Cranial neural crest (CNC) cells are a transient population of stem cells that originate at the border of the neural plate and the epidermis, and migrate ventrally to contribute to most of the facial structures including bones, cartilage, muscles and ganglia. ADAM13 is a cell surface metalloprotease that is essential for CNC cell migration. Here, we show in Xenopus laevis embryos that the Wnt receptor Fz4 binds to the cysteine-rich domain of ADAM13 and negatively regulates its proteolytic activity in vivo. Gain of Fz4 function inhibits CNC cell migration and can be rescued by gain of ADAM13 function. Loss of Fz4 function also inhibits CNC cell migration and induces a reduction of mature ADAM13, together with an increase in the ADAM13 cytoplasmic fragment that is known to translocate into the nucleus to regulate gene expression. We propose that Fz4 associates with ADAM13 during its transport to the plasma membrane to regulate its proteolytic activity.
The cranial neural crest (CNC) is a stem cell population that gives rise to the craniofacial structures. Normal craniofacial development depends on the correct induction, migration and differentiation of these cells. Wnt signaling pathways play crucial roles at multiple levels during CNC development. Whereas the canonical Wnt/β-catenin pathway is known to mediate CNC induction (Stuhlmiller and García-Castro, 2012), in Xenopus the non-canonical Wnt/PCP (planar cell polarity) pathway is implicated in CNC cell migration (Mayor and Theveneau, 2014). In contrast, knockout of the membrane protein Vangl2, one of the PCP components, does not affect neural crest migration in mouse, suggesting that Wnt/PCP signaling is not crucial for neural crest cell migration in mammals (Pryor et al., 2014). The canonical and non-canonical Wnt pathways are generally regulated by specific combinations of Wnt ligands and their Frizzled (Fz) receptors, together with specific co-receptors and secreted modulators that compete for Wnt ligand binding (Niehrs, 2012). This, ultimately, leads to activation or inhibition of either β-catenin-mediated transcription (canonical) or PCP-signaling-mediated remodelling of the actin cytoskeleton (non-canonical).
Frizzled-4 (Fz4) is a Wnt receptor that is essential for CNC cell migration in Xenopus laevis (Gorny et al., 2013). Two variants encoding a secreted and a transmembrane form have been reported. The secreted variant, Fz4-v1 (previously called Fz4S), is generated by intron retention and possesses the cysteine-rich domain (CRD) of the receptor, but not the intracellular-signaling domain and the seven transmembrane domains (Sagara et al., 2001; Swain et al., 2005). Fz4-v1 has strong similarity to secreted frizzled-related proteins (SFRPs), which are well-known Wnt modulators that also contain a Fz-like CRD (Kawano and Kypta, 2003). These proteins can have both positive and negative regulatory effects on the Wnt pathway, depending on their concentration. At low concentration SFRPs can potentiate Wnt signaling, whereas at high concentration they are thought to sequester the Wnt ligand and inhibit pathway activation (Mii and Taira, 2009; Uren et al., 2000). Both secreted Fz4-v1 in human and Xenopus have been shown to activate Wnt/β-catenin signaling under specific conditions (Gorny et al., 2013; Sagara et al., 2001; Swain et al., 2005). Furthermore, it has been reported, that the Xenopus Fz4-v1 has, like the SFRPs, also a biphasic activity. In gain-of-function experiments, Fz4-v1 synergistically activates Wnt/β-catenin with low levels of Wnt ligands; but it inhibits the pathway when expressed at high levels (Gorny et al., 2013). In addition, Fz4-v1 can enhance activation of the PCP pathway when expressed with non-canonical Wnts such as Wnt11 (Gorny et al., 2013). Morpholino knockdown of Fz4 and Fz4-v1 inhibits CNC cell migration, and this can be rescued by re-expression of Fz4-v1 alone. However, whether the function of Fz4-v1 in the neural crest involves the Wnt pathway has not yet been shown.
In mice, SFRP1 and SFRP2 (hereafter referred to as SFRP1/2) have a Wnt-independent function in the retina, where they bind and inhibit ADAM10 (Esteve et al., 2011). ADAM (a disintegrin and metalloprotease) proteins are transmembrane proteases that regulate cell adhesion, as well as multiple signaling pathways that require proteolysis of ligands or receptors to initiate activation or termination of signaling. ADAM10 is the primary activator of the Notch pathway by performing the initial cleavage of the Notch receptor (Brou et al., 2000; Pan and Rubin, 1997). In the mouse retina, the absence of SFRP1/2 leads to a defect in retinal neurogenesis due to hyper activation of ADAM10, thereby increasing Notch signaling (Esteve et al., 2011). In Xenopus, at least three ADAMs, ADAM9, ADAM13 and ADAM19, are involved in the induction and migration of CNC cells (Alfandari et al., 1997; Cousin et al., 2011; Cousin et al., 2012; McCusker et al., 2009; Neuner et al., 2009). ADAM13 stimulates migration by both cleaving the adhesion molecule Cadherin-11 at the cell surface and regulating gene expression through its cytoplasmic domain, which is cleaved by γ-secretase and then translocates to the nucleus. Given the similarities between SFRPs and the secreted Fz4 variant, we investigated whether Fz4-v1 modulates the activity of ADAM13 in the CNC.
