The Neurofibromatosis 2 Protein, Merlin, Regulates Glial Cell Growth in an ErbB2- and Src-Dependent Manner

Cell culture.Forebrain glial cell cultures from postnatal day 3 Nf2^flox/flox mice (18) were generated as previously described (24, 49). Briefly, forebrains were isolated and enzymatically digested with 0.25% trypsin for 10 min at 37°C. Cells were then placed in modified Eagle's medium with 10% fetal bovine serum and grown for 2 weeks to generate cultures composed of >97% GFAP-immunoreactive cells (glia). Adenovirus type 5 (Ad5) viruses, namely, Ad5-LacZ and Ad5-Cre (University of Iowa Gene Transfer Core, Iowa City, IA), were used to produce control (WT) and Nf2-deficient (Nf2^−/−) glia, respectively, according to protocols described previously in our laboratory (10, 49). Merlin loss was confirmed by immunoblot analysis using a rabbit polyclonal NF2 antibody (NF2 C-18; Santa Cruz Biotechnology, Santa Cruz, CA) (1:1,000 dilution) or the laboratory-generated WA30 antibody (50) (1:2,000 dilution).

Pharmacological inhibitors.Glial cell cultures were treated with the following experimentally determined concentrations of inhibitors for 24 h prior to all analyses: Src inhibitor PP2, 0.5 nM (Calbiochem, San Diego, CA); ErbB2 inhibitor AG825, 4.0 μg/ml (Calbiochem); FAK/paxillin inhibitor echistatin, 2.0 μg/ml (Sigma, St. Louis, MO).

shRNA constructs and lentiviral delivery.Mouse gene-specific lentiviral plasmids for Src (GenBank accession no. NM_009271; TRCN0000023595 and TRCN0000023598), Fak (NM_007982), paxillin (NM_133915), and ErbB2 (NM_109715) small hairpin RNAs (shRNAs) were commercially purchased (Sigma). Corresponding lentiviruses were produced as previously described (54), and the most effective silencing construct was selected for further study. Briefly, 293T cells were transfected with Src.shRNA, FAK.shRNA, Pxn.shRNA, and ErbB2.shRNA lentiviral plasmids or a control vector (pLKO1-GFP plasmid; courtesy of Sheila Stewart, Washington University), using Fugene HD (Roche, Mannheim, Germany). Lentiviruses were harvested at 48 and 72 h posttransfection. WT and Nf2^−/− glia were infected for a total of 48 h. WT and Nf2^−/− glia were serum starved for 24 h prior to each experiment.

Thymidine incorporation assay.Glial cells were seeded at 5 × 10⁴ cells per well for 24 h of serum starvation. [³H]thymidine (1 μCi per well) was added in serum-free medium, and the number of counts per minute was determined after 24 h.

Cell attachment, apoptosis, and cell adhesion.Cell attachment was measured by seeding 5 × 10⁴Nf2^flox/flox glial cells treated with Ad5-LacZ (WT) and Ad5-Cre (Nf2^−/−) into 96-well plates precoated with 10 μg/ml fibronectin (Gibco) following 24 h in serum-free medium. After 4 hours, the wells were washed with 1× phosphate-buffered saline (PBS) and incubated in 0.5% crystal violet for 30 min. The number of attached cells was determined by extraction in 1% sodium dodecyl sulfate (SDS) overnight and spectrophotometric analysis at 540 nm.

Apoptosis was measured by seeding WT and Nf2^−/− glial cells into 24-well plates, followed by preincubation in serum-free medium in the presence and absence of 5 nM staurosporine (Sigma) for 24 h. Cells were then immunostained with cleaved caspase-3 antibody (1:500 dilution; Cell Signaling Technology) and 1% bis-benzimide-PBS solution. The number of apoptotic cells was determined by the fraction of caspase-3-positive cells relative to the total number of DAPI (4′,6-diamidino-2-phenylindole)-positive cells. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) was also used to detect apoptosis levels in vivo. TUNEL labeling of Nf2^GFAPCKO and Nf2^flox/flox mouse brain sections was performed using a commercially available fluorescein-based in situ cell death detection kit (Roche) according to the manufacturer's instructions. The number of TUNEL-positive cells in the hippocampus was determined by direct cell counting.

The formation of adherens junctions was assessed by β-catenin immunostaining as previously described (28). WT and Nf2^−/− glial cells were grown to confluence and serum starved for 24 h. The cells were then permeabilized and fixed by incubation in 3% paraformaldehyde-0.1% Triton X-100 in PBS for 10 min. Cells were blocked in PBS plus 10% goat serum and incubated with anti-β-catenin antibodies (1:200 dilution; Chemicon) and then with secondary antibody (1:200 dilution; Alexa). A 1.0% bis-benzimide solution was used to counterstain nuclei. Images were acquired on an inverted fluorescence microscope (Eclipse TE300 inverted microscope; Nikon, Japan) equipped with a digital camera (Optronics, Goleta, CA).

Merlin reexpression.To reintroduce merlin into WT and Nf2-deficient glia, murine stem cell virus (MSCV) retroviral transduction was used to express either WT merlin or an inactive mutant merlin protein containing a nonfunctional patient missense mutation (L64P), as previously described (49, 62). MSCV-merlin and MSCV-merlin-L64P viral constructs were generated by cloning the human NF2 cDNA into the MSCV-green fluorescent protein (MSCV-GFP) vector. MSCV-merlin, MSCV-merlin-L64P, and the control (empty MSCV-GFP vector) retrovirus were generated in 293T cells. Transduced glia were evaluated by fluorescence microscopy using a microscope equipped with a digital camera. More than 90% of glial cells displayed GFP expression, indicative of effective transduction. Merlin expression was confirmed by immunoblotting with a commercially available merlin antibody (NF2 C-18; 1:1,000 dilution).

