|Home | About | Journals | Submit | Contact Us | Français|
The immunoglobulin superfamily recognition molecule L1 plays important functional roles in the developing and adult nervous system. Metalloprotease-mediated cleavage of this adhesion molecule has been shown to stimulate cellular migration and neurite outgrowth. We demonstrate here that L1 cleavage is mediated by two distinct members of the disintegrin and metalloprotease family, ADAM10 and ADAM17. This cleavage is differently regulated and leads to the generation of a membrane bound C-terminal fragment, which is further processed through γ-secretase activity. Pharmacological approaches with two hydroxamate-based inhibitors with different preferences in blocking ADAM10 and ADAM17, as well as loss of function and gain of function studies in murine embryonic fibroblasts, showed that constitutive shedding of L1 is mediated by ADAM10 while phorbol ester stimulation or cholesterol depletion led to ADAM17-mediated L1 cleavage. In contrast, N-methyl-d-aspartate treatment of primary neurons stimulated ADAM10-mediated L1 shedding. Both proteases were able to affect L1-mediated adhesion and haptotactic migration of neuronal cells. In particular, both proteases were involved in L1-dependent neurite outgrowth of cerebellar neurons. Thus, our data identify ADAM10 and ADAM17 as differentially regulated L1 membrane sheddases, both critically affecting the physiological functions of this adhesion protein.
The neural cell adhesion molecule L1 (NCAM-L1) is a 200- to 220-kDa type I membrane glycoprotein, which consists of six immunoglobulin-like domains, followed by five fibronectin-like repeats (type III), a transmembrane domain, and a small conserved cytoplasmic tail (16, 49). L1 is critically involved in the development of the nervous system promoting neuronal migration, neuronal survival, neurite outgrowth, and myelination, as well as axon guidance, fasciculation, and regeneration (8, 25, 30, 60). The importance of L1 function is also illustrated by mutations in the human L1 gene, resulting in CRASH syndrome (corpus callosum agenesis, retardation, adducted thumbs, shuffling gait, and hydrocephalus) (19).
L1 mediates homophilic cell-cell interaction but also binds heterophilic ligands like axonin-1 (39), CD24 (36), and several integrins (35). It has been shown previously that L1 can undergo membrane-proximal cleavage, leading to the release of the soluble extracellular domain and the generation of a membrane-bound stub (2, 22, 23, 37, 48). The soluble extracellular domain remains intact and has been suggested to serve as a substrate for integrin-mediated cell adhesion, thereby stimulating cellular motility and cell migration (48). In the nervous system, soluble L1 released from cultured neurons promotes neurite outgrowth and influences neuronal differentiation (12, 37, 68).
Recently, ADAM10, a family member of the disintegrin and metalloprotease family (ADAM) has been implicated in the release of soluble L1 in human tumor cell lines (22, 23, 48). ADAMs are involved in the constitutive and stimulated ectodomain shedding of various membrane-bound proteins (6, 63, 65). ADAM10 has been shown to play important roles in development and in the central nervous system by influencing Notch/Delta activation and N-cadherin-dependent β-catenin signaling (24, 51, 54, 55, 58). ADAM10 has high sequence similarity with the tumor necrosis factor alpha (TNF-α)-converting enzyme (TACE/ADAM17), which has been implicated in the ectodomain shedding of several substrates like TNF-α, TNF-α receptors I and II, transforming growth factor-α, interleukin 6 receptor, and fractalkine (4, 31, 59).
Recently, it has been shown that the proteolytic release of the extracellular domain of transmembrane proteins can be followed by intramembrane γ-secretase-mediated cleavage, leading to the release of the cytoplasmic domain into the cytosol. This process, named regulated intramembrane proteolysis (RIP) (7), is discussed as a novel mechanism for signal transduction (71) and has been implicated in the processing of an increasing number of proteins like Notch (66), the amyloid precursor protein (APP) (20), and N-cadherin (44, 55).
In this study, we have analyzed the proteolytic processing of the L1 adhesion molecule in detail. We provide multiple lines of evidence that two independent and differentially regulated members of the ADAM family, ADAM10 and ADAM17, are critically involved in the release of the extracellular domain of L1. We demonstrate that the remaining C-terminal fragment is further processed through γ-secretase activity, suggesting the possibility that L1 may also contribute to intracellular signal transduction pathways. Finally, we show that ADAM10 and ADAM17 play an important functional role by regulating L1-dependent neuronal cell adhesion, cell migration, and neurite outgrowth.
The monoclonal antibody (MAbs) against the cytoplasmic domain of L1 (NCAM-L1) was purchased from Covance Research Products, Inc. (HiSS Diagnostics GmbH, Freiburg, Germany). The antibodies MAb 324 and MAb 555 against the ectodomain of the mouse L1 adhesion molecule were described before (2, 14, 34). ADAM10 was detected by using a polyclonal antiserum B42.1, described previously (24). ADAM17 antibody was from Chemicon International, Inc. (Temecula, CA). Microtubule-associated protein-2 antibodies were from Sigma (Deisenhofen, Germany). The recombinant human NCAM L1/Fc chimera and the recombinant mouse ADAM10 were obtained from R&D Systems (Wiesbaden, Germany). Reagents were obtained as follows: phorbol-12-myristate-13-acetate (PMA), EDTA, and N-methyl-d-aspartate (NMDA) were from Sigma; methyl-β-cyclodextrin (MCD) was from Research Biochemicals International (Natick, MA); and GM6001 and γ-secretase inhibitor L-685,458 were from Calbiochem (Darmstadt, Germany). Hydroxamate-based inhibitors GW280264X and GI254023X were described previously (31). Complete EDTA-free protease inhibitor mixture was obtained from Roche Molecular Biochemicals (Mannheim, Germany).
