Insulin Enhances Growth Hormone Induction of the MEK/ERK Signaling Pathway*

Growth hormone (GH) plays an important role in growth and metabolism by signaling via at least three major pathways, including STATs, ERK1/2, and phosphatidylinositol 3-kinase/Akt. Physiological concentrations of insulin promote growth probably by modulating liver GH receptor (GHR) levels in vivo, but the possible effects of insulin on GH-induced post-GHR signaling have yet to be studied. We hypothesized that short-term insulin, similar to the fluctuations that occur following feeding, affects GH-induced post-GHR signaling. Our present studies suggest that, in rat H4IIE hepatoma cells, insulin (4 h or less) selectively enhanced GH-induced phosphorylation of MEK1/2 and ERK1/2, but not GH-induced activation of STAT5 and Akt. Although insulin pretreatment altered GH-induced formation of Shc·Grb2·SOS complex, it did not significantly affect GH-induced activation of other signaling intermediates upstream of MEK/ERK, including JAK2, Ras, and Raf-1. Immunofluorescent staining indicated that insulin pretreatment facilitated GH-induced cell membrane translocation of MEK1/2. Insulin pretreatment also increased the amount of MEK association with its scaffolding protein, KSR. In summary, short-term insulin treatment of cultured, liver-derived cells selectively sensitized GH-induced MEK/ERK phosphorylation independent of JAK2, Ras, and Raf-1, but likely resulted from increased cell membrane translocation of MEK1/2. These findings suggest that insulin may be necessary for sensitization of cells to GH-induced ERK1/2 activation and provides a potential cellular mechanism by which insulin promotes growth.

A primary effect of growth hormone (GH) 2 is to stimulate growth of the skeleton and soft tissues. Binding of GH to its receptor (GHR) results in dimerization of GHR followed by tyrosine phosphorylation of GHR and Janus activating kinase (JAK) 2 (1,2). Activated JAK2 leads to the direct phosphorylation of signal transducer and activator of transcrip-tion (STATs) proteins, controlling the transcription of many GH target genes, such as insulin-like growth factor (IGF)-I (2)(3)(4). Activation of JAK2 by GH also activates the extracellular signal-regulated kinases 1 and 2 (ERK1/2) pathway, which plays an important role in gene expression, cell proliferation, and cross-talk between the GH and other growth factor signaling pathways (5)(6)(7)(8)(9).
STAT5 activation requires JAK2-mediated phosphorylation of tyrosine residues located toward the C-terminal of the cytoplasmic domain of the GHR (10). By contrast, activation of ERK1/2 pathway is dependent on a 46-amino acid stretch of the GHR located adjacent to the cell membrane (11). GH activates ERK1/2 by both JAK2-dependent and JAK2-independent mechanisms. The JAK2-dependent mechanism is the well known cascade of Shc, Grb2, son-of-sevenless (SOS), Ras, Raf-1 and MAPK kinase (MEK) (12)(13)(14)(15), possibly involving multiple docking proteins such as insulin receptor substrate-1, Gab-1, and the epidermal growth factor receptor (6,16,17). Activation of GHR/JAK2 results in the tyrosine phosphorylation of Shc, which then interacts with the adapter protein Grb2. Pre-associated with Grb2 is SOS, a guanine nucleotide exchange factor, which is recruited to the cell membrane via Shc⅐Grb2 association, stimulates formation of active GTP-bound Ras, which then initiates a sequence of phosphorylation events that activate a cascade of protein serine/threonine kinases. Ras activates Raf-1 kinase leading to phosphorylation and activation of MEK1/2, which in turn phosphorylates and activates ERK1/2. A JAK2-independent mechanism of ERK1/2 activation via Src-dependent activation of Ral and phospholipase D was also reported (18). GH-induced ERK1/2 activation can activate downstream gene expression and transcription factors, including c-Fos, Egr-1, Jun-B, and Elk-1 (5).
Responsiveness to GH can be regulated by other hormones, such as insulin. A previous report from our laboratory found that, in rat H4IIE hepatoma cells, prolonged insulin treatment (8 -24 h) inhibits GH signaling via the GHR/JAK2/STAT5B pathway, probably by down-regulating GHR levels (19). This suggests that hepatic GH resistance may develop when a patient exhibits chronic hyperinsulinemia, a condition often observed in patients with obesity and in the early stage of Type 2 diabetes. However, physiological levels of insulin appear to be necessary for normal liver GH responsiveness, possibly by maintaining liver GHR levels (20 -23). In addition to the role of insulin in maintaining liver GHR levels, insulin may also affect GH-induced post-receptor signaling. In the present study, we investigated the effects of short-term insulin pretreatment, to mimic the fluctuations that occur following feeding, on GH signaling in cultured rat H4IIE hepatoma cells. We found that insulin selectively enhanced GH-induced phosphorylation of ERK1/2, but not GH-stimulated phosphorylation/activation of STAT5 or Akt. ERK1/2 phosphorylation induced by GH, alone or following insulin pretreatment, required MEK1/2 activation. Insulin pretreatment increased GH-induced Shc⅐Grb2 association and accelerated GH-in-duced Grb2⅐SOS dissociation, leading to increased SOS dissociation from Shc and Grb2. GH-induced activation of other signaling intermediates located upstream of MEK1/2, such as JAK2, Ras, and Raf-1, was not significantly changed following insulin pretreatment. However, there was an increase in GH-induced phosphorylation of MEK1/2 following insulin pretreatment, which correlated with increased cell membrane translocation of MEK1/2. Furthermore, insulin increased the association of MEK and its scaffolding protein, kinase suppressor of Ras (KSR), and GH-induced tyrosine phosphorylation of KSR.
