Growth Hormone-induced Alteration in ErbB-2 Phosphorylation Status in 3T3-F442A Fibroblasts*

The growth hormone receptor (GHR), a cytokine receptor superfamily member, requires the JAK2 tyrosine kinase for signaling. We now examine functional interactions between growth hormone (GH) and epidermal growth factor (EGF) in 3T3-F442A fibroblasts. Although EGF enhanced ErbB-2 tyrosine phosphorylation, GH, while causing retardation of its migration on SDS-polyacrylamide gel electrophoresis, decreased ErbB-2's tyrosine phosphorylation. GH-induced retardation was reversed by treatment of anti-ErbB-2 precipitates with both alkaline phosphatase and protein phosphatase 2A, suggesting that GH induced serine/threonine phosphorylation of ErbB-2. Both GH-induced shift in ErbB-2 migration and GH-induced MAP kinase activation were unaffected by a protein kinase C inhibitor but were blocked by the mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1 (MEK1) inhibitor, PD98059. Notably, leukemia inhibitory factor, but not interferon-γ, also promoted ErbB-2 shift and mitogen-activated protein kinase activation. Cotreatment with EGF and GH versus EGF alone resulted in a 35% decline in acute ErbB-2 tyrosine 1248 autophosphorylation, a marked decline (approximately 50%) in DNA synthesis, and substantially decreased cyclin D1 expression. We conclude that in 3T3-F442A cells, 1) the GH-induced decrease in ErbB-2 tyrosine phosphorylation correlates with MEK1/mitogen-activated protein kinase activity and 2) GH antagonizes EGF-induced DNA synthesis and cyclin D1 expression in a pattern consistent with its alteration in ErbB-2 phosphorylation status.

regulatory effects in various tissues (1). The particular GH responses depend on the target cell type and the context in which the stimulation occurs. GH action is initiated by its interaction with the GH receptor (GHR), a transmembrane glycoprotein member of the cytokine receptor superfamily (2,3). GH promotes dimerization of the GHR and activation of the cytoplasmic tyrosine kinase, JAK2, which physically and functionally associates with the receptor via the GHR Box1 region (4 -11). Following GH-induced GHR dimerization and JAK2 activation, several signaling pathways are engaged, including those involving signal transducers and activators of transcription, insulin receptor substrates/phosphatidylinositol 3-kinase, and MAP kinases and their upstream activators in the Ras-Raf-MEK1 pathway (Refs. 12 and 13 and references therein).
Murine 3T3-F442A fibroblasts are endowed with GHRs, and this cell line has been a useful model system for biochemical and functional studies of GH signaling (6, 14 -17). Treatment with GH promotes in these cells the robust acute activation of JAK2 and the tyrosine phosphorylation of the GHR, JAK2, signal transducers and activators of transcription, insulin receptor substrates 1 and 2, MAP kinases, and numerous other substrates as well as the activation of c-fos transcription (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26). Prolonged treatment of 3T3-F442A cells with GH in a mixture of other growth factors results in their differentiation from fibroblasts (preadipocytes) to adipocytes (14). GH, unlike other growth factors found in serum, does not promote proliferation of the fibroblast form.
Epidermal growth factor (EGF), however, is a potent mitogen for 3T3-F442A fibroblasts, which have EGF receptors (EGFRs) that are readily detectable immunologically (27). The EGFR (ErbB-1) is a transmembrane glycoprotein with intrinsic tyrosine kinase activity in its cytoplasmic domain and is a member of a receptor family that includes the structurally related ErbB-2, ErbB-3, and ErbB-4. Though no ligand is known for ErbB-2, it is activated upon binding of EGF or transforming growth factor-␣ to the EGFR or following the binding of neuregulins or heregulins to ErbB-3 and/or ErbB-4 (28 -31). It is believed that ErbB-2 participates in signaling by heterodimerizing with other ErbB family members upon binding of those receptors to their respective ligands (32). Further, ErbB-2, when overexpressed (as it is in many cancer cells (33)) or when mutated (34), can homodimerize to confer ligand-independent signaling. ErbB-2 signaling can have various biological results, including cellular proliferation (35).
Several recent reports point to biologically relevant crosstalk between signaling via ErbB family members and signaling promoted by GH and other cytokines and non-EGF family growth factors (27, 36 -41). For instance, the EGFR and ErbB-2 have been implicated as signaling components in pathways activated by GH and interleukin-6 (IL-6), respectively, in the absence of EGF or other ErbB family ligands (36 -38).
Other reports indicate that in various cell types EGF signaling via EGFR and/or ErbB-2 can be inhibited by GH, prolactin, platelet-derived growth factor (PDGF), or the protein kinase C (PKC) activator, PMA (27, 39 -41). In several instances, such inhibition has been speculated to be dependent on the activation of PKC, although the mechanisms involved are incompletely understood.
