Adam11.L is expressed in the central nervous system and the cranial neural crest (Cai et al., 1998). The mRNA for both S and L allo-alleles is expressed zygotically at low levels. Adam11.S is the major form expressed during gastrulation, while Adam11.L is the main form during neural tube closure and neural crest cell migration (Xenbase RRID: SCR_003280, st17 to 25) (Session et al., 2016; Fortriede et al., 2020). To test the role of Adam11, we designed two sets of morpholino to either block translation of both the Adam11.L and.S allo-alleles (MO11, L and S) or block the splicing of the first intron of Adam11.L which is the main form expressed within the cranial neural crest cells (MO11spl) (Figure 1A). The efficiency of these morpholinos was tested by quantitative capillary Western blot (Figures 1B, C) and RT-qPCR (Figure 1E), we also confirmed that MO11 blocks translation of 1 ng of mRNA encoding for Adam11.L (Figure 1G, lane2, A11.L) but not the translation of an mRNA lacking the 5′UTR corresponding to the morpholino binding site (Figure 1G, lane4, A11.L_R). We used time-lapse imaging to follow potential developmental defects after the onset of zygotic transcription (Stage 8). While we did not see any defect or delay during gastrulation, we observed that the neural tube closed earlier in morphant embryos compared to control embryos (Figures 1D, F). This observation was true for both splicing and translation-blocking morpholino and is a very rare phenotype as injections typically result in a slight delay of development but rarely an apparent acceleration (DA personal observation). We were able to rescue this phenotype by injecting Adam11.L mRNA lacking the 5′ untranslated region (Adam11.L_R), a form that is not affected by the morpholino (Figures 1D, F), confirming that the phenotype was indeed specific (Figures 1F, G). Given that both Morpholinos gave the same phenotypes, we focused the rest of the study on the translation-blocking morpholino that blocked both allo-alleles.

Because of Adam11.L expression in the cranial neural crest cells (CNC), we tested the ability of CNC with reduced Adam11 protein to migrate in vivo and in vitro (Figure 2). In vivo, we tested the migration capacity by targeted injection (Figures 2A, B) as well as grafting experiments (Figures 2D, E) using previously described assays (Cousin et al., 2011; Cousin et al., 2012; Abbruzzese et al., 2016). In both experiments, we found that the loss of Adam11 perturbed CNC migration significantly in approximately 50% of the embryos (Figures 2B, E). In the grafting experiments, we found significant defects in the CNC migration into the Branchial segments but not in the mandibular pathway. The inhibition of CNC migration was rescued by injection of the Adam11.L rescue mRNA in the CNC precursor cells (Figures 2A, B). As we have shown before, inhibition of CNC migration in vivo can be due to either a loss of cell motility (Alfandari et al., 2003) or a more complex effect of the cellular environment (3D ECM with multiple structural and signaling proteins) in vivo (Cousin et al., 2012). To test this possibility, we placed CNC explants in vitro on fibronectin-coated substrates and performed time-lapse video microscopy. Using this assay, we have previously shown that CNC explant dissected at stage 17 migrates collectively during the first 5 h before individual cells break away, a process also observed in vivo (Alfandari et al., 2001; Alfandari et al., 2003). We found that CNC cells in both control and Adam11 KD were able to spread and migrate on the substrate efficiently, but that cells lacking Adam11 migrated individually even during the first 5 h (Figure 2F). Furthermore, we found that the average velocity and persistence of migration was higher in cells with reduced Adam11 (Figure 2G). Collectively, these results show that Adam11 is essential for proper CNC migration and may influence the timing of neural tube closure and single neural crest cell migration. We then tested if neural and neural crest cell markers were expressed and localized properly in embryos lacking Adam11.

Given the distribution of Adam11.L mRNA in the CNC and neural tube, and the phenotypes observed following Adam11 KD, we focused our analysis on neural and neural crest markers at stages when the first phenotype is visible (Neurula stage). We used the well-characterized neural crest marker Slug/Snai2 (Figure 3A), Sox8 (Figure 3B), and Sox9 (Figure 3C) in embryos injected unilaterally at the 2-cell stage with MO11 and a fluorescent lineage tracer. This allows KD in one-half of the embryo (right or left side). In-situ hybridization revealed that all three markers were expressed at the appropriate position. Interestingly, we found that these markers were increased on the injected side (Right and Figure 3F). Expansion of the neural crest territory can, in some cases, be done at the expense of the neural ectoderm. We tested the position of two neural markers, Sox2 (Figure 3D) by in situ hybridization and Sox3 by immunostaining (Figure 3E). In both cases, we found that the domain of expression on the injected side was also wider, suggesting that the increase in neural crest territory was not done at the expense of the neural tissue.

