Gp130-mediated signal transduction in embryonic stem cells involves activation of Jak and Ras/mitogen-activated protein kinase pathways.

The leukemia inhibitory factor/interleukin 6 (LIF/IL6) family of cytokines promotes cell type-specific pleiotropic effects by engaging multimeric receptor complexes that share the common affinity converter/signal transducing subunit gp130. While the maintenance of embryonic stem (ES) cell self-renewal is an activity unique to this family of cytokines, the intracellular signaling events mediated by gp130 remain largely unknown. Here we show a rapid and transient increase in the specific activity of the Src-related kinase Hck as well as of the Janus kinases Jak1, Jak2, and Tyk2 following treatment of ES cells with LIF or a combination of IL6 plus a soluble form of the IL6 receptor. Within 2 min of stimulation, we also observed increased tyrosine phosphorylation of SHC, activation of the guanidine nucleotide exchange activity on p21ras, and an electrophoretic mobility shift of MAP kinase. Functional involvement of Hck and p21ras activation in gp130-mediated signaling is supported by the finding that the introduction of constitutively activated Hck or v-Ha-ras partially alleviates the requirement of ES cells for LIF to remain undifferentiated. In contrast, suppression of Jak1 in ES cells by antisense technology increased the amount of LIF required to retain their pluripotentiality. These results are consistent with the notion that gp130-mediated suppression of ES cell differentiation depends on signaling through at least two cascades, namely a p21ras-dependent pathway that possibly involves Hck, as well as a Jak kinase-dependent pathway.

The leukemia inhibitory factor (LIF) 1 /interleukin-6 (IL6) family of cytokines elicits a wide variety of cellular responses including survival, proliferation, differentiation, and the regulation of end-cell function. Mice that contain targeted disruptions of genes encoding members of the LIF/IL6 family of cytokines, or the corresponding ligand binding chains, have confirmed the importance of these cytokines during fetal development and in the adult (1)(2)(3)(4). In particular, LIF has been implicated in early embryogenesis (5,6) by promoting the sur-vival and proliferation of primordial germ cells in vitro (7) and maintaining the pluripotency of embryonic stem (ES) cells.
Cytokines like LIF and IL6 exert their biological effects through the formation and activation of multimeric hemopoietin receptors. These complexes frequently comprise a ligand-specific binding subunit and an affinity converting/signal transducing subunit common to a particular family of cytokine receptors. The competition of cytokines of the LIF/IL6 family for the shared and ubiquitously expressed signal transducing subunit gp130 defines a group of cytokines with overlapping functions. These include LIF, IL6, IL11, oncostatin M (OncM), ciliary neurotrophic factor (CNTF) (8), and cardiotrophin CT-1 (9). While the cytoplasmic portion of the CNTF-and IL6-ligand binding chains (IL6R) is not required for the generation of intracellular signals (10), the intracellular portion of the LIF-ligand binding chain (LIFR) has been implicated in signal transduction in some cell types (11). By analogy to tyrosine kinase receptors, the current model proposes that each receptor complex must contain two membrane-spanning subunits. This may be fulfilled either by a LIFR/gp130 heterodimer (in the case of LIF, OncM, and CNTF) or a gp130 homodimer (in the case of IL6) (12).
Despite the fact that hemopoietic receptors lack conspicuous catalytic subdomains, the appearance of tyrosine-phosphorylated proteins is among the earliest events following ligandinduced receptor oligomerization. For the LIF/IL6 family of cytokines, it has recently been shown that these effects can be accounted for by the association of the gp130 receptor subunit with members of the cytoplasmic tyrosine kinases of the Src (10) and Janus kinase (Jak) families (13). Indeed, the introduction of a constitutively active version of the Src-related kinase Hck (hckY499F) (10) partially abrogates the requirement of LIF to suppress differentiation of ES cells in vitro. Since tyrosine phosphorylation appears to be an obligatory early event in gp130-mediated signaling, additional cytoplasmic tyrosine kinases must be activated in ES cells in response to LIF either upstream of Hck or as part of Hck-independent signaling cascades. The Jak family of kinases represent likely candidates for the complementation of gp130-dependent signaling in ES cells, since at least 3 family members can be activated in response to LIF and IL6 in a variety of cellular systems (13). However, it appears that activation of individual Jak kinases may well be cell type-specific (13) with an apparent restriction to Jak1 and Jak2 activation in response to LIF in ES cells (14). Whether the differential utilization of Jak kinases (or indeed cytoplasmic kinases in general) may account for the different cell-type specific biological effects exerted by the LIF/IL6 family of cytokines has not yet been established.
