Ephrin-B2-induced Cleavage of EphB2 Receptor Is Mediated by Matrix Metalloproteinases to Trigger Cell Repulsion*

EphB receptors provide crucial adhesive and repulsive signals during cell migration and axon guidance, but it is unclear how they switch between these opposing responses. Here we provide evidence of an important role for matrix metalloproteinases (MMPs) in repulsive EphB2 signaling. We found that EphB2 is cleaved by MMPs both in vitro and in vivo, and that this cleavage is induced by interaction with its ligand ephrin-B2. Our findings demonstrate that MMP-2/MMP-9-specific inhibition or cleavage-resistant mutations in the ectodomain of EphB2 can prevent EphB2-mediated cell-cell repulsion in HEK293 cells, and block ephrin-B1-induced growth cone withdrawal in cultured hippocampal neurons. Transient expression of wtEphB2, but not noncleavable EphB2–4/5 mutant, restored ephrin-B1-induced growth cone collapse and withdrawal in EphB-deficient neurons. The inhibition of EphB2 cleavage also had potent regulatory effects on EphB2 activity. This study provides the first evidence that MMP-mediated cleavage of EphB2 is induced by receptor-ligand interactions at the cell surface and that this event triggers cell-repulsive responses.

Eph 2 receptors are a unique family of receptor tyrosine kinases that play important roles at key stages of neural development, including cell migration, axon guidance, and synaptogenesis (1)(2)(3)(4). Activation of Eph receptors by their specific ligands, ephrins, can trigger either cell-repulsive or -adhesive responses. Because both Eph receptors and ephrins are membrane-bound, Eph/ephrin trans-interactions require cell-cell adhesion between cells expressing Eph receptors and those expressing ephrins. It is unclear how high affinity trans-cellular interactions between Ephs and ephrins are broken to convert adhesion into repulsion. One possibility is that repulsive cell responses are induced by endocytosis of the EphB-ephrinB complex (5,6). Ephrin-B has also been shown to induce EphB2 receptor cleavage by ␥-secretase following endocytosis (7). Alternatively, Eph-ephrin interactions can be broken by proteolytic cleavage of ephrin-A2 and ephrin-B2 by metalloproteinases, subsequent to binding EphA3 or EphB receptors, respectively (8 -10). Moreover, shedding of the EphB2 receptor ectodomain can occur at the cell surface in response to N-Methyl-D-aspartate receptor activation independently of ephrin-B2 interactions (7). However, until now it has not been known whether metalloproteinase cleavage of EphB receptors could influence signaling events that regulate ephrin-induced cell repulsion.
Matrix metalloproteinases (MMPs) belong to the metzincin clan of metalloproteinases that can collectively cleave all extracellular matrix components in addition to a number of cell surface proteins and growth factors (11). As MMP cleavage of extracellular matrix, membrane, and pericellular proteins results in complex changes to cellular homeostasis, these proteinases are likely to play important roles in neuronal development and plasticity. The expression of many MMPs and their inhibitors, tissue inhibitors of metalloproteinases, is developmentally regulated in the mouse brain and spinal cord (12). We have recently shown that MMPs directly impact dendritic spine development and stability in hippocampal neurons (13,14). Other studies also suggest roles for MMPs in long term potentiation and hippocampus-dependent learning (15,16). Furthermore, abnormal MMP expression is associated with several neurodegenerative disorders (17). Therefore, MMPs are well positioned to affect Eph/ephrin signaling and may play a role in the conversion of adhesive cellular responses into repulsion.
Here we have investigated the relationships between ephrin-B-induced EphB2 receptor activation, cell repulsion, and MMP-mediated shedding of the EphB2 receptor ectodomain. We examined the effects of several MMPs and MMP inhibitors on EphB2 shedding and identified two MMP cleavage sites in the ectodomain EphB2 receptor. Modification of these sites did not affect ligand binding but had significant effects on EphB2 activation and ephrin-B-induced cell repulsion. These findings demonstrate for the first time that MMP-mediated cleavage plays an important role in the repulsive signaling of the EphB2 receptor.
Live Cell Cleavage Assay-HEK293T cells transiently transfected with pcDNA3 expression vector containing cDNA of mEphB2 or various EphB2 mutants and 5 DIV hippocampal neurons were treated with 12.5 M of the metalloproteinase inhibitor, Galardin/GM6001 (Chemicon), or with 40 g/ml active MMP-9 for 4 h, together with 10 M lactacystin (proteasome inhibitor; Calbiochem).
