Mitogenic synergy through multilevel convergence of ... - Nature

Oncogene (2002) 21, 2201 ± 2211 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Mitogenic synergy through multilevel convergence of hepatocyte growth factor and interleukin-4 signaling pathways Regina M Day1,2,3, Lilian Soon1,2,3, Diane Breckenridge1,2,3, Benjamin Bridges1,2,3, Bharvin KR Patel1,2,3, Ling Mei Wang1,2,3, Seth J Corey1,2 and Donald P Bottaro*,1,2,3 1

Laboratory of Cellular and Molecular Biology, Division of Basic Sciences, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA; 2Department of Pediatrics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA; 3 Department of Pharmacology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA

Hepatocyte growth factor (HGF) regulates various physiological and developmental processes in concert with other growth factors, cytokines and hormones. We examined interactions between cell signaling events elicited by HGF and the cytokine interleukin (IL)-4, in the IL-3-dependent murine myeloid cell line 32D transfected with the human HGF receptor, c-Met. HGF was a potent mitogen in these cells, and prevented apoptosis in response to IL-3 withdrawal. IL-4 showed modest anti-apoptotic activity, but no signi®cant mitogenic activity. IL-4 synergistically enhanced HGFstimulated DNA synthesis, whereas only additive prevention of apoptosis was observed. IL-4 did not enhance HGF-dependent tyrosine phosphorylation of c-Met or Shc. In contrast, HGF-stimulated activation of MAP kinases was enhanced by IL-4, suggesting that the IL-4 and HGF signaling pathways converge upstream of these events. Although phosphatidylinositol 3-kinase (PI3K) inhibitors diminished HGF-induced mitogenesis, antiapoptosis, and MAP kinase activation, IL-4 enhanced HGF signaling persisted even in the presence of these inhibitors. IL-4 enhancement of HGF signaling was partially blocked in 32D/c-Met cells treated with inhibitors of MEK1 or c-Src kinases, completely blocked by expression of a catalytically inactive mutant of Janus kinase 3 (Jak3), and increased in 32D/c-Met cells overexpressing STAT6. Our results suggest that the IL-4 and HGF pathways converge at multiple levels, and that IL-4-dependent Jak3 and STAT6 activities modulate signaling events independent of PI3K to enhance HGF-dependent mitogenesis in myeloid cells, and possibly other common cellular targets. Oncogene (2002) 21, 2201 ± 2211. DOI: 10.1038/sj/ onc/1205289 Keywords: growth factor; cytokine; signal transduction; cell proliferation

*Correspondence: DP Bottaro, Department of Cell and Molecular Biology, EntreMed, Inc., Rockville, Maryland, MD 20850 USA; E-mail: [email protected] Received 2 May 2000; revised 3 January 2002; accepted 7 January 2002

Introduction Hepatocyte growth factor (HGF) stimulates mitogenesis, motogenesis, and morphogenesis in a wide range of cellular targets including epithelial and endothelial cells, hematopoietic cells, neurons, melanocytes, as well as hepatocytes (reviewed in Michalopoulos and DeFrances, 1997; Rubin et al., 1993; Zarnegar and Michalopoulos, 1995). These pleiotropic eects play important roles during development, organogenesis and tissue regeneration. HGF is essential for the normal development of both liver and placenta (reviewed in Birchmeier and Gherardi, 1998), contributes to neural development (reviewed in Streit and Stern, 1997), branching morphogenesis in various organs (reviewed in Birchmeier et al., 1997), and promotes kidney and lung regeneration (Yanagita et al., 1993; Balkovetz and Lipschutz, 1999). Among hematopoietic tissues, HGF is produced by bone marrow stromal cells (Weimar et al., 1998), and cMet is expressed in CD34+ hematopoietic progenitor cells (HPC) derived from bone marrow, peripheral blood, and cord blood (Weimar et al., 1998; Kmiecik et al., 1992; Ikehara, 1996). HGF promotes the proliferation, adhesion, and survival of CD34+ HPC (Weimar et al., 1998; Kmiecik et al., 1992), and acts synergistically with interleukin (IL)-3 or granulocyte-macrophage colony stimulating factor (GM-CSF) to increase colony forming units in culture by CD34+ HPC in vitro (Kmiecik et al., 1992; Ikehara, 1996). Erythropoietin (EPO) and HGF also act synergistically to induce the formation of erythroid burst-forming unit colonies from CD34+ HPC (Ikehara, 1996; Iguchi et al., 1999). HGF compensates for de®ciencies in stem cell proliferation observed in mice harboring dominant mutations in the genes encoding stem cell factor (SCF) or its receptor, ckit (Yu et al., 1998). HGF enhances antibody production in mature B cells (Delaney et al., 1993), and stimulates rapid cytoskeletal changes, adhesion, and migration in peripheral blood T cells (Adams et al., 1994). In addition to roles in normal development and adult homeostasis, HGF signaling is strongly linked to cancer, including colon, breast, lung, thyroid and renal carcinomas, melanoma and several sarcomas, as well as glioblastoma (reviewed in Vande Woude et al., 1997).

