Regulated ADAM10-dependent Ectodomain Shedding of γ-Protocadherin C3 Modulates Cell-Cell Adhesion*

γ-Protocadherins (Pcdhγ) are type I transmembrane proteins, which are most notably expressed in the nervous system. They are enriched at synapses and involved in synapse formation, specification, and maintenance. In this study, we show that Pcdhγ C3 and Pcdhγ B4 are specifically cleaved within their ectodomains by the disintegrin and metalloprotease ADAM10. Analysis of ADAM10-deficient fibroblasts and embryos, inhibitor studies, as well as RNA interference-mediated down-regulation demonstrated that ADAM10 is not only responsible for the constitutive but also for the regulated shedding of these proteins in fibroblasts and in neuronal cells. In contrast to N-cadherin shedding, which was activated by N-methyl-d-aspartic acid receptor activation in neuronal cells, Pcdhγ shedding was induced by α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid hydrate stimulation, suggesting differential regulation mechanisms of cadherin-mediated functions at synapses. Cell aggregation assays in the presence or absence of metalloprotease inhibitors strongly suggest that the ectodomain shedding events modulate the cell adhesion role of Pcdhγ. The identification of ADAM10 as the protease responsible for constitutive and regulated Pcdhγ shedding may therefore provide new insight into the regulation of Pcdhγ functions.

␥-Protocadherins (Pcdh␥) are type I transmembrane proteins, which are most notably expressed in the nervous system. They are enriched at synapses and involved in synapse formation, specification, and maintenance. In this study, we show that Pcdh␥ C3 and Pcdh␥ B4 are specifically cleaved within their ectodomains by the disintegrin and metalloprotease ADAM10. Analysis of ADAM10-deficient fibroblasts and embryos, inhibitor studies, as well as RNA interference-mediated down-regulation demonstrated that ADAM10 is not only responsible for the constitutive but also for the regulated shedding of these proteins in fibroblasts and in neuronal cells. In contrast to N-cadherin shedding, which was activated by N-methyl-D-aspartic acid receptor activation in neuronal cells, Pcdh␥ shedding was induced by ␣-amino-3-hydroxy-5-methylisoxazole-4-propionic acid hydrate stimulation, suggesting differential regulation mechanisms of cadherin-mediated functions at synapses. Cell aggregation assays in the presence or absence of metalloprotease inhibitors strongly suggest that the ectodomain shedding events modulate the cell adhesion role of Pcdh␥. The identification of ADAM10 as the protease responsible for constitutive and regulated Pcdh␥ shedding may therefore provide new insight into the regulation of Pcdh␥ functions.
Cadherins are calcium-dependent cell adhesion molecules that play a fundamental role in embryonic and tissue development (1,2). Protocadherins (Pcdhs) 3 are cadherin-related adhesion molecules with six or seven extracellular cadherin motifs. Their cytoplasmic domains diverge from each other and from those of the classical cadherins, indicating different binding properties and intracellular functions. With 80 members, the Pcdhs represent the largest subgroup of the cadherin superfamily in mammals (3). More than 50 of these genes are arranged in three clusters: Pcdh␣, Pcdh␤, and Pcdh␥. The Pcdh␤ cluster comprises tandemly arrayed single-exon genes flanked by individual promoters, whereas the ␣and ␥-clusters additionally contain at the distal 3Ј end three small exons coding for a cluster-specific constant domain. The variable domains of the Pcdh␣ and Pcdh␥ proteins are encoded by large exons and encompass most of the transmembrane protein, including a short cytoplasmic tail. The Pcdhs derived from these gene clusters are predominantly expressed in the nervous system and localized to synaptic junctions (4 -6). Functionally connected neurons are often characterized by the expression of a distinct set of cadherins, which may thus provide an adhesive code regulating neuronal differentiation and synaptogenesis (7)(8)(9). The family-specific constant domains of Pcdh␣-and Pcdh␥-proteins could transduce different extracellular informations into a common intracellular signaling pathway or other executive functions. The large number of distinct Pcdhs possibly contributes to the complexity of neuronal connections by significantly increasing the number of adhesive and/or signaling combinations (7,10).