Simian virus large T-antigen-immortalized ADAM10−/− mouse embryonic fibroblasts (MEFs) and cell lines derived from presenilin 1/2−/− (PS1/2−/−), ADAM9−/−, ADAM15−/−, and respective wild-type (WT) animals were generated and characterized as described previously (24, 28, 29, 53, 59, 73). TACE−/− EC2 cells and TACE−/− EC2 cells stably retransfected with full-length TACE were described previously (5, 47). Human neuronal H4 cells were kindly provided by J. Wilfang (Erlangen, Germany). All cells were grown in Dulbecco's modified essential medium (DMEM; PAA Laboratories, Cölbe, Germany) supplemented with 10% fetal calf serum (FCS) and 1% penicillin-1% streptomycin.
cDNA-encoding mouse ADAM10 was cloned into the XhoI/BamHI site of the expression vector pcDNA3.1 (Invitrogen, Karlsruhe, Germany) and confirmed by DNA sequencing. Cells were transfected in six-well tissue culture plates (Sarstedt, Nümbrecht, Germany) with FuGENE 6 (Roche) according to the manufacturers' instructions and analyzed 48 h after transfection.
For analysis of constitutive L1 shedding, cells were washed with phosphate-buffered saline (PBS) and incubated in DMEM with GI254023X (5 μM), GW280264X (5 μM), EDTA (5 mM), or dimethylsulfoxide (DMSO; Roth, Karlsruhe, Germany) for 3 h. For stimulation, FCS-free medium containing metalloprotease inhibitors or DMSO was added. After 30 min, the cells were stimulated with PMA (200 ng/ml) for 3 h or with MCD (10 mM) for 60 min. Afterward, the conditioned media were harvested, and a protease inhibitor cocktail (Complete; Roche) was added. Supernatants were concentrated (30 fold) through centrifugation in Microcon Centrifugal filter devices (YM-10; Millipore, Bedford, MA). Cells were washed with 2 ml PBS and harvested in 1 ml PBS. Cell pellets were lysed in 100 μl radioimmunoprecipitation assay buffer (50 mM Tris-HCl [pH 8], 150 mM NaCl, 0.1% sodium dodecyl sulfate [SDS], 1% NP-40, 0.5% sodium deoxycholate) containing a protease inhibitor cocktail (Complete; Roche).
The mammalian expression vector pSUPER, kindly provided by T. R. Brummelkamp, was used for expression of short hairpin RNA in H4 cells. Since short hairpin RNAs are processed into small interfering RNA (siRNA)-like molecules in the cell, this method is referred to as ADAM10 siRNA in the following text. The sequence of the human ADAM10 siRNA was 5′-GACAUUUCAACCUACGAAU-3′. The sequence of the human ADAM17 siRNA was 5′-CACAUGUAGAAACACUACU-3′. Both sequences were separated by a 9-nucleotide noncomplementary 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.
A 96-well plate (Microlon; Greiner, Frickenhausen, Germany) was coated overnight with 4 μg/ml mouse anti-L1 (MAb 324) in 50 mM Na2CO3 (pH 9.3), subsequently washed three times with PBS with 0.05% Tween (PBS-T), and blocked with PBS-T containing 2% bovine serum albumin (BSA) for 2 h. Samples (50 μl per well) were added, and the plate was incubated at room temperature (RT) for 2 h. Following washing, 100-ng/ml biotinylated anti-L1 MAb 555 in PBS-T containing 1% BSA was added to each well, and the plate was incubated at RT for 1 h. After being washed, 100-mU/ml streptavidin-peroxidase (POD) conjugate (Roche) in PBS-T containing 1% BSA was added, followed by 1 h of incubation at RT. After being washed, chromogenic POD substrate (BM Blue; Roche) was added. The reaction was stopped after 20 min of incubation at RT by the addition of 1.8 M H2SO4 before the optical density was determined at 450 nm.
Same amounts of protein were loaded on 10% SDS-polyacrylamide gel electrophoresis gels. The samples were electrotransferred onto polyvinylidene difluoride membranes (Hybond-P; Amersham), 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. Primary antibodies were detected using affinity purified POD-conjugated secondary antibodies for 1 h at room temperature. Detection was carried out using the ECL detection system (Amersham, Freiburg, Germany). Signals were recorded by a luminescent image analyzer (image reader LAS1000; Fujifilm, Tokyo, Japan) and analyzed with image analyzer software (Gel-ProAnalyser; Media Cybernetics). 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 containing 0.1% Tween, membranes were blocked in blocking buffer for 1 h at room temperature, and immunodetection was repeated.