Given the early onset of single CNC cell migration and increased Slug expression, we hypothesized that β-catenin controlled EMT could be dysregulated following Adam11 KD. To test this hypothesis, we dissected CNC explants from embryos injected into one cell of the two-cell stage and compared the explants from the control and the injected side after migration on fibronectin-coated slides (Figure 4). Membrane Cherry was co-injected with the morpholino to identify the injected side and confirm that the CNC had been targeted. We found that while the overall fluorescence of β-catenin was similar on the control and injected side, the percentage of nuclei with β-catenin protein was expanded on the injected side (Figures 4A, C compare NI to MO11). In control explants, the β-catenin protein was present near the plasma membrane and in the nuclei of the cells that are at the leading edge of the explant but was mostly restricted to the plasma membrane in cells within the explant (Figure 4A, NI) (Alfandari et al., 2010). In contrast, explants missing Adam11 show nuclear β-catenin localization all the way through the explant (Figure 4A MO11). In addition, it is also clear that the nuclei of the control explants are more packed (Figure 4A DAPI) than the one from the Adam11 KD, suggesting that the cells are denser on the FN substrate (Figure 4D). Nuclear β-catenin is known to activate gene expression, including slug (Figure 3A) and cyclin D1. We, therefore, tested the expression of CyclinD1 in CNC explants (Figure 4B). Again, while the expression of CyclinD1 is mostly restricted to the periphery of the CNC explant from the control side, the expression is present throughout the explant from the injected side, and the overall intensity is significantly higher than in the control side (Figure 4E). It is important to note that these images are maximum projections so that protein expression levels should not be affected by their relative Z-position within the explant.

To directly test if the β-catenin transcriptional activity was affected by the loss of Adam11, we used the well-characterized Top-Flash luciferase reporter (Chuang et al., 2010; Baarsma et al., 2011) in vivo. We targeted the Top flash reporter to the precursor CNC territory at the 16-cell stage (Moody, 1987) in either control embryos or embryo KD for Adam11. We found a statistically significant increase in Top flash activity in embryos KD for Adam11 (Figure 5B).

Given the known role of CyclinD1 in controlling the G1 to S transition (Stacey, 2003), we tested the proliferation of CNC cells using Edu labeling. Embryos were incubated in Edu for 2 hours at the beginning of neurulation (stage 15/16) when the expression of Adam11 mRNA starts accumulating (Session et al., 2016). CNC explants were dissected at stage 17 and allowed to migrate on fibronectin for 3 h prior to fixing and staining for Edu-positive nuclei (Figure 5C). We chose this timing to allow enough cells to divide in the CNC since we only have 300 cells per explant on average (2 h labeling, Fixed 4 h after the end of labeling). We observed a significant increase in Edu-positive cells in CNC with lower Adam11 when compared to control explants (Figure 5D). To account for variability in explant size and labeling stage between experiments, the number of Edu-positive nuclei was normalized to the size of each explant and set to one for the control explants. Loss of E-cadherin and gain of N-cadherin is a classical hallmark of EMT in CNC and cancer (Araki et al., 2011; Scarpa et al., 2015; Huang et al., 2016; Cousin, 2017), we therefore, tested the expression of E-cadherin and N-cadherin via quantitative capillary Western blot on protein extracts from stage 17 CNC explants. When normalized using Ribophorin 1 (Rpn1 as a loading control), we found that CNC lacking Adam11 expressed less E-cadherin and more N-cadherin than the control explants (Figures 5E, F). Together, these results show that loss of Adam11 increases β-catenin transcriptional activity and causes earlier EMT, suggesting that Adam11’s role is to reduce and/or delay this activity in the neural crest.

While the apparent early EMT pointed to β-catenin, we wanted to use an unbiased approach to identify additional functions for Adam11 in neural and neural crest cells. Adam11 is a non-proteolytic ADAM with no known role during embryogenesis. Protein interaction studies have identified putative binding partners for Adam11 in mouse, including integrins and LGl proteins in neurons (Takahashi et al., 2006a; Takahashi et al., 2006b; Sagane et al., 2008; Kole et al., 2015; Wang et al., 2018). These putative protein interactions cannot explain the effect of Adam11 KD on β-catenin activation. In the absence of an antibody to Xenopus Adam11 at the time of this experiment, we expressed the Flag-tagged Adam11.L in Xenopus Cranial neural crest cells and human Hek293T cells and pulled down interacting proteins (Figure 6). The flag-tag experiment also has the advantages that no part of the proteins is blocked by the antibody. In embryos, we injected at the 8-cell-stage to target the CNC and extracted the proteins at either the neurula or tailbud stages (4 samples from two independent experiments). We considered proteins with at least two unique high-confidence peptides (95% peptide threshold, 99% protein threshold) that were present in at least two of the 4 samples as potential candidates. We also performed Flag pull down on human Hek293T cells transfected with Adam11-Flag or RFP-Flag. Among the proteins that were found in both data sets (Supplementary Table S1), we identified the secreted protein BMP4. We also found multiple members of the LRP family of protein and the Wnt receptor FzD6, but these were not present in both data sets (LRP2 in embryos, LRP5-6, and Fzd6 in Hek293T cells). We also found the BMP4 receptor BMPR1A in Hek293T cells. Given the critical role of the Wnt/β-catenin and BMP4 signaling pathway to both neural and neural crest cell development and our results showing increased β-catenin activity, we focused on these putative binding partners for this study. In addition, other ADAMs have been shown to interact with Wnt receptors or small secreted forms of the receptor that act as antagonists (Esteve et al., 2011; Abbruzzese et al., 2015); thus, we also tested the interaction with Wnt receptor Fzd. We were able to confirm the interaction of Adam11 with BMP4 (Figure 6C) and the BMP receptor 1A (Figure 6D) by co-immunoprecipitation. We also identified the interaction of Adam11 with the Wnt receptor Fzd7, a key component of the β-catenin signaling pathway in embryos (Figure 6E).