Recent reports showed that Pcdh␥ proteins undergo proteolytic processing events. A first cleavage is executed by a metalloprotease activity in between the 18 amino acids NH 2 -terminal to the transmembrane segment, releasing a soluble fragment. The precise nature of the Pcdh␥ shedding protease is unknown. The proteolytic ectodomain cleavage also leads to a membranebound carboxyl-terminal (COOH-terminal) fragment, which is a substrate for regulated intramembrane proteolysis (11,12). In this process the COOH-terminal membrane-bound fragment becomes a substrate for a presenilin/␥-secretase-mediated proteolysis, resulting in the release of a cytoplasmic COOH-terminal fragment with nuclear localization establishing a role for Pcdh␥ as cell surface signaling molecules. The regulation of regulated intramembrane proteolysis as shown for substrates like Notch, APP, and CD44 is thought to occur at the level of the first ectodomain shedding processing step (13)(14)(15). This observation and the fact that other regulated intramembrane proteolysis substrates including classical cadherins are cleaved by similar proteases (16,17) suggested to us that a member of the disintegrin and metalloprotease (ADAM) family is involved in the ectodomain cleavage of Pcdh␥.
The ADAMs are a family of type I transmembrane proteins and combine features of both cell adhesion molecules and proteases. They play important roles in fertilization, neurogenesis, and angiogenesis (18) and are involved in the shedding of various membrane-bound proteins including cytokines, growth factors, and adhesion molecules (19,20). ADAM10 and ADAM17 (TACE, TNF␣-converting enzyme) have been studied in particular in the context of ectodomain shedding. They are involved in the proteolysis of various substrates such as Notch, epidermal growth factor ligands, and fractalkine (21)(22)(23).
In the present study we focused on the potential role of different ADAMs in Pcdh␥ shedding. We demonstrate that ADAM10 is the prominent protease responsible for Pcdh␥ C3 ectodomain cleavage in fibroblasts, neuronal cells, and mouse embryos. This shedding process modulates Pcdh␥ C3-mediated cell adhesion and is likely to contribute to the suggested intracellular signaling functions of Pcdh␥.
siRNA Transfection-The mammalian expression vector pSUPER, kindly provided by Dr. Brummelkamp, was used for expression of siRNA in neuronal H4 cells. The sequence of the human ADAM10 siRNA was as follows: 5Ј-GACAUUUCAAC-CUACGAAU-3Ј. The sequence was separated by a 9-nucleotide non-complementary spacer (tctcttgaa) from the corresponding reverse complement of the same 19-nucleotide sequence. These sequences were inserted into the pSUPER backbone after digestion with BglII and HindIII. H4 cells were transfected with pSuper-ADAM10-siRNA vector with the use of FuGENE 6 (Roche Applied Science) according to manufacturer's recommendations.
SDS-PAGE and Western Blotting-Cells were lysed in Triton lysis buffer (1% Triton X-100, 5 mM Tris, 1 mM EGTA, 250 mM saccharose, pH 7.4). Protein concentrations were determined with a BCA protein assay (Pierce). Equal amounts of protein were loaded on 10% SDS gels for analysis of Pcdh␥ C3 stubs. The samples were electrotransferred onto polyvinylidene difluoride membranes (Hybond-P; GE Healthcare, Freiburg, Germany), which were blocked overnight with 5% skim milk in Tris-buffered saline (TBS). After incubation with the indicated antibody in blocking buffer, membranes were washed three times in TBS containing 0.1% Tween (TBS-T). Primary antibodies were detected using affinity-purified peroxidase-conjugated secondary antibodies for 1 h at room temperature. Detection was carried out using the ECL detection system (Amersham Biosciences). Signals were recorded by a luminescent image analyzer (Image Reader LAS1000, Fujifilm, Tokyo, Japan) and analyzed with the image analyzer software (Gel-ProAnalyser, Media Cybernetics, Silver Spring, MD). For reprobing blots, polyvinylidene difluoride membranes were incubated in stripping buffer (100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris-HCl, pH 6.7) for 1 h at 60°C with occasional shaking. After three washes in large volumes of TBS-T, membranes were blocked in blocking buffer for 1 h at room temperature, and immunodetection was repeated.
