Intrinsic Signaling Functions of the β4 Integrin Intracellular Domain*

A key issue regarding the role of α6β4 in cancer biology is the mechanism by which this integrin exerts its profound effects on intracellular signaling, including growth factor-mediated signaling. One approach is to evaluate the intrinsic signaling capacity of the unique β4 intracellular domain in the absence of contributions from the α6 subunit and tetraspanins and to assess the ability of growth factor receptor signaling to cooperate with this domain. Here, we generated a chimeric receptor composed of the TrkB extracellular domain and the β4 transmembrane and intracellular domains. Expression of this chimeric receptor in β4-null cancer cells enabled us to assess the signaling potential of the β4 intracellular domain alone or in response to dimerization using brain-derived neurotrophic factor, the ligand for TrkB. Dimerization of the β4 intracellular domain results in the binding and activation of the tyrosine phosphatase SHP-2 and the activation of Src, events that also occur upon ligation of intact α6β4. In contrast to α6β4 signaling, however, dimerization of the chimeric receptor does not activate either Akt or Erk1/2. Growth factor stimulation induces tyrosine phosphorylation of the chimeric receptor but does not enhance its binding to SHP-2. The chimeric receptor is unable to amplify growth factor-mediated activation of Akt and Erk1/2, and growth factor-stimulated migration. Collectively, these data indicate that the β4 intracellular domain has some intrinsic signaling potential, but it cannot mimic the full signaling capacity of α6β4. These data also question the putative role of the β4 intracellular domain as an “adaptor” for growth factor receptor signaling.

A key issue regarding the role of ␣6␤4 in cancer biology is the mechanism by which this integrin exerts its profound effects on intracellular signaling, including growth factor-mediated signaling. One approach is to evaluate the intrinsic signaling capacity of the unique ␤4 intracellular domain in the absence of contributions from the ␣6 subunit and tetraspanins and to assess the ability of growth factor receptor signaling to cooperate with this domain. Here, we generated a chimeric receptor composed of the TrkB extracellular domain and the ␤4 transmembrane and intracellular domains. Expression of this chimeric receptor in ␤4-null cancer cells enabled us to assess the signaling potential of the ␤4 intracellular domain alone or in response to dimerization using brain-derived neurotrophic factor, the ligand for TrkB. Dimerization of the ␤4 intracellular domain results in the binding and activation of the tyrosine phosphatase SHP-2 and the activation of Src, events that also occur upon ligation of intact ␣6␤4. In contrast to ␣6␤4 signaling, however, dimerization of the chimeric receptor does not activate either Akt or Erk1/2. Growth factor stimulation induces tyrosine phosphorylation of the chimeric receptor but does not enhance its binding to SHP-2. The chimeric receptor is unable to amplify growth factor-mediated activation of Akt and Erk1/2, and growth factorstimulated migration. Collectively, these data indicate that the ␤4 intracellular domain has some intrinsic signaling potential, but it cannot mimic the full signaling capacity of ␣6␤4. These data also question the putative role of the ␤4 intracellular domain as an "adaptor" for growth factor receptor signaling.
The ␣6␤4 integrin is a structural and functional anomaly among the integrin family of receptors. This integrin, which is expressed primarily on the basal surface of epithelia and in a few other cell types, is defined as an adhesion receptor for most of the known laminins (1)(2)(3). The distinguishing structural feature of ␣6␤4 is the atypical intracellular domain of the ␤4 subunit. Two pairs of fibronectin type III repeats separated by a connecting segment characterize this domain, and it is distinct both in size (ϳ1000 amino acids) and structure from any other integrin subunit (4). Although the ␣6␤4 integrin provides a well characterized adhesive function in normal epithelial cells by anchoring the epithelium to its underlying basement membrane, the carcinoma-associated functions of this integrin are becoming increasingly recognized (5). Importantly, the expression of this integrin is often maintained as epithelial structures dissociate during the initiation and progression of carcinomas, and, consequently, many carcinomas express ␣6␤4 (6,7). Numerous studies by our groups and others have revealed that ␣6␤4 can facilitate the ability of carcinoma cells to migrate, invade, and resist apoptotic stimuli (8 -16). More recently, ␣6␤4 has been implicated in the genesis of squamous and breast carcinomas (17)(18)(19). The ability of ␣6␤4 to impact these diverse functions results largely from its effects on multiple signaling pathways, including phosphatidylinositol 3-kinase/Akt and MAPK, 2 a process that may result from its association with specific growth factor receptors, tetraspanins, and possibly other molecules (5). The dichotomy of ␣6␤4 function is summarized best by the hypothesis that ␣6␤4 switches from a mechanical adhesive device into a signaling competent receptor during the progression from normal epithelium to invasive carcinoma (5,20).
