The kit-ligand (steel factor) and its receptor c-kit/W: pleiotropic roles in gametogenesis and melanogenesis

Receptor tyrosine kinases (RTK) and their ligands function in the transduction of extracellular signals and are known to control cell proliferation, cell survival, motility and differentiation; consequently, RTK’s play key roles in embryonic development, in organogenesis and in the adult life of invertebrate and vertebrate animals. RTK’s that govern developmental processes, defined by mutant phenotypes, are known in Drosophila, the nematode C. elegans, and in mammals. In mice, several mutations in RTK genes with developmental consequences have been described. The c-kit receptor and the kit -ligand, KL, are allelic with the white spotting (W) and the steel (Sl) loci respectively; the platelet derived growth factor receptor-α chain (PDGFRα) with the patch (Ph) locus, and the macrophage colony stimulating factor (CSF-1) with the osteopetrosis (op) locus (Besmer, 1991; Pawson and Bernstein, 1990). The W and SI genes function in several unrelated cell types and have been of interest to developmental biologists for a long time. In this paper insights into c-kit receptor function will be discussed that arise from the molecular characterization of c-kit and its ligand, and from the analysis of various alleles at the IF and SI loci, with particular emphasis on gametogenesis and melanogenesis. The detailed knowledge of the functional significance of the c-kit receptor system in vivo, provided through analysis of mutant phenotypes, should facilitate future elucidation of mechanisms underlying cell proliferation, cell adhesion/migration, cell survival and other postmitotic functions in various cell systems during development and in adult life.

The proto-oncogene c-kit is the cellular homolog of v-kit, the oncogene of the Hardy-Zuckerman 4 -feline sarcoma virus, an acute transforming feline retrovirus (Besmer et al., 1986). c-kit encodes a receptor tyrosine kinase of the platelet derived growth factor (PDGF) receptor subfamily. The c-kit protein has an extracellular domain containing live immunoglobulin repeals and a cytoplasmic kinase domain, which is divided into two subdomains by the so-called kinase insert segment (Qiu et al., 1988; Yarden et al., 1987). A normal variant of the c-kit protein (Kit A⁺). formed as the result of alternate usage of 3′ splice sites, contains a four amino acid insert in the extracellular domain between amino acids 512 and 513 of the known murine c-kit sequence (Hayashi et al., 1991; Reith et al., 1991). Studies of c-kit signaling in several cell systems indicate that the c-kit receptor is autophosphorylated in response lo KL and that the activated receptor binds to and phosphorylates several known cytoplasmic proteins thought to represent intermediates in signaling pathways (Roltapel ct al.. 1991; Funasaka et al., 1992).

The c-kit gene maps to human chromosome 4 and mouse chromosome 5 in the vicinity of the PDGF receptor α chain gene and the flkl receptor kinase gene, and consists of 21 exons covering 65 kb (Qiu et al., 1988; Yarden et al., 1987; Stephenson et al., 1991; Matthews et al., 1991; Gokkel et al., 1992: Chu, Pritzer and Besmer. unpublished data). The realization that the c-kit proto-oncogene resides on mouse chromosome 5 in the vicinity of the while spoiling locus (W) initially raised the question of whether c-kit is encoded by the VV locus. Subsequently, the identity of the c-kit protooncogene with the while spotting locus was established by linkage analysis (Chabot et al., 1988). the demonstration of rearrangements in the c-kit gene in some IV alleles (Gcissler et al., 1988), and the finding of missense IV mutations that inactivate the c-kit kinase (Nocka et al., 1989; Tan et al., 1990).