In summary, short-term insulin treatment of cultured, liver-derived cells, selectively sensitized GH-induced MEK/ERK phosphorylation, most likely by facilitating the formation of a MEK⅐KSR complex and subsequent cell membrane translocation of MEK1/2. These findings suggest that insulin may sensitize liver GH-induced signaling via the MEK/ERK pathway and provide a potential mechanism by which insulin promotes growth.

EXPERIMENTAL PROCEDURES
Materials-Bovine GH was purchased from the National Institutes of Health NIDDK National Hormone & Pituitary Program. Fetal bovine serum, calf serum, and horse serum were purchased from Invitrogen. Other materials were purchased from Sigma and Fisher Scientific unless otherwise noted.
Cell Culture and Treatments-Rat H4IIE hepatoma cells (from ATCC) were cultured in Swim's medium supplemented with 10% serum (5% horse serum, 3% newborn calf serum, and 2% fetal calf serum) and 100 units/ml penicillin and 100 g/ml streptomycin. 3T3-F442A cells (25), kindly provided by Dr. H. Green (Harvard University, Boston, MA) and C. Carter-Su (University of Michigan, Ann Arbor, MI), were cultured in Dulbecco's modified Eagle's medium containing 4.5 g/liter glucose (Cellgro, Inc.), supplemented with 10% calf serum, 50 g/ml gentamicin sulfate, 100 units/ml penicillin, and 100 g/ml streptomycin (all from Biofluids, Rockville, MD). At ϳ50% confluence, cells were removed from serum and maintained in serum-free (and bovine serum albumin-free) medium for 48 h prior to the start of experimental treatments. PD 98059 was obtained from Cell Signaling Technology (Beverly, MA) and LY 294002 was purchased from Biomol Research Laboratories, Inc. (Plymouth Meeting, PA).
Immunoprecipitation and Immunoblotting-For immunoprecipitation, 1 mg of protein in lysis buffer was incubated with primary antibody overnight at 4°C. Protein A/G-agarose (fast flow, Amersham Biosciences) was then added, and incubations were continued for 4 h at 4°C. Both immunoprecipitated proteins or whole cell lysate protein (50 g) were resolved by SDS-PAGE (Bio-Rad MiniProtean II System) and transferred to Schleicher & Schuell Protran BA85 membranes. The Western transfers were blocked in Tris-buffered saline containing 0.05% Tween (TBST) and 5% milk, rinsed in TBST, and incubated 2 h at room temperature or overnight at 4°C with the primary antibody, washed in TBST, and then incubated with horseradish peroxidase-labeled conventional secondary antibody or TrueBlot TM for 1 h at room temperature. Proteins were visualized by ECL Plus (Amersham Biosciences) or SuperSignal West Femto Maximum Sensitivity Substrate reagents (Pierce) and autoradiography.
Determination of GTP-bound Ras-According to the manufacturer's instructions, cells were lysed with magnesium-containing lysis buffer (MLB) containing 25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Igepal CA-630, 10% glycerol, and 25 mM NaF. 10 g of Raf-1 Ras-binding domain-agarose (Upstate Biotechnology, Inc.) was added to 1 mg of cell lysate and gently rocked at 4°C for 30 min. The agarose beads were washed three times with MLB and boiled in Laemmli sample buffer for 5 min, the supernatant was then resolved by SDS-PAGE and transferred to Protran BA85 membrane. GTP-bound Ras was determined by immunoblotting with total Ras antibody.
Raf-1 Kinase Assay-Raf-1 proteins were immunoprecipitated from 1 mg of cell lysate with mouse Raf-1 antibody (Santa Cruz Biotechnology, Inc.). Immune complexes were washed three times with lysis buffer and once with kinase assay buffer (20 mM MOPS, pH 7.2, 25 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol, and 5 g/ml aprotinin). Pellets were resuspended in 40 l of kinase assay buffer with 1.5 mM MgCl 2 and 7.5 M ATP along with 0.4 g of GST-MEK1 (Upstate Biotechnology, Inc.), and the reaction was incubated for 30 min at 30°C. The kinase reaction was terminated by adding 40 l of 2ϫ Laemmli buffer, boiled for 5 min, resolved by SDS-PAGE, and transferred to Protran BA85 membrane. Raf-1 activity was evaluated by immunoblotting with anti-phospho-MEK1/2 antibody (BIOSOURCE International, Inc.).
Immunofluorescence-Serum-starved cells were treated with various stimuli and then washed with PBS and fixed in 3% paraformaldehyde in PBS for 45 min. The cells were permeabilized with 0.5% Triton X-100 in PBS for 3 min and blocked in 1% bovine serum albumin in PBS for 30 min. Thereafter, the cells were incubated with rabbit anti-MEK1/2 antibody (1:50, Santa Cruz Biotechnology, Inc.) for 1 h. Cells were then washed with PBS and incubated with secondary anti-rabbit antibodies conjugated with Alexa Fluor 488 (1:200, Molecular Probes, Eugene, OR) for 1 h. Cells were mounted in VECTASHIELD Mounting Medium with 4Ј,6-diamidino-2phenylindole (Vector Laboratories, Inc., Burlingame, CA) and analyzed by confocal microscopy using a Leica DMIRBE Inverted SP1 confocal microscope (Bannockburn, IL) with a 100ϫ oil immersion objective.
Densitometric and Statistical Analysis-Chemiluminescent images of immunoblots were analyzed by scanning densitometry. Multiple exposures of each blot were used to obtain gray-scale images of each chemiluminescent band and were quantified with the Fluorchem FC imager system (Alpha Innotech, San Leandro, CA). All data were analyzed by analysis of variance using the InStat statistical program (version 3) by GraphPad Software, Inc. (San Diego, CA), with * or # ϭ p Ͻ 0.05, ** or ## ϭ p Ͻ 0.01, and *** or ### ϭ p Ͻ 0.001, upon comparison to the corresponding control group.
