Genetic Analysis of Kit Ligand Functions During Mouse Spermatogenesis
Department of Genetics, University of Georgia, Athens, GA (e-mail: email@example.com).
Kit ligand (KITL), also known as mast cell growth factor (MGF), stem cell factor (SCF), and Steel factor,1 and its receptor, KIT, are essential for embryonic and postnatal development of both male and female germ cells. While germ cells express KIT at multiple stages in their development, somatic cells that support the growth and differentiation of germ cells express KITL. Numerous studies have revealed that KITL promotes proliferation and suppresses apoptosis of differentiated spermatogonia and has a role in either initiating or maintaining meiosis. However, many details of the molecular mechanisms by which KITL signaling affects proliferation, apoptosis, and differentiation of postnatal male germ cells are not known. This review summarizes the structure, organization, and expression of the Kitl gene in mice. In addition, the functions of KITL and its receptor during spermatogenesis are reviewed, with an emphasis on information gained through the analysis of mouse mutants. In mice, KITL and KIT are encoded by the Steel (Sl) locus and the Dominant White Spotting (W) locus, respectively, which are two classical mutant loci. Long before the Sl and W gene products were identified, numerous studies revealed that they are essential for spermatogenesis. Readers are encouraged to read the earlier literature; however, in this review we summarize research findings that were published since 1990, the year in which KITL was shown to be the product of the Sl locus. Although a great deal of information has been revealed about KITL functions during spermatogenesis, we raise several important questions that remain to be addressed.
[Wild-type and mutant KITL proteins (membrane-bound is indicated by MB; precursor, Pre; and soluble, S forms). The mRNAs encoding each form are described briefly in the text. The original reference describing each mutation is as follows: KitlSl-39R and KitlSl-36R (Rajaraman et al, 2002a), KitlSl-17H (Brannan et al, 1992), and KitlSl-d (Brannan et al, 1991; Flanagan et al, 1991). For each mutant, MB*, Pre*, and S* refer to KITL proteins that have altered sequences. For wild-type and mutant proteins, the following symbols are used: solid black rectangles are the 4 α-helical domains; the light grey rectangle is the alternately spliced exon 6 that contains the proteolytic cleavage site, shown as a jagged line; black rectangles with dots are transmembrane domains, the ovals and numbers to the right of each protein are the sites of N-linked glycosylation sites, and the amino acid number (with 1 at the N-terminus), respectively. For KITLSl-39R, the white asterisk indicates the position of the S122F missense mutation. For KITLSl-17H, KITLSl-d, and KITLSl-136R, the numbers corresponding to the C-terminal KITL sequence plus the numbers of out-of-frame residues are shown with outof-frame sequences indicated as cross-hatched rectangles.]
Kitl Gene and Gene Products
Kitl is located on chromosome 10 of mice and its coding region comprises 9 exons. Of these 9 exons, only exon 6 is known to be alternatively spliced. In testes, Sertoli cells, but not germ cells, were shown to express Kitl messenger RNA (mRNA) as well as biologically active KIT ligand (KITL) (Motro et al, 1991; Rossi et al, 1991; Tajima et al, 1991). Although Kitl expression in Sertoli cells is under developmental, tissue-specific, stage-specific, and hormonal regulation (Manova et al, 1993; Rossi et al, 1993; Tajima et al, 1993; Hakovirta et al, 1999; Yan et al, 1999), little is known of its transcriptional and posttranscriptional regulation. In studies using cultured Sertoli cells from mice, rats, and humans, sequences in the adjacent 5′-flanking region of Kitl have been identified that mediate transcriptional activation in response to stimulation by follicle-stimulating hormone (FSH) or cyclic adenosine monophosphate (cAMP) analogues (Taylor et al, 1996; Jiang et al, 1997; Grimaldi et al, 2003). However, it is not clear whether these cultured cells reflect the transcriptional state of the gene in vivo. For example, DNA sequences from a 10-kilobase (kb) region upstream of the Kitl transcriptional start site were not sufficient to drive lacZ reporter gene expression in the gonads of transgenic mice (Yoshida et al, 1996). This absence of expression in the gonads occurred even though this same fragment, or an even smaller fragment containing only 2 kb of sequence upstream to the start site, was able to direct reporter gene expression to neural tissue and skin in a pattern identical to that of the endogenous gene. Furthermore, chromosomal rearrangements located up to 200 kb away from the Kitl transcription start site exert tissuespecific effects on Kitl mRNA expression (Bedell et al, 1995), indicating that transcriptional regulation of this gene is complex and may require sequences located a large distance from the gene.
A potentially important aspect of Kitl mRNA expression in the adult testis is that it is stage-specific, and as such, may reflect requirements for KITL function during different aspects of spermatogenesis. The first indication of this came from a study that used in situ hybridization of testes sections from mice at different ages (Manova et al, 1993). In this study, all tubules from testes of mice younger than postnatal day 9 (P9) had high levels of Kitl mRNA, but by P13, Kitl mRNA levels were greatly reduced in about half the tubules. It is interesting that Kitl mRNA expression in P13 testes was maximal in tubules that contain proliferating spermatogonia but lack meiotic cells, and was greatly reduced in tubules that contained meiotic cells but no proliferating spermatogonia. Stage-specific expression of Kitl mRNA was confirmed by analysis of rat testis in which maximal expression was again observed in tubules (specifically stages II-VI) that have proliferating spermatogonia (Hakovirta et al, 1999; Yan et al, 1999). Addition of FSH to culture media of dissected tubules from rat testes resulted in a stage-specific increase in Kitl mRNA levels, with maximal induction occurring in stages II-VI (Yan et al, 1999). In this same study, the FSH-induced increase in Kitl mRNA was shown to result from both increased transcription initiation and increased mRNA stability. Such dual regulatory mechanisms may provide a way for Kitl mRNA levels to be tightly fine-tuned in response to developmental needs during spermatogenesis. In support of the idea that regulation of Kitl mRNA expression may relate to the proliferative capacity of the epithelium, addition of recombinant KITL to culture media of dissected stage XII tubules from rat testes resulted in a dramatic increase in spermatogonial DNA synthesis (Hakovirta et al, 1999).