A key issue regarding the role of ␣6␤4 in carcinoma biology is the mechanism by which this integrin exerts its profound effects on intracellular signaling. Given that ␣6␤1 and ␣6␤4 exhibit substantial differences in their known signaling functions, it is reasonable to postulate that the unique signaling properties of ␣6␤4 derive largely from the ␤4 intracellular domain (21). Despite its large size, however, the ␤4 intracellular domain lacks intrinsic kinase activity. A current hypothesis argues that the ␤4 intracellular domain functions as a "signaling adaptor" that facilitates signaling through growth factor receptors such as Met (22). Another viable, though not mutually exclusive, hypothesis is that the association of ␣6␤4 with tetraspanins and its localization in tetraspanin-enriched membrane microdomains enhances its signaling capacity (23). These hypotheses are complicated by the finding that ␣6␤4 signaling can be either dependent on engagement of its ligands (laminins) or independent of such ligation (22,24).
One approach to understanding the nature of ␣6␤4 signaling in more detail is to evaluate the intrinsic signaling capacity of the unique ␤4 intracellular domain itself in the absence of con-* This work was supported by National Institutes of Health Grants CA80789 tributions from the ␣6 subunit and tetraspanins, and to assess the ability of growth factor receptor signaling to cooperate with this intracellular domain. To execute this approach, we generated a chimeric receptor composed of the TrkB extracellular domain and the ␤4 transmembrane and intracellular domains. Expression of this chimeric receptor in ␤4-null cancer cells enabled us to assess the signaling potential of the ␤4 intracellular domain either alone or in response to dimerization of the chimeric receptor using brain-derived neurotrophic factor (BDNF), the ligand for TrkB (25). The data obtained indicate that the ␤4 intracellular domain has some intrinsic signaling potential but that it cannot mimic the full signaling capacity of the intact ␣6␤4 integrin. These data also highlight the need to re-evaluate the putative role of the ␤4 intracellular domain as an "adaptor" for growth factor receptor signaling.

EXPERIMENTAL PROCEDURES
Cells-The MDA-MB-435 cancer cell line, which has been reported to be derived from the melanoma cell line M14 (26), was obtained from the Lombardi Breast Cancer Depository at Georgetown University (Washington, D. C.) and maintained in Dulbecco's modified Eagle's medium (DMEM) containing 1 g/liter glucose, 5% fetal bovine serum, 1% penicillin-streptomycin, and 10 mM HEPES. MDA-MB-435 cells stably expressing the ␣6␤4 integrin were described previously (21).
MDA-MB-435 cells stably expressing full-length TrkB, Trk-Bextra, and TrkB␤4 were generated as follows. For generation of a full-length TrkB retroviral expression construct, a 2.5-kb EcoRI/SalI fragment was removed from the rat TrkB expression construct pEYFP-N1TrkB (a gift from Dr. Luis Parada, University of Texas Southwestern Medical Center, Dallas, TX), bluntended with T4 DNA polymerase (New England Biolabs, Beverly, MA), and subcloned into XhoI-digested and blunt-ended pCLXSN (Imgenex, San Diego, CA). To generate a TrkBextra retroviral expression construct, the extracellular and transmembrane regions (amino acids 1-458) of TrkB were PCRamplified from pEYFP-N1TrkB with the following primers: Primer 1 (forward) 5Ј-CCGCTCGAGCGGATGTCGCCCTG-GCCGAGGT and Primer 2 (reverse) 5Ј-CCGCTCGAGCGG-CTAATGTCTCGCCAACTTGA. Purified PCR products were then digested with XhoI and subcloned into XhoI-digested pCLXSN. For generation of a retroviral TrkB␤4 expression construct, the TrkB extracellular domain (amino acids 1-429) was first amplified from pEYFP-N1TrkB by PCR, adding 18 bp of sequence complementary to the coding region for the first 6 amino acids of the ␤4 transmembrane domain (710 -715) to the 3Ј-end of the product, using the following primers: Primer 1 (forward) and Primer 3 (reverse) 5Ј-GAGCCACCAGAAGGA-ATGCTCCCGATTGGTT. Next, the transmembrane and intracellular domains of ␤4 were amplified from pRc/CMV␤4 (27), adding 18 bp of sequence complementary to the coding region of the final 6 amino acids of the TrkB extracellular domain (424 -429) to the 5Ј-end of the product, using the following primers: Primer 4 (forward) 5Ј-AACCAATCGGGAG-CATTCCTTCTGGTGGCTC and Primer 5 (reverse) 5Ј-CCG-CTCGAGCGGTCAAGTTTGGAAGAACTGT. Finally, these two PCR products were used together as template with Primer 1 (forward) and Primer 5 (reverse) to PCR-amplify the TrkB␤4 insert. This insert was then digested with XhoI and subcloned into pCLXSN. All plasmids were sequenced to confirm that the inserts were correctly inserted and contained no mutations.