To determine whether insulin pretreatment would affect the other two signaling pathways activated by GH, GH-induced tyrosine phosphorylation of STAT5 (PY-STAT5) and serine phosphorylation of Akt (PS-Akt) were also characterized. The magnitude of GH-induced PY-STAT5 was not significantly altered by insulin pretreatment (Fig. 1A, row 1, compare lanes 1-5 to lanes 8 -12). Induction of PS-Akt by GH was much weaker than by insulin and, due to the short exposure time, the PS-Akt induced by GH alone was difficult to observe (Fig. 1A, row 2 and lanes 1-6). However, with longer autoradiography exposure, the GH effect on PS-Akt is clearly evident (data not shown) (26). After insulin pretreatment for 120 min, phosphorylation of Akt (PS-Akt) was still highly elevated (Fig. 1A, row 2), and there was little effect of GH on FIGURE 1. Short-term insulin pretreatment selectively enhanced GH-induced ERK1/2 phosphorylation. H4IIE cells were treated without or with 10 nM insulin for 4 h or less followed by 500 ng/ml GH or vehicle for the indicated times. Western blot analysis was performed with antibodies for PY-STAT5, PS-Akt, and P-ERK1/2. The blot was reprobed with total ERK1/2 (T-ERK1/2) antibody as a loading control. A, representative Western blots of at least three separate experiments when insulin pretreatment was 120 min. B and C, densitometric analysis of autoradiographs from three or more separate experiments to quantify P-ERK1/2 levels when insulin pretreatment was 120 min (B), or 30 min (C), 60 min (C), and 240 min (C), respectively. The data are expressed as mean Ϯ S.E. The P-ERK1/2 levels in untreated samples were arbitrarily set to 1 (see "Experimental Procedures" for definition of the levels of statistical significance). top of this elevated PS-Akt. Therefore, the enhanced activation of GHinduced signaling following insulin pretreatment was selective to the ERK1/2 pathway and was not observed for GH-induced activation of STAT5 or Akt.
Similar to insulin pretreatment for 120 min (Fig. 1, A and B), pretreatment with insulin for 30, 60, and 240 min also significantly enhanced GH-induced P-ERK1/2 (Fig. 1C). In the following experiments, to investigate the mechanism by which insulin enhanced GH-induced ERK1/2 phosphorylation, we chose insulin pretreatment for 120 min. After insulin pretreatment for 120 min, insulin-induced P-ERK1/2 had returned almost to basal levels (not true following 30 -60 min of insulin pretreatment) and the enhanced GH-induced P-ERK1/2 by insulin 120 min was greater than or equal to insulin pretreatment for 240 min (Fig. 1, B and C).
Phosphorylation of ERK1/2 Induced by Insulin and GH, Alone or Combined, Depended on MEK1/2 Activation-Two major cellular signaling pathways activated by insulin are the PI3K/Akt and MEK/ERK pathways (27). Activation of PI3K is required for ERK1/2 activation by GH in 3T3F442A cells, and GH-induced ERK1/2 activation is significantly augmented in insulin receptor substrate-1-expressing 32D-rGHR cells (6,28). This suggests that the insulin receptor substrate-1/ PI3K pathway may play an important role in GH signaling to ERK1/2. In H4IIE cells, PI3K signaling was not a major pathway activated by GH (Fig. 1A, row 2), but it could be significantly activated by insulin and remained activated at all insulin pretreatment time points used in the present studies. Thus, it was possible that insulin induction of PI3K was necessary for the enhanced GH-induced ERK1/2 activation following insulin pretreatment. To investigate the pathways involved in the enhanced GH-induced ERK1/2 activation following insulin pretreatment, H4IIE cells were treated with the MEK inhibitor (PD 98059) or the PI3K inhibitor (LY 294002) before insulin pretreatment. The MEK inhibitor (PD 98059) blocked P-ERK1/2 induced by GH and insulin, alone or combined (Fig. 2, A and B), indicating that MEK1/2 activation is an absolute requirement for GH-and insulin-induced activation of ERK1/2. However, the PI3K inhibitor (LY 294002) did not significantly block the effect of GH alone to induce P-ERK1/2. Most important for the present studies, the addition of LY 294002 did not alter the effect of insulin to greatly augment GH-induced P-ERK1/2 levels (Fig. 2, A and  B). This implies that PI3K does not play a major role in the enhancement of GH-induced P-ERK1/2 following insulin pretreatment.
Insulin Did Not Alter GH-induced Tyrosine Phosphorylation of JAK2-Following GH binding to GHR, JAK2 is phosphorylated and transphosphorylates GHR. To investigate the mechanisms by which insulin enhanced GH-induced ERK1/2 signaling, we first characterized the effects of insulin on GH-induced tyrosine phosphorylation of JAK2 (PY-JAK2). In H4IIE cells, GH significantly induced tyrosine phosphorylation of JAK2 by 5 min, reached a peak by 10 -20 min (3-fold), and returned to basal levels by 60 min or later (Fig. 3, A and B). It has been reported that insulin can directly activate JAK2 in some cell types (29,30), but insulin did not induce PY-JAK2 in H4IIE cells ( Fig. 3A and data not shown). After insulin pretreatment, there was no significant change in GH-induced PY-JAK2 (Fig. 3, A and B), suggesting that insulin may affect GH signaling either independent from or downstream of JAK2 activation.