For other hemopoietin receptors, such as those for IL2 and G-CSF, it has been shown that different regions of the cytoplasmic domain are required for the transduction of proliferative signals and those related to end-cell function (15,16). Thus, the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 61-3-9347-3155; Fax: 61-3-9347-6957; E-mail: dunn@licre.ludwig.edu.au. 1 The abbreviations used are: LIF, leukemia inhibitory factor; LIFR, leukemia inhibitory factor receptor; CNTF, ciliary neurotrophic factor; ES, embryonic stem; G-CSF, granulocyte colony-stimulating factor; IL6, interleukin 6; sIL6R, soluble interleukin 6 receptor; OncM, oncostatin M; PAGE, polyacrylamide gel electrophoresis; STAT, signal transducer and activator of transcription; FCS, fetal calf serum; INF␣, interferon ␣; PIPES, 1,4-piperazinediethanesulfonic acid. activation of more than one intracellular signaling cascade may also explain the pleiotropic responses observed for the LIF/IL6 family of cytokines. Most studies addressing gp130-mediated intracellular signaling events have been carried out by overexpressing individual receptor chains in heterologous cell systems, including COS and hepatoma cells (11,(17)(18)(19). While such transfection experiments aim at reconstituting the normal regulatory systems in operation, they may often yield artificially exaggerated responses to extracellular signals as a direct consequence of overexpressing experimentally introduced molecules. Indeed, for many hemopoietin receptor signaling pathways it has been shown that the relative concentrations of intermediary signaling molecules are critical in specifying transcriptional regulation of target genes (20,21). To overcome these potential difficulties, we have investigated the impact of LIF/IL-6-dependent activation of membrane-proximal signaling molecules on preventing cellular differentiation in ES cells where the signal transducing molecules are at normal physiological concentrations. We provide insights into the signaling events which permit unrestricted selfrenewal of ES cells in vitro in response to the LIF/IL-6 family of cytokines. Based on "guilt by association and/or activation," we identify Hck, Jak1, and p21 ras as key intracellular signaling molecules. We also present functional evidence for the involvement of these molecules in promoting ES cell self-renewal by enforcing their activation in a LIF/IL6-independent manner. Finally, our data suggest that Hck and Jak kinases activate at least two intracellular signaling cascades which are required for the effective regulation of target genes that suppress differentiation of ES cells.

EXPERIMENTAL PROCEDURES
Plasmids-The expression cassette containing an activated version of v-Ha-ras (G12V,A59T) was constructed by polymerase chain reactionmediated introduction of XbaI linkers into a full-length cDNA (22) and subsequent ligation into the XbaI site of the mammalian expression vector pEF-BOS (23). The inducible Jak1 antisense construct was obtained by cloning a 2.7-kilobase BglII fragment, derived from pCDHJ5 and encompassing the site of translational initiation of a human Jak1 cDNA (24) in the antisense orientation into the BglII site immediately downstream of the human 6 -16 promoter in the expression plasmid p1OX (25). Expression construct were co-transfected with the resistance marker plasmids PGKneo (26) or PGKhprt (10).
Cell Lines-All experiments were carried out in the ES cell line E14TG2a which is defective in the hypoxanthine phosphoribosyltransferase (hprt) gene (27). ES cells were routinely subcultured in ES cell medium (Dulbecco's modified Eagle's medium containing 15% fetal calf serum, 0.1 mM 2-mercaptoethanol, and supplemented with nonessential amino acids and nucleotides) plus 500 pM (1000 units/ml) LIF (Esgro from AMRAD Operations Pty Ltd.) in the absence of feeder cell lines.
Stable transfected ES cell lines were obtained by co-electroporating 15 g of the linearized expression plasmids together with 1.5 g of the selectable marker plasmid PGKneo or PGKhprt (500 microfarads; 270 V, Bio-Rad Gene Pulser) into 1.5 ϫ 10 7 ES cells. The cells were plated into five 100-mm diameter culture dishes (Nunc) and selected for 7 days in ES cell medium containing G418 (Geneticin, Life Technologies, Inc.; 175 g/ml) or HAT (hypoxanthine, 120 M; aminopterine, 0.4 M; thymidine, 20 M) beginning 24 h after the electroporation. Individual resistant colonies were expanded, and the integration of the expression construct was verified by Southern blotting.
Cell Culture Assays-The extent to which gp130-mediated signaling prevents ES cell differentiation was determined by plating cells at a density of 500 cells/cm 2 in gelatin-coated 6-well multiwell plates (Nunc) in ES cell medium containing the indicated concentration of LIF. Five days after inoculation, the proportion of undifferentiated cells was determined by scoring ES cell colonies that consisted entirely of densely packed cells ("undifferentiated") and colonies consisting of a mixture of densely packed cells and cells of flattened morphology or entirely of flattened morphology ("differentiated") (10). The proportion of undifferentiated colonies was calculated after scoring the morphology of 300 randomly chosen colonies in triplicate culture dishes.