To understand better how the Adam11 KD changed the state of the CNC cells, we performed mass spectrometry on the dissected CNC explants (Figure 7A). After analysis of triplicate experiments, we focused on proteins that were significantly up or downregulated and identified with a minimum of one (Supplementary Figure S1; Supplementary Table S2; Supplementary Table S3) or two (Figure 7A) high confidence peptides. Selecting one peptide rather than two reduces the confidence that the protein is present but allows to identify proteins that are less abundant. While the majority of proteins identified were common (386 at 2 peptides and 2020 at 1 peptide), we found 80 proteins that were upregulated in the morphant and 40 proteins that were downregulated in the CNC KD for Adam11. In particular, we found that Hsp90ab1 was only detected in Adam11 KD CNC. Hsp90ab1 has been shown to interact with the β-catenin destruction complex (Cooper et al., 2011) as well as LRP5 (Wang et al., 2019) protein to increase β-catenin signaling and promote EMT. It is important to note that HSP expression is not a general response to Morpholino as only nine HSP proteins were identified in our MS, three were upregulated (Hsp90ab1, Hsp90b1 and Hspa4). In addition, Hsp90b1 and Hspa4 were also upregulated in CNC KD for Adam13/33 but not Hsp90ab1 showing a clear specificity (Supplementary Figure S2). To confirm the result from the Mass spectrometry experiments, we performed immunofluorescence on CNC explants using an antibody to Hsp90ab1 (Figures 7B, C). These results are strikingly similar to those observed for CyclinD1 (Figure 4B) and confirm the upregulation of Hsp90ab1 in CNC lacking Adam11.

As shown above, Adam11 interacts with both BMP4 and its receptor BMPR1A (Figure 6). We also found that secreted BMP4 interacted with shed Adam11 in the conditioned media of co-transfected cells (data not shown). We, therefore, tested if Adam11 binding to BMP4 increased or decreased BMP4 signaling. For this, we used the animal cap (AC) assay. The intact animal cap represents a naive ectoderm with high BMP4 signaling and no expression of Adam11 based on published RNAseq data (Xu et al., 1995; Sive et al., 2007; Haas et al., 2019). Dissociation of the animal cap dilutes BMP4 signaling and induces neural tissue that expresses a high level of Sox3 (Figure 8A). We repeated these experiments using AC expressing injected Adam11.L. Following animal cap dissociation into single cells, we found that Sox3 was significantly reduced in explants from embryos overexpressing Adam11 (Figure 8A, A11AC). In the same samples, we found that the phosphorylated Smad1/5 was increased by Adam11 overexpression (Figure 8B), consistent with an activating role of Adam11 toward BMP4. To further test the activity of Adam11 towards BMP4 signaling during neurulation, we performed a luciferase assay using the pGL2-15xGCCCG-lux BMP reporter (Alkobtawi et al., 2021) targeted at the 8-cell stage (Figure 8C). The luciferase activity was decreased with Adam11 KD further confirming that Adam11 increases BMP4 activity. BMP4 plays a critical role during neural induction by defining the ventral limit of the neural tissue, while BMP4 inhibitors are essential for neural induction (Weinstein and Hemmati-Brivanlou, 1997). To test if Adam11 interaction with BMP4 was important for the neural tube closure phenotype (Figure 1D), we repeated the assay using low doses of BMP4 mRNA (12.5 pg) to rescue the neural tube closure phenotype (Figures 8D, E). Indeed, injection of BMP4 in the dorsal part of the embryo rescued the neural tube closure timing to a level indistinguishable from the non-injected control (Figure 8E). Taken together, these results show that Adam11 interacts with the BMP4 ligand and receptor to increase local signaling to a level critical for proper neural tube closure. The role of Adam11 in controlling BMP4 signaling levels is also consistent with our observation that Adam11 KD increases the domain of expression of neural markers such as sox2 and Sox3 (Figure 3).