GH, Alone or Combined with Insulin, Did Not Induce an Interaction between Insulin and GH Receptors in H4IIE Cells-We recently reported that GH induces association of GHR and the insulin-like growth factor-1 (IGF-1) receptor (IGF-1R) in murine 3T3-F442A preadipocytes, and GH and IGF-1 synergize in signaling via STAT5 and ERK1/2 pathways (31). Insulin pretreatment did not affect GH activation of JAK2 (Fig. 3), but it is possible that GH might utilize the insulin receptor (IR) to enhance its activation of intracellular signaling, similar to the use of the IGF-1R, and due to the high level of similarity between insulin and IGF-1 receptors. As a control for the techniques used, we were able to find, as previously reported (31), that treatment of 3T3-F442A cells with GH resulted in the appearance of a diffuse tyrosine phosphorylation of roughly 100 -120 kDa in the anti-IGF-1R precipitate (Fig. 4, rows 1 and 2 and lane 2 versus 1), which is the GHR. GH did not induce association of phosphorylated GHR and IR in 3T3-F442A cells (Fig. 4, rows 1 and 2, lanes 3 and 4 versus 2), possibly due to the low abundance of the IR protein in these cells (Fig. 4, row 3, lanes 3 and 4).   In rat H4IIE hepatoma cells, tyrosine phosphorylation of IR was strongly induced by insulin at 5 min and still remained highly elevated at 120 min (Fig. 4, rows 1 and 2, lanes 6 and 8 versus 5). GH did not induce tyrosine phosphorylation of IR, alone, or following insulin 120 min (Fig.  4, rows 1 and 2, lanes 7 and 9 versus 5). No GHR was co-immunoprecipitated with IR in response to GH, alone or following insulin pretreatment (Fig. 4, rows 1 and 2, lanes 5-9). IR was not detected in the immunoprecipitates with anti-GHR antibody under these conditions in H4IIE cells (data not shown), further suggesting a lack of interaction between insulin and GH receptors in the H4IIE cells used in the present study.
Effects of Insulin on GH-induced Formation of the Shc⅐Grb2⅐SOS Complex-When activated, both JAK2 and the IR can phosphorylate Shc. Activated Shc recruits Grb2⅐SOS, bringing SOS to the cell membrane, resulting in activation of the Ras/Raf-1/MEK/ERK pathway (32)(33)(34). In H4IIE cells, insulin alone induced strong and persistent phosphorylation of IR (Figs. 4 (rows 1 and 2) and 5A (row 1)), the 46-kDa isoform of Shc (Fig. 5A, row 2) and Raf-1 (Fig. 5A, row 3). Phosphorylation of all three proteins was increased by 5 min and were still highly elevated above basal values even 4 h following the continuous exposure to insulin and decreased between 4 and 19 h. All three isoforms of Shc (46, 52, and 66 kDa) were present in H4IIE cells, but the 46-and 52-kDa isoforms were more abundant compared with the 66-kDa isoform (Fig.  5, B and D, row 2). Of the three isoforms, only the 46-kDa isoform of Shc was strongly activated by insulin in H4IIE cells (Fig. 5A, row 2). Unlike phosphorylation of the IR, Shc, and Raf-1, phosphorylation of MEK1/2 (Fig. 5A, row 4) and ERK1/2 (Fig. 5A, rows 5 and 6) induced by insulin was more transient, reaching maximal levels by 5 min and returning toward basal levels thereafter.
Next, using immunoprecipitation, an increased association of Shc and Grb2 following insulin treatment was observed. This association was rapid but transient, with a detectable increase at 2.5 min, a maximal effect at 5 min, and a reduction to basal levels by 30 min (Fig. 5B, row 1,  and C). The association of Grb2⅐SOS did not change significantly during the 30-min insulin treatment period (Fig. 5B, row 3). GH-induced Shc⅐Grb2 association was weaker than that of insulin (Fig. 5, C versus E), with maximal association by 5-7.5 min, and Shc⅐Grb2 dissociation thereafter (Fig. 5, D (row 1, lanes 1-7) and E), with a less noticeable but consistent dissociation of SOS from Shc over this same time period (Fig.  5, D (row 3, lanes 1-7) and F).
Following insulin pretreatment for 120 min, GH-induced association of Shc⅐Grb2 was greater and more persistent than after GH alone (Fig. 5,  D (row 1, lanes 8 -14) and E), probably due to the increased basal level of Shc⅐Grb2 association following insulin pretreatment for 120 min (Fig. 5,  D (row 1, lane 8) and E (dashed line, 0 time point)). However, there was no significant change in Shc-associated SOS between 2.5 and 7.5 min and then a loss of Shc-associated SOS (Fig. 5, D (row 3, lanes 8 -14) and F). Following insulin pretreatment for 120 min, basal levels of Grb2⅐SOS association was only ϳ30% of that in untreated cells (Fig. 5, D (row 4,  lane 8 versus 1) and G (dashed line, 0 time point)), but the rate of Grb2⅐SOS dissociation in the first 10 min following GH addition was approximately parallel to that of cells untreated with insulin, suggesting no significant change (Fig. 5G).