Ligand-induced EphB2 Receptor Cleavage-5 DIV hippocampal cultures or HEK293 cells transiently transfected with pcDNA3 plasmids containing cDNA of mEphB2 or EphB2-4/5 were incubated with or without 3 M MMP-2/MMP-9-specific inhibitor, SB-3CT (Calbiochem). SB-3CT has been shown to specifically target gelatinases MMP2 and MMP-9 (20) and to inhibit MMP-9 activity in neurons (21). Cells were then stimulated with 2 g/ml preclustered ephrin-B2-Fc or 2 g/ml control Fc for 4 h, together with 10 M lactacystin (proteasome inhibitor; Calbiochem). Pre-clustered oligomers of ephrin-B2-Fc or Fc were generated by preincubation of ephrin-B2-Fc or Fc with goat anti-human IgG (Jackson ImmunoResearch) for 1 h at 4°C at a ratio of 1:2. The cultured cells or brain tissues were lysed in RIPA buffer containing 10 mM Tris, pH 7.4, 0.15 M NaCl, 1% Triton X-100, 2% SDS, 1 mM EDTA, 0.5 mM pervanadate, and protease inhibitor mixture (Sigma). To remove ephrin-B2-Fc, the cell lysates were preincubated with protein A-agarose beads (Sigma) for 1.5 h at 4°C. Supernatants were collected, and EphB2 cleavage was determined by immunoblotting. Protein concentrations were determined using the BCA protein assay kit (Pierce). The samples were resolved on an 8 -16% SDS-polyacrylamide gel, transferred onto nitrocellulose, and probed with an antibody to mEphB2 (SAM domain). EphB2-N fragment was immunoprecipitated from culture media of treated 5 DIV hippocampal cultures and immunodetected with an antibody against EphB2 extracellular domain (R & D Systems).
RhoA Activation-To investigate RhoA activation, 4 DIV hippocampal neurons or HEK293 cells were treated with preclustered ephrin-B2-Fc or Fc for 20 min. Cell lysates were prepared in 10 mM Tris buffer, pH 7.4, containing 0.15 M NaCl, 1% Triton X-100, 1 mM EDTA, 20 mM MgCl 2 , 0.5 mM pervanadate, and protease inhibitor mixture (Sigma). GTP-RhoA (active form of RhoA) was pulled down from the cell lysates by GST-RBD beads (Upstate Cell Signaling). The beads were washed three times, and the bound proteins were eluted with Laemmli buffer and analyzed by Western blot. The blots were probed with anti-RhoA antibody (Upstate Cell Signaling Solution). The levels of GTP-RhoA were quantified by densitometry and normalized to total RhoA levels in the cell lysates (GTP-RhoA/total RhoA).
Adhesion Assays-HEK293 cells (2 ϫ 10 5 ) stably expressing ephrin-B2 were plated on coverslips and maintained in Dulbecco's modified Eagle's medium containing 10% FBS at 37°C and 5% CO 2 overnight. Then HEK293 cells expressing GFP and mEphB2 (WT or 4/5 mutant) (1 ϫ 10 3 ) were plated on top of the ephrin-B2-expressing cells and allowed to adhere for 1 h. The coverslips were rinsed with PBS to remove nonadherent cells and fixed with 2% paraformaldehyde. Images of the cells were taken under a Nikon TE2000 inverted fluorescent microscope with a 4ϫ air objective and a Hamamatsu ORCA-AG 12-bit CCD camera using Image-Pro software. Ten fields were randomly selected, and the numbers of attached HEK293 cells expressing GFP, GFP and mEphB2, or GFP and EphB2-4/5 were counted. Data represent the average percentage of attached cells Ϯ S.D. from three independent experiments. Statistical differences for multiple groups were assessed by oneway ANOVA followed by Newman-Keuls post hoc tests.
Repulsion Assay-1,1Ј-Dioctadecyl-3,3,3Ј,3Ј-tetramethylindocarbocyanine perchlorate-labeled HEK293 cells expressing wtEphB2 or EphB2-4/5 mutant and 3,3Ј-dioctadecyloxacarbocyaninine perchlorate-labeled HEK293 cells expressing ephrin-B2 were mixed together and plated on coverslips at 2 ϫ 10 4 cells/ml in Dulbecco's modified Eagle's medium with 10% FBS. Images of cell-cell contact sites were taken under an inverted fluorescence microscope (DIC optics; model TE2000; Nikon) with a ϫ40 air Fluor objective and monitored by a 12-bit CCD camera (model ORCA-AG; Hamamatsu, Hamamatsu City, Japan) using Image-Pro software (Media Cybernetics, Silver Spring, MD). During imaging, the cultures were maintained in Hanks' solution (supplemented with 1.8 mM CaCl 2 , 0.45% glucose, and 0.1% bovine serum albumin) at 37°C and 5% CO 2 . Frequencies of the repulsions between the cells expressing EphB2 and ephrin-B2 were analyzed for each group. Cell-cell repulsion was observed when more than 50% of the cell-cell contact areas had separated and developed membrane ruffles and filopodia (5) (supplemental Fig. S2). At least 100 contact sites were randomly selected, and three independent experiments were performed for each condition. Data represent the average percentage of repelled cells Ϯ S.D. from three independent experiments for each condition. Statistical differences for multiple groups were assessed by one-way ANOVA followed by Newman-Keuls post hoc tests.
