/04/$15.00/0 Molecular Endocrinology 18(6): Copyright 2004 by The Endocrine Society doi: /me

Transcription

1 /04/$15.00/0 Molecular Endocrinology 18(6): Printed in U.S.A. Copyright 2004 by The Endocrine Society doi: /me Physical and Functional Interaction of Growth Hormone and Insulin-Like Growth Factor-I Signaling Elements YAO HUANG, SUNG-OH KIM, NING YANG, JING JIANG, AND STUART J. FRANK Department of Medicine (Y.H., S.-O.K., N.Y., J.J., S.J.F.), Division of Endocrinology, Diabetes, and Metabolism, and Department of Cell Biology, University of Alabama at Birmingham, Birmingham, Alabama 35294; and Endocrinology Section (S.J.F.), Medical Service, Veterans Affairs Medical Center, Birmingham, Alabama GH and IGF-I are critical regulators of growth and metabolism. GH interacts with the GH receptor (GHR), a cytokine superfamily receptor, to activate the cytoplasmic tyrosine kinase, Janus kinase 2 (JAK2), and initiate intracellular signaling cascades. IGF-I, produced in part in response to GH, binds to the heterotetrameric IGF-I receptor (IGF- IR), which is an intrinsic tyrosine kinase growth factor receptor that triggers proliferation, antiapoptosis, and other biological actions. Previous in vitro and overexpression studies have suggested that JAKs may interact with IGF-IR and that IGF-I stimulation may activate JAKs. In this study, we explore interactions between GHR-JAK2 and IGF-IR signaling pathway elements utilizing the GH and IGF-I-responsive 3T3-F442A and 3T3-L1 preadipocyte cell lines, which endogenously express both the GHR and IGF-IR. We find that GH induces formation of a complex that includes GHR, JAK2, and IGF-IR in these preadipocytes. The assembly GH IS AN important regulator of growth and metabolism that is largely derived from the anterior pituitary gland. All known actions of GH require its specific binding to the GH receptor (GHR), a transmembrane glycoprotein member of the cytokine receptor superfamily (1, 2). The GHR is devoid of intrinsic tyrosine kinase activity, but is physically and functionally coupled to Janus kinase 2 (JAK2), a nonreceptor cytoplasmic tyrosine kinase member of the JAK family (3, 4). GH-induced JAK2 activation results in engagement of several intracellular signaling pathways, including STATs (signal transducers and activators of transcription) -1, -3, and -5, ERKs, and phosphatidylinositol 3-kinase (PI3K) (reviewed in Refs. Abbreviations: EGFR, Epidermal growth factor receptor; GHR, GH receptor; IGF-IR, IGF-I receptor; JAK, Janus kinase; LIF, leukemia-inhibitory factor; MEK, MAPK kinase; PI3K, phosphatidylinositol 3-kinase; ptyr, phosphotyrosine; SDS, sodium dodecyl sulfate; STAT, signal transducer and activator of transcription. Molecular Endocrinology is published monthly by The Endocrine Society ( the foremost professional society serving the endocrine community. of this complex in intact cells is rapid, GH concentration dependent, and can be prevented by a GH antagonist, G120K. However, it is not inhibited by the kinase inhibitor, staurosporine, which markedly inhibits GHR tyrosine phosphorylation. Moreover, complex formation does not appear dependent on GH-induced activation of the ERK or phosphatidylinositol 3-kinase signaling pathways or on the tyrosine phosphorylation of GHR, JAK2, or IGF-IR. These results suggest that GH-induced formation of the GHR-JAK2-IGF-IR complex is governed instead by GH-dependent conformational change(s) in the GHR and/or JAK2. We further demonstrate that GH and IGF-I can synergize in acute aspects of signaling and that IGF-I enhances GH-induced assembly of conformationally active GHRs. These findings suggest the existence of previously unappreciated relationships between these two hormones. (Molecular Endocrinology 18: , 2004) 2, 5, and 6). Activation of STAT5b is prominent in GH signaling and is critical for regulation of several GH-responsive hepatic genes, including the IGF-I gene (7 9). Activation of ERKs has been linked to GH-induced activation of c-fos gene expression (10) and fosters cross-talk between the GHR and epidermal growth factor receptor family signaling pathways (11, 12). The mechanisms by which GH exerts its somatogenic and metabolic effects have been studied intensively but are still incompletely understood. A major physiological effector of GH is IGF-I (also known as somatomedin-c), the expression and secretion of which is induced by GH in various target tissues. As originally articulated, the somatomedin hypothesis of GH action postulated that GH stimulated the hepatic secretion of IGF-I, which then functioned in an endocrine manner to interact with IGF-I receptors (IGF-IR) in tissues that responded with growth (13, 14). Recent studies of mice with unrestricted targeted deletion of the IGF-I gene validated the importance of IGF-I in growth mediation but also suggested that GH may exert growth-promoting activity in some tissues even in the absence of IGF-I (15, 16). Further, liver-specific 1471