In this study, we use 3T3-F442A fibroblasts to explore biochemical and functional effects of GH on EGFR and ErbB-2. We observe that upon GH treatment, there is an increase in the tyrosine phosphorylation of EGFR, whereas ErbB-2 undergoes a shift in electrophoretic mobility that can be reversed by alkaline phosphatase or protein phosphatase 2A (PP 2A) treatment and is accompanied by a decrease in basal and EGFinduced ErbB-2 tyrosine phosphorylation. This GH-induced mobility shift of ErbB-2 is not affected by the PKC antagonist GF109203X, but it is completely blocked by the MEK1 inhibitor, PD98059. Furthermore, although two other cytokines, namely leukemia inhibitory factor (LIF) and interferon-␥ (IFN-␥), can activate JAK2 in these cells, only LIF (and not IFN-␥) can promote both MAP kinase activation and a shift in ErbB-2 migration. Finally, we demonstrate that GH and LIF can substantially inhibit EGF-induced DNA synthesis and cyclin D1 expression in these cells. Our results point to a potentially relevant modulation of ErbB-2 phosphorylation and EGF signaling by a GH-induced pathway.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human GH was kindly provided by Lilly (to S. J. F.) and Genentech, Inc. (South San Francisco, CA) (to P. J. B.). Recombinant human EGF (Collaborative Biomedical Products, Bedford, MA and Upstate Biotechnology Inc., Lake Placid, NY), murine LIF (R & D Systems, Minneapolis, MN), and murine IFN-␥ (Sigma) were purchased commercially. PP 2A was purchased from Upstate Biotechnology. GF109203X, okadaic acid, and AG1478 (Calbiochem) and PD98059 (New England Biolabs, Beverly, MA) were obtained commercially. Routine reagents were purchased from Sigma unless otherwise noted.
Inhibitor Pretreatment, Cell Stimulation, and Protein Extraction-Serum starvation of 3T3-F442A cells was accomplished by substitution of 0.5% (w/v) bovine serum albumin (fraction V, Roche Molecular Biochemicals) for serum in the culture medium for 16 -20 h prior to experiments, except as indicated below in the description of the [ 3 H]thymi-dine incorporation and cyclin D1 protein expression experiments. Pretreatments and stimulations were carried out at 37°C in binding buffer (consisting of 25 mM Tris-HCl (pH 7.4), 120 mM NaCl, 5 mM KCl, 1.2 mM MgCl 2 , 0.1% (w/v) bovine serum albumin, and 1 mM dextrose). GF109203X (0.5 M) or PD98059 (100 M) (or Me 2 SO, as a vehicle control) were incubated with serum-starved cells for 15 or 60 min, respectively, prior to treatment with GH (500 ng/ml). The inhibitors were diluted from Me 2 SO-dissolved stock solutions (GF109203X (1 mM) and PD98059 (50 mM)); final Me 2 SO concentration was always 0.2% (v/v) or less.
Immunoprecipitation, Enzymatic Dephosphorylation, Electrophoresis, and Immunoblotting-For immunoprecipitation, the rabbit antisera described above were used at the following volumes or amounts per precipitation: anti-JAK2 AL33 (directed at residues 746 -1129), 3 l; anti-ErbB-2, 1 g; anti-EGFR, 1 g. Protein A-Sepharose (Amersham Pharmacia Biotech) was used to adsorb immune complexes, and, after extensive washing with lysis buffer, Laemmli sample buffer eluates were resolved by SDS-PAGE and immunoblotted as indicated.
For enzymatic dephosphorylation with alkaline phosphatase, anti-ErbB-2 precipitates were washed once with PBS and then incubated with 20 units of calf intestinal phosphatase (CIP) (New England Biolabs) in 60 l of CIP reaction buffer (25 mM Tris-HCl (pH 7.4), 50 M CaCl 2 , 0.1 mM MgCl 2 , 50 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 10 mM benzamidine, 5 g/ml aprotinin, 1 g/ml leupeptin) for 1 h at 37°C. An equal volume of twice concentrated Laemmli sample buffer was added to stop the reaction, and the eluates were resolved by SDS-PAGE and immunoblotted as below. Dephosphorylation of immunoprecipitated ErbB-2 with the serine/threonine-specific phosphatase, PP 2A, was accomplished in a similar fashion by incubation with 0.5 units of PP 2A in 60 l of PP 2A reaction buffer (20 mM Hepes (pH 7.0), 1 mM dithiothreitol, 7 mM MgCl 2 , 0.01 mM MnCl 2 , 100 g/ml bovine serum albumin, 5 g/ml aprotinin, 1 g/ml leupeptin) for 1 h at 37°C. As indicated, okadaic acid (50 M) (or its vehicle, Me 2 SO) was added during the incubation.
[ 3 H]Thymidine Incorporation-3T3-F442A fibroblasts were cultured in 24-well plates (Falcon, Lincoln Park, NJ) and allowed to grow for 2 days in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum at 37°C in 5% CO 2 and then serum-starved for 48 h in Dulbecco's modified Eagle's medium supplemented with 0.5% bovine calf serum under the same conditions. The cells were treated in triplicate with control buffer (40 M HEPES, pH 7.4), EGF (10 nM), GH (220 ng/ml), or LIF (25 ng/ml) or combinations of GH or LIF and EGF, as indicated, for 18 h. [ 3 H]thymidine (2 Ci/well) was added, and the cells were further incubated for 3.5 h. After removal of media, cells were treated with ice-cold 10% trichloroacetic acid for 30 min and washed once with ice-cold trichloroacetic acid. Precipitated proteins were then solubilized with hot 0.5 N NaOH, 0.1% SDS solution. After neutralization with 1 N HCl, solutions were diluted with water, and [ 3 H]thymidine incorporation was determined by scintillation counting.