HGF and IL-4 mitogenic synergy RM Day et al

2202

Oncogene

Among hematopoietic tumors, many human leukemia cell lines secrete HGF in vivo and in vitro, including those derived from adult T-cell leukemia, acute lymphoblastic leukemia, multiple myeloma (MM), acute undierentiated leukemia, chronic myelocytic leukemia, acute myeloblastic leukemia, and acute promyelocytic leukemia (Gohda et al., 1995). HGF promotes the adhesion of B-lymphoma cells to the extracellular matrix (Weimar et al., 1997), and the overproduction of HGF associated with MM has been linked to the extensive bone marrow angiogenesis and bone destruction characteristic of that disease (Borset et al., 1999). An emerging theme in our understanding of HGF action is the modulation of its biological activities by other cytokines. Among the many cytokines whose actions coincide spatially and temporally with those of HGF, few are as ubiquitous as Interleukin (IL)-4. IL-4 is produced predominantly by T cells, mast cells and basophils (Paul, 1991). The IL-4 receptor, IL4Ra, a member of the hematopoietin/cytokine receptor superfamily, is expressed on both hematopoietic and nonhematopoietic cells (Cosman, 1993). IL-4 stimulates the proliferation and dierentiation of B cells, the proliferation of T cells and mast cells, and regulates the lymphokine producing phenotype of CD4+ T lymphocytes (Keegan et al., 1994). In non-hematopoietic cells such as ®broblasts and epithelial cells, IL-4 modulates extracellular matrix production, causes changes in morphology and cytoskeletal organization, and stimulates chemotaxis (Postlethwaite and Seyer, 1991; Postlethwaite et al., 1992). IL-4 can also stimulate the dierentiation of human carcinoma cells, and thereby block their proliferation (Hoon et al., 1991; Morisaki et al., 1992, 1994). The biological eects of IL-4 are also in¯uenced by other cytokines and growth factors. IL-4 complements IL-2 in T-cell tumor lines for ex vivo autocrine growth and for in vivo tumorigenicity (Hassuneh et al., 1997), acts synergistically with platelet-derived growth factor (PDGF) in mitogenesis (Patel et al., 1996), and inhibits HGFstimulated migration of colon carcinoma cells (Uchiyama et al., 1996). We investigated the eects of IL-4 on HGF biological activities in the myeloid cell line 32D transfected with human HGF receptor (32D/c-Met cells; Day et al., 1999). HGF stimulated mitogenesis and prevented apoptosis in response to IL-3 withdrawal. IL-4, in contrast, showed modest antiapoptotic activity and no mitogenic activity. However, IL-4 synergistically enhanced HGF-stimulated DNA synthesis, which was re¯ected biochemically at the level of MAP kinase, but not phosphatidylinositol 3-kinase (PI3K) activation. HGF/IL-4 mitogenic synergy was partially dependent on Src kinases, required the activation of Jak3, and was enhanced by overexpressing STAT6. Our results suggest that the HGF and IL4 signaling pathways converge at multiple levels downstream of HGF-stimulated PI3K activation, and that IL-4-activated Jak3 and STAT6 are critical contributors to HGF/IL-4 mitogenic synergy in myeloid cells.

Results Mitogenic synergy between HGF and IL-4 The mitogenic response of 32D/c-Met cells to HGF is shown in Figure 1a. Increased DNA synthesis was detectable at 50 ng/ml (0.55 nM) HGF, and reached maximal levels at 1 mg/ml (11 nM; ED50=3.3 nM). 32D/c-Met cells also express low levels of IL-4 receptor, but IL-4 is not mitogenic in these cells (Wang et al., 1993). As shown in Figure 1b, IL-4 at concentrations of up to 100 ng/ml did not stimulate signi®cant [3H]-thymidine incorporation. However, IL4 signi®cantly enhanced HGF-stimulated [3H]-thymidine incorporation in a dose-dependent manner (Figure 1b). These results show that the eects of HGF with IL-4 far exceeded the sum of either mitogen administered alone, indicating mitogenic synergy between IL-4 and HGF in 32D/c-Met cells. We estimate that IL-4 (1 ng/ml) increased the ED50 of HGF approximately 3 ± 5-fold. Similar eects have been observed in epithelial cells, such as Balb/MK keratinocytes (JS Rubin and WG Taylor, personal communication). Heparan sulfate proteoglycan (HSPG) has been shown to modulate HGF signaling in several target cell types, including Baf3 and 32D, which lack abundant cell surface HSPG (Day et al., 1999; Schwall et al., 1996; Sakata et al., 1997). Soluble heparin (10 mg/ml), which may substitute for cell-surface HSPG to some extent, enhanced HGF mitogenic activity in 32D/c-Met cells at least threefold, but did not impart mitogenic activity to the truncated HGF isoform, HGF/NK2 (Figure 2). The eects of heparin were most striking at low levels of HGF (2 and 20 ng/ml), and decreased at very high (200 ng/ml) HGF concentrations, as maximal mitogenic stimulation was reached (Figure 2a). The simultaneous addition of both soluble heparin and IL-4 to 32D/c-Met cells increased the mitogenic activity of HGF to a greater extent than either factor alone. As shown in Figure 2a, IL-4 (1 ng/ml) or heparin (10 mg/ml) alone enhanced mitogenic activity of HGF (2 ng/ml) about fourfold. In the presence of both IL-4 and heparin, HGF-dependent mitogenic activity was more than 20-fold higher than that of HGF alone (Figure 2a). In addition to dierences in their presumed sites of action, the ability of soluble heparin to further enhance the HGF/IL-4 mitogenic synergy supports the hypothesis that IL-4 and heparin facilitate HGF mitogenicity by distinct mechanisms: heparin at the level of HGF-c-Met interaction, and IL-4 at the level of intracellular signaling. Interestingly, the combination of IL-4 and high concentrations (1 mg/ml) of HGF/NK2 was mitogenic, while neither factor alone stimulated signi®cant DNA synthesis (Figure 2b). In most cell lines, HGF/NK2 stimulates increased cell motility, but not mitogenesis (Stahl et al., 1997; Chan et al., 1991). In 32D/c-Met cells, HGF and HGF/NK2 potently stimulate the activation of c-Met kinase, phosphatidylinositol 3kinase (PI3K), and MAP kinase, yet the biological response to HGF/NK2 is limited to increased cell motility (Day et al., 1999). Thus, HGF/NK2 appears