Brains from adult wild-type mice were homogenized in 1 mM NaHCO3, 0.2 mM CaCl2, 0.2 mM MgCl2, and 1 mM spermidine (pH 7.9) and centrifuged at 600 × g for 15 min at 4°C. After recentrifugation of the supernatant at 25,000 × g for 45 min at 4°C, the crude membrane fraction of one mouse brain was resuspended in DMEM (PAA Laboratories) and subdivided into four 20-μl aliquots. These aliquots were incubated at 37°C for different times with recombinant mouse ADAM10 (600 ng/aliquot; R&D). As a control, aliquots were incubated without recombinant mouse ADAM10 for 60 min. Samples were resuspended in sample buffer and loaded on 10% SDS-polyacrylamide gel electrophoresis gels.
Microexplant cultures from mouse cerebella were prepared basically as described previously (18). Briefly, cerebella were taken from 6-day-old wild-type mice and transferred to ice-cold PBS. The tissue was freed from meninges, choroid plexus, and blood vessels and forced through a Nitrex net with a pore width of 100 μm. The small tissue pieces were washed twice with Hanks' balanced salt solution, followed by a washing step with culture medium (minimum Eagle's medium containing 10% horse serum, 10% FCS, 6 mM glucose, 200 μM l-glutamine, 50 units/ml penicillin, 50 μg/ml streptomycin, 10 μg/ml human transferrin, 10 μg/ml insulin, and 10 ng/ml selenium [Sigma]). The tissue pieces were collected by sedimentation. A total of 30 to 40 explants were plated onto poly-l-lysine (PLL; Sigma-Aldrich)-coated glass coverslips (15-mm diameter) or coverslips additionally coated with 2 μg/ml recombinant human NCAM L1/Fc chimera (R&D). Coverslips were rinsed three times with distilled water before the explants were plated. At 16 h after plating, 1 ml of culture medium lacking FCS was added to the explants containing the metalloprotease inhibitor GW280264X or GI254023X at 5 μM. As control, explants were incubated with DMSO only in parallel. After an incubation time of 24 h, the explants were fixed with 2% glutaraldehyde and 2% paraformaldehyde and stained with Coomassie brilliant blue G-250 solution (Serva, Heidelberg, Germany). The effect of the different protease inhibitors on neurite outgrowth was analyzed and quantitated by measuring the length of the 10 longest neurites of 10 aggregates in three independent experiments (37) with Axio Vision Release 4.2 software (Kontron, Zeiss, Germany).
Cerebella were taken from 6-day-old wild-type mice and transferred to ice-cold PBS. The tissue was freed from meninges, choroid plexus, and blood vessels. Sagittal sections of the cerebella were cut on a high-frequency vibratome (Bio-Rad, München, Germany) at 400 μm and collected in a petri dish with fresh buffer. The EGL was selectively dissected from the slices with fine spatulas under a stereomicroscope and collected in separate vials before dissociation (56). Tissues were dissociated mechanically with Pasteur pipettes and enzymatically by limited trypsin digestion. Cells were seeded on PLL-coated six-well dishes (Sarstedt). For proper identification of neurons, cells were stained with anti-microtubule-associated protein-2 in parallel assays (data not shown).
For analyzing cell substrate adhesion, 96-well plates (Sarstedt) were coated with 2 μg/ml recombinant human NCAM L1/Fc chimera (R&D) or 2 μg/ml BSA in 100 μl PBS at 4°C overnight. Fluorescently labeled cells were seeded at a density of 5 × 104 cells/well and incubated for 20 min. For fluorescent labeling, cells were suspended at 2 × 106/ml in PBS-0.1% BSA and incubated with 2.5 μM fluorescent dye (calcein AM; Molecular Probes, Leiden, The Netherlands) at 37°C for 30 min. Excess dye was removed by washing cells twice with 15 ml PBS. Afterwards the cells were resuspended in growth medium and preincubated for 30 min with or without metalloprotease inhibitor GW280264X or GI254023X (5 μM) or vehicle control (DMSO). After incubation for 60 min with or without PMA (200 ng/ml), the cells were rapidly seeded onto the plates by centrifugation for 1 min at 100 × g. The plate was incubated at 37°C for 20 min and then repeatedly washed with 200 μl PBS per well. The fluorescence signal from the adherent cells was measured before and after washing with a fluorescence plate reader (Lambda Fluoro 230; MWG Biotech, München, Germany) at an excitation wavelength of 480 nm and an emission wavelength of 530 nm. The differences in fluorescence before and after washing are depicted as percentages of adherent cells.
For the haptotactic cell migration assay, recombinant human NCAM L1/Fc chimera (2 μg/ml) or BSA (2 μg/ml) for a control was coated for 90 min to the back sides of transwell chambers (6.5-mm diameter, 5-μm pore size; Costar). A total of 3 × 105 dissociated EGL cells from 6-day-old mice were seeded in growth medium into the upper chamber and were allowed to transmigrate to the lower compartment for 48 h at 37°C. To quantitate transmigrated cells, the inner chamber was removed and cleaned with a cotton swab to remove the nonmigrated cells. Migrated cells adherent to the back side of the membrane were stained with crystal violet solution. The membranes were extensively washed in distilled water, and the remaining stain was eluted with 10% acidic acid. The eluted dye was measured at 595 nm with an ELISA reader (Sunrise; Tecan, Crailsheim, Germany).