2 1472 Mol Endocrinol, June 2004, 18(6): Huang et al. Interaction of GH and IGF-I Signaling Elements knockout of IGF-I, although dramatically lowering serum IGF-I levels, does not prevent normal growth, suggesting that autocrine/paracrine IGF-I (rather than liver-derived IGF-I) may predominate for normal postnatal growth (17, 18). Interestingly, a recent report of the effect of combined knockout of the GHR and IGF-I genes compared with individual knockout of each suggested that GH and IGF-I may independently contribute to growth in the mouse (19). IGF-I signals through the IGF-IR, a cell surface heterotetramer with similarity to the insulin receptor that has intrinsic kinase activity embedded in its -subunit cytoplasmic domains (20, 21). IGF-IR activation engages the ERK and PI3K pathways via Src homology and collagen domain protein and insulin receptor substrate-1 phosphorylation to effect proliferation, antiapoptosis, and other biological actions (22, 23). Recent reports suggest that, in addition to causing activation of their own receptor kinases, IGF-I and insulin may activate members of the JAK family, including JAK1 and JAK2 (24 27). The impact of JAK activation by these growth factors is not clear in all cases and in most instances has been observed in the setting of overexpression of at least one of the components. In one study, recombinant proteins were used in vitro to demonstrate physical association of JAK1 with insulin and IGF-I receptors, which was dependent on tyrosine phosphorylation of the receptors (28). We now explore the potential for interaction between GHR-JAK2 and IGF-IR signaling pathway elements. In these studies, we utilize the GH and IGF-Iresponsive 3T3-F442A and 3T3-L1 preadipocyte cell lines, which endogenously express both the GHR and IGF-IR. We uncover evidence for the GH-induced formation of a complex that includes GHR, JAK2, and IGF-IR in these preadiopocytes. The assembly of this complex in intact cells can be prevented by a GH antagonist, G120K, but is not inhibited by the kinase inhibitor, staurosporine. Moreover, complex formation does not appear dependent on GH-induced activation of the ERK or PI3K signaling pathways or on the tyrosine phosphorylation of GHR, JAK2, or IGF-IR. These results suggest that GH-induced formation of the GHR-JAK2-IGF-IR complex is governed instead by GH-dependent conformational change(s) in the GHR and/or JAK2. We further demonstrate that GH and IGF-I can synergize in acute aspects of signaling and that IGF-I enhances GH-induced assembly of conformationally active GHRs. These findings suggest the existence of previously unappreciated relationships between these two hormones. RESULTS GH-Induced Formation of a Complex that Includes the IGF-IR and the Tyrosine- Phosphorylated GHR and JAK2 Interactions between JAKs and the IGF-I and insulin receptors have previously been observed in overexpression systems (28). Given the important physiological relationships between GH and IGF-I action, we asked whether cross-talk might also exist between elements of these two signaling systems. The 3T3- F442A preadipocyte is an appealing system in which to study GH signaling, in that it endogenously expresses GHR and responds to GH with detectable activation of multiple pathways (3, 10 12, 29 33). Specific binding sites for radioiodinated IGF-I have also been detected on 3T3-F442A preadipocytes, with an estimated 380,000 IGF-I receptors per cell (34). Our initial studies used these cells. To examine their effects on tyrosine phosphorylation of proximal GH and IGF-I signaling elements, GH, IGF-I, or vehicle (-) were incubated with serum-starved 3T3-F442A cells for 15 min before detergent extraction. Extracted proteins were immunoprecipitated with anti-ghr cyt-al47 (Fig. 1A), anti-jak2 AL33 (Fig. 1B), or anti-igf-i receptor -chain (anti-igf-ir; Fig. 1C) anti- Fig. 1. GH Acutely Induces Appearance of a Tyrosine Phosphoprotein Complex in the Anti-IGF-IR Precipitates A C, Serum-starved 3T3-F442A cells were stimulated with GH (500 ng/ml), IGF-I (20 ng/ml), or vehicle ( ) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-ghr cyt-al47 (A), anti-jak2 AL33 (B), and anti-igf-ir (C) antibodies, respectively. Eluted proteins were analyzed by immunoblotting with anti-ptyr (A C, lanes 1 3), anti- GHR cyt-al47 (A, lanes 4 6), anti-jak2 AL33 (B, lanes 4 6), and anti-igf-ir (C, lanes 4 6). Migration of prestained molecular weight markers is indicated. Also indicated are the expected positions of GHR (bracket), JAK2 (arrow), and IGF-IR -chain (arrowhead). The experiments shown are representative of six such experiments.

3 Huang et al. Interaction of GH and IGF-I Signaling Elements Mol Endocrinol, June 2004, 18(6): bodies, resolved by SDS-PAGE, and immunoblotted with antiphosphotyrosine (anti-ptyr; Fig. 1, A C, lanes 1 3). As expected, anti-ghr cyt-al47 precipitation revealed GH-induced tyrosine phosphorylation of a collection of proteins in the 100- to 125-kDa range, which included a diffuse lower band ( kda) closely approximated with a sharper band of 125 kda (Fig. 1A, lane 2 vs. 1; bracket and arrow, respectively). Their migration and appearance in response to GH were consistent with these bands being the tyrosine-phosphorylated GHR and JAK2, as we have previously detected in anti-ghr precipitates from these cells (31, 35). Notably, this complex of tyrosine phosphoproteins was not detected in response to IGF-I (Fig. 1A, lane 3 vs. 1). Reprobing of this blot with anti-ghr cyt- AL47 revealed the GHR as the diffuse band, the migration of which was retarded in response to GH [as seen previously (31)], but not in response to IGF-I (Fig. 1A, lanes 4 6). Also as expected (11, 31, 35), GH caused tyrosine phosphorylation of JAK2, as detected in the direct anti-jak2 AL33 precipitates (Fig. 1B, lane 2 vs. 1). IGF-I treatment, in contrast, did not result in JAK2 tyrosine phosphorylation (lane 3 vs. 1). Reprobing with anti-jak2 AL33 verified that equivalent amounts of JAK2 were precipitated from each sample (Fig. 1B, lanes 4 6). In contrast to its inability to promote GHR and JAK2 tyrosine phosphorylation, IGF-I treatment did cause IGF-IR -chain (arrowhead at 95 kda) tyrosine phosphorylation (Fig. 1C, lane 3 vs. 1), reflecting the productive IGF-I binding in these cells. Interestingly, treatment with GH resulted in the appearance of a diffuse tyrosine phosphoprotein of roughly kda in the anti-igf-ir precipitate (Fig. 1C, lane 2 vs. 1), without evident tyrosine phosphorylation of the IGF-IR -chain itself. Reprobing with anti-igf-ir confirmed similar IGF-IR abundance in each precipitate (Fig. 1C, lanes 4 