Here, we show that both Fz4 and Fz4-v1 bind to the cysteine-rich domain of ADAM13, and decrease its proteolytic activity in a cell autonomous manner. In vitro, overexpression of Fz4 and Fz4-v1 decrease proteolysis of cadherin-11 and the paraxial protocadherin PAPC by ADAM13. In vivo, overexpression of Fz4 or Fz4-v1 in the CNC inhibits cell migration, which can be rescued by co-expressing ADAM13 or the cadherin-11 extracellular cleavage fragment EC1-3. We show that knockdown of Fz4 in the CNC results in a decrease of the mature form of ADAM13 and an increase of the cytoplasmic fragment, suggesting that Fz4 is important for the stability of ADAM13 at the cell surface and may control the timing of the cytoplasmic domain processing. These gain- and loss-of-function experiments indicate that both the full-length receptor and the secreted variant interact with ADAM13 to regulate the level of protease activity and the processing of the cytoplasmic domain fragment.
ADAM13 cysteine-rich domain binds to Fz4
SFRP1/2 have been shown to interact with and inhibit ADAM10 proteolytic activity (Esteve et al., 2011). Therefore, we tested whether such an interaction existed between Fz4 and ADAM13. Indeed, ADAM13 and Fz4-v1, as well as the Fz4 receptor, co-precipitate when transfected into HEK293T cells (Fig. 1A). In addition to the disintegrin and metalloprotease domains, ADAMs also contain a cysteine-rich domain, an EGF-like domain, and a transmembrane and cytoplasmic domain (see Fig. 1B). To identify which of these domains is responsible for binding to Fz4, we tested variants of ADAM13 that express either individual or combinations of domains. We used Fz4-v1 in these experiments because both proteins seem to have a similar binding, and the potential surface contact in Fz4-v1 is smaller. In these experiments, we found that the cysteine-rich domain of ADAM13 is sufficient to promote Fz4-v1 interaction, whereas the disintegrin domain is not (Fig. 1B). Interestingly, this domain of ADAM13 is also responsible for interacting with the second heparin-binding domain of fibronectin (Gaultier et al., 2002) and is thought to control proteolytic specificity (Smith et al., 2002).
Whereas the functional analyses – showing that Fz4/Fz4-v1 combined knockdown inhibits CNC cell migration – strongly suggest that these proteins are present in the CNC (Gorny et al., 2013), the published expression pattern reported for Fz4 and Fz4-v1 (Shi and Boucaut, 2000) do not closely match neural crest cells, suggesting that Fz4 and Fz4-v1 could act on the CNC from adjacent tissues (for example, the placodes). To confirm that mRNA of either Fz4 or Fz4-v1 is present, we performed quantitative real-time PCR with dissected CNC explants using primers designed to selectively amplify each form. The result shows that both Fz4 and Fz4-v1 are, indeed, expressed in the CNC, but their mRNAs are not restricted to these cells, confirming the in situ hybridization (ISH) pattern. When compared with the Zinc finger protein SNAI2 (also known as and hereafter referred to as Slug), a typical CNC marker, the overall expression levels of Fz4 and Fz4-v1 in whole embryos at stage 17 are very similar and represent between 2% (Fz4-v1) and 4% (Slug, Fz4) of GAPDH expression. At the same stage, Slug is expressed in the CNC at 57% of the GAPDH level, making it very obvious by ISH. However, Fz4 and Fz4-v1 are expressed at 5% and 2% of GAPDH, respectively, a level similar to the global expression in the embryo. We also performed double ISH using a Fz4-receptor-specific full-length probe and a short locked nucleic acid (LNA)-probe specifically recognizing Fz4-v1 but not Fz4 (Gorny et al., 2013) to distinguish between the two Fz4 splice variants (Fig. 1D and supplementary material Fig. S1, blue) and the neural crest markers Sox10 and Twist (both red). With this technique it is clear that Fz4-v1 is present in the CNC at both stage 20 and stage 24, whereas Fz4 appears to be expressed in placodes as well as in the extreme tips of the CNC segments (Fig. 1D, arrowheads). Together these data show that ADAM13, Fz4 and Fz4-v1 are all expressed in the CNC, and that ADAM13 and Fz4-v1 expression patterns clearly overlap.