All values are expressed as means ± standard error of the mean (SEM). 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 SigmaStat 3.1 software (Erkrath, SYSSTAT, Germany). The analysis of variance was performed with one- or two-way analysis of variance. Multiple parametric statistical comparisons between experimental groups versus a control group were accomplished by Dunnett's method. All pairwise multiple comparison procedures were performed with Tukey's test. Nonparametric statistics were made using the Mann-Whitney rank sum test. P values of <0.05 were classified as statistically significant.
The extracellular domain of the full-length 220-kDa type I protein L1 can be released through membrane-proximal cleavage, leading to the generation of a 32-kDa C-terminal fragment 1 (CTF1) and a soluble N-terminal fragment of 200 kDa (NTF) (Fig. (Fig.1A).1A). To assess the involvement of the most likely candidate sheddase ADAM10 and its close homologue ADAM17, we performed inhibitor studies with two hydroxamate-based compounds which differ in their capacity to block the activities of these two proteases (1, 31). The inhibitor GW280264X has been shown to preferentially block ADAM17 and, to a lesser extent, ADAM10, while the compound GI254023X blocked ADAM10 with >100-fold-more potency than ADAM17. Western blot analysis of supernatants of inhibitor-treated WT MEFs demonstrated that the release of L1-NTF could be strongly diminished by the ADAM10 inhibitor GI254023X. In contrast, treatment with the inhibitor GW280264X affected L1 release only slightly (Fig. (Fig.1B).1B). The reduction with GI254023X was comparable to the effect of the broad-spectrum metalloprotease inhibitor EDTA, indicating that the majority of metalloprotease released L1 in MEFs can be attributed to ADAM10. To normalize variations in transfection efficiency, the cell pellet was also analyzed for L1 expression (Fig. (Fig.1B,1B, bottom), and shedding was calculated as the percentage of soluble L1 in relation to total L1 (full-length protein and L1 fragment). The densitometric analysis of three independent experiments confirmed that ADAM10 contributes to the constitutive shedding of L1 (Fig. (Fig.1C1C).
To further characterize the role of different ADAMs, we examined L1 shedding in a genetically defined system. We compared L1 release in a panel of different L1-transfected MEFs deficient for either ADAM9, ADAM10, ADAM15, or ADAM17 with wild-type cells. Western blot analysis with a C-terminal L1 antibody revealed that the generation of L1-CTF1 was unaffected in ADAM9−/−, ADAM15−/−, and ADAM17−/− fibroblasts and comparable with WT MEFs. In contrast, ADAM10−/− cells showed a strikingly reduced generation of this C-terminal fragment (Fig. (Fig.2A).2A). This was also confirmed by densitometric analysis of several independent experiments, also taking into account the different transfection efficiencies (Fig. (Fig.2A,2A, right). Notably, the generation of the CTF1 correlated well with the release of the soluble NTF in the cell culture medium in all analyzed cell lines (Fig. (Fig.2A,2A, bottom).
To exclude heterogeneity in cell culture (24), we also confirmed the essential role of ADAM10 for constitutive L1 shedding in different independently derived ADAM10−/− cell lines (Fig. (Fig.2B).2B). Moreover, we compared the release of soluble L1 in wild-type and ADAM10-deficient cells by ELISA (Fig. (Fig.2C).2C). Corresponding to the data obtained by Western blot analysis, the release of the extracellular domain of L1 in ADAM10-deficient cells was severely reduced compared to that of wild-type cells. L1 release could also be inhibited with the broad-spectrum metalloprotease inhibitor GM6001 in wild-type cells but not in ADAM10-deficient cells (Fig. (Fig.2C).2C). Finally, we confirmed by gain-of-function experiments that the defect in shedding in ADAM10−/− cells was indeed due to the lack of ADAM10. Transient retransfection of wild-type mouse ADAM10 rescued the constitutive L1 shedding in ADAM10-deficient fibroblasts (Fig. (Fig.2D2D).
To evaluate the in vivo relevance of ADAM10-mediated L1 shedding, we analyzed protein extracts from embryonic day 9 wild-type and ADAM10-deficient embryos. The absence of ADAM10 in the embryos was confirmed by Western blotting with an antiserum raised against murine ADAM10 (Fig. (Fig.2E,2E, bottom). When the same Western blot was reprobed with C-terminal L1 antibodies, we found that the L1-CTFs were diminished but not absent in the ADAM10 knockout embryos. These data indicate that even though ADAM10 is critically involved in constitutive shedding of L1 it is not the exclusive sheddase in vivo. Since the embryo represents a very complex system of different cell types and complex interactions, it is not surprising that L1 can also be cleaved by one or even more distinct proteases. Of note, we also found a C-terminal fragment of about 28 kDa, which we called L1-CTF2. Comparable to CTF1, this fragment was prominent in the WT embryos but reduced in the knockout embryos, indicating that CTF1 production is the rate-limiting step for further processing through another protease. In particular, this obvious sequence of proteolytic events suggested to us that L1 may also be processed by γ-secretase cleavage.