We have shown that Xenopus Adam11 restricts β-catenin activity in the CNC. Loss of Adam11 increases single CNC cell migration in vitro, resembling an epithelium to mesenchyme transition (EMT). Given the role of β-catenin in cancer and the known relation between EMT and cancer metastasis, we looked at human ADAM11 gene expression in available databases (Bartha and Gyorffy, 2021). Out of 56,938 unique multilevel quality-controlled samples, ADAM11 expression was found to be significantly different in 18 out of 23 tumor types (Figure 9A*). It was significantly reduced in 16 of the 18 tumor types. Given the previous observation that ADAM11 was mutated in two breast cancer samples (Emi et al., 1993), we further focused our analysis on data obtained from invasive breast carcinoma (Figure 9B). The expression level was significantly lower in the samples from cancer patients (1,097 patients) when compared to normal tissues (403 patients, mean FC = 0.35, p = 10⁻⁸⁴). Interestingly in the same samples, CYCLIND1 was found to be significantly increased (mean FC = 4.15, p = 2.29 10⁻⁷⁶). Together, these results strongly suggest that Adam11’s ability to restrict β-catenin activity and delay EMT in the CNC is conserved in adult tissues and that loss of ADAM11 may contribute to β-catenin hyperactivation and CYCLIND1 overexpression, leading to tumor progression and metastasis.

One type of tumor in which ADAM11 is increased rather than decreased is neuroblastoma (Figure 9C, 190 normal and 149 tumor samples, mean FC = 61.52 p = 6.94 10⁻⁵⁶). In these tumors, we found that SMAD1 and SMAD5 were significantly increased (mean FC = 4.46 p = 5.8 10⁻⁵² and FC = 1.31 p = 8.55 10⁻⁸ respectively) while SOX3 was significantly decreased (mean FC = 0.13 p = 4.29 10⁻⁴⁹). An observation that is consistent with our results showing that the overexpression of Adam11 in the naïve animal cap ectoderm increases BMP4 signaling, resulting in increased Smad1-5 phosphorylation and decreased Sox3 (Figure 8), while loss of Adam11 increases Sox3 expression (Figures 3E, F). It is important to note that for human cancer, we do not have phosphorylation data concerning Smad and that the increase in mRNA expression, while indicative, is not a direct indicator of an increase in BMP4 signaling.

The human cancer analysis indicates a strong correlation between ADAM11 expression and CYCLIND1 levels. To further test if Adam11 expression in cancer cells could modulate the Wnt signaling pathway, we used mouse B16 melanoma cells (Poste et al., 1980). These cells are derived from neural crest cells, express a relatively high level of Adam11, and are highly invasive in vivo when injected into the bloodstream but not subcutaneously. Furthermore, in skin cutaneous melanoma, published RNA-seq data shows a decreased level of ADAM11 and an increased level of CYCLIND1 between tumors (103 patients) and normal samples (474 patients, Figure 10A). We used shRNA mediated inhibition of Adam11 in these cells to test the effect on cell proliferation. Based on the CNC results, we expected that loss of Adam11 would increase Wnt signaling and CyclinD1 expression (Figure 4), resulting in increased cell proliferation (Figure 5). While we obtained only a modest but reproducible reduction in Adam11 expression (35%, Figure 10B), we were able to see an increase in total cell number (40%) in cells with reduce Adam11 expression (Figure 10C). In order to understand the increase in cell number (Figure 10D), we used the well-characterized FUCCI cell cycle reporter (Sakaue-Sawano and Miyawaki, 2014), transfected in the context of Adam11KD and used FACS to follow cell cycle progression. Following successful transfection, the cells are red when they are in G1, yellow during the transition from G1 to S, or green if they are in S, G2, or M phase (Figure 10D). We found that the decrease in Adam11 resulted in a significant decrease in the number of cells in the G1-S transition and increased number of cells in S, G2 and M phases (Figure 10G). Like our finding in CNC, we found that B16 with reduced Adam11 have increased β-catenin (Figure 10F). We then tested by RT-qPCR the expression of multiple markers of EMT and target of β-catenin. We observed a statistically significant increase in CyclinD1 and Snail levels (Figure 10E), and while there was a small decrease in E-cadherin and an increase in N-cadherin and Vimentin that fit with our EMT expectation, those were not significant.