Growth Cone Assay-Dissociated hippocampal cultures were prepared from E15 to 17 wild type, mice as described previously, plated at 8 ϫ 10 4 cells/ml on glass coverslips, pre-coated with poly-DL-ornithine (0.5 mg/ml in borate buffer) and laminin (5 g/ml in PBS), and maintained in Neurobasal medium with 25 M glutamine and B27 supplement (Invitrogen), under 5% CO 2 , 10% O 2 at 37°C. HEK293 cells expressing GFP and ephrin-B2 or GFP were added to 2 DIV hippocampal cultures. Mixed cultures of hippocampal neurons and HEK293 cells were maintained for 1 h at 37°C and 5% CO 2 , incubated with or without 3 M MMP-2/MMP-9 specific inhibitor, SB-3CT (Calbiochem) for another 3 h, and then fixed in 2% paraformaldehyde and permeabilized in 0.1% Triton X-100. F-actin was visualized by rhodamine-coupled phalloidin (Invitrogen). HEK293 cells were randomly selected and neurites/growth cones located within 40 m of HEK293 cell were imaged and analyzed (see supplemental Fig. S3). The number of neurites, attached or nonattached to HEK293 cells, were counted within a 40-m radius of HEK293 cells. The percentage of attached neurites was calculated for each condition. Nonattached neurites were further subdivided into neurites with collapsed and growing growth cones based on their morphology. Three independent experiments were performed, and at least 300 neurites were analyzed for each condition. Statistical differences for multiple groups were assessed by one-way ANOVA followed by Newman-Keuls post hoc tests.
For the rescue experiment, EphB1 Ϫ/Ϫ /EphB3 Ϫ/Ϫ or EphB1 Ϫ/Ϫ /EphB2 Ϫ/Ϫ /EphB3 Ϫ/Ϫ hippocampal neurons were transfected with GFP or GFP plus EphB2 (WT or 4/5-mutant) at 1 DIV. HEK293 cells expressing ephrin-B1 were added to 2 DIV hippocampal cultures for 3 h. Mixed cultures were processed for immunostaining as described above, and the images were taken under an inverted fluorescent microscope (model TE300; Nikon) with a ϫ40 air Fluor objective or a confocal laser scanning microscope (model LSM 510; Zeiss) with a ϫ63 Fluor objective. Three independent experiments were performed, and at least 150 neurites/growth cones were counted for each condition. Statistical differences for multiple groups were assessed by one-way ANOVA followed by Newman-Keuls post hoc tests.
Binding Assay-3 ϫ 10 5 of HEK293 cells stably expressing mEphB2 (WT or 4/5 mutant) were detached with trypsin/ EDTA (Invitrogen) and resuspended in ice-cold FACS buffer (20 mM HEPES, pH 7.4, 2% FBS, and 2 g/ml goat IgG in PBS). Cells were incubated at 37°C for 10 min and then on ice for 3 h in FACS buffer with increasing concentrations of biotinylated ephrin-B2-Fc or a fixed amount of biotinylated ephrin-B2-Fc (64 M) with increasing concentrations of competitive unlabeled ephrin-B2-Fc and anti-FLAG antibody (M2; Sigma) to determine the levels of EphB2 expression. Cells were washed twice with ice-cold FACS buffer and incubated with Alexa Fluor 488-conjugated streptavidin (0.5 g; Invitrogen) and Alexa Fluor 594-conjugated goat anti-mouse IgG (0.5 g; Invitrogen) for 30 min on ice. Cells were again washed twice with ice-cold PBS and analyzed using a BD FACScan cytometer and Cellquest software (BD Biosciences) as follows. The geometric mean fluorescence intensity (GMean) value was calculated for each sample and represented the average amount of biotinylated ephrin-B2-Fc bound to the cell surface of a single cell. The GMean values were further normalized against the amount of surface EphB2 (WT or 4/5 mutant) and control values representing nonspecific binding of biotinylated ephrin-B2-Fc to HEK293 cells without mEphB2 receptor. Three independent reactions were performed for each condition. Statistical differences for multiple groups were assessed by oneway ANOVA followed by Newman-Keuls post hoc tests.