To determine whether ADAM13 colocalizes with either form of Fz4, we C-terminally tagged ADAM13 with GFP, and Fz4 and Fz4-v1 with RFP, and co-expressed these constructs in Xenopus CNC and cell lines (Fig. 2 and supplementary material Fig. S2). In the CNC, Fz4 is detected in two main compartments, vesicles and the plasma membrane associated with the fibronectin substrate (Fig. 2; supplementary material Movie 1). ADAM13 protein is detected in all membranes but enriched in vesicles as well as in cellular protrusions that are in contact with the fibronectin substrate. ADAM13 colocalizes with Fz4 in a subset of vesicles (Fig. 2 upper and middle rows, arrowheads) and at the membrane that is in contact with the substrate (Fig. 2 middle row, arrows). The localization of Fz4-v1 is clearly distinct, surrounding the nucleus in a pattern resembling the endoplasmic reticulum. In time-lapse movies at the level of the substrate, Fz4-v1 appears to be closely associated with ADAM13 in some membrane protrusions (Fig. 2 lower row, boxed areas, and supplementary material Movie 2). Thus, the subcellular localization of ADAM13, Fz4 and Fz4-v1 is compatible with a potential interaction of these proteins in migrating CNC cells.
In the Xenopus fibroblast cell line (XTC), we used sub-cellular markers for the lysosome (Lysotracker), endoplasmic reticulum (ER-tracker), Golgi complex (WGA) to identify the compartment in which ADAM13 colocalizes with Fz4 and Fz4-v1. Fz4 expression was found in both the endoplasmic reticulum and lysosomes (supplementary material Fig. S2), but was hardly detected in the Golgi complex (supplementary material Fig. S2). ADAM13 was associated with all three markers but appeared to only colocalize with Fz4 in the ER in these cells (supplementary material Fig. S2). The vesicles in which ADAM13 and Fz4 colocalized did not stain with any of the markers tested, suggesting that they were neither Golgi nor lysosomal in origin. The staining obtained for Fz4-v1 was much more obvious and appeared to be restricted to the ER where it perfectly colocalized with ADAM13 (supplementary material Fig. S2). Again, in WGA-positive vesicles ADAM13 appeared to be mostly free of Fz4-v1 suggesting that the proteins do not transit together to the cell surface in XTC cells (supplementary material Fig. S2). Unfortunately, these markers could not be used in a CNC explant, as they all bind very strongly to the yolk-rich platelets (data not shown). A similar distribution of ADAM13-GFP, Fz4-RFP and Fz4-v1-RFP was found in transfected human osteosarcoma (U2OS) cells, when using antibodies against calnexin and GM130 to identify the ER and Golgi respectively (data not shown).
Fz4 inhibits ADAM13 proteolytic activity and migration of CNC cells
Having established that all three mRNAs are expressed in the CNC and that, upon expression of fluorescent forms of the protein, they are expressed in overlapping compartments of the CNC, we next tested whether the interaction between Fz4 and ADAM13 affects the proteolytic activity of the latter. We tested the ability of Fz4 and Fz4-v1 to inhibit ADAM13 cleavage of the paraxial protocadherin (PAPC) (Fig. 3A, Shed PAPC) and found full-length Fz4 more efficient (∼100% inhibition) than secreted Fz4-v1 (∼50% inhibition). However, neither form effectively prevented self-shedding of ADAM13 (Fig. 3A, Shed A13). We also found that Fz4-v1 was able to inhibit ADAM13 cleavage of cadherin-11 in HEK-293T cells (supplementary material Fig. S3). We could not test Fz4 in this assay because the expression of both Fz4 and cadherin-11 drastically reduced the level of ADAM13 in transfected cells (data not shown), making the interpretation of the absence of cleavage impossible. To test whether Fz4 simply prevented the interaction of ADAM13 with PAPC, we performed co-immunoprecipitation of PAPC with ADAM13 in the presence or absence of Fz4. The result shows that presence of Fz4 or Fz4-v1 does not prevent binding of ADAM13 to PAPC, and suggests that Fz4 and Fz4-v1 inhibit ADAM13 activity using a different mechanism (Fig. 3B).
Because ADAM13 proteolytic activity is essential for CNC cell migration, we hypothesized that overexpression of Fz4 should interfere with this process if both proteins interact in vivo. Indeed, targeted overexpression of Fz4 or Fz4-v1 inhibited CNC cell migration (Fig. 4A,B). To test whether Fz4 overexpression acts by inhibiting ADAM13 activity we attempted to rescue CNC cell migration by overexpressing either ADAM13 or the extracellular cleavage fragment of cadherin-11 (EC1-3). Interestingly, whereas ADAM13 and EC1-3 were capable of partially rescuing both forms of Fz4 overexpressed, the rescue of full-length Fz4 was much more efficient in our assay than that of the secreted Fz4-v1 (Fig. 4A,B).