ADAM11 is well conserved between Xenopus, Mouse, and Human both in the extracellular and cytoplasmic domains (Supplementary Figure S3). This conservation is especially striking in the short cytoplasmic domain that does not contain any putative functional attribute. The least conserved region is the pro-domain, which in proteolytic ADAM controls the activation. Thirty years ago, ADAM11 was found to be mutated in two mammary gland tumors and was postulated to be a tumor suppressor (Emi et al., 1993). Surprisingly, no other study tested this hypothesis. Given our results showing that Adam11 regulates β-catenin activity in Xenopus embryos (Figure 4), we analyzed the expression of ADAM11 in available tumor expression data. The analysis shows that ADAM11 expression is significantly lower in most solid tumors (Figure 9A), including the neural crest-derived melanoma, and this is even more pronounced in invasive breast carcinoma, a type of tumor where ADAM11/MDC was originally found mutated, supporting the hypothesis that ADAM11 may be a tumor suppressor gene. We find that in addition to the significant decrease in ADAM11 (Figures 9, 10), there was also a significant increase in CYCLIND1. While there is no causal relation in this observation, our data from Xenopus embryos and mouse B-16 melanoma cells shows that decreasing Adam11 protein level is sufficient to increase β-catenin activity and CyclinD1 expression, suggesting a conserved function of Adam11 in cancer. It will be of interest to test if human ADAM11 can also regulate β-catenin and rescue neural tube closure timing and CNC migration in Xenopus embryos. On the other hand, it would be interesting to test if increasing ADAM11 protein levels in cancer cells is sufficient to reduce β-catenin activity, CYCLIND1 expression, and cell proliferation. While this seems like an interesting strategy, we have also found that ADAM11 expression is increased in Neuroblastoma, a childhood cancer thought to be derived from neural crest cells. In Neuroblastoma, the increase in ADAM11 expression is associated with increased expression of Smad1 and 5 (Here mRNA expression rather than Phosphorylation) and decreased expression of Sox3, similar to what we observed when overexpressing ADAM11 in naïve ectoderm in Xenopus embryos (Figure 8) suggesting that in Neuroblastoma ADAM11 positive regulation of BMP4 signaling could lead to the repression of neural differentiation markers (Sox2 and 3) resulting in more naïve cells. These naïve or “stem” cells could contribute significantly to the malignancy. Therefore, our observation suggests that a fine control of ADAM11 expression is critical for normal development as well as tissue homeostasis. A careful structure/function analysis will be required to understand the role of each of the functional domains to the overall function of ADAM11.

We have found that CNC migrated well in vitro but not in vivo. This is certainly not unique, as we found this to be the case in most ADAM inhibition, highlighting the essential role of ADAM interaction with the extracellular environment (McCusker et al., 2009; Cousin et al., 2012). For example, we have previously shown that Adam13/33 activity is no longer required when the CNC cells are physically separated from the underlying mesoderm in vivo (Cousin et al., 2012). In vitro, CNC migrates on top of a pure fibronectin substrate, while in vivo, the CNC are in a three-dimensional extracellular matrix that is crosslinked and composed of multiple proteins in addition to fibronectin (e.g., Laminins and collagens). In addition to this difference in structure, in vivo, CNC are presented with both attractive and repulsive signals that are absent in vitro.

While β-catenin is essential for proper CNC induction, studies have shown that β−catenin activity needs to be reduced for proper migration and that abnormal activation within the CNC leads to inhibition of migration (Maj et al., 2016). This suggests that the excess activation of β−catenin following Adam11 KD could be responsible for the perturbation of CNC migration in vivo. In contrast, both chemical inhibition of GSK3 and activation of TcF3 also inhibit migration in vitro, preventing single-cell dispersion (Maj et al., 2016). This is opposite to the phenotype that is observed in the Adam11 morphant embryos where single cell number increase, and both speed and persistence are higher. This suggests that the loss of Adam11 is not simply a hyperactivation of β−catenin and that the decrease in BMP4 signaling may also influence CNC migration. Certainly, the activation of β−catenin following the loss of Adam11 is much more subtle as the overall level of the protein is unchanged, and only the nuclear localization is increased.

Adam11 knock-out mice show normal mechanical and noxious heat responses but have significantly decreased chemical nociception, indicating that Adam11 is necessary for the formation or the relay of specific set of sensory information (Takahashi et al., 2006a). The sensory neurons are derived mostly from the cranial placodes and cranial neural crest (cranial ganglia) anteriorly and from the trunk neural crest posteriorly (Vermeiren et al., 2020). The loss of specific type of nociception in Adam11 knock-out mice (Takahashi et al., 2006a) raises the possibility that the defect in these neurons might be a consequence of abnormal specification or migration of the neural crest. Adam11 transcripts have been detected in the dorsal root ganglia and the neural crest in the mouse (Rybnikova et al., 2002; Diez-Roux et al., 2011) and the frog (Cai et al., 1998). While no obvious anomalies were reported in the DRG of Adam11 KO mice, it is likely that the absence of the small nociceptive neurons would not be detected without using cell-specific markers. Alternatively, it is also possible that these neurons exist but fail to properly organize their synapse, as shown for the cerebellar basket cell (Kole et al., 2015). Further analysis of the Adam11 morphant embryos at a later stage could help identify subtle DRG defects.