KITL is expressed as either a membrane-bound protein or as a soluble protein, and these two forms arise from alternative RNA splicing and posttranslational processing (Flanagan et al, 1991; Huang et al, 1992). Two Kitl mRNAs that differ by the presence or absence of exon 6 [(+)E6 and (−)E6, also called KL-1 and KL-2, respectively (Huang et al, 1992)] are expressed in a developmentally regulated fashion in the testis (Manova et al, 1993; Rossi et al, 1993; Mauduit et al, 1999). The expression patterns of these alternatively spliced products and the potential significance of their differential expression are discussed in the next section. Both Kitl mRNAs are translated into proteins that contain four functionally important regions (see Figure and reviews by Besmer et al, 1993 and Lev et al, 1994): a signal sequence, an extracellular domain, a transmembrane domain, and a cytoplasmic domain. The N-terminal signal sequence of 25 amino acids is removed during intracellular trafficking and is similar to that of many other proteins targeted for the cell surface. The extracellular domain contains the major structural elements of KITL and contains sequences that bind to KIT with high specificity. A hydrophobic domain of 26 amino acids functions as the transmembrane domain of KITL and is typical of other single-pass integral membrane proteins. The cytoplasmic tail of MB-KITL is highly conserved, and two studies revealed that intracellular trafficking and stability of KITL require sequences in its cytoplasmic domain (Tajima et al, 1998; Wehrle-Haller and Weston, 1999). Although it was previously believed that the KITL cytoplasmic domain has no recognizable motifs shared with other transmembrane proteins, a recent study has revealed a new motif in this domain that targets KITL and other transmembrane growth factors to the basolateral surface of polarized cells (Wehrle-Haller and Imhof, 2001). For KITL in the testes, such basolateral targeting may have important functional consequences.
The (+)E6 Kitl mRNA encodes a 248-amino acid transmembrane precursor (Pre-KITL) that is processed at the primary cleavage site in exon 6 to produce a soluble isoform (S-KITL) of 165 amino acids (Figure). Unlike other transmembrane cytokines such as transforming growth factor-α, normal proteolysis of KITL does not require sequences in the cytoplasmic domain (Cheng and Flanagan, 1994). Even though Pre-KITL cleavage could be an important mechanism for regulating KITL function, little information exists about KITL processing enzymes. Currently, the only proteins known to cleave KITL are a chymase expressed on mast cells (Longley et al, 1997) and matrix metalloproteinase-9, which is found in bone marrow (MMP-9, [Heissig et al, 2002]). The alternatively spliced Kitl mRNA, (−)E6 Kitl mRNA, lacks the primary cleavage site and encodes a predominantly membranebound KITL (MB-KITL) of 220 amino acids (Figure). It is clear from many in vitro studies that both S-KITL and MB-KITL are biologically active; these studies were conducted with primordial germ cells (PGCs) (see reviews by De Felici, 2000 and Donovan, 1994) and spermatogonia (Tajima et al, 1991; Rossi et al, 1993; Feng et al, 2002; Yan et al, 2000b), as well as with other cells that require KITL signaling, such as oocytes, hematopoietic cells, and melanocytes. However, studies with cultured hematopoietic cells have shown that MB-KITL has greater bioactivity in vitro and promotes a more persistent activation of KIT than does S-KITL (Miyazawa et al, 1995; Kapur et al, 1998). Furthermore, genetic studies with mice suggest that MB-KITL plays the predominant role in many tissues in vivo, including the testis. While both S-KITL and MB-KITL form noncovalently linked dimers (Arakawa et al, 1991; Hsu et al, 1997; Tajima et al, 1998) and are heavily glycosylated at both N- and O-linked sites (Lu et al, 1991, 1992; Huang et al, 1992), the roles of dimerization and glycosylation in KITL functions are currently not known.
Crystallographic studies revealed that KITL is a member of the short-chain subgroup of helical cytokines (Jiang et al, 2000; Zhang et al, 2000), a family that shares significant structural similarities despite minimal conservation in primary amino acid sequence (Hill et al, 2002). The major structural features of these helical cytokines are 4 α-helical regions in the extracellular domain (Figure) that are oriented in an up-up-down-down topology with long loops between helices A and B and between helices C and D (Hill et al, 2002). Structural studies have so far been conducted only with S-KITL, and it will be of interest to determine whether the extracellular domains of S-KITL and MB-KITL have the same tertiary structure. Alternatively, the presence of additional sequences, tethering via its transmembrane domain, or both, may affect the folding of the extracellular domain of KITL. If so, such structural differences in the extracellular domains of MB-KITL and S-KITL could contribute to their functional differences.