Cyclin D1 Protein Expression-3T3-F442A fibroblasts were cultured in 100-mm tissue culture plates (Falcon) and serum-starved as described for [ 3 H]thymidine incorporation experiments. The cells were treated with control buffer (40 M HEPES, pH 7.4), EGF (10 nM), GH (220 ng/ml), or LIF (25 ng/ml) or with combinations of either GH or LIF and EGF, as indicated, for 18 h. After washing twice with ice-cold Hanks' buffered saline (containing 10 mM HEPES, pH 7.4, and 1 mM sodium orthovanadate), cells were collected and lysed in twice-concentrated boiling SDS-PAGE sample buffer (10 mM Tris-HCl, pH 8.0, 2 mM EDTA, 2% (w/v) SDS, 2 mM dithiothreitol, 1 mM sodium orthovanadate, 20% (v/v) glycerol) and further boiled for 5 min. Samples were sonicated for 10 s, and the total protein concentration of each sample was assessed by the Micro-BCA method (Pierce). Total cellular proteins (50 g/sample) were separated by SDS-PAGE, and the separated proteins were transferred to a polyvinylidene difluoride membrane (Sigma) at 0.5 mA for 90 min. The membrane was then blocked in Tris-buffered saline with Tween 20 (TBST) (10 mM Tris, pH 8.0, 150 mM NaCl, 0.05% Tween 20) with 5% nonfat dry milk. Cyclin D1 protein expression was assessed by immunoblotting with anti-cyclin D1 antibody, as described above. The same membrane was stripped and reprobed with anti-MAP kinase, as above, to assess protein loading equivalence. The effects of GH and LIF on EGF-induced cyclin D1 expression were quantified by densitometry, as below. The percentage of EGF-induced cyclin D1 protein expression (Fig. 10) was calculated by subtracting the amount of cyclin D1 protein expression in the control-treated cells and dividing by the amount of cyclin D1 protein expressed in the presence of EGF alone (EGF-induced cyclin D1 expression was defined as 100%).
Densitometric Analysis-Densitometry of ECL immunoblots was performed using a solid state video camera (Sony-77, Sony Corp.) and a 28-mm MicroNikkor lens over a light box of variable intensity (Northern Light Precision 890, Imaging Research Inc., Toronto, Canada). Quantification was performed using a Macintosh II-based image analysis program (Image 1.49 (or 1.62), developed by W. S. Rasband, Research Services Branch, NIMH, National Institutes of Health, Bethesda, MD). Relative ErbB-2 tyrosine phosphorylation in detergent extracts of cells treated with EGF and GH or with EGF alone (as in the experiment in Fig. 8B) was estimated in each instance by the ratio of densitometrically determined anti-Tyr(P)-ErbB-2 immunoblotting signal (tyrosine-phosphorylated ErbB-2) to the densitometric signal of the same band after stripping and reprobing with anti-ErbB-2 (total ErbB-2). The relative ErbB-2 tyrosine phosphorylation, determined as such, was then compared for EGF/GH costimulation versus EGF stimulation alone.
We first tested whether EGF-and GH-induced changes in the phosphorylation state of EGFR and ErbB-2 were detectable. As expected, EGF promoted increase in the tyrosine phosphorylation of the EGFR (Fig. 1A). EGFR immunoprecipitation from detergent-treated cell extracts using anti-EGFR antibodies followed by APT immunoblotting revealed maximal EGFR tyrosine phosphorylation after 2-5 min of EGF stimulation. As reported by others (36), we also detected EGFR tyrosine phosphorylation in response to GH stimulation (Fig. 1B). GH-induced EGFR tyrosine phosphorylation was also maximal at 2-5 min but was consistently less robust than that promoted by EGF (compare relative APT and anti-EGFR immunoblot signal intensities in Fig. 1A versus Fig. 1B). As anticipated, EGF promoted a decrease in EGFR abundance (Fig. 1A, lower panel); no such EGFR loss was observed in response to GH (Fig.  1B, lower panel).
We next compared the basal and EGF-induced tyrosine phosphorylation states of the EGFR and ErbB-2 in these cells ( Fig.  2A). Serum-starved cells were exposed to either EGF (ϩ) or vehicle only (Ϫ) for 15 min and detergent-solubilized. ErbB-2 and EGFR present in the cell extracts were each specifically immunoprecipitated with anti-ErbB-2 and anti-EGFR, respectively, and immunoblotted with the 4G10 APT ( Fig. 2A, upper panel). As in Fig. 1, the EGFR exhibited little basal tyrosine phosphorylation, but the APT signal was substantially augmented in response to EGF. In contrast to the EGFR, appreciable tyrosine phosphorylation of ErbB-2 was observed in the absence of ligand. The level of ErbB-2 tyrosine phosphorylation was increased by EGF treatment, but to a lesser degree than that observed for EGFR tyrosine phosphorylation. As expected, ErbB-2 migrated more slowly than EGFR after SDS-PAGE ( Fig. 2A, upper panel). Notably, reprobing of anti-ErbB-2 precipitates with anti-ErbB-2 revealed that EGF treatment of cells even further retarded ErbB-2 migration ( Fig. 2A, lower panel).
