Kit ligand (Kitl), the ligand for the Kit receptor tyrosine kinase, plays important roles in hematopoiesis, gametogenesis and melanogenesis. Kitl is synthesized as a membrane-anchored precursor that can be processed to produce the soluble growth factor. Here, we evaluated the role of ADAM (a disintegrin and metalloprotease) metalloproteases in ectodomain shedding of Kitl. We found that both ADAM17 and ADAM19 affect Kitl1 shedding, albeit in different ways. Overexpression of ADAM19 resulted in decreased levels of Endo-H-resistant mature Kitl1, thereby reducing the amount of Kitl that is shed from cells following stimulation with phorbol esters. ADAM17 was identified as the major phorbol-ester-stimulated sheddase of Kitl1, whereas ADAMs 8, 9, 10, 12 and 15 were not required for this process. ADAM17 also emerged as the major constitutive and phorbol-ester-stimulated sheddase of Kitl2 in mouse embryonic fibroblasts. Mutagenesis of the juxtamembrane domain of Kitl2 showed no stringent sequence requirement for cleavage by ADAM17, although two nonadjacent stretches of four amino acid residues were identified that are required for Kitl2 shedding. Taken together, this study identifies a novel sheddase, ADAM17, for Kitl1 and Kitl2, and demonstrates that ADAM19 can reduce ADAM17-dependent phorbol-ester-stimulated Kitl1 ectodomain shedding.
Proteolytic processing of transmembrane proteins to release their soluble extracellular domains from the membrane, a process termed ectodomain shedding, is emerging as an important post-translational mechanism to regulate the function of membrane proteins. A variety of structurally and functionally distinct cell surface proteins are subjected to protein ectodomain shedding, including cytokines such as tumor necrosis factor-α (TNFα), growth factors such as transforming growth factor-α (TGFα) and epidermal growth factor (EGF), receptors such as the tumor necrosis factor receptor (TNFR) 1 and TNFR2 and Notch, and adhesion molecules such as L-selectin (reviewed by Blobel, 2005; Hooper et al., 1997; Schlöndorff and Blobel, 1999; Seals and Courtneidge, 2003). The functional consequences of ectodomain shedding are diverse and include regulation of availability or active range of membrane-bound growth factors and the activation or inactivation of their receptors.
The membrane growth factor Kit ligand (Kitl), also referred to as stem cell factor (SCF), and its tyrosine kinase receptor Kit are encoded at the mouse White spotting (W) and Steel (Sl) loci, respectively (Chabot et al., 1988; Copeland et al., 1990; Geissler et al., 1988; Huang et al., 1990; Zsebo et al., 1990). Analyses of phenotypes of mutant mice have shown that the Kit receptor tyrosine kinase functions in distinct cell populations during embryonic development and in hematopoiesis, gametogenesis, melanogenesis in the postnatal animal and in interstitial cells of Cajal in the intestinal tract (Russell, 1979; Silvers, 1979). Kitl binds to its cognate receptor Kit leading to receptor dimerization, activation of its kinase activity, autophosphorylation and activation of various signaling cascades leading to cell proliferation, differentiation, migration and survival (Blume-Jensen et al., 1991; Lev et al., 1992). Although Kitl was originally identified as a soluble protein (Nocka et al., 1990), Kitl is synthesized as a transmembrane protein that may be processed to produce soluble Kitl (Huang et al., 1990). Alternative splicing generates two Kitl RNA transcripts that encode two cell-associated Kitl proteins, designated Kitl1 and Kitl2 (Flanagan et al., 1991; Huang et al., 1992). Whereas the Kitl1 protein is rapidly cleaved at a major cleavage site located in exon 6, releasing a soluble protein, by contrast the Kitl2 protein lacks the exon 6 sequences and is processed less efficiently (Huang et al., 1992). Although both membrane-bound and soluble Kitl appear to be biologically active in vitro, the phenotypes of mice carrying the Sld allele, which produces only a secreted soluble Kitl protein as a result of an intragenic deletion, imply that membrane Kitl and presumably juxtacrine signaling are crucial for normal function in vivo (Brannan et al., 1991; Flanagan et al., 1991). In agreement with this conjecture, mice that express Kitl2 exclusively have only minor deficiencies in hematopoiesis and mast cell development (Tajima et al., 1998).