Neural tube closure is a complex morphogenetic event that involves hundreds of different genes. In Xenopus, like in mammals, the neural folds elevate toward the center of the neural plate and then fuse. It is important to note that the neural plate, as defined by Sox3 expression, is wider on the side missing Adam11. Defects in the non-canonical Wnt planar cell polarity (PCP) pathway induce a wider neural plate, but contrary to Adam11 KD, the neural fold fails to reach the midline (Goto and Keller, 2002; Wallingford and Harland, 2002). While the role of the PCP pathway is clearly critical, β-catenin has been shown to regulate the expression of Pax3 and Cdx2, two transcription factors critical for neural tube closure. In the absence of β-catenin, neural tube closure is inhibited but can be rescued by the expression of Pax3 (Zhao et al., 2014). Therefore, an increase or early activation of β-catenin in Adam11 morphant embryos could lead to the activation of Pax3 and the early neural tube closure. In mouse, FGF3 mutants have elevated levels of BMP signal, increased neural cell proliferation, and delayed neural tube closure (Anderson et al., 2016). This would be consistent with the decreased BMP4 signaling and accelerated neural tube closure in Adam11 morphant embryo and could explain the rescue observed using low levels of BMP4 mRNA (Figures 8D, E). Careful analysis of β-catenin and BMP4 signaling levels prior to neural tube closure would be critical to determine if they are indeed responsible for this Adam11 phenotype. In the future, it would be interesting to determine if the early closure of the neural tube affects the proper specification of the various neuronal cell types and their organization. This could partly explain the neurological disorder observed in mice lacking Adam11. It is clearly much more difficult to evaluate neural tube closure timing in mouse embryos, and as such, it is understandable that those have not been reported.

Model of ADAM11 function in Cancer and Neural crest cells (Figure 11). We have shown that Adam11 binds to BMP4 and its receptor BMPR1A (Figure 6). Our results using the animal cap dissociation experiment (Figure 8) show that Adam11 increases BMP4 signaling, possibly by retaining BMP4 at the plasma membrane and presenting the protein to the receptor (Figure 11). Thus, Adam11 expression at the surface of the CNC would allow them to amplify the gradient of BMP4 locally restricting the expression of neural markers such as Sox2 and Sox3 to define the CNC zone (Figure 3). At the same time, Adam11 can also bind to Wnt receptors Fzd as well as multiple LRP proteins (Figure 6). We propose that the interaction of Adam11 with these Wnt receptor and co-receptor reduce their capacity to either bind to the Wnt ligand or to induce the cytoplasmic cascade that inhibit the destruction complex of β-catenin (Figure 11). Given the capacity of several ADAM protein to bind to the cysteine rich domain of Fzd or Sfrp proteins (Esteve et al., 2011; Abbruzzese et al., 2015), the usual binding site of the Wnt ligand (Dijksterhuis et al., 2015), Adam11 could simply act as a competitive inhibitor of the Wnt/Fzd complex. Alternatively, Adam11 could wedge itself between the Fzd and LRP proteins reducing the activation of the complex. In CNC lacking Adam11, we find a much higher Hsp90ab1 protein expression. This protein has been shown to stabilize Lrp proteins, thus reinforcing the Wnt signaling pathway. Refining the composition of the Adam11 protein complex and the binding sites of each protein will be required to understand the exact mechanism of Adam11 inhibition of β-catenin activation.

The following antibodies were used in this study: monoclonal antibody to Ribophorin-1 (Rpn1) was used as loading control, RRID:AB_2687673 (Khedgikar et al., 2017). The rat monoclonal to flag tag (Agilent Cat# 200474, RRID:AB_10597743). The monoclonal to X. laevis Sox3 (DSHB Cat# DA5H6, RRID:AB_2876376) was developed in the laboratory (Horr et al., 2023) against the full length X. laevis protein produced in Hek293T cells. It has been characterized by Western blot and immunofluorescence (R24OD21485). The rabbit polyclonal to HA tag (Applied Biological Materials Cat# G166, RRID:AB_2813867). The Rabbit monoclonal to phospho-Smad 1/5 (Cell Signaling Technology Cat# 9516, RRID:AB_491015). Mouse monoclonal to BMPR1A (Thermo Fisher Scientific Cat# MA5-17036, RRID:AB_2538508). Rabbit polyclonal to β-catenin (Abcam Cat# ab16051, RRID:AB_443301). Rabbit polyclonal to Cyclin-D1 (Bioss Cat# bs-0623R, RRID:AB_10856925). Mouse monoclonal to Hsp90ab1 (Thermo Fisher Scientific Cat# 37-9400, RRID:AB_2533349). The anti-myc antibody 9E10 (DSHB Cat# 9E 10, RRID:AB_2266850). Rat monoclonal against N-cadherin (DSHB Cat# MNCD2, RRID:AB_528119). Mouse monoclonal to X. laevis Adam11.L (3C5) was developed in the lab by injecting a bacterial fusion protein containing the DC (disintegrin and cystein rich) domain and isolating the clone by the methods described in (Horr et al., 2023).