Expression and Function of KITL and KIT During Spermatogenesis
In postnatal testes, KITL is expressed by Sertoli cells, whereas KIT is expressed by germ cells and by Leydig cells (reviewed by Besmer et al, 1993). Examination of the expression patterns of Kitl mRNA and subcellular localization of KITL in Sertoli cells has provided important clues to the functions of KITL during spermatogenesis. Using cocultures of mast cells with Sertoli cells as an assay system for KITL bioactivity, Tajima et al (1991) showed that mast cell proliferation occurred only if there was direct cell-cell contact between the 2 cell types. This requirement for cell-cell contact suggested that the biologically active form of KITL expressed from these Sertoli cells, which were from P16-P20 mice, was predominantly membrane-bound. Additional studies revealed that the ratios of the 2 alternatively spliced Kitl mRNAs, as well as total amount of Kitl mRNA, are altered at distinct developmental stages of the testis (Manova et al, 1993; Marziali et al, 1993; Mauduit et al, 1999). These findings suggest that there is a preferential requirement for either S-KITL or MB-KITL at different stages of spermatogenesis. Splicing of exon 6 is affected by pH, such that the (−)E6 Kitl mRNA, which expresses MB-KITL, is expressed in an acidic environment at a much higher level than (+) E6 Kitl mRNA (Mauduit et al, 1999). As discussed by Mauduit et al (1999), it is conceivable that the pH of the Sertoli cell environment could influence KITL function by altering the relative amounts of alternativelyspliced Kitl mRNA in favor of the (−) E6 Kitl mRNA. Based on severe defects in spermatogenesis observed with a mouse mutant that lacks MB-KITL (KitlSl-d, see below), it has often been suggested that MB-KITL is functionally more important than S-KITL in the postnatal testis. However, as discussed below, it seems unlikely that KitlSl-d encodes an S-KITL that is normal in sequence levels, expression levels, or both. Furthermore, the functions of S-KITL during later stages of spermatogenesis could be obscured by the clearly predominant and earlier acting functions of MB-KITL. Indeed, S-KITL has been shown to promote both proliferation (Rossi et al, 1993; Packer et al, 1995) and meiotic progression (Vincent et al, 1998; Feng et al, 2002) of cultured primary spermatogonia. The localization of KITL in Sertoli cells also suggests that KITL functions may vary at different stages of spermatogenesis. Prior to P5, KITL is concentrated in the apical cytoplasm of Sertoli cells but is concentrated in the basal region of Sertoli cells in juvenile testes and in specific stages of tubules of mature testes (Manova et al, 1993). Furthermore, a radial staining pattern for KITL in these cells was observed in stages VII and VIII of mature testes, with signal extending from the periphery to the adluminal compartment (Vincent et al, 1998). These studies suggest that the localization of KITL could be altered in response to a specific requirement during different stages of spermatogenesis (ie, basally concentrated KITL would be available for promoting spermatogonial proliferation, whereas radially concentrated KITL would be available for promoting the initiation of meiosis).
An important reagent for identifying KIT-expressing cells and for dissecting KITL and KIT function is the ACK2 monoclonal antibody (Nishikawa et al, 1991). This antibody is specific for an epitope in the extracellular domain of KIT and neutralizes signaling of KITL by blocking its binding to KIT. Nishikawa and coworkers demonstrated conclusively that injection of ACK2 into mice causes a depletion of differentiating type A spermatogonia but does not affect other types of spermatogonia (undifferentiated, intermediate, and type B), or spermatocytes, spermatids, or spermatozoa (Yoshinaga et al, 1991). The specificity of ACK2-neutralizing activity was related directly to expression of KIT in these various cell types (ie, in adults, all spermatogonia except undifferentiated spermatogonia are positive for ACK2-immunoreactivity, whereas spermatocytes and spermatids are negative for ACK2-immunoreactivity).
Many studies have confirmed that KIT is expressed on differentiating but not undifferentiated spermatogonia and that S-KITL promotes the proliferation and suppresses apoptosis of cultured spermatogonia (Packer et al, 1995; Schrans-Stassen et al, 1999; Yan et al, 2000b). The distinction between undifferentiated and differentiated spermatogonia is particularly important because the former are believed to be spermatogonial stem cells. Brinster and colleagues developed a powerful functional test for spermatogonial stem cell activity that involves transplantation of testes cells into recipient testes (for recent reviews see Brinster, 2002; Johnston et al, 2000). In one such study, flow cytometry was used to purify specific populations of spermatogonia from adult testes, and the results demonstrated conclusively that only KIT(−) spermatogonia, and not KIT(+) spermatogonia, have stem cell activity (Shinohara et al, 2000b). It is interesting that the age of either donor or recipient for transplantation has an effect on the repopulating ability of testis cells. The testes of mice at age P5 to P12 were found to provide a much better environment for repopulation than were testes from adult mice at 14–20 weeks of age (Shinohara et al, 2001). A possible explanation for this age-related difference is that about 80% of the total Kitl mRNA in the young testes is (−)E6 Kitl mRNA, whereas the proportion of this mRNA relative to (+)E6 Kitl mRNA is greatly decreased in mature mice (Manova et al, 1993). Thus, more MB-KITL should be present in the young testis environment than in the older testis environment; however, that possibility remains to be tested directly. Recently, it was reported that unlike KIT(+) spermatogonia from adults, KIT(+) spermatogonia from neonatal mice have repopulating ability when transplanted into adult recipients (Ohbo et al, 2003). Because KIT expression is considered a marker of differentiating spermatogonia, it will be of considerable interest to determine whether the repopulating ability of neonatal KIT(+) spermatogonia reflects a reversion of at least some of these cells to an undifferentiated state. Alternatively, neonatal KIT(+) spermatogonia may be true stem cells whose properties differ from those of adult KIT(−) spermatogonial stem cells, or KIT may have significant functional differences in the neonatal versus adult testes.
In addition to the now well-established role in differentiating spermatogonia, evidence has been accumulating that KITL signaling is involved in initiating meiosis, maintaining meiosis, or both. However, there have been conflicting reports regarding whether KIT is expressed on meiotic and postmeiotic cells. These differences may depend on the methods of detection (ie, in situ hybridization versus immunostaining, and the type of antibody used for immunostaining) or on the preparation of germ cells (ie, testes sections versus purified cell populations). For example, early studies used in situ hybridization (Manova et al, 1990) and immunostaining with ACK2 (Yoshinaga et al, 1991) to detect Kit mRNA and KIT protein, respectively, in mouse testis sections. These results are in basic agreement, with the highest levels found in differentiating spermatogonia and early primary spermatocytes. In both cases, KIT expression was not detected in late primary spermatocytes, secondary spermatocytes, spermatids, or spermatozoa. Despite this apparent lack of KIT expression in meiotic and postmeiotic cells, experiments with cultured spermatogonia have provided compelling evidence that KITL and KIT have roles in inducing meiosis. Through use of reverse transcriptase-polymerase chain reaction and flow cytometry of purified testis cell populations, the presence of both Kit mRNA and KIT protein in pachytene spermatocytes was demonstrated (Vincent et al, 1998). In addition, progression of germ cells through meiosis in culture could be achieved using an immortalized Sertoli cell line (15P-1), and this meiotic progression was shown to be dependent on KITL and KIT (Vincent et al, 1998). An important advance in understanding spermatogenesis has come from the generation of a cell line of type A spermatogonia that was immortalized by overexpression of the catalytic subunit of telomerase (Feng et al, 2002). These authors showed that S-KITL induces the appearance of both meiotic and postmeiotic germ cells from the immortalized spermatogonia. Although both studies provide convincing evidence that KITL can initiate meiosis, there is a formal possibility that the purified cells or the culture conditions used in these in vitro experiments may not precisely mimic the in vivo situation.