Immunoblotting.Untreated or reagent-treated primary glia and brains were lysed in NP-40 lysis buffer (50 mM Tris [pH 7.0], 150 mM NaCl, 0.5% NP-40, 1 mM dithiothreitol [DTT]) with protease inhibitors (leupeptin, benzamidine, aprotinin, and phenylmethylsulfonyl fluoride). Proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride membranes (Millipore, Billerica, MA) prior to detection with antibodies against phosphorylated Src-Y416, FAK-Y576/577 (Cell Signaling Technology, Beverly, MA) (1:1,000 dilution), ErbB2-Y877 (Abcam, Cambridge, MA) (1:1,000 dilution), and paxillin-Y118 (Cell Signaling Technology, Beverly, MA) (1:1,000 dilution). Antibodies against total Src, FAK, ErbB2, and paxillin (from the same manufacturers and at the same dilutions as the phospho-specific antibodies), as well as α-tubulin (Sigma) (1:5,000 dilution), were used as controls for equal protein loading and quantitation. Antibodies against phosphorylated MAPK p44/p42, AKT-T308, and EGFR-Y845 as well as total MAPK p44/p42, AKT, and EGFR were purchased from Cell Signaling Technology and used at a 1:1,000 dilution. Densitometric analysis was performed using Gel-Pro Analyzer 4.0 software (MediaCybernetics, Silver Spring, MD) with anti-tubulin or non-phospho-Src, -paxillin, -FAK, and -ErbB2 antibodies as internal loading controls.

Immunoprecipitation.Primary glia were lysed by sonication in precipitation lysis buffer (20 mM Tris [pH 7.5], 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na₃VO₄, 1 mg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride). Antibodies against Src (U.S. Biological) (1:100 dilution) or ErbB2 (Upstate, Temecula, CA) (1:100 dilution) were added to the lysates and incubated with gentle rocking overnight at 4°C. Protein A-agarose beads (Calbiochem) were then added to the mixture and rotated for 4 hours. Beads were washed with precipitation lysis buffer and resuspended in 2× Laemmli buffer. Each sample was then boiled and resolved by SDS-PAGE, followed by immunoblot analysis as described above.

Tyrosine phosphorylation was detected following incubation of cell lysates with agarose-conjugated monoclonal anti-phosphotyrosine (PY20) antibody (Sigma) according to the manufacturer's protocols. Following washes, proteins were eluted by boiling in gel loading buffer and detected by immunoblotting using antibodies against Ack1 (Cell Signaling Technology) (1:1,000 dilution), p130CAS (Sigma) (1:1,000 dilution), CASPR (Santa Cruz Biotechnology) (1:500 dilution), ErbB3 (Cell Signaling Technology) (1:1,000 dilution), and ErbB4 (Cell Signaling Technology) (1:1,000 dilution).

Src and ErbB2 kinase activity assays.Primary glial cells were lysed by sonication in precipitation lysis buffer. Antibodies against Src (Cell Signaling) (1:50 dilution) or ErbB2 (Cell Signaling) (1:50 dilution) were added to the lysates and incubated with gentle rocking overnight at 4°C. Protein G agarose beads (Calbiochem) were then added to the mixture and rotated overnight. Beads were washed and resuspended in 50 μl of 1× kinase buffer containing signal transduction protein Tyr160-biotinylated peptide substrate (1.5 μM; Cell Signaling), 20 μM ATP, and 1.25 μM DTT for Src or FLT3 Tyr589-biotinylated peptide (1.5 μM; Cell Signaling), 20 μM ATP, and 1.25 μM DTT for ErbB2. After 30 min, each reaction was stopped by the addition of EDTA. Twenty-five microliters of each reaction mix was added to a 96-well streptavidin-coated plate containing 75 μl distilled H₂O/well and incubated at room temperature for 60 min. After the plates were washed with PBS-Tween 20 (PBS-T), 100 μl of diluted detection antibody (phosphotyrosine monoclonal antibody [MAb] P-Tyr-100 [1:500 dilution] in PBS-T with 1% bovine serum albumin) was added to each well for 2 h at 37°C. Next, 100 μl of horseradish peroxidase-labeled mouse secondary antibody (1:500 in PBS-T with 1% bovine serum albumin; Cell Signaling) was added, and the plates were incubated at room temperature for 30 min, washed with PBS-T, and developed in 100 μl TMB solution (Cell Signaling) for 10 min at 37°C. Following the addition of stop solution (Cell Signaling), the absorbance was read at 450 nm.

Rac1 activation assay.Rac activation in WT and Nf2^−/− glial cell lysates was determined using a Rac1 activation assay kit (Upstate Biotechnologies, Temecula, CA) according to the manufacturer's instructions, as previously described (17).

In vitro binding assays.A commercially available TnT quick coupled transcription/translation system (Promega, Madison, WI) was used as previously described (20) to transcribe and translate WT merlin (NF2.WT), mutant merlin (NF2.L64P), a C-terminal merlin fragment containing residues 300 to 595 (NF2.C-term), an N-terminal merlin fragment containing residues 1 to 300 (NF2.N-term), a merlin fragment containing residues 300 to 557 (NF2.300-557), and WT Src (Src) in the presence of 2 μCi [³⁵S]methionine (1,000 Ci/mmol at 10 mCi/ml; Perkin-Elmer, Waltham, MA). The first set of experiments were direct binding experiments in which 75% of each TnT preparation containing radiolabeled merlin or Src was incubated with 1.0 μg recombinant His-tagged ErbB2 or Src protein (Invitrogen, Carlsbad, CA) overnight at 4°C. ErbB2.His/merlin, ErbB2.His/Src, and Src.His/merlin reaction mixtures were then incubated with His-Select nickel affinity gel (Sigma) for 4 h at 4°C. The nickel affinity gel was washed and eluted with 2× Laemmli buffer. Bound proteins along with the remaining original 25% of each TnT reaction mixture were then separated by SDS-PAGE and analyzed by autoradiography. Densitometric analysis was performed using Gel-Pro Analyzer 4.0 software (MediaCybernetics), with 25% of each TnT preparation serving as the input control for normalization. The second set of experiments were competition binding assays in which 1.0 μg of recombinant His-tagged ErbB2 protein was incubated with radiolabeled Src in the presence of increasing amounts of nonradiolabeled in vitro TnT-generated merlin or merlin fragments (none or a 1×, 5×, or 10× excess of merlin or merlin fragments relative to Src). Complementary experiments using increasing amounts of nonradiolabeled Src (none or a 1×, 5×, or 10× excess of Src relative to merlin) were also performed. As described above, each reaction mix was incubated with His-Select nickel affinity gel, followed by SDS-PAGE, autoradiography, and densitometric analyses.