Morpholino antisense oligonucleotides (Gene Tools, Philomath OR) were designed to either block the translation of Adam11L and S (MO11 Figure 1, CAT​CAT​TCA​TAC​ATA​ATC​ATT​CAG​C) or the splicing of Adam11L (MO11spl, CCA​ATG​GAG​GCA​TAG​TGG​CAC​TG). Adam11.L was cloned in pCS2+ using cDNA made from RNA extracted from stage 18 Xenopus laevis embryos. Flag and HA tags were added using Takara infusion cloning according to the manufacturer’s instructions. BMP4-Flag in pCS2 was a generous gift from Dr. Chenbei Chang (University of Alabama). Top Flash was previously described in (Chuang et al., 2010). All constructs were sequenced and transfected in 293T cells to validate protein translation, size and proper cell compartment expression.

SP6 polymerase was used for capped mRNAs synthesis after digesting using Not1 (Cousin et al., 2011). Injectors were calibrated using a 1 µL capillary needle (Microcaps, Drumond, PA, United States). The injection pressure was set at 15 psi and the injection time set between 50 and 200 ms to obtain a 5 nL delivery. Embryos were injected at 1-cell, 2-cell, 8-cell and 16-cells as described previously (Khedgikar et al., 2017). Embryos were raised at 14°C–15°C until scoring for neural tube closure or CNC migration. For each injection, percentage of open NT was normalized to control embryos and was set to 100%, same was done for CNC migration using Membrane Cherry injected embryos. CNC grafts were performed as described (Alfandari et al., 2001; Cousin, 2018). CNC explants migration assays were performed as described in (Alfandari et al., 2003). All experiments were performed at least three times using different females to determine statistical significance.

Hek293T (ATCC, CRL-3216 RCB2202) cells were cultured in RPMI media supplemented with 10 U/mL Pen/Strep, 2 mM L-glutamine, 0.11 mg/mL sodium pyruvate, and 10%FBS (Hyclone, South Logan, UT). Transfections were performed using Polyethylenimine (PEI, Polysciences Inc.). For each transfection, 1 µg of DNA was mixed with 10 µg of PEI in 200 µL of Optimem media (Hyclone) at room temperature for 15 min prior to adding to the cells dropwise. The media was changed after 24 h post transfection.

Mouse melanoma B16/F10 (CRL-6475) cells were a generous gift from Dr. Leonid pobezinsky, the cells were cultured in DMEM media with 10 U/mL Pen/Strep and 10%FBS. Transfections were performed as described for Hek293T.

Animal caps dissociation and reaggregation was done as described in (Sive et al., 2007). Briefly, 10 to 20 animal caps were dissected in 1XMBS (1XMBS: 88.0 mM NaCl, 1.0 mM KCl, 2.4 mM NaHCO₃, 15.0 mM HEPES [pH 7.6], 0.3 mM CaNO₃-4H₂O, 0.41 mM CaCl₂-6H₂O, 0.82 mM MgSO₄) containing 50 μg/mL of gentamycin. They were then transferred to Calcium, Magnesium free MBS (CMF) supplemented with 1 mg/mL of BSA for 2 h with gentle rocking to dissociate the animal cap into single cells suspension. The cells were then transferred into conical 96 well (Thermofisher) plate precoated with PBS 1% BSA and incubated overnight in Danilchik media (53 mM NaCl, 11.7 mM Na₂CO₃, 4.25 mM potassium gluconate, 2 mM MgSO₄, 1 mM CaCl₂, 17.5 mM Bicine, 1 mg/mL BSA, pH 8.3) until control embryos reached stage 20.

Quantitative real time PCR was done as previously described (Neuner et al., 2009). All primers were tested for efficiency. Embryos were collected at stage 15–18, B-16 cells were collected after 48 h of transfection and RNA was extracted using (Roche, RNA isolation Kit). Total RNA was quantified on a nanodrop (Thermofisher) at 260 nm. The cDNA was made using (Invitrogen Superscript III) according to manufacturer’s instructions. Quantitative PCR was performed using SYBR green (Takara, Kyoto, Japan) to measure mRNA levels of Adam11.L and GAPDH was used to normalize total cDNA amount, in case of B-16 cells beta-actin was used to normalize the total cDNA amount. The relative gene expression was calculated using the ΔΔCT method as described (Livak and Schmittgen, 2001). The result is represented in fold change compared to non-injected control. Oligos used are described in the table below (xl indicate X. laevis primer, all other primer are mouse).

