JAK2, Ras, and Raf are required for activation of extracellular signal-regulated kinase/mitogen-activated protein kinase by growth hormone.

Protein-tyrosine kinases (PTKs) of the JAK family have been characterized on the basis of their ability to mediate the rapid induction of transcription of interferon-responsive genes through the stimulation of a class of latent cytoplasmic transcription factors known as signal transducers and activators of transcription (STATs). STAT activation, which has been described as being Ras-independent, requires tyrosine phosphorylation, but STAT transactivating activity is enhanced by phosphorylation on serine as well, probably by extracellular signal-regulated kinase/mitogen-activated protein kinase(s) (ERK/MAPK). STATs can be activated upon binding of ligands to receptor PTKs, to G-protein-linked receptors, and to cytokine receptors. Whether JAKs are required for the activation of signaling pathways other than that leading to STAT activation is not known. The binding of growth hormone (GH) to its receptor (GHR) activates JAK2 and STATs as well as ERK/MAP kinases. We have used a transient transfection system in 293 cells to evaluate the requirement for JAK2 in the activation of ERK2/MAPK by GH. We found that JAK2 is required for GH-simulated activation of ERK2/MAPK. Employing the transient expression of dominant negative forms of H-Ras and Raf-1, we determined that the GHR/JAK2-mediated activation of ERK2/MAPK is dependent on both Ras and Raf. Thus, JAK protein-tyrosine kinases may represent a common component in the activation of the ERK2/MAPK and STAT signaling pathways, which appear to bifurcate upstream of Ras activation but converge with ERK/MAPK phosphorylation of STATs.

kinases in cell signaling was first defined by the observation that the expression of receptor-specific JAK proteins is required for the rapid induction of interferon (IFN)-responsive genes. This activity is mediated by the activation of STAT proteins that promote transcription via binding to interferonspecific response elements (1). In addition to mediating transcription of interferon-responsive genes, JAK/STAT signaling pathways (2) are now thought to mediate similar responses to angiotensin II, which binds to a G-protein-coupled receptor (3), and to several cytokines, including growth hormone (GH), prolactin, erythropoietin (EPO), leukemia inhibitory factor, ciliary neurotrophic factor, granulocyte-CSF and granulocyte-macrophage-CSF, oncostatin M, and several interleukins (4,5). The persistence of the IL-6-induced DNA binding activity of STAT3 (6) and the PDGF-stimulated DNA binding activity of STAT1␣ (7) in nuclear extracts of cells expressing dominant negative Ras suggests that at least some aspects of STAT activation are Ras-independent.
The binding of GH to its receptor (GHR) induces receptor homodimerization (8,9) and stimulates the tyrosine phosphorylation of JAK2 (10) and the DNA binding activity of STATs 1 (11), 3 (12), and 5 (13). Like other cytokines, GH can also activate ERK/MAP kinases (MAPK) (14 -16). The activation of the ERK/MAPK pathway, which in most systems requires Ras and Raf, is a universal downstream response to the activation of most receptor PTKs. It is frequently the case that ligandstimulated receptor autophosphorylation results in the creation of receptor phosphotyrosine docking sites for the binding of the SH2 domain-containing proteins Grb2 or Shc, which can bind Grb2, bound to effectors of Ras activation. By analogy, then, it has been presumed that cytokine activation of the ERK2/MAPK pathway would first require the recruitment by ligand activated cytokine receptors of one or more PTKs in order to mimic, at least qualitatively, this role of receptor PTKs.
The maximal activation of STAT1␣and STAT3-mediated transcription requires phosphorylation on both tyrosine and serine residues (2). That JAK/STAT pathways might be regulated in part by ERK2/MAPK is suggested by the observations that the activating serine phosphorylation site in STATs 1 and 3 is in an ERK2/MAPK phosphorylation consensus sequence, that IFN␤-activated STAT1␣ coimmunoprecipitates with ERK2/MAPK, and that the kinase activity of ERK/MAPK is required for IFN␤-stimulated gene transcription (18). However, the issue of whether JAKs are involved in the regulation of the ERK2/MAPK pathway remains unresolved (19). To test the possibility that JAK proteins might be required for the activation of the ERK/MAPK pathway by GH, we transiently expressed the GHR with or without JAK1, JAK2, or a kinaseinactive form of JAK2 in E1A-immortalized human embryonic kidney (293) cells, and the effect on the activity of an epitopetagged form of ERK2/MAPK was evaluated. Our results indi-cate that JAK2 is required for the activation of ERK2/MAPK by GH. We evaluated the involvement of Ras and Raf in JAK2mediated activation of ERK2/MAPK by GH by transiently coexpressing dominant negative forms of either H-Ras or Raf-1 with the GHR and JAK2 and found that the GHR/JAK2-mediated activation of ERK2/MAPK was inhibited both by dominant negative Ras and dominant negative Raf.