The mechanisms responsible for Kitl processing are poorly understood. Early studies of Kitl processing showed that the production of soluble Kitl in COS cells is inducible and may be regulated by the protein kinase C activator, phorbol 12-myristate 13-acetate (PMA) (Huang et al., 1992). Studies using different protease inhibitors implied that Kitl1 and Kitl2 are processed by distinct proteolytic activities (Pandiella et al., 1992). Some proteases, including matrix metalloproteinase (MMP) 9 (Heissig et al., 2002), mast cell chymase (chymotrypsin-like protease) (Longley et al., 1997) and several members of the ADAMs (a disintegrin and metalloprotease) family of metallopeptidases (Amour et al., 2002; Chesneau et al., 2003; Mohan et al., 2002; Roghani et al., 1999; Zou et al., 2004) (Table 1) have been suggested to have activities that mediate Kitl shedding. Thus, tissue expression of differentially spliced mRNAs encoding Kitl1 and Kitl2 and the tissue-specific expression of proteolytic activities capable of processing Kitl to produce soluble Kitl may provide mechanisms for the regulation of Kitl function.
|ADAMs .||Kitl1 peptide cleavage sites .||References .|
|ADAM8||PPVAA↓SSLRN||(Amour et al., 2002)|
|ADAM9||PPVA↓A↓S↓SLRN||(Roghani et al., 1999)|
|ADAM10||No cleavage||(Amour et al., 2002)|
|ADAM17||PPVAA↓SSLRN||(Mohan et al., 2002)|
|ADAM19||PPVAA↓SSLRN||(Chesneau et al., 2003)|
|ADAM33||PPVAA↓SSLRN||(Zou et al., 2004)|
|ADAMs .||Kitl1 peptide cleavage sites .||References .|
|ADAM8||PPVAA↓SSLRN||(Amour et al., 2002)|
|ADAM9||PPVA↓A↓S↓SLRN||(Roghani et al., 1999)|
|ADAM10||No cleavage||(Amour et al., 2002)|
|ADAM17||PPVAA↓SSLRN||(Mohan et al., 2002)|
|ADAM19||PPVAA↓SSLRN||(Chesneau et al., 2003)|
|ADAM33||PPVAA↓SSLRN||(Zou et al., 2004)|
ADAMs are a large family of transmembrane metallopeptidases that have crucial roles in ectodomain shedding (for reviews, see Blobel, 2005; Seals and Courtneidge, 2003; White, 2003). ADAMs consist of an N-terminal signal sequence followed by a pro domain, metalloprotease domain, disintegrin domain, cysteine-rich domain, epidermal growth factor-like domain, transmembrane domain and cytoplasmic domain. ADAMs have been implicated in diverse functions, including fertilization (Primakoff and Myles, 2000), neurogenesis (Fambrough et al., 1996; Hattori et al., 2000), angiogenesis (Hartmann et al., 2002; Horiuchi et al., 2003), adipogenesis (Kawaguchi et al., 2003; Kawaguchi et al., 2002), heart development (Horiuchi et al., 2005; Jackson et al., 2003; Kurohara et al., 2004; Zhou et al., 2004), asthma (Van Eerdewegh et al., 2002) and cancer (Kveiborg et al., 2005; Peduto et al., 2005). A conserved zinc-binding catalytic site sequence motif (HEXXH) (Stocker et al., 1995) is found in approximately half of the known ADAMs. Of these, ADAM17 (also referred to as tumor necrosis factor-α converting enzyme, TACE) is one of the best-characterized members of this gene family and is known to be required for PMA-stimulated ectodomain shedding of a broad number of transmembrane protein substrates, including the pro-inflammatory cytokine TNFα (Black et al., 1997; Moss et al., 1997; Zheng et al., 2004) and the EGF-receptor (EGFR) ligands amphiregulin, heparin-binding EGF-like growth factor (HB-EGF) and TGFα (Merlos-Suarez et al., 2001; Peschon et al., 1998; Sahin et al., 2004; Sternlicht et al., 2005; Sunnarborg et al., 2002). Mice lacking ADAM17 die shortly after birth, most probably because of defects in the functional activation of the EGFR ligands (Jackson et al., 2003; Peschon et al., 1998; Sternlicht et al., 2005). Recently, ADAM17 was shown to mediate the shedding of the Kit receptor tyrosine kinase ectodomain, the only known receptor for Kitl (Cruz et al., 2004).