6). The size and pattern of migration of the GH-induced tyrosine phosphoprotein detected in the anti-igf-ir precipitate suggested that it might contain GHR and JAK2. These same findings were obtained when an independently derived GH- and IGF-I-responsive preadipocyte cell line, 3T3-L1, was used (as in Figs. 3, 5, and 7, below) To better characterize the tyrosine phosphoproteins seen in response to GH and IGF-I, we exploited the fact that both the GHR and IGF-IR are glycoproteins. Like many glycoproteins, the particular patterns of glycosylation dictate particular alterations in the migration of each receptor in SDS-PAGE. Thus, enzymatic deglycosylation of immunoprecipitated proteins enhances the specificity of identification of the immunoisolated tyrosine phosphoproteins. Enzymatic deglycosylation experiments are shown in Fig. 2. Serumstarved cells were stimulated with GH, IGF-I, or vehicle control for 15 min before detergent extraction. Extracted proteins were then immunoprecipitated with anti-igf-ir (Fig. 2, lanes 1 6 and 9 12). As a control for the migration pattern of the GHR, anti-ghr cyt-al47 immunoprecipitation was also carried out (Fig. 2, lanes 7 and 8). Precipitated proteins were either treated with the combination of endoglycosidase F/N and glycosidase F ( ; lanes 2, 4, 6, 8, 10, and 12) or glycosidase buffer only ( ; lanes 1, 3, 5, 7, 9, and 11), as indicated, resolved by SDS-PAGE, and immunoblotted with antiptyr (lanes 1 8) or anti-igf-ir -chain (lanes 9 12). As we have previously observed (31), the directly precipitated tyrosine-phosphorylated GHR appearing in response to GH migrated more rapidly after deglycosylation, changing from kda to kda (lanes 7 8, labeled GHR and degly GHR, respectively). Notably, the sharp 125-kDa tyrosine phosphoprotein present in the anti-ghr precipitates of GH-treated cells did not change its migration with deglycosylation, consistent with its being the coimmunoprecipitated JAK2 (lanes 7 and 8, indicated by arrow). The IGF-IR -chain directly precipitated with anti- IGF-IR antibody also displayed enhanced migration Fig. 2. Characterization of GH-Induced Tyrosine Phosphoprotein Complex in the Anti-IGF-IR Precipitates by Enzymatic Deglycosylation Serum-starved 3T3-F442A cells were stimulated with GH (500 ng/ml; lanes 3, 4, and 7 10), IGF-I (20 ng/ml; lanes 5, 6,11, and 12), or vehicle (lanes 1 and 2) for 15 min. Detergent extracts (1000 g) were immunoprecipitated with anti-igf-ir (lanes 1 6, 9 12) and anti-ghr cyt-al47 (lanes 7 and 8). Precipitates were treated with ( ) or without (-) endoglycosidase F/N-glycosidase F, as described in Materials and Methods, before resolution by SDS-PAGE and immunoblotting with anti-ptyr (lanes 1 8) and anti-igf-ir (lanes 9 12). Positions of JAK2 (arrow), glycosylated and deglycosylated GHRs (brackets), and IGF-IRs (arrowheads) are indicated. Note the effects of deglycosylation on the pattern of SDS-PAGE migration of GH-induced tyrosine phosphoproteins in IGF-IR precipitates (lanes 3 and 4) compared with that of the GH-induced phospho-ghr-jak2 complex in GHR precipitates (lanes 7 and 8). The experiment shown is representative of three such experiments.

4 1474 Mol Endocrinol, June 2004, 18(6): Huang et al. Interaction of GH and IGF-I Signaling Elements Fig. 3. Effects of GH Antagonist and GH Concentration on Formation of anti-igf-ir-precipitated Tyrosine Phosphoprotein Complex A, G120K antagonism. Serum-starved 3T3-L1 cells were stimulated with vehicle (lane 1), GH (500 ng/ml, lane 2), IGF-I (20 ng/ml, lane 3), G120K (500 ng/ml ( 1), lane 4; 2500 ng/ml ( 5), lane 5), or GH (500 ng/ml) plus G120K (2500 ng/ml) (lane 6) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-jak2 AL33 (upper panel) and anti-igf-ir (lower panel), respectively. Immunoprecipitates were analyzed by immunoblotting with anti-ptyr. B, GH concentration dependence. Serum-starved 3T3-L1 cells were stimulated with various concentrations of GH for 15 min, as indicated. Detergent extracts (500 g) were subjected to immunoprecipitation with anti-igf-ir and immunoblotting with anti-ptyr. Note decreased coprecipitation of tyrosine phosphoproteins at 1000 ng/ml GH vs. 500 ng/ml (lane 6 vs. 5). The experiments shown in panels A and B are representative of two such experiments. after deglycosylation, shifting from 95 kda (lanes 5, 9, and 11; labeled IGF-IR ) to approximately 75 kda (lanes 6, 10, and 12; labeled degly IGF-IR ), whether detected by anti-ptyr after IGF-I treatment (lanes 5 and 6) or by anti-igf-ir after either GH or IGF-I treatment (lanes 9 12). Notably, when subjected to glycosidase treatment, the GH-induced tyrosine phosphoproteins precipitated with anti-igf-ir (lanes 3 and 4) adopted a pattern of SDS-PAGE migration very similar to the tyrosine phosphoproteins induced by GH and precipitated with anti-ghr cyt-al47 (compare lane 4 with lane 8). These findings furthered the conclusion that the IGF-IR antibody precipitated GH-induced tyrosine phosphoproteins that included the GHR and JAK2. The active signaling conformation of the GHR induced by GH is as a dimer bound to a single GH molecule (36). Recent studies suggest that receptor is likely a preformed dimer in the absence of ligand and that the active assemblage arises by a GH-induced conformational change in receptor dimer partners, which somehow allows more productive interaction between the GHR and JAK2 and resultant triggering of signaling (37 41). A class of recombinant GH antagonists harboring mutations at residues known to be critical for promotion of the active receptor dimer conformation have been developed (42, 43). One of these, hgh-g120k (G120K), does not itself induce GHR and JAK2 tyrosine phosphorylation in 3T3-F442A cells and antagonizes the ability of GH to induce these effects; additionally, GH-induced