To generate stable cell lines expressing full-length TrkB, Trk-Bextra, and TrkB␤4, the above retroviral expression constructs were transfected along with expression plasmids for the vesicular stomatitis virus glycoprotein envelope glycoprotein and Gag-Pol packaging proteins into 293T cells using Lipofectamine 2000 (Invitrogen). Three days post-transfection, viral supernatants were harvested, diluted in serum-containing media supplemented with 8 g/ml Polybrene (Sigma), and used to infect MDA-MB-435 cells. Following 24 h of infection, cells were selected with Geneticin (2.5 mg/ml) to yield stable cell lines, and cells were then maintained in 1.0 mg/ml Geneticin. Immunoblotting was performed, as described below, to confirm protein expression in the stable cell lines.
For the generation of cell lines with stable knockdown of the SHP-2 protein, an shRNA cloned into the pSUPER retroviral expression construct was obtained from Ben Neel (Ontario Cancer Institute, Toronto), and viral production and infection were performed as described above. Puromycin (0.5 g/ml) was used for cell selection.
To generate stable HCC1937 cell lines expressing ␤4 shRNA-pFSIPPW the above construct was transfected along with the ViraPower lentiviral packaging mix (Invitrogen) into 293T cells using Lipofectamine 2000 following the manufacturer's instructions. Viral supernatants were harvested 3 days posttransfection, diluted in serum-containing medium supplemented with 8 g/ml Polybrene, and used to infect HCC1937 cells. Stable cell lines were generated by selection with puromycin (1 g/ml). Cells were maintained in 0.5 g/ml puromycin. Suppression of ␤4 expression was confirmed by immunoblotting as described below.
A875 melanoma cells were a generous gift from Alonzo Ross (University of Massachusetts Medical School, Worcester, MA) and were maintained in DMEM containing 1g/liter glucose, 10% fetal bovine serum, 1% penicillin-streptomycin, and 10 mM HEPES.
Biochemical Analyses-Biotinylation of cell surface proteins was performed using EZ-Link Sulfo-NHS-LC-Biotin (Pierce) following the manufacturer's instructions with a few modifications, as follows. Cells were washed and suspended in PBS (pH 8.0) at a concentration of 25 ϫ 10 6 cells/ml and labeled with 0.5 mg/ml biotin for 30 min at room temperature. Following the labeling, cells were washed with PBS containing 100 mM glycine to quench and remove excess biotin and then extracted in a Triton X-100 buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100) supplemented with 1 mM sodium fluoride, 1 mM sodium orthovanadate, and complete-Mini protease inhibitor mixture (Roche Applied Science).
For co-immunoprecipitation studies, cells were extracted with either the Triton X-100 buffer or an Nonidet P-40 buffer (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, 1 mM sodium fluoride, 1 mM sodium orthovanadate, and complete-Mini protease inhibitor mixture). Cell extracts were clarified by centrifugation at 16,000 ϫ g for 10 min and preabsorbed for 2 h using protein G-Sepharose beads. After centrifugation at 5,000 ϫ g for 5 min to pellet the beads, extracts were incubated with primary Abs for 1 h, and immune complexes were then precipitated with protein G-Sepharose overnight. For precipitation of biotinylated proteins, pre-absorbed extracts were incubated with a streptavidin agarose-conjugate overnight. Precipitates were washed two times with lysis buffer, one time with PBS, and then eluted in 1ϫ reducing SDS sample buffer while boiling for 5 min.
For immunoblotting, cell extracts were prepared as described above, and protein concentrations determined using the Bradford assay (Bio-Rad). These extracts, or eluted immune complexes from the immunoprecipitations, were separated by SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad). Membranes were then blocked in TBS-T (Tris-Cl, pH 7.5, 150 mM NaCl, 0.05% Tween 20) containing 5% nonfat dry milk for at least 30 min, except for immunoblots for phosphorylated proteins, in which case TBS-T containing 5% BSA was used. Following blocking, membranes were incubated overnight with primary Ab in blocking buffer, washed 3ϫ with TBS-T, incubated with horseradish peroxidase-conjugated secondary Ab in blocking buffer, washed 3ϫ with TBS-T, and detected using SuperSignal West Pico Chemiluminescent substrate (Pierce).
For assays where BDNF was used to dimerize the TrkB␤4 chimera, cells were washed once with PBS and serum-deprived for 24 h in DMEM containing 1 g/liter glucose, 1% penicillinstreptomycin, 10 mM HEPES, and 0.1% BSA. Following serum starvation, cells were incubated with fresh serum-free medium supplemented with 100 ng/ml BDNF for the indicated time periods at 37°C. Treatment of cells with HGF was performed in a similar manner, using HGF at a final concentration of 100 ng/ml.