In this study, we evaluated the contribution of catalytically active ADAMs to Kitl shedding. Although previous studies that tested whether purified ADAMs can cleave Kitl1 peptides covering the major cleavage site revealed a possible role of several ADAMs in Kitl1 shedding (Amour et al., 2002; Chesneau et al., 2003; Mohan et al., 2002; Roghani et al., 1999; Schlomann et al., 2002) (see Table 1), the actual role of those ADAMs in Kitl shedding in cells or in vivo has not yet been evaluated. To address this issue, we isolated primary mouse embryonic fibroblasts (mEFs) from mice that are deficient in several widely expressed, catalytically active ADAMs (8, 9, 10, 12, 15, 17 and 19), and evaluated how the loss of these ADAMs affects Kitl1 as well as Kitl2 shedding. Moreover, we evaluated the mechanism underlying the previously observed function of ADAM19 as a negative regulator of PMA-stimulated Kitl1 shedding (Chesneau et al., 2003). Our results demonstrated that both ADAM17 and ADAM19 affect Kitl1 processing in cell-based assays, and identified ADAM17 as the major sheddase for Kitl2. Interestingly, whereas processing by ADAM17 resulted in ectodomain shedding of Kitl1 from cells, overexpression of ADAM19 decreased PMA-stimulated Kitl1 ectodomain shedding by reducing the levels of mature Kitl1 available for release, thus implicating an ADAM in a processing event that decreases stimulated release of a membrane-anchored substrate.
ADAM19 reduces stimulated processing of Kitl1
The ability of ADAM19 to cleave Kitl1 peptides in vitro (Table 1) prompted us to evaluate the potential role of ADAM19 in shedding of Kitl1 in cell-based assays. To increase the sensitivity of detection of the precursor and shed forms, we used expression plasmids encoding full-length Kitl1 fused to an alkaline phosphatase (AP) module at the N-terminus, such that the AP module is present in the extracellular domain of Kitl1 and was released into the culture medium after shedding. Since a previous study suggested that overexpression of ADAM19 strongly reduced the PMA-stimulated shedding of Kitl1 (Chesneau et al., 2003), and because a robust response to short treatment with PMA (30-60 minutes) is considered a hallmark property of the related ADAM17 (Horiuchi et al., 2007), we first tested whether overexpression of ADAM19 also affects the shedding of TGFα, the PMA-stimulated release of which depends on ADAM17. We found that under conditions where overexpression of ADAM19 strongly reduced PMA-stimulated shedding of Kitl1 compared with the catalytically inactive E>A mutant of ADAM19 (Fig. 1A, top panel, compare lanes 2 and 4) (see also Chesneau et al., 2003), or compared with cells expressing only Kitl1 (data not shown), overexpression of ADAM19 did not affect the PMA-dependent shedding of TGFα compared with ADAM19E>A (Fig. 1A, bottom panel, compare lanes 2 and 4). These results argue against a direct effect of ADAM19 on the activity of ADAM17, and demonstrate that the catalytic activity of ADAM19 is necessary for its effect on Kitl1 shedding. When we examined the lysates of cells expressing Kitl1 together with either ADAM19 or ADAM19E>A, we found similar expression levels of a 200 kDa pro-form of Kitl1. However, the levels of a slower migrating form of Kitl1 of 240 kDa were strongly reduced in cells expressing ADAM19 compared with controls expressing the inactive ADAM19E>A mutant (Fig. 1A, top panel, lanes 5 and 6, and Fig. 1B). The expression of pro-TGFα was comparable in cells expressing wild-type or E>A mutant ADAM19 (Fig. 1A, lower panel, lanes 5 and 6). Essentially identical results were obtained when these experiments were repeated in Chinese hamster ovary (CHO) cells (data not shown).
ADAM17, but not ADAM8, 9, 10, 12 or 15 is required for stimulated Kitl1 shedding in mEFs
The ability of several purified catalytically active ADAMs to cleave a peptide corresponding to the major constitutive cleavage site of Kitl1 in vitro (shown in Table 1) (Amour et al., 2002; Chesneau et al., 2003; Mohan et al., 2002; Roghani et al., 1999; Schlomann et al., 2002) raised the possibility that members of the ADAM family of membrane-anchored metallopeptidases function as Kitl sheddases. To test this hypothesis, we performed Kitl1 shedding assays in primary mEFs derived from different ADAM-deficient mice at embryonic day E13.5 [Adam8–/–, Adam9/12/15–/– (homozygous mutant form of Adam9, Adam12 and Adam15), Adam17–/– and wild-type controls]. In the case of Adam10–/– and Adam10+/– cells, only immortalized mEFs could be used since Adam10–/– mice do not survive past E9.5 (Hartmann et al., 2002). As seen in Fig. 2A, shedding of Kitl1 from wild-type mEFs can be stimulated by addition of PMA or pervanadate (PV) for 1 hour. Both PMA- and PV-stimulated shedding in Adam8–/–, Adam9/12/15–/– and Adam10–/– mEFs was similar to that seen in the corresponding wild-type control mEFs. Even though there was some variability in the level of stimulation by PMA or PV from experiment to experiment, the control cells that were analyzed at the same time together with Adam8–/– or Adam9/12/15–/– or Adam10–/– cells always displayed a comparable level of stimulation as the Adam–/– cells (Fig. 2A). However, the PMA-dependent increase in Kitl1 shedding was abolished in Adam17–/– mEFs, and the response to PV was significantly reduced (Fig. 2A, bottom panel, lanes 3, 4, 7 and 8). These results provide the first evidence for an essential role of ADAM17 in PMA- and PV-stimulated Kitl1 shedding in mEFs. The BB94-dependent reduction of Kitl1 shedding was not affected in any of the Adam–/– mEFs tested here compared with wild-type controls, arguing against a role for ADAMs 8, 9, 10, 12, 15, 17 and 19 in constitutive BB94-sensitive Kitl1 shedding. Finally, PMA- and PV-stimulated shedding of Kitl1 in Adam17–/– cells could be restored by reintroduction of wild-type mouse ADAM17 (Fig. 2B, PV-stimulated shedding not shown).