2203

Figure 1 Dose-dependent modulation of HGF mitogenic activity by IL-4. 32D/c-Met cells were starved for 36 h in RPMI+10% FBS with the indicated concentrations of growth factor. [3H]-Thymidine incorporation into DNA was measured as described in Materials and methods. (a) HGF alone; (b) increasing concentrations of IL-4 in the absence (open circles) or presence (solid squares) of HGF (100 ng/ml). Values represent mean+standard deviation of triplicate samples; results are representative of at least three experiments. Where no standard deviations are visible, the error is smaller than the symbol size

synthesis. The eects of IL-4 on signaling by HGF and HGF/NK2 suggest that the critical elements not activated in response to HGF/NK2 could be common to both HGF and IL-4 signaling pathways. In contrast to some other cultured cell models (Schwall et al., 1996), the combination of HGF/NK2 and soluble heparin did not stimulate signi®cant mitogenesis in 32D/c-Met cells, even at very high concentrations of HGF/NK2 (Figure 2b). Moreover, the mitogenicity observed in response to the combination of HGF/NK2 and IL-4 was not further enhanced in the presence of heparin (Figure 2b). HGF and IL-4 protect 32D/c-Met cells from apoptosis

Figure 2 IL-4 and soluble heparin independently enhance HGF mitogenic activity in heparan sulfate proteoglycan-de®cient 32D/ c-Met cells. Intact cells were treated with either HGF (a) or HGF/ NK2 (b) at the indicated concentrations in the presence or absence of and/or soluble heparin. [3H]-Thymidine incorporation into DNA was measured as described in Materials and methods. Open bars: HGF or HGF/NK2 alone; black bars:+IL-4 (1 ng/ ml); gray bars:+soluble heparin (10 mg/ml); stippled bars:+IL-4 and soluble heparin. Values represent mean+standard deviation of triplicate samples; results are representative of at least three experiments. Where no standard deviations are visible, the error was too small to indicate graphically

to activate a subset of the intracellular pathways that are activated by full-length HGF, but fails to activate some as yet unidenti®ed eector(s) critical to DNA

32D cells have been used extensively to study the antiapoptotic eects of IL-3 and IL-4 (Wang et al., 1992, 1993; Zamorano et al., 1996). We observed that within 36 h of IL-3 deprivation, approximately 31% of 32D/cMet cells enter apoptosis, and that the addition of 1 ng/ml IL-3 reduced this to 0.2% (Figure 3). The addition of HGF at concentrations of 50 and 200 ng/ ml prevented 75 and 92% of apoptosis stimulated by IL-3 deprivation, respectively (Figure 3). Consistent with previous reports showing that IL-4 is nonmitogenic but anti-apoptotic (Wang et al., 1992, 1993), this cytokine (1 ng/ml) reduced apoptosis associated with IL-3 withdrawal in 32D/c-Met cells by approximately 25% (Figure 3). IL-4 treatment enhanced the anti-apoptotic eects of HGF, reducing apoptosis 38% (from 8.2 to 5.1%) for 50 ng/ml HGF, and 66% (from 2.6 to 0.9%) for 200 ng/ml HGF (Figure 3). The anti-apoptotic eects of HGF in combination with IL-4 were approximately additive, suggesting some degree of independence in their signaling pathways. IL-4 caused no further reduction of apoptosis in the presence of IL-3 (data not shown). HGF/NK2 had no anti-apoptotic activity on 32D/cMet cells, even in combination with IL-4, above the eects of IL-4 alone (data not shown). Oncogene


2204

Figure 3 IL-4 modulates HGF anti-apoptotic activity. 32D/cMet cells were treated with the indicated concentrations of HGF with or without IL-4 (1 ng/ml), or with IL-3 (1 ng/ml). FACS analysis was used to determine the percentage of apoptotic cells as described in Materials and methods. Open bars: without IL-4; gray bars:+IL-4. Values represent mean+standard deviation of triplicate samples; results are representative of at least three experiments. Where no standard deviations are visible, the error was too small to indicate graphically