Hek293T, embryos, Animal caps and CNC explants were extracted in 1XMBS-1% Triton-X100, Protease phosphatase inhibitor cocktail 1X (Thermoscientific) and 5 mM EDTA. Immunoprecipitations were performed using 1-2 ug of antibody bound to protein A/G magnetic beads (Thermofisher), incubated overnight at 4°C. The beads were washed 3 times for 5 min with extraction buffer at room temperature then eluted in 2X reducing laemmli sample buffer. All proteins were separated in 5%–22% gradient SDS-PAGE gels and transferred to polyvinylidene fluoride membranes (PVDF, Millipore, Billerica, MA) using a semi-dry transfer apparatus (Hoeffer). Quantitative capillary western blots were performed using Biotechne automatic western machine (Wes) according to manufacturer’s instructions https://www.bio-techne.com/resources/videos/meet-wes-protein-characterization-problem-solver. The data obtained using WES is a graph corresponding to the light intensity from the chemiluminescence. This graph can be shown as a conventional gel image as shown in Figure 1B or as an intensity histogram as shown in Figure 1C.

Whole mount in situ hybridization was done using previously described protocol (Harland, 1991). The probes for Sox2, Snai2/Slug, Sox9, Sox8 were synthesized using diogoxigenin-rUTP–label.

CNC cells were dissected at stage 17 and placed on fibronectin coated glass bottom plates (20ug/mL) for 1–3 h at 18°C (Cousin and Alfandari, 2018). The explants were fixed using MEMFA (0.1 M MOPS pH 7.4, 2 mM EGTA pH 8, 1 mM MgSO4 and 4% paraformaldehyde) for 1 h, permeabilized using 0.5% TX100 in 1XMBS with 100 mM glycine for 1 h. The explants were then blocked using PBS containing 10% heat inactivated goat serum, 1% BSA, 0.1% Tween for 1 h prior to incubation in the same buffer with the primary antibodies overnight at 4°C. CNC explants were washed in PTw (PBS-tween 0.1%) 3 times 15 min, blocked again for 1 h at room temperature using blocking solution prior to incubation with the secondary antibody and Hoest33342 for 1 h at room temperature. The explants were washed 3 times in PTw and imaged using Nikon confocal microscope (A1RHD25).

The Immunofluorescence images were analyzed using Fiji (Schindelin et al., 2012) the mean gray value was calculated for the antibody specific channel and was normalized using the mean gray value for DAPI. The images were taken at same gain and power for all explants within the same experiment. To account for the differences in-between biological repeats, in each repeat the NI was set to 1 and the values for MO11 were calculated relative to it. Edu + cells were counted manually using Fiji and normalized to the relative size of the explant.

Protein extracted from 10 CNC explants as described above or immunoprecipitated with the Flag antibody from embryos or Hek293T cells were diluted in 8M urea and processed with Trypsin/Lysine-C according to manufacturer instruction (Promega). Tandem mass spectrometry was performed using a Thermo orbitrap Fusion (Mass spectral data were obtained at the University of Massachusetts Mass Spectrometry Core Facility, RRID:SCR_019063). All MS/MS samples were analyzed using Sequest (Thermo Fisher Scientific, San Jose, CA, United States; version IseNode in Proteome Discoverer 2.4.1.15). Sequest was set up to search XenbaseProteinXL2020.fasta (unknown version, 72,266 entries) or the human protein database uniprot-human containing the sequences for Xenopus Adam11 (126,358 entries) assuming the digestion enzyme trypsin. Sequest was searched with a fragment ion mass tolerance of 0.60 Da and a parent ion tolerance of 10.0 PPM. Carbamidomethyl of cysteine was specified in Sequest as a fixed modification. Oxidation of methionine, acetyl of the n-terminus and phospho of serine threonine and tyrosine were specified in Sequest as variable modifications.

Criteria for protein identification: Scaffold (version Scaffold_5.0.1, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Peptide Prophet algorithm (Keller et al., 2002) with Scaffold delta-mass correction. Protein identifications were accepted if they could be established at greater than 99.0% probability and contained at least 2 unique identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Proteins sharing significant peptide evidence were grouped into clusters.

Flow cytometry was performed on B-16F10 cells that were transiently transfected with the various plasmids. Only the cells successfully transfected with FUCCI were fluorescent and analyzed. The cells were harvested 36–48 h post transfection using a cell scraper and resuspended in PBS 1% BSA. Flow cytometry was performed on a BD DUAL LSRFortessa and the data was analyzed using FACSDiva 8.0. 50,000 cells were acquired for each sample in three independent experiments.

If not specified in the figure legends, for two samples two-tail unpaired Student’s t-test was performed. For more than 2 samples one-way ANOVA was performed.