EXPERIMENTAL PROCEDURES
Cell Culture, Antibodies, and Reagents-Human embryonic kidney (293) cells were maintained in DMEM containing 10% FBS. Human pituitary growth hormone, which was used at 20 nM in all experiments, was generously provided by Dr. Jack Kostyo. A monoclonal antibody (12CA5) to the hemagglutinin epitope tag of ERK2/MAPK (20) was used both for detection of HA-ERK2/MAPK by immunoblotting and for immunoprecipitation of HA-ERK2/MAPK for in vitro kinase assays. The expression of the transfected GHR was confirmed by immunoblotting with a rabbit polyclonal antibody to the cytoplasmic domain of the GHR fused to GST. The increase in JAK2 levels following transfection was evaluated by immunoprecipitation/immunoblotting with a polyclonal antibody to a C-terminal peptide of JAK2 (Santa Cruz).
Expression Vectors and Transfection-The full-length murine GHR cDNA, provided in a pBluescript vector by Dr. Frank Talamantes (21), was subcloned into a pSLX-CMV mammalian expression vector. The full-length wild-type murine JAK2 and JAK1 cDNAs, provided by Dr. Rikiro Fukunaga, or a kinase-inactive form of the murine JAK2 cDNA, provided by Dr. Don Wojchowski (22), had been subcloned into pEF-Bos expression vectors (23). The HA-tagged p42 ERK2/MAPK cDNA, provided by Dr. Michael Weber, was in a pLNCAL7 vector (20). The dominant negative (S17N) H-Ras cDNA in pSR␣, the dominant negative (K375W) Raf-1 and wild-type Raf-1 cDNAs in pRSV, and the cDNA for the constitutively active MEKK1 in pCMV5 (24) were provided by Dr. Michael Karin. For experiments, cells were cultured in 6-cm dishes precoated with poly-L-lysine and, when subconfluent, were transfected with Lipofectamine, as per the manufacturer's instructions (Life Technologies, Inc.), and harvested 36 h later following 16 h of incubation in serum-free medium.
For immunoprecipitation/kinase assays, cells were lysed in ice-cold RIPA, and lysates were clarified by centrifugation at ϳ25,000 ϫ g at 4°C. Supernatant protein levels were assayed, and 100-g aliquots were incubated with ϳ2 g of 12CA5 monoclonal antibody for 60 min at 4°C. Antibody-antigen complexes were captured on protein A-bound beads (Repligen) during a 30-min incubation at 4°C. Beads were washed three times with lysis buffer and two times with kinase reaction buffer (20 mM Hepes, pH 7.4, 5 mM MgCl 2 , 1 mM dithiothreitol). Washed beads were incubated for 10 min at 30°C in 20 l of kinase reaction buffer containing 20 M ATP, 10 Ci of [␥-32 P]ATP (3000 Ci/mmol; Amersham), and 10 g of myelin basic protein (Sigma), and reactions were terminated by the addition of 2 ϫ sample buffer and boiling. Kinase reaction proteins were resolved by SDS-PAGE, and the phosphorylation of myelin basic protein was confirmed by phosphorimaging (Molecular Dynamics) and quantitated by Cerenkov counting.