Ephrin-B-EphB2 Receptor Interactions Induce Shedding of the EphB2
Receptor-We first asked if EphB2 receptor interaction with its ligand ephrin-B2 induces shedding of the EphB2 receptor in primary hippocampal neuron cultures. Full-length EphB2 and two additional EphB2-immunoreactive products, of ϳ65 kDa (EphB2-LF) and 45 kDa (EphB2-ICF), were detected in cell lysates of 5 DIV hippocampal neurons using an antibody specific for the intracellular SAM domain of EphB2 receptor (Fig. 1A). Activation of EphB2 receptor with pre-clustered ephrin-B2-Fc significantly increased levels of EphB2-LF in hippocampal neurons, suggesting that ephrin-B-EphB2 receptor interaction may induce cleavage of the EphB2 receptor. A soluble EphB2-immunoreactive product of ϳ40 kDa (EphB2-N) was also detected in culture medium of ephrin-B2-Fc treated hippocampal neurons using an antibody specific for the extracellular N-terminal portion of EphB2, indicating that EphB2 cleavage most likely occurs at the cell membrane (Fig. 1B). EphB2-LF and EphB2-ICF were also detected in embryonic day 14 (E14) brains of wild-type mice, but not EphB2 Ϫ/Ϫ mice, which confirmed the specificity of anti-EphB2 antibodies used and established that these cleavage fragments occur in vivo during embryonic brain development (Fig. 1C). Interestingly, SB-3CT, an inhibitor of gela-tinaseA/MMP-2 and gelatinaseB/MMP-9, blocked ephrinB-in-duced cleavage of EphB2 in hippocampal neurons (Fig. 1, A and B). These findings indicate that EphB2 receptor is cleaved within the ectodomain both in vitro and in vivo. Furthermore, this cleavage is induced by ephrinB-EphB2 interaction and inhibited by MMP inhibitor SB-3CT.

MMP-7 and MMP-9 Cleave EphB2
Receptor Ectodomain-We next tested if MMP-7 and MMP-9 could cleave EphB2 in vitro and in vivo. Recombinant mEphB2-Fc, consisting of mouse EphB2 ectodomain (residues 1-548) and human Fc, was bound to protein A-agarose beads and incubated with active MMP-7 or MMP-9. Both reactions released an ϳ40-kDa fragment from the beads and reduced the size of protein A-agarose-bound EphB2-Fc (ϳ60 kDa) ( Fig. 2A). The released fragment was detected with an antibody that recognizes the N-terminal portion of EphB2 and was similar in size to EphB2-N that was generated by ephrin-B2 treatment. The size of EphB2-N indicated a possible MMP cleavage site in the first fibronectin (FN) type III domain of EphB2 (Fig.  2E). Recombinant GST-EphB2 protein containing full-length chicken EphB2 was also cleaved by MMP-7 and MMP-9, generating a shorter GST-EphB2 product (ϳ75 kDa) containing GST (ϳ30 kDa) and a similar N-terminal portion of EphB2 (ϳ40 kDa, Fig. 2, B and E). When expressed in HEK293 cells, membrane-bound, full-length EphB2 was cleaved by active MMP-9 to generate two C-terminal EphB2 cleavage products that were similar in size to EphB2-LF and EphB2-ICF, which were produced in response to ephrin-B2 treatment (Fig. 2, C and E). Membrane-bound EphB2-LF was unstable in HEK293 cells and rapidly underwent a subsequent cleavage to produce more EphB2-ICF. Finally, endogenous mouse EphB2 was also cleaved by MMP-9 in hippocampal neuron cultures to generate the same fragments, EphB2-LF and EphB2-ICF (Fig. 2, D and E). Interestingly, EphB2-LF was more stable in hippocampal neurons than in HEK293 cells. Taken together, these findings demonstrate that MMPs can cleave within the extracellular region of EphB2 in recombinant form and when expressed on the surface of live cells.  3 and 4), and the antibody is specific to the N-terminal region of EphB2. C, Western blot detects full-length EphB2 and EphB2 cleavage products, EphB2-LF and EphB2-ICF, in cell lysates from E14 brain of wild type, but not EphB1 Ϫ/Ϫ / EphB2 Ϫ/Ϫ /EphB3 Ϫ/Ϫ mice, detected with an antibody against the SAM domain of EphB2 (upper and middle panels). The blot was also re-probed with anti-GAPDH antibody to confirm equal loading (lower panel