Integrin Clustering-Cells were removed from their dishes with trypsin and washed twice with RPMI medium containing 1% BSA (RH/BSA). After washing, the cells were resuspended in the same buffer at a concentration of 10 6 cells/ml and incubated for 30 min with either ␣6or ␤4-integrin-specific Abs (2 g/ml) or in buffer alone. The cells were washed once, resuspended in RH/BSA, and added to plates that had been coated overnight at 4°C with anti-mouse IgG (100 g/10-cm plate). The plates were blocked with RH/BSA for 30 min prior to the addition of the cells. Inhibitors were added to the cells for 15 min prior to plating the cells in the antibody-coated plates. After incubation at 37°C for 15-60 min, the cells were washed twice with PBS and solubilized at 4°C for 10 min in the Nonidet P-40 lysis buffer. Nuclei were removed by centrifugation at 12,000 ϫ g for 8 min.
Migration Assays-Migration assays were performed using Corning (Corning, NY) Transwell chambers (8.0-m pore size). Membranes were prepared by coating the upper and lower surfaces with 15 g/ml collagen (Cohesion, Palo Alto, CA) overnight at 4°C, and then blocking with DMEM containing 0.25% heat-inactivated BSA for 1 h at 37°C. Cells were then trypsinized, counted, and resuspended in DMEM containing 0.25% heat-inactivated BSA. A total of 1 ϫ 10 6 cells was added to the upper chamber of the Transwell, and HGF (50 g/ml) was added to the bottom wells as a chemoattractant. Migration was allowed to proceed for 2.5 h at 37°C at which time nonmigrating cells were removed mechanically from the upper chamber using a cotton swab. Cells that migrated to the lower surface of the Transwell membrane were fixed in methanol for 10 min at room temperature, and membranes were mounted on glass slides using Vectashield mounting medium containing 4Ј,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Migration was quantified by counting the number of stained nuclei in five fields of view in each Transwell, in triplicate.

Generation and Characterization of a TrkB␤4 Chimeric
Receptor-To assess the intrinsic signaling capacity of the ␤4 integrin intracellular domain, we generated an expression construct consisting of the extracellular domain of the neuronal TrkB receptor fused to the transmembrane and intracellular domains of ␤4 (Fig. 1A). We used this system as a model of ␤4 lateral clustering following adhesion of ␣6␤4 to its extracellular matrix ligand by inducing dimerization of these chimeric mol-ecules with the TrkB ligand, BDNF. We also generated a truncated, signaling incompetent TrkB construct consisting of only the extracellular and transmembrane regions of TrkB to use as a negative control (Fig. 1A). This control enabled us to discount the possibility that endogenous receptors mediate BDNF signaling. These constructs, as well as full-length TrkB, were expressed in MDA-MB-435 cancer cells, which express neither the ␣6␤4 integrin nor the TrkB receptor, as shown in immunoblots of extracts from subclones expressing backbone vector (pCLXSN) alone (Fig. 1B). We chose these cells because they have been used in numerous studies of ␤4 function and signaling, allowing us to compare the signaling capability of the TrkB␤4 chimera to that of wild-type ␣6␤4 (12,21,22,29,30). Expression of the transfected proteins in stable clones was confirmed by immunoblotting using an Ab against the extracellular domain of TrkB (anti-TrkB), which recognizes all three pro-teins, and one against the intracellular domain (carboxyl 20 amino acids) of ␤4 (505), which recognizes only the TrkB␤4 chimera (Fig. 1B). To confirm cell surface localization of these expressed proteins, we performed biotinylation followed by immunoprecipitation with streptavidin-agarose and immunoblotting for TrkB. As shown in Fig. 1C, the TrkB␤4 chimera, as well as the fulllength TrkB and TrkBextra proteins, were all immunoprecipitated, indicating that they are localized at the cell surface. The TrkB␤4 chimera, unlike the intact ␣6␤4 integrin, should not associate with the ␣6 subunit, because the association of the ␣ and ␤ subunits of integrin heterodimers occurs through their extracellular domains (31). Indeed, as shown in Fig. 1D, TrkB␤4 was not detected by immunoblotting in ␣6 immunoprecipitates, although the full-length ␤4 subunit was detected. Moreover, we were unable to detect any association of the TrkB␤4 chimera with tetraspanins (data not shown), a result that is consistent with the observation that this association occurs though the ␣6 subunit of ␣6␤4 (32,33).
The neurotrophin, BDNF, not only binds to the TrkB receptor but also binds to the non-selective p75 neurotrophic receptor (p75NTR) with a lower affinity (34). Expression of p75NTR has been observed in several non-neuronal cell types and cancers (35,36), including melanoma cells (37). Given that the MDA-MB-435 cell line may possess properties of melanoma cells (26), we examined whether these cells express p75NTR, to exclude the possibility of signaling through p75NTR following BDNF treatment. Expression of p75NTR mRNA was examined by reverse transcription-PCR. Although the melanoma cell line, A875, expresses a significant amount of mRNA for p75NTR, the MDA-MB-435 TrkB␤4 cell line did not express mRNA for this receptor (Fig. 1E).