IL-4 enhances HGF mitogenicity downstream of c-Met, Shc, and PI3K To explore the mechanism(s) by which IL-4 synergized with HGF, we investigated the activation of c-Met, Shc and PI3K. After treatment of 32D/c-Met cells with HGF, IL-4 or IL-3, cell lysates were immunoprecipitated with anti-phosphotyrosine and immunoblot analysis was performed using anti-c-Met. As shown in Figure 4a, signi®cantly more c-Met was immunoprecipitated with anti-phosphotyrosine after treatment with HGF, indicating HGF-stimulated c-Met kinase autophosphorylation and activation in these cells. Neither IL-4 nor IL-3 alone stimulated signi®cant cMet tyrosine phosphorylation, and neither cytokine enhanced HGF-stimulated c-Met activation (Figure 4a), suggesting that IL-4 did not enhance the activation of c-Met by HGF. Shc is an adapter protein implicated in the mitogenic pathways of several growth factors, and is thought to participate in the activation of the PI3K and SOS-RasMAP kinase pathways (Bon®ni et al., 1996). In 32D/cMet cells, Shc was tyrosine phosphorylated in response to IL-3, and, to a lesser extent, in response to HGF (Figure 4b). IL-4 alone did not increase Shc phosphorylation on tyrosine, and IL-4 did not appear to enhance HGF-stimulated Shc tyrosine phosphorylation, suggesting that IL-4 acts independently or downstream of Shc in HGF signaling. Upon its activation by HGF, c-Met has been shown to associate with, and activate, PI3K in epithelial cells (Graziani et al., 1991), as well as in 32D/c-Met cells (Day et al., 1999). PI3K activation is required for HGF-induced mitogenesis in SP1 mammary carcinoma cells and Mv1Lu lung epithelial cells (Rahimi et al., 1996). We investigated the eects of IL-4 on HGFstimulated PI3K activity in 32D/c-Met cells by measuring PI3K activity in anti-phosphotyrosine immunoprecipitates prepared from intact cells treated with each factor alone or in combination. As shown in Oncogene

Figure 4 IL-4 does not enhance HGF stimulated c-Met, Shc or PI3K activation. 32D/c-Met cells were treated with HGF, IL-4, HGF+IL-4 or IL-3. Cell lysates were immunoprecipitated with anti-pY and immunoblotted for either c-Met (a) or shc (b), or assayed for PI3K activity (c). HGF concentrations are indicated in mg/ml; `+' indicates HGF at 200 ng/ml. IL-3 and IL-4 were added at 100 ng/ml, ®nal concentration. The results are representative of three or more experiments

Figure 4c, HGF stimulated PI3K activation while IL-4 did not, and IL-4 did not appear to enhance PI3K activation by HGF. Higher levels of PI3K activity were observed in response to higher concentrations of HGF, indicating the level of PI3K activity observed in response to both HGF and IL-4 was submaximal (data not shown). These data are consistent with a point of convergence between the HGF and IL-4 mitogenic signaling pathways downstream of PI3K activation. Several HGF-stimulated biological activities are dependent on PI3K in epithelial HGF target cell types (Day et al., 1999; Graziani et al., 1991; Rahimi et al., 1996). To assess the PI3K dependence of HGFstimulated mitogenic and anti-apoptotic activities in 32D/c-Met cells, and to further explore the contribution of IL-4 to these activities, each was examined in the presence and absence of speci®c PI3K inhibitors (Figure 5). Neither Wortmannin (100 nM) nor LY294002 (10 mM) exhibited strong inhibitory eects on IL-3-stimulated 32D/c-Met cell growth (Figure 5a). In contrast, DNA synthesis elicited by HGF was strongly suppressed: Wortmannin inhibited 95% of HGF-induced mitogenesis, while LY294002 inhibited 88% (Figure 5a). Thus HGF-stimulated mitogenesis in 32D/c-Met cells exhibited a degree of PI3K dependence similar to those reported for other HGF target cell types (Rahimi et al., 1996). Immunoblotting with antiactive Akt antibody con®rmed that the activation of targets directly downstream of PI3K was completely blocked under these conditions (data not shown). In the presence of both HGF and IL-4, Wortmannin inhibited 84% of HGF-stimulated mitogenesis, while LY294002 blocked 60% of this activity (Figure 5a). Close scrutiny of these data revealed that despite potent blockade of HGF mitogenic activity by PI3K


inhibitors, signi®cant enhancement of HGF mitogenic signaling by IL-4 was still observed, again suggesting that convergence between the HGF and IL-4 mitogenic signaling pathways occurred downstream of PI3K. Wortmannin did not signi®cantly inhibit IL-3dependent anti-apoptotic activity, but HGF-dependent anti-apoptotic activity was strongly suppressed by this drug (Figure 5b). The percentage of cells entering apoptosis in the presence of HGF with Wortmannin increased ®vefold over the percentage treated with HGF alone (Figure 5b). IL-4 enhanced HGF-mediated protection from apoptosis even in the presence of Wortmannin, reducing the level of apoptosis observed in cells treated with both HGF and Wortmannin by approximately 30% (Figure 5b). These data are consistent with previous ®ndings that IL-4 inhibited apoptosis via a PI3K-independent pathway (Yashiro et al., 1996).

examined the eects of HGF and IL-4 on MAP kinase activation by immunoblot analysis using antibody raised against active (phosphorylated at Y204) p44/42 MAP kinases. HGF and IL-3 stimulated the activation of these MAP kinases above the level of phosphorylation observed in untreated control 32D/c-Met cells (Figure 6a). Increased activation of MAP kinase was seen primarily in the level of phosphorylation of ERK-2. IL-4 enhancement of HGF-stimulated ERK-2 phosphorylation was observed, while IL-4 alone had no detectable eect on MAP kinase activation (Figure 6a). To con®rm the apparent increase in MAP kinase activity indicated by immunoblotting experiments, the time course of MAP kinase activity toward a synthetic myelin basic protein (MBP) peptide in vitro was analysed in anti-MAP kinase immunoprecipitates prepared from control and growth factor treated 32D/c-Met cells. MAP kinase-catalysed MBP peptide phosphorylation in vitro increased threefold above