In Vitro Kinase Assay and Immunoblotting-For each time point, approximately 4 ϫ 10 7 cells were starved in serum-free ES cell medium in the absence of LIF for 12 h. The cultures were then stimulated with 500 pM LIF or with 64 nM human sIL6R (encompassing amino acids 1-344 of the extracellular domain (28)) plus 36 nM IL6 (29) for the indicated period of time. After harvesting in ice-cold PBS, cells were lysed in 500 l of KALB buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris, pH 7.5, 1 mM EDTA; 1 mM phenylmethylsulfonyl fluoride, 1% Trasylol, 0.1 mM Na 3 VO 4 ) for 30 min on ice before removing the nuclei by centrifugation. The protein concentration of the supernatant was determined using a BCA kit (Pierce), and aliquots of 150 g of protein were precleared with nonimmune serum and immunoprecipitated with either the Hck-specific antiserum 1077 (1:500 dilution, gift of C. Lowell, University of California, San Francisco), the Jak1-specific antiserum M7 (1:100 dilution (30)), a Jak2-specific antiserum (1:200 dilution, UBI), or a Tyk2-specific antiserum (1:200 dilution, Transduction Laboratories) for 2 h at 4°C with Protein A-Sepharose as a carrier. p21 ras was immunoprecipitated with the monoclonal antibody Y13259 (1:50 dilution of hybridoma supernatant (31)) and protein G-Sepharose as a carrier. The immunoprecipitates were washed twice in KALB buffer and twice in kinase buffer (50 mM NaCl, 10 mM Hepes, pH 7.0, 5 mM MgCl 2 , 5 mM MnCl 2 , 0.5 mM dithiothreitol, 0.1 mM Na 3 VO 4 ). The autophosphorylation reaction was carried out in 50 l of kinase buffer containing 10 Ci of [␥-32 P]ATP (3000 Ci/mmol, Bresatec, Australia) by incubating the reaction for 10 min (for Hck) or 30 min (for Jak kinases) at 25°C. All samples were separated on reducing 7.5% (for Jak kinases), 10% (for Hck), or 12.5% (for p21 ras ) SDS-PAGE gels which were subsequently treated with 1 M KOH for 2 h at 55°C in order to reduce binding of 32 P to serine and threonine residues. The intensity of radiolabeled bands was quantified using a PhosphorImager (Molecular Dynamics).
The co-existence of Jak1 and Hck in the same receptor complexes was investigated in cell lysates obtained after mild lysis in the presence of 1% digitonin (10) of parental E14TG2a ES cells stimulated with 500 pM LIF for 10 min. After preclearing of the samples with Pansorbin (Calbiochem) for 30 min at 4°C, the lysates were immunoprecipitated with the anti-Hck antiserum 1077 or the anti-Jak1 antiserum M7 and protein A-Sepharose as a carrier. The immunoprecipitated pellets were washed four times in digitonin lysis buffer before adding 100 l of kinase buffer containing 10 Ci of [␥-32 P]ATP and incubating the reaction for 30 min at 25°C. Following separation of the denatured samples on SDS-PAGE gels, the radiolabeled bands of approximately 56/59-kDa molecular mass were excised from the dried gel and subjected to partial digestion with staphylococcal V8 protease (Sigma) during the course of electrophoretic separation through a 15% SDS-PAGE gel (10).
For the detection of tyrosine-phosphorylated SHC, cell lysates were prepared from either LIF or IL6 plus sIL6R-stimulated confluent cultures that had been LIF-starved and serum-starved for 40 h prior to the experiment. Four hundred g of lysates were immunoprecipitated with the anti-phosphotyrosine antibody 4G10 (UBI) at a concentration of 1 g/ml for 2 h at 4°C with rabbit anti-mouse immunoglobulins and protein A-Sepharose as carriers. The immunoprecipitates were separated on a 10% SDS-PAGE gel and blotted with an antiserum directed against SHC (1:500 dilution; UBI). The same cell lysates were also used for the MAP kinase mobility shift assay, by separating 60 g of total cell lysates on a 10% SDS-PAGE gel and subsequent blotting with an anti-MAP kinase antibody (1:3000 dilution; UBI).