The observations that ErbB-2 tyrosine phosphorylation was detected basally and was augmented by EGF treatment were confirmed by immunoblotting anti-ErbB-2 precipitates with a well characterized state-specific antibody (anti-Tyr(P)-ErbB2) (45) that specifically recognizes ErbB-2 molecules that are phosphorylated at tyrosine residue 1248 ( Fig. 2A, middle panel). Tyrosine 1248 is an important ErbB-2 autophosphorylation site (46,47); reactivity with anti-Tyr(P)-ErbB-2 has been correlated with ErbB-2 tyrosine kinase activation. Thus, EGF promoted both tyrosine phosphorylation and retardation of migration of ErbB-2 in 3T3-F442A cells; in other experiments (not shown), both of these phenomena were observed with as little as 2 min exposure to EGF and persisted for at least 60 min.
Treatment of the cells with GH also resulted in the retarded migration of ErbB-2 in SDS-PAGE (Figs. 2, B and C, lower panels). However, in contrast to the effects of EGF, the more slowly migrating ErbB-2 induced by GH exhibited decreased tyrosine phosphorylation, as assessed by both APT immunoblotting of immunoprecipitated ErbB-2 ( Fig. 2B, upper panel) and anti-Tyr(P)-ErbB-2 immunoblotting of detergent extracts (Fig. 2C, upper panel). This loss of ErbB-2 phosphotyrosine content in response to GH was not accompanied by significant loss of ErbB-2 mass (Figs. 2, B and C, lower panels). Rather, GH treatment apparently diminished the ErbB-2 activation state. In studies of the response of serum-starved cells to treatment with either serum or platelet-derived growth factor (PDGF), others have also used APT and anti-Tyr(P)-ErbB-2 antibodies to correlate retardation of ErbB-2 SDS-PAGE migration with reduction in specific ErbB-2 tyrosine phosphorylation and activation (39). The deactivation and shift of ErbB-2 in that study were attributed to serum-and PDGF-induced ErbB-2 serine/ threonine phosphorylation. In the experiment shown in Fig.  3A, anti-ErbB-2 immunoprecipitates from cells exposed to GH (ϩ) or vehicle (Ϫ) were treated with (ϩ) or without (Ϫ) CIP, as described under "Experimental Procedures." CIP is a general purpose alkaline phosphatase capable of removing phosphate from tyrosine, serine, or threonine residues. After SDS-PAGE of the precipitated and treated ErbB-2, anti-ErbB-2 immunoblotting revealed that the GH-induced retardation of migration of ErbB-2 (lane 3 versus lane 1) was reversed by CIP treatment (lane 4 versus lane 3). No change in ErbB-2 migration was detected when the precipitate from cells not exposed to GH was treated with CIP (lane 2 versus lane 1), indicating the lack of significant proteolytic activity associated with CIP under these conditions. These findings suggest that the GH-induced shift of ErbB-2 electrophoretic mobility is related to a GH-induced change in ErbB-2 phosphorylation.
To further test this issue, we performed similar experiments in which we assessed the effect of treating ErbB-2 immunoprecipitates with PP 2A, a serine/threonine-specific phosphatase ( . Interestingly, we also observed that PP 2A treatment resulted in broadening of the ErbB-2 band following GH or PMA stimulation, perhaps as a consequence of differential sensitivity of multiple ErbB-2 phosphorylation sites. From the experiments in Figs. 1 and 2, we conclude that GH treatment retards the migration of ErbB-2; however, unlike the effect of EGF, the GH-induced ErbB-2 shift is accompanied by decreased ErbB-2 tyrosine phosphorylation. In this context, the reversal of the GH-induced ErbB-2 shift by CIP (Fig. 3A) and by PP 2A (Fig. 3B) strongly supports the conclusion that GH promotes serine/threonine phosphorylation of ErbB-2. Together, these findings suggest that, similar to the effects of PDGF and serum observed by others, this GH-induced phosphorylation may dampen the ErbB-2 basal activation state.

GH-and Cytokine-induced Retardation of Migration of ErbB-2 in SDS-PAGE Correlates with MAP Kinase
Activation-PDGF-and PMA-induced serine/threonine phosphorylation and diminished activation of ErbB-2 have been attributed to the activity of PKC (39). To address the potential involvement of PKC in the mediation of the GH-induced shift of ErbB-2 in 3T3-F442A cells, we tested the effects of the PKC inhibitor GF109203X on this process (Fig. 4). GF109203X is a well characterized selective PKC inhibitor that acts by competing with ATP for binding to the kinase catalytic domain (48); it most potently inhibits the classical PKC isoforms, but it also affects the novel isoforms. Cells were pretreated with GF109203X (0.5 M) or its vehicle (Me 2 SO) for 15 min prior to exposure to GH, PMA, or vehicle controls. Anti-ErbB-2 immunoblotting of detergent cell extracts (Fig. 5, top panel) indicated that the GH-induced shift of ErbB-2 (lane 3 versus lane 1) was not affected by GF109203X pretreatment (lane 4 versus lane 3).
The same extracts were also evaluated for their abundance of the activated form of the MAP kinases, ERK1 and ERK2 (Fig.  4, middle panel) by immunoblotting with a state-specific antibody (anti-active MAPK) that specifically recognizes the phosphorylated threonine 183 and tyrosine 185 residues in the MAP kinase molecule that correlate to its enzymatic activation. As we and others have previously observed in these cells (17)(18)(19)(20)27), GH promoted marked MAP kinase activation (lane 3 versus lane 1); this GH-induced MAP kinase activation, like the GH-induced shift of ErbB-2, was not affected by pretreatment with GF109203X (lane 4 versus lane 3). As a positive control for the efficacy of GF109203X, we tested its ability in the same experiment to inhibit effects of the pharmacologic activator of PKC, PMA. As anticipated, PMA induced a substantial shift of ErbB-2 and activation of MAP kinases. Both of these effects were at least partially inhibited by 0.5 M GF109203X pretreatment of the cells (upper and middle panels, lane 6 versus lane 5). From these experiments, it appears that the shift of ErbB-2 observed in response to GH is not attributable to GHinduced activation of GF109203X-sensitive PKC isoforms. Further, these results point to a correlation between GH-induced (or PMA-induced) MAP kinase activation and the shift of ErbB-2.