The realization that c-kit was encoded at the W-locus accelerated the quest for the ligand of the c-kit receptor. The known function of c-kit/W in bone marrow-derived mast cells provided an assay for the isolation of a soluble form of the ligand of the c-kit receptor. KL (Nocka et al., l990a.c). However. KL was also isolated as a factor that promotes the formation of colonies from early hematopoietic progenitors (Zsebo et al. 1990a.b). and as a mast cell growth factor (Williams et al., 1990). Two alternatively spliced KL RNA transcripts encode two cell-associated KL protein products. KL-I and KL-2. that differ in their sequences N-terminal of the transmembrane segment (Flanagan cl al., 1991; Huang cl al., 1992). The KL-I and KL-2 RNA transcripts arc expressed in a tissue-specific fashion. The KL-2 protein lacks sequences that include the major proteolytic cleavage site for the generation of the soluble KL protein from KL-I (Fig. 1). The KL-I protein is efficiently processed by proteolytic cleavage to produce soluble KL; by contrast KL-2 is also processed to form soluble KL. but not as effectively. KL-2 therefore represents a differentially more stable cell-associated form of KL (Huang et al., 1992). The protease activities facilitating cleavage of KL-1 and KL-2 in COS-1 cells were shown to be distinct by using a panel of protease inhibitors (Pandiella ct al., 1992). Interestingly, the protein kinase C inducer PM A accelerates proteolytic cleavage of both KL-I and KL-2. suggesting that this process is subject to regulation (Huang et al., 1992). Consequently, differential expression of variant cell membrane associated KL molecules and their proteolytic cleavage to generate soluble forms of KL provide different means to control and modulate c-kit function.

Mutations at the murine while spoiling and steel loci generate deficiencies in three cell systems during embryogenesis and in the postnatal animal: the pigmentary system, germ cells and hematopoiesis (for reviews see: Russell. 1979; Silvers. 1979). During normal development, melanoblasts arise from the neural crest and migrate to the periphery where they enter the epidermis, colonize hair follicles and poslnatally differentiate to become pigmented melanocytes (Silvers. 1979). W and SI mutations affect several aspects of melanogenesis causing varying degrees of depigmentation (see below). Primordial germ cells are generated from (he posterior primitive streak and migrate from the base of the allantois and the hind gut Io the genital ridges; spermatogenesis and oogenesis then proceed following different developmental programs. W and SI mutations affect the survival, migration and proliferation of primordial germ cells as well as steps in spermatogenesis and oogenesis causing impaired fertility (see below).

In hematopoiesis VV and SI mutations affect cells within the stem cell hierarchy, distinct cell populations in the erythroid cell lineage and mast cells, during early development as well as in the adult animal (for a review see Russell. 1979). Mutant animals suffer from macrocytic anemia and they lack tissue mast cells. While VV mutations are cell autonomous, SI mutations affect the microenvironment of the cellular targets of the mutations (McCulloch et al., 1964, 1965). These findings were a strong indication that the VV and SI gene products function in the same biochemical pathway, possibly as receptor and ligand. The defects in W and SI mutant mice are consistent with a role of the c-kit receptor system in facilitating cell proliferation and survival of precursor cells as well as promoting cell migration and other functions in differentiated cells (Besnier. 1991; Williams et al., 1992).

The synchronous expression of the c-kit receptor and its ligand in close cellular environments is a good predictor for sites at which c-kit functions in vivo. Therefore, the examination of c-kit and KL expression during embryonic development and in the adult animal using RNA blot analysis, in situ hybridization and immunohistochemistry has provided important insights into understanding c-kit function. In agreement with the cell autonomous nature of W mutations. c-kit is expressed in cellular targets of IV’ and SI mutations during embryogenesis and in the postnatal animal in melanogenesis, gametogenesis and in cells of the hematopoietic system (Nocka et al., 1989; Orr-Urtreger et al., 1990; Manova et al., 1990: Manova and Bachvarova, 1991: Ogawa et al., 1991; Nishikawa et al., 1991; Yoshinaga et al., 1991). Expression of KI. has been shown to be associated with migratory pathways of melanoblasts and germ cells, and homing sites of both germ cells and hematopoietic progenitors during embryonic development, i.e. the genital ridges and the fetal liver (Malsui et al., 1990).

An easily recognizable phenotype, coal color spotting, has made possible the isolation of many distinct mutations al the IV and the SI loci (Russell. 1979; Silvers. 1979; Lyon and Searle. 1989). These mutants provided an opportunity to characterize both the molecular basis of these mutations as well as their effects on different cell lineages and tissues, thus furthering our understanding of c-kit function. IV and SI mutations vary in their degree of severity on the affected cellular targets. In the homozygous state, several alleles, including the original W and SI alleles, cause perinatal lethality while others arc viable and semi-fertile. Several different IV and SI alleles have been characterized at the molecular level (see Tables 1 and 2).