If KITL does play a role in meiosis in vivo, then an important issue to be resolved is whether or not KIT is expressed in meiotic cells. Strong evidence that KIT is indeed expressed in meiotic and postmeiotic germ cells in testis sections has been provided by two reports that used methods of detection that were different from those used previously. Through use of a polyclonal antibody and electron microscopy, KIT expression was found on type A spermatogonia, round spermatids, and spermatozoa (Sandlow et al, 1999). This demonstration of KIT-immunoreactivity in cells that were previously shown to lack ACK2-immunoreactivity suggests that the KIT epitope recognized by ACK2 may be masked or otherwise not expressed on meiotic and postmeiotic cells. The most definitive evidence that KIT is indeed expressed at high levels on these cells in vivo has been provided by genetic evidence. Guerif et al (2002) used mice in which one allele of the Kit gene was disrupted by a targeted, inframe insertion of a lacZ reporter gene. Thus. expression of the reporter was under all of the cis-acting elements in their proper context for Kit transcriptional regulation and so should precisely mimic the pattern of endogenous Kit transcription. In the adult testis, intense reporter expression was observed in type A spermatogonia as well as in late primary spermatocytes, secondary spermatocytes, and round spermatids. Furthermore, detailed histological analyses of the testes of these mice, which have reduced KIT function because they are heterozygous for the Kit null mutation caused by insertion of the lacZ gene, revealed a significant delay in meiosis (Guerif et al, 2002). Whether this indicates a delay in initiation or a delay in progression of meiosis remains to be determined.
Collectively, all available data indicate that KITL and KIT are required for proliferation and survival of spermatogonia and for either initiation or progression of meiosis (or both) in the testis. The challenge now is to understand the molecular basis for these quite disparate processes. Do KITL and KIT have different functions in mitotic, meiotic, and postmeiotic cells? If so, are these functions mediated through differences in downstream KIT signaling pathways? Or are other factors expressed in these cells or in somatic cells of the testes that directly modulate the activities of KITL and KIT? Clearly, some of these questions could be addressed through a detailed understanding of the KITL/KIT signaling pathway and of the structure and function of these proteins. The next section briefly summarizes the current information on the KITL/KIT signaling pathway, with an emphasis on this signaling pathway in spermatogonia.
KITL/KIT Signaling Pathway
KITL signaling is required for the normal development of germ cells, erythroid cells, mast cells, and melanocytes. Although KITL is a helical cytokine (see above), KIT is a type III receptor tyrosine kinase (RTK) that has sequence and structural similarities to the platelet-derived growth factor receptor, beta polypeptide (PDGFRB) family of RTKs (Besmer et al, 1993; Lev et al, 1994). When KITL binds, KIT undergoes ligand-induced dimerization and activation of its intracellular kinase domain. Activated KIT is autophosphorylated on tyrosine residues and phosphorylates or binds to a number of cytoplasmic signaling molecules. A wide variety of such downstream signaling molecules have been identified in KIT-dependent cell types (see Feng et al, 2000; Kissel et al, 2000; Dolci 2001; De Miguel et al, 2002 and references cited therein), including phosphatidylinositol 3′-kinase (PI 3-K), the SRC nonreceptor tyrosine kinase, the mitogen-activated protein kinase kinase (MEK, also known as MAPKK), the Janus-activated kinase 2 (JAK2), the SHP1 and SHP2 protein phosphatases, phospholipase Cγ1, the GRB2 adaptor protein, the RAS protooncogene, SHC (an adaptor protein), and VAV (a GDP/GTP exchange factor). Based on the complexity of these downstream signaling pathways, it has been difficult to determine which pathway or pathways are most critical to KITL/KIT functions in vivo and to determine whether events downstream of activated KIT are the same or different in different cell types. However, important information is emerging from recent in vitro and in vivo studies (see below), and many new resources have been developed that will be useful for understanding these complex processes.
Chemicals that specifically inhibit different signaling molecules have been valuable reagents to dissect complex signaling pathways in a variety of cultured cells. This approach was used in two recent investigations of KITL signaling in cultures of purified mouse spermatogonia (Feng et al, 2000; Dolci et al, 2001). In both studies, addition of S-KITL to spermatogonial cultures resulted in stimulation of DNA synthesis. Furthermore, both studies showed that the proliferative response of these cells to S-KITL was abolished by addition of chemicals that specifically inhibit signaling through PI 3-K and that AKT, a serine threonine kinase, is required downstream of PI 3-K activation. Retinoblastoma protein (RB), a well-documented regulator of the G1/S checkpoint, was shown in both studies to be phosphorylated in response to cyclin D3 activity in response to S-KITL. Thus, the importance of PI3-K and other major downstream players in activating spermatogonial cell cycle regulators in response to KITL are indicated by both studies. Despite their general agreement, there are some differences in the two reports. In particular, Dolci et al (2001) found that S-KITL induced a proliferative response in spermatogonia from P8 mice but not spermatogonia from P5 mice. This observation is consistent with previous observations about the expression of KIT in germ cells at these ages (Manova et al, 1990; Yoshinaga et al, 1991) (ie, P5 spermatogonia are predominantly undifferentiated and so should be KIT(−) and not responsive to S-KITL, while P8 spermatogonia are predominantly differentiated and so should be KIT(+) and responsive to S-KITL). Feng et al (2000) used spermatogonia from P5 animals for their study, and these cells would not be expected to be responsive to KITL if they really were KIT(−). Because all available evidence indicates that a cellular response to KITL requires its interaction with KIT, it is likely that the P5 spermatogonia used by Feng et al (2000) were in fact KIT(+). In support of this supposition, the proliferation of spermatogonia from P5 mice (but not younger mice) was shown to be dependent on KIT (Tajima et al, 1994). Several factors, such as the strain of mice used, the method of purifying spermatogonia, and the culture conditions could affect the numbers of KIT(+) and KIT(−) cells in these testis cell preparations from young mice. In future studies of KITL signaling, it should be possible to avoid potentially contradictory results if spermatogonial preparations are enriched for KIT(+) cells or if the specific responses of KIT(−) and KIT(+) cells are examined.