2205

IL-4 enhances HGF-stimulated MAP kinase activation The p44 and p42 MAP kinases (ERK-1 and -2) are activated by a wide variety of growth factors and cytokines, and are critical for many mitogenic pathways (Cowley et al., 1994; Mansour et al., 1994). We

Figure 5 HGF-stimulated mitogenic and anti-apoptotic activities are PI3K-dependent in 32D/c-Met cells. (a) DNA synthesis was measured in 32D/c-Met cells treated with IL-3 (1 ng/ml), HGF (200 ng/ml), or HGF+IL-4 (1 ng/ml), as indicated below the x-axis. Open bars represent no drug addition, black bars represent treatment with Wortmannin (0.1 mM), and gray bars represent treatment with LY-294002 (10 mM). Values shown are mean+standard deviation of triplicate samples. The results are representative of three or more experiments. (b) Apoptosis of 32D/c-Met cells that had been left untreated (control), or treated with HGF, HGF+IL-4, or IL-3 at the same ®nal concentrations listed in panel (a) was measured after 36 h as described in Materials and methods. Open bars represent no drug addition, black bars represent treatment with Wortmannin (0.1 mM). Values shown are mean+standard deviation of triplicate samples. The results are representative of three or more experiments

Figure 6 IL-4 enhances MAP kinase activation by HGF. (a) 32D/c-Met cells were treated for 10 min at 378C with HGF at the indicated concentrations (mg/ml), or IL-4 (100 ng/ml), or IL-3 (100 ng/ml), or left untreated (lane 6), and cell lysates were immunobloted with anti-phospho-MAP kinase. (b) 32D/c-Met cells were treated for the indicated times at 378C with HGF (200 ng/ml; open squares) or HGF+IL-4 (200 ng/ml and 100 ng/ ml, respectively; open circles). Anti-MAP kinase immunoprecipitates were then assayed for MAPK activity in vitro using myelin basic protein (MBP) peptide as substrate, as described in Materials and methods. Values shown are mean+standard deviation of triplicate samples; where error bars are not visible the standard deviation is smaller than the symbol size. (c) 32D/cMet cells were treated for 10 min at 378C with HGF (200 ng/ml), IL-4 (100 ng/ml), IL-3 (100 ng/ml), or HGF+IL-4, in the presence and absence of Wortmannin (0.1 mM) as indicated. Lysates were immunobloted with monoclonal anti-phospho-MAP kinase. The results are representative of three experiments Oncogene


2206

basal levels in response to HGF treatment, and the combination of HGF with IL-4 increased MAP kinase activity more than sevenfold (Figure 6b). These experiments also revealed that IL-4 enhanced primarily the magnitude of MAPK activation, without aecting the kinetics or duration of activation (Figure 6b). The results of both immunoblot analysis and in vitro kinase assays suggest that the modulation of HGF signaling by IL-4 can occur, at least in part, upstream of MAP kinase. HGF-stimulated MAP kinase activation is thought to be PI3K-dependent in epithelial cells (Rahimi et al., 1996), and immunoblot analysis of growth factortreated 32D/c-Met cells using the anti-active MAP kinase antibody revealed that MAP kinase activation induced by HGF, but not IL-3, was strongly inhibited by Wortmannin (Figure 6c). Consistent with the eects of Wortmannin on DNA synthesis driven by HGF and IL-4, evidence of IL-4 enhancement of HGF-mediated MAP kinase activation persisted in the presence of Wortmannin, although the total MAP kinase activity detected was signi®cantly decreased (Figure 6c). Buers used in these assays contained dimethyl sulfoxide (DMSO) to increase the solubility of Wortmannin, and the presence of DMSO was associated with an equal activation of both p42 and p44 MAP kinases by HGF and IL-3 (Figure 6c), in contrast to the predominant activation of p42 by these factors in the absence of DMSO (Figure 6a). The mechanism by which this occurs is unknown. Nevertheless, our results show unambiguously that HGF-stimulated ERK-2 activation is PI3K-dependent in 32D/c-Met cells, and that this activation was enhanced by IL-4. The persistence of IL-4 enhancement of HGF-stimulated MAP kinase activation in the presence of Wortmannin also indicates convergence of the HGF and IL-4 signaling pathways upstream of MAP kinase activation.

and data not shown). In contrast, partial inhibition (approximately 35%) of the increase in DNA synthesis stimulated by the combination of HGF and IL-4 was consistently observed (Figure 7a), suggesting that part of the synergistic eect of IL-4 might occur through MEK1 activation. Src kinases mediate the biological actions of a variety of cytokines and growth factors, including HGF (Sunitha et al., 1999). Although we have observed increased tyrosine phosphorylation of pp60 c-Src in 32D/c-Met cells in response to HGF treatment (data not shown), the individual roles of the several Src kinase family members expressed in HGF target cells have not yet been thoroughly characterized. The recent development of selective Src kinase inhibitors has provided a means to rapidly evaluate the impact of Src kinases on speci®c biological activities. The Src inhibitors PP1 and PP2 are eective on members of