Dimerization of the ␤4 Intracellular Domain Induces SHP-2 Binding and Activation-A recent report demonstrated that signaling through the Met receptor can promote the association of the tyrosine phosphatase SHP-2 with the ␤4 intracellular domain (29). A key issue that has not been addressed, however, is whether clustering of the ␤4 intracellular domain by itself is sufficient to induce SHP-2 binding in the absence of growth factor stimulation. The TrkB␤4 chimera provided an ideal model system for testing this issue. SHP-2 was immu- noprecipitated from MDA-MB-435 cells expressing TrkB␤4, and the presence of TrkB␤4 in these immunoprecipitates was assessed by immunoblotting. As shown in Fig. 2A, BDNF stimulation resulted in a significant increase in the amount of TrkB␤4 that co-immunoprecipitated with SHP-2. To verify this finding, we performed the reverse immunoprecipitation to determine whether SHP-2 co-immunoprecipitated with TrkB␤4. Indeed, SHP-2 co-immunoprecipitated with TrkB␤4, but the relative amount did not increase significantly upon BDNF stimulation. We attribute this result to the fact that the expression of TrkB␤4 is much higher than that of SHP-2 and that only a relatively small fraction of TrkB␤4 engages SHP-2. Together, the above studies demonstrate that the ␤4 intracellular domain, in the absence of any cooperative signaling from growth factor receptors, is able to bind SHP-2 and that this effect is enhanced significantly by dimerization of the ␤4 intracellular domain.
Subsequently, we assessed whether SHP-2 itself is activated following BDNF treatment of cells expressing TrkB␤4. Phosphorylation of SHP-2 on Tyr-542 activates the molecule by releasing it from its basal conformation that inhibits its phosphatase activity (38,39). BDNF stimulation of cells expressing the TrkB␤4 chimera resulted in an increase in SHP-2 (Tyr-542) phosphorylation that peaked at 15 min but remained higher than that observed for unstimulated cells for 60 min (Fig. 2B). No increase in SHP-2 phosphorylation was observed in cells expressing TrkBextra (Fig. 2B) or vector alone (data not shown) indicating that the observed effects could be attributed to the ␤4 intracellular domain and not to nonspecific effects of BDNF treatment. Interestingly, basal levels of phosphorylated SHP-2 (0-min BDNF) were slightly higher in the cells expressing TrkB␤4 as compared with cells expressing TrkBextra (Fig. 2B). This result suggests that expression of the ␤4 intracellular domain alone may allow for low levels of activation of SHP-2 and is consistent with the observation that some binding of ␤4 to SHP-2 occurs in the absence of BDNF ( Fig. 2A). To demonstrate that the intact ␣6␤4 integrin can activate SHP-2 in the absence of exogenous growth factor stimulation, we examined SHP-2 phosphorylation in MDA-MB-231 cells, which express significant amounts of endogenous ␣6␤4. As shown in Fig. 2C, the phosphorylation of SHP-2 on Tyr-542 increased over time in response to ␣6␤4 ligation. These results indicate that SHP-2 activity can be regulated by engagement of this integrin independently of growth factor signaling.
Dimerization of the ␤4 Intracellular Domain Activates Src Family Kinases-To identify signaling events other than SHP-2 activation that are stimulated following dimerization of the TrkB␤4 chimera, MDA-MB-435 cells expressing this chimera were treated with BDNF for varying times and the global pattern of tyrosine phosphorylation was assessed by immunoblotting using a phospho-tyrosine-specific Ab. TrkBextra cells, which lack the ␤4 intracellular domain, were used as a control. The most noticeable effect of BDNF stimulation was a time-dependent increase in the intensity of a phosphoprotein that migrated between 50 and 60 kDa (Fig. 3A). This band was not detected in the control TrkBextra cells (Fig.  3A) or cells expressing vector alone (data not shown).
The Src family kinases (SFKs), proteins with molecular masses of ϳ60 kDa, are key mediators of integrin signaling (40,41), including signaling through ␤4 (18,29,42,43). Thus, we hypothesized that the 60-kDa phospho-protein detected in Fig.  3A could be an SFK. Activation of SFKs is coordinated by two tyrosine residues: dephosphorylation of the inhibitory Tyr-529, which allows for a conformational change in the molecule exposing the activation loop and phosphorylation of Tyr-418 within this loop that results in Src-kinase activation (44). BDNF treatment of cells expressing TrkB␤4 resulted in a time-dependent increase in phosphorylation of Tyr-418 and a concomitant decrease in the phosphorylation of Tyr-529 (Fig. 3B). No significant change in the phosphorylation of either site was observed in response to BDNF stimulation of TrkBextra cells (Fig. 3B) or cells expressing vector alone (data not shown), substantiating the conclusion that the observed increase in Src activation can be attributed to the ␤4 intracellular domain. As observed for SHP-2 (Tyr-542), the basal level of Tyr-418 phosphorylation (0 min BDNF) was higher in cells expressing TrkB␤4 than in the TrkBextra cells, but the basal level of Tyr-529 phosphorylation was slightly higher in the TrkBextra cells (Fig. 3B). These results suggest that expression of the ␤4 intracellular domain by itself results in some activation of an SFK. Together, these findings imply that the ␤4 intracellular domain has the intrinsic ability to mediate the activation of a SFK, and that this activation is enhanced following its dimerization.
