The growth and differentiation of mast cells and melanocytes require stem cell factor (SCF), the ligand for the kit receptor tyrosine kinase. SCF may exist as a membrane-bound or soluble molecule. Abnormalities of the SCF-kit signaling pathway, with increased local concentrations of soluble SCF, have been implicated in the pathogenesis of the human disease cutaneous mastocytosis, but have not yet been shown to play a causal role. To investigate both the potential of SCF to cause mastocytosis and its role in epidermal melanocyte homeostasis, we targeted the expression of SCF to epidermal keratinocytes in mice with two different transgenes controlled by the human keratin 14 promoter. The transgenes contained cDNAs that either produced SCF, which can exist in both membrane-bound and soluble forms, or SCF, which remains essentially membrane bound. Murine epidermal keratinocyte expression of membrane-bound/ soluble SCF reproduced the phenotype of human cutaneous mastocytosis, with dermal mast cell infiltrates and epidermal hyperpigmentation, and caused the maintenance of a population of melanocytes in the interadnexal epidermis, an area where melanocytes and melanin are found in human skin but where they are not typically found in murine skin. Expression of membrane-bound SCF alone resulted in epidermal melanocytosis and melanin production, but did not by itself cause mastocytosis. We conclude, first, that a phenotype matching that of human mastocytosis can be produced in mice by keratinocyte overproduction of soluble SCF, suggesting a potential cause of this disease. Second, we conclude that keratinocyte expression of membrane-bound SCF results in the postnatal maintenance of epidermal melanocytes in mice. Since the resulting animals have skin that more closely approximates human skin than do normal mice, their study may be more relevant to human melanocyte biology than the study of skin of normal mice.
Melanocytes are present in the interadnexal epidermis in human skin. In contrast, melanocytes in adult murine skin are generally confined to hair follicles, with the exception of rare epidermal melanocytes found in the ears, footpads, and tail (1). A few dermal melanocytes may also be found in mice, mostly in the ears. Melanocyte migration and development, as well as the survival of melanocytes and mast cells, are dependent on expression of the kit protein, a receptor tyrosine kinase encoded by the c-kit protooncogene (2–6). The ligand for kit, known as stem cell factor (SCF1; also called mast cell growth factor, steel factor, and kit ligand) may be produced locally in human skin by epidermal keratinocytes, fibroblasts, and endothelial cells (7, 8). Definitive studies of SCF production in murine skin have not been reported, but transgenic studies using the SCF gene promoter region and β-galactosidase as a reporter gene suggest that, unlike in human skin, postnatal murine cutaneous SCF expression is limited to the dermis and hair follicles, and not found in epidermal keratinocytes (9). The difference in SCF expression between human and murine epidermis could explain the difference in melanocyte distribution in these two species.
SCF may be produced in two isoforms by alternate splicing of exon 6. One isoform lacks exon 6–encoded sequences and exists predominantly as a membrane-bound molecule. The other isoform contains exon 6–encoded sequences, which include a protease-sensitive site (10–19). Cleavage at the protease-sensitive site causes the release of a soluble, bioactive form of SCF. The membrane-bound and soluble forms of SCF have differential effects on melanocyte precursor dispersal and survival (20) and exogenous soluble SCF may produce cutaneous mast cell hyperplasia and cutaneous hyperpigmentation (21–23). In addition, local high concentrations of soluble SCF have been found in lesions of human cutaneous mastocytosis, a disease characterized by dermal accumulations of mast cells and increased epidermal melanin (7, 8, 24). These observations have led to the hypothesis that cutaneous mastocytosis represents a hyperplastic response to locally increased soluble SCF (25). However, clonal proliferations of mast cells containing mutations of c-KIT, which result in constitutive activation of kit and a selective growth advantage for the mast cells, have been identified in lesions of some clinical forms of mastocytosis (26), and in the peripheral blood of patients with mastocytosis and hematologic abnormalities (27). Similar mutations have been found in several mast cell lines (28– 32). Analysis of these latter data together have suggested that cutaneous mastocytosis may occur as a true primary neoplasm of mast cells (26).