Overexpression of ADAM19 decreases levels of mature Kitl1
The results shown in Fig. 2 demonstrate that PMA- and PV-stimulated shedding of Kitl1 depend on ADAM17. However, since overexpression of ADAM19 did not affect the stimulated shedding of TGFα, which also depends on ADAM17, this suggested that the effect of overexpressed ADAM19 on Kitl1 shedding is not a consequence of inactivation of ADAM17. To gain further insights into the mechanism underlying the decrease in PMA- and PV-stimulated Kitl1 shedding in cells overexpressing ADAM19, we performed a pulse-chase analysis to determine whether ADAM19 might affect intracellular maturation or processing of Kitl1. COS-7 cells were transfected with Kitl1 alone, or Kitl1 with either wild-type ADAM19 or the ADAM19E>A mutant, then labeled with [35S]-methionine for 30 minutes and chased for different time periods as indicated in Fig. 3. The cell lysates were collected and immunoprecipitated with antibodies against mouse Kitl. In cells expressing only Kitl1 and in cells coexpressing catalytically inactive ADAM19E>A or wild-type ADAM19 together with Kitl1, similar amounts of core-glycosylated forms of Kitl1 were visible at all time points throughout the chase period (Fig. 3, lanes 1–4, indicated by asterisks). However, after 1, 3 and 6 hours of chase, very little of the slowest migrating form of Kitl1, indicated by an black arrow in Fig. 3, was visible in cells coexpressing wild-type ADAM19, whereas much higher levels of this mature form of Kitl1 accumulated in cells transfected with the ADAM19E>A mutant and in cells expressing only Kitl1. To assess whether ADAM19 reduces the amount of Kitl1 at the cell surface, we used a membrane-impermeable biotinylation reagent to label proteins on the surface of cells expressing Kitl1 and ADAM19 or ADAM19E>A. Kitl1 was immunoprecipitated with an antibody against its AP-tag, transferred to nitrocellulose and biotinylated material was detected with horseradish peroxidase (HRP)-labeled streptavidin. This corroborated that overexpression of ADAM19 reduces the amount of Kitl1 on the cell surface (Fig. 3B, lower panel), but does not reduce the core-glycosylated form of Kitl1 detected by western-blot analysis with an anti-AP antibody (Fig. 3B, upper panel, marked by asterisk). The proteasome inhibitor MG132 did not block the effect of wild-type ADAM19 on appearance of mature pro-Kitl1, arguing against a role of proteasomal degradation in this process (Fig. 3A, lower panel).
ADAM17 is required for stimulated shedding of Kitl2
Kitl2 is a splice variant of Kitl lacking exon 6, which contains the major physiological cleavage site of Kitl1 (Fig. 4A). Kitl2 is therefore more likely to remain as a membrane-associated form. When we investigated whether ADAM19 might also affect Kitl2 shedding, we found no evidence for an effect of overexpressed ADAM19 on PMA-stimulated shedding of Kitl2 compared with cells co-transfected with the ADAM19E>A mutant or cells expressing only Kitl2 (Fig. 4B). Thus, ADAM19 only affects the shedding of Kitl1, but not that of Kitl2. However, in Adam17–/– cells, both PMA- and PV-stimulated shedding of Kitl2 were almost completely abolished (Fig. 4C, PMA data not shown). Overexpression of mouse ADAM17 rescued constitutive and stimulated shedding in Adam17–/– cells (Fig. 4C), confirming that the defect of shedding in Adam17–/– cells is because of the loss of ADAM17. This result demonstrates that ADAM17 is the major constitutive and PMA-stimulated sheddase for Kitl2 in mEF cells.