Nucleotide Exchange Activity on p21 ras -Cultures of undifferentiated cells grown in 35-mm diameter culture dishes were starved of LIF and FCS for 12 h prior to the experiment. Cultures were then washed with warm phosphate-buffered saline and 0.8 ml of permeabilization buffer (150 mM KCl, 37.5 mM NaCl, 6.25 mM MgCl 2 , 0.8 mM EGTA, 1 mM CaCl 2 , 1.24 mM ATP, 12.5 mM PIPES, pH 7.5) was added (32). Following the addition of 0.2-ml volume of streptolysin O (2 units/ml in water; Wellcome), the cells were permeabilized for 5 min. LIF (500 pM), IL6R (36 nM) plus sIL6R (64 nM) or FCS (10%) were added together with 5 Ci of [␣-32 P]GTP (3000 Ci/mmol, Bresatec), and the reaction was stopped at the indicated time by removing the supernatant and lysing the cells on ice in 1 ml of lysis buffer (100 mM NaCl, 5 mM MgCl 2 , 1 mg/ml bovine serum albumin, 1% Triton X-100, 30 mM Hepes (pH 7.4), 1 mM phenylmethylsulfonyl fluoride, 1% Trasylol, 0.1 mM Na 3 VO 4 ) containing 0.1 mM unlabeled GTP and the anti-p21 ras antibody Y13259 (1:20 dilution of hybridoma supernatant). The lysis buffer was collected and the immunoprecipitation was continued for 1 h at 4°C before adding protein G-Sepharose. After washing the pellets 3 times in lysis buffer, nucleotides were eluted in 1 M KH 2 PO 4 (pH 3.4) for 3 min at 95°C and separated on thin layer chromatography plates using unlabeled GDP and GTP as standards.

RESULTS
The pluripotentiality of ES cells can be maintained in vitro by culturing in the presence of LIF which triggers the formation of a receptor complex containing a LIFR/gp130 heterodimer. While ES cells do not express IL6R, the same biological effect can be produced by IL6 if cells are exposed to a truncated, soluble form of the IL6R (sIL6R) that lacks intracellular and transmembrane sequences (10,33). In this case, transmembrane signaling is mediated by a gp130 homodimer. To test for a possible contribution of the LIFR subunit to signaling, we compared the potency of LIF and IL6 to maintain self-renewal of ES cells. The concentration-dependent effects of LIF showed half-maximally effective concentrations (EC 50 ) in the range of 4 -10 pM (Fig. 1). IL6, in the presence of saturating concentrations of sIL6R, was about 100 -300-fold less effective than LIF, while IL6 alone had no biological activity. Furthermore, the dose-response curves for LIF and OncM were comparable and consistent with suggestions by others (33) that both cytokines transduce the differentiation-retarding signal through the same LIFR/gp130 heterodimer. Thus, our results suggest that a LIFR/gp130 heterodimer may be a more effective signaling complex than a gp130 homodimer, although both types of receptor complex transmit signals that ultimately produce the same biological effect.
Several Tyrosine Kinases Are Activated by and Associated with gp130 -Receptor complexes containing gp130 have been shown to activate members of the Jak family of tyrosine kinases in a cell type-specific manner (13). We therefore compared the time course of activation of Jak1, Jak2, Tyk2, and Hck in ES cells in response to LIF or IL6 in the presence of its soluble receptor (sIL6R). Since autophosphorylation of cytoplasmic kinases appears to be an important prerequisite for their catalytic activation in vitro, autophosphorylation experiments were carried out on lysates of cells stimulated with either LIF or IL6 plus sIL6R and immunoprecipitated with antibodies directed against Hck, Jak1, Jak2, or Tyk2 (Fig. 2).
As shown previously (10), we observed transient activation of the two isoforms of Hck, p56 hck and p59 hck (34) in response to LIF. In addition, Jak1 and Jak2 were also activated within 2-5 min of stimulation; the autophosphorylation activity of these kinases returned to basal levels within 120 min. In contrast, Tyk2 activity was only moderately stimulated 10 min after exposure of ES cells to ligands; however, this increase was maintained throughout the 120-min duration of the experiment. We confirmed that the ligand-dependent increase in kinase autophosphorylation reflected an increase in specific kinase activity, since Western blotting showed that the protein levels for all kinases investigated were unaffected over the time course of the experiment. Furthermore, the similar kinetics of p56/59 hck and Jak1 autophosphorylation is consistent with the activation of these kinases independently of each other. This notion is supported by our finding that the kinetics and extent of Jak1 autophosphorylation in ES cells expressing a constitutively activated form of Hck (hckY499F) was unaltered (data not shown). The observed overall pattern of kinase activation was similar in cells treated with either LIF or IL6 plus sIL6R, but stimulation with IL6 plus sIL6R tended to produce delayed and less pronounced autophosphorylation of these cytoplasmic kinases. Thus, while LIFR/gp130 receptor heterodimers (in the case of LIF stimulation) as well as gp130 homodimers (in the case of IL6 stimulation) cause activation of Jak1, Jak2, Tyk2, and p56/59 hck , our data suggest a functional advantage conferred by the heterodimeric receptor complex.