In vivo Src inhibition. Nf2 ^GFAPCKO mice were generated by successive intercrossing of Nf2^flox/flox and Nf2^flox/wt; GFAP-Cre mice, and merlin expression in the brain was detected by Western blotting using merlin antibodies (NF2 C-18; 1:1,000 dilution). Ten-day-old Nf2^flox/flox (WT) and Nf2^GFAPCKO (Nf2^−/−) mice (n = 4 per experimental group) received 2 mg/kg of body weight of PP2 (Calbiochem) in dimethyl sulfoxide or the same volume of dimethyl sulfoxide alone by intraperitoneal injection three times a week for 2 weeks, as previously described (15).

Immunohistochemistry.To analyze proliferation in vivo, 7- to 10-day-old Nf2^flox/flox and Nf2^GFAPCKO mice (n = 3) were injected with 5-bromo-2-deoxyuridine (BrdU; Sigma) (50 μg/g of body weight) as previously described (24). After 2 hours, mice were euthanized and their brains were removed. A quarter of each brain was separated for immunoblot analyses, and the remainder of each brain was fixed in 4% paraformaldehyde and paraffin sections prepared. Brain sections were immunolabeled with BrdU (Abcam, Cambridge, MA) (1:100 dilution) and GFAP (Sigma) (1:200 dilution) antibodies followed by Alexa 488 (BrdU; Abcam) (1:200 dilution)- and Cy3 (GFAP; Sigma) (1:500 dilution)-conjugated secondary antibodies. All sections were photographed with a digital camera (Optronics) attached to an inverted microscope (Nikon). The number of BrdU-positive cells in the hippocampus for each set of three littermates was determined by direct cell counting.

To quantitate glial cell numbers in vivo, GFAP immunohistochemistry (Sigma) (1:200 dilution) with Vectastain ABC development (Vector Laboratories, Burlingame, CA) was performed as previously reported (2). The number of GFAP-positive cells in the hippocampus for each set of three littermates was determined by direct cell counting.

Merlin regulates glial cell growth in vitro and in vivo.To study merlin growth regulation in CNS glia, we employed primary neonatal forebrain glial cultures from Nf2 conditional knockout (Nf2^flox/flox) mice, in which the expression of merlin is eliminated by adenoviral delivery of Cre recombinase (Ad-Cre). Following Cre recombinase transduction, Nf2^−/− (Nf2^flox/flox + Ad-Cre) glia consistently demonstrated a >95% decrease in merlin expression compared to Nf2^flox/flox glia transduced with Ad-LacZ virus (WT) (Fig. 1A). As previously reported for other cell types, Nf2^−/− glia exhibited a 2-fold increase in proliferation compared to WT glia by thymidine incorporation assay (mean = 2.3-fold increase) (Fig. 1B). Identical results were obtained using BrdU incorporation in vitro (data not shown). In contrast, Nf2^−/− glia were indistinguishable from WT glia with respect to cell attachment (Fig. 1C), apoptosis (Fig. 1D), and adherens junctions (Fig. 1E).

• Open in new tab
• Download powerpoint

FIG. 1.

Merlin regulates glial cell growth in vitro. (A) Immunoblot analysis of merlin expression in primary glial cell lysates demonstrates a >95% reduction in merlin expression in Nf2^−/− glia (Ad5-Cre treatment) relative to that in WT glia (Ad-LacZ treatment), as assessed by scanning densitometry. α-Tub, α-tubulin. (B) Nf2^−/− glia exhibited a 2.3-fold increase in proliferation relative to WT glia, as determined by [³H]thymidine incorporation. *, P < 0.001 (unpaired t test). (C) WT and Nf2^−/− glial cells were seeded in fibronectin-coated plates, and the number of adherent cells was determined using a spectrophotometric assay. No differences in cell attachment were observed between WT and Nf2^−/− glial cells. OD, optical density. (D) WT and Nf2^−/− glial cells were immunostained with cleaved caspase-3 antibody in the presence and absence of staurosporine (5 nM). No differences in apoptosis were observed between WT and Nf2-deficient glial cells in either the presence or absence of staurosporine. (E) β-Catenin staining along cell contact junctions in WT and NF2^−/− glial cells revealed similar patterns by immunofluorescence microscopy.

To determine whether our findings using Nf2-deficient glia in vitro were also observed in the intact brain in vivo, we analyzed glial cell proliferation in Nf2^GFAPCKO mice. Nf2^GFAPCKO mice were generated by the successive intercrossing of Nf2^flox/flox mice with Nf2^flox/wt; GFAP-Cre mice to yield mice with Nf2 loss in GFAP-expressing cells (glia). Nf2^GFAPCKO mice were born at the expected Mendelian frequencies, were phenotypically normal, and demonstrated a >90% decrease in merlin expression in the brain compared to Nf2^flox/flox mice (Fig. 2A). Despite efficient Nf2 inactivation in glia throughout the neuroaxis, these mice did not develop nervous system tumors, even after 15 months of age (n = 15 mice to date).

• Open in new tab
• Download powerpoint

FIG. 2.

Merlin regulates glial cell growth in vivo. (A) Immunoblot analysis of merlin expression in Nf2^flox/flox and Nf2^GFAPCKO mouse brain lysates also demonstrates a >90% reduction in merlin expression in Nf2^GFAPCKO mice, as assessed by scanning densitometry. α-Tub, α-tubulin. (B) Representative immunofluorescence photomicrographs of hippocampi from BrdU-injected Nf2^flox/flox and Nf2^GFAPCKO mice labeled with both BrdU (green) and GFAP (red) antibodies. Nf2^GFAPCKO mouse brains had 2.2-fold more BrdU-positive cells than Nf2^flox/flox mouse brains. *, P < 0.001 (unpaired t test). (C) Representative photomicrographs of hippocampi from Nf2^flox/flox and Nf2^GFAPCKO mice labeled with GFAP antibodies. Nf2^GFAPCKO mouse brains had a 2.4-fold increase in the number of GFAP-expressing cells compared to Nf2^flox/flox mouse brains. *, P = 0.001 (unpaired t test). (D) Representative photomicrographs of in vivo TUNEL staining of hippocampi from Nf2^flox/flox and Nf2^GFAPCKO mouse brains, revealing no significant differences in the number of apoptotic cells between Nf2^GFAPCKO and Nf2^flox/flox mice.