EphB2-4/5 Mutant Resists MMP-9
Cleavage-We identified seven potential MMP cleavage sites (22) in the cysteine-rich region and FN type III domains of EphB2 receptor (Fig. 3A), and then used site-directed mutagenesis to modify each site in EphB2 expression plasmids. Single mutations at sites 4 (I358A in chicken and I395A in mouse) and 5 (L396A in chicken and L433A in mouse) made EphB2 somewhat resistant to MMP cleavage (Fig. 3B). However, the double EphB2-4/5 mutant was virtually noncleavable by MMPs even at high concentrations (Fig. 3C). Interestingly, besides preventing cleavage of EphB2 at the first FN type III domain, double mutations at sites 4 and 5 also prevented secondary shedding of EphB2-4/5 in HEK293 cells. EphB2-ICF was detected in 293 cells expressing wtEphB2, but not EphB2-4/5 mutant, in response to both MMP-9 and ephrin-B2-Fc treatments (Fig. 3, C and D). Cleavage site 5 (V/ISDL) is highly conserved among all Eph receptors (supplemental Fig. S1). However, site 4 is unique to EphB2 and EphB3 receptors but is not found in EphB1 receptor, suggesting that differences in MMP cleavage susceptibility may affect the specificity of repulsive responses triggered by different EphB receptors. These findings demonstrate that ephrinB-induced shedding of EphB2 results from MMP-mediated cleavage at two sites within the extracellular domain.
In another assay we evaluated cell adhesion between HEK293 cells expressing ephrin-B2 ligand and HEK293 cells expressing EphB2 receptor. Ephrin-B2-expressing HEK293 cells were plated on glass coverslips and grown to confluence. Next, EphB2 (wtEphB2 or EphB2-4/5)-expressing HEK293 cells were transfected with a GFP-expressing vector and plated on top of the confluent cells. Loose cells were removed after 1 h, and the number of GFP-expressing cells that adhered to the confluent layer was counted. Expression of wtEphB2 reduced the adhesion of HEK293 cells to ephrin-B2-expressing cells (14.1 Ϯ 3.1%) but not to untransfected HEK293 cells (26.1 Ϯ 4.8%; Fig. 4B). However, cells expressing EphB2-4/5 were more adhesive to ephrin-B2-expressing cells (38.9 Ϯ 6.8%) than control cells (27.1 Ϯ 4.1%) and 3-fold more adhesive than cells expressing wtEphB2 (14.1 Ϯ 3.1%; Fig. 4B). As sites 4 and 5 lie within the first FN type III domain of EphB2, we tested if the EphB2-4/5 mutations affect binding affinity for ephrin-B2-Fc, compared with wtEphB2. Biotinylated ephrin-B2-Fc was bound to HEK293 cells expressing EphB2-4/5 and wtEphB2 and then displaced with increasing concentrations of unlabeled ephrin-B2-Fc. Both wtEphB2 and EphB2-4/5 showed high affinities for ephrin-B2 (supplemental Fig. S4). These results demonstrate that mutations at MMP cleavage sites 4 and 5 in EphB2 ectodomain do not change the ligand affinity of EphB2, but do affect the ability of EphB2 receptor to induce cell-cell repulsion following the ephrin-B2-EphB2 receptor interactions.