p56/59 hck are physical components of the functional LIFR-gp130 complex in ES cells (10), and various Jak kinases can be immunoprecipitated with gp130 receptor complexes in many transfected cell systems (13,14). Therefore, we investigated whether Jak1 exists in the same physical complex as p56/59 hck in cell lysates (containing 1% digitonin) prepared from LIFstimulated ES cells. Jak1 immunoprecipitates subjected to an in vitro kinase assay showed a tyrosine-phosphorylated species of approximately 116 kDa (corresponding to Jak1) and two additional products with identical electrophoretic mobilities as p56/59 hck immunoprecipitated from the same ES cell lysates using an Hck specific antiserum (Fig. 3A). To investigate whether the phosphorylated 56-kDa and 59-kDa products observed in the Hck and Jak1 immunoprecipitates were the same, the two species of radiolabeled bands were recovered from the gels and subjected to partial digestion with V8 protease. As shown in Fig. 3B, the pattern of V8-generated products was identical suggesting that in ES cells p56/59 hck exists in the same immunological complex as Jak1, possibly utilizing gp130 as a molecular bridge. Intriguingly, the anti-Hck antiserum did not co-immunoprecipitate Jak1 suggesting the possibility that the amino-terminal epitope recognized by the anti-Hck antiserum may be involved in the formation of the protein complex between Hck and Jak1.
Jak1 Activation Is Required for Maintaining ES Cell Pluripotentiality-To gauge the extent to which the biological effect of LIF/IL6 stimulation of ES cells could be attributed to activated Jak1 kinase, we generated ES cell lines expressing Jak1 antisense RNA under the control of the interferon ␣-inducible human 6 -16 promoter. Following a preincubation period of 6 days in the absence or presence of INF␣ (3000 units/ml), ES cell lines stably expressing the inducible antisense construct showed a suppression of Jak1 expression by approximately 80% as assessed by Western blot analysis using a Jak1 antiserum (Fig. 4). By contrast, INF␣ had no effect on Jak1 protein levels in mock-transfected cells. We next compared the proportion of ES cell colonies remaining undifferentiated in the presence of various concentrations of LIF in cultures either pretreated with INF␣ for 6 days or in untreated control cultures. We observed a shift in EC 50  INF␣ had no effect on this parameter in mock-transfected ES cells (clone 18) (Fig. 4). This observation most likely reflects a direct involvement of Jak1 in signal transduction from the LIFR-gp130 receptor complex that contributes to maintenance of the stem cell characteristics of ES cells in vitro.
The ras/MAP Kinase Pathway Is Involved in gp130-mediated Signaling-While the signaling cascades initiated through the Jak family of tyrosine kinases are believed to be distinct from those converging at the level of p21 ras , it has been suggested that the activation of at least some of the Src kinases results in the activation of Ras guanine nucleotide exchange factor (35). We therefore investigated whether LIF/IL6-dependent inhibition of ES cell differentiation involved the activation of p21 ras . In preliminary experiments, we found that LIF stimulation of 32 P ipreloaded, serum-starved ES cells led to a moderate increase in the ratio of [ 32 P]GTP-p21 ras to [ 32 P]GDP-p21 ras of approximately 45% to 55% (data not shown). The high proportion of GTP-p21 ras in unstimulated cells suggested that the majority of serumstarved undifferentiated ES cells still progressed through the cell cycle. In order to investigate the LIF-dependent activation of p21 ras in more detail, we determined the GDP/GTP exchange activity on p21 ras according to the GTP-loading protocol described by Buday and Downward (32). For this purpose, cultures of undifferentiated ES cells were starved of LIF and FCS for 12 h FIG. 2. Time course of ligand-induced increase in hck, Jak1, Jak2, and Tyk2 autophosphorylation activity. Cultures of undifferentiated ES cells were stimulated for the indicated period of time with maximally effective concentrations of either LIF (32 nM) or IL6 (36 nM) plus sIL6R (64 nM) following 12 h of cytokine starvation. The cell lysates were then analyzed for in vitro autophosphorylation activity following immunoprecipitation with antibodies directed against p56/59 hck , Jak1, Jak2, or Tyk2. The reaction products were separated by SDS-PAGE. Western blot analysis was carried out in order to assess for equal amounts of kinase proteins by using 20% of the crude cell lysates (for Hck) or of the immunoprecipitates (for Jak1, Jak2, and Tyk2). before stimulation with either LIF, IL6 plus sIL6R, or FCS in the presence of [␣-32 P]GTP. Fig. 5 shows that experimental conditions that prevent ES cell differentiation in vitro (i.e. stimulation with LIF or IL6 plus sIL6R) led to a rapid and transient 3-4-fold increase in nucleotide exchange activity on p21 ras . The stimulation of GDP/GTP exchange activity is comparable to the increase observed after mitogenic stimulation of ES cells with FCS, and the onset of nucleotide exchange activity occurred in less than 2 min after stimulation with FCS, LIF, or IL6 plus sIL6R. By contrast, treatment of ES cells with LIF or IL6 plus sIL6R had no effect on p120 GAP -mediated hydrolysis of GTP-p21 ras (data not shown). These observations suggest that the LIFR-gp130 heterodimeric or gp130 homodimeric receptor complex mediated activation of p21 ras reflects increased GDP/GTP exchange activity rather than decreased GTP hydrolysis of p21 ras .