To measure glial cell proliferation in vivo, postnatal day 10 Nf2^GFAPCKO pups were injected with BrdU, and the numbers of BrdU- and GFAP-positive glia were quantitated. Consistent with our findings in vitro, we observed a 2-fold increase in the numbers of BrdU-positive cells and GFAP-expressing cells in Nf2^GFAPCKO mice compared to those in Nf2^flox/flox controls (mean = 2.2- and 2.4-fold increases, respectively) (Fig. 2B and C). No in vivo changes in apoptosis were observed in the hippocampus for Nf2^GFAPCKO mice compared to Nf2^flox/flox controls (Fig. 2D).

To demonstrate that the observed increase in cell proliferation was directly related to merlin expression, we restored merlin expression in Nf2-deficient glia by MSCV retroviral transduction. For these experiments, we reintroduced either WT merlin (NF2) or an inactive merlin protein containing a nonfunctional patient missense mutation (NF2.L64P) (4) into WT and Nf2^−/− glia. Following transduction, the levels of merlin expression in Nf2^−/− glia were similar to those observed in WT glia (Fig. 3A). Consistent with the notion that merlin directly regulates cell growth in glia, the reintroduction of WT but not mutant merlin reduced the proliferation of Nf2^−/− glia to the levels observed in WT glia (Fig. 3B). No effect of merlin expression was observed on WT glial cell proliferation. Collectively, these results demonstrate that merlin is a direct regulator of glial cell growth in vitro and in vivo.

• Open in new tab
• Download powerpoint

FIG. 3.

Expression of WT but not mutant merlin reduces Nf2^−/− glial cell proliferation in vitro. (A) Immunoblot analysis of merlin expression in WT and Nf2^−/− glia infected with empty MSCV, MSCV containing WT merlin (MSCV.NF2), or MSCV containing mutant merlin (MSCV.NF2.L64P). α-Tubulin (α-Tub) protein expression was used as an internal control for equal protein loading in all Western blot analyses. (B) Merlin reintroduction reduces the increased proliferation observed in Nf2^−/− glia to WT levels, whereas the reintroduction of mutant (L64P) merlin has no effect. *, P < 0.003 (unpaired t test).

Merlin regulates glial cell growth in a Src-dependent fashion.Previous studies using different primary cell types and established cell lines have shown that merlin regulates several distinct signaling pathways. To identify the signaling pathway responsible for merlin growth regulation in glia, we employed activity assays and activation (phospho-)-specific antibodies to determine which of these previously implicated NF2-associated signaling pathways were deregulated in Nf2^−/− glia. In contrast to previous reports using non-CNS cell types, no changes in Rac1, MAPK, and AKT activation were observed in Nf2^−/− compared to WT glia (Fig. 4A). However, we found that Nf2^−/− glia exhibited hyperphosphorylation of the Src nonreceptor tyrosine kinase oncoprotein on tyrosine 416, a residue located within the Src kinase domain activation loop (45) (Fig. 4B). No changes in Src phosphorylation were seen on another common activation site (tyrosine 215) or on the deactivation site (tyrosine 527). We also examined Src phosphorylation in lysates from Nf2^GFAPCKO mice and found that Src tyrosine 416 phosphorylation was likewise increased in Nf2^GFAPCKO mice relative to that in control mice in vivo (Fig. 4C). To directly demonstrate that Src activity was increased in Nf2^−/− glia, we assayed Src activity following Src immunoprecipitation. Similar to the results obtained with the Src Y416 phospho-antibody, we observed a 4.7-fold increase in Src activity in Nf2^−/− glia relative to that in their WT counterparts (Fig. 4D). Finally, to demonstrate that merlin directly regulates Src activity, we assessed Src Y416 phosphorylation following the reexpression of merlin. In these experiments, WT but not mutant (NF2.L64P) merlin reexpression in Nf2^−/− glia resulted in a reduction of Src Y416 phosphorylation to the levels observed in WT glia (Fig. 4E).

• Open in new tab
• Download powerpoint

FIG. 4.

Merlin regulates Src activation in Nf2^−/− glial cells. (A) Rac1 activation assay shows similar Rac1 activities in WT and Nf2^−/− glial cell lysates. Immunoblot analyses using phospho-specific MAPK and AKT antibodies demonstrated no change in MAPK or AKT activation in Nf2^−/− glia compared to WT glia in vitro. Total Rac1, MAPK, and AKT expression levels were used as internal controls for equal protein loading. α-Tub, α-tubulin. (B) Immunoblot analysis of Src activation in WT and Nf2-deficient (Nf2^−/−) glia, using phospho-specific Src antibodies (specific for phosphorylation at Y416, Y215, and Y527). Nf2^−/− glia exhibited a 6.7-fold increase in Src activation at tyrosine 416 compared to WT glia, with no change in Src phosphorylation at tyrosine 215 or tyrosine 527. (C) Nf2^GFAPCKO mouse brains also show increased activation of Src (2.9-fold) compared to Nf2^flox/flox mouse brains, as assessed by immunoblot analyses and scanning densitometry. (D) In a Src kinase activity assay, Nf2^−/− glia exhibited a 4.7-fold increase in Src activity relative to WT glia. *, P < 0.001 (unpaired t test). OD, optical density. (E) WT but not mutant merlin reduces Src activation in Nf2^−/− glia.

To determine whether Src hyperactivation resulting from merlin loss was responsible for the increased proliferation seen in Nf2-deficient glia, we treated WT and Nf2^−/− cells with PP2, a potent inhibitor of the Src family of kinases (23). PP2 treatment resulted in a >90% reduction in Src activity, as assessed using Src phospho-specific antibodies (Fig. 5A), and was sufficient to reduce the proliferation of Nf2^−/− glia to WT levels (Fig. 5B). No effect of PP2 treatment on WT glial cell proliferation was observed. To complement these findings, we utilized shRNA interference to silence Src expression. Nf2^−/− glia infected with Src.shRNA lentivirus exhibited a >90% reduction in Src expression, as assessed using total and Src activation-specific antibodies (Fig. 5C), and similar to the case for PP2 treatment, the shRNA decreased Nf2^−/− glial cell proliferation to WT levels (Fig. 5D). Similar results were obtained using a second, independent Src.shRNA lentivirus (data not shown). In all cases, no effect of Src.shRNA on WT glial cell proliferation was observed. Together, these results demonstrate that merlin regulation of glial cell growth requires Src Y416 phosphorylation and activation.

• Open in new tab
• Download powerpoint

FIG. 5.