Cell Aggregation Assays-For cell aggregation assays either wild-type K562 cells or K562 cells stably expressing Pcdh␥ C3 were used as described previously (5). Briefly, single cell suspensions were produced by two passages through yellow  AUGUST 4, 2006 • VOLUME 281 • NUMBER 31

JOURNAL OF BIOLOGICAL CHEMISTRY 21737
pipette tips (10 l, Sarstedt). Viability of cells was determined by trypan blue dye exclusion and shown to be Ͼ95% in every assay. One million cells of each transfectant clone or wild-type cells were incubated in 3 ml of DMEM medium containing 10% fetal bovine serum and penicillin/streptomycin at 37°C. For analysis under constitutive conditions, cells were incubated in DMEM with GM6001 (10 M) or Me 2 SO (Roth). For stim-ulation, metalloprotease inhibitor GM6001 or Me 2 SO were added. After 16 h, the cells were incubated with PMA (200 ng/ml) for a further 4 h. Afterward, cells were photographed using an inverted phasecontrast microscope (Carl Zeiss) and counted manually for both single cells and number of cells in aggregates. Average standard deviations of one representative experiment were calculated.
Statistical Analysis-Values are expressed as S.D. values or means Ϯ S.E. The standard error values indicate the variation between mean values obtained from at least three independent experiments. The assumptions for normality (Kolmogorov-Smirnov test) and equal variance (Levene median test) were verified with the SigmaStat 3.1 software (SYSSTAT, Erkrath, Germany). The analysis of variance were performed with oneor two-way analysis of variance. Multiple parametric statistical comparisons between experimental groups versus a control group were accomplished with DunnettЈs method. All pairwise multiple comparison procedures were performed with Tukey's test. p values Ͻ0.05 were classified as statistically significant.
both known to inhibit a wide spectrum of metalloproteases, led to a significant reduction of the 25-kDa CTF (Fig. 1, B and C, left  panel).
Previously, we have described two hydroxamate-based compounds that differ in their capacity to block the activities of the two disintegrin-like metalloproteases, ADAM17 and ADAM10 (28). The inhibitor GW280264X has been shown to preferentially block ADAM17 and to a lesser extent ADAM10, while the compound GI254023X preferentially blocked ADAM10. Western blot analysis of cell lysates of inhibitor-treated wild-type MEFs demonstrated that the ADAM10 inhibitor GI254023X caused a reduction in Pcdh␥ C3-CTF1 levels in a dose-dependent manner, which was not observed with GW280264X (Fig.  1B, middle and right panels). The effect of GI254023X was comparable with that of the broad-spectrum metalloprotease inhibitor TAPI, indicating that the majority of metalloprotease-mediated Pcdh␥ C3 proteolysis in MEFs can be attributed to ADAM10. The densitometric analysis of three independent experiments suggested that ADAM10 may be responsible for the cleavage of Pcdh␥ C3 (Fig. 1C).
Loss of Function and Reconstitution Experiments Provide Evidence for the Specific Role of ADAM10 in Constitutive ␥-Protocadherin Shedding-To more precisely define whether ADAM10 is involved in Pcdh␥ C3 shedding, we used a panel of established MEFs that were generated from mouse embryos with a targeted deletion of different ADAM proteases as described previously (21,24,29). As judged by the presence of both the 25 kDa CTF1 (Fig. 2A, middle panel) and a soluble Pcdh␥ C3 NH 2 -terminal fragment, NTF, released into the medium of the cultured cells ( Fig. 2A, lower  panel), shedding of Pcdh␥ C3 appeared to be unaffected in wildtype, ADAM9 Ϫ/Ϫ , ADAM15 Ϫ/Ϫ , ADAM17 Ϫ/Ϫ , and BACE1 Ϫ/Ϫ MEFs. In contrast, shedding of Pcdh␥ C3 was almost completely abolished in ADAM10-deficient cells as confirmed by densitometric analysis of six independent experiments ( Fig. 2A, right panel).