Since phosphorylation of the SH2-adaptor protein SHC often provides a functional link between p21 ras -dependent signaling cascades and membrane-associated tyrosine kinases, we next investigated the appearance of tyrosine phosphorylation of SHC in response to stimulation of ES cells with LIF or IL6 plus sIL6R. Within 2 min after stimulation, we observed a transient increase in the abundance of the phosphorylated p46/52 Shc isoforms (Fig. 6A), while no tyrosine-phosphorylated p66 Shc was detected. Therefore, the time course of SHC phosphorylation is consistent with a possible function of SHC upstream of the GDP/GTP exchange factor in mediating gp130-dependent FIG. 5. p21 ras specific guanine nucleotide exchange activity in ES cells. Undifferentiated ES cells were starved of LIF and FCS for 12 h before being permeabilized for 5 min. [␣-32 P]GTP and LIF (32 nM), IL6 (36 nM) plus sIL6R (64 nM), or FCS (10%) was then added for the indicated period of time after which p21 ras was immunoprecipitated. Guanine nucleotides bound to p21 ras were separated by thin layer chromatography, and the associated radioactivity was visualized and quantified using a PhosphorImager. The radioactivity associated with [␣-32 P]GTP is expressed as a percentage of total radioactivity associated with guanine nucleotides, and the exchange activity in cells mock-treated for 2 min was taken as 100%. Results are given as means Ϯ S.E. of triplicate cultures.

gp130-dependent Signal Transduction in ES Cells
signaling in ES cells. Further circumstantial evidence for an involvement of p21 ras -mediated pathway(s) in maintaining ES cell pluripotentiality is suggested from our analysis of MAP kinase activation in growth-arrested ES cells in response to LIF and IL6. Since it is well established that activation of the two highly related isoforms (p42/p44) of MAP kinase is the result of combined tyrosine and threonine phosphorylation, we investigated the electrophoretic mobility shift pattern of MAP kinase in response to the mitogenic stimulus of FCS or the differentiation-inhibiting activity of LIF or IL6. Fig. 6B reveals that treatment of ES cells with either LIF or IL6 plus sIL6R induced phosphorylation of MAP kinase as does the treatment of cultures with FCS. However, the peak of cytokine-induced stimulation of MAP kinase appeared to lag behind the peak of MAP kinase activation observed after FCS treatment, suggesting that the kinetics of the activation of the Ras/MAP kinase pathway may be different when contributing to the transduction of mitogenic or differentiation inhibiting signals.
Finally, we assessed the functional consequences of p21 ras activation by introducing a constitutively activated version of p21 ras (v-Ha-ras) into ES cells. The viral oncoprotein v-Ha-ras contains two point mutations (G12V, A59T) which permanently "lock" the molecule in the GTP-bound form, thereby mimicking a constitutively activated c-Ha-ras. The expression of v-Ha-ras in ES cells was confirmed by its in vitro autophosphorylation activity which is not observed for its cellular counterpart (36) (Fig. 7). The EC 50 of LIF required to maintain the undifferentiated characteristics of ES cells expressing v-Ha-ras (E14-R2) was decreased by approximately 8-fold compared with that required by the parental control cell lines (E14). Furthermore, the reduction in LIF requirement observed in E14-R2 cells was comparable with that observed in cells harboring the activating hckY499F mutation (F10). These data suggest a maximal activation of p21 ras in F10 cells and a crucial involvement of p21 ras in mediating LIF/IL6 signal(s) in ES cells.

DISCUSSION
In this report, we provide insights into the intracellular signaling events that take place upon stimulation of ES cells with members of the LIF/IL6 family of cytokines. A number of lines of evidence suggest that suppression of differentiation by members of the LIF/IL6 family is mediated through at least two intracellular signaling cascades. We propose that one pathway depends on the activation of the Jak family of kinases, while the other is initiated through engagement of Hck, an event likely to precede the phosphorylation of SHC and activation of the Ras/MAP kinase cascade. We have shown that expression of constitutively activated v-Ha-ras partially suppresses ES cell differentiation in vitro to an extent comparable with that observed in ES expressing constitutively activated Hck (10). Interestingly, while expression of activated versions of these molecules reduces the LIF requirement of ES cells to retain their stem cell character, their influence on the retardation of differentiation is only partial. Similarly, inhibition of the Jak pathway following expression of Jak1 antisense RNA only partially abolishes LIF signaling. These observations strongly imply that a full LIF signal involves the activation of molecules that operate in non-overlapping pathways. Indeed, our results suggest that signaling through the Ras/MAP kinase cascade is additive to signaling through Jak kinases, consistent with the notion that these molecules contribute to relieving ES cell dependence on LIF independently of each other. Notwithstanding this observation, our results do not preclude the involvement of molecules, in addition to those operating downstream of Ras/MAP kinase or Jak kinases which are required to elicit a full LIF/IL6 response.