To test whether secreted Fz4-v1 inhibits migration in a cell autonomous manner we grafted CNC expressing Fz4-v1 into wild type embryos, as well as wild type CNC into Fz4-v1-expressing embryos (Fig. 4C). Our results show that CNC cell migration is only inhibited when Fz4-v1 is expressed by the CNC. Similarly, we found that mixing cells that express Fz4-v1 with those that express ADAM13 and PAPC did not inhibit PAPC shedding, suggesting that the secreted Fz4-v1 is unable to inhibit ADAM13 proteolytic activity on adjacent cells (Fig. 4D). This result, taken together with the data showing colocalization of ADAM13 with Fz4 and Fz4-v1 in mostly internal compartments (Fig. 2 and supplementary material Fig. S2), and the co-immunoprecipitation with the pro-form of ADAM13 (Fig. 1B), suggests that the functional interaction is occurring during the transit of ADAM13 to the cell surface. Another possibility is that PAPC is already cleaved before both proteins (ADAM13 and PAPC) reach the surface.
Loss of Fz4 inhibits CNC cell migration and increases ADAM13 processing
Our results suggest that both forms of Fz4 can interact with ADAM13 and inhibit its proteolytic activity. To test whether this is important in vivo, we performed loss-of-function experiments by knocking down Fz4 (and Fz4-v1). As previously reported (Gorny et al., 2013), Fz4 morpholino (MOFz4) inhibits CNC cell migration both in grafting experiments (Fig. 5A,B, and supplementary material Movie 3) and targeted injection (Fig. 5C). In targeted injections we found that we could not rescue CNC cell migration by overexpressing ADAM13 in embryos lacking Fz4 (Fig. 5C). In embryos, decrease of Fz4 or Fz4-v1 (the morpholino does not distinguish between the two) results in a decrease of the mature form of ADAM13 and an increase in the ADAM13 cytoplasmic domain during CNC cell migration (Fig. 6A, stage 24). We have previously shown that the cleavage of the cytoplasmic domain by γ-secretase is preceded by an initial self-cleavage of ADAM13 within its cysteine-rich domain (Cousin et al., 2011). Our results here suggest that, in the absence of Fz4 or Fz4-v1, ADAM13 self-processing is increased, resulting in the decrease in the surface form of ADAM13 and a corresponding increase in the cytoplasmic fragment. In gain-of-function experiments, when Fz4 or Fz4-v1 are overexpressed in the embryo, we observed a reduction in the cleaved cytoplasmic fragment of ADAM13 at stage 19 (Fig. 6B). This reduction persists at stage 30 only for the overexpression of Fz4-v1. In the case of Fz4, decrease of the mature ADAM13 is observed together with that of the cytoplasmic fragment; however, this was not the case when Fz4-v1 was overexpressed.
To confirm the role of endogenous Fz4 in ADAM13 longevity or stability, we injected ADAM13-GFP together with membrane-bound mCherry (mb-Cherry) in embryos, and visualized the GFP and mCherry signals in dissected CNC explants (Fig. 6C). Whereas mb-Cherry was unaffected by Fz4 knockdown, we found that ADAM13-GFP was drastically reduced following Fz4 knockdown. Given the finite amount of mRNA encoding ADAM13-GFP, it is understandable that the results are more obvious than those observed for endogenous ADAM13, whose mRNA is constantly produced.
These results suggest that expression of ADAM13 should alleviate part of the Fz4-knockdown phenotype by, at least, providing a larger pool of ADAM13 protein during the onset of CNC cell migration. To test this hypothesis in more detail, we used in situ hybridization with CNC markers (Sox10 and Twist) to visualize the position of the CNC towards the end of migration (Fig. 7). As seen for both graft and targeted injections we found that the loss of Fz4 significantly affected the position of the CNC (Fig. 5). Whereas targeted injections were scored by counting the number of embryos that had no visible fluorescent cells in the CNC pathway, the in situ hybridization was scored for the relative lengths of segments (inhibited weak or strong), as well as the position and the separation of each segment (fused). Using this latter assay, we found that ADAM13 overexpression was, indeed, capable of reducing the severity of the CNC defects in embryos lacking Fz4 (Fig. 7). Accordingly, EC1-3 – the recombinant cleavage fragment of cadherin-11 by ADAM13 – was also capable of restoring the quality of CNC cell migration. This was true when the injections were targeted to one dorsal blastomere, either left or right, at the 4-cell stage (Fig. 7A) or when targeted to one dorsal animal blastomere at the 8-cell stage (Fig. 7B). Together, these data support the model by which one of the roles of Fz4 or Fz4-v1 during CNC cell migration is to regulate the proteolytic activity of ADAM13.