Insulin Did Not Affect GH-induced GTP-bound Ras-Phosphorylated Shc recruits Grb2⅐SOS to the cell membrane where SOS can convert Ras from the GDP-bound (inactive) to the GTP-bound (active) form. As described above, the effects of insulin on GH-induced formation of Shc⅐Grb2⅐SOS were complex. Next we examined whether Ras activation followed the association and dissociation of Shc⅐Grb2⅐SOS complex. Insulin caused a rapid increase in the amount of GTP-bound Ras by 2.5 min, which peaked by 5 min, and then decreased by 7.5 min but remained elevated at 30 min (Fig. 6, A and B). GH-induced GTPbound Ras was more transient, with an increase by 5 min, a maximal 2-fold increase by 7.5 min and decreased by 30 min (Fig. 6, C and D). Compared with GH alone, there was a slight (but statistically insignificant) decrease of GH-induced GTP-bound Ras following insulin pretreatment for 120 min (Fig. 6, C and D). FIGURE 6. Effect of insulin on GH-induced GTP-bound Ras. H4IIE cells were treated: A, without or with 10 nM insulin for the indicated times, or C, pretreated without or with 10 nM insulin for 120 min followed by 500 ng/ml GH or vehicle for the indicated times. Whole cell lysates were affinity precipitated with Raf-1 Ras-binding domain agarose and then subjected to Western blot analysis by total Ras antibody to detect the activated Ras (upper panel), or whole cell lysate was directly resolved to SDS-PAGE and transferred to membrane, followed by probing with total Ras antibody as a loading control (lower panel). A and C, representative Western blots of three separate experiments. B and D, densitometric analysis of autoradiographs from three separate experiments to quantify activated Ras (Ras-GTP) levels. The data are expressed as mean Ϯ S.E. The Ras-GTP levels in untreated samples were arbitrarily set to 1.

Effects of Insulin on GH-induced Raf-1 Phosphorylation and Kinase
Activation-We next examined the effects of insulin on the ability of GH to activate the Raf-1 kinase, which is immediately downstream of Ras. Phosphorylation of Ser-338 on Raf-1 (PS338-Raf-1) is required for Raf-1 kinase activation (35). In H4IIE cells, GH alone induced a 2-fold increase of PS338-Raf-1 by 10 min (Fig. 7, A and B). As shown in Fig. 5A  (row 3), insulin-induced PS338-Raf-1 was still highly elevated 120 and 240 min after insulin addition. Following insulin pretreatment for 120 min, there was a further increase by GH addition, and the time-course curve of GH-induced PS338-Raf-1 was strikingly similar to GH alone, but shifted up due to the insulin pretreatment-induced increase of basal PS338-Raf-1 levels (Fig. 7B). If the elevated PS338-Raf-1 levels induced by 120 min insulin were subtracted before GH exposure, the PS338-Raf-1 induced by GH following insulin pretreatment was not significantly different from GH alone (Fig. 7B).
Even though P-ERK1/2 levels were not significantly elevated compared with control values following insulin pretreatment for 120 min (Fig. 1, A-C), the elevated PS338-Raf-1 level may be involved with the enhanced GH-induced ERK1/2 phosphorylation following insulin pretreatment. To test this possibility, Raf-1 kinase activity induced by GH, alone or following insulin pretreatment was examined by in vitro kinase assay. Raf-1 kinase activation induced by GH for 10 min was almost 2-fold of control levels (Fig. 7C), similar to the induction of PS338-Raf-1 (Fig. 7, A and B). After insulin pretreatment for 120 min, Raf-1 kinase activation was not significantly different from basal levels (Fig. 7C) even though PS338-Raf-1 was still elevated (ϳ3.5-fold, Fig. 7, A and B). More importantly for the present study, GH-induced Raf-1 kinase activity was not significantly altered by insulin pretreatment (Fig. 7C).
Insulin Enhanced GH-induced MEK1/2 Phosphorylation-GH-induced ERK1/2 phosphorylation was enhanced by insulin pretreatment, but the activation of signaling intermediates located upstream of MEK1/2, including JAK2, Shc⅐Grb2⅐SOS, Ras, and Raf-1, did not appear to be the cause of this enhancement. GH-induced ERK1/2 phosphorylation, alone or following insulin, required MEK1/2 activation (Fig. 2). Therefore, MEK1/2 kinase, which lies between Raf-1 and ERK1/2 would likely be involved with the enhanced GH-induced ERK1/2 phosphorylation by insulin. Therefore, we next examined whether insulin enhanced GH-induced MEK1/2 activation. In H4IIE cells, GH-induced serine phosphorylation of MEK1/2 (PS-MEK1/2) peaked (4-fold) by FIGURE 7. Effects of insulin on GH-induced serine 338 phosphorylation of Raf-1 and kinase activity. H4IIE cells were pretreated without or with 10 nM insulin for 120 min, followed by 500 ng/ml GH or vehicle for the indicated times. Western blot analysis was performed with PS338-Raf-1 antibody, followed by stripping and reprobing with total Raf-1 antibody. A, representative Western blots of three separate experiments. B, densitometric analysis of autoradiographs from three separate experiments to quantify PS338-Raf-1 levels. The data are expressed as mean Ϯ S.E. The PS338-Raf-1 levels in untreated samples were arbitrarily set to 1. C, whole cell lysates were immunoprecipitated with anti-Raf-1 antibody, and kinase assays were performed using GST-MEK1 as the substrate as described under "Experimental Procedures." Densitometric analysis of autoradiographs from three separate experiments was performed to quantify Raf-1 kinase activation (PS-GST-MEK1) levels. The data are expressed as mean Ϯ S.E. The Raf-1 kinase activation levels in untreated samples were arbitrarily set to 1.
7.5-10 min, decreased by 15 min and returned to near basal levels by 30 min (Fig. 8, A and B (solid line)). After insulin pretreatment for 120 min, insulin-induced MEK1/2 phosphorylation was near basal levels, but the maximal GH-induced MEK1/2 phosphorylation was increased to 9-fold after insulin pretreatment (Fig. 8, A and B (dashed line)).