To identify the nature of the L1-CTF2-generating protease, we analyzed L1 shedding in the presence of the γ-secretase inhibitor L-685,458 in WT MEFs, ADAM10−/− MEFs, and fibroblasts deficient for the γ-secretase complex components presenilin 1 and 2 (PS1/2) (Fig. (Fig.3).3). The CTF2 fragment of L1 was less abundant in MEFs than in embryo extracts. After overexposure of the immunoblot, a faint band of the CTF2 could be visualized in WT MEFs, which was missing in ADAM10−/− MEFs (Fig. (Fig.3,3, bottom). The application of γ-secretase inhibitor increased the amount CTF1 and diminished CTF2 production in WT cells, indicating a role of γ-secretase in L1 processing. The absence of CTF2 and the accumulation of CTF1 in PS1/2−/− MEFs confirmed this assumption. Thus, these data provide evidence that L1 is a substrate for RIP and suggest the possibility that L1 cleavage might also influence intracellular signal transduction pathways.
L1 shedding has been shown to be a regulated process which could be stimulated by treatment with different stimuli like protein kinase C agonist PMA (23). To clarify whether this stimulation could also be attributed to ADAM10, we analyzed the effect of GI254023X and GW280264X on PMA-treated wild-type fibroblasts. PMA treatment increased the release of L1 into the supernatant (Fig. (Fig.4A).4A). Time-dependent analysis of PMA stimulation revealed that 2 h of incubation were sufficient to detect a considerable increase in L1 shedding, while a maximum was reached after 6 h (data not shown). The L1 release could be strongly diminished when cells were preincubated with the broad-spectrum metalloprotease inhibitor EDTA before stimulation. Since the inhibitor was applied before PMA stimulation and present during the time of PMA treatment, additional influence on the constitutive L1 shedding decreased the amount of soluble fragments even below the level of DMSO-treated control cells. In contrast to the constitutive conditions (Fig. (Fig.1B),1B), the ADAM17/ADAM10 inhibitor GW280264X had an effect similar to that of the broad-spectrum inhibitor, while the ADAM10 inhibitor GI254023X affected L1 release to a lesser extent. This also became apparent after densitometric quantification of three independent experiments (Fig. (Fig.4A,4A, right), suggesting that ADAM17 is involved in the PMA-stimulated cleavage of L1.
To clarify whether this effect was indeed due to ADAM17 activity, we analyzed PMA-stimulated shedding in the panel of different ADAM-deficient cells and ADAM17-deficient cells, which were stably retransfected with wild-type ADAM17. As shown in Fig. Fig.4B,4B, PMA increased the generation of the L1-NTF not only in wild-type cells but also in ADAM9−/−, ADAM10−/−, and ADAM15−/− fibroblasts. In contrast, PMA was not able to stimulate L1 release in ADAM17-deficent cells. Moreover, this effect was rescued in retransfected cells, clearly identifying ADAM17 as a major protease responsible for PMA-stimulated shedding in MEFs. The quantification of constitutively released soluble L1 (Fig. (Fig.4B,4B, lower left) and the increase of soluble L1 after PMA stimulation (Fig. (Fig.4B,4B, lower right) underlined the distinct roles of ADAM10 and ADAM17 for constitutive and PMA-stimulated shedding, respectively.
Cholesterol depletion has been shown to induce cleavage of L1 in breast tumor cells (22, 48). To examine if this stimulus would also induce ADAM17-mediated L1 shedding, ADAM10−/−, ADAM17−/−, and wild-type cells were treated with MCD or PMA. The release of soluble L1 into the cell culture medium was analyzed by ELISA. Both PMA and MCD induced L1 shedding in wild-type cells and also in ADAM10-deficient cells. In contrast, neither PMA nor MCD was able to increase the constitutive L1 release in ADAM17-deficient cells (Fig. (Fig.4C),4C), indicating that MCD stimulation also leads to ADAM17-mediated L1 shedding in fibroblasts.
Due to the early embryonic lethality of ADAM10-deficient mice (24), analysis of L1 processing was restricted to fibroblast cell lines. To further explore the role of ADAM10 in cells which express L1 endogenously, we used human neuroglioma H4 cells. To examine whether ADAM10 or ADAM17 is involved in L1 shedding, we knocked down the expression of these proteases by transfection of vector-based siRNA. By Western blot analysis, we found decreased expression of ADAM10 and ADAM17 after 48 h and almost undetectable levels after 72 h, compared with expression of mock-transfected control cells (Fig. (Fig.5A,5A, top). While ADAM17 suppression did not affect L1 processing, the reduced expression of ADAM10 correlated with strongly decreased L1-CTF1 generation (Fig. (Fig.5A,5A, bottom) in a densitometric analysis of three independent experiments. These data indicate that ADAM10 also represents the major sheddase responsible for constitutive processing of L1 in these cells.