To exclude heterogeneity in cell culture (21), we analyzed the essential role of ADAM10 for constitutive Pcdh␥ C3 shedding also in different independently derived ADAM10deficient cell-lines (Fig. 2B). We found that the Pcdh␥ C3-CTF generation was strongly diminished, although not completely absent in one of three tested ADAM10-deficient cell lines. These results support our inhibitor data and indicate that ADAM10 is the major protease responsible for the constitutive shedding of Pcdh␥ C3.
Accordingly, transient transfection of wild-type mouse ADAM10 rescued the Pcdh␥ C3 ectodomain proteolysis in ADAM10-deficient fibroblasts (Fig. 2C). Taken together, the disappearance of the shedding-dependent 25-kDa Pcdh␥ C3 fragment as well as the significantly reduced capacity to generate soluble Pcdh␥ C3 (Fig. 2, A-C) in ADAM-deficient cells strongly argues for a crucial role of ADAM10 as the major Pcdh␥ C3 sheddase.
To further analyze whether ADAM10 mediates shedding of Pcdh␥ family members in general, we used wild-type fibroblasts and ADAM10-deficient cells expressing different Pcdh␥ proteins fused COOH-terminally to GFP. Whereas cleavage of the Pcdh␥ C3-GFP protein was observed in wildtype fibroblasts, a clear reduction in the generation of the Pcdh␥ C3-GFP-CTF1 was noted in ADAM10-deficient cells (Fig. 2D) confirming our findings on the endogenous Pcdh␥. Analysis of Pcdh␥ B4-GFP, another Pcdh␥ family member, showed the same pattern: a strikingly reduced generation of COOH-terminal fragments in ADAM10-deficient cells compared with wild-type fibroblasts (Fig. 2E). This provides evidence for a general role of ADAM10 in mediating the extracellular cleavage of the different members of the Pcdh␥ family.
ADAM10 Is Also Crucial for Induced Shedding of ␥-Protocadherin-In general, shedding of surface membrane proteins can occur in a constitutive and in a regulated fashion (16,26,30). Therefore we set out to analyze in more detail the regulated shedding of Pcdh␥ C3 (Fig. 3A). Stimulation of protein kinase C using the phorbol ester PMA (31) strongly induced Pcdh␥ C3 shedding in wild-type fibroblasts. Staurosporine, which  AUGUST 4, 2006 • VOLUME 281 • NUMBER 31

JOURNAL OF BIOLOGICAL CHEMISTRY 21739
induces apoptosis and has been previously implicated in the activation of metalloprotease-mediated cleavage of E-cadherin (17) and N-cadherin (16), also stimulated Pcdh␥ C3 shedding. Also ionomycin, an agent that promotes shedding of cadherins through stimulation of calcium influx (16), increased CTF1 production in wild-type cells. Depletion of cholesterol using cyclodextrin (MCD) stimulated the shedding of Pcdh␥ C3 only slightly (Fig. 3A, left panel).
This clearly shows that Pcdh␥ C3 cleavage is controlled by several signaling pathways, which include protein kinase C-and calcium-dependent regulation. We next addressed whether this inducible cleavage is dependent on ADAM10 by analyzing the effects of these different compounds in ADAM10-deficient cells (Fig. 3A, middle panel). If another protease (e.g. TACE) was responsible for the induction of Pcdh␥ shedding, we expected to observe CTF1 generation in these cells, which was not the case as shown by analysis of three independent experiments (Fig. 3A). We therefore conclude that ADAM10 is also responsible for the stimulated shedding of Pcdh ␥C3.