Insulin Facilitated GH-induced Cell Membrane Translocation of MEK1/2-The present data indicate that GH-induced activation of kinases upstream of MEK1/2 was not changed following insulin pretreatment while phosphorylation of MEK1/2 and ERK1/2 was increased. One possible explanation is that the availability of MEK1/2 for Raf-1 was increased by insulin pretreatment. MEK1/2 moves to the cell membrane, where it can then be activated by Raf-1, leading to activation of ERK1/2 (36). We then asked whether insulin pretreatment affected GH-induced cellular localization of MEK1/2, which was determined by immunofluorescent staining. To make sure the results of immunofluorescent staining were reproducible, multiple fields with multiple cells in each field were examined, and similar results were obtained in three independent experiments. In untreated H4IIE cells, MEK1/2 was homogeneously distributed throughout the cytosol (Fig. 9,  A (panel a-1) and B (panels a-2 and a-3)). Treatment with insulin for 5 min caused a redistribution of most cytoplasmic MEK1/2 to the cell periphery (Fig. 9A, panel b). Some MEK1/2 protein was still located on the cell periphery following insulin pretreatment for 60 min (Fig. 9A,  panel c), but it was diffusely distributed in the cytosol after insulin for 120 min (Fig. 9A, panel d), resembling the pattern of MEK1/2 distribution in untreated cells. Addition of GH caused the MEK1/2 protein to group into small clusters scattered throughout the cell (Fig. 9, A (panel  e-1) and B (panels e-2 and e-3)), with little localization of MEK1/2 to the cell periphery.
Like insulin treatment for 5 min, when GH was added following insulin pretreatment for 120 min, a substantial fraction of MEK1/2 translocated to the cell periphery (Fig. 9, A (panel g-1) and B (panels g-2 and  g-3)), although a portion of MEK1/2 remained in the cytosol clustered similarly to GH addition alone. Similar results were observed when GH addition was following 60 min of insulin pretreatment (Fig. 9A, panel f). Because phosphorylation of ERK1/2 induced by GH following insulin pretreatment depended on MEK activation (Fig. 2), we next asked whether the MEK inhibitor (PD 98059) could affect the cell membrane translocation of MEK. As shown in Fig. 9A (panel h), PD 98059 did not prevent MEK protein from moving to the cell periphery. It is thought that PD 98059 interacts specifically with the dephosphorylated (inactive) conformation of MEK and blocks access to activating enzymes, such as Raf-1 (37). Our data suggest that, although PD 98059 binds to inactive MEK, MEK may still be able to move to the cell membrane upon stimulation. However, this membrane localized MEK cannot be activated by Raf-1, because PD 98059 blocks the availability of MEK for Raf-1. Therefore, we suggest that insulin pretreatment favored a GHinduced translocation of MEK1/2 to the cell membrane, where it was available for phosphorylation/activation by the Raf-1 kinase, leading to an enhancement of ERK1/2 activation.
Insulin-increased Association of MEK⅐KSR and GH-induced Tyrosine Phosphorylation of KSR-The mechanisms by which insulin facilitated GH-induced cell membrane translocation of MEK1/2 were next investigated. Kinase suppressor of Ras (KSR) is a scaffolding protein that associates with MEK1/2 and aids in the translocation of MEK from the cytosol to the plasma membrane (36, 38 -40). We therefore asked whether insulin affected the interaction between KSR and MEK1/2. Following the addition of insulin for 5 min or GH for 10 min, the amount of MEK that co-immunoprecipitated, and therefore was associated, with KSR was increased by ϳ2and 1.5-fold, respectively (Fig. 10,  A (row 1) and B). Following insulin pretreatment for 120 min, KSRbound MEK1/2 further increased to ϳ3.5-fold of control levels, with no further increase by a subsequent addition of GH (Fig. 10, A (row 1) and B). In contrast, insulin did not increase the association between KSR and 14-3-3 protein, which also binds to KSR (data not shown).
We also examined the effects of insulin on phosphorylation of KSR. Insulin and GH, alone or combined, induced tyrosine phosphorylation (PY) of KSR (Fig. 10A, row 2), but only the effects of insulin alone or GH following insulin pretreatment reached statistical significance (Fig.  10C). Therefore, insulin pretreatment induced PY-KSR and increased association of the MEK⅐KSR complex. . Insulin enhanced GH-induced MEK1/2 phosphorylation. H4IIE cells were pretreated without or with 10 nM insulin for 120 min, followed by 500 ng/ml GH or vehicle for the indicated times. Western blot analysis was performed with PS-MEK1/2 antibody, followed by stripping and reprobing with total MEK1/2 antibody. A, representative Western blots from three separate experiments. B, densitometric analysis of autoradiographs from three separate experiments to quantify PS-MEK1/2 levels. The data are expressed as mean Ϯ S.E. The PS-MEK1/2 levels in untreated samples were arbitrarily set to 1.

DISCUSSION
Although GH and insulin induce distinct signaling cascades, these two hormones can elicit some common physiological responses. For instance, in adipose cells, following an absence of GH, GH has insulinlike effects such as inhibition of lipolysis and stimulation of lipogenesis (41). Some evidence indicates cross-talk between insulin and GH intracellular signaling. Insulin was found to be necessary for maintenance of GHR levels in vivo (20 -23). Insulin induces tyrosine phosphorylation of JAK2 and STAT5 in some cell types, and direct protein-protein interaction between the insulin receptor (IR) and STAT5B has been demonstrated in the yeast two-hybrid system (42,43). The level of insulininduced tyrosine phosphorylation of STAT5A and STAT5B is markedly enhanced in skeletal muscle cells overexpressing IR (43). However, the possible effects of insulin on GH-induced post-GHR signaling have not been studied.