The studies by Dolci et al (2001) also provided evidence that KITL-mediated suppression of apoptosis in spermatogonia does not occur through pathways involving PI3-K, MEK, or JAK2. Although the apoptotic pathway was not revealed in this study, the results suggest that KITL signaling in spermatogonia may proceed through different pathways to suppress apoptosis than to promote cell proliferation. Apoptosis is known to be a critical aspect of normal spermatogenesis and its control is subject to a number of paracrine and endocrine signals (reviewed by Print and Loveland, 2000). Although a number of known regulators of apoptosis have been shown to function during spermatogenesis, little is known of the molecular interactions leading from KITL signaling to apoptosis in testis cells. Recently, addition of S-KITL to cultures of rat seminiferous tubules was shown to cause increased expression of BCL-xL (also known as Bcl2-like, BCL2L) and BCL-w (also known as Bcl2-like 2, BCL2L2), which have prosurvival functions, and to cause decreased expression of BAX, which has proapoptotic functions (Yan et al, 2000a). In addition, genetic evidence suggests that the increased apoptosis that occurs in the testes of mice with Kit mutations is dependent on p53 (Jordan et al, 1999), a tumor suppressor protein (also known as transformation-related protein 53, TRP53), and FAS (Sakata et al, 2003), a death receptor family member (also known as tumor necrosis factor receptor superfamily, member 6, TNFRSF6). However, using transplantation of p53-deficient spermatogonia, Ohta and colleagues provided evidence that loss of germ cells from KitlSl-d testes is independent of p53 function (Ohta et al, 2003). Although these studies provide good starting points for more detailed studies, much more work will be required to gain a complete understanding of the mechanisms by which KITL affects the timing or frequency of apoptosis in the testes.
An interesting aspect of KITL signaling is that the pathway used to promote proliferation of PGCs may be slightly different from that used to promote proliferation of spermatogonia. Even though both PGCs and spermatogonia are dependent on AKT for their proliferative response to KITL, PI3-K does not appear to play a role in activating AKT in PGCs (De Miguel et al, 2002) as it does in spermatogonia and in other cell types. De Miguel et al (2002) showed that after stimulation of PGCs with S-KITL, AKT was activated by an SRC-dependent pathway and not by a PI3-K-dependent pathway. Furthermore, proliferation of PGCs was inhibited with a specific chemical inhibitor of SRC, and not with an inhibitor of PI3-K. Importantly, the in vitro studies of KITL-induced proliferation in PGCs and spermatogonia are consistent with in vivo studies described below, and it is now clear that signaling through PI3-K is important for this process in spermatogonia but not PGCs. The use of chemical inhibitors to dissect signaling pathways in cultured germ cells should continue to be a productive approach in future studies as in the studies described here. Important issues to be examined in future studies of KITL signaling are the one or more pathways used to suppress apoptosis of PGCs and spermatogonia, the stage in germ cell development in which the proliferative response to KITL switches from PI3-K independence (in PGCs) to PI3-K dependence (in spermatogonia), and the significance of that switch to the biological responses of PGCs and spermatogonia to KITL.
Genetic evidence for differences in KITL/KIT signaling between PGCs and spermatogonia has come from studies of mice carrying mutations in the Kit gene. The KitW-f mutation is one of only a few Kit or Kitl mutations that allow fertility in homozygous mutants of both sexes. Koshimizu et al (1992) showed that normal spermatogenesis occurs in KitW-f/KitW-f mutants even though the mice are born with only about 50% the normal number of gonocytes. Because the reduced number of gonocytes is likely caused by deficient PGC development, the preferential effect of the KitW-f mutation on gonocytes may reflect differences in KITL/KIT signaling between PGCs and spermatogonia. The nucleotide alterations in the KitW-f have not been identified to our knowledge but could provide important clues about KIT signaling in these cell types.
Two studies that used mice carrying targeted mutations in KIT provided conclusive evidence that PI3-K is an integral aspect of the KITL/KIT signaling pathway in spermatogenesis (Blume-Jensen et al, 2000; Kissel et al, 2000). Both laboratories independently generated mice in which tyrosine 719 in the KIT cytoplasmic domain was mutated to phenylalanine (Y719F). This mutation abolished PI3-K binding to activated KIT but left intact binding sites for other signaling molecules. Unlike mice that carry kinase-defective KIT mutations and consequently have white coat color, reduced red blood cell counts, reduced skin mast cells, and reduced numbers of PGCs (see below and reviews by Besmer et al, 1993; Lev et al, 1994), the KitY719F homozygous mice have normal coat pigment, normal peripheral blood, normal skin mast cells, and normal PGC numbers. While fertility of mutant females and oogenesis were relatively unaffected, the mutant males were sterile. Testicular development in the KitY719F mutants was normal up until P8, after which decreased proliferation and increased apoptosis of spermatogonia were observed and no meiotic or postmeiotic cells were observed. Thus, these studies clearly demonstrate that signaling through PI3-K is required for KIT function in proliferation, survival, and differentiation of spermatogonia in vivo. In addition, KITL signaling through PI3-K may affect the steroidogenic capacity of Leydig cells but not the numbers or morphology of these cells, and this steroidogenic alteration may have a deleterious effect on spermatogenesis (Rothschild et al, 2003). Although it is tempting to speculate that PI3-K signaling is not required for KIT functions in melanocytes, red blood cells, mast cells, or PGCs, the possibility of redundant or compensatory factors that take the place of PI3-K in these cells cannot be excluded.