To further pursue this correlation, we employed the MEK1 inhibitor, PD98059 (Fig. 5). This drug has previously been shown to powerfully inhibit MAP kinase activation by binding to the inactive form of MEK1, an important upstream activat- ing kinase of ERK1 and ERK2 (49). In contrast to GF109203X, pretreatment with PD98059 prevented the GH-induced shift of ErbB-2, as detected by anti-ErbB-2 immunoblotting of detergent cell extracts (Fig. 5A, upper panel, lane 4 versus lane 3). Immunoblotting of the same extracts with anti-active MAPK revealed, as expected, that the inhibition by PD98059 of the GH-induced ErbB-2 shift was accompanied by dramatic inhibition of GH-induced activation of ERK1 and ERK2 (Fig. 5A,  middle panel, lane 4 versus lane 3). To verify that the PD98059 effects on GH-induced MAP kinase activation and ErbB-2 shift were not due to nonspecific inhibition of proximal aspects of GH signaling, we evaluated GH-induced tyrosine phosphorylation of JAK2 in the presence or absence of the inhibitor. JAK2 was immunoprecipitated from aliquots of the same cell extracts evaluated in the experiments shown in Fig. 5A, and precipitated proteins were probed with APT. This analysis indicated that PD98059 did not affect GH-induced JAK2 tyrosine phosphorylation (Fig. 5B, upper panel, lane 4 versus lane 3). Thus, GH-induced MAP kinase activation is further correlated with GH-induced ErbB-2 shift in these cells.
In this experiment, EGF also promoted a substantial retardation of ErbB-2 migration (Fig. 5A, upper panel, lane 5 versus lane 1), which in Fig. 2 was shown to be associated with increased ErbB-2 tyrosine phosphorylation, in distinction from the GH-induced ErbB-2 shift. Interestingly, PD98059 pretreatment, while it blunted EGF-induced MAP kinase activation (Fig. 5A, middle panel, lane 6 versus lane 5), did not reverse the EGF-dependent shift of ErbB-2. Thus, the GH-induced ErbB-2 shift, associated with decreased ErbB-2 tyrosine phosphorylation and related to serine/threonine phosphorylation (Figs. 2 and 3), is a PD98059-sensitive process, whereas the EGFinduced shift of ErbB-2, principally associated with ErbB-2 tyrosine phosphorylation and activation, appears largely PD98059-independent.
Because GH-induced MAP kinase activation (which correlates with the shift and deactivation of ErbB-2) is believed to be dependent on JAK2 activation, we asked whether other activators of JAK2 in these cells might also affect ErbB-2 migration. In addition to GHRs, 3T3-F442A cells express functional receptors for two other cytokines that activate JAK2, LIF and IFN-␥ (6,50). In the experiments shown in Fig. 6, we tested the effects of treatment of cells with GH, LIF, and IFN-␥ on the electrophoretic migration of ErbB-2. Anti-ErbB-2 immunoblotting of detergent cell extracts revealed the GH-induced ErbB-2 shift (Fig. 6A, lane 2 versus lane 1), as already observed. (We note the presence in some immunoblots of anti-ErbB-2-reactive bands migrating slightly more rapidly than the more dominant retarded ErbB-2 band induced by GH; whether this reflects less complete phosphorylation of ErbB-2 in some experiments is presently not known.) Notably, LIF treatment promoted a shift in the migration of ErbB-2 similar to that seen in response to GH (Fig. 6A, lane 3 versus lane 2). In contrast to LIF and GH, IFN-␥ failed to promote such an ErbB-2 migration shift (Fig.  6A, lane 4 versus lanes 1-3).
As a positive control for stimulation in this experiment, JAK2 in the cell extracts from each sample was immunoprecipitated and subjected to APT immunoblotting (Fig. 6B, upper  panel). As has been observed by others using this same cell type (6,50), GH-induced JAK2 tyrosine phosphorylation (which correlates with JAK2 activation) was more pronounced than that induced by either LIF or IFN-␥ (lanes 3 and 4 versus lane 2); however, IFN-␥-induced JAK2 tyrosine phosphorylation was quite similar to that seen in response to LIF (lane 4 versus lane 3). Thus, the lack of IFN-␥-induced shift in ErbB-2 migration cannot be attributed to a lack of IFN-␥ receptor-coupled JAK2 activation. Notably, however, evaluation of the same cell ex-tracts by SDS-PAGE and anti-active MAPK blotting (Fig. 6C) revealed that both GH and LIF, but not IFN-␥, promoted substantial activation of ERK1 and ERK2.