The original W allele Wⁿ and W^19H are c-kit null or loss of function mutations (Chabot et al., 1988; Nocka et al., 1990b; Tsujimura et al., 1993). Null mutations have severe phenotypes when homozygous but display only minor effects when heterozygous, i.c. a white belly spot, no anemia, no infertility. Therefore, null mutations have primarily recessive characteristics. Several mutations of W (W^{42,37,V,55,41}) are known in which heterozygotes are affected more severely than W/+: they vary in severity in the homozygous state and affect the three principal cell systems to comparable degrees (Geissler et al., 1981). These alleles contain c-kit missense mutations (see Table 1; Fig. 2), which impair c-kit kinase activity to differing degrees

(Tan et al., 1990; Reith et al., 1990; Nocka et al., 1990b). Work on the mechanism of activation of several tyrosine kinase receptors, including the members of (he PDGF receptor family and the c-kit receptor, implicates receptor dimers or oligomers as intermediates (Heldin cl al., 1989; Blume-Jensen et al., 1991). The dominant phenotypes of these mutations indicate that the mutant c-kit proteins in receptor heterodimers interfere with KI. induced signal transmission, effectively reducing the number of active receptor dimer/oligomers on the cell surface. Consequently these mutations give rise to more severe heterozygous mutant phenotypes than those of null mutations and have the hallmarks of dominant negative mutations.

Several severe SI alleles (Sl,Sl^{j,gb.8H,10H.12H,18H}) have been shown to contain deletions that include (he KI. gene, and therefore are KI. loss-of-function mutations (Copeland et al., 1990; Nocka et al., 1990c; Zsebo et al., 1990b). Homozygotes for the Steel-Dickie allele (SI’¹) are viable and less severely affected, implying some residual functional activity of KI., but they display all of the plciolropic effects normally associated with steel mutations. Molecular analysis indicates that the Sl¹ allele arose as a result of an intragenic deletion including the transmembrane domain and C terminus, generating a secreted KL protein product with normal biological activity (Fig. 1; Flanagan et al., 1991; Brannan et al., 1991; Huang et al., 1992). The biological characteristics of homozygous Sl^d/Sl^d mice and of Sl/Sl^d mice suggest that, although the Sl^d KL protein sustains some activity, it is largely defective in facilitating proliferation and survival of target cells. This indicates that the membrane-bound form of KL plays a critical role in c-kit function.

Melanocytes originate from neural crest cells migrating at embryonic day 9–10 (E9-10) along a dorsolateral pathway over the somites. They then spread through the dermis to all regions of the body by E12 ½ (see, LeDouarin, 1982; Rawles, 1947; Mayer, 1965; Serbedzija et al., 1990). Subsequently they move from the dermis into the epidermis (Rawles, 1947; Mayer, 1973), colonize hair follicles, and differentiate to become melanocytes. The c-kit receptor is expressed in melanoblasts from the time they leave the neural crest throughout development, as well as in differentiated melanocytes located over the papilla in hairbulbs; KL expression has been demonstrated in the microenvironment of melanoblasts and melanocytes during development and in the postnatal animal, implying a role for c-kit at several stages of melanogenesis (Manova and Bachvarova, 1991; Matsui et al., 1990; Duttlinger et al. 1993; Manova, unpublished data). The unpigmented skin of W mutant mice is devoid of melanocytes (Silvers, 1956, 1979). The absence of melanocytes has been interpreted as a failure of the precursor cells to migrate, to proliferate, or to survive. In the postnatal animal c-kit function is thought to be essential for pigment formation during the cycles of active hair growth (Nishikawa et al., 1991). During embryonic development c-kit function is necessary at around E14 ½, when melanocyte precursors migrate from the dermis to the epidermis (Mayers, 1973; Nishikawa et al., 1991). An earlier function for c-kit was suggested by the finding of synchronous expression of c-kit in melanoblasts located dorsally and laterally of the somites and KL in the somitic dermatome at around embryonic day 10½ (Manova and Bachvarova, 1991; Matsui et al., 1990). This notion is supported by the evidence that melanocyte precursors in the head disappear between E11½ and E12½ in Sl^d/Sl^d embryos (Steel et al., 1992).