GHR disulfide linkage, a reflection of the receptor conformational change, is also antagonized by G120K (35). We used G120K to further characterize the GHinduced accumulation of tyrosine phosphoproteins in IGF-IR immunoprecipitates (Fig. 3A). Like 3T3-F442A, the independently derived 3T3-L1 preadipocyte bears both GHR (33) and IGF-IR (44, 45). In these experiments, we used 3T3-L1, first demonstrating that, as in 3T3-F442A cells, treatment of 3T3-L1 induced tyrosine phosphorylation of JAK2 and coimmunoprecipitation of tyrosine phosphoproteins with anti-igf-ir -chain (Fig. 3A, upper and lower panels, respectively, lanes 2 vs. 1). This illustrates the similarity of these findings in the two preadipoctyes. G120K alone induced neither effect (lanes 4 and 5 vs. 1). Moreover, G120K substantially antagonized both effects of GH at a GH to G120K ratio of 1:5 (lanes 6 vs. 2), strongly suggesting that the coimmunoprecipitated tyrosine phosphoproteins were GHR and JAK2. The GH concentration dependency of the coimmunoprecipitation was examined in Fig. 3B. The association of tyrosine-phosphorylated GHR and JAK2 with IGF-IR -chain was detected with as little as 20 ng/ml GH (lane 2 vs. 1) and declined at a concentration greater than 500 ng/ml (lane 6 vs. 5). This fall off in coprecipitation at high GH concentrations is reminiscent of that observed for other GH-induced phenomena and likely relates to the inability of GH to effectively induce the conformational change(s) necessary for receptor triggering (35, 40, 42). GH-Induced GHR-JAK2-IGF-IR Complex Formation Does Not Require GHR Tyrosine Phosphorylation The data in Figs. 1 3 provide evidence for GH-induced association of tyrosine-phosphorylated GHR and JAK2 with the IGF-IR. However, they do not address whether association of the molecules is constitutive or occurs only in response to hormonal stimulation and/or tyrosine phosphorylation. This was initially ap-

5 Huang et al. Interaction of GH and IGF-I Signaling Elements Mol Endocrinol, June 2004, 18(6): Fig. 4. GH Induces Association of Tyrosine-Phosphorylated GHR and JAK2 with IGF-IR A, GH induces the formation of GHR-JAK2-IGF-IR complex. Serum-starved 3T3-F442A cells were stimulated with vehicle (lane 1), GH (500 ng/ml; lane 2), or IGF-I (20 ng/ml; lane 3) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-igf-ir. The immunoprecipitates were analyzed by immunoblotting with anti-ptyr, anti- GHR cyt-al47, anti-jak2 AL33, and anti-igf-ir, respectively. The experiments shown are representative of five such experiments. B and C, Time course of GH-induced GHR-JAK2- IGF-IR complex formation. Serum-starved cells were stimulated with GH (500 ng/ml) for 0 30 min as indicated. Detergent extracts (500 g) were immunoprecipitated with anti-jak2 (B, upper panel), anti-ghr cyt-al47 (B, lower panel), or anti-igf-ir (C). The immunoprecipitates were analyzed by immunoblotting with anti-ptyr (B and C, upper panel), anti- proached by immunoblotting anti-igf-ir precipitates from cells treated with vehicle, GH, or IGF-I sequentially with anti-ptyr, anti-ghr cyt-al47, anti-jak2 AL33, and anti-igf-ir (Fig. 4A). GHR and JAK2 were coprecipitated with anti-igf-ir only in response to GH treatment (lane 2 vs. 1) and not in response to IGF-I treatment (lane 3 vs. 2). In concert with the data in Fig. 3, these results indicated that the conformational change(s) in GHR and/or JAK2 induced by GH treatment are apparently required for their inclusion in a complex with IGF-IR -chain. We note that IGF-IR -chain was not detected in immunoprecipitates with our available anti-ghr or anti-jak2 reagents either in the absence or presence of GH stimulation (data not shown), further suggesting conformational sensitivity for detection of the intact complex. The time course of GH-induced GHR-JAK2-IGF-IR complex formation was examined in Fig. 4, B and C. Cells stimulated with GH from 0 30 min (lanes 2 5 vs. 1) were lysed and proteins were immunoprecipitated with either anti-jak2 AL33 and anti-ghr cyt-al47 (Fig. 4B) or anti-igf-ir (Fig. 4C). Precipitated proteins were immunoblotted with anti-ptyr, anti-ghr AL-47, or anti- JAK2 AL33, as indicated. Consistent with previous data (3, 31), this revealed that GH rapidly induced robust tyrosine phosphorylation of the GHR and JAK2. Likewise, association of GHR and JAK2 with the IGF-IR was detectable after only 1 min of exposure to GH. Thus, the time courses of GH-induced JAK2 activation/ghr tyrosine phosphorylation and GHinduced formation of the GHR-JAK2-IGF-IR complex were similar. To determine whether complex formation was dependent on GH-induced tyrosine phosphorylation, we sought to uncouple GHR engagement from GHR tyrosine phosphorylation. We previously demonstrated that pretreatment of cells with the kinase inhibitor, staurosporine (1.25 M), markedly decreased GHinduced tyrosine phosphorylation of the JAK2, GHR, and STAT5, but did not prevent GH-induced GHR disulfide linkage (a proxy for GH-induced conformational change of the receptor) or enhancement of the GHR-JAK2 association (35, 46). Cells were treated with ( ) or without ( ) GH in the presence of staurosporine ( ) or its vehicle ( ) and evaluated for GHinduced GHR tyrosine phosphorylation (Fig. 5, A and B) and coprecipitation of GHR with anti-igf-ir (Fig. 5, C and D). Consistent with previous results, staurosporine dramatically inhibited GH-induced GHR tyrosine phosphorylation (Fig. 5A, lane 3 vs. 2). However, this did not prevent GH-induced association of the GHR with IGF-IR (Fig. 5C, lane 3 vs. 2). These data strongly suggest that the formation of the GHR-JAK2- IGF-IR complex does not require GHR tyrosine phos- GHR cyt-al47 (C, middle panel), and anti-jak2 AL33 (C, lower panel), respectively. The experiments shown are representative of two such experiments.