Previous work had suggested the involvement of SHP-2 in Src activation (45). To assess the potential involvement of SHP-2 in regulating Src activation by the ␤4 intracellular domain, we evaluated the ability of calpeptin, an inhibitor of the catalytic activity of SHP-2, to impede Src activation in response to dimerization of the TrkB␤4 chimera. Calpeptin was originally characterized as an inhibitor of the calpain proteases, but was more recently shown to also inhibit SHP-2 function (46 -48). As shown in Fig. 3C, calpeptin prevented the phosphorylation of Tyr-418 on Src that is induced by BDNF stimulation of cells expressing TrkB␤4. To confirm that the effects of calpeptin were the result of inhibiting SHP-2 specifically, the BDNF-dependent activation of Src in these cells was also examined in the presence of ALLN, an inhibitor of calpain activity that also inhibits the proteasome (49). ALLN did not block phosphorylation of Tyr-418 on Src in response to dimerization of TrkB␤4 with BDNF (Fig. 3C).
We also confirmed the ability of the intact ␣6␤4 receptor to activate Src. For this purpose, we examined Src activation in response to Abmediated ligation of either ␣6␤1, in mock transfected MDA-MB-435 cells, or ␣6␤1 and ␣6␤4 in MDA-MB-435 cells that had been transfected with the full-length ␤4 integrin subunit. Although engagement of ␣6␤1 with an ␣6 Ab stimulated Src activation, the level of activation in response to ␣6␤4 ligation was markedly higher (Fig. 3D). In the presence of calpeptin, ␣6␤4dependent activation of Src was inhibited, whereas ␣6␤1-dependent activation of Src was unaffected by the inhibition of SHP-2 (Fig. 3D). These results indicate that the mechanisms by which ␣6␤1 and ␣6␤4 activate Src are distinct. The ␣6␤4-dependent activation of Src was also examined in the presence of ALLN and a peptide derived from calpastatin that is a calpain-specific inhibitor (50). Neither ALLN nor the calpastatin peptide inhibited Src activation in response to ligation of the ␣6␤4 integrin, demonstrating that the reduction in Src activation observed using calpeptin resulted from SHP-2 inhibition (Fig. 3E). Furthermore, stable expression of an shRNA against SHP-2 to reduce SHP-2 protein levels also inhibited Src activation significantly in response to antibody-mediated ligation of ␣6␤4 (Fig. 3F). MDA-MB-435 cells expressing either TrkBextra or TrkB␤4 were treated with 100 ng/ml BDNF for the indicated periods of time. A and B, equal amounts of total protein from cell extracts were immunoblotted with a phospho-tyrosine antibody (A) or with an Ab against tyrosine 418-phosphorylated Src (pSrc (Tyr-418)), tyrosine 529-phosphorylated Src (pSrc (Tyr-529)), total Src (Src), or actin (B), as described. C, cells were treated with BDNF and either calpeptin (50 g/ml) or ALLN (50 M), as indicated, and equal amounts of total protein were immunoblotted for ␤4 (505, top panel), tyrosine 418-phosphorylated Src (Src (Tyr-418), middle panel), or total Src (lower panel). D, MDA-MB-435 cells that had been transfected with the ␤4-integrin subunit (␤4) or empty vector (mock) were incubated without (IgG) or with an ␣6-specific antibody (␣6) in either the absence or presence of 50 g/ml calpeptin (CP). Cells were allowed to adhere to anti-mouse IgG-coated plates for 30 min. Cell extracts that contained equivalent amounts of total protein were immunoblotted for tyrosine 418-phosphorylated Src (pSrc (Tyr-418), top panel), or total Src (lower panel). E, MDA-MB-435 cells transfected with the ␤4 subunit were incubated without (IgG) or with an ␣6-specific Ab (␣6) in either the absence or presence of 50 g/ml calpeptin (CP), 50 M ALLN or 5 M calpastatin peptide (CS) and allowed to adhere to anti-mouse IgG coated plates for 30 min. Equal amounts of total protein from cell extracts were then immunoblotted for tyrosine 418-phosphorylated Src (pSrc (Tyr-418), top panel), or total Src (lower panel). F, MDA-MB-435 cells transfected with the ␤4 subunit and either empty vector alone (pSUPER), a vector expressing a scrambled shRNA (scr), or a vector expressing an shRNA for SHP-2 (SHP-2) were incubated without (IgG) or with an ␣6-specific antibody (␣6). Cells were allowed to adhere to anti-mouse IgG-coated plates for 30 min. Cell extracts that contained equivalent amounts of total protein were immunoblotted for tyrosine 418-phosphorylated Src (pSrc (Tyr-418), top panel), total Src (middle panel), or total SHP-2 (bottom panel). OCTOBER 12, 2007 • VOLUME 282 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 30327
Dimerization of the ␤4 Intracellular Domain Is Unable to Promote Activation of Key Downstream Signaling Pathways Associated with Ligation of the ␣6␤4 Integrin-Ab-mediated ligation of ␣6␤4 in MDA-MB-435 cells that express the intact ␣6␤4 integrin resulted in a time-dependent increase in the activation of both Akt and Erk1/2 (Fig. 4A). This finding confirms previous studies, which concluded that the expression and ligation of the ␣6␤4 integrin result in a robust activation of the PI3-K/Akt and MAPK signaling pathways (21,51). As shown in Fig. 4B, however, dimerization of the TrkB␤4 chimeric receptor with BDNF failed to induce activation of either Akt or Erk1/2 as assessed by phospho-immunoblotting. Dimerization of the full-length TrkB receptor in the same cells, however, resulted in activation of both pathways (Fig. 4A), in accord with previous studies (52,53). These data indicate that expression and dimerization of the ␤4 intracellular domain are not sufficient to promote activation of Akt and Erk1/2.