RESULTS AND DISCUSSION
To evaluate the role of the PTK JAK2 in GH-stimulated MAPK activity, we expressed the full-length mGHR and HA-tagged p42 MAPK (ERK2) in human embryonic kidney (293) cells, in the presence or absence of coexpressed JAK proteins. JAK2 is ubiquitously expressed; however, the level of expression in 293 cells appears to be sufficiently low that some level of overexpression of JAK2 is required in order to detect activation (25). We found the level of immunoprecipitable JAK2 in GHR/JAK2 transfected cells to be approximately 10-fold higher than that in untransfected cells (data not shown). As shown in Fig. 1A, expression of the GHR alone failed to support GHstimulated activation of ERK2/MAPK, whereas coexpression of the GHR and wild-type JAK2 resulted in a GH-stimulated decrease in the mobility of a fraction of the expressed ERK2/ MAPK, which is known to reflect the dual phosphorylation and activation of the enzyme. We confirmed that GH stimulated ERK2/MAPK activity by assaying HA-ERK2/MAPK immunoprecipitates from cells coexpressing the GHR and different JAKs for myelin basic protein (MBP) phosphorylating activity. The observed GH-stimulated increase in activity in GHR/ JAK2-expressing cells was approximately 4-fold (Fig. 1B). Wild-type JAK2 expressed in the absence of the GHR did not activate this pathway (data not shown). JAK2 kinase activity was required for the activation of ERK2/MAPK, since activation failed to occur in GH-treated cells in which GHR was coexpressed with a mutant form of JAK2 (JAK2ki) rendered kinase-inactive by two point mutations in the PTK subdomain VIII (22) (Fig. 1, A and B). This is consistent with the recent report by Frank et al. (26) that a form of JAK2 truncated in the functional C-terminal PTK domain can associate with the GHR but not support a GH-stimulated mobility shift in ERK2/MAPK.

JAK2/Ras/Raf-dependent Activation of ERK2/MAP Kinase by GH 30838
A significant role for JAK1 in GH-stimulated ERK2/MAPK activity seems unlikely from the observations that GH stimulates the tyrosine-phosphorylation of JAK2 but not JAK1 in IM-9 cells (27) and that in HeLa cells transiently expressing receptor chimerae containing cytoplasmic domains of the GHR, EPOR, or gp130, JAK1 failed to coprecipitate with either the GHR or EPOR chimera (28). We found that coexpression of JAK1 with the GHR in 293 cells resulted in only a weak GH-stimulated mobility shift in ERK2/MAPK on immunoblotting (Fig. 1A). In cells coexpressing the GHR and JAK1 the stimulation by GH was less than 2-fold, and the level of GHstimulated ERK2/MAPK kinase activity was less than 30% of that stimulated by GH in GHR/JAK2-transfected cells (Fig.  1B). We tested whether the GH-stimulated ERK/MAPK activation mediated by JAK2 required Ras by coexpressing the GHR and JAK2 in 293 cells with or without coexpression of a dominant negative form of H-Ras (S17N; Ref. 29). As shown in Fig. 2, A and B, GH stimulation of ERK2/MAPK activity was inhibited by dominant negative Ras (dn Ras). That this was not due to a nonspecific effect of dominant negative Ras on ERK2/ MAPK activation is evident from the observation that ERK2/ MAPK activity was not inhibited in 293 cells transiently expressing constitutively active MEKK1 (⌬MEKK, Fig. 2A), a specific activator of Jun kinases (JNK) that, at higher concentrations, can also activate ERK2/MAPK (24). The multiple mobility shifted forms of HA-ERK2/MAPK in ⌬MEKK-expressing cells may reflect phosphorylation of sites in ERK2/MAPK in addition to Thr-183 and Tyr-185. In contrast to the inhibition of GH-stimulated ERK2/MAPK activity by dominant negative Ras, activation by ⌬MEKK appeared to be enhanced in cells expressing dominant negative Ras by a presently unknown mechanism.
Certain cytokine receptors are tyrosine-phosphorylated when coexpressed with JAK PTKs in insect cells, suggesting that JAK-mediated phosphorylation could provide the binding sites for Ras effectors in cytokine receptors lacking intrinsic PTK activity. In the case of the GHR, however, Wang et al. (32) found several cellular responses to GH to be preserved in mutated receptors lacking phosphorylatable tyrosine residues.
Moreover, ERK2/MAPK can be activated in cells expressing a truncated rat GHR in which only four of the 10 cytoplasmic domain tyrosine residues are retained (33), and MAP kinase activation is not inhibited by mutation of two of these tyrosine residues to phenylalanine (34). Thus, the GHR itself may not be the binding site for Ras-activating effectors.