6 1476 Mol Endocrinol, June 2004, 18(6): Huang et al. Interaction of GH and IGF-I Signaling Elements Fig. 5. Tyrosine Phosphorylation of GHR Is Not Required for GH-Induced GHR-JAK2-IGF-IR Complex Formation A D, Serum-starved 3T3-L1 cells were pretreated with vehicle (lanes 1 and 2) or 1.25 M staurosporine (lane 3) for 15 min before stimulation with vehicle (lane 1) or GH (500 ng/ml; lanes 2 and 3) for 15 min, as indicated. Detergent extracts (500 g) were immunoprecipitated with anti-ghr cyt-al47 (A and B) or anti-igf-ir (C and D). The immunoprecipitates were analyzed by immunoblotting with anti-ptyr (A), anti-ghr cyt- AL47 (B and C), and anti-igf-ir (D), respectively. (Note that the retarded migration of the GHR in GH-treated samples relative to those not treated with GH (B) and those treated with GH in the presence of staurosporine (B and C) is likely related to its tyrosine phosphorylation.) The experiments shown are representative of three such experiments. phorylation, but does require GH-induced GHR conformational change, as implicated by the data in Fig. 3. Specificity of GH-Induced Formation of the GHR- JAK2-IGF-IR Complex Fig. 6. Specificity of GH-Induced Formation of GHR-JAK2- IGF-IR Complex A C, Serum-starved 3T3-F442A cells were stimulated with vehicle (lanes 1 and 6), GH (500 ng/ml; lanes 2 and 4), or IGF-I (20 ng/ml; lanes 3 and 5) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-igf-ir (lanes 1 3) and anti-egfr (as an immunological specificity control; lanes 4 6), respectively. Eluted proteins were analyzed by immunoblotting with anti-ptyr (A), anti-igf-ir (B), or anti-egfr (C). The GH dependency of GHR-JAK2-IGF-IR -chain complex formation seen in Figs. 4 and 5 argued that the interaction is specific. However, we pursued further experiments to validate this conclusion. Because GH-induced GHR and JAK2 tyrosine phosphorylation in 3T3-F442A and 3T3-L1 cells is extensive (see Fig. 1, A and B, and Fig. 5 and Refs. 3 and 31), their coprecipitation could, in principle, reflect a nonspecific association with the reagents used for immunoprecipitation. As an immunological specificity control, extracted proteins from cells stimulated with GH, IGF-I, or vehicle were precipitated with anti-epidermal growth factor receptor (EGFR) antibody in comparison with anti-igf-ir antibody (Fig. 6). Both are affinity purified rabbit antibodies, and both were adsorbed by protein A-sepharose. The anti-igf-ir precipitate of the GH-treated cells again contained the tyrosine-phosphorylated GHR-JAK2 complex (Fig. 6A, lane 2 vs. 1), whereas that from the IGF-I-treated cells contained the tyrosine-phosphorylated IGF-IR -chain and coprecipitated tyrosine phosphoproteins, the migration of which was consistent with being insulin receptor substrate-1 and the p66 and p52 isoforms of Src homology and collagen domain protein (lane 3), as expected. In contrast, none of these GH- or IGF-I-induced tyrosine phosphoproteins were detected in anti-egfr precipitates of the same samples (Fig. 6A, lanes 4 6). In particular, we note that GH-induced tyrosine phosphorylation of the EGFR was observed (lane 4 vs. 6), consistent with previous findings (11, 12, 47), but that the tyrosine-phosphorylated GHR and JAK2 were not detected in this same precipitate. Anti- IGF-IR (Fig. 6B) and anti-egfr (Fig. 6C) blotting verified that each of these molecules was similarly abundant in its respective specific immunoprecipitates. Thus, the GHR and JAK2 tyrosine phosphorylated in response to GH did not associate with the GH-induced tyrosine-phosphorylated EGFR or the reagents used to specifically immunoprecipitate it. This favors the interpretation that their inclusion in IGF-IR precipitates in response to GH reflects a specific association. This interpretation is even more compelling in that the

7 Huang et al. Interaction of GH and IGF-I Signaling Elements Mol Endocrinol, June 2004, 18(6): control antibody used (anti-egfr) does directly recognize its target (the EGFR) and is thus an even more powerful specificity control than would be a nonimmune antibody. JAK2 is activated by a wide range of cytokines and hormones in addition to GH (reviewed in Refs. 2 and 48). As data from overexpression systems point to an association of JAKs with the insulin and IGF-I receptors (28), we investigated whether another stimulus that activates JAK2 can also lead to its association with the IGF-IR. Leukemia inhibitory factor (LIF) is known to cause JAK2 kinase activation and tyrosine phosphorylation in both 3T3-F442A and 3T3-L1 cells (3, 11, 49). As seen in Fig. 7A (middle panel), JAK2 tyrosine phosphorylation was easily detected in 3T3-L1 cells in response to both GH and LIF (lanes 2 and 3 vs. 1). As expected, GH induced coimmunoprecipitation of tyrosine-phosphorylated GHR and JAK2 with IGF-IR (Fig. 7A, upper panel; lane 2 vs. 1). However, despite LIF s induction of JAK2 tyrosine phosphorylation, no tyrosine-phosphorylated JAK2 was detected by coprecipitation with the IGF-IR, even when the anti-ptyr immunoblot was subjected to prolonged exposure (Fig. 7A, upper panel; and Fig. 7B, lanes 3 vs. 2). Along with the data in Fig. 5, these results indicated that the complex formation observed in response to GH is more specifically related to the presence of the GH-engaged GHR rather than the ability of JAK2 to become activated. Although inhibition of GH-induced GHR tyrosine phosphorylation was dramatic (Fig. 5), the inhibition of GH-induced JAK2 tyrosine phosphorylation by staurosporine was incomplete (Ref. 35 and data not shown). Thus, we examined whether the formation of the GH-induced GHR-JAK2-IGF-IR complex was dependent on GH-induced JAK2-mediated signals. In particular, we focused on the ERK and PI3K activation pathways. Cells were pretreated with either PD98059 (Fig. 8A) or LY (Fig. 8B) to inhibit the ERK or PI3K pathways, respectively. In neither case did pathway inhibition prevent GH-induced coimmunoprecipitation of tyrosine-phosphorylated GHR and JAK2 with anti-igf-ir (Fig. 8, A and B, upper panels; lanes 4 vs. 2). Immunoblotting with the anti-perk and anti-pakt activation-specific antibodies verified that each inhibitor substantially diminished the activation of the pathways (Fig. 8, A and B, lower panels; lanes 4 vs. 2). Collectively, these findings further supported the conclusion that GH-induced GHR conformational changes, rather than GH-induced signaling, are responsible for allowing formation of the GHR-JAK2- IGF-IR complex. Synergistic Signaling Effects of GH and IGF-I Previous reports have suggested that GH and IGF-I may act synergistically in promoting cellular responses such as gene activation and proliferation (50, 51). Our findings that GH acutely induces formation of a GHR, JAK2, and IGF-IR complex encouraged us to examine aspects of signaling in response to GH or IGF-I in comparison with that in response to the combination of GH plus IGF-I. We first examined activation of a luciferase reporter gene driven by tandem copies of the serine protease inhibitor (Spi) 2.1 -interferon-activated sequence-like element (Spi-GLE-luc). Spi 2.1 is a GH-responsive gene (52, 53), and we have used this reporter in the past to monitor GH-induced gene activation (32, 54, 55). 3T3-F442A cells transiently transfected with Spi-GLE-luc were divided into replicates and treated with vehicle, GH (100 ng/ml), IGF-I (20 ng/ml), or GH plus IGF-I for 16 h, after which luciferase activity was measured. Compared with vehicle, GH induced substantial luciferase activity (varying from Fig. 7. LIF, a JAK2 Activator, Does Not Induce the Formation of GHR-JAK2-IGF-IR Complex A and B, Serum-starved 3T3-L1 cells were stimulated with vehicle (lane 1), GH (500 ng/ml; lane 2), or LIF (20 ng/ml; lane 3) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-igf-ir (A, upper panel), or anti-jak2 AL33 (A, middle and lower panels). Eluted proteins were analyzed by immunoblotting with anti-ptyr (A, upper and middle panels) or anti-jak2 AL33 (A, lower panel). B, A longer exposure of the immunoblot presented as the upper panel of A.