Fz4 regulates ADAM13 maturation and stability
ADAM13 and Fz4 are both essential for CNC cell migration. ADAM13 controls CNC cell migration first by cleaving cadherin-11 in order to release an extracellular fragment, but also by sending its own cytoplasmic domain into the nucleus to regulate gene expression (Cousin et al., 2011; McCusker et al., 2009). Thus, there are at least two functional forms of ADAM13, the active protease at the cell surface (100 kDa) and the cytoplasmic fragment within the nucleus (17 kDa). Although we know that both forms are essential, it remains unclear how and when the cytoplasmic fragment is generated. We know that cleavage by γ-secretase releases the cytoplasmic domain from the membrane and that this follows an initial cleavage within the cysteine-rich domain by ADAM13 itself (Cousin et al., 2011). Loss of Fz4 appears to increase the rate or induce precocious processing of ADAM13, whereas overexpression reduces this processing (Fig. 6). In cell culture, transfected Fz4 associates preferentially with the pro-form of ADAM13, and appears to colocalize with ADAM13 in the ER, vesicles and at the plasma membrane (Figs 1,2; supplementary material Fig. S2). These observations suggest that Fz4 regulates ADAM13 processing. This regulation involves a physical interaction of ADAM13 with Fz4 through the cysteine-rich domain of ADAM13 during transit and at the cell surface. Although this domain is also involved in binding ADAM13 substrate (e.g. fibronectin) (Gaultier et al., 2002), binding of ADAM13 to Fz4 does not compete for binding to one of its substrates PAPC (Fig. 3B), suggesting that the proteolytic inhibition is not due to competitive binding with the substrate. Given that ADAM13 also binds to secreted Fz4 (Fz4-v1), this suggests that interaction of the two proteins involves the extracellular domain of Fz4 (common to both splicing variants). However, our results indicate that full-length Fz4 protein is more effective at inhibiting ADAM13 self-processing and proteolytic activity than secreted Fz4-v1. To which extent and how the cytoplasmic domain or the seven transmembrane domains of Fz proteins are involved in the regulation of ADAM processing and activity should be addressed in future studies. Recently, it has been shown that furin-mediated maturation of ADAM17 and its trafficking to the cell surface requires binding of another seven-pass transmembrane protein, iRhom2 (also known as Rhbdf2) (Adrain et al., 2012). In addition, tetraspanin18, another multi-spanning transmembrane protein, prevents cadherin-6B degradation prior to neural crest cell migration (Fairchild and Gammill, 2013), a degradation that is performed by two ADAM proteins (ADAM10 and ADAM19) expressed in the neural crest (Schiffmacher et al., 2014). Thus, it is possible that regulation by multiple transmembrane-containing proteins is a common feature of ADAM metalloproteases.
Regulation of ADAM13 is not the sole function of Fz4 in the CNC
Our results show that loss of Fz4 produces premature cleavage of the cytoplasmic domain of ADAM13, together with a reduction of the mature metalloprotease at the cell surface. In addition, loss of Fz4 inhibits CNC cell migration. On the one hand, our targeted injections suggest that ADAM13 cannot rescue CNC cell migration in embryos that lack Fz4. This suggests that Fz4 is involved in several other roles within the CNC that remain to be investigated. On the other hand, the in situ hybridization data (Fig. 7) suggest that the Fz4-knockdown phenotype can be partially rescued by ADAM13 or the EC1-3 fragment of cadherin-11, which mimics the cleavage by ADAM13. This apparent contradiction of results can be explained by the nature of each technique. In targeted injections and in grafts, CNC cells that have received the morpholinos are fluorescently labeled so that their migration can be directly monitored. A loss of cell migration can be easily observed even if this inhibition results in a loss of CNC markers. If non-targeted cells migrate from a different location, even from the contra-lateral side, they will not be visible in this assay, whereas they would be detected as migrating CNC cells in an in situ hybridization assay. For example, knockdown of ADAM13 results in a clear inhibition of CNC cell migration when investigated by using targeted injection but not by using graft or in situ hybridization (Abbruzzese et al., 2014). This is because the grafting procedure and, in particular, the separation of the CNC from the underlying mesoderm, releases the CNC cells and rescues the phenotype (Cousin et al., 2012). However, the use of in situ hybridization allows comparison of the non-injected with the injected side of the same embryo, sometimes allowing to measure more subtle effects of CNC positioning. They also provide a single snap shot at a specific time by showing a position of the CNC that may be perturbed but will later recover.
Given that none of the used techniques yields a complete rescue of the phenotype, it is clear that ADAM13 is not the only target affected by knockdown of Fz4. As an example, other ADAMs, such as ADAM9 and ADAM19 and their respective targets, could also be affected by Fz4 KD and, therefore, cause part of the phenotype that is not rescued by ADAM13/EC1-3. It should be noted that Fz4-v1 can also inhibit cleavage of cadherin-11 by ADAM9 in vitro (our data not shown). Therefore, more detailed analyses of the phenotypes will be required to understand the complete molecular pathway that depends on Fz4 during CNC cell migration.