Src inhibition restores Nf2^−/− glial cell proliferation to WT levels in vitro. (A) The PP2 Src inhibitor blocks Src activation in Nf2^−/− glia. (B) PP2 treatment eliminates the growth advantage observed in Nf2^−/− glia, as measured by [³H]thymidine incorporation. *, P < 0.001 (unpaired t test). (C) Immunoblot analysis of Src expression and activation in WT and Nf2^−/− glia following infection with Src.shRNA or vector (pLK01-GFP) demonstrates >90% inhibition of Src expression and activation after Src.shRNA treatment, as assessed by scanning densitometry. Total Src or tubulin (α-Tub) protein expression was used as an internal control for equal protein loading. (D) Src.shRNA treatment eliminates the growth advantage observed in Nf2^−/− glia, as measured by [³H]thymidine incorporation *, P < 0.001 (unpaired t test).

Src-mediated merlin regulation of glial cell growth requires FAK and paxillin.To determine how Src regulates glial cell growth, we next examined the activation of known Src downstream effector molecules, including FAK, paxillin, p130CAS, Ack1, and CASPR. While we found no change in the phosphorylation status of p130CAS, Ack1, and CASPR in Nf2^−/− versus WT glia (Fig. 6A), we found increased FAK tyrosine 576/577 and paxillin tyrosine 118 phosphorylation in Nf2^−/− glia relative to their WT counterparts (Fig. 6B, left panels). Similarly, immunoblot analysis of Nf2^GFAPCKO mouse brain lysates demonstrated increased FAK and paxillin phosphorylation relative to that in control littermates (Fig. 6B, right panels).

• Open in new tab
• Download powerpoint

FIG. 6.

Merlin regulates phosphorylation of the Src effectors FAK and paxillin. (A) Immunoblot analyses of WT and Nf2^−/− glial lysates following phosphotyrosine (pTyr) affinity binding with antibodies against p130CAS, Ack1, and CASPR show no change in the phosphorylation state of these Src effectors. Merlin and α-tubulin (α-Tub) expression levels are included as internal controls. (B) Nf2-deficient (Nf2^−/−) glia demonstrated a 4.9-fold increase in FAK (Y576/577) phosphorylation and a 5.5-fold increase in paxillin (Pxn) phosphorylation (Y118) (left). Nf2^GFAPCKO mouse brains also showed increased activation of FAK (2.2-fold) and paxillin (2.5-fold) compared to Nf2^flox/flox mouse brains, as assessed by immunoblot analyses and scanning densitometry (right). (C) WT but not mutant merlin inhibits both FAK and paxillin phosphorylation in Nf2^−/− glia. (D) The Src inhibitor PP2 (left) as well as Src.shRNA silencing (right) reverses the hyperphosphorylation of both FAK and paxillin in Nf2^−/− glia.

To demonstrate that merlin loss and Src activation were directly responsible for the increases in FAK and paxillin phosphorylation, we transduced WT and Nf2^−/− glia with WT or mutant (L64P) merlin, using MSCV retroviral delivery. In these experiments, reexpression of WT but not mutant merlin in Nf2^−/− glia reduced FAK and paxillin phosphorylation to WT levels (Fig. 6C). To demonstrate that Src activation was responsible for FAK and paxillin activation in Nf2^−/− glial cells, we employed PP2 pharmacologic Src inhibition and shRNA Src silencing (Src.shRNA). Both PP2 (left panels) and Src.shRNA (right panels) inhibition reduced FAK and paxillin phosphorylation in Nf2^−/− glial cells to WT levels (Fig. 6D). No effect of merlin expression, PP2 treatment, or Src.shRNA lentiviral infection was observed on FAK and paxillin phosphorylation in WT glia.

To determine whether FAK and paxillin were responsible for the increased proliferation observed in Nf2-deficient glia, we initially used a known FAK/paxillin pharmacological inhibitor, echistatin (11). Treatment of Nf2^−/− glia with echistatin resulted in inhibition of FAK and paxillin phosphorylation but had no effect on Src phosphorylation (Fig. 7A, upper panel). Furthermore, echistatin treatment reduced the proliferation of Nf2^−/− glia to WT levels (Fig. 7A, lower panel). No effect of echistatin treatment on WT glial cell proliferation was observed. We next performed complementary experiments using specific FAK (FAK.shRNA) and paxillin (Pxn.shRNA) shRNA reagents to circumvent the problems inherent in using relatively nonselective pharmacological inhibitors. In these experiments, lentiviral FAK.shRNA expression in Nf2^−/− glia inhibited paxillin phosphorylation but had no effect on Src phosphorylation (Fig. 7B, upper panel). FAK shRNA inhibition in Nf2^−/− glia also reduced the proliferation to levels observed in WT glia (Fig. 7B, lower panel). In addition, we inhibited paxillin expression in Nf2^−/− glia with lentiviral Pxn.shRNA and observed no effect on either Src or FAK phosphorylation (Fig. 7C, upper panel); however, this reduced the proliferation of Nf2^−/− glia to WT levels (Fig. 7C, lower panel). In both cases, no effects of FAK and paxillin inhibition on WT glial cell proliferation were observed. These experiments demonstrate that merlin negatively regulates glial cell proliferation in a Src-, FAK-, and paxillin-dependent manner and place paxillin downstream of Src and FAK in the merlin glial cell growth control pathway.

• Open in new tab
• Download powerpoint

FIG. 7.

Merlin growth regulation requires FAK and paxillin signaling. (A) (Top) Echistatin inhibits the phosphorylation of both FAK and paxillin but has no effect on Src activation. (Bottom) Echistatin treatment eliminates the growth advantage observed in Nf2^−/− glia, as determined by [³H]thymidine incorporation. *, P < 0.001 (unpaired t test). (B) (Top) FAK.shRNA inhibits the expression of FAK (>92% reduction) as well as the phosphorylation of both FAK and paxillin. No effect on Src activation was observed. (Bottom) FAK.shRNA treatment eliminates the growth advantage observed in Nf2^−/− glia, as assessed by [³H]thymidine incorporation. *, P < 0.003 (unpaired t test). (C) (Top) Pxn.shRNA treatment inhibits the expression of paxillin (>95% reduction) as well as the phosphorylation of paxillin but has no effect on FAK and Src activation. Total tubulin, Src, FAK, and paxillin expression levels were used as internal controls for equal protein loading. (Bottom) Pxn.shRNA treatment eliminates the growth advantage observed in Nf2^−/− glia, as determined by [³H]thymidine incorporation. *, P < 0.001 (unpaired t test).