Evaluation of the effect of deletions on Kitl shedding
Whereas soluble Kitl forms non-covalent homodimers, each Kitl monomer contains two intra-chain disulfide bonds, C4-C89 and C43-C138 (Arakawa et al., 1991; Jiang et al., 2000; Langley et al., 1992; Zhang et al., 2000). The Kitl1-141 sequences represent the functional core of Kitl that contains the dimer interface and the domain involved in Kit receptor binding and are sufficient to mediate cell proliferation activity (Langley et al., 1994). In addition, previous reports have identified a major cleavage site at A164 or A165 (Zsebo et al., 1990), and two additional minor secondary cleavage sites near S142 and KAAK177-180 (Cheng and Flanagan, 1994; Majumdar et al., 1994). In the light of these observations, we hypothesized that the cleavage site for ADAM17 in Kitl must be located C-terminally of C138, and thus potentially around the boundary region of exon 6. In order to test this hypothesis, we prepared eight deletion mutants in this region (Fig. 5A) and evaluated their effect on Kitl shedding.
In order to rule out that these deletions cause improper protein folding, which would result in retention by endoplasmic reticulum (ER) chaperones and intracellular degradation, we used acquisition of resistance to endodeglycosidase H (Endo H) as an indicator that the expressed mutant glycoproteins were able to fold, had been released from the ER and had traversed the medial Golgi complex. If a mutant did not acquire resistance to Endo H, this would suggest that the mutation prevented proper protein folding, resulting in retention in the ER. Since an unfolded mutant protein would be unable to encounter a functional sheddase such as ADAM17, which only becomes active after its pro-domain is removed in the trans-Golgi network, any mutation(s) that affect(s) protein folding would not be informative in terms of how the mutation influences processing by ADAM17. COS-7 cells expressing wild-type Kitl1 were metabolically labeled with [35S]-methionine, chased for 6 hours and their cell lysates were either treated with Endo H or peptide N-glycanase F (PNGase F), which removes all N-linked carbohydrate residues, or mock treated (Fig. 5B, top panel, lanes 1–3). The slowest migrating form of Kitl1 (indicated by asterisks) was resistant to Endo H, indicating that it had passed through the medial Golgi complex, whereas the faster migrating intermediate form (indicated by the black arrowhead) was sensitive to Endo H treatment. Wild-type Kitl2 as well as all Kitl2 mutants gave rise to comparable relative levels of Endo-H-resistant forms, confirming that all Kitl2 mutants are released from the ER and can pass through the medial Golgi complex. Immunofluorescence analysis of COS-7 cells that were either permeabilized with Triton X-100 or not permeabilized showed that the staining pattern of all mutants was indistinguishable from that of wild-type Kitl2 (data not shown). The only exception was the deletion mutant of Kitl1, KL1DM, which did not acquire resistance to Endo H, suggesting that this very membrane-proximal deletion mutation prevented proper protein folding. The subsequent shedding studies therefore focused on Kitl2 and Kitl2 mutants that were not retained in the ER.
First, we tested the shedding profile of Kitl2 mutants with sequential four amino acid deletions: SSTL141-144, PEKG146-149 or KAAK150-153 (KL2SSTL, KL2PEKG and KL2KAAK). Surprisingly, none of these deletions abolished constitutive or PV-stimulated shedding (Fig. 5C). Therefore, we next tested mutants with larger deletions of eight or nine amino acid residues. Interestingly, the shedding of a mutant carrying two non-adjacent deletions of four amino acids each (SSTL141-144 and KAAK150-153; KL2STTL/KAAK) was completely abolished, whereas the shedding of two mutants containing a deletion of nine contiguous amino acid residues covering either the SSTL141-144 or KAAK150-153 sequence (KL2SSTLGPEKG or KL2GPEKGKAAK) was strongly stimulated by PV (Fig. 5C). When we tested the shedding of the mutants described above in the presence of PMA, we obtained essentially the same results as following stimulation with PV (data not shown).
Proteolytic release of Kitl from its membrane anchor is known to have an important role in regulating juxtacrine versus paracrine functions of Kitl. This prompted us to identify and characterize novel proteases that are able to process Kitl1 as well as the splice variant Kitl2, which lacks several cleavage sites for proteolytic enzymes that can participate in ectodomain shedding of Kitl1. Our results identified two ADAMs, ADAM17 and ADAM19, as enzymes that can affect Kitl processing. We will first discuss how ADAM19 might affect the intracellular maturation or degradation of Kitl1, thereby reducing its stimulated shedding into the supernatant. Then we will discuss the identification of ADAM17 as a major phorbol-ester- and phosphatase-inhibitor-stimulated sheddase of Kitl1 and Kitl2, and will consider the implications of these results for the function of Kitl1 and Kitl2 in vivo.