Partial blockade of IL-4 enhanced HGF signaling by MEK1 and Src inhibitors To identify eectors at the convergence of the HGF and IL-4 mitogenic signaling pathways that might act downstream of PI3K and upstream of MAP kinases, we characterized the eects of two selective kinase inhibitors on DNA synthesis stimulated by HGF and IL-4: the MEK1 inhibitor PD98059 (Alessi et al., 1995), and the Src inhibitor PP2 (Hanke et al., 1996). MEK1 has been shown to be an important activator of MAP kinases, the latter being critical for HGFstimulated mitogenesis and motogenesis in certain epithelial HGF target cell types (Vande Woude et al., 1997). In 32D/c-Met cells, however, PD98059 was not found to be a potent inhibitor of HGF-stimulated cell migration (Day et al., 1999). Consistent with those observations, PD98059 had no signi®cant eect on HGF-stimulated DNA synthesis in 32D/c-Met cells at concentrations up to 20 mM, despite complete blockade of MAP kinase activation con®rmed by immunoblotting with anti-active MAP kinase antibody (Figure 7a Oncogene

Figure 7 Mek1 and Src kinase inhibitors diminish the enhancement of HGF mitogenic signaling by IL-4. (a) DNA synthesis by 32D/c-Met cells left untreated (open bars) or treated with HGF (50 ng/ml; gray bars) or HGF+IL-4 (50 ng/ml and 10 ng/ml, respectively; black bars). The Mek1 inhibitor PD98059 was added at a concentration of 10 or 20 mM, as indicated. Controls for the presence of low levels of DMSO in PD98059-containing media consisted of samples treated with identical amounts of DMSO alone, as indicated. (b) DNA synthesis by 32D/c-Met cells treated with the Src kinase inhibitor PP2 at 1, 3, or 10 mM concentrations 15 min prior to treatment with HGF (30 ng/ml; open bars) or HGF+IL-4 (30 ng/ml and 5 ng/ml, respectively; gray bars), as indicated. Controls for the presence of low levels of DMSO in PP2-containing media consisted of samples treated with identical amounts of DMSO alone, as indicated. For both panels, values represent mean+standard deviation of triplicate samples; where no standard deviation is visible, the value was too small to indicate graphically. The results are representative of at least three experiments


this kinase family at concentrations of 0.1 ± 10 mM, whereas eects on other tyrosine kinase families such as EGF receptor require at least 10-fold higher concentrations (Hanke et al., 1996). We found that HGF-stimulated DNA synthesis in 32D/c-Met cells was only modestly inhibited by PP2 over a range of 1 ± 10 mM (Figure 7b). However, PP2 caused more substantial inhibition of HGF+IL-4 stimulated DNA synthesis (Figure 7b). While the inhibition of combined HGF+IL-4 mitogenic activity was partial, maximal eects of 40 ± 50% at 10 mM PP2 were consistently seen. These results, together with those obtained with the MEK1 inhibitor PD98059, suggest that the HGF and IL-4 signaling pathways converge at multiple levels. HGF/IL-4 mitogenic synergy mediated by Jak3 and STAT6 In parallel with our analysis of HGF-activated pathways for biochemical evidence of HGF/IL-4 mitotic synergy, we examined IL-4 activated pathways to characterize their potential roles in this phenomenon. IL-4Ra activation can lead to the tyrosine phosphorylation of Jak1, Jak3, insulin receptor substrate (IRS) 1, IRS-2, STAT6, and various other eector molecules (Keegan et al., 1994; Witthuhn et al., 1994; Wang et al., 1992, 1993). In 32D cells, which do not express IRS-1 or IRS-2, IL-4 is unable to stimulate the activation of PI3K and its downstream eectors, and lacks mitogenic activity (Figure 1b). However, IL-4 does stimulate the activation of Jak1, Jak3, and STAT6 in 32D cells (Patel et al., 1996). We examined the role of Jak1 activation in HGF/IL-4 mitogenic synergy by comparing ligand-stimulated DNA synthesis in 32D/c-Met cells with that observed in the same cells transfected with a dominant-negative, kinase dead (kd) Jak1 mutant. Although expression of the mutant Jak1 protein was well in excess of endogenous Jak1, no signi®cant eect on HGF/IL-4 mitogenic synergy was observed (data not shown). We examined the role of Jak3 in HGF/IL-4 mitogenic synergy using the same strategy outlined for Jak1. 32D/c-Met cells were transfected with a kinase dead mutant of human Jak3, and clonal cell lines expressing mutant Jak3 in excess of endogenous Jak3 were selected for further analysis. A comparison of ligand-stimulated STAT6 tyrosine phosphorylation in parental 32D/c-Met cells and a representative c-Met/ Jak3kd cell line revealed that it was severely reduced in the presence of the dominant negative Jak3 mutant (Figure 8). Associated with the strong suppression of STAT6 tyrosine phosphorylation was the complete loss of mitogenic synergy between HGF and IL-4 in c-Met/ Jak3kd cells (Table 1). These results suggest that Jak3 is more critical than Jak1 for HGF+IL-4 signaling in 32D/c-Met cells, and that IL-4 stimulated Jak3 activation is required for IL-4 enhancement of HGF mitogenic signaling. Tyrosine phosphorylation and activation of STAT proteins is an important consequence of Jak kinase