The ␤4 Intracellular Domain Itself Does Not Amplify HGF Signaling and Function-The chimeric TrkB␤4 receptor provided an opportunity to evaluate the role of the ␤4 intracellular domain in amplifying growth factor-mediated signaling and function in carcinoma cells. For this purpose, we focused on HGF/Met signaling, because cooperation between Met and ␣6␤4 signaling in carcinoma cells has been established and a significant amount of these data were derived from the use of MDA-MB-435 cells (29). Initially, we assessed the ability of HGF to stimulate tyrosine phosphorylation of the ␤4 intracellular domain in comparison to dimerization of this domain with BDNF. As shown in Fig. 5A, HGF stimulation resulted in a substantially greater increase in ␤4 phosphorylation than did BDNF stimulation. Interestingly, no synergic increase in phosphorylation was evident when both factors were added together (Fig.  5A). Based on these data, we compared the ability of HGF and BDNF to promote the association of ␤4 with SHP-2. The data shown in Fig.  5B reveal that HGF stimulation resulted in only a slight increase in ␤4/SHP-2 association in comparison to BDNF stimulation, and no synergy was evident when both factors were added together. These data argue that dimerization of the ␤4 intracellular domain is a much more potent inducer of SHP-2 association than is HGF stimulation. We  also assessed the ability of the TrkB␤4 chimera to amplify HGFmediated activation of Akt and Erk1/2 using TrkBextra cells and cells that express the TrkB␤4 chimera. Although HGF was able to stimulate activation of both kinases in TrkBextra cells as determined by phospho-immunoblotting, expression of the ␤4 intracellular domain did not enhance this activation (Fig. 5C). Moreover, HGF-induced activation of Erk1/2 in HCC1937 breast carcinoma cells, which endogenously express ␤4, was similar in cells expressing an shRNA against ␤4 as compared with cells expressing a control shRNA against GFP (Fig. 5D).
The signaling data, as well as previous reports on the ability of ␤4 to enhance HGF-mediated migration (22,54), prompted us to examine the ability of HGF to stimulate the chemotactic migration of cells that expressed either the TrkB␤4 chimera or TrkBextra. As shown in Fig. 5E, no difference was seen in the ability of HGF to stimulate the migration of TrkBextra cells as compared with cells that express the TrkB␤4 chimera. The conclusion from these results is that expression of the ␤4 intracellular domain itself is not sufficient to enhance HGF-mediated migration.

DISCUSSION
This study was predicated on numerous reports suggesting that the atypical intracellular domain of the ␤4 integrin subunit exhibits considerably more autonomy in signaling and function than other integrin intracellular domains (5,55). The data reported in this study reveal that the ␤4 intracellular domain does indeed possess intrinsic signaling function but that it also is unable by itself to transduce all signaling events and functions that have been attributed to ␣6␤4. The ␤4 intracellular domain can function independently of the ␣6 subunit, tetraspanins, and growth factor signaling to bind and activate SHP-2, as well as to activate Src. In contrast, expression and dimerization of this intracellular domain are not sufficient to activate two of the major kinases known to be involved in ␣6␤4 signaling: Akt and Erk1/2 (21,51). Also, expression of this intracellular domain does not enhance HGF-mediated activation of Akt and Erk1/2 or HGF-stimulated migration, as previously reported for the intact ␣6␤4 receptor (22). One hypothesis that can be derived from these data is that the association of the ␤4 subunit with the ␣6 subunit and tetraspanins is necessary for complete ␣6␤4 signaling and function.