Whether mastocytosis could be caused by overexpression or altered expression of SCF, with mutations occurring as secondary events, or whether c-KIT mutations are primary events and neoplastic mast cells induce secondary alterations in the local metabolism of SCF, has not been determined experimentally. Likewise, the reason for a lack of melanocytes in the interadnexal epidermis of murine skin is not known, but may be related to SCF expression.
To reproduce mastocytosis experimentally in the mouse, and to investigate the effects of various forms of SCF on melanocyte migration and development in the epidermis, we developed two types of transgenic mice. One type contained a transgene using the human keratin 14 gene promoter to express epidermal membrane-bound SCF from which the soluble form is spontaneously produced (referred to herein as membrane/soluble SCF). The other type used the same promoter to produce epidermal SCF that normally exists almost exclusively in a membrane-bound form. We found that keratinocyte expression of membrane/soluble SCF resulted in the accumulation of mast cells within the dermis as well as epidermal melanocyte maintenance and pigment production, thereby reproducing the phenotype of mastocytosis without inducing detectable c-kit mutations. In contrast, expression of only membrane-bound SCF by epidermal keratinocytes resulted in the maintenance of melanocytes in murine epidermis, thereby mimicking melanocyte growth in human skin, but did not spontaneously produce the mastocytosis phenotype.
Materials And Methods
Two murine SCF cDNAs were cloned into constructs containing the human cytokeratin 14 upstream region (33; gift of Dr. E. Fuchs, Howard Hughes Medical Institute, University of Chicago, Chicago, IL; Fig. 1). This promoter causes expression in the skin limited to the basal layers of the interadnexal epidermis and the follicular epithelium. The cDNAs were both full-length clones, containing exon 6–encoded sequences. One cDNA (transgene [TG]1) was unmodified and therefore could produce a membrane-bound protein with the exon 6–encoded protease sensitive site, from which a soluble, bioactive form of SCF could be efficiently generated (10, 11, 34). The product of this transgene will be referred to as membrane/soluble SCF. The second cDNA (TG2) had been previously modified by site-directed mutagenesis, deleting the primary high efficiency cleavage site (between amino acids 164 and 165) and an alternate exon 7–encoded low efficiency cleavage site (found in murine SCF between amino acids 180 and 181). The SCF produced by this transgene therefore exists predominantly as a membrane-bound molecule (membrane SCF; reference 35). Both cDNAs have been previously shown to produce biologically active SCF (35, 36).
Generation and Analysis of Transgenic Animals.
2 μg/ml transgenic DNA in 10 mM Tris (pH 7.5), 0.1 mM EDTA was injected into fertilized oocytes collected from pseudopregnant mice as described (37). At birth, most transgene expressing mice could be identified by distinctive pigmentary phenotypes, as described in Results. Integration of transgenes was verified by PCR of genomic DNA with transgene-specific primers and copy number estimated by Southern blotting of PCR products, followed by autoradiography and densitometry. Skin-specific expression of transgene messenger RNA was confirmed by Northern blotting and by reverse transcription PCR with transgene-specific primers using RNA extracted from representative animals. Transgene expression was quantitated by a ribonuclease protection assay kit (RPA II; Ambion, Austin, TX) according to the manufacturer's directions. In brief, total RNA extracted from mouse skin was hybridized with digoxigenin-labeled single-stranded RNA probes for 23 h at 42°C and digested with RNAseA and RNAseT1, electrophoresed through 5% polyacrylamide/7 M urea. Protected fragments were transferred to Hy+ membrane (Boehringer Mannheim, Indianapolis, IN), bands were detected by chemiluminescences, and band density was determined by densitometry. Preliminary studies of RNA preparations from each transgenic line were performed to measure β-actin, and the amounts of RNA for SCF mRNA determinations were adjusted for comparison. RNA was also used with reverse transcription and PCR for direct amplimer sequencing of c-kit mRNA sequences in regions that could contain known activating mutations, as previously described (26). The primers used were 5′ CAAATC/GCATCCC/ TCACACCCTGTTCAC and 5′ CCATAAGCAGTTGCCTCAAC, which bind to nucleotides 1568→ 1593 and 1854→ 1835, and 5′ TGTATTCACAGAGACTTGGC and 5′ AAAATCCCATAGGACCAGAC binding to nucleotides 2384→ 2403 and 2595→ 2576. These regions contain the codons with both of the activating mutations, codon 559 and codon 814, respectively, which have been described in human mastocytosis and in a murine mast cell line (5, 28).