These findings were extended in the experiment shown in Fig. 7. Cells were either pretreated (ϩ) or not (Ϫ) with PD98059 and then stimulated with GH, LIF, or IFN-␥ or left untreated (Ϫ). Anti-ErbB-2 blotting (Fig. 7, upper panel) revealed, as expected, the shift of ErbB-2 migration caused by GH and LIF (lanes 3 and 5, respectively, versus lane 1 failed to promote both ErbB-2 shift and MAP kinase activation (lane 7 versus lane 1). Although we do not know the reason for the inability of IFN-␥ to activate ERKs in these cells, these data, along with those in Figs. 5 and 6, strongly support the conclusion that the ability of GH and LIF to promote retardation of migration of ErbB-2 correlates with their ability to activate MAP kinases and suggest that cytokine-induced modulation of ErbB-2 phosphorylation may be a MEK1-dependent phenomenon.

GH-induced Modulation of ErbB-2 Phosphorylation Correlates with Inhibition of EGF-induced Mitogenic Signaling in
3T3-F442A Cells-The effects of GH and LIF on the basal phosphorylation state of ErbB-2 suggest that these cytokines might potentially modulate an important growth factor receptor signaling system. ErbB-2 has no known ligand, but it has been shown to dramatically potentiate signaling promoted by ligands for the EGF receptor (e.g., EGF, transforming growth factor-␣) and for ErbB-3 and/or ErbB-4 (e.g. heregulin, neuregulin) by heterodimerizing with these other ErbB family receptors in a ligand-dependent fashion (32). We thus sought to determine if the GH-and LIF-induced modulatory effects on ErbB-2 phosphorylation that we have observed might be reflected in alteration of ErbB-2's potentiation of EGF-induced cellular responses.
In the experiment shown in Fig. 8, the effect of cotreatment with EGF and GH versus EGF alone on the the tyrosine phosphorylation state of ErbB-2 was assessed. ErbB-2 was specifically immunoprecipitated from detergent extracts of cells treated with EGF alone for 15 min or with EGF for 15 min in the presence of GH, which was added 5 min prior to the EGF (Fig. 8A, upper panel). APT immunoblotting of the immunoprecipitated ErbB-2 revealed that the abundance of tyrosine-phosphorylated ErbB-2 induced by EGF was markedly diminished when GH was present during the acute stimulation. Stripping and reprobing of this blot with anti-ErbB-2 (Fig. 8A, lower  panel) indicated that this decreased APT reactivity could not be explained by a difference in the recovery of ErbB-2 in the immunoprecipitates of the two extracts.
This GH-dependent antagonism of EGF-induced ErbB-2 tyrosine phosphorylation was also detected using the state-specific anti-Tyr(P)-ErbB-2 antibody directed at phosphorylated tyrosine 1248 of ErbB-2 (Fig. 8B). Cells were treated, as in the experiment in Fig. 8A, with EGF alone or EGF in the presence of GH. Detergent-extracted proteins were resolved by SDS-PAGE and immunoblotted sequentially with anti-Tyr(P)-ErbB-2 (Fig. 8B, upper panel) and, for normalization, anti-ErbB-2 (Fig. 8B, lower panel). Densitometric analysis (as described under "Experimental Procedures") of this and three other independent experiments indicated that cotreatment with GH and EGF in comparison with EGF alone yielded on average a 35% (range Ϯ 4%, p Ͻ 0.001) decline in phosphorylation of tyrosine 1248, when normalized for ErbB-2 abundance. This decrease in EGF-induced anti-Tyr(P)-ErbB-2 signal imparted by cotreatment with GH (without an apparent GHinduced loss of immunodetectable ErbB-2) suggests that EGFinduced ErbB-2 tyrosine kinase activation, along with overall ErbB-2 tyrosine phosphorylation, may be dampened by concomitant activation of GHR signaling pathway(s).
To examine whether GH-or LIF-induced changes in ErbB-2 phosphorylation might influence downstream EGF signals, we tested the effects of cytokine treatment on EGF-induced mitogenesis in these cells (Fig. 9). Serum-deprived cells were incubated for 21 h with vehicle (control), GH, LIF, or EGF alone or with EGF combined with either GH or LIF. Incorporation of a pulse of [ 3 H]thymidine was then measured for each as an index of DNA synthesis (Fig. 9A). As has been observed previously (27), GH does not promote DNA synthesis in these cells. Like GH, LIF did not promote [ 3 H]thymidine incorporation, despite the ability of each to acutely activate JAK2 (as shown earlier in   6B). In contrast, EGF treatment resulted in a robust increase in [ 3 H]thymidine uptake in comparison with control, GH, or LIF treatment. Notably, cotreatment with EGF and either GH or LIF markedly diminished the EGF-induced DNA synthesis observed in this assay by up to 50%. This same pattern of antagonism of EGF-induced mitogenic signaling by GH and LIF was noted when cells treated in the same manner with these cytokines and growth factors were lysed and their total cellular proteins were resolved by SDS-PAGE and immunoblotted with anti-cyclin-D1 (Fig. 10). Again, GH and LIF significantly inhibited the EGF-induced increase in the abundance of this important cell cycle regulator. Thus, the pattern of inhibition by GH and LIF of these two markers of EGFinduced mitogenic signaling correlated with the ability of the cytokines to cause a change in ErbB-2 migration. DISCUSSION It is now understood that the EGFR (ErbB-1), an intrinsic tyrosine kinase growth factor receptor, is one member of a family of four similar, but distinct, receptors: ErbB-1, -2, -3, and -4. Although it has no known ligand, emerging evidence indicates that ErbB-2, by being a heterodimerization partner for each of the other ErbB members, augments signaling by EGF or EGF-like ligands that bind to EGFR, ErbB-3, and/or ErbB-4 (28 -32). The importance of ErbB-2 in such dimerization-driven signaling is evidenced by several major findings, including the following: 1) blockade of cell surface expression of ErbB-2 by expression of intracellular single-chain anti-ErbB-2 antibodies substantially blunts various cellular responses to EGF-like ligands that are mediated via ErbB family members (51); 2) overexpression of normal ErbB-2 is a frequent feature of human cancers (33); and 3) an oncogenic form of ErbB-2, Neu*, which has a mutation in its transmembrane domain, is believed to be oncogenic by virtue of its propensity to form homodimers in the absence of ligand stimulation (34).