In the present work, insulin did not induce tyrosine phosphorylation of JAK2 and STAT5 in H4IIE cells, indicating that the activation of JAK2/STAT5 by insulin may be cell-type-specific. No effect of insulin pretreatment on GH-induced JAK2 phosphorylation was observed, which likely explains the lack of effect of insulin on the magnitude of GH-induced STAT5 phosphorylation. More importantly, we found that insulin pretreatment selectively enhanced GH signaling via ERK1/2. This occurred when insulin pretreatment was for 30 -240 min prior to GH exposure. Autophosphorylation of the insulin receptor further phosphorylates insulin receptor substrates, which serve as "docking" molecules, favoring the generation of at least two main intracellular signaling: the PI3K/Akt pathway and the Ras/Raf/MEK/ERK pathway (27,44,45). The present studies indicate that the MEK inhibitor (PD 98095) prevented the ERK1/2 phosphorylation induced by GH, alone or following insulin pretreatment, whereas the PI3K inhibitor (LY294002) did not. This suggests that MEK activation is required for ERK1/2 signaling induced by GH, alone or combined with insulin. In addition, the PI3K/Akt pathway plays little if any role in the enhancement of GHsignaling following insulin pretreatment.
GH induces association of the GHR and the IGF-1R in 3T3-F442A preadipocytes and synergizes with IGF-1 to induce activation of ERK1/2 and STAT5 (31). Although insulin and IGF-1 receptors share high levels of similarity, no interaction between insulin and GH receptors was induced by GH, alone or combined with insulin, in the H4IIE cells used in the present study. This suggests that the GHR-IGF-1R interaction may have specific determinants not shared by the related IR. Further studies will define the cell-type specificity of such interactions.
GH-induced association of Shc and Grb2 was more transient than that induced by insulin. In addition, GH resulted in a rapid dissociation of the Grb2⅐SOS complex by 30 min, but insulin did not significantly affect the Grb2⅐SOS complex during the same treatment period. Interestingly, following GH stimulation, Grb2⅐SOS complex dissociated faster than the Shc⅐SOS complex (Fig. 5, F and G). This is consistent with a previous study by Carter-Su and colleagues (15) in 3T3-F442A fibroblasts. Because this was not a major point of emphasis of the present study, and was not relevant to our main findings, we did not further pursue the underlying mechanisms. However, it is possible that there are two pools of Grb2⅐SOS activated by GH. Following GH treatment, one pool of Grb2⅐SOS is recruited to activated Shc, whereas another pool of Grb2⅐SOS is not. The second pool of Grb2⅐SOS (non-Shcbound) may dissociate faster than that of the first pool (Shc-bound Grb2⅐SOS). Therefore, the average dissociation rate of total Grb2⅐SOS, including both the Shc-bound Grb2⅐SOS and the non-Shcbound Grb2⅐SOS pools, may be faster than the Shc-bound SOS (Shc⅐Grb2⅐SOS) complex following GH treatment. The effects of insulin pretreatment on GH-induced formation of Shc⅐Grb2⅐SOS were complex, increasing both the association of Shc⅐Grb2 and the dissociation of Grb2⅐SOS. However, the final outcome was that insulin pretreatment slightly reduced the amount of GH-induced SOS recruited to Shc. This suggests that the insulin pretreatment increases GH-induced Grb2⅐SOS dissociation, and this overcomes the insulin effect to enhance GH-induced Shc⅐Grb2 association. The decreased association of SOS with Shc correlated with the slightly decreased (but statistically insignificant) activation of Ras induced by GH following insulin pretreatment.
Previous measurements of Raf-1 phosphorylation with the Ser-338 phosphospecific antibody suggests that phosphorylation of this site by growth factors, integrins, and Ras closely parallels the kinetics of kinase activation (46,47). However, when H4IIE cells were treated with insulin for 120 min, Raf-1 kinase activity was close to basal levels while phosphorylation of Ser-338 of Raf-1 remained elevated. In addition, although Ser-338 phosphorylation induced by GH following insulin pretreatment was significantly higher than GH alone, Raf-1 kinase activity was not changed, suggesting that Ser-338 phosphorylation was not representative of Raf-1 kinase activity induced by insulin or GH in H4IIE cells. This  -1) or treated as follows: b, insulin for 5 min; c, insulin for 60 min; d, insulin for 120 min e-1, GH for 10 min; f, insulin 60 min followed by GH for 10 min; g-1, insulin 120 min followed by GH for 10 min; and h, PD 98059 (50 M) for 30 min plus insulin 120 min followed by GH for 10 min. The concentrations of insulin and GH were 10 nM and 500 ng/ml, respectively. After fixation and permeabilization, the cells were stained with anti-MEK1/2 antibody followed by Alexa Fluor 488-conjugated antibody as described under "Experimental Procedures." Green: MEK1/2. Blue: nuclei. The cell periphery is indicated by arrows. B, similar results were obtained in the other two independent experiments. To indicate the reproducibility of these findings, H4IIE cells were treated with vehicle (a-2 and a-3), GH for 10 min (e-2 and e-3), or insulin 120 min followed by GH for 10 min (g-2 and g-3).
is consistent with other reports that, although Ser-338 phosphorylation is required for Raf-1 activation, the levels of phosphorylation do not always correlate with Raf-1 kinase activity (48,49). Therefore, Ser-338 phosphorylation may not be used as a surrogate marker of Raf-1 activation.
Further studies found that insulin enhanced GH-induced MEK/ERK activation by facilitating cell membrane translocation of MEK1/2, possibly involving its scaffolding protein, KSR. The intracellular localization of MEK1/2 is important for its activation and function. MEK1/2 moves to the cell membrane where it can be activated by Raf-1 and activate ERK1/2 (36). KSR has been identified as a scaffolding protein for MEK1/2 and plays integral roles in membrane-associated scaffolding of the Ras/Raf-1/MEK/ ERK pathway proteins. For instance, KSR constitutively binds to MEK1/2 (50,51), and, after growth factor stimulation, KSR translocates from cytosol to the plasma membrane, thus bringing MEK1/2 into close proximity with both Raf-1 and ERK1/2 (36, 38 -40). In the KSR knockout mouse, Ras and Raf-1 activation was normal, but MEK1/2 and ERK1/2 activation was impaired, indicating KSR is needed for proper activation of MEK1/2 and ERK1/2 in the Ras/Raf/MEK/ERK pathway (52).