The results obtained from the analysis of mice carrying the lacZ insertion in Kit, the KitW-f mutation, and the KitY719F mutation provide excellent examples of the power of a genetic approach in dissecting complex biological processes. Although the KitW-f mutation arose spontaneously, the lacZ insertion in Kit and the KitY719F mutations were generated through “reverse genetics” using gene targeting in embryonic stem cells. In addition to KitW-f, a large number of mouse mutants in KIT and KITL were identified through “forward genetics,” often referred to as classical genetics. In the next section, we describe such mutants and summarize the information gained about spermatogenesis through analysis of Kitl mutant mice.
Mutations in the Kitl and Kit Genes of Mice
In mice, KITL is encoded by the Steel (Sl) locus, and KIT is encoded by the Dominant White Spotting (W) locus (see reviews by Besmer et al, 1993; Lev et al, 1994). Over the last 50 years, classical genetic techniques have been used to identify spontaneous, radiation-induced, and chemically induced mutations at both loci. More information on KitlSl, KitW, and many other mouse mutants may be found online at http:www.informatics.jax.org (Mouse Genome Database, The Jackson Laboratory) and at http:bio.lsd.ornl.govmouse (Mutant Mouse Stock Database, Oak Ridge National Laboratory). Although these mutations were revealed on the basis of semidominant pigmentation phenotypes, KitlSl and KitW mutations disturb normal development of several diverse cell types. The characteristic phenotype of a KitlSl or KitW mutant is identical, with sterility in either or both sexes, macrocytic anemia, mast cell deficiency, and reduced coat pigmentation, as well as perinatal lethality in the case of homozygous null mutations. These phenotypic defects are due to deficiencies in development of germ cells, hematopoietic cells (particularly hematopoietic stem cells, erythroid cells, and mast cells), and melanocytes, and the extent of deficiencies in each cell type are allele-specific (see below). Once the products of the Sl and W loci were identified, expression studies in wild-type mice revealed that both genes are expressed in many more tissues than would be expected from the mutant phenotypes. Accordingly, it is now known that several additional cell types are deficient in KitlSl and KitW mice, including enteric neurons (Huizinga et al, 1995; Torihashi et al, 1995) and T lymphocytes (Di Santo and Rodewald, 1998). In addition, these mutant mice have been reported to have learning and memory deficiencies (Motro et al, 1996).
The molecular and phenotypic diversity of KitlSl mutants provide a rich genetic resource for dissecting KITL functions. There are more than 80 different KitlSl mutant alleles, all of which exert semidominant phenotypes. This semidominance, in which heterozygous mutants display a milder phenotype than homozygous mutants, reveals that gene dosage is critical to KITL function. Another important aspect of KitlSl mutations that allows for their use in detailed functional studies is that different KitlSl mutations produce phenotypes that are graded with respect to severity (ie, some KitlSl mutations produce very severe phenotypes whereas other KitlSl mutations produce very mild phenotypes). These milder phenotypes are often useful for fine-dissection of gene functions that may be quite subtle or may be required at later developmental stages.
DNA sequence alterations have been identified in many KitlSl mutant alleles (see Rajaraman et al, 2002a for references), including alleles that have deletions of the entire Kitl coding region, intragenic mutations in the Kitl coding region, and intergenic mutations that occur outside of the Kitl coding region. KitlSl alleles with coding mutations include intragenic deletions, splicing defects, or nonsense mutations that result in the absence of specific KITL domains, whereas other alleles have point mutations that cause missense mutations. KitlSl-null mutations, such as the complete Kitl deletions, some of the nonsense and missense mutations, and an intragenic duplication, cause prenatal or perinatal lethality with severe anemia (Rajaraman et al, 2002b, 2003; Chandra et al, 2003). In addition, embryos homozygous for KitlSl-null mutations have severe deficiencies in PGC development (A. Mahakali Zama, F.P. Hudson III, and M.A. Bedell, in preparation) but they are not useful for studies of postnatal germ cell development because they fail to live beyond the first few days after birth. On the other hand, homozygous hypomorphic KitlSl mutations allow viability but have milder effects on RBCs (Rajaraman et al, 2002b) and slightly milder effects on PGCs (Brannan et al, 1992; A. Mahakali Zama, F. P. Hudson III, and M. A. Bedell, in preparation). Two KitlSl mutations, KitlSl-d and KitlSl-17H, both of which contain coding mutations that affect whole domains of KITL and are hypomorphic, have been used in a number of studies to define the role of KITL during early spermatogenesis.
The KitlSl-d allele contains an intragenic deletion of 4 kb from genomic DNA that removes the transmembrane and cytoplasmic domains and potentially encodes a normal S-KITL (Figure) but completely lacks MB-KITL (Brannan et al, 1991; Flanagan et al, 1991). Although KitlSl-d homozygous mice are viable and severely anemic, their anemia is milder than that caused by null KitlSl mutations (Rajaraman et al, 2002b). This indicates that the KitlSl-d allele encodes a partially functional KITL protein. In vitro studies have revealed that the KITLSl-d protein or proteins are biologically active (Brannan et al, 1991; Dolci et al, 1991). However, little is known about the in vivo expression levels or posttranslational modification of KitlSl-d encoded proteins. On the basis of nucleotide sequence, three different KITLSl-d proteins may be expressed, of which all three lack the transmembrane and cytoplasmic domains and two contain additional sequences at the truncated C-terminus (Figure). The 5′ end of the intragenic deletion in KitlSl-d gene is in exon 7, and so two mRNAs that differ by the presence or absence of exon 6, as in wild-type Kitl, could be produced that express two truncated proteins. It is likely that the absence of the transmembrane domain or the presence of the abnormal sequences (or both) would result in greatly reduced expression of both of these proteins at the cell surface or abnormal folding and glycosylation (or both). Significantly, if the abnormal Pre-KITLSl-d protein does reach the cell surface, proteolysis could occur at the normal cleavage site to produce a S-KITL whose sequence is identical to that of wild-type. However, spacing between the primary cleavage site and the transmembrane domain of Pre-KITL is critical for efficient proteolysis (Cheng and Flanagan, 1994), and so it is not clear whether this normal S-KITL would even be produced by the KitlSl-d allele in vivo.