W-sash is a particularly interesting allele at the W locus that affects primarily mast cells and melanogenesis but not other cellular targets of W and Si mutations (Lyon and Glenister, 1982). Thus, mice are fertile and not anemic, but they lack mast cells in their skin and intestine and they are almost entirely unpigmented (Stevens and Loutit, 1982) (Fig. 3). Heterozygotes are black with a broad white sash or belt in the lumbar region (Fig. 3). The restricted display of W mutant characteristics typical of W mutations in W^sh/W^sh mice suggests a mutation that affects c-kit expression in a cell-type-specific manner. In addition, the more severe pigment defect in heterozygous as compared to W/+ mice indicates an enhanced dominant effect of this mutation.

c-kit RNA and protein expression patterns in adult mice and during embryonic development have been investigated to elucidate the basis for the phenotypes of W-sash mice (Duttlinger et al., 1993; Tono et al., 1992). c-kit expression was absent in bone marrow-derived mast cells, the fetal and the adult lung, and the digestive tract at El3½; all of these tissues normally express c-kit. In addition, at E13½ W^sh/W^sh embryos lacked the c-kit positive melanocyte precursors normally found in the skin. However, c-kit was expressed normally at numerous other sites in embryos and adults. In E1014 mutant embryos, as in normal embryos, a low number of c-kit positive presumptive melanoblasts are present in the skin. Unexpectedly, in E10½ W^sh/W^sh embryos, ectopic c-kit expression was observed in the dermatome of the somites, the mesenchyme around the otic vesicle and the floorplate of the neural tube (Fig. 4). These structures are known to express KL in wild-type embryos (Fig. 4). In W^sh+ embryos, similar to W^sh/W^slt embryos, ectopic c-kit expression was observed in the head mesenchyme and at similar levels in all dermatomes along the axis of the embryo. This ectopic expression continues at E13½ W^sh/W^sh in homozygous and heterozygous mutant embryos (data not shown).

The inappropriate c-kit expression in the dermatome of mutant embryos provides an explanation for the dominant pigmentation defect in adult mutant mice, c-kit receptor expression in cells of the dermatome of W^sh / W^sh and W^sh+ mice may bind kit-ligand and reduce its concentration in the extracellular space. If so. the available KL may be limiting for the c-kii-cx pressin g mclanoblasts migrating over the dermatome, thus reducing their survival and/or proliferation. Alternatively, the co-expression ol c-kit and its ligand may activate dcrmatomal cells causing changes in the extracellular matrix. In either model, the W^sh / W^sh and W^sh,/+ melanocyte precursors would die or fail to proliferate between day E11½ and day E13½. In agreement with these models, migration of melanocyte precursors over the somites and through the dermis throughout the trunk region normally occurs at E10½−13½ and thus coincides with the lime of inappropriate c-kit expression in the dermatome.

Also. c-kit-expressing melanocyte precursors are observed in E10½−11½ W^sh / W^sh embryos, apparently on a dorsolateral pathway, whereas E13½ embryos contain no c-kit-positive cells in the skin. Importantly, these results imply that c-kit is required in melanogenesis between E10½−13½. This is a period of substantial proliferation of mclanoblasts from a sparsely to a more numerously distributed cell population (Manova and Bachvarova. 1991; Steel et al., 1992).