Interaction of ADAM13 and Fz4 in the CNC – a model
Our data support a model in which ADAM13 is associated with Fz4 during the initial stage of CNC cell migration to maintain the proper level of ADAM13 activity at the cell surface. At this time it could also control which substrates may interact with ADAM13. Clearly, full-length cadherin-11 is required (Kashef et al., 2009) and only about 20% of cadherin-11 needs to be cleaved by ADAM13 to generate the EC1-3 fragment (McCusker et al., 2009). In addition, proteolysis of ADAM13 generates a cytoplasmic fragment that has a specific function in the nucleus to regulate the expression of several genes (Cousin et al., 2011). Again, Fz4 appears to control the level of this fragment during migration. Interestingly, the two forms of Fz4 do not behave the same way with regard to ADAM13. Fz4 is most efficient at inhibiting ADAM13 activity but also reduces the ADAM13 protein level (Figs 3 and 6). This may explain why this form of overexpression is rescued most efficiently following the expression of additional ADAM13 or the cadherin-11 EC1-3 fragment. However, Fz4-v1 overexpression has little effect on mature ADAM13 levels but does substantially reduce the processing of the ADAM13 cytoplasmic domain (Fig. 6B). The results obtained for Fz4 knockdown, showing an increase in the fragment of the cytoplasmic domain of ADAM13, suggest that the main form associated with ADAM13 in the CNC is the secreted variant Fz4-v1, at least at stage 24. This would also be consistent with the Fz4-v1 expression pattern and the fact that Fz4-v1 can rescue CNC cell migration in Fz4 morphant embryos. The expression pattern of Fz4 shows that it increases at the edges of the migration path towards the end of their progression (Fig. 1D; supplementary material Fig. S1). This increase could block ADAM13 proteolytic activity so that cadherins and other cell-adhesion molecules can accumulate in order to promote the integration of the CNC into the target tissue. This process might be coordinated by expression of Fz4, which would directly inhibit the activity of ADAM13, reduce ADAM13 protein levels and, also indirectly, shut down ADAM13 by decreasing the activity of GSK3, a kinase required for the nuclear function of ADAM13 (Abbruzzese et al., 2014).
Is ADAM13 involved in the Wnt signaling pathway?
During gastrulation, PAPC directly interacts with the extracellular domain of another Wnt receptor Fz7. This interaction stabilizes PAPC and is increased by Wnt11 (Kraft et al., 2012). In the absence of Wnt11 or Fz7, PAPC is internalized and the protein decreases. Although we have not tested whether PAPC interacts with Fz4, Fz7 does bind to ADAM13 as well (data not shown). It is, therefore, possible that the interaction of Fz with ADAM13 is important in order to protect PAPC from degradation during internalization. It should be noted that Wnt11, Fz4, Fz7, PAPC and ADAM13 are all expressed in the CNC cells during migration (Alfandari et al., 1997; De Calisto et al., 2005; Matthews et al., 2008; Schneider et al., 2014; Swain et al., 2005). Furthermore, ADAM13 is also regulated by one of the kinases GSK3 that control the Wnt pathway. Whereas GSK3 phosphorylates β-catenin to inhibit its activity, phosphorylation of ADAM13 through GSK3 is required for its correct function during CNC cell migration (Abbruzzese et al., 2014). Thus, in theory, the Wnt pathway, by decreasing GSK3 activity and by interaction of the Fz receptors with ADAM13, could inhibit ADAM13 activity. It is unclear at the moment whether Fz4 binding to a Wnt ligand increases, decreases or has no effect on its ability to inhibit ADAM13. In addition, it is also unclear whether binding of ADAM13 to Fz4 modifies its ability to respond to Wnt and to signal. In Xenopus tropicalis, loss of ADAM13 results in a severe decrease in Wnt signaling during CNC cell induction (Wei et al., 2010). This is attributed to the loss of cleavage of ephrinB, resulting in an abnormal accumulation and increased signaling that, in turn, decreases canonical Wnt signaling. In Xenopus laevis, although knockdown of ADAM13 also increases the level of ephrinB protein, it has no effect on Wnt signaling. In contrast, knockdown of the two paralogues ADAM13 and MDC13 does result in a substantial decrease in β-catenin levels during gastrulation (data not shown). In this model, however, it is unclear whether this change is due to cleavage of ephrinB or a more direct effect on the Wnt pathway. Our new results show a direct interaction of ADAM13 with Fz receptors, suggesting a more direct role of ADAM13 in regulating the Wnt pathway, a role that will require further investigation.
MATERIALS AND METHODS
The following antibodies were used: g821, goat polyclonal antibody against the ADAM13 cytoplasmic domain affinity purified using ADAM13 peptide corresponding to amino acids 821–914 (Cousin et al., 2011); g877, goat polyclonal antibody against the ADAM13 cytoplasmic domain affinity purified using ADAM13 peptide corresponding to amino acids 877–914; 6615F, rabbit polyclonal antibody against the ADAM13 cytoplasmic domain (Alfandari et al., 1997); 4A7, mouse monoclonal against the ADAM13 cytoplasmic domain; 7C9, mouse monoclonal against the ADAM13 cysteine-rich domain (Gaultier et al., 2002); PAPC, mouse monoclonal against the extracellular domain (Chen and Gumbiner, 2006); 8C8, monoclonal antibody against β1 integrin (Gawantka et al., 1992); 9E10, mouse anti-Myc monoclonal (ATCC); α-his, anti-His mouse monoclonal (GE Healthcare). Rabbit polyclonal antibodies against Fz4-v1 were obtained by immunizing two rabbits with a peptide (STNAQLTRRPYSYA conjugated to LPH) specifically recognizing the v1 transcript. The antibodies were affinity purified with the peptide before use.