EphB2-mediated Growth Cone Collapse and Withdrawal in E15 Hippocampal Neurons Are Regulated by MMPs-To determine whether MMP-mediated cleavage of EphB affects neuronal process repulsion or growth cone collapse, we assayed hippocampal neurons from wild type and EphB knock-out mice. We first assayed growth cone collapse and withdrawal by adding ephrin-B2 expressing HEK293 cells to 2 DIV wild-type hippocampal neurons for 3 h (Fig. 5A). Quantitative analysis revealed that a significantly lower proportion of the neurites were attached to HEK293 cells expressing ephrin-B2 (293-eB2, 22.7 Ϯ 2.4%) than control HEK293 cells (293, 40.6 Ϯ 0.8%), indicating higher growth cone withdrawal (Fig. 5B). We found a higher percentage of collapsed growth cones in culture with ephrin-B2-expressing HEK293 cells (293-eB2, 63.7 Ϯ 3.9%) as compared with cultures with control HEK293 cells (293, 47.7 Ϯ 4.0%; Fig. 5C and supplement Fig. S3). However, the effects of ephrin-B2-expressing HEK293 cells (293-eB2-SB-3CT, 57.7 Ϯ 3.3%) were not significantly different from control HEK293 The amount of cleavage product EphB2-ICF (ϳ45 kDa) was reduced in HEK293 cells that overexpressed mutants 4 or 5, indicating higher resistance to MMP-9 proteolysis. C, EphB2-ICF cleavage product (lower band) was detected in a concentration-dependent manner in HEK293 cells expressing mouse wtEphB2 (mEphB2, WT) but not EphB2-4/5 (Mut 4/5). This Western blot shows that EphB2-4/5 is noncleavable by MMP-9 in HEK293 cells, even at concentrations as high as 40 ng/ml (upper panel). The blot was re-probed with anti-GAPDH antibody to confirm equal protein loading (lower panel). D, Western blot shows that EphB2 cleavage is induced with ephrin-B2-Fc in HEK293 cells expressing wtEphB2, but not noncleavable EphB2-4/5 (upper panel). The blot was re-probed with anti-GAPDH antibody to confirm equal protein loading (lower panel). cells (293-SB-3CT, 52.0 Ϯ 3.6%) in the presence of the MMP-2/MMP-9 inhibitor SB-3CT, suggesting that ephrin-B2-induced growth cone withdrawal and collapse are regulated by MMPs. Surprisingly, the proportion of collapsed growth cones in 293-eB2 cultures treated with SB-3CT (293-eB2-SB-3CT, 57.7 Ϯ 3.3%) was only slightly, but not significantly, lower as compared with untreated 293-eB2 cultures (293-eB2, 63.7 Ϯ 3.9%). The lack of significance may be explained by the fact that this MMP inhibitor also slightly increased the number of collapsed growth cones in control 293 cultures, suggesting that it may have additional effects on growth cones independent of EphB2 receptor signaling. It is also possible that the MMP-2/MMP-9 inhibitor did not completely block EphB2 cleavage and only partially inhibited EphB2-mediated growth cone collapse.
ated growth cone withdrawal and collapse in cultured hippocampal neurons.
MMP Cleavage Regulates EphB2 Receptor Activity-Interestingly, the repulsive responses from ephrin-B-EphB receptor interactions coincide with the cleavage of only a small fraction of total EphB2, suggesting that the cleavage of EphB2 receptor has potent regulatory effects on receptor signaling. Therefore, we examined whether noncleavable EphB2-4/5 could be activated by its ligand and induce intracellular signaling events similar to wtEphB2. Pre-clustered ephrin-B2-Fc induced rapid tyrosine phosphorylation of both wtEphB2 and EphB2-4/5 in HEK293 cells at 5 min and recruitment of signaling proteins, such as nonreceptor tyrosine kinase Src and focal adhesion kinase (FAK; Fig. 7, A and B). However, phosphorylation levels of EphB2-4/5 decreased at 10 min and returned to basal levels by 15 min. Moreover, a significant decrease in the association of EphB2-4/5 with FAK and Src was noted at 15 min following its activation with ephrin-B2-Fc. These responses contrasted with the sustained phosphorylation of wtEphB2 and its association with FAK and Src, which lasted for at least 15 min (Fig. 7B). In primary cultures of hippocampal neurons, the inhibition of MMP-2/MMP-9 activity with SB-3CT reduced the levels of EphB2 phosphorylation and inhibited the recruitment of FAK and Src to EphB2 receptor following its activation with ephrin-B2-Fc (Fig. 7C). Taken together, these results demonstrate that the inhibition of EphB2 cleavage with MMP inhibitor or cleavage-resistant mutation in the EphB2 ectodomain has potent regulatory effects on EphB2 receptor activity.

EphrinB2-induced RhoA Activation Is Regulated by MMP Activity-As
Eph receptor-repulsive responses are shown to be mediated through RhoA GTPase, we asked whether MMP cleavage of EphB2 influences its ability to activate RhoA GTPase. For this, we tested the ability of ephrin-B2-Fc to activate RhoA GTPase in HEK293 cells expressing wtEphB2 or noncleavable EphB2-4/5. Significantly higher levels of active RhoA were detected in HEK293 cells expressing wtEphB2, but not EphB2-4/5, in response to ephrin-B2-Fc, as compared with control Fc (Fig. 8A). Moreover, ephrinB2-induced RhoA activation was also observed in untreated hippocampal cultures but not in cultures treated the MMP-2/ MMP-9 inhibitor SB-3CT (Fig. 8B). These findings demonstrate that MMP-mediated cleavage of EphB2 receptor regulates its ability to elicit RhoA activation.