The W^sh mutation blocks c-kit expression in mast cells and mesenchymal cells in the lung and the digestive tract al E1314, while expression in other tissues is normal. This implies that positive elements regulating c-kit expression in mast cells, lung and digestive tract are affected by the W^xh mutation. Negative elements are also affected, since c-kit is expressed in additional tissues: interestingly, these are tissues which normally express KL. DNA blot analysis revealed no alteration of c-kit exon and intron sequences in the W^sh allele (Dultlinger et al., 1993). in agreement with the recent demonstration that the c-kit coding sequence in W^sh is unaltered (Tono et al., 1992). However, pulsed-field gel electrophoresis showed that the W^sh mutation involves a rearrangement within the vicinity of the c-kit gene Duttlinger et al., 1993), consistent with (he idea (hat several control elements arc affected by this mutation. Recent experiments indicate that the W^sh mutation involves sequences 5′ to c-kit. between c-kit and the PDGFRα gene (Duttlinger et al., unpublished data).

To investigate the formation of the ‘sash’ in W^sh+ mice, the distribution of melanoblasts in E11½ and E13½ normal and W^sh+ embryos has been analyzed (Duttlinger et al., 1993). Interestingly, in the epidermis of normal E11½ embryos, the melanoblasts are distributed in a graded fashion with high densities in the rostral and caudal regions and a minimal density in the lumbar region (Figs 5, 6). In W^sh/+ embryos mclanoblast numbers arc reduced at all levels, with very few present in the lumbar region. In normal E11½ embryos melanoblasts are already distributed al varying densities along the body axis and their numbers arc reduced in W^sh/+ embryos. Therefore, the melanocyte deficit in W^sh/+ embryos. which results in the sash. is established between E10½ and E13½ and may stem from a reduction of cells in the lumbar region essentially to zero, or below the minimal density required for pigmentation. Therefore, the sash appears to arise from a generalized effect of the mutation on mclanoblast number (presumably due to ectopic c-kit expression) in combination with an uneven distribution of melanoblasts in normal embryos. The relatively low density of melanocyte precursors in the thorax and rump Hanking the future sash is maintained at E15½ (unpublished observations). These cells are apparently able to colonize and contribute pigment to large contiguous areas. This is consistent with the observation that early melanocyte precursors produce clones that remain coherent during development, resulting in pigmented areas (patches) with sharp boundaries (Mintz. 1974). The low density and perhaps time-dependent restriction of migration of these melanocytes may account for their lack of colonization of the .sash/belt region.

Further support for the proposed cellular mechanism of .sash/belt formation in W^sh/+ mice comes from white spoiling phenotypes in transgenic mice expressing the c-kit protein of the dominant W⁴² allele. In these mice the W⁴² allele was expressed cctopically under the instruction of the human actin promotor. The transgenic mice displayed irregular while dorsal and ventral spots, in agreement with the dominant negative characteristics of the c-kit^W42 protein product (Ray et al., 1991). Presumably, these coat color patterns resulted from transgene expression in c-kit expressing melanoblasts as well as in the microenvironment of migrating melanoblasts. However, the coat color patterns in these mice were irregular and apparently not a stably inherited genetic trait. Since the transgenic mice had been derived in (CBA/J x C57BL6/J) F₁ mice, the variable penetrance of the pigmentation phenotype in these mice could have been the result of differences in genetic background among the off springs. To address this issue, one transgenic line (line 485 h/W-c-kit^W42) was backcrossed onto a C57BL6/J genetic background. A stable pigmentation pattern emerged (Fig. 3) similar to that seen in heterozygous W^sh/+ mice, but with a somewhat broader sash and a head spot. In these mice, the number of melanoblasts is presumably reduced all along the body axis as a result of uniform transgenc expression within the melanoblasts and/or in the microenvironment of migrating melanoblasts and this effect is superimposed on the graded distribution of mclanoblast precursors along the body axis. These results suggest that early melanoblasts in C57BL6/J are less numerous/viable than in CBA mice, possibly as a result of modifier genes affecting the melanoblasts or the microenvironment of melanoblasts. In agreement with this prediction, melanoblasts were more numerous at E11½ in an outbred mouse strain (ICR) than in C57BL6/J mice (K. Manova, unpublished data).

W/+ mice provide another case involving the c-kit receptor in which a belled phenotype is obtained. On different genetic backgrounds, the pigmentation pattern of W/+ mice ranges from almost fully pigmented through belted to almost white (Dunn and Charles, 1937).