In this study, we show that 3T3-F442A cells, in addition to their already established ability to respond biochemically and functionally to GH, exhibit readily detectable EGF-induced enhancement of EGFR and ErbB-2 tyrosine phosphorylation and, as previously shown (27), significant EGF-induced mitogenesis. Because the receptors for GH (GHR) and EGF (EGFR and ErbB-2) are homologously expressed in these cells and do not require transfection and/or overexpression for their biochemical and functional detection, this cell line was particularly appealing for studies of the influence of GH signaling on the responsiveness to EGF. Our principal findings are that 1) GH can lead to diminished basal and EGF-induced activation and tyrosine phosphorylation of ErbB-2; 2) GH induces, via a PD98059-sensitive pathway, retardation of migration of ErbB-2 on SDS-PAGE, which is sensitive to treatment in vitro with alkaline phosphatase and PP 2A; 3) LIF, but not IFN-␥, also induces PD98059-sensitive retardation of ErbB-2 migration; and 4) this cytokine-induced change in ErbB-2 electrophoretic migration is correlated with GH-and LIF-induced decrease in EGF-induced mitogenesis.
Our observations regarding this potentially important interaction between the cytokine receptor and ErbB-2 signaling pathways in 3T3-F442A cells can be viewed in the context of several other recent studies pointing to the interplay of certain cytokines and non-EGF-like growth factors on ErbB family signaling. Epstein et al., using a murine 3T3 fibroblast derivative overexpressing the rat ErbB-2, observed that treatment of serum-deprived cells with serum, PDGF, or PMA rapidly promoted retardation of SDS-PAGE migration, decreased tyrosine phosphorylation and activation, and increased serine/threonine phosphorylation of ErbB-2 (39). Effects of PDGF or serum on EGF responsiveness were not directly tested in that study, but the findings are otherwise conceptually similar to our observed GH and LIF effects on the endogenous ErbB-2 in 3T3-F442A cells. It may be notable that in both studies (Ref. 39 The experiment shown is representative of three experiments. B, the data in A, along with those obtained in two other experiments, were subjected to densitometric analysis, as described under "Experimental Procedures." In each case, the abundance of cyclin D1 expression induced by each treatment is expressed as a percentage of that induced by EGF alone. Error bars represent S.E. of the determinations in the three independent experiments (p Ͻ 0.05 both for samples treated with EGF plus GH and samples treated with EGF plus LIF compared with EGF alone). and the current study), the cells in which the modulatory effects on ErbB-2 were observed are not tumor cells or tumor derivatives.
Similarly, Quijano and Sheffield (41) observed that treatment of normal murine mammary gland epithelium (NMuMG) cells with prolactin, which activates several signaling pathways in common with those activated by GH (reviewed in Ref. 52), resulted in decreased EGF-induced EGFR tyrosine phosphorylation and kinase activation. This prolactin-induced desensitization of the EGFR was accounted for by increased threonine phosphorylation of the EGFR and was reversed by in vitro dephosphorylation of the isolated receptor by treatment with alkaline phosphatase. In other studies, the same group showed that EGF-induced mitogenesis of NMuMG cells was markedly inhibited by cotreatment with prolactin (53).
It is not yet clear whether the mechanism(s) by which prolactin desensitizes EGFR signaling in mammary cells and GH affects ErbB-2 phosphorylation and EGF-induced mitogenic signaling in 3T3-F442A cells are the same. We note that biochemical and signaling effects on ErbB family members induced by PMA, prolactin, serum, or PDGF, when attributed to serine/threonine phosphorylation of the ErbB family protein, have been variably considered to depend on the activity of protein kinase C and/or MAP kinase(s) (39 -41). Our findings point to the possibility of a MEK1 dependence for the GH-and LIF-induced retardation of migration of ErbB-2 observed in 3T3-F442A cells. The ErbB-2 shift was seen in response to GH and LIF, but not IFN-␥, in correspondence with each cytokine's ability to elicit MAP kinase activation in these cells. Further, GH-and LIF-induced ErbB-2 shift was completely inhibited by pretreatment with PD98059. The PKC inhibitor, GF109203X, however, had no effect on GH-induced ErbB-2 shift or MAP kinase activation, despite being active against PMA-induced MAP kinase activation. As a tentative model that incorporates our experimental findings, we propose that the well characterized activation of the Ras-Raf-MEK1-MAP kinase pathway by GH (Refs. 54 and 55; reviewed in Refs. 12 and 13) results in changes in ErbB-2 (possibly including increased serine/threonine phosphorylation and diminished basal tyrosine phosphorylation) that render it less able to deliver EGF-induced mitogenic signals. While we cannot yet know which particular molecule(s) is involved with conferring these GH-induced changes in ErbB-2, MEK1 itself or MEK1-dependent kinase(s) (most notably ERKs) are at this point leading candidates.