activation by IL-4R, and overexpression of STAT6 can enable mitogenic signaling by IL-4 in this system (Patel et al., 1996). To further explore the impact of STAT6 on the mitogenic synergy between HGF and IL-4, DNA synthesis was compared among clonal 32D cell lines expressing both human c-Met and STAT6 at various expression levels (Table 1). Even very high expression of STAT6, relative to endogenous levels, was not associated with transformation, and all cell lines remained IL-3-dependent for proliferation (data not shown). Relative to 32D/c-Met cells, c-Met/STAT6 cells expressed approximately twofold more STAT6 protein, c-Met/STAT6.1 cells expressed about 20-fold more STAT6 protein, and the STAT6 protein content of c-Met/STAT6.2 was about 30-fold over endogenous STAT6 levels (data not shown). Tyrosine phosphorylation of STAT6 in c-Met/STAT6 cells was more potently stimulated by IL-4, but phosphorylation in response to HGF was not detected (Figure 8). Moreover, HGF did not signi®cantly enhance IL-4-stimulated STAT6 phosphorylation in any of these cell lines (Figure 8 and data not shown). However, DNA synthesis stimulated by the combination of HGF and IL-4 was substantially increased in these cells relative to the parental 32D/c-Met cell line (Table 1). Twofold overexpression of STAT6 protein was associated with a modest but consistent increase in the HGF/IL-4

2207

Figure 8 Jak3 and STAT6 contribute to IL-4 enhanced HGF signaling. 32D/cMet cells (left four lanes), and two 32D/cMetderived clones transfected with a Jak3 kinase-dead mutant (cMet/Jak3kd; center four lanes), or human STAT6 (c-Met/STAT6; right four lanes) were treated with HGF (200 ng/ml), IL-4 (100 ng/ml), HGF+IL-4, or left untreated, as indicated above each lane. Cells were lysed and equal amounts of protein were immunoprecipitated with anti-STAT6. STAT6 immunoprecipitates were resolved by SDS ± PAGE and immunoblotted with anti-pY (top panel), or anti-STAT6 (bottom panel). Note that because human STAT6 is not as well-recognized as murine STAT6 by the anti-STAT6 antibody used for immunoblotting, the rightmost four lanes in the bottom panel underestimate the total STAT6 protein content of these samples Oncogene


2208

Table 1 HGF+IL-4-stimulated DNA synthesis by 32D/cMet cell lines Cell line 32D/c-Met c-Met/Jak3kd c-Met/STAT6 c-Met/STAT6.1 c-Met/STAT6.2

DNA synthesisa 162+8 80+9 200+17 340+11 510+16

a 3

[ H]-Thymidine incorporation into DNA was measured as described in Materials and methods. 32D cells expressing human c-Met, kinasedead JAK3, or STAT6 at various expression levels were treated with HGF (100 ng/ml) or HGF+IL-4 (100 ng/ml and 2 ng/ml, respectively). Values shoen are mean+standard error of the mean of triplicate samples, and are expressed as a percentage of DNA synthesis stimulated by HGF treatment alone. Results are representative of at least three experiments

mitogenic synergy: 162 vs 200%, relative to HGF treatment alone (Table 1). Higher levels of STAT6 expression were associated with greater increases in HGF/IL-4 synergy: 340 and 510% for c-Met STAT6.1 and c-Met/STAT6.2 cells, respectively, relative to HGF treatment alone (Table 1). These results con®rm that STAT6 is an important contributor to the IL-4 signaling pathway, and suggest that STAT6 also mediates, at least in part, the HGF/IL-4 mitogenic synergy observed in this system. Discussion Various growth factors, cytokines and hormones acting in concert regulate complex physiological processes, and one of the current challenges of cell biology and physiology is understanding the integration of dierent incoming signals by target cells. HGF is involved in embryogenesis, organogenesis, homeostasis and oncogenesis, yet relatively little is known regarding the modulation of HGF-stimulated biological activities by other signaling molecules. HGF stimulates motility in melanocytes, but HGF-stimulated mitogenesis in that system appears to require the presence of other factors, such as basic ®broblast growth factor (bFGF), mast cell growth factor (MGF), or dibutyryl cAMP (Rubin et al., 1993). HGF in combination with other growth factors may support the growth and migration of primary melanoma cells (Rubin et al., 1993). HGF appears to cooperate with nerve growth factor to enhance axonal outgrowth from cultured dorsal root ganglion neurons (Streit and Stern, 1997). Mitogenic synergy has also been observed between HGF and insulin-like growth factor 1 (IGF-1) in cultured keratinocytes and mammary epithelial cells (WG Taylor and JS Rubin, personal communication). In addition to positive or cooperative interactions between HGF and other signaling pathways, evidence of negative modulation also exists. Activation of PKC can lead to the phosphorylation of c-Met on serine 985, and in turn, inhibition of the receptor's intrinsic tyrosine kinase activity (Gandino et al., 1994). IL-4 blocked the HGF-stimulated migration of colon Oncogene

