing rat. Development 116, 387±395. Galli, S.J., Zsebo, K.M., Geissler, E.N., 1993. The kit ligand. Stem cell factor. Adv. Immunol. 55, 1±96. Geissler, E.N., Ryan, M.A., Houseman, D.E., 1988. The dominant-white spotting (W) locus of the mouse encodes the c-kit proto-oncogene. Cell 55, 185±192. Halaban, R., Rubin, J.S., Funasaka, Y., Cobb, M., Boulton, T., Faletto, D.,
77
Rosen, E., Chan, A., Yoko, K., White, W., Cook, C., Moellman, G., 1992. Met and hepatocyte growth factor/scatter factor signal transduction in normal melanocytes and melanoma cells. Oncogene 7, 2195± 2206. Hirobe, T., 1982. Genes involved in regulating the melanocyte and melanoblast-melanocyte populations in the epidermis of newborn mouse skin. J. Exp. Zool. 223, 257±264. Hirobe, T., 1984. Histochemical survey of the distribution of the epidermal melanoblasts and melanocytes in the mouse during fetal and postnatal periods. Anat. Rec. 208, 589±594. Hirobe, T., Abe, H., 1999. Genetic and epigenetic control of the proliferation and differentiation of mouse epidermal melanocytes in culture. Pigment Cell Res. 12, 147±163. Hiscox, S., Jiang, W.G., 1999. Association of the HGF/SF receptor, c-met, with the cell-surface adhesion molecule, E-cadherin, and catenins in human tumor cells. Biochem. Biophys. Res. Commum. 261, 406±411. Honda, S., Kagoshima, M., Wakana, A., Tohyama, M., Matsumoto, K., Nakamura, T., 1995. Localization and functional coupling of HGF and c-Met/HGF receptor in rat brain: implication as neurotrophic factor. Mol. Brain Res. 32, 197±210. Hori, Y., Takayama, O., 1988. Circumscribed dermal melanoses: classi®cation and histologic features. Dermatol. Clin. 6, 315±326. Hosoda, K., Hammer, R.E., Richardson, J.A., Baynash, A.G., Cheung, J., Giaid, A., Yanagisawa, M., 1994. Targeted and natural (Piebald-lethal) mutations of endothelin B receptor gene produce megacolon associated with spotted coat color in mice. Cell 79, 1276±1297. Imokawa, G., Yada, Y., Morisaki, N., Kimura, M., 1998. Biological characterization of human ®broblast-derived mitogenic factors for human melanocytes. Biochem. J. 330, 1235±1239. Kimura, S., Niwa, H., Moriyama, M., Araki, K., Abe, K., Miike, T., Yamamura, K., 1995. Improvement of germ cell line transmission by targeting b-garactosidase to nuclei in transgenic mice. Dev. Growth Differ. 38, 87±97. Kos, L., Aronzon, A., Takayama, H., Maina, F., Ponzetto, C., Merlino, G., Pavan, W., 1999. Hepatocyte growth factor/Scatter factor-MET signaling in neural crest-derived melanocyte development. Pigment Cell Res. 12, 13±21. Kunisada, T., Lu, S-Z., Yoshida, H., Nishikawa, S., Nishikawa, S.-I., Mizoguchi, M., Hayashi, S.I., Tyrrell, L., Williams, D.A., Wang, X., Longley, B.J., 1998a. Murine cutaneous mastocytosis and epidermal melanocytosis induced by keratinocyte expression of transgenic stem cells factor. J. Exp. Med. 187, 1565±1573. Kunisada, T., Yoshida, H., Yamazaki, H., Miyamoto, A., Hemmi, H., Nishimura, E., Shultz, L.D., Nishikawa, S.-I., Hayashi, S.I., 1998b. Transgene expression of steel factor in the basal layer of epidermis promotes survival, proliferation, differentiation and migration of melanocyte precursors. Development 125, 2915±2923. Lee, M.G., Sharrow, S.O., Farr, A.G., Singer, A., Udey, M.C., 1994. Expression of the homotypic adhesion molecule E-cadherin by immature murine thymocytes and thymic epithelial cells. J. Immunol. 