Merlin regulates glial cell proliferation through Src signaling in vivo.To demonstrate that merlin regulates glial cell proliferation in a Src-dependent manner in vivo, postnatal day 10 Nf2^GFAPCKO pups were injected with either the Src inhibitor PP2 or vehicle three times a week for 2 weeks (14). Nf2^GFAPCKO pups were then injected with BrdU, and the numbers of BrdU- and GFAP-positive glia were quantitated as previously reported (3, 24). Consistent with our findings in vitro, PP2 treatment of Nf2^GFAPCKO mice resulted in reductions in the numbers of BrdU (Fig. 8A) - and GFAP (Fig. 8B)-expressing cells to levels observed in Nf2^flox/flox mice. Similarly, PP2 treatment decreased Src, FAK, and paxillin phosphorylation in Nf2^GFAPCKO mice to WT levels (Fig. 8C). PP2 treatment had no effect on glial cell numbers, proliferation, or Src pathway activity in Nf2^flox/flox mice. Collectively, these results demonstrate that merlin also regulates glial cell growth in a Src-dependent manner in vivo.

• Open in new tab
• Download powerpoint

FIG. 8.

Merlin regulation of glial cell growth is Src dependent in vivo. (A) Representative immunofluorescence photomicrographs of hippocampi from BrdU-injected Nf2^flox/flox and Nf2^GFAPCKO mice, with or without treatment with the Src inhibitor PP2 (2 mg/kg). Images were labeled with both BrdU (green) and GFAP (red) antibodies. A twofold decrease in the number of BrdU-positive cells was observed in the PP2-treated Nf2^GFAPCKO brains compared to vehicle-treated mouse brains. *, P = 0.001 (unpaired t test). (B) Representative photomicrographs of hippocampi labeled with GFAP antibodies. PP2-treated Nf2^GFAPCKO mice exhibited a twofold decrease in the number of GFAP-labeled cells compared to vehicle-treated Nf2^GFAPCKO littermates. No effect of PP2 treatment on Nf2^flox/flox mice was seen. *, P < 0.002 (unpaired t test). (C) Immunoblot analysis demonstrates amelioration of the increased Src, FAK, and paxillin activation following PP2 treatment. The expression levels of tubulin, total Src, FAK, and paxillin were used as internal controls for equal protein loading.

Merlin regulates Src by modulating ErbB2 activation.Several receptor tyrosine kinase proteins have been shown to regulate Src activity, including ErbB2 (erythroblastic leukemia viral oncogene homolog 2) (36, 53) and EGFR (13, 19). Based on previous studies implicating EGFR in merlin growth regulation in non-nervous-system cells (9), we first examined EGFR in Nf2^−/− glia in vitro and in Nf2^GFAPCKO mice in vivo. We found no changes in EGFR expression or phosphorylation in Nf2^−/− glia relative to that in WT glia (Fig. 9A). Next, we examined the activation status of other members of the EGFR family, including ErbB2, ErbB3, and ErbB4, using phospho-specific antibodies. Whereas we observed no changes in ErbB3 or ErbB4 tyrosine phosphorylation (Fig. 9B), we observed a significant increase in ErbB2 phosphorylation on tyrosine 877 in Nf2^−/− glial cells compared to that in WT glia (Fig. 9C). In contrast, no change in ErbB2 tyrosine 1248 phosphorylation was observed in Nf2^−/− glial cells relative to that in WT glia (Fig. 9C). As expected, we also detected increased ErbB2 tyrosine 877 phosphorylation in Nf2^GFAPCKO compared to Nf2^flox/flox mouse brains (Fig. 9D). To directly demonstrate that ErbB2 activity was increased in Nf2^−/− glia, we assayed ErbB2 activity following ErbB2 immunoprecipitation. Similar to the results obtained with the ErbB2 Y877 phospho-antibody, we observed a 3.2-fold increase in ErbB2 activity in Nf2^−/− glia relative to that in their WT counterparts (Fig. 9E). Lastly, to show that merlin is directly responsible for ErbB2 activation, we found that reexpression of WT but not L64P mutant merlin in Nf2^−/− glia reduced ErbB2 Y877 phosphorylation to WT levels (Fig. 9F). No effect of merlin expression on ErbB2 Y877 phosphorylation was observed in WT glia.

• Open in new tab
• Download powerpoint

FIG. 9.

Merlin regulates ErbB2 activation in glial cells. (A) Immunoblot analysis of WT and Nf2^−/− glial lysates, using a phospho-specific EGFR antibody (Y845), demonstrates no change in EGFR activation. Total EGFR expression was used as an internal control for equal protein loading. (B) Immunoblot analyses of WT and Nf2^−/− glial lysates following phosphotyrosine (pTyr) affinity binding with antibodies against ErbB4 and ErbB3 show no change in phosphorylation state for the other two ErbB family members. Merlin and tubulin expression levels were included as internal controls. (C) Nf2^−/− glia exhibited a 7.2-fold increase in ErbB2-Y877 activation compared to WT controls in vitro, with no change in ErbB2-Y1248 phosphorylation. (D) Nf2^GFAPCKO mouse brains exhibited a 2.3-fold increase in ErbB2 (Y877) activation compared to WT controls in vivo, as assessed by immunoblotting analysis and scanning densitometry. (E) In an ErbB2 kinase activity assay, Nf2^−/− glia exhibited a 3.2-fold increase in ErbB2 activity relative to WT glia. *, P < 0.001 (unpaired t test). OD, optical density. (F) WT but not mutant merlin expression reduces ErbB2 activation in Nf2^−/− glia.

Consistent with ErbB2 activation as the initiating signal in the merlin growth control pathway, inhibition of Src or Src effectors in vitro or in vivo had no effect on ErbB2 Y877 phosphorylation. In these studies, no changes in ErbB2 tyrosine 877 phosphorylation were seen following PP2 treatment of Nf2^−/− glia in vitro (Fig. 10A, left panels), PP2 treatment of Nf2^GFAPCKO mouse brains in vivo (Fig. 10A, right panels), inhibition of FAK and paxillin by echistatin (Fig. 10B), or shRNA genetic inhibition of Src (Fig. 10C), FAK (Fig. 10D), and paxillin (Fig. 10E).

• Open in new tab
• Download powerpoint

FIG. 10.