DISCUSSION
Both MMPs and EphB receptors play important roles in the development and plasticity of the central nervous system, but the consequences of their interactions have not been established. Here we provide conclusive evidence for the role of MMPs in regulating repulsive EphB2 receptor signaling. Our findings indicate that EphB receptor cleavage is induced by ephrin-B2 interaction and mediated by MMPs. MMP inhibitors blocked ephrininduced EphB2 cleavage and ephrinB2-triggered growth cone collapse. Furthermore, cleavage-resistant mutations at two MMP cleavage sites in the ectodomain of EphB2 receptor inhibited its ability to induce cell-cell repulsion and growth cone withdrawal.
Recent studies have revealed that ephrin-B binding can lead to EphB2 receptor cleavage within the transmembrane region by a ␥-secretase mechanism that produces a C-terminal cytoplasmic fragment of EphB2 (7); however, a corresponding N-terminal fragment of EphB2 containing the entire ectodomain was not detected, suggesting that an additional shedding step may occur within the EphB2 ectodomain. Our results show that ephrin-B2 binding induces EphB2 receptor cleavage within the first FN type III domain at the cell surface to produce a long fragment (EphB2-LF) that is subsequently cleaved within the transmembrane domain to release a shorter fragment (EphB2-ICF). This secondary EphB2-ICF cleavage product is similar to the fragment produced by ␥-secretase. Our findings demonstrate that EphB2 is cleaved first by MMPs at the cell surface in response to interaction with ephrin-B2, and a subsequent secondary cleavage within the transmembrane domain, likely mediated by ␥-secretase, produces EphB2-ICF. Interestingly, ligand-induced ectodomain shedding is also required for ␥-secretase cleavage of Notch and the central Alzheimer factor, amyloid precursor protein (APP). Ligand binding to Notch diminishes steric hindrance and permits ectodomain shedding by A-disintegrin-and-Ametalloproteinase (ADAM)-10 (24) or ADAM-17 (25) to produce NEXT. NEXT processing by ␥-secretase releases Notch intracellular domain (26). Prior to the release of ␤-amyloid by ␥-secretase, the extracellular region of APP is shed by ␤-amyloidcleaving enzymes to produce an intermediate C-terminal fragment of APP (residues 672-770) (27). It is not clear if ␤-amyloidcleaving enzyme cleavage is liganddependent or even what the relevant APP ligands may be, but findings presented here add to evidence that ligand-dependent ectodomain shedding is a common feature of ␥-secretase cleavage substrates.
Our analysis of putative MMP cleavage sites in the ectodomain of EphB2 suggests that other Eph receptors may also serve as MMP substrates. The cleavage site 5 (ISDL) is highly conserved at P3 (Val, Met, or Ile) and P1Ј (Leu) in all Eph receptors, with the exception of EphA6 (supplemental Fig. S1). On the other hand, site 4 (VYNI) is found only in EphB2 and EphB3 receptors, suggesting that differences in MMP-cleavage susceptibilities of EphB receptors may affect the specificity of repulsive responses triggered by the different EphB receptors (28). The appearance of EphB2 cleavage products in embryonic brain (Fig. 1C) indicates these EphB2 cleavages also occur in vivo and may play important roles in embryonic development. Indeed, early studies on the expression of chicken EphB2 (Cek5) in developing skeletal muscle identified two EphB2-immunoreactive products of 65 and 40 kDa that were similar in size to EphB2-LF and EphB2-ICF and were detected only at specific developmental stages (29). Because Eph-ephrin interactions can elicit different, sometimes opposing, cel- Treatment of HEK293 cells with ephrin-B2-Fc induced rapid tyrosine phosphorylation of both wtEphB2 and EphB2-4/5 at 5 min and recruitment of signaling proteins, such as nonreceptor tyrosine kinase Src and FAK. Whereas tyrosine phosphorylation of wtEphB2 remained high at 10 and 15 min, the phosphorylation level of EphB2-4/5 decreased after 10 min and returned to basal level at 15 min. Moreover, a significant decrease in association of EphB2-4/5 with FAK and Src was noted at 15 min following its activation with ephrin-B2-Fc (B). C, 4 DIV hippocampal neurons were treated with ephrin-B2-Fc for 15 min to activate EphB receptors with or without MMP-2/MMP-9 inhibitor SB-3CT. Control cultures were treated with Fc for 15 min. The lysates were immunoprecipitated (IP) with anti-EphB2 antibody and immunoblotted (IB) with anti-phosphotyrosine antibody (PY20). The levels of the EphB2 phosphorylation or association of FAK and Src with EphB2 were quantified by densitometry and normalized to total EphB2 levels. MMP inhibitor SB-3CT blocked ephrin-B2-induced recruitment of Src and FAK to the EphB2 receptor. The error bars indicate S.E. lular responses, it is possible that EphB2 cleavage may play an important role in regulating these responses.