Morpholinos and DNA constructs
Morpholino antisense oligonucleotides (Gene Tools) against Fz4/Fz4-v1 and ADAM13 have been described previously (Cousin et al., 2011; Gorny et al., 2013). pCS2-Fz4/Fz4-v1 with and without Myc-tag were linearized with NotI (Fermentas) and in vitro transcribed using Sp6 RNA Polymerase (Ambion) (Gorny et al., 2013). Fz4-RFP was made by cloning RFP-Flag-tag downstream of Fz4 between the EcoRI and XbaI sites previously used to introduce the Myc-tag. ADAM13 and cadherin-11 constructs have been previously described (Alfandari et al., 2001; Gaultier et al., 2002; McCusker et al., 2009). ADAM13-GFP was made by replacing the Myc-tag of ADAM13-MT (Alfandari et al., 2001) with GFP from the pCS2mt-UGP cut at Cla1 and Not1 (generous gift from Mike Klymkowsky, University of Colorado, Boulder, CO).
Embryo injections and manipulations
All experiments complied with local and international guidelines for the use of experimental animals. Xenopus laevis embryos were obtained, and cultured as described previously (Gorny et al., 2013). For Fz4 loss-of-function grafts, 250 pg of GFP mRNA was injected alone or together with 18.75 ng of MOFz4 into one cell at the 2-cell stage. For gain-of-function grafts, 334 ng of GFP mRNA was injected alone of together with 666 ng of Fz4-v1 mRNA into one cell at the 2-cell stage. Fluorescently labeled CNC cells were grafted into non-injected host embryos at stage 15, prior to migration. For targeted injections, embryos were injected into a single CNC precursor cell (dorsal-animal blastomere) at the 8-cell stage. For gain-of-function, 200 pg RFP was injected alone or with 100 pg Fz4 or 300 pg of Fz4-v1 mRNA, plus 300 pg ADAM13 or EC1-3 mRNA. In the loss-of-function targeted injections, 200 pg RFP mRNA, 10 ng MOFz4, and 300 pg ADAM13 mRNA were used. All embryos were raised at 15°C until tail-bud stage (stage 24–27) when they were scored for inhibition of CNC cell migration by the absence of GFP- or RFP-positive cells within the migration pathways. For in situ hybridizations, embryos were injected unilaterally in one dorsal blastomere at the 4-cell or 8-cell stage with 25 ng of MOFz4, and 250 pg ADAM13 or EC1-3 mRNA. When the embryos reached stage 30, they were fixed in MEMFA (100 mM MOPS, 2 mM EGTA, 1 mM MgSO4, 3.7% formaldehyde) and whole-mount in situ hybridization was performed as described (Gorny et al., 2013). To examine ADAM13 protein levels in Fz4-knockdown or -overexpressing embryos, 25 ng MOFz4 or 10 ng MO13 was injected at the 1-cell stage, or 300 ng RFP, Fz4 or Fz4-v1 mRNA was injected into both dorsal cells at the 4-cell stage. Embryos were raised at 15°C and frozen at various stages for later processing. To observe ADAM13-GFP fluorescence in CNC explants, embryos were injected at the 2-cell stage with 200 pg ADAM13-GFP and 200 pg membrane-bound mCherry (mbCherry) mRNA, and 12.5 ng MOFz4.
Non-injected embryos were raised at 15°C until stage 17, at which point CNC cells were dissected from ten embryos. Total RNA was extracted from ten CNC or three sibling embryos at the same stage (High Pure RNA Tissue kit, Roche). PolyA mRNA was purified on oligo-dT cellulose (Qiagen) and cDNA was synthesized instruction using qScript cDNA kit (Quanta), both according to manufacturer's. Fz4 and Fz4-v1 primers were tested for specificity. Primer sequences are as follows (5′-3′): gapdh- forward TTAAGACTGCATCAGAGGGCCCAA and reverse GGGCAATTCCAGCATCAGCATCAA; slug- forward AAACACTTCAACACGACCAAAAA and reverse GCTGACCCGACCTAAAGATGAA; fz4- forward ACACGAGCTACAAGCAGATG and reverse GGACATACACTGAGCAGAGAAA; fz4-v1- forward CTCCAGCTCACCACCTTTAC and reverse AAGTTGCGCGTTGGTAGA.