TG1, containing the full-length unmodified SCF cDNA (membrane/soluble SCF), was injected into 100 F1 oocytes (C57 BL6 × SLJ), which were implanted into six host mothers, resulting in four independent hyperpigmented mice, all of which were positive for the transgene, and 40 other littermates that were pigmentary phenotype– and transgene–negative by PCR.
Oocytes for TG2 (membrane SCF) were F1 (C57BL/6J female × SLJ/J male) and the offspring could be black, agouti, or white. Injection of 40 embryos and implantation into 6 host mothers generated 48 pups, 21 of which were positive for integration by PCR. Of the 25 founder mice identified by PCR with the transgene-specific primers, 3 were black, 13 were agouti, and 9 were white. Five PCR-positive mice (three agouti and two black) showed a clearly identifiable pigmentary phenotype. Given the inability of white mice to produce normal cutaneous pigment, it is possible that there were also white founders that expressed the transgene without the production of an obvious change in pigment. Backcrossing of phenotype-positive black and agouti founders to C57 BL/6 mice produced uniform pigmentary changes, described in the Results section.
Histology, Immunohistochemistry, and Electron Microscopy.
Tissues from transgenic and littermate mice were fixed in formalin and embedded in paraffin or polyester wax, sectioned, and stained with hematoxylin and eosin, azure blue, alcian blue, or Giemsa's stain according to standard techniques (37–39). Immunofluorescence studies were performed on polyester wax-embedded sections or frozen sections, also using standard techniques. Antibodies included anti-S100 (rabbit anti–cow S100, prediluted; Dako Corp., Carpinteria, CA), and the ACK2 and ACK4 monoclonals (rat anti–mouse c-kit  at 20 μg/ml). Controls included omission of the primary antibody or the use of isotype-matched monoclonal antibodies of irrelevant specificity. Electron microscopy was done as previously described (41).
Dermal Mast Cells Accumulate in the Presence of Membrane/ soluble Keratinocyte SCF.
Sections of skin from all mice producing membrane/soluble SCF (TG1) showed increased mast cells in the dermis (Fig. 2). In newborn TG1-positive mice, the mast cells were superficial near the dermal–epidermal junction, close to the epidermal source of soluble SCF (Fig. 2,a). In older mice, the mast cells filled the papillary dermis in some areas, but were also present in the reticular dermis, in a pattern identical to that of human mastocytosis (Fig. 2, b–d). Electron microscopic analysis confirmed the presence of numerous mast cells with characteristic granules within the dermis of the TG1-positive animals and also showed that some of the heavily pigmented cells within the dermis of TG1-positive mice were melanocytes (Fig. 3). Mast cells were relatively rare and dermal melanocytes were not detected in the body wall skin of nontransgenic littermates and in TG2-positive animals of equivalent age. These observations were true across a wide range of copy numbers and levels of SCF mRNA expression (Fig. 4). Since the keratin 14 promoter is properly expressed in the skin only by keratinocytes, and since the production of only membrane-bound keratinocyte SCF did not spontaneously result in increased dermal mast cells in TG2-positive animals, keratinocyte production of the soluble form of SCF appears to be able to cause cutaneous mastocytosis in mice.
Known Activating c-kit Mutations Are Not Identified in Transgenic Mice.