Previously, we had found that overexpression of ADAM19, which is able to process the cleavage site peptide for Kitl1 in vitro, decreased the stimulated shedding of Kitl1 from cells (Chesneau et al., 2003). Since overexpression of ADAMs usually increases the release of their substrates into the cell supernatant, this result raised the possibility that ADAM19 either cleaves and inactivates the sheddase for Kitl1 or activates an inhibitor of Kitl1 shedding. In the current study, we found that increased expression of ADAM19 did not affect the release of TGFα, a substrate of ADAM17. This argues against an effect of overexpressed ADAM19 on the catalytic activity of ADAM17. Instead, we observed that overexpression of ADAM19 strongly reduced the ratio of mature membrane-associated pro-Kitl1 in cell lysates compared with immature forms of pro-Kitl1, whereas overexpression of the catalytically inactive ADAM19E>A mutant did not. These findings were inconsistent with a model in which ADAM19 releases an inhibitor of Kitl1 shedding, as this should have resulted in increased levels of mature pro-Kitl1 in cells. Finally, when we performed loss-of-function studies in mEFs lacking candidate sheddases of the ADAM protease family, we found that ADAM17 is the main enzyme responsible for phorbol-ester- and phosphatase-inhibitor-stimulated release of Kitl1.
In the light of these results, an alternative hypothesis for how overexpression of ADAM19 might decrease ADAM17-dependent PMA-stimulated shedding of Kitl1 is that ADAM19 cleaves Kitl1 intracellularly, resulting in the intracellular degradation of the ectodomain instead of its release from the cell. It can also not be ruled out that overexpression of ADAM19 leads to a slight increase in constitutive Kitl1, but such an increase might not be evident in comparison with normal amounts of constitutive Kitl1 shedding. Moreover, it is conceivable that ADAM19 somehow prevents maturation of Kitl1, although this would have to depend on both its catalytic activity and on the presence of the ADAM19 cleavage site, since inactive ADAM19 does not affect Kitl1 maturation, and active ADAM19 does not reduce shedding of Kitl2, which differs from Kitl1 only by the absence of the ADAM19 cleavage site in exon 6. Nevertheless, both possibilities are consistent with the results of a pulse-chase analysis, in which overexpressed ADAM19 led to lower levels of mature Kitl1 without affecting biosynthesis of the immature precursor. The consequence of overexpressing ADAM19 is thus a reduction of the amount of mature, Endo-H-resistant Kitl1 that is available for ADAM17-dependent PMA-stimulated shedding. Since ADAM19 can release other substrates into the supernatant when it is overexpressed, including TNFα and TNFα-related activation-induced cytokine (TRANCE), also referred to as osteoprotegerin ligand (OPGL) (Chesneau et al., 2003; Zhou et al., 2004), it will now be interesting to determine why Kitl1 shedding is not increased by overexpressed ADAM19. Perhaps overexpressed ADAM19 processes Kitl1 in a different subcellular location within the secretory pathway compared with other ADAM19 substrates. Moreover, it cannot be ruled out that ADAM19 activates another enzyme that is responsible for the intracellular degradation of Kitl1. As there is an increased release of Kitl1 from Adam19–/– cells compared with wild-type controls, endogenous levels of ADAM19 are evidently also able to influence the regulation of Kitl1 shedding from mEFs (Chesneau et al., 2003).
Although the predicted cleavage site for ADAM17 in Kitl1 is lacking in Kitl2 (Table 1), we found that ADAM17 is required for PMA- and PV-dependent release of Kitl2 from mEFs. ADAM17 must therefore use a different cleavage site in Kitl2 than the one used by purified ADAM17 in the Kitl1 cleavage site peptide in vitro. To further define the cleavage site for ADAM17 in Kitl2, we tested how mutants in the juxtamembrane domain of Kitl2 impinge on its ectodomain shedding in cell-based assays. Although Majumdar et al. had previously reported that Kitl1 lacking the sequence KAAK178-181 together with the mutation in the major proteolytic cleavage site (VLES163-166 instead of VAAS163-166) remains membrane associated without producing soluble Kitl (Majumdar et al., 1994), the KAAK150-153 deletion in Kitl2, in which the major proteolytic cleavage site is lacking and therefore expected to have an effect as previously shown, was not enough to abolish PMA- or PV-stimulated Kitl shedding completely in our study. Only a Kitl2 mutant lacking both the KAAK150-153 and SSTL141-144 sequences as well as a mutant lacking 13 amino acid residues (G149-Q161) did not produce detectable amounts of soluble Kitl. The deletion of 13 amino acid residues in KL2DM might reduce the distance between the cleavage site and the plasma membrane to a point where ADAM17 can no longer gain access to the site. However, this is probably not the case with the KL2SSTL/KAAK mutation, in which only eight amino acid residues were deleted, since two mutants in which nine amino acid residues were deleted are shed efficiently (KL2SSTLGPEKG and KL2GPEKGKAAK). One possible explanation for the lack of processing of KL2SSTL/KAAK is that ADAM17 has an extended recognition sequence in Kitl2 for processing that includes these two sequences, but requires only one to be present. This notion is consistent with the crystal structure of ADAM17, which shows an extended peptide-binding groove in the catalytic site (Maskos et al., 1998). Finally, it is possible that ADAM17 uses binding sites in the ectodomain of Kitl as a determinant for its recognition, and then proceeds to cleave the substrate in a membrane proximal position without a strong cleavage site preference, but nevertheless requires either SSTL141-144 or KAAK150-153 to be present in Kitl2.