A belted pigmentation pattern similar to that in W^sh /+ mice is also seen in mice heterozygous for the patch mutation (Fig. 3). When homozygous, Ph/Ph mice die during embryonic development. The Patch locus encodes the PDGF receptor α -chain and is closely linked to c-kit W (Smith et al., 1991; Stephenson et al., 1991; Duttlinger, unpublished data). The patch mutation consists of a deletion that includes the entire PDGF-A receptor. The 3 ′ -deletion endpoint lies in between the 3 ′ end of the PDGFR a gene and the 5 ′ end of the c-kit gene, but its location is not known precisely. Because of the similar pigmentation phenotypes of heterozygous VP⁷’ and patch mice, it is reasonable to speculate that the pigmentation defect in Ph/+ mice results from inappropriate c-kit expression as in W^slt/+ mice, rather than from a half dose of PDGFR α and that both W^sh and Ph affect common elements in the 5’ control region of the c-kit gene. In agreement with these predictions, preliminary results have revealed ectopic c-kit expression in Ph/+ embryos (R. Duttlinger, unpublished data) and this ectopic c-kit expression may lead to an overall reduction in melanoblast numbers along the body axis of the embryo.

Taken together, the coat color patterns of W^sh /+. W/+ and Ph/+ mice, and of transgenic mice expressing the dominant c-kit^W42 protein, as well as some other spotting mutations including piebald (Charles, 1938; Silvers, 1979), can be explained on the basis of general effects on melanoblasts and the graded melanoblast distribution at E13½ and this is in agreement with an earlier conjecture (Charles, 1938; Silvers, 1979).

An open question is why different regions of the body are unevenly populated by melanocyte precursors. First, the graded distribution may result from differential production of melanoblasts from neural crest along the body axis, i.e. more may be produced from cranial than from trunk neural crest. Second, it could arise from differential growth of body regions, i.e. the trunk region expands more than the head region from E10 to El3, or from different target sizes, i.e. the surface of the tail is smaller than that of the trunk. A third possibility is that the different melanoblast densities could be the result of differences in the cellular microenvironment through which melanocyte precursors migrate. Interestingly, the equivalent of the somitic dermatome along which melanocyte precursors migrate is absent in somito-meres and the dermis in the head originates entirely from the neural crest (Couly et al., 1992). In the tail, the neural tube and somites arise from the tail bud blastema, again raising the possibility of intrinsic and environmental differences for melanocyte precursors.

Germ cells are formed from cells of the epiblast that move through the posterior primitive streak and first appear in the allantois at E7½ (Ginsburg et al., 1990). They then move into the hindgut, up the dorsal mesentery and laterally to the genital ridges, where they arrive by E11½. In normal mice the c-kit receptor is expressed highly in primordial germ cells (PGC) during their proliferative phase from E714 to El 3; thereafter, as the germ cells enter a quiescent period or meiotic prophase, c-kit expression is no longer detected (Manova and Bachvarova, 1991; Bachvarova et al., 1993). Early investigations by Bennett (1956) and by Mintz and Russell (1957) indicated that in mice with severe SI and W mutations PGCs fail to increase in number during early development (E8½−E12) and that as a consequence very few are present in the fetal gonads. However, PGCs are formed in mutant animals and therefore c-kit does not appear to be involved in the initial determination of PGCs.

In mice, ovarian oocytes enter the diplotene stage around the time of birth and primordial follicles are subsequently formed. In a first wave, follicles in the central region of the ovary begin growth immediately and oocytes reach full size by 16 days of age. Thereafter, growing follicles are continually recruited from primordial oocytes throughout fertile life. The c-kit receptor is expressed in oocytes at high levels at all stages of postnatal development, starting in the diplotene stage (Manova et al., 1990). In contrast, KL is expressed in follicle cells of growing follicles and the expression increases during follicle development to high levels in the three layered follicles (Manova et al., 1993). Taken together these results suggest a role for c-kit in oocyte growth.

In some weak Sl alleles, significant numbers of germ cells are found in the gonad, but either female (Sl^Pan, Sl^t, Sl^con) or male mice (Sl^l7H) are sterile, implying that the c-kit receptor is essential for postnatal development of oocytes and spermatogonia, but that different properties of KL or its expression are important for development of female and male germ cells.