It has been demonstrated that calcium influx is temporally coincident with synapse formation and leads to a redistribution of synaptic proteins during synapse maturation (32,33). Extracellular influx of calcium also induces rapid cleavage of classical cadherins (e.g. E-cadherin and N-cadherin) (16,17). Therefore we investigated this aspect in more detail. In wild-type cells, ionomycin induced an accumulation of the CTF1 within minutes (Fig. 3B, left  panel). This effect was completely blocked with the ADAM10 inhibitor GI254023X in wild-type cells (Fig. 3B, middle panel) and also absent in ADAM10-deficient cells (Fig. 3B, right panel). Taken together, these results demonstrate that ADAM10 is the major protease responsible for the constitutive and for the inducible Pcdh␥ C3 shedding.
Upon ADAM10 overexpression in these cells (Fig. 4A, upper panel) an increased level of Pcdh␥ C3-CTF1 was observed (Fig. 4A, lower panel). We confirmed the role of endogenously expressed ADAM10 in Pcdh␥ C3 shedding in neuronal cells using vector based RNA interference (36). ADAM10 levels dropped to 15%, and Pcdh␥ C3-CTF1 levels dropped to 35%, as compared with the levels seen in mock transfected cells (Fig.  4B). To clarify whether PMA stimulates ADAM10-mediated shedding also in H4 cells, we analyzed the effect of PMA treatment on Pcdh␥ C3 cleavage in the presence or absence of the specific inhibitors GI254023X and GW280264X in these cells. PMA stimulation increased the amount of Pcdh␥ C3-CTF1 generation (Fig. 4C). By contrast, no increase in the amount of Pcdh␥ C3-CTF was seen when the ADAM10 inhibitor GI254023X was added to the cells prior to PMA stimulation. The treatment with the ADAM10/ADAM17 inhibitor GW280264X showed only a minor effect. These results were confirmed by statistical analysis of three different experiments (Fig. 4C, right panel) suggesting that ADAM10 is also involved in the PMA-stimulated cleavage of Pcdh␥ C3 in neuronal cells.
ADAM10-mediated ␥-Protocadherin C3 Shedding Cannot Be Compensated in Vivo-To further elucidate the in vivo relevance of Pcdh␥ C3 cleavage by ADAM10 we analyzed extracts of E9.5 wild-type and ADAM10-deficient embryos by Western blotting. It has been shown recently that the processing of Pcdh␥ involves a presenilin-dependent downstream processing of Pcdh␥ C3-CTF1, resulting in a second COOH-terminal cleavage product, termed Pcdh␥ C3-CTF2 (12). In addition to the Pcdh␥ C3 full-length protein, we found both fragments, CTF1 and a smaller fragment, most likely corresponding to CTF2 in the wild-type embryos. The generation of both Pcdh␥ C3 fragments was almost, if not completely, abolished in the ADAM10-deficient embryos even though the full-length protein was expressed in equal amounts (Fig. 5). These findings indicate that the ADAM10-dependent formation of Pcdh␥ C3-CTF1 is a prerequisite for further processing by the ␥-secretase complex. Thus, these results show that ADAM10 is essentially involved in Pcdh␥ C3 processing and signaling in vivo.
Glutamate Stimulation Induces ADAM10-mediated ␥-Protocadherin C3 and N-cadherin Shedding in Neuronal Cells-Dynamic regulation of synaptic efficacy is thought to play a critical role in the formation of neuronal circuitry during development and learning (37). To evaluate whether stimulation of neural activity is able to induce ADAM10-mediated processing of Pcdh␥ C3, we incubated H4 cells with KCl and glutamate to mimic depolarization and glutamate receptor stimulation, respectively. Application of both stimuli in H4 cells resulted in an increased generation of Pcdh␥ CTF fragments (Fig. 6A). Recently, we have shown that ADAM10-mediated N-cadherin shedding is also stimulated by the activation of NMDA-type glutamate receptors (16). In particular, the application of NMDA to H4 cells led to an increased shedding of N-cadherin (Ref. 16 and Fig. 6A, right panel). In contrast NMDA stimulation did not increase the ADAM10-mediated shedding of Pcdh␥ C3 in H4 cells leading to the conclusion that the glutamate-activated cleavage of Pcdh␥ C3 is independent of NMDAtype glutamate receptor activation.