Alternative mechanisms for Ras activation are suggested by the findings that GH stimulates the tyrosine phosphorylation of IRS-1 (35) and of Shc proteins (36). GH also promotes an increase in the association of Grb2 with Shc and the association of JAK2 with a GST-Shc SH2 domain fusion protein (36). Although in vivo association of Shc with JAK2 was not detected, the latter finding suggested the possibility that JAK2 itself might provide the phosphotyrosine binding site(s) for effector molecules on the ERK/MAPK signaling pathway (see also below). Indeed, Shc has been shown to associate in vivo with JAK2 in cells expressing a truncated EPOR lacking tyrosine residues (38).
To test the requirement for Raf in GH activation of ERK2/ MAPK, the GHR and JAK2 were coexpressed in 293 cells with or without coexpression of a dominant negative form of Raf-1, which harbors a point mutation (K375W) in the ATP binding site. As with dominant negative Ras, GH stimulation of ERK2/ MAPK activity was inhibited by coexpression with dominant negative Raf (dn Raf; Fig. 3, A-C). Although we found TPA to be a relatively weak and finicky activator of ERK2/MAPK activity in the 293 cell line used in these studies, the almost complete inhibition of this activity by dominant negative Raf (Fig. 3A) is nevertheless consistent with the report of its Raf JAK2/Ras/Raf-dependent Activation of ERK2/MAP Kinase by GH 30839 dependence in 293 cells (39). Activation of ERK2/MAPK by ⌬MEKK, on the other hand, was not inhibited by dominant negative Raf (Fig. 3A). Further support for a role of Raf in GH-stimulated ERK2/MAPK activity is suggested by the observation that overexpression of wild-type Raf-1, which when expressed alone did not activate ERK2/MAPK, could reproducibly increase ERK2/MAPK activation when coexpressed with the GHR and JAK2 (Fig. 3C), although the magnitude of this effect varied between experiments (Fig. 3B).
Of relevance to the JAK2-mediated activation of Raf is the recent finding that Raf-1 physically associates with JAK2 in insect cells in which Raf-1 and JAK2 are coexpressed, in EPOtreated CTLL cells expressing the EPOR and in IFN-treated fibroblasts and HeLa cells. 2 Raf-1 activation appears to require multiple inputs, including but not necessarily limited to recruitment to the plasma membrane by activated Ras (40) and phosphorylation on both serine (41) and tyrosine residues (40,41). Raf-1 is tyrosine-phosphorylated in NIH 3T3 cells overexpressing v-Src (41). However, it remains to be determined whether c-Src is a physiological Raf-1 kinase and, if so, whether the phosphorylation of Raf-1 by c-Src is direct or whether other kinases, such as JAKs, might phosphorylate Raf-1 in vivo.
In summary, we have found that GH-stimulated activation of ERK2/MAPK in 293 cells requires JAK2, Ras, and Raf. There is growing evidence that ERK/MAP kinases play an essential role in promoting the growth or differentiation of cells. Studies in which the number of cellular PTK receptors has been manipulated by overexpression have demonstrated that increased receptor number facilitates sustained ligand-dependent activation and enhanced nuclear translocation of ERK/MAP kinases, which correlate with cellular differentiation (42), and expression of the constitutively active MAPK kinase, MEK, promotes the transformation of fibroblasts (41)(42)(43) and the differentiation of PC-12 cells (44). GH has been shown to promote both the differentiation of 3T3-preadipocytes (45) and the proliferation of IM-9 lymphocytes (46). However, the specific role of ERK/ MAP kinases in these processes remains to be identified.
Although STAT activation has been described as being Rasindependent, the observation that STAT-mediated transcription is enhanced by serine phosphorylation, probably by ERK/ MAP kinase(s), implies that STAT activity may at the very least be quantitatively regulated by both Ras-independent and -dependent pathways. Given the requirement for JAK proteins in both STAT and ERK/MAP kinase activation, JAK proteins likely act as common upstream activators of these two signaling pathways that bifurcate upstream of Ras but converge with STAT phosphorylation by ERK/MAPK. Although STAT proteins have not yet been shown to in turn influence the ERK/ MAP kinase pathway, an intriguing possibility is that STATs, which can bind the sis-inducible factor response element (SIE), and ERK/MAP kinase-activated ternery complex transcription factors, which bind the serum response element (SRE), might cooperate to enhance the transcription of single genes regulated by both the SIE and SRE, akin to that which has been shown to occur for PDGF-induced transcription of the c-fos gene (47).