In the present studies, insulin pretreatment did not affect GH-induced activation of Ras and Raf-1, but activation of MEK1/2 and ERK1/2 was enhanced, which correlated with increased translocation of MEK1/2 to the cell membrane. Further investigation suggested that insulin increased the amount of MEK bound to KSR. Insulin alone, GH alone, or the combination of insulin pretreatment and GH all moderately increased tyrosine phosphorylation of KSR. However, only shortterm insulin or pretreatment with insulin followed by GH treatment resulted in a significant induction of KSR tyrosine phosphorylation. Interestingly, we found an increased MEK translocation to the cell membrane only under these same conditions that resulted in a significant increase of KSR tyrosine phosphorylation. Previous studies found that, following epidermal growth factor, a serine residue of KSR, Ser-392, was dephosphorylated, followed by a rapid translocation of KSR to the cell membrane (53,54). However, the present study is the first to suggest a potential role of tyrosine phosphorylation of KSR in its cell membrane translocation and a possible role in regulated MEK co-trans-location to the cell membrane. Thus, we hypothesize that, in the absence of insulin, the ability of GH to induce MEK1/2 movement to the cell membrane is weak. However, following insulin pretreatment, possibly due to increased MEK⅐KSR association and GH-induced tyrosine phosphorylation of KSR, GH promotes a significant translocation of MEK1/2 to the plasma membrane, allowing interaction of MEK1/2 with activated Raf-1 and phosphorylation of ERK1/2, resulting in enhanced signal transduction through the Ras/Raf-1/MEK/ERK pathway. ERK1/2 activation mediates GH-induced expression of genes, including c-Fos, Egr-1, and Jun-B (5). Male-specific expression of hepatic cytochrome P450 isoforms is thought to be induced by GH via activation of STAT5. However, a recent report suggests that ERK1/2 may also be involved in episodic GH regulation of CYP2C11 (55). GH-induced ERK1/2 activation may also be involved in growth. The role of GH in promoting growth is also thought to be mediated by STAT5 activation. However, several recent reports indicate that STAT5 may not be the only pathway involved. A mutant GHR found in a family with markedly short stature shows increased ability to activate STAT5 with only minimal activation of the ERK1/2 pathway, compared with wild-type GHR after GH stimulation (56). Most recently, a GH variant manifesting normal activation of STAT5, but reduced activation of ERK1/2, was determined in a child with short stature (57). These reports suggest that ERK1/2 might also play an important role in mediating some of the growth-promoting effects of GH.
The biological effect of GH-induced ERK1/2 activation may not be fully realized without previous exposure to insulin. In normal human subjects, basal insulin concentrations in the hepatic portal circulation are ϳ0.2 nM (58), whereas peak post-prandial insulin concentrations reach ϳ3 nM in the portal circulation (58,59). In rats, basal hepatic portal insulin concentrations are similar, ϳ0.5 nM, whereas post-prandial insulin concentrations in the portal circulation were not measured (60). If the ratio of post-prandial to basal insulin in rats is similar to that found in humans, portal post-prandial insulin concentrations in rats would be expected to reach ϳ10 nM, the concentration of insulin used in the current experiments. Normal post-prandial concentrations of insulin may sensitize liver GH signaling via the MEK/ERK pathway. How- FIGURE 10. Insulin increased the association of MEK and KSR and GH-induced tyrosine phosphorylation of KSR. A, H4IIE cells were treated with or without insulin (10 nM) for the indicated times, followed by GH (500 ng/ml) or vehicle for the indicated times. Whole cell lysates were immunoprecipitated with anti-KSR antibody and then subjected to Western blot analysis with antibodies for MEK, PY, or KSR as indicated. Representative Western blots of three separate experiments are shown. B and C, densitometric analysis of autoradiographs from three similar experiments to quantify MEK⅐KSR association (B) and PY-KSR (C). The data are expressed as mean Ϯ S.E. The levels in untreated samples were arbitrarily set to 1. ever, in patients with deficient insulin secretion, such as malnutrition or Type 1 diabetes, the biological effects of GH via ERK1/2 activation may be weakened. Nutrition exerts an important effect upon pubertal growth. When malnutrition occurs, growth is inhibited (61). This may be due to the decreased insulin secretion and thereby insufficient ERK activation by GH. Type 1 diabetes is well known to adversely affect linear growth (62,63). However, with the more recent insulin treatment regimens, growth has been substantially improved and height in children with Type 1 diabetes today is similar to the height of their unaffected peers (62,64). Normal hepatic portal insulin concentrations are needed to modulate the hepatic GHR expression and indirectly promote growth (20 -23). Our current findings suggest that the role of insulin in growth may also involve changes in post-GHR signaling.
In summary, we found that insulin pretreatment increased GH signaling in H4IIE cells. Insulin pretreatment 1) specifically enhanced GHinduced ERK1/2 phosphorylation; 2) affected GH-induced formation of Shc⅐Grb2⅐SOS complex, but did not significantly alter the activation of other signaling intermediates upstream of MEK1/2, including JAK2, Ras, and Raf-1; 3) facilitated GH-induced cell membrane translocation and phosphorylation of MEK1/2; and 4) increased the association of MEK and KSR and tyrosine phosphorylation of KSR. The present study implies normal physiological insulin secretion may be necessary to enhance hepatic GH signaling, specifically GH activation of the MEK/ ERK signaling pathway, and this may be one of the cellular mechanisms by which insulin works in concert with GH to promote growth.