To further explore the influence of ADAM10-mediated L1 shedding on the functional properties of neuroglioma cells, we performed a substrate adhesion assay. The adhesion of mock-treated H4 cells was significantly increased on L1 substrate compared to the BSA control (Fig. (Fig.5B,5B, left). While there was no effect from using BSA as a substrate (not shown), preincubation with GI254023X or GW280264X further increased cell adhesion on recombinant L1-Fc protein (Fig. (Fig.5B,5B, left). In particular, the ADAM10 inhibitor had the strongest effect and increased cell adhesion twofold in comparison to DMSO. Moreover, overexpression of ADAM10 resulted in decreased L1-dependent adhesion, while siRNA-mediated suppression of ADAM10 increased substrate adhesion (Fig. (Fig.5B,5B, middle). The respective effect on ADAM10 expression was additionally controlled by immunoblot analysis (Fig. (Fig.5B,5B, middle, insert). Thus, these data indicate that ADAM10-mediated constitutive cleavage of L1 is an important mechanism for regulating L1-dependent cell adhesion.
To examine whether ADAM17 would also play a role under nonconstitutive conditions, we analyzed cell adhesion with additional PMA stimulation before seeding. PMA-decreased cell adhesion by 50% compared to DMSO-treated cells (Fig. (Fig.5B,5B, right). This decrease could be abolished when cells were preincubated with GI254023X or GW280264X before stimulation. In contrast to the constitutive conditions, the ADAM17/ADAM10 inhibitor showed a stronger ability to enhance cell-substrate adhesion under PMA-stimulated conditions than the ADAM10 inhibitor. This finding indicates an influence of ADAM17 on L1-dependent adhesion after protein kinase C activation. Since both inhibitors were applied before PMA stimulation and during PMA treatment, parts of the increased cell adhesion needed to be attributed to an effect on constitutive ADAM10-mediated L1 shedding (Fig. (Fig.4A).4A). Taken together, these findings demonstrate that ADAM10- and ADAM17-mediated shedding of L1 is involved in regulating L1-dependent cell adhesion.
The neuronal cell adhesion molecule L1 is well known to exert important functions during development of the central nervous system (CNS) but also in the adult brain. To analyze the role of the predominantly neuronal expressed ADAM10 in more detail, we used an in vitro cleavage assay. For this assay, crude membrane fractions prepared from adult mouse brain were incubated with recombinant ADAM10 for different time periods at 37°C. Western blot analysis showed that incubation with ADAM10 led to a time-dependent increase in the generation of L1-CTF1 (Fig. (Fig.6A).6A). In the developing cerebellum, L1 is expressed only by postmitotic neurons (52). Therefore, we also analyzed the influence of ADAM10 on primary neurons of early postnatal mouse cerebellum. We found that expression of ADAM10 led to an increase in the generation of L1-CTF1, indicating that this protease is indeed capable of cleaving L1 in primary cells of the mouse cerebellum (Fig. (Fig.6B6B).
To further explore the functional significance of ADAM10- and ADAM17-mediated L1 shedding, we analyzed haptotactic migration of cerebellar granule neurons. For this study, we used a modified Boyden chamber assay in which cells migrate from top to bottom chambers through filters coated on the bottom side with BSA as a control or with L1-Fc protein. To determine the different effects of the proteases, cells were seeded in the absence or presence of GI254023X and GW280264X in transwell inserts. As shown in Fig. Fig.6C,6C, neurons displayed greater migration towards L1 substrate in 48 h than in their random migration towards BSA. This haptotactic migration was strongly inhibited by the ADAM10 inhibitor GI254023X, while the effect of the mixed ADAM17-ADAM10 inhibitor GW280264X was less strong. These findings indicate that ADAM10 is involved in regulating L1-dependent haptotactic migration of cerebellar neurons.
It has been suggested previously that the presynaptic protein L1 might play important roles in synaptic plasticity (42, 46). Cell adhesion molecules thought to be involved in regulating synaptic functions by modulating synaptic contacts in response to neural activity (3, 69). To evaluate if stimulation of neural activity would induce ADAM10- or ADAM17-mediated processing of L1, we incubated primary neurons of early postnatal mouse cerebellum with the glutamate receptor agonist NMDA in the absence or presence of GI254023X and GW280264X. As shown in Fig. Fig.6D,6D, NMDA stimulation increased the generation of L1-CTFs. The increase of L1 proteolysis was strongly diminished by the ADAM10 inhibitor GI254023X, while GW280264X only showed weak effects. These results suggest that ADAM10 is involved in the cleavage of L1 after stimulation with NMDA in cerebellar neurons. Of note, NMDA stimulation led to a decrease in the mature form of ADAM17, while there was an increase in the mature form of ADAM10, further supporting that NMDA stimulation leads to activation of ADAM10-mediated L1 shedding (Fig. (Fig.6E6E).