The KitlSl-17H allele contains a point mutation that affects splicing (Brannan et al, 1992) such that an MB-KITL and a Pre-KITL with abnormal cytoplasmic do-mains are encoded (Figure), but the mutation does not affect the proteolytic cleavage site for the release of S-KITL. Although the normal cytoplasmic domain is not required for dimerization of KITL, the abnormal cytoplasmic domain in the KITLSl-17H protein interferes with its dimerization (Tajima et al, 1998). In addition, there is evidence that the KitlSl-17H mutation is a gain-of-function mutation, because the KITLSl-17H cytoplasmic domain contains sequences for abnormal routing of the mutant protein to lysosomes (Wehrle-Haller and Imhof, 2001). It is interesting that Wehrle-Haller and Weston (1999) reported that of the KITLSl-17H that reaches the cell surface, most of it localizes to the apical compartment rather than to the basolateral compartment of polarized epithelial cells. These authors suggested that such mislocalization could also occur in Sertoli cells, and because less KITL would be available at the basolateral surface for presentation to spermatogonia, such a model provides a plausible explanation for spermatogenesis defects in KitlSl-17H homozygotes. A possible alternative explanation for reduced KITL function in these mutants is that the normal KITL cytoplasmic domain may elicit one or more signals within Sertoli cells.
KitlSl-d and KitlSl-17H homozygous mice are born with greatly reduced numbers of germ cells due to defects in development of PGCs (reviewed by Donovan, 1994). Because so few germ cells are present in these mutants at birth, it is difficult to separate direct effects of these mutations on spermatogenesis from indirect effects on this process that result from germ cell deficiency. Nonetheless, examination of the testes of these mutants has revealed important information about KITL function during spermatogenesis. In all aspects of spermatogenesis, the phenotypes reported are slightly milder in KitlSl-17H mutants than in KitlSl-d mutants. This suggests that there must be more KITL function in the former mice. Despite the severe paucity of germ cells, some tubules in the KitlSl-17H prepubertal testes contain germ cells that undergo one wave of fairly normal spermatogenesis (Brannan et al, 1992). Complete cessation of spermatogenesis occurs by 8 weeks of age in the KitlSl-17H mutants, suggesting that spermatogenesis is regulated differently in prepubertal mice compared with adult mice. Indeed, transplantation studies (described above) have provided conclusive evidence that prepubertal and adult testes differ with respect to functional properties of spermatogonia and the environment in which these cells proliferate and differentiate. In KitlSl-17H and KitlSl-d mutants at about 3 months or older, nearly 40% and 15%, respectively, of the tubules contain spermatogonia but no meiotic or postmeiotic cells (Ohta et al, 2000). These results confirm the conclusions drawn from the use of the ACK2 neutralizing antibody that KITL and KIT are essential to the survival and proliferation of spermatogonia. However, in the case of the KitlSl mutants, none of the persisting spermatogonia were shown to be KIT(+), indicating that they are undifferentiated spermatogonia. This observation is consistent with a previous report that in mature KitlSl-17H mutants, all spermatogonia were arrested at the Aaligned (Aal) stage, which is believed to be the last stage of undifferentiated spermatogonia (de Rooij and Grootegoed, 1998). It is interesting that proliferation of undifferentiated spermatogonia in the KitlSl mutants is at a much higher rate than in wild-type mice (de Rooij and Grootegoed, 1998; Ohta et al, 2000), but the numbers of these cells in the mutants do not increase because of a high rate of apoptosis (de Rooij and Grootegoed, 1998). Another growth factor, called glial cell line-derived neurotrophic factor (GDNF), is also expressed by Sertoli cells and regulates spermatogonial proliferation (Meng et al, 2000). It has been proposed that the increased proliferation of undifferentiated spermatogonia in KitlSl mutants is due to GDNF, because the mutant testes express a greatly increased amount of GDNF mRNA and the proliferation of these cells in the mutant testes was blocked by administration of an anti-GDNF neutralizing antibody (Tadokoro et al, 2002).
Transplantation of germ cells into seminiferous tubules has been a powerful method for elucidating functional and molecular properties of spermatogonial stem cells. By using donor and recipient mice of different genotypes with respect to KitlSl and KitW mutations, this procedure has provided another level of information about the functions of KITL and KIT during spermatogenesis. Transplantation of undifferentiated spermatogonia from KitlSl-d donors into the testes of wild-type recipients resulted in complete spermatogenesis and spermiogenesis of the donor-derived germ cells (Ohta et al, 2000; Shinohara et al, 2000a). The same result occurred when Kitl complementary DNAs were introduced into KitlSl-d testes by electroporation (Yomogida et al, 2002) and recombinant viral infection (Ikawa et al, 2002; Kanatsu-Shinohara et al, 2002), indicating that the restoration of spermatogenesis in the mutant testes is not an artifact due to the transplantation procedure. These results demonstrate that true stem cells are present in KitlSl-d testes, and their arrested differentiation is reversible as long as functional KITL is provided. However, transplantation experiments revealed the unexpected finding that KitlSl-d testes have fewer stem cells than expected, even after accounting for the reduced numbers of PGCs (Shinohara et al, 2000a). This observation raises the interesting possibility that KITL function is required to maintain spermatogonial stem cells in situations, such as in the KitlSl mutant testes, in which the numbers of these stem cells are abnormally low. Such a possibility does not contradict the large body of evidence that spermatogonial stem cells are KITL-independent when the cells are present in normal numbers. Indeed, a multitude of abnormal molecular and cellular interactions are expected in germ cell-deficient testes, any of which could result in either altered KITL/KIT signaling or altered responsiveness of germ cells to KITL/KIT signaling.