8 1478 Mol Endocrinol, June 2004, 18(6): Huang et al. Interaction of GH and IGF-I Signaling Elements Fig. 8. The GH-Induced Formation of GHR-JAK2-IGF-IR Complex Is Independent of GH-Induced Activation of ERK and PI3K Pathways A and B, Serum-starved 3T3-F442A cells were preincubated with PD98059 (100 M) for 60 min (A, lanes 3 and 4) or LY (50 M) for 30 min (B, lanes 3 and 4) before stimulation with vehicle (lanes 1 and 3) or GH (500 ng/ml; lanes 2 and 4) for 15 min. Detergent extracts (500 g) were immunoprecipitated with anti-igf-ir (A and B, upper and middle panels). Eluted proteins were immunoblotted with anti-ptyr (A and B, upper panels) or anti-igf-ir (A and B, middle panels). Detergent extracts (30 g), as used for immunoprecipitation in A and B, upper and middle panels, were directly resolved by SDS-PAGE and immunoblotted with antiactive ERK (A, lower panel) and anti-pakt (B, lower panel) to 13.9-fold, depending on the particular experiment). A representative experiment is shown in Fig. 9A, in which GH induced 6.1-fold increase in luciferase activity over basal and IGF-I-induced transactivation was negligible. However, the combination of GH plus IGF-I induced significantly greater Spi-GLE-luc transactivation (10.1-fold over basal) than did GH alone, suggesting a synergistic effect of the addition of IGF-I to GH. Pooled data from three independent experiments were evaluated and are shown in Fig. 9B. In this display, GH-induced transactivation within each experiment was considered as 100%. IGF-I cotreatment augmented GH-induced transactivation by % (P 0.05). In other experiments, augmentation by IGF-I was also observed when GH was used at Fig. 9. GH and IGF-I Cotreatment Augments GH-Induced Spi-GLE-Luciferase Transactivation A, As detailed in Materials and Methods, 3T3-F442A cells were transfected with Spi-GLE-luc reporter plasmid. Serumstarved cells were then stimulated with vehicle (control), GH (100 ng/ml), IGF-I (20 ng/ml), or GH (100 ng/ml) plus IGF-I (20 ng/ml) in triplicate for 16 h, after which luciferase activity was measured. The data shown in panel A are from a representative experiment, in which the basal luciferase activity (treated with vehicle) was considered as 1 and the mean SE of triplicates is indicated for each condition. B, Pooled data from three independent experiments such as those in A are plotted, in which GH-induced transactivation within each experiment was considered as 100%. IGF-I cotreatment augmented GH-induced Spi-GLE-luc transactivation by % (P 0.05). either 20 ng/ml or 500 ng/ml (data not shown). This synergistic effect of the addition of IGF-I to GH is reminiscent of that reported previously for c-fos gene induction in these cells (50). We also examined acute aspects of signaling by assessing the activation of ERK1 and ERK2 (Fig. 10, A and B). In these experiments, 3T3-L1 cells were treated with vehicle, GH, IGF-I, or GH plus IGF-I, and ERK activation was monitored by immunoblotting with an antibody that recognizes the activated (phosphorylated) form of ERKs (anti-active ERK). GH and IGF-I each individually caused acute activation of ERK phosphorylation (Fig. 10A, lanes 2 and 3 vs. 1), as expected. Notably, treatment with GH plus IGF-I caused a marked increase in active ERK signal (lane 4 vs. 2 and 3). The degree of ERK activation relative to ERK abundance induced by each treatment regimen was estimated densitometrically from three such experiments. In the analysis displayed in Fig. 10B, we sought to determine whether the augmented signal

9 Huang et al. Interaction of GH and IGF-I Signaling Elements Mol Endocrinol, June 2004, 18(6): Fig. 10. GH and IGF-I Cotreatment Augments the Activation of ERK 1/2 A, Serum-starved 3T3-L1 cells were treated with vehicle (lane 1), GH (500 ng/ml; lane 2), IGF-I (20 ng/ml; lane 3), or GH (500 ng/ml) plus IGF-I (20 ng/ml) (lane 4) for 5 min. Total cell lysates (extracted in the presence of 1% SDS, 15 g) were resolved by SDS-PAGE and immunoblotted with antiactive ERK1/2 (upper panel) and anti-erk1/2 (lower panel). B, Data shown in panel A, along with those obtained from two other experiments, were subjected to densitometric analysis. Comparison of the level of ERK1/2 activation induced by GH and IGF-I cotreatment (referred to as GH/IGF-I) with the sum of those induced by GH alone plus IGF-I alone (referred to as GH alone IGF-I alone) is shown, in which the sum of ERK1/2 activation induced by GH alone IGF-I alone within each experiment was considered as 100%. GH and IGF-I cotreatment augmented the ERK1/2 activation by % (P 0.01), as indicated. induced by the cotreatment reflected simply the summation of GH-induced plus IGF-I-induced ERK activation or instead reflected synergistic ERK activation (more than the summation). With the sum of the levels of activation induced by GH alone plus IGF-I alone set to 100% (light bar), the observed ERK activation induced by the combination of GH plus IGF-I (dark bar) was % (P 0.01), indicating synergy of GH plus IGF-I. To examine the mechanism of these synergistic effects on signaling, we treated 3T3-L1 cells with each hormone alone or in combination and assessed the degree of GHR tyrosine phosphorylation by anti-gh- R cyt-al47 precipitation and anti-ptyr blotting (Fig. 11A). GH alone induced the characteristic acute GHR tyrosine phosphorylation (lane 2 vs. 1). IGF-I failed to induce GHR tyrosine phosphorylation (lane 3 vs. 1), as expected. Notably, cotreatment with GH plus IGF-I resulted in augmented GHR tyrosine phosphorylation compared with GH alone (lane 4 vs. 2) (2.3-fold increase; n 3; P 0.056). We further examined the effect of cotreatment on GH-induced GHR disulfide linkage. We previously demonstrated in multiple cell types that GH induces the formation of disulfide-linked GHRs that can be detected as a high-m r receptor form by anti-ghr immunoblotting of proteins electrophoretically resolved under nonreducing conditions (35, 46, 56, 57). GHR disulfide linkage reflects GH-induced conformational change in the receptor that allows its activation (35). 