L1-induced neurite outgrowth is known to be mainly mediated by homophilic interaction dependent on the processing of L1 via metalloproteases (37). Therefore, we investigated the influence of ADAM10 and ADAM17 in mediating neurite outgrowth of cerebellar neurons. We cultured microexplants of early postnatal mouse cerebellum in the presence of GI254023X or GW280264X either on PLL or on L1-Fc/PLL protein-coated coverslips, respectively. As demonstrated in Fig. Fig.7,7, both inhibitors affected neurite length on PLL substrate only slightly, compared to the DMSO control which was set to 100% for quantification (Fig. 7A and B). In contrast, when microexplants were seeded on L1-coated surfaces, both inhibitors significantly inhibited neurite outgrowth. Additionally, the movement of small neuronal cell bodies from the explants was obviously decreased by both inhibitors applied (Fig. (Fig.7A).7A). After GW280264X treatment, the extension of neuronal processes from the explant core was almost absent in the vast majority of explants. These data indicate that ADAM10 (and even more, ADAM17) is involved in L1-dependent neurite outgrowth. Microexplants seeded on PLL and treated with the inhibitors in parallel were harvested and analyzed by Western blotting with C-terminal anti-L1 antibodies (Fig. (Fig.7C).7C). According to the neurite outgrowth assay, GW280264X had the strongest effect in reducing the generation of L1-CTF in the microexplants.
The neural cell adhesion molecule L1 is involved in the regulation of cell adhesion and cell migration. Therefore, posttranslational cleavage of L1 is important for regulating its function, but little is known about the proteolytic system responsible. Our present study demonstrates that two distinct metalloproteases, ADAM10 and ADAM17, are critically involved in controlling the release of soluble L1 by membrane-proximal cleavage, thereby regulating L1 function.
Initial evidence that ADAM10 is a candidate for constitutive L1 cleavage was obtained in inhibition studies with two hydroxamate-based compounds with different preferences in blocking ADAM10 and ADAM17. In accordance with other reports implicating the role of ADAM10 in shedding of L1 in tumor cells (21, 48), the ADAM10 inhibitor reduced constitutive L1 shedding in wild-type fibroblasts. We confirmed these findings by demonstrating that L1 shedding was nearly completely abolished in ADAM10-deficient fibroblasts. Thus, our data represent the first evidence in a genetically defined system that ADAM10 is responsible for the constitutive proteolysis of L1.
Analyzing the in vivo relevance of ADAM10 for L1 shedding, we found that L1 C-terminal fragments were clearly diminished in extracts from ADAM10-deficient embryos compared to the wild-type mice. This finding confirmed that ADAM10 is also of relevance for L1 proteolysis in vivo. However, in contrast to the almost-complete block of constitutive L1 shedding in ADAM10-deficient fibroblasts, which could not be compensated by other metalloproteases, L1-CTFs were still generated in the knockout embryos, demonstrating that ADAM10 is not the exclusive metalloprotease cleaving L1 in vivo. This is further supported by several findings indicating that L1 release seems to be a regulated process triggered by different extracellular signals (23, 27, 57). In particular, stimulation by phorbol esters or pervanadate, which activates cellular protein kinases, may regulate L1 cleavage (23). In agreement with these findings, our data demonstrate that L1 proteolysis can be induced through PMA stimulation in fibroblasts. Moreover, we identified ADAM17 as the protease responsible for this stimulated L1 proteolysis. Of note, PMA-stimulated L1 cleavage could not be compensated in ADAM17-deficient fibroblast by other proteases. The stimulation of ADAM17 activity by PMA is in agreement with numerous reports describing phorbol ester activation of ADAM17-mediated ectodomain shedding (13, 50). Interestingly, ADAM10 and ADAM17 have been reported to share other substrates as well. Fractalkine, the cellular prion protein, and APP are shed constitutively by ADAM10 and in a PMA-stimulated manner predominantly by ADAM17 (9, 31, 40, 72).
MCD treatment leading to the membrane extraction of cholesterol, has also been shown to activate L1 proteolysis (22). Cholesterol depletion has been implicated in the ADAM17-mediated shedding of the interleukin 6 receptor (47) but also in the ADAM10-mediated cleavage of N-cadherin (55). Our data suggest that the release of soluble L1 through cholesterol depletion is indeed ADAM17 dependent in fibroblasts. Taken together, our findings demonstrate that constitutive L1 shedding is mediated by ADAM10, while ADAM17 is responsible for PMA- but also MCD-stimulated shedding.
Interestingly, our data demonstrate for the first time that the metalloprotease-generated C-terminal fragment is a substrate for RIP. It is further processed through γ-secretase activity leading to the intracellular release of another L1 fragment (CTF2). This proteolytic sequence has been implicated in the processing of an increasing number of proteins like Notch (66) or APP (20), which are able to directly regulate signal transduction via translocation to the nucleus. For other proteins, like N-cadherin, intramembrane cleavage has been shown to indirectly influence cell signaling by enhancing the degradation of transcription factors (44). Therefore, our data suggest the interesting possibility that L1-CTF2 might also contribute to cellular signaling. It should be noted that L1-dependent gene regulation involving sustained extracellular signal-regulated kinase activation was recently reported (67). Further studies will have to clarify whether a link exists between L1-CTF2 cleavage and gene regulation.