The reciprocal transplantation experiment, in which cells from wild-type testes were transplanted into KitlSl mutant recipients, has also yielded important information. When such an experiment was conducted with KitlSl-d and KitlSl-17H mutant recipients, the donor spermatogonia proliferated and expanded in number (Ohta et al, 2000). Significantly, no differentiation of the donor cells was observed in these mutant testes, providing important confirmation that KITL function is required for spermatogonial differentiation. However, the increased numbers of proliferating donor spermatogonia in the KitlSl-d recipients appears to contradict the observation that endogenous spermatogonia in these mutant testes proliferate but do not increase their numbers because of increased apoptosis (de Rooij and Grootegoed, 1998). If this apparent contradiction holds true, it raises the following question: If the numbers of endogenous spermatogonia in the KitlSl-d mutant testes cannot increase, then why can the numbers of transplanted spermatogonia in the KitlSl-d mutant testes increase? Although many issues remain to be resolved about KITL functions during spermatogenesis, it is clear that KITL has a significant role in the intricate balance between survival, proliferation, and differentiation of spermatogonia and that this balance may be directly or indirectly affected by the numbers of undifferentiated germ cells.
New Hypomorphic KitlSl Mutations
We recently described the nucleotide sequence alterations in 10 KitlSl mutant alleles, of which 8 have little or no functional activity for mouse survival or development of peripheral blood cells, while 2 of the alleles (KitlSl-36R and KitlSl-39R) are hypomorphic for these activities (Rajaraman et al, 2002a,b, 2003). In the KitlSl-36R allele, a point mutation in exon 5 was identified that creates a nonsense mutation in codon 147 (Rajaraman et al, 2002a). Thus, a mutant product of only 146 amino acids of S-KITL (Figure) would be produced by premature termination of both (+)E6 Kitl and (−)E6 Kitl messenger RNAs. In addition, abnormal splicing (skipping) of exon 5 was observed and is expected to produce a second KITLSl-36R isoform of only the first 96 amino acids of S-KITL with 25 C-terminal amino acids out-of-frame (Figure). Evidence to date indicates that this second isoform is likely to be functionally null (Rajaraman et al, 2002b), and so it is likely that the 146-amino acid form is the only biologically active form expressed by the KitlSl-36R allele. The KitlSl-39R mutation results from a point mutation that causes an S122F missense mutation in the fourth α-helix and disrupts one of the four N-linked glycosylation sites (Asn120) identified in mouse KITL (Figure). While KITL expressed from tissue culture cells is known to be glysosylated at Asn120 (Lu et al, 1991, 1992; Huang et al, 1992), the effect, if any, that the absence of this glycan has on KITL function in vivo is not known.
Our recent studies have revealed that the KitlSl-17H, KitlSl-36R, and KitlSl-39R mutations exert equivalent effects on the numbers of postmigratory germ cells in the embryo that result in reduced number of germ cells in the postnatal testes (A. Mahakali Zama, F.P. Hudson III, and M.A. Bedell, in preparation). It is interesting that the testes of these mutants differ with respect to the severity of defects in spermatogenesis (A. Mahakali Zama and M.A. Bedell, in preparation). At 8 weeks of age, all mice homozygous for the KitlSl-17H, KitlSl-36R, or KitlSl-39R mutations display a complete cessation in spermatogenesis due to the depletion of differentiating germ cells. However, analysis of younger mice (3 to 5 weeks) revealed that the KitlSl-36R and KitlSl-17H mutations have nearly identical effects, but both mutations have effects on spermatogenesis that are more severe than that of KitlSl-39R. In the latter, many tubules lack germ cells, but the numbers of tubules with germ cells is greater in this mutant than in KitlSl-36R or KitlSl-17H mutants. Also, unlike the KitlSl-36R and KitlSl-17H mutants, some tubules from the young KitlSl-39R mutants contain all stages of mitotic, meiotic, and postmeiotic cells (A. Mahakali Zama and M.A. Bedell, in preparation). The milder effect of the KitlSl-39R mutation on the testis provides another genetic resource from which it may be possible to dissect further aspects of KITL function in the testes.
It is interesting that the relative effects of the hypomorphic KitlSl mutations in the testis are different from their effects on peripheral blood cells (A. Mahakali Zama and M.A. Bedell, in preparation). For example, while the KitlSl-17H mutation has a milder effect on red blood cells than the KitlSl-39R mutation, the KitlSl-39R mutation has a milder effect on spermatogenesis than the KitlSl-17H mutation. Although the basis for these preferential effects on different cells are not currently known, these differences suggest that in the bone marrow and testis, either the cells expressing the mutant proteins (stromal cells versus Sertoli cells) differ with respect to their expression or posttranslational modification of KITL (or both) or that hematopoietic cells and spermatogonia respond in different ways to the mutant KITL proteins.
Conclusions and Future Directions
Although it is clear that the primary functions of KITL in spermatogenesis are in differentiated spermatogonia and in meiosis, several interesting questions remain to be resolved about these functions. What are the molecular mechanisms that underlie the KITL functional differences between promoting proliferation and survival of differentiating spermatogonia? What role, if any, does KITL play in spermatogonial stem cells? Is KITL not required at all in these cells, or is it required only under certain circumstances, such as in neonatal mice or in germ celldeficient adults? Does KITL directly regulate initiation and progression of meiosis? If so, what are the molecular differences that underlie its functions in mitotic versus meiotic germ cells? Are the signaling pathways downstream of activated KIT different in male germ cells from other KITL/KIT requiring cells, such as hematopoietic cells or melanoblasts? If so, what are those differences and what effect do those molecular differences have on the biological response of the cells?
Further characterization of KITL function using existing or newly generated mouse strains with mutations in either KITL or KIT, combined with in vitro culture and testis cell transplantation, will ultimately allow dissection of these and other aspects of KITL functions.
The authors thank Merrill C. Morris and two anonymous reviewers who made helpful suggestions that improved the manuscript.
Supported by grant GM065393 from the National Institute of General Medical Sciences.
In using “Kit ligand” and “Kitl” instead of other names, we are following the guidelines established by the International Committee on Standardized Genetic Nomenclature for mice (Maltais et al, 2002). According to these guidelines, the mRNA and protein products of the Sl locus are Kitl and KITL, respectively, and mutation symbols become the allele symbols, which are shown as superscripts to the gene symbol (eg, the Sl-d mutation is designated KitlSl-d). The guidelines for nomenclature and all recommended gene names and allele symbols for mice are available at the following Web address: www.informatics.jax.orgmgihomenomenindex.shtml.