These studies thus define two mutants of Kitl that are produced at levels comparable to wild-type Kitl in cells and do not affect transport to the cell surface, but nevertheless abolish Kitl shedding. This information might be useful for generating knock-in mutations in mice that abolish shedding of Kitl in order to dissect the roles of soluble and membrane-associated Kitl in mouse development and disease models. Similar knock-in mutations have corroborated the essential role for soluble HB-EGF in heart valve development (Yamazaki et al., 2003), and helped dissect the contribution of membrane-anchored versus soluble TNFα in the immune response (Ruuls et al., 2001).
In summary, these results provide the first evidence that one ADAM can compete with another ADAM for substrate processing by reducing the amount of available substrate. Moreover, we identified ADAM17 as the major PMA- and PV-stimulated sheddase for Kitl1 and Kitl2 in mEF cells. These results have implications for the functional regulation of Kitl1 and Kitl2 by ectodomain shedding, and might be particularly relevant for Kitl2 shedding, as ADAM17 is its only currently known sheddase. Since in vivo analysis of the role of ADAM17 was restricted because of the perinatal lethality of Adam17–/– mice (Peschon et al., 1998) (no defects in hematopoiesis were found in newborn mice, data not shown), future studies with conditional Adam17–/– mice as well as with mice carrying a knockin inactivating mutation of the Kitl-cleavage sites or deletion of ADAM17 activity in Kitl2 knockin mice should help to address the physiological role of ADAM17 and Kitl shedding in mouse development and in disease models.
Materials and Methods
Reagents and antibodies
All chemicals and reagents were purchased from Sigma unless indicated otherwise. Concanavalin A (ConA)-Sepharose and GammaBind-G Sepharose were purchased from Amersham Biosciences. Endo H and PNGase F were purchased from New England BioLabs. The hydroxamate-based metalloprotease inhibitor Batimastat (BB94) was kindly provided by J. David Becherer (GlaxoSmithKline, Research Triangle Park, NC). The anti-ADAM17 cytoplasmic domain antibody (Schlöndorff et al., 2000) and anti-ADAM19 antibody (Chesneau et al., 2003) have been previously described. The anti-Kitl cytoplasmic domain antibody was obtained by immunizing rabbits with a C-terminal Kitl peptide (YMLQQKEREFQEV) and used after affinity purification (Manova et al., 1992).
The cDNA encoding AP-tagged Kitl2 was constructed by inserting full-length mouse Kitl2 cDNA into the XhoI and XbaI sites of pAPtag5 vector (GenHunter Corp.) in frame, as described previously for the AP-Kitl1 construct (Chesneau et al., 2003). In both cases, an AP-tag was attached to the N-terminus of the full-length wild-type protein, and a 6xHis-Myc tag was attached to the C-terminus. In order to prepare deletion mutants, site-directed mutagenesis was performed with either AP-Kitl1 or AP-Kitl2 as a template using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). All constructs were sequenced to confirm that the desired mutations or deletions had been obtained, and to rule out other mutations.