ErbB2 activation functions upstream of Src/FAK/paxillin in the merlin glial control growth pathway. (A) PP2 treatment of Nf2^−/− glial cells has no effect on ErbB2 hyperactivation in vitro or in vivo. (B) Echistatin treatment of Nf2^−/− glial cells has no effect on ErbB2 hyperactivation in vitro. Genetic inhibition of Src (C), FAK (D), or paxillin (E) in Nf2^−/− glial cells, using Src.shRNA, FAK.shRNA, or Pxn.shRNA, respectively, has no effect on ErbB2 hyperactivation. Total tubulin and total ErbB2 expression levels were used as internal loading controls.

To determine whether merlin-dependent ErbB2 activation was responsible for Src-FAK-paxillin pathway activation as well as increased proliferation in Nf2^−/− glia, we initially employed the ErbB2-specific inhibitor tyrphostin (AG825) (39). Following treatment of Nf2^−/− glia with AG825, Src, FAK, and paxillin phosphorylation in Nf2^−/− glial cells was reduced to WT levels (Fig. 11A, top panel). Similarly, ErbB2 inhibition also decreased Nf2^−/− glial cell growth to the levels observed in WT glia (Fig. 11A, bottom panel). To provide additional evidence for the role of ErbB2 in merlin growth regulation, we employed ErbB2 silencing with lentiviral ErbB2.shRNA. In these experiments, loss of ErbB2 expression likewise reduced Src, FAK, and paxillin phosphorylation in Nf2^−/− glia to WT levels (Fig. 11B, top panel). As observed following AG825 treatment, ErbB2 shRNA inhibition also decreased Nf2^−/− glial cell growth to WT levels (Fig. 11B, bottom panel). These findings demonstrate that merlin selectively regulates ErbB2 activation, which is responsible for the increased Src signaling and proliferation observed in Nf2^−/− glial cells.

• Open in new tab
• Download powerpoint

FIG. 11.

Merlin regulation of glial cell growth requires ErbB2 activation. (A) (Top) Inhibition of ErbB2 activation with AG825 in Nf2^−/− glia reduces Src, FAK, and paxillin hyperactivation to WT levels. (Bottom) AG825 treatment of Nf2^−/− glia reduces cell proliferation to WT levels, as assessed by [³H]thymidine incorporation. No effect of AG825 treatment was observed in WT glia. *, P < 0.001 (unpaired t test). (B) (Top) Inhibition of ErbB2 expression and activation with ErbB2.shRNA (>98% reduction) reduces the hyperactivation of Src, FAK, and paxillin in Nf2^−/− glia to WT levels. Total tubulin, Src, FAK, and ErbB2 expression levels were used as internal loading controls. (Bottom) Treatment with ErbB2.shRNA but not with vector eliminates the growth advantage observed in Nf2^−/− glia, as assessed by [³H]thymidine incorporation. No effect of ErbB2.shRNA was observed in WT glia. *, P < 0.002 (unpaired t test).

Merlin regulates the interaction between Src and ErbB2.Based on the observation that ErbB2 can interact with both merlin (15) and Src (27, 36, 42), we next sought to determine whether merlin loss is associated with increased ErbB2-Src binding. In these experiments, ErbB2 was precipitated from WT and Nf2^−/− glia using total ErbB2 antibodies. First, we confirmed that merlin binds to ErbB2 in WT glia and demonstrated that merlin loss in glia leads to a dramatic increase in both total and Y416-phosphorylated Src binding to ErbB2 (Fig. 12A). Second, we showed that the reintroduction of WT but not mutant merlin inhibited ErbB2-Src binding (Fig. 12B). Third, we found that inhibition of ErbB2 activation using the AG825 inhibitor blocked the interaction of ErbB2 with Src in Nf2^−/− glia (Fig. 12C), whereas inhibition of Src activation with PP2 had no effect on ErbB2-Src binding (Fig. 12D). These findings demonstrate that only functional merlin can inhibit Src binding to ErbB2 and that the association between Src and ErbB2 is dependent on ErbB2 but not Src activity.

• Open in new tab
• Download powerpoint

FIG. 12.

Merlin directly regulates the interaction between ErbB2 and Src. (A) Immunoprecipitation of ErbB2 from WT and Nf2^−/− glia shows that the association between ErbB2 and Src as well as ErbB2 and p-Src (Y416) occurs only in Nf2-deficient glia. (B) WT but not mutant merlin expression inhibits ErbB2 and Src interaction in Nf2^−/− glia. (C) AG825 treatment inhibits the association between ErbB2 and Src in Nf2-deficient glia. (D) PP2 treatment has no effect on the association between ErbB2 and Src in Nf2-deficient glia. Total ErbB2 was used as an internal control for ErbB2 precipitation. (E) Src and merlin directly associate with ErbB2 following immunoprecipitation of [³⁵S]methionine-radiolabeled merlin (NF2.WT or NF2.L64P) or Src proteins with recombinant His-tagged ErbB2 (ErbB2.His). (Top) No binding to ErbB2 was observed with mutant merlin. (Middle) In the absence of the ErbB2 recombinant protein, there was no binding of radiolabeled NF2.WT, NF2.L64P, or Src protein to the His affinity gel. (Bottom) Input levels of radiolabeled NF2.WT, NF2.L64P, and Src proteins in the ErbB2.His immunoprecipitation analysis were used as normalization controls in scanning densitometric analyses to determine the percentages of bound proteins. A total of 32% of radiolabeled NF2.WT and 28% of Src proteins were bound to recombinant ErbB2 protein, whereas <1% of NF2.L64P was bound to ErbB2. (F) (Top) ErbB2-Src binding was decreased 67% in the presence of equal amounts of merlin and completely eliminated in the presence of a fivefold excess of merlin. (G) (Top) ErbB2-merlin binding was decreased 52% in the presence of equal amounts of Src, decreased 94% in the presence of a 5-fold excess of Src, and completely eliminated in the presence of a 10-fold excess of Src (top). (Bottom [F and G]) Immunoblotting was used to verify the amounts of nonradiolabeled merlin and Src added to each reaction mix.