3T3-L1 cells were treated with vehicle, GH, IGF-I, or the combination of GH plus IGF-I for 5 and 30 min, and detergent-extracted proteins were resolved under nonreducing conditions and anti-ghr cyt-al37 immunoblotted. As expected, GH treatment caused the appearance of the disulfide-linked GHR (Fig. 11B, upper panel; lanes 2 and 5 vs. 1), but IGF-I itself did not do so (lanes 3 and 6 vs. 1). Interestingly, cotreatment with GH plus IGF-I substantially increased the abundance of the disulfide-linked GHR in comparison with treatment with GH alone (lanes 4 vs. 2 and 7 vs. 5). Further, reprobing of this blot with anti-ptyr confirmed that the disulfide-linked GHR appearing in response to GH is tyrosine phosphorylated (Fig. 11B, lower panel; lanes 2 and 5 vs. 1) and that cotreatment with GH plus IGF-I enhanced the abundance of the tyrosine-phosphorylated disulfide-linked receptor form (lanes 4 vs. 2 and 7 vs. 5). This enhancement of tyrosine phosphorylation of the disulfide-linked GHR form was estimated densitometrically, and the pooled data from three independent experiments are graphically displayed in Fig. 11C, in which the level of GH-induced disulfide-linked GHR tyrosine phosphorylation was considered as 100% for each time point. Treatment for 5 min with GH plus IGF-I induced 4-fold greater abundance of tyrosine-phosphorylated GHR than did treatment with GH alone for 5 min (P 0.01), and treatment for 30 min with GH plus IGF-I enhanced the phosphorylation by 1.8-fold over GH alone (P 0.05). We conclude that cotreatment with IGF-I enhances the ability of GH to promote attainment of a GHR conformation that is competent for JAK2 activation and initiation of downstream signaling. DISCUSSION With increasing knowledge of the complexity of signaling pathways and the utilization of common pathways by diverse receptors, cross-talk between receptor signaling systems has emerged as an important and interesting area of signaling research. Our previous work has focused on interactions between GHinduced signaling and signaling mediated by intrinsic tyrosine kinase growth factor receptors, most notably members of the EGF receptor family. We previously showed that GH treatment of murine preadipocytes resulted in phosphorylation of both the EGFR (ErbB-1)

10 1480 Mol Endocrinol, June 2004, 18(6): Huang et al. Interaction of GH and IGF-I Signaling Elements Fig. 11. GH and IGF-I Cotreatment Augments GH-Induced Disulfide Linkage and Tyrosine Phosphorylation of GHR A, Serum-starved 3T3-L1 cells were stimulated with vehicle (lane 1), GH (500 ng/ml; lane 2), IGF-I (20 ng/ml; lane 3), or GH (500 ng/ml) plus IGF-I (20 ng/ml) (lane 4) for 15 min. Detergent extracts (300 g) were subjected to immunoprecipitation with anti-ghr cyt-al47 and eluted proteins were analyzed by immunoblotting with anti-ptyr (upper panel) and anti-ghr cyt-al47 (lower panel). The experiment shown is representative of three such experiments. B, Serum-starved 3T3-L1 cells were stimulated with vehicle (lane 1), GH (100 ng/ml; lanes 2 and 5), IGF-I (20 ng/ml; lanes 3 and 6), or GH (100 ng/ml) plus IGF-I (20 ng/ml) (lanes 4 and 7) for 5 and 30 min, respectively. Detergent extracts (100 g) were resolved without immnoprecipitation by SDS-PAGE under nonreducing conditions and immunoblotted with anti-ghr cyt-al37 (upper panel) and anti-ptyr (lower panel). The position of disulfide-linked murine GHR (dsl-ghr) is indicated. C, Enhancement of tyrosine phosphorylation of disulfide-linked GHR. Data from three independent experiments such as those shown in B (lower panel) were subjected to densitometric analysis. In each experiment, the level of GH-induced and ErbB-2 (also known as c-neu) at a residue(s) recognized by a state-specific antibody that specifically reacts with proteins at ERK consensus phosphorylation sites (11, 12). These GH-induced phosphorylation events were dependent on MAPK kinase 1 (MEK1) activity and were associated with alterations in EGFR and ErbB-2 signaling and/or trafficking. In the current study, we examined potential crosstalk between GH and IGF-I. We detected the GHinduced formation of a protein complex including the GHR, JAK2, and IGF-IR. GH concentration dependence and GH antagonist experiments suggested that its ability to induce formation of this complex was related to induction by GH of conformational change(s) in its receptor and/or JAK2. The use of the kinase inhibitor staurosporine to inhibit GH-induced GHR tyrosine phosphorylation, although not preventing GHinduced GHR conformational change (35, 46), further indicated that receptor tyrosine phosphorylation was likely not required for the interaction. Likewise, the inability of LIF to promote JAK2-IGF-IR association, despite causing ample JAK2 tyrosine phosphorylation, indicated that JAK2 tyrosine phosphorylation per se is unlikely to be required to support the GH-induced GHR-JAK2-IGF-IR complex. Finally, deglycosylation of the proteins within the complex allowed us to determine that the IGF-IR -chain associated with the GHR and JAK2 was also not detectably tyrosine phosphorylated. Thus, although we cannot completely rule out a requirement for tyrosine phosphorylation of any of the members of the complex for their association, our results do not favor such a requirement. In that respect, our findings differ from those of Gual et al. (28), which suggested that associations between the IGF-IR or insulin receptor and JAKs required tyrosine phosphorylation of the receptors. The associations in that study were dependent on in vitro overexpression of both binding partners and did not include analysis of cytokine receptors with which the JAK can associate in a normal cellular context. Although other studies have indicated that stimulation of insulin or IGF-I receptors can activate JAKs (24 27), the current report is the first of which we are aware that has studied interaction of cytokine receptors/jaks with the IGF-IR in cells that do not overexpress any of the components. As such, we find relevant the results of our experiments aimed at probing the specificity of the interactions. In contrast to the GH-induced association of GHR/JAK2 with the IGF-IR, we have observed no such association with the EGFR in the same cellular setting (Fig. dsl-ghr tyrosine phosphorylation was considered as 100% for each time point. The data are plotted as a percentage of tyrosine phosphorylation level of dsl-ghr relative to that induced by GH alone and are mean SE of three such experiments. The P values for each time point are indicated.