Cell culture, transfection and shedding assays
mEFs were prepared from E13.5-day embryos isolated from wild-type or ADAM-deficient mice [Adam8–/– (Kelly et al., 2005), Adam9/12/15–/– (Sahin et al., 2004), Adam17–/– (Peschon et al., 1998) and Adam19–/– (Zhou et al., 2004)] as described previously (Weskamp et al., 2002). Adam10+/– heterozygote or Adam10–/– knockout immortalized fibroblast cell lines derived from E9.5 embryos have been described previously (Hartmann et al., 2002). Both mEFs and COS-7 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, 100 IU/ml penicillin G and 100 μg/ml streptomycin. CHO cells were grown in F-12 media supplemented with 5% FCS, 2 mM L-glutamine, 100 IU/ml penicillin G and 100 μg/ml streptomycin. Cells were transfected with plasmids encoding the various Kitl1 or Kitl2 constructs alone or together with mouse wild-type ADAM17, mouse wild-type ADAM19 or the catalytically inactive E>A mutant ADAM19 using Lipofectamine 2000 (Invitrogen Life Technologies). The expression of ADAM17 and ADAM19 were confirmed by western blotting as described previously (Chesneau et al., 2003; Schlöndorff et al., 2000). The AP-tagged Kitl1 or Kitl2 ectodomains released in the cell supernatants were collected and analysed in an 8% SDS-PAGE, either directly or after concentration using ConA-Sepharose lectin beads (Amersham Biosciences) followed by elution with 0.5 M α-methyl-D-mannoside in 50 mM Tris-HCl, pH 7.4 for 2 hours at 37°C (Sahin et al., 2006). The shedding of the AP-tagged ectodomain into the cell supernatant was assessed using an in-gel AP-assay as previously described for other substrates (Chesneau et al., 2003; Sahin et al., 2004; Zheng et al., 2002). Ectodomain shedding was stimulated for 1 hour with 25 ng/ml PMA or with 100 μM of the phosphatase inhibitor pervanadate (PV). Cell lysates were prepared in phosphate buffered saline (PBS), 1% Triton X-100, supplemented with a protease inhibitor cocktail (Roche). The lysates were incubated on ice for 30 minutes, and then nuclei and cell debris were removed by centrifugation at 16,000 g for 15 minutes in an Eppendorf tabletop centrifuge. 2× sample loading buffer was added prior to electrophoresis and the samples were heated to 95°C in preparation for western-blot analysis, or not heated in cases in which the AP-tag was detected after renaturation for an in-gel assay of AP activity (Sahin et al., 2006).
Pulse-chase, immunoprecipitation and enzymatic deglycosylation
Transfected COS-7 cells were used for pulse-chase experiments after overnight recovery following transfection. Cells were washed twice with PBS and then incubated for 1 hour with DMEM lacking cysteine and methionine that was supplemented with 10% dialyzed FCS. 200 μCi/ml of [35S]Pro-mix (Amersham Bioscience) was then added for a 30-minute pulse-labeling period. After pulse labeling, the cells were washed twice with PBS, then chased in Opti-MEM (Invitrogen) for the indicated amounts of time, and supernatants were collected. The cells were washed twice and immediately lysed in lysis buffer [1% Triton X-100, 20 mM Tris (pH 7.5), 150 mM NaCl, 20 mM EDTA, 10% glycerol supplemented with protease inhibitor cocktail (Sigma) and 1 mM phenylmethanesulfonyl fluoride]. Cell debris was removed by centrifugation at 16,000 g for 15 minutes. The supernatants were incubated with anti-Kitl antibodies conjugated to GammaBind-G Sepharose beads (Amersham BioScience) for 2 hours. The beads were then washed once in buffer B (50 mM Tris, 500 mM NaCl, 5 mM EDTA, 0.2% Triton X-100), three times in buffer C (50 mM Tris, 150 mM NaCl, 0.1% Triton X-100, 0.1% SDS, 5 mM EDTA) and once in buffer D (10 mM Tris, 0.1% Triton X-100). Bound material was released by incubating the beads in 5× sample buffer for 5 minutes at 95°C, separated on an 8% SDS-PAGE and visualized by autoradiography. For deglycosylation experiments, pulse-labeled cells were chased for 6 hours and then lysed and immunoprecipitated as described above. The immunoprecipitates were denatured in 60 μl of glycoprotein denaturing buffer containing 5% SDS and 10% β-mercaptoethanol by boiling for 10 minutes and divided into three tubes. For PNGase F treatment, the incubation was performed at a final concentration of 1% Nonidet P-40. The radiolabeled samples were then either mock treated or incubated with 500 units of Endo H or 500 units of PNGase F (New England BioLabs) at 37°C overnight, following the manufacturer instructions.
We thank Jacques Peschon and Roy Black (Amgen Inc., Seattle, WA) for providing Adam17–/– mice, Andy J. P. Docherty (Celltech R&D, Slough, UK) for Adam 8–/– mice, Dieter Hartmann (Leuven and Flanders Interuniversity Institute for Biotechnology, Leuven, Belgium) and Paul Saftig (Christian-Albrechts University, Kiel, Germany) for immortalized Adam10–/– mEFs, and Yasemin Yozgat for excellent technical assistance. This study was supported by NIH-R01 GM64750 (to C.P.B.) and NIH-HL/DK55748 (to P.B.).