Double in situ hybridization
For double in situ hybridization (double ISH) Digoxigenin (DIG)- and Fluorescein-labeled antisense RNA probes were synthesized the using the respective RNA polymerases and Labeling Mix from Roche according to manufacturer's instructions. Double ISH was carried out as described for normal ISH (Gorny et al., 2013) with the following modifications. Embryos were hybridized simultaneously by using the DIG- and the Fluorescein-labeled antisense probes. First, the DIG-probe was detected and stained in blue as described. When the desired staining was achieved, the reaction was stopped by washing in PBS, 0.1% Tween-20 and embryos were refixed for 1 h at room temperature (RT), followed by 1 h EDTA (5 mM in PBS) treatment at 60°C and six 10-min washes at RT in PBS, 0.1% Tween-20. Blocking and second antibody reaction were carried out as described for the DIG-probe but by using an anti-Fluorescein antibody (Roche). After washing, the Fluorescein-labeled probe was visualized using the FastRed tablets (Sigma) according to the manual. Pictures were taken with an Olympus SZX12 stereomicroscope (with U-RFL-T UV-light) using a CC12 camera and software analySIS (Soft Imaging Systems).
Cell culture and transfection
Co-immunoprecipitation experiments were performed in HEK293T obtained from ATCC and transfected with XtremeGENE HP (Roche) according to manufacturer's instructions. Total cellular protein was extracted in TBS containing 1% Triton X-100, 5 mM EDTA, 1× Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific) and immunoprecipitated for either ADAM13 or Fz4. PAPC shedding and cadherin-11 cleavage assays were performed in COS-7 and HEK293T cells (ATCC). For PAPC shedding, COS-7 cells were incubated with medium containing 2% serum 24 h after transfection, and conditioned supernatants were collected at 48 h. The cellular proteins were extracted in reducing Laemmli buffer and blotted for ADAM13, Fz4 or PAPC. The shed extracellular domains were purified using concanavalin-A–agarose (Vector) overnight, eluted in reducing Laemmli buffer and blotted using antibodies against PAPC or 7C9. To test the cleavage of PAPC in trans, HEK293T cells were transfected with ADAM13 and PAPC, or either RFP or Fz4-v1. At 24 h after transfection, cells were resuspended in medium containing 2% serum, and the ADAM13 plus PAPC cells were mixed with the cells expressing RFP or Fz4-v1. At 48 h the conditioned supernatants were collected and purified for shed PAPC or ADAM13 as described above.
CNC explants from embryos injected with the various mRNAs (ADAM13-GFP, Fz4-RFP, Fz4-v1-RFP, mbCherry) were dissected and placed on a glass-bottomed dish (Matek Corp) coated with gelatin (1 mg/ml) followed by fibronectin (20 µg/ml). Low magnification photographs (Fig. 5) were obtained using an Axiovert200M fluorescence microscope (Zeiss) with a 20× objective, 5 h after dissection of the explants. High-magnification photographs (Fig. 2) were obtained using a Nikon spinning disc confocal with a 60× objective. To identify subcellular compartments, Xenopus XTC cells were transfected with the various fluorescent constructs, plated on fibronectin-coated (10 µg/ml) glass-bottomed dishes (In vitro scientific) and incubated for 30 min in 1× MBS with ER-tracker (blue), Lyso-tracker (blue) or WGA (blue), all from Molecular probes. Cells were rinsed once with 1× MBS and once with complete XTC medium (67% L15, 10% FBS, 2 mM L-Glutamine, pen/strep, sodium pyruvate). Photographs were taken using a Zeiss 200M inverted microscope equipped with an Apotome and a 63× oil-immersion lens to obtain optical sections. For in vivo CNC cell migration assays, embryos were imaged using a Zeiss Stereo Lumar fluorescent stereoscope.
The authors dedicate this work in memory of Dr Herbert Steinbeisser.
↵* These authors contributed equally to this work
The authors declare no competing or financial interests.
All authors contributed to the design of the experiments. The original observation that Fz4 inhibited ADAM13 was made by A.K.G. and H.S.; G.A. and D.A. wrote the manuscript with input from all the authors. G.A., I.K., A.K.G., L.T.K., H.S. and D.A. performed ADAM13 and Fz4/Fz4-v1 interaction experiments. G.A. performed targeted injections and H.C. performed the grafts. A.K.G. and L.T.K. performed the in situ hybridizations. All authors read and approved the final manuscript.
This work was supported by grants from National Institutes of Health USPHS [grant numbers F31-DE023275 to G.A. and RO1-DE016289 to D.A.] and by a research grant of the Deutsche Forschungsgemeinschaft (DFG) (STE 613/8-2) to H.S. Deposited in PMC for release after 12 months.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.163063/-/DC1
- Received September 10, 2014.
- Accepted January 16, 2015.
- © 2015. Published by The Company of Biologists Ltd