carcinoma cells in a cell-matrix invasion assay (Uchiyama et al., 1996). In that study, IL-4 also inhibited HGF-stimulated expression of matrix metalloproteinases (MMP)-1, -2, and -9. Results from our own experiments (not shown), where HGF-stimulated cell migration was measured using a modi®ed Boyden chamber system (Day et al., 1999), indicated that IL-4 had no eect on HGF- or HGF/NK2-stimulated migration of 32D/c-Met cells. It is important to note that comparative analysis of HGF signaling in epithelial and myeloid cells suggests that while some intracellular eectors are common to both cell types, certain events leading to HGF-stimulated mitogenesis and migration in these two cell types are distinct (Day et al., 1999). Despite emerging evidence that modulation of HGF signaling may occur frequently in vivo, the intracellular mechanisms responsible remain to be elucidated. The results presented here suggest that IL-4 does not aect HGF-stimulated c-Met autophosphorylation, Shc phosphorylation, or the overall level of PI3K activation. IL-4 did enhance HGF-stimulated MAP kinase activation in 32D/c-Met cells, and consistent with this, the MEK1 inhibitor PD98059 partially blocked HGF/ IL-4 mitogenic synergy. In contrast, the mitogenic synergy between HGF and bFGF, dbcAMP, or MGF reported earlier was not accompanied by synergistic or additive increases in MAP kinase activity (Rubin et al., 1993). Interestingly, we found that the MEK1 inhibitor PD98059 had little signi®cant eect on DNA synthesis stimulated by HGF alone, suggesting that in myeloid cells such as 32D, MAP kinase activation might be mediated by other factors such as PKC, PLC-g and/or Src family kinases (Machide et al., 1998; van Dijk et al., 1997; Torigoe et al., 1992). In any event, the substantial reduction in HGF-stimulated MAP kinase activation associated with Wortmannin treatment suggests that the mediators of HGF-stimulated MAP kinase activation lie downstream of PI3K. To identify molecules downstream of PI3K that might mediate the observed mitogenic synergy between HGF and Il-4, we investigated the role of Src kinases in the IL-4 and HGF mitogenic pathways, since both factors are known to activate the Src kinase family members. We found that the Src inhibitor PP2 blocked approximately 50% of the HGF/IL-4-stimulated DNA synthesis above that observed in response to HGF alone, suggesting that a portion of the synergy was the result of Src-related kinase activity. Hematopoietic stem cell lines are known to express four of the eight known Src-family members; Lyn, Fyn, Yes and Hck (Torigoe et al., 1992). Of these kinases, Fyn (but not Lyn) has been shown to be activated by IL-4 in a human Burkitt lymphoma B-cell line, DND39 (Ikizawa et al., 1994). Indeed, c-Met is known to associate with c-Src and Fyn (Bardelli et al., 1992). Studies are underway to further de®ne the role of Src-kinase activation by IL-4 and HGF in mitogenic signaling using the 32D/c-Met model system. Through the activation of its receptor, and in turn, Jak kinases and STAT proteins, IL-4 can protect 32D


cells from apoptosis in a PI3K-independent manner (Wang et al., 1992, 1993; Zamorano et al., 1996). To investigate the role of these PI3K-independent events in the mitogenic synergy between HGF and IL-4, we compared ligand-stimulated DNA synthesis by parental 32D/c-Met cells with that of cells expressing dominant negative Jak mutants, and separately with cells overexpressing STAT6. We found that expression of a kinase-dead Jak1 mutant well in excess of endogenous Jak1 levels had no eect on the mitogenic synergy between HGF and Il-4 (data not shown). In contrast, similar relative expression levels of a mutant Jak3 protein completely blocked the ability of IL-4 to synergize with HGF. The Jak3 mutant also strongly suppressed IL-4-stimulated tyrosine phosphorylation of STAT6, providing a clear biochemical link to this intracellular eector. STAT6 activation is critical for IL-4-stimulated mitogenicity, which occurs in myeloid cells and ®broblasts that express PI3K binding proteins such as IRS-1 and IRS-2 (Patel et al., 1996; Wang et al., 1993). The absence of IRS family members in 32D cells, and thus a physical link between PI3K and the IL-4 receptor, is responsible for the failure of IL-4 to elicit mitogenicity in this cell line (Wang et al., 1992, 1993; Zamorano et al., 1996). In 32D/c-Met cells HGF stimulates PI3K activation, and IL-4-stimulated STAT6 activation may provide a complementary mitogenic signal that contributes to the observed synergy between these two factors. In support of this hypothesis, overexpression of STAT6 in 32D/c-Met cell lines resulted in enhanced STAT6 tyrosine phosphorylation and HGF/IL-4 mitogenic synergy, relative to the parental cell line. The mechanism by which IL-4stimulated STAT6 activation complements HGF signaling is not yet known. In 32D cells, mitogenic synergy between IGF-1 and IL-4 mediated by the Jak/ STAT6 pathway is associated with upregulation of cmyc expression (Soon et al., 1999). Others have shown that STAT proteins control lymphocyte proliferation by downregulating the expression of p27Kip1, an inhibitor of cell cycle progression (Kaplan et al., 1998). Clearly much more experimentation will be required to identify the targets of STAT6 signaling that contribute critically to the mitogenic synergy between HGF and IL-4. Recently, STAT3 was shown to be activated by HGF in human hepatocytes, hepatoma cells and MDCK epithelial cells (Boccaccio et al., 1998; Schaper et al., 1997). Blockade of HGF-stimulated STAT3 activation in epithelial cells resulted in the inhibition of tubulogenesis, but not HGF-stimulated cell motility or proliferation (60 Boccaccio et al., 1998). The time course of STAT3 DNA