To more precisely define the pathway of glutamate-dependent induction of Pcdh␥ C3 shedding in H4 cells and primary neurons, we compared the effect of glutamate in the presence or absence of the selective AMPA (␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid)/kainate receptor antagonist CNQX, which blocks non-NMDA-type glutamate receptors. Interestingly, the increase of Pcdh␥ C3 proteolysis after stimulation with glutamate or KCl was strongly diminished in the presence of CNQX (Fig. 6B,  left panel). Thus, we conclude that the glutamate-induced response is mediated by non-NMDA/AMPA-type receptors. These findings were confirmed in primary neurons, where the application of AMPA, but not of NMDA, led to an increased processing of Pcdh␥ C3 (Fig. 6B, right panel).
To further elucidate the functional relevance of metalloprotease-mediated Pcdh␥ C3 shedding for cell-cell aggregation, we used GM6001 as a broad-spectrum metalloprotease inhibitor and the shedding-inducing agent PMA. Cell aggregation of stably transfected K562 cells was strongly increased by the inhibitor GM6001 (Fig. 7F), whereas GM6001 treatment showed only a slight effect on K562 cell aggregation in untransfected cells (Fig. 7B). To examine whether increased metalloprotease-mediated shedding, evoked by PMA stimulation, leads to an anti-adhesive effect, we analyzed cell aggregation after a 4-h presence of PMA. PMA decreased cell adhesion by 20% compared with Me 2 SO-treated Pcdh␥ C3 cells, while minor changes were seen in wild-type K562 cells (Fig. 7, C and G). Pretreatment with GM6001 prior to PMA stimulation abolished the PMA-induced effect in Pcdh␥ C3 K562 cells (Fig. 7, D and H). Taken together, these findings suggest that metalloprotease-mediated shedding of Pcdh␥ C3 is involved in regulating Pcdh␥ C3-dependent cell-cell aggregation.

DISCUSSION
Pcdh␥ are widely expressed during the development of the central nervous system. Roles in cell-cell adhesion and signaling have been attributed to these cadherin-like molecules (3,34). Important contributions of Pcdh␥ proteins to the proper development of the central nervous system are apparent by the strong neurological phenotype and neonatal death of mice lacking a functional Pcdh␥ cluster (11), but details of the involved functional pathways are largely unknown.
Pcdh␥ are endoproteolytically cleaved by a metalloprotease to generate a membrane-bound CTF and a soluble NTF, which is released to the extracellular environment. The metalloprotease cleavage is a prerequisite for subsequent intramembrane preseni-lin-mediated processing and release of a smaller CTF with nuclear localization and potential signaling functions (11,12). In this report we show that the metalloprotease ADAM10 is critically involved in the ectodomain shedding of Pcdh␥ proteins. We find that constitutive shedding of Pcdh␥ C3 is strongly reduced in ADAM10deficient fibroblasts, whereas the shedding is unchanged in fibroblast deficient for other members of the ADAM family. Accordingly, the ADAM10 inhibitor compound GI2540223X is able to block constitutive shedding of Pcdh␥ C3, yielding similar or even stronger effects than the broad-spectrum metalloprotease inhibitors TAPI and GM6001, which we and others have used previously (11,12). In line with these findings ectodomain cleavage of Pcdh␥ C3 is also absent in ADAM10 knockout embryos. The Pcdh␥ cluster encodes 22 different family members, and a first analysis of several family members indicated that Pcdh␥ in general is a substrate for regulated proteolytic processing. This would allow the propagation of different types of intracellular signals initiated by ectodomain shedding (12).