Reverse transcriptase PCR analysis of mRNA extracted from the skin of both types of transgenic mice showed normal murine c-kit sequences with no evidence of activating mutations in two of the TG1 mice and one of the TG2 mice. Since the majority of the c-kit mRNA molecules in the TG2 mouse samples were derived from mast cells and melanocytes, this technique would identify mutations if they were expressed by a significant percentage of these cells (26).
SCF Transgenic Mice Are Hyperpigmented.
Targeted expression of each of the SCF transgenes in murine skin caused a similar, distinctive pigmented phenotype. The pigment responsible for the coat color of normal mice resides in the hair follicles and hair shafts, not in the epidermis. The transgenic mice, however, developed prominent epidermal pigmentation (Fig. 5). Transgene-positive animals could be identified by increased pigment at birth. By ∼21 d of age, the phenotypes were well established; phenotype-positive animals showed pigmentation of most of the skin as well as increased coat pigment. Extensive pigmentation was noted in a number of areas including the nose, mouth, ears, paws, and external genitalia when compared with normal littermate controls. There was enough individual variation in pigmentation so that no clear correlation between the level of pigmentation and the levels of transgenic expression could be shown. All transgenic animals showed similar degrees of pigmentation regardless of transgene type, copy number, or levels of SCF mRNA expression. In addition to the epidermal pigmentation, the three TG2-positive agouti founders showed thin black transverse strips, consistent with the pigment distribution of the allophenic mice described by B. Mintz (pictures not shown; reference 42).
Numerous Melanocytes Are Maintained in the Skin of Transgenic Mice.
The increased pigmentation of the skin of the transgene-positive mice of both types is attributable to the presence of intraepidermal melanocytes, and to the epidermal melanin produced by those cells. Intraepidermal melanocytes can be identified in hematoxylin and eosin–stained sections as cells in the basalar layers surrounded by clear halos (Fig. 6, a and b) or in immunoperoxidase preparations by their expression of S-100 protein. Immunohistochemical analysis of animals expressing each of the transgenes showed numerous S-100(+) intraepidermal melanocytes (also see Fig. 2,a). These melanocytes can be differentiated from Langerhans cells, which also express S-100 protein, because melanocytes are in the basal layers and Langerhans are in the suprabasal layers. Melanocytes can also be differentiated from Langerhans cells by their expression of the kit protein, the receptor for SCF, which is not expressed by Langerhans cells. Staining of transgenic animal skin with anti-kit antibody identified well-developed dendritic cells within the basalar layers of the epidermis and follicular epithelium, consistent with melanocytes (Figs. 6 c and 2 b).
Histologic examination confirmed the presence of pigment within the epidermis of both TG1 and TG2 phenotype–positive mice from all sites examined, including the ears, tail, footpads, and body wall (Fig. 6 a). In addition, TG1-positive mice showed many pigmented cells within the dermis. Pigmentary abnormalities were not observed in transgene-negative littermates. Only slight epidermal pigment was identified in these control mice, and mostly in non–hair-bearing areas like the footpad and tail. Although pigment patterns were stable throughout much of the adult life of the mice, an occasional TG1 (membrane/soluble SCF) mouse developed patchy areas of depigmentation, mostly in the ears, associated with loss of epidermal melanocytes and increased pigment incontinence. This phenomenon was not observed in the membrane SCF mice.
Electron microscopy confirmed the presence of numerous melanocytes within the epidermis of both types of transgenic mice (Fig. 7). Pigmented keratinocytes, similar to those seen in the epidermis of humans, were also present in the interadnexal epidermis of the transgenic mice. Intraepidermal melanocytes and pigmented keratinocytes were extremely rare in control mice.