Release of soluble L1 was observed in several cell lines of mouse and human origin (2), suggesting a conserved mechanism of metalloprotease-mediated L1 cleavage among different cell types and species. Accordingly, we were also able to demonstrate that ADAM10 is involved in the constitutive release of soluble L1 in the human neuroglioma cell line H4, which expresses L1 endogenously. In agreement with the presented biochemical data, our functional analysis demonstrated that inhibition or suppression of ADAM10 increased L1-dependent cell adhesion. In contrast, reduced cell adhesion after PMA stimulation could be reversed by the ADAM17 inhibitor. Thus, our data imply a role of ADAM10, as well as ADAM17, in the down-regulation of L1-mediated cell adhesion.
Both ADAM10 and ADAM17 are widely and differentially expressed in the CNS. ADAM10 is mainly expressed in neuronal cells (38, 45), while ADAM17 can be predominantly found in glia (32). During early postnatal development of the cerebellar cortex, granule cells migrate from their germinating zone, forming the external granular layer and ending at their final location in the internal granular layer. In the postnatal cerebellum, L1 is expressed by postmitotic, premigratory, and migrating granule cells at sites of neuron-neuron contact but not by stellate and basket cells or on glia (52). L1 antibodies have been shown to modify cell migration of granule cell neurons in early postnatal mouse cerebellum (41). In this study, we found that the ADAM10-specific inhibitor was able to strongly decrease the L1-dependent haptotactic migration of cerebellar granule cells. This finding suggests that ADAM10-mediated disruption of L1-dependent contacts might be an important mechanism for the regulation of the adhesion of migrating neurons.
Cell adhesion molecules play important roles in regulating synaptic plasticity (3, 69). Increased neural activity is discussed to induce the rapid changes in synaptic morphology that are required for controlling synaptic functions (15, 43, 70). Previously, L1 was thought to be involved in this process (42). Since NMDA stimulation led to increased L1 shedding in primary neurons, it is tempting to speculate that the ADAM10-mediated proteolysis may also be involved in regulating synaptic plasticity. NMDA-stimulated neural activity triggered a rapid increase in the mature form of ADAM10, while we found an opposite effect on ADAM17 expression. In agreement with these observations, the increase of L1 proteolysis due to NMDA treatment could be inhibited with the ADAM10 inhibitor. Interestingly, the NMDA subtype of glutamate receptor is responsible for a substantial fraction of Ca2+ influx in response to synaptic activity, especially early in development (75). This Ca2+ influx is known to affect activity-dependent synaptic plasticity (10). Recently, it was reported that Ca2+ influx activates ADAM10 by triggering the dissociation of calmodulin from pro-ADAM10 (50). Therefore, NMDA-stimulated Ca2+ influx might be the cause of enhanced ADAM10 activity, leading to increased L1 shedding. This hypothesis is also in accordance with a previous report demonstrating that calmodulin inhibitors enhance membrane proximal cleavage of L1 (37). NMDA stimulation has also been shown to induce ADAM10-mediated shedding of the neural adhesion molecule N-cadherin (55), which is also involved in synaptogenesis, thereby influencing synaptic plasticity.
It is well documented that L1 promotes neurite growth in vitro and in vivo and that a number of second-messenger systems are involved. ADAM10 has been implicated in multiple functions in axon extension and guidance (17). Moreover, regulated cleavage of the contact-mediated axon repellent ephrin-A2 (26) and the regulation of midline crossing of CNS axons (62) have been attributed to ADAM10. Therefore, we hypothesized that this protease might also influence L1-dependent neurite outgrowth. Indeed, the ADAM10-specific inhibitor strongly decreased the outgrowth of cerebellar neurons grown on L1 substrate. Even more severe effects on neurite outgrowth, as well as on L1-CTF generation, were caused by the ADAM17-specific inhibitor. On the one hand, this is not surprising, since microexplants do not present a single culture system but rather a complex interaction system. Moreover, L1 homophilic or heterophilic binding has been shown to induce the clustering of the molecule at the plasma membrane, which activates a mitogen-activated protein kinase signaling cascade through the intermediates Src, phosphoinositide-3 kinase, Rac1, and p21-activated kinase, leading to neurite growth (33, 61, 64). Thus, this signaling cascade might contribute to the activation of ADAM17 instead of ADAM10. The ADAM17 inhibitor also slightly affected neurite outgrowth on PLL substrate. Since ADAM17 has been shown to play a critical role in ectodomain shedding of the p75 neurotrophin receptor (74), which is also important for neuronal survival, and in the proteolysis of the neurotrophin receptor TrkA (11), it seems plausible that the processing of additional substrates by ADAM17 might contribute to the severe effect.
Taken together, our data highlight the crucial role of ADAM10 and ADAM17 in the physiological processing of L1. These results have implications for brain development and for the function of the adult brain, since L1 has key roles in neuronal cell adhesion, cell migration, synaptic plasticity, and neurite outgrowth.
We thank S. Jessen for her excellent technical assistance and J. Hedderich for help in perfoming statistical analysis.
This work was supported by Deutsche Forschungsgemeinschaft Sonderforschungsbereich 415 to P.S. and K.R., DFG LU869/1-2 to A.L., Interuniversity Attraction Poles Program P5/19 of the Belgian Federal Science Policy Office, and the APOPIS EU Network. K.R. was supported by the Stiftung zur Förderung der medizinischen Forschung, CAU Kiel. P.A. was supported by a grant from the Deutsche Krebshilfe (10-1307-3A1).