Ephrin-A has been previously identified as a substrate for membrane-bound ADAMs (8,9). Here we demonstrate that the EphB2 receptor is cleaved by secreted MMPs, MMP-2 and MMP-9, and this cleavage is induced by ephrinB-EphB receptor interaction. MMP-2 and MMP-9 are the most abundant MMPs in the brain, and their expression/activities are regulated during neural development, in response to brain injury, and in several neurological disorders (14,30). Therefore, changes in the expression and/or activation of MMP-2/MMP-9 may directly regulate EphB2-mediated cellular responses. Indeed, we show that cleavage-resistant mutations at two MMP cleavage sites in the ectodomain of EphB2 or blocking MMP-2/MMP-9 activity inhibit EphB2-mediated cell-cell repulsion and growth cone withdrawal.
To gain further insight into the mechanism of MMP regulation of EphB2-mediated cellular responses, we investigated the effects of MMP inhibition on EphB2 receptor activity. Our findings suggest that MMP-mediated cleavage may elicit EphB2mediated repulsive cellular responses by facilitating prolonged EphB2 receptor activation. In contrast to the sustained phosphorylation of wtEphB2, cleavage-resistant mutations in the EphB2 ectodomain induced rapid dephosphorylation of noncleavable EphB2-4/5 following its initial phosphorylation/activation with ephrin-B2. MMP-mediated cleavage of EphB2 receptor may enhance the duration of EphB2 phosphorylation with the help of EphB2 cleavage products, EphB2-LF and EphB2-ICF. These fragments lack the ephrin-binding domain, but retain the entire cytoplasmic portion of the receptor, and may sequester phosphatases from full-length EphB2 receptor, thus preventing its dephosphorylation. Several tyrosine phosphatases have been shown to be involved in regulating Eph-mediated repulsive responses (31)(32)(33). Daniel and coworkers (31) reported that EphB1 can bind low molecular weight protein-tyrosine phosphatase, an interaction that is blocked by the Y929F mutation, indicating that the SAM domain may be responsible for protein-tyrosine phosphatase recruitment and receptor dephosphorylation. Protein-tyrosine phosphatase receptor type O (Ptpro) has also been shown to dephosphorylate both EphA and EphB receptors and to inhibit ephrin-mediated repulsion of retinal axons (33). Therefore, changes in EphB2 receptor phosphorylation can directly influence EphB2-mediated signaling cascades and consequent cellular responses.
The intracellular EphB2 fragment may also signal on its own by interacting with nonreceptor tyrosine kinases through its Src homology domain 2 or by recruiting adaptor proteins. There is also an intriguing possibility that the EphB2 intracellular fragment may be translocated to the nucleus and directly affect transcription similar to NICD of Notch. In contrast to NICD, the cytoplasmic portion of the EphB2 receptor does not contain an apparent nuclear localization signal sequence (27,34,35). However, it is possible that the intracellular fragment of EphB2 may indirectly affect transcription through its Src homology domain 2 and kinase activity.
Here we demonstrate that MMP cleavage of EphB2 receptor is important for regulating EphB2 signaling because blocking EphB2 receptor cleavage with MMP inhibitor or cleavage-resistant mutations inhibits ephrinB-induced EphB2 receptor phosphorylation, recruitment of FAK, and RhoA activation (Figs. 7 and 8). Our previous studies have shown that ephrinB1-induced EphB2 receptor activation, in primary cultures of hippocampal neurons, can enhance the levels of RhoA activity through recruitment and activation of FAK (37), suggesting that RhoA activation may also contribute to ephrinB1-induced growth cone withdrawal and collapse. RhoA activation has been shown to mediate ephrinA-induced growth cone collapse (38) and the retraction of cell processes and cell rounding in HEK293 and melanoma cells (39). The effects of RhoA on growth cone collapse and cell repulsion are most likely attributed to its ability to promote actin-myosin contractility (40). Beyond regulating the activity of Rho GTPases, EphB receptors may also trigger neurite retraction (41) by inhibiting R-Ras function through phosphorylation (42) or reducing R-Ras activity through the GTPase-activating protein, p120 RasGAP (36). Future studies will determine whether MMP cleavage of EphB2 receptor can also regulate the Ras/mitogen-activated protein kinase cascade, which plays an important role in cell migration and axon guidance.
In conclusion, our findings demonstrate that ephrinB-induced cleavage of EphB2 is regulated by MMPs and facilitates receptor activity to produce strong cytoskeletal responses, such as cell-cell repulsion and growth cone collapse.