In order to elucidate the function of c-kit in oocyte development we investigated mice carrying the Sl-panda allele. Homozygous Sl^Pan / Sl^Pan mice are black-eyed whites with pigmented ears and scrotum and have mild macrocytic anemia; females are sterile, whereas males are fertile (Beechey et al., 1986). Molecular analysis indicated that the KL coding sequences are normal in the SIP™ allele, but that the levels of the KL transcripts are consistently reduced in most tissues analyzed; therefore, the Sl^Pan mutation affects KL gene expression (Huang et al., 1993). Histological analysis of ovaries from homozygous Sl^Pan mice showed that the number of primordial oocytes in neonatal animals was reduced to 20% of normal, indicating an effect of the Sl^Pan mutation on PGC’s (Fig. 7). Furthermore, in juvenile and adult mice ovarian follicle development in homozygous mutant animals was arrested at the one-layered cuboidal stage (Fig. 7). Therefore, a reduced level of KL in SlP^an/SlP^an ovarian follicle cells appears to arrest ovarian follicle development, implying an essential role for c-kit in oocyte growth/maturation. Whereas KL is limiting in oogenesis, a reduced level of KL does not appear to affect spermatogenesis.

Two other steel alleles, SI¹ and Sl^con, specifically affect female fertility. In Sl/Sl¹ females, follicles are arrested at a stage similar to that of SIP™/SIP™, but more germ cells are present (Kuroda et al., 1988). Unlike Sl¹ and SIP™, in Sl^con/Sl^con females, follicle development appears to be normal, but they have only a few germ cells which are quickly depleted in adults, while in males spermatogenesis appears to he capable of some postnatal regeneration (Beechey and Searle. 1983).

In summary the c-kit receptor system is essential at several stages in gametogenesis. During development c-kit provides a proliferative and/or cell survival signal for primordial germ cells in a defined time period. In spermatogenesis c-kit is thought to facilitate proliferalion/survival of spermatogonia during the first four cycles of spermatogonia proliferalion/differentiation (Manova ct al.. 1990: Yoshinaga et al., 1991). In oogenesis c-kit appears to function in the diplotene stage of prophase of meiosis to facilitate oocyte growth in the early stages of follicle development (Huang et al., 1993).

The discovery of the identity of the c-kit gene with the W locus has brought to light interesting pleiotropic roles of this gene in developmental processes, particularly melanogenesis. gametogenesis and hematopoiesis. Knowledge of c-kit function has facilitated the identification and characterization of the ligand for the c-kit receptor and the demonstration of allelism between kit-ligand and Steel provided a molecular basis for the relationship between IV and SI mutations in mice. A role for c-kit is now known in many cell types and lineages through studies of (he numerous mutations that arc available in this system. Interestingly, in both melanogenesis and gametogenesis, as well as in hematopoiesis. c-kit plays a critical role in both the earliest precursors during embryonic development, and in postnatal differentiating cells.

A corollary of recent studies of c-kit is the realization that the gene may function in cell types that are not known targets of IV and SI mutations (Moteo et al., 1991; Manova et al., 1992). During embryonic development c-kit expression is seen in portions of the developing central nervous system, the olfactory epithelium and other tissues. This expression is typically observed in cells that have ceased to divide and have begun their differentiation. In the adult animal c-kit and KL expression are prominent in the lung and in specific cells in the brain, in the hippocampus and the cerebellum (Manova et al., 1992; Motroetal., 1991; Mori et al., 1992). Redundant signaling mechanisms in these cell systems may compensate for the lack of c-kit function in these cell systems in W and SI mutant mice. The elucidation of a role for the c-kit receptor system in the central nervous system is clearly an important task of the future.

We would like to thank Drs. Tony Brown, Elizabeth Lacy, Prabir Ray, Karl Nocka, Tang-Yuan Chu, Ellen Pritzer and Nelson Yee for numerous discussions and comments on this manuscript. Support by grants from the American Cancer Society, the National Cancer Institute, National Institute of Child Health and Human Development and from the National Science Foundation is acknowledged.