Our data provide insight into the relationship between ␣6␤4 and growth factor signaling. The ability of the ␤4 intracellular domain to associate with SHP-2 in response to Met signaling has been reported recently (29). In this study, the conclusions were drawn that SHP-2, when bound to ␤4, enhances Src activation, which then phosphorylates Gab1 leading to MAPK activation. This study examined the association between the ␤4 subunit and SHP-2 in response to HGF stimulation, but it did not address the independent functions of the ␤4 subunit. Our data using the chimeric TrkB␤4 receptor confirm the ability of SHP-2 to associate with ␤4. A critical difference, however, is our observation that dimerization of the ␤4 intracellular domain with BDNF promotes SHP-2 association to a greater extent than does transactivation of ␤4 by HGF stimulation. In fact, we detected only a slight increase in SHP-2/␤4 association in response to HGF stimulation. These data argue that SHP-2 binding and activation are intrinsic functions of the ␤4 intracellular domain and that they can occur independently of growth factor signaling. Although the mecha-nism for SHP-2-mediated activation of Src has not been established in our model system, recent work suggests that SHP-2 may regulate Src indirectly via de-localization of Csk, which is responsible for tyrosine phosphorylation of Src on its inhibitory site, and not via direct dephosphorylation of this inhibitory site by SHP-2 itself (45).
Our data on ␤4 association with SHP-2, along with the failure of the intracellular domain of ␤4 to enhance growth factor-mediated activation of Akt and Erk1/2 in MDA-MB-435 cells, suggest that the role of the ␤4 intracellular domain as an adaptor for growth factor signaling needs to be evaluated more rigorously. This conclusion is substantiated by the inability of full-length ␤4 to enhance HGF-induced activation of Erk1/2 in HCC1937 cells.
An important question that arises from our study is why the ␤4 intracellular domain itself is unable to activate either Akt or Erk1/2, kinases that have been implicated in ␣6␤4 signaling. In this respect, our data differ from other studies that have suggested that activation of these kinases is intrinsic to the ␤4 intracellular domain (11,22). A likely explanation for our data is that the TrkB␤4 chimera is unable to associate with tetraspanins, integral membrane proteins that have been implicated in a wide variety of functions, including cancer progression (56,57). The ␣6␤4 integrin, as well as the ␣6␤1 and ␣3␤1 integrins, interact directly with several tetraspanins, and this interaction occurs through their ␣-subunits (57). Moreover, the concept of a "tetraspanin web" has been proposed, which is a membrane microdomain distinct from lipid rafts and formed by tetraspanin-tetraspanin interactions (23,58). Such webs have been shown to contain not only tetraspanins and integrins, but other signaling molecules as well. Disruption of this tetraspanin web through suppression of tetraspanin expression significantly impairs the function of integrins, such as ␣6␤4, that associate with these domains. For example, CD151 null mice have defects in both angiogenesis and epithelial wound healing, and cells derived from these mice are deficient in signaling in response to laminin adhesion (59,60). Thus, the argument can be made that the incorporation of ␣6␤4 into tetraspanin webs, or possibly other membrane microdomains, is necessary for realizing its full signaling potential. Interestingly, a previous study on endothelial cells using an interleukin-2 receptor/␤4 intracellular domain chimera observed that the ␤4 intracellular domain can localize to "fibrillar" structures on the basal surface and associate with HD-1/plectin in the absence of ␣6 (61). It appears, therefore, that the localization and signaling properties of the ␤4 intracellular domain can be regulated differentially. The possibility also needs to be considered that the ␤4 extracellular domain, possibly through its ability to interact with ␣6, is required to allow conformational changes in the ␤4 intracellular domain that facilitate signaling. Finally, BDNFinduced dimerization of TrkB␤4 may not be sufficient to mimic the lateral clustering of ␣6␤4 molecules completely following ligand binding. However, this possibility is diminished by our observation that Ab-mediated clustering of TrkB␤4 did not enhance Src activation to levels observed following Ab-mediated clustering of ␣6␤4, or promote activation of Akt and Erk1/2 (data not shown).
Our data provide some insight into how growth factor receptor signaling impacts ␣6␤4. A current hypothesis is that the ␤4 ␤4 Integrin Intracellular Domain Signaling OCTOBER 12, 2007 • VOLUME 282 • NUMBER 41 intracellular domain functions as an adaptor for growth factor receptor signaling and that transactivation of ␤4 by such signaling results in ␤4 phosphorylation, the recruitment of signaling intermediates such as SHP-2 and the consequent activation of downstream signaling pathways (29,62). In agreement with this hypothesis, we observed that HGF stimulated a marked increase in the tyrosine phosphorylation of the TrkB␤4 chimera. This increase in phosphorylation, however, was not manifested in a concomitant increase in SHP-2 binding or activation of either Akt or Erk1/2. Nor, was the expression of the ␤4 intracellular domain able to enhance HGF-mediated migration. These findings are in contrast to the reports that expression of the ␤4 intracellular domain tagged with c-myc in MDA-MB-435 cells is able to amplify HGF-mediated signaling and function (22,29). Given that the c-myc/␤4 receptor does not associate with ␣6, we cannot explain the discrepancy in the two sets of data. However, our data demonstrate clearly that the ␤4 intracellular domain is not simply a passive amplifier of growth factor signaling.