The existence of multiple, often independent intracellular signaling pathways is an emerging theme for many hemopoietin receptors. For instance, the membrane-proximal region of the gp130-related A1C2B chain, the common signal-transducing molecule for the receptors for G-CSF, IL3, and IL5, is essential for c-myc induction by G-CSF, IL3, and IL5, while the distal portion is required for the activation of the Ras/MAP kinase pathway and the induction of c-fos/c-jun activity (15). Similarly, mutational analysis of the IL2R ␤-chain in factordependent BaF cells indicates a membrane-proximal origin for Jak kinase-dependent pathway(s) culminating in the induction of c-myc and a membrane distal origin for a Src kinase-dependent pathway leading to c-fos/c-jun activation while a third unidentified pathway involves bcl-2 (16). Our results on LIF signaling in ES cells extend findings that IL6-dependent gene expression depends on the p21 ras -dependent MAP kinase cascade for the activation of NF-IL6 (37) and the Jak-dependent activation of the signal transducer and activator of transcription (STAT) 3 complexes (38).
A number of cytokines, including IL6 (39) and LIF (2) activate components of the ras/MAP kinase pathway, notably p21 ras , Raf-1, and p42/44 MAPK (40). Furthermore, cytokinemediated activation of p21 ras is often preceded by tyrosine phosphorylation of SHC and its association with Grb2 (41). In an IL6-dependent B-cell line, it has been shown that Grb2 may be constitutively associated with gp130, whereas SHC appears in the same immunological complex as gp130 following stimulation of cells with IL6 (40). While neither gp130, LIFR nor p56/59 hck contain the SHC consensus binding sequence NXXpY (42), several lines of circumstantial evidence link the activation of p56/59 hck to the Ras/MAP kinase pathway. They include the observations that Hck-mediated signaling from the Fc␥RI receptor leads to the activation of p42/44 MAPK (43), that p120 GAP can serve as a substrate for Hck in a baculovirus system (44), and that p95 vav , a nucleotide exchange factor for p21 ras (45), is co-immunoprecipitated with Hck in myelomonocytic U937 cells (43).
It is likely that the proliferative response of ES cells in vitro to serum depends on the activation of the Ras/MAP kinase pathway, since we found an increase in GDP/GTP nucleotide exchange activity and of SHC and p42/44 MAPK phosphorylation following serum stimulation. While it could be argued that partial suppression of differentiation by p21 ras is a consequence of a decrease in transit time through the cell cycle, no alteration in cell doubling time was detected during the first 96 h after withdrawal of LIF. 2 Thus, our results strongly suggest a dual role for the Ras/MAP kinase pathway in ES cells by transducing intracellular signals affecting proliferation as well as differentiation. Hence, the Ras/MAP kinase pathway may play a role in ES cells similar to that reported in the neuronal PC12 cell line, where EGF-induced mitosis as well as nerve growth factor-induced differentiation critically depend on the activation of the Ras/MAP kinase pathway (46). While SHC has been shown to associate with Jak2 following stimulation with erythropoietin (47), the contribution of Jak kinases to activation of molecules upstream of p21 ras remains unclear. For instance, it has been reported that a truncation in the human A1C2B chain results in the loss of Ras/MAP kinase signaling without affecting ligand-dependent activation of Jak2 or of the STAT proteins (15). By contrast, there is evidence for partial convergence of the Jak/STAT and p21 ras pathway(s) on the level of MAP kinase, since MAP kinase-dependent serine phosphorylation of STAT proteins appears to be critical for their binding to DNA (48).
The search for ligand-dependent association and/or activation of cytoplasmic tyrosine kinases is a classical biochemical method for identifying components of non-tyrosine kinase receptor complexes. Out of the eight currently recognized families of non-receptor kinases (49), the ligand-dependent phosphorylation and/or physical association of gp130 has been reported for members of the Src (10), Jak (13), Tec/Btk (50), and Fes families (51) of kinases. We have previously provided evidence for the physical and functional involvement of Hck in the gp130-mediated suppression of ES cell differentiation (10). Using Jak1 antisense RNA experiments, we now present strong and direct evidence for the functional involvement of Jak1 activation in LIF-dependent prevention of ES cell differentiation. Surprisingly, the decreased levels of Jak1 protein in these cells did not affect the ligand-dependent activation of Jak2 (data not shown). While these results are consistent with the finding that the absence of one Jak kinase does not prevent activation of the others in response to stimulation with members of the LIF/IL6 family of cytokines (52), they also support the notion of functional differences among Jak1, Jak2, and Tyk2 in signaling from gp130 (13). Indeed it has been shown that the absence of Jak1 most markedly reduced the IL6-dependent tyrosine phosphorylation of intracellular proteins including the cytosolic portion of gp130 (52).