152, 5653±5659. Lindner, G., Menrad, A., Gherardi, E., Merlino, G., Eelker, P., Handjiski, B., Roloff, B., Paus, R., 2000. Involvement of hepatocyte growth factor/ scatter factor and Met receptor signaling in hair follicle morphogenesis and cycling. FASEB J. 14, 319±332. Matsumoto, K., Nakamura, T., 1996. Emerging multipotent aspects of hepatocyte growth factor. J. Biochem. 119, 591±600. Matsumoto, K., Nakamura, T., 1997. Hepatocyte growth factor (HGF) as a tissue organizer for organogenesis and regeneration. Biochem. Biophys. Res. Commun. 239, 639±644. Matsumoto, K., Tajima, H., Nakamura, T., 1991. Hepatocyte growth factor is a potent stimulator of human melanocyte DNA synthesis and growth. Biochem. Biophys. Res. Commun. 176, 45±51. Matsumoto, K., Date, K., Shimura, H., Nakamura, T., 1996. Acquisition of invasive phenotype in gallbladder cancer cells via mutual interaction of stromal ®broblasts and cancer cells as mediated by hepatocyte growth factor. Jpn. J. Cancer Res. 87, 702±710.
78
T. Kunisada et al. / Mechanisms of Development 94 (2000) 67±78
Mishima, Y., 1960. New technique for comprehensive demonstration of melanin, premelanin, and tyrosinase sites; combined DOPA-premelanin reaction. J. Invest. Dermatol. 34, 355±360. Molleman, G., Halaban, R., 1998. Growth-factor receptors and signal transduction regulating the proliferation and differentiation of melanocytes. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 135±149. Nakamura, T., Nawa, K., Ichihara, A., 1984. Partial puri®cation and characterization of hepatocyte growth factor from serum of hepatectomozed rats. Biochem. Biophys. Res. Commun. 122, 1450±1459. Nakamura, T., Nishizawa, T., Hagiya, M., Seki, T., Shimonishi, M., Sugimura, A., Tashiro, K., Shimizu, S., 1989. Molecular cloning and expression of human hepatocyte growth factor. Nature 342, 440±443. Nishikawa, S., Kusakabe, M., Yoshinaga, K., Ogawa, M., Hayashi, S.I., Kunisada, T., Era, T., Sakakura, T., Nishikawa, S.-I., 1991. In utero manipulation of coat color formation by a monoclonal anti-c-kit antibody: two distinct waves of c-kit-dependency during melanocyte development. EMBO J. 10, 2111±2118. Nishimura, E., Yoshida, H., Kunisada, T., Nishikawa, S.-I., 1999. Cadherin expression associated with melanocyte migration and diversi®cation re¯ects the microenvironmental setting. Dev. Biol. 215, 155±166. Ohmichi, H., Koshimizu, U., Matsumoto, K., Nakamura, T., 1998. Hepatocyte growth factor (HGF) acts as a mesenchyme-derived morphogenic factor during fetal lung development. Development 125, 1315± 1324. Ortonne, J.-P., Nordlund, J.J., 1998. Mechanisms that cause abnormal skin color. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 489±502. Otsuka, T., Takayama, H., Sharp, R., Celli, G., Larochelle, W.J., Bottaro, D.P., Ellmore, N., Vieira, W., Owens, J.W., Anver, M., Merlino, G., 1998. c-Met autocrine activation induces development of malignant melanoma and acquisition of the metastatic phenotype. Cancer Res. 15, 5157±5167. Reedy, M.V., Parichy, D.M., Erickson, C.A., Mason, K.A., Frost-Mason, S.K., 1998. The regulation of melanoblast migration and differentiation. In: Nordlund, J.J., Boissy, R.E., Hearing, V.J., King, R.A., Ortonne, J.P. (Eds.). The Pigmentary System: Physiology and Pathophysiology, Oxford University Press, New York, pp. 75±95. Shimosaka, S., Tsuboi, R., Jindo, T., Imai, R., Takamori, K., Rubin, J.S., Ogawa, H., 1995. Hepatocyte growth factor/scatter factor expressed in follicular papilla cells stimulates human hair growth in vitro. J. Cell. Physiol. 165, 333±338.