To determine whether ErbB2 binding to merlin and Src represents a direct interaction, we immunoprecipitated radiolabeled in vitro-transcribed/translated merlin and Src proteins with a recombinant ErbB2 protein containing the activation region of ErbB2 known to interact with Src (27). Both merlin and Src directly interacted with ErbB2 (Fig. 12E). Importantly, we observed no binding of mutant merlin (L64P) to ErbB2, consistent with its inability to restore normal growth regulation in Nf2^−/− glia. We next sought to determine whether merlin and Src compete for binding to ErbB2. In these experiments, excess amounts of merlin or Src were added to interfere with Src-ErbB2 or merlin-ErbB2 binding, respectively. We show that merlin inhibited ErbB2-Src binding and Src inhibited ErbB2-merlin binding in a concentration-dependent manner (Fig. 12F and G).

To validate the Src-ErbB2 association results, we performed reciprocal immunoprecipitation studies, using Src to detect ErbB2 binding. In these experiments, we found that merlin loss in glia led to a dramatic increase in Src binding to ErbB2 (Fig. 13A) and that reexpression of WT but not mutant merlin inhibited ErbB2-Src binding (Fig. 13B). Moreover, the Src-ErbB2 interaction was dependent only on ErbB2 activation (Fig. 13C) and not Src activity (Fig. 13D). Interestingly, we also noticed that merlin was found in the Src immunoprecipitates from WT glial cells (Fig. 13A). However, unlike ErbB2-Src binding, the interaction between Src and merlin was not dependent on Src or ErbB2 activity.

• Open in new tab
• Download powerpoint

FIG. 13.

Merlin interacts with both ErbB2 and Src. (A) Immunoprecipitation of Src from WT and Nf2^−/− glia shows that the association between Src and ErbB2 occurs only in Nf2-deficient glia. (B) WT but not mutant merlin expression inhibits Src and ErbB2 interaction in Nf2^−/− glia. (C) AG825 treatment inhibits the association between Src and ErbB2 in Nf2-deficient glia. (D) PP2 treatment has no effect on the association between Src and ErbB2 in Nf2-deficient glia. Total Src was used as an internal control for Src precipitations. In WT glia, merlin and Src associate; however, this binding is not affected by SrbB2 or Src activation.

The finding that merlin can associate with either Src or ErbB2 suggests two non-mutually exclusive models for merlin function. First, merlin may compete with Src for binding to ErbB2, such that merlin-ErbB2 binding precludes an association between Src and ErbB2. In this fashion, merlin loss allows Src to bind to ErbB2, which culminates in Src activation and increased glial cell growth. Second, merlin may bind to Src and sequester Src from binding to ErbB2. The loss of merlin would release Src to bind to ErbB2 and initiate Src-dependent signaling to increase glial cell proliferation.

To distinguish between these two possibilities, we sought to define the merlin residues important for Src and ErbB2 binding. In these experiments, we employed an in vitro binding assay using recombinant ErbB2 and Src proteins to determine which regions of merlin were necessary for merlin-ErbB2 and merlin-Src interactions. In vitro-transcribed/translated radiolabeled full-length merlin (WT) as well as merlin fragments containing residues 1 to 300 (NF2.N-term), 300 to 595 (NF2.C-term), and 300 to 557 were immunoprecipitated with recombinant ErbB2 or Src protein. Merlin binding to ErbB2 required C-terminal domain residues 300 to 557 (Fig. 14A), whereas merlin binding to Src required C-terminal domain residues 557 to 595 (Fig. 14B). These findings demonstrate that the residues important for merlin binding to Src and ErbB2 are distinct and separable.

• Open in new tab
• Download powerpoint

FIG. 14.

Merlin binding to ErbB2 precludes the association of Src with ErbB2. (A) In an in vitro binding assay, [³⁵S]methionine-radiolabeled full-length merlin containing residues 1 to 595 (NF2.WT), NF2.C-term containing residues 300 to 595, and NF2.C-term containing residues 300 to 557, bound to recombinant His-tagged ErbB2 (ErbB2.His). In contrast, no binding to ErbB2 was observed using NF2.N-term containing residues 1 to 300. Input levels for the reactions are shown and were used as normalization controls in scanning densitometric analyses to determine the percentages of bound proteins. Thirty percent of radiolabeled NF2.WT, 25% of NF2.C-term, and 24% of NF2.300-557 bound to recombinant ErbB2 protein, whereas <1% of NF2.N-term bound to ErbB2. (B) In an in vitro binding assay, [³⁵S]methionine-radiolabeled full-length merlin containing residues 1 to 595 (NF2.WT), and NF2.C-term containing residues 300 to 595, bound to recombinant His-tagged ErbB2 (ErbB2.His). In contrast, no binding to ErbB2 was observed using either NF2.C-term containing residues 300 to 557, or NF2.N-term containing residues 1 to 300. Input levels for the reactions are shown and were used as normalization controls in scanning densitometric analyses to determine the percentages of bound proteins. Thirty-one percent of radiolabeled NF2.WT and 19% of NF2.C-term bound to recombinant Src protein, whereas <1% of NF2.N-term and NF2.300-557 bound to Src. (C) (Top) ErbB2-Src binding was decreased 48% in the presence of equal amounts of NF2.300-557, decreased 96% in the presence of a 5-fold excess of NF2.300-557, and completely eliminated in the presence of a 10-fold excess of NF2.300-557. (Bottom) Coomassie staining was used to verify the amount of nonradiolabeled NF2.300-557 added to each reaction mix. (D) (Top) NF2.300-557-ErbB2 binding was not altered by equal or excess amounts of Src. (Bottom) Immunoblotting was used to verify the amount of nonradiolabeled Src added to the reaction mixtures. (E) Potential molecular model for merlin growth regulation in glial cells. In WT cells, merlin associates with ErbB2 and precludes Src binding to ErbB2. In the absence of functional merlin, Src binds to activated ErbB2, leading to ErbB2-mediated Src phosphorylation/activation, Src effector protein phosphorylation, and increased glial cell growth.

Since the merlin C-terminal fragment containing residues 300 to 557 binds only to ErbB2 and not to Src, we took advantage of this selective binding to discriminate between the two possible models of merlin function. We found that increasing amounts of the merlin C-terminal 300-557 fragment inhibited Src binding to ErbB2 (Fig. 14C). In contrast, increasing amounts of Src did not inhibit the association between the merlin C-terminal 300-557 fragment and ErbB2 (Fig. 14D). These results indicate that merlin binding to ErbB2 precludes the ability of Src to associate with ErbB2 and that the association between merlin and ErbB2 likely interferes with ErbB2-mediated activation of Src.