In this report we show that LIF, OncM, and IL6 are functionally equivalent in preventing ES cell differentiation in vitro, thereby confirming that activation of gp130 is sufficient for the maintenance of their pluripotentiality (33). Indeed, the pluripotency of long term cultures of ES cells maintained in growth medium supplemented with OncM has been demonstrated (53), and chimeric mice have been produced from ES cells propagated in the presence of CNTF (54) or a combination of IL6 plus sIL6R (33). However, these studies have not investigated possible quantitative differences in the signaling capacity between the LIFR-gp130 heterodimeric receptor core complex (in the case of LIF, OncM, CNTF) and the gp130 homodimer receptor core complex (in the case of IL6). Our data on ES cell renewal, tyrosine phosphorylation of Hck, Jak1, Jak2, Tyk2, and SHC, and the activation p21 ras and MAP kinase strongly suggests that the LIFR binding chain in heterodimeric receptor complexes may produce an amplification of the intracellular signal(s) over those generated by the gp130 homodimer receptor complex. The suggestion of an independent signaling contribution of the LIFR binding chain is consistent with the observation that the forced homodimerization of the cytoplasmic tails of the LIFR as part of a chimeric receptor is capable of transducing some biological signals in transfected cells (11,17). 2 Since ligand-bound LIFR monomers fail to elicit biological responses, it has been proposed that signals originating from the LIF/IL6 family of hemopoietin receptors may be initiated by the ligand-induced oligomerization of at least two transmembrane receptor chains (55). The resulting heteromeric (for example in the case of LIFR) or homodimeric receptor complexes (for instance in the case IL6R) are thought to bring about a juxtapositioning of constitutively associated cy-2 M. Ernst, unpublished observations. FIG. 7. Expression of v-Ha-ras prevents ES cell differentiation. Cell lines were derived by transfecting a cDNA encoding v-Ha-ras (G12V,A59T) in wild type ES cells (E14), and the expression of v-Ha-ras was assessed by in vitro autophosphorylation assays using [␥-32 P]ATP on anti-ras immunoprecipitates prepared from cell lysates of the indicated cell lines. The radiolabeled reaction products were separated on a 12% SDS-PAGE gel (inset). Wild-type ES cells (E14), v-Ha-ras expressing Hck-wild type cells (E14-R2), or cells harboring the hckY499F mutation (F10) were cultured in the presence of the indicated concentration of LIF. After 5 days, the morphological appearance of colonies was scored as described under "Experimental Procedures." Results are given as means Ϯ S.E. of triplicate cultures. Similar results were obtained with 3 independently derived clonal cell lines. toplasmic kinases, which leads to their reciprocal transphosphorylation and activation. As a consequence, tyrosine residues within distinct regions of gp130 and the LIFR become phosphorylated, thereby providing SH2 domain binding sites for STAT and other intracellular signaling proteins that may be selectively required for the generation of cell-type specific responses. For instance, the membrane-proximal Box1 and Box2 regions found in most hemopoietin receptors are required for the association of receptors with Jak kinases as well as for the generation of proliferative signals (11,17,56). In contrast, a membrane distal domain (Box3) which is unique to gp130, LIFR, and the receptor for G-CSF appears to be essential for the transduction of nonproliferative signals in transfected hepatoma and neuroblastoma cell lines (11) since these signals are not generated from receptor truncations lacking Box3. Consistent with these observations, we have observed that activation of the transfected receptor for G-CSF partially prevents ES cell differentiation. 2 Conversely, the introduction and activation of the related thrombopoietin receptor that contains Box1 and Box2 while lacking Box3 failed to promote ES cell self-renewal despite the ligand-dependent activation of Jak kinases. 2 Similarly, the stimulation of the INF␣ receptor led to activation of Jak1 and Tyk2 in ES cells without affecting their responsiveness to LIF. These results support the notion that the cytokine response of a particular cell type is determined by the binding specificity of signaling proteins to individual phosphotyrosine residues within the receptor as well as the association and activation of one or other Jak kinases (21). Indeed, recent findings have indicated that the C-terminal portions (including Box3) of gp130, the LIFR, and the G-CSFR are required for the activation of STAT3 (57). Thus, the appearance of STAT3containing DNA binding activity is readily observed following G-CSF, LIF, or IL6 stimulation, while the extent of STAT3 activation is much less pronounced after engagement of the INF␣ receptor (19). It is tempting to speculate that Jak kinasemediated STAT3 activation may be a critical component in the intracellular signaling cascade activated in ES cells undergoing self-renewal.