The human disease mastocytosis is a heterogeneous group of conditions characterized by increased numbers of mast cells in various organs, most commonly the dermis (24). Mastocytosis involving the skin is also characterized by increased epidermal melanin, produced by melanocytes. Since melanocytes are neuroectodermally derived cells that migrate to the epidermis through patterns distinct from those of mesodermally derived mast cells, the colocalization of pigmentary abnormalities and mast cells in lesions of cutaneous mastocytosis implies the involvement of local factors. It has been suggested that the proliferation of mast cells in human mastocytosis is a reactive phenomenon rather than a true neoplastic process (7, 25) based upon the following three observations: (a) both melanocytes and mast cells express KIT and respond to SCF, (c) the injection of soluble SCF can cause epidermal pigmentation and dermal mast cell accumulation (21, 23, 43), and (c) increased amounts of soluble SCF have been identified in lesions of cutaneous mastocytosis (7, 8). In apparent conflict with this hypothesis is the fact that c-KIT mutations resulting in constitutive activation of KIT have been identified in multiple lesions of human mastocytosis, evidence that the mast cells in some cases of human mastocytosis represent a true proliferating neoplastic clone (26). However, recent studies suggest that only some clinical forms of human mastocytosis are associated with these mutations (Longley, B.J., unpublished observations), so the possible causes of mastocytosis in the clinical forms not associated with c-KIT mutations remain to be elucidated. In this study, we were able to reproduce the phenotype of mastocytosis in mice by expression of soluble SCF by murine epidermal keratinocytes. None of the known activating c-kit mutations were identified in the proliferating mast cells, showing that activating mutations are not necessary to produce the phenotype of this disease. These observations highlight a potential cause of mastocytosis in the absence of activating c-KIT mutations. Although the work reported in this manuscript fulfills one of Koch's criteria for establishing the cause of a disease, e.g., the reproduction of the disease in animals, it should be noted that murine mast cells differ considerably from human mast cells with regard to their growth requirements and that there are other potential growth factors and cytokines besides SCF that could contribute to the development of mastocytosis.
Melanocytes are maintained in human epidermis throughout life. In normal mice DOPA reaction–positive cells (melanoblasts and melanocytes) are found in the epidermis at birth, but their number decreases from postnatal day 4 and is severely reduced after 1 mo of age (44). One possible explanation for the maintenance of epidermal melanocytes in human skin, and the difference between the distribution of melanocytes in adult human and murine skin, could be expression of epidermal SCF. Human epidermal keratinocytes produce SCF (7, 8, 45), but the SCF gene does not appear to be expressed in murine epidermis (9). The results presented here show that SCF expression by murine epidermal keratinocytes causes the maintenance and stimulation of epidermal melanocytes throughout life. These data support the hypothesis that the decrease in melanocyte numbers in the postnatal mouse epidermis is due to a lack of local SCF expression. In combination with the fact that the soluble SCF produced by Sl/Sld mice is insufficient to support normal melanocyte survival and the observations that membrane-bound SCF promotes longer lasting kit activation and increased survival of kit-dependent cells in the hematopoietic system (1, 2, 10–12, 35), our data suggest that it is specifically the membrane-bound form of SCF that is crucial for melanocyte survival and function.
It is interesting to note that none of the animals expressing either of the transgenes described in this paper have developed melanoma to date, a finding that supports previous observations that stimulation of the kit tyrosine kinase receptor does not appear to promote the development of melanocytic tumors (46). It also seems likely that the animals described in this paper, or animals derived from them, will be useful in the study of cutaneous mastocytosis and epidermal melanocyte biology.
We thank Dr. L. Schultz and R. Halaban for critical discussions, and also thank Drs. E. Fuchs for providing hk 14 promoter and V. Hearing for providing anti-TRP2 antibody.
This work was supported by National Institutes of Health grants RO1AR3356 and SP30041942 (to B.J. Longley) and by grants from the Special Coordination funds of the Science and Technology Agency of Japan, from the Ministry of Education, Science and Culture of Japan, and from the Cellular Technology Institute, Otsuka Pharmaceutical Co., Ltd., Tokushima, Japan (to T. Kunisada).
Abbreviations used in this paper
Address correspondence to B. Jack Longley, Section of Dermatopathology, College of Physicians and Surgeons of Columbia University, 630 West 168th St., VC 5-578, New York, NY 10032. Phone: 212-305-2155; Fax: 212-927-9704; E-mail: firstname.lastname@example.org