Interestingly, using the same cells, VanderKuur et al. (56) recently provided evidence for rapid feedback inhibition of GH signaling that may reflect activation of GH-induced pathways that also lead to the change in ErbB-2 phosphorylation state that we observe. In that study, GH promoted within 5 min a transient quantitative retardation in SDS-PAGE mobility of Son of Sevenless (SOS) that, by virtue of its reversal when immunoprecipitated SOS was subjected to alkaline phosphatase treatment, was attributed to serine/threonine phosphorylation. This GH-induced shift in SOS mobility was blocked by pretreatment of the cells with PD98059. Because SOS serine/ threonine phosphorylation was concomitant with dissociation of Grb2 from SOS and inactivation of GH-induced Ras-Raf-MEK1-MAP kinase signaling, it is speculated that these findings reflect MEK1-dependent homologous desensitization of this aspect of GH signaling by either MEK1 itself or a downstream SOS kinase. Future studies comparing SOS and ErbB-2 with regard to their particular patterns of phosphorylation in response to GH will be necessary to determine whether these biochemical findings are indeed related.
The exciting recent findings of Yamauchi et al. (36,37), uncovering a potentially significant role for the EGFR in GH signaling, complement our current findings in underscoring cross-talk between the GHR and ErbB family member activation pathways. That work indicates that in a number of cells (including 3T3-F442A) GH promotes tyrosine phosphorylation of the EGFR without associated activation of the EGFR tyrosine kinase. It is proposed that JAK2 activated by GH directly phosphorylates the EGFR at tyrosine 1068 and thereby provides a docking site for Grb-2. This GH-induced Grb-2 association with the EGFR, independent of EGFR intrinsic tyrosine kinase activity, is proposed to be required for full activation of the MAP kinase pathway in response to GH. While complete confirmation of these results is not yet available, our observation of GH-induced tyrosine phosphorylation of the EGFR (Fig.  1B) is in agreement with these findings. We have also observed GH-induced EGFR tyrosine phosphorylation when cells were pretreated with the tyrosine kinase inhibitor, AG1478, which inhibited EGF-induced EGFR tyrosine phosphorylation (data not shown); this finding further supports the idea that EGFR can function as a docking molecule in the GHR signaling pathway. Our observation that GH can decrease ErbB-2 tyrosine phosphorylation, while furthering the concept of relevant interplay between the GHR and ErbB family member pathways, stands in contrast to GH's effects on EGFR (enhanced, rather than diminished, GH-induced EGFR tyrosine phosphorylation). We note, however, that we have not yet intensively studied whether GH may also induce serine/threonine phosphorylation of EGFR in these cells and, if so, whether such modification may also dampen EGF-induced EGFR activation and signaling events. This is a worthy topic for further investigation.
Other intriguing findings have recently been reported concerning a potential role for ErbB-2 in mediation of IL-6-induced proliferation of human prostate cancer cells (38). It was found that IL-6 promoted tyrosine phosphorylation and activation of ErbB-2 in those cells and that intracellular expression of single-chain antibodies against ErbB-2 abrogated IL-6-induced growth of the cells. The effect of IL-6 on EGF-induced signaling was not examined. The similarities in receptor components and signaling pathways between IL-6 and LIF (both receptors incorporate the gp130 subunit, and JAK1 and -2 are activated by both cytokines) make it relevant to consider our own findings in the context of these data. While LIF reliably promoted in 3T3-F442A cells a PD98059-sensitive shift in ErbB-2 migration (Figs. 6 and 7), we note that LIF's effects on ErbB-2 tyrosine phosphorylation, though not dramatic, were more variable (data not shown). Like GH, LIF's negative effects on EGFinduced mitogenic signaling and cyclin D1 expression (Figs. 9 and 10) were substantial and correlated with the cytokineinduced shift in ErbB-2 migration. Further studies will be necessary to determine to what degree the mechanism(s) by which GH and LIF affect ErbB-2 are similar. It is conceivable, for example, that both GHR and gp130-containing receptors can promote serine/threonine phosphorylation of ErbB-2 but that pathways activated by gp130-containing receptors can also lead to enhanced tyrosine phosphorylation of ErbB-2. Further, the degree and/or signaling significance of such gp130mediated tyrosine phosphorylation could be more manifest in some cells (e.g. cancer cells) compared with others (e.g. nontumorous fibroblasts).
Our observations that GH promotes a change in phosphorylation status of ErbB-2 and inhibits EGF-induced mitogenic signaling raise several important issues for further study. While a MEK1-dependent pathway may be involved in mediating the GH-induced ErbB-2 shift, it will be interesting to determine if other molecules thought to be important in both EGF and GH signaling (e.g. SHP-2 and SIRP-␣/SHPS-1 (15,50,57,58)) are required for GH's effects on ErbB-2 and/or EGFinduced mitogenesis. Similarly, it will be important to determine if GH (or other MEK1-activating cytokines) can also affect EGF signaling in cancer cells. Since ErbB-2 is enriched in many cancers and may confer resistance to chemotherapeutics (59), such a finding might suggest potential novel strategies to inhibit tumor progression and/or to enhance chemosensitivity of tumor cells.