Shirayoshi, Y., Nose, A., Iwasaki, K., Takeichi, M., 1986. N-linked oligosaccharides are not involved in the function of a cell-cell binding glycoprotein E-cadherin. Cell Struct. Funct. 11, 245±252. Silvers, W.K., 1979. The Coat Colors of Mice: A Model for Mammalian Gene Action and Interaction, Springer±Verlag, New York. Steel, K.P., Davidson, D.R., Jackson, I.J., 1992. TRP-2/DT, a new early melanoblast marker, shows that steel growth factor (c-kit ligand) is a survival factor. Development 115, 1111±1119. Takayama, H., LaRochelle, W.J., Anver, M., Bockman, D.E., Merlino, G., 1996. Scatter factor/hepatocyte growth factor as a regulator of skeletal muscle and neural crest development. Proc. Natl. Acad. Sci. USA 93, 5866±5871. Takayama, H., LaRochelle, W.J., Sharp, R., Otsuka, T., Kriebel, P., Anver, M., Aaronson, S.A., Merlino, G., 1997. Diverse tumorigenesis associated with aberrant development in mice overexpressing hepatocyte growth factor/scatter factor. Proc. Natl. Acad. Sci. USA 94, 701±706. Tang, A., Amagai, M., Granger, L.G., Stanley, J.R., Udey, M.C., 1993. Adhesion of epidermal Langerhans cells to keratinocytes mediated by E-cadherin. Nature 361, 82±85. Tang, A., Eller, M.S., Hara, M., Yaar, M., Hirohashi, S., Gilchrest, B.A., 1994. E-cadherin is the major mediator of human melanocyte adhesion to keratinocytes in vitro. J. Cell. Sci. 107, 983±992. Tsukamoto, K., Jackson, I.J., Urabe, K., Montague, P.M., Hearing, V.J., 1992. A second tyrosinase-related protein. TRP-2, is a melanogenic enzyme termed DOPAchrome tautomerase. EMBO J. 11, 519±526. Vassar, R., Rosenberg, M., Ross, S., Tyner, A., Fuchs, E., 1989. Tissuespeci®c and differentiation-speci®c expression of a human K14 keratin gene in transgenic mice. Proc. Natl. Acad. Sci. USA 86, 1563±1567. Watabe, M., Matsumoto, K., Nakamura, T., Takeichi, M., 1993. Effect of hepatocyte growth factor on cadherin-mediated cell-cell interaction. Cell Struct. Funct. 18, 117±124. Wehrle-Haller, B., Weston, J.A., 1997. Receptor tyrosine kinase-dependent neural crest migration in response to differentially localized growth factors. BioEssays 19, 337±345. Zarnegar, R., Michalopoulos, G.K., 1995. The many faces of hepatocyte growth factor: from hepatopoiesis to hematopoiesis. J. Cell. Biol. 129, 1177±1180. Yamane, T., Hayashi, S.I., Mizoguchi, M., Yamazaki, H., Kunisada, T., 1999. Derivation of melanocytes from embryonic stem cells in culture. Dev. Dyn. 216, 450±458. Yoshida, H., Nishikawa, S.-I., Okamura, H., Sakakura, T., Kusakabe, M., 1993. The role of c-kit proto-oncogene during melanocyte development in mouse. In vivo approach by the in utero microinjection of anti-c-kit antibody. Dev. Growth Differ. 35, 209±220.