11 Huang et al. Interaction of GH and IGF-I Signaling Elements Mol Endocrinol, June 2004, 18(6): and our unpublished observations). GH-induced GHR-EGFR association has been detected by others (47), but again only in the setting of transient overexpression of both proteins. Further studies will be required to understand whether any structural similarities underlie the abilities of GH to promote association of its receptor with the IGF-IR and to engage in cross-talk with the EGFR signaling system. Inherent in such studies will be mapping of the GHR and/or JAK2 region(s) required for each effect. We note, however, that in contrast to the requirement for GH-induced MEK/ERK signaling for cross-talk with the EGFR-family systems (11, 12), GH-induced formation of the GHR-JAK2-IGF-IR complex was not affected by inhibition of the GHinduced MEK/ERK or PI3K pathways. We probed the potential signaling impact of formation of the complex in response to GH by examining the effects of cotreatment with GH plus IGF-I. Notably, we found that acute induction by GH of GHR tyrosine phosphorylation was augmented when cells were coincubated with IGF-I. Importantly, IGF-I alone had no apparent effect on GHR tyrosine phosphorylation. Further, the GH-induced formation and tyrosine phosphorylation of disulfidelinked GHRs, in particular, were enhanced by cotreatment with IGF-I. These are important observations, in that the disulfide linkage of GHRs reflects for receptor dimers the attainment of conformational competence for signaling (35). We are intrigued by the mechanistic implications of these findings. Existing data suggest that GH engagement of the GHR causes changes in receptor dimers that allow achievement of an active signaling conformation. Our data suggest that this also allows association of GHR/JAK2 with the IGF-IR, independent of the activation status of IGF-IR. We interpret our findings to further indicate that IGF-I engagement of the IGF-IR in such a complex facilitates or stabilizes the GH-activated dimeric GHR signaling conformation (reflected by GHR disulfide linkage). We do not yet know whether this enhancement of the GHR conformational change induced by IGF-IR engagement depends on IGF-IR tyrosine kinase activation or, rather, kinaseindependent changes in the IGF-IR itself induced by IGF-I binding. Regarding the enhancement of GHR tyrosine phosphorylation observed in the cotreatment situation, several mechanistic possibilities, in addition to solely achievement or stability of the activated conformation, can be considered. The GHR, by virtue of either direct or indirect interactions with the IGF-IR, could, in response to IGF-I, become a better substrate for tyrosine phosphorylation by the GH-activated JAK2 kinase. Alternatively, JAK2, by being in close proximity to the IGF-I-activated IGF-IR kinase, could be rendered a more active kinase and thus more robustly phosphorylate the GHR. Yet another possibility is that the IGF-I-activated IGF-IR kinase, upon being brought into proximity of the GHR in response to GH, itself acts as a GHR kinase, further increasing the abundance of tyrosine phosphorylated GHR. At present, our data do not allow us to discriminate between these or other possible mechanisms. In preliminary experiments, we have yet to detect enhanced JAK2 or IGF-IR kinase activities in the GH plus IGF-I-cotreated vs. individually treated situations; similarly, we have yet to observe synergy in IGF-I-induced Akt activation (our unpublished observations). However, we realize that such changes could be subtle and still produce changes in GHR tyrosine phosphorylation. Conceivably, induction by GH of the GHR-JAK2-IGF-IR complex allows multiple mechanisms to collaborate to produce signaling synergy. A more thorough understanding of the mechanistic details of this interaction will await a better understanding of the structural determinants necessary for GHR-JAK2-IGF-IR complex formation. In particular, reconstitution experiments using kinase-inactive molecules (JAK2 and IGF-IR) and experiments in which JAK2 levels are altered may be useful in discriminating among these possibilities. Independent of the mechanisms involved, we also observed synergy in acute activation of the ERK pathway for the combination of GH and IGF-I vs. each hormone individually in our cell system. Synergy between GH and IGF-I for signaling responses has also been observed by others. Ashcom et al. (50), examined the abilities of GH and IGF-I to induce c-fos gene expression in 3T3-F442A cells. They observed that treatment of serum-starved cells with GH plus IGF-I induced c-fos mrna levels nearly 30% greater than the summed response of both individually. Further, run-on transcription experiments suggested nearly 2-fold synergy of c-fos transcription by the combination of hormones compared with the sum of each. Our findings of synergy in ERK activation are particularly noteworthy in light of the established influence of the MEK/ERK pathway on GH-induced c-fos gene activation in preadipocytes (10). Further, we also observe enhanced transactivation of a reporter driven by enhancer elements of Spi 2.1, another GH-regulated gene, suggesting that the enhancement of GH signaling afforded by IGF-I costimulation may affect various aspects of GH action. Indeed, Edmondson et al. (51) examined the influence of GH and IGF-I on the growth of human melanocytes in vitro. Despite expressing the GHR (58), these cells were unable to grow in response to GH alone. IGF-I or des(1 3)IGF-I (an IGF-I analog with reduced affinity for IGF-binding proteins) alone each caused increased cell numbers, but the addition of GH to IGF-I or des(1 3)IGF-I caused an increase of nearly 1.5-fold or 2-fold, respectively, in the response. The mechanisms responsible for these examples of synergy in c-fos gene induction in preadipocytes and growth responsiveness of melanocytes have not been uncovered and may be quite complicated. However, in neither case is it likely that GH induction of IGF-I gene expression is at the