ADAM10 activity has been implicated in the constitutive shedding of several substrates. However, shedding induced upon stimulation may be attributed to different members of the ADAM family, e.g. to ADAM17 (22). For Pcdh␥, we demonstrate that both constitutive and inducible shedding are dependent on the presence of ADAM10. For example, calcium influx triggered by the application of ionomycin led to a strong stimulation of ADAM10-mediated Pcdh␥ C3 shedding. This finding is of particular interest in the context of synapse formation since modulation of intracellular calcium levels is important for synapse shaping and regulation of synaptic efficacy (42,43). Induced shedding of cadherin and protocadherin ectodomains may provide an elegant mechanism to combine the local control of synaptic adhesion with intracellular signaling as a result of downstream processing of the membrane-bound CTFs via the ␥-secretase complex. Regulated processing of E-and N-cadherin leads to a cascade of distinct signaling events in the cytosol (16,17). In contrast, processing of Pcdh␥ C3 by the ␥-secretase generates a CTF2, which is analogous to the situation described for Notch-ICD and localizes to the nucleus (11,12). Notch controls transcriptional activation or repression of its direct target genes by binding to downstream transcription factors (44). Our analysis of

ADAM10 Affects ␥-Protocadherin Shedding
ADAM10-deficient embryo extracts demonstrate that the ADAM10-dependent cleavage of Pcdh␥ is a prerequisite, thus a rate-limiting step for intramembrane proteolysis of Pcdh␥-C3 also in vivo and suggests an important function of ADAM10 in the regulation of Pcdh␥-mediated signaling. Further analysis of neuronal cell lines and primary neurons supported this view since we found a specific activation of Pcdh␥ C3 shedding upon activation of non-NMDA-type glutamate receptors by the agonist AMPA. In contrast, activation of NMDA-type glutamate receptors had no effect on Pcdh␥ C3 shedding but stimulated N-cadherin shedding as we have also reported previously (16). These findings indicate that regulation of the interactions of ADAM10 with its substrates, e.g. Pcdh␥ C3 or N-cadherin, is able to specifically determine downstream signaling.
As a result of ADAM10-mediated cleavage accumulation of soluble Pcdh␥-NTF in the culture medium is observed (Ref. 12 and Fig. 2A). Regulated processing and nuclear translocation of Pcdh␥-CTF promoted the idea that Pcdh␥ proteins may function as cell surface receptors; however, bi-directional signaling mechanisms, which involve Pcdh␥-NTF, could be envisioned as well. In particular, Pcdh␥-NTFs may engage in heterophilic binding to neuronal receptor molecules, thus acting as ligands in a paracrine or autocrine fashion, similar to the situation shown for other ADAM10 substrates, e.g. the various ligands of the epidermal growth factor receptor (22). In addition, modulation of cell motility or cell adhesion may result from (homophilic) binding of soluble Pcdh␥-NTF to membrane-tethered Pcdh␥ complexes, as we demonstrated previously for E-cadherin-NTF generated by ADAM10 activity (17).
Classical cadherins were initially identified as the molecules that mediate cell-cell adhesion, and knock-out mice experiments have proven the role of classical cadherins in embryogenesis (45). They provide strong and usually homophilic cell-cell adhesion, which is important for tissue formation during embryogenesis or cell differentiation, e.g. in the nervous system (8). Similarily, protocadherins, including the Pcdh␥ family, were shown to mediate cell-cell adhesion, which is, however, much weaker than that of classical cadherins (3,5). Metalloprotease-mediated ectodomain shedding may limit the amount of adhesive moieties present on the cell surface for a variety of cell adhesion molecules (16,17,26) and thus may also regulate Pcdh␥-mediated cell adhesion. Our experiments in a lymphoid cell line, which was used previously to test cadherin and protocadherin adhesion (5,41), confirmed this idea. Pcdh␥ C3-mediated cellcell adhesion was increased upon metalloprotease inhibition and decreased under conditions inducing ADAM10 ectodomain shedding.
In conclusion, our results demonstrate that ADAM10 is critical for constitutive as well as for induced ectodomain shedding of Pcdh␥. This cleavage step is required for the regulated processing of Pcdh␥ by the ␥-secretase complex but negatively regulates Pcdh␥-mediated adhesion. Our finding that ectodomain shedding of Pcdh␥ and N-cadherin by ADAM10 is induced by different effector mechanisms strongly supports the view that ADAM10 is crucially involved in the regulation of Pcdh␥-me- diated signaling and cell adhesion. This opens novel perspectives to study Pcdh␥ functions in tissue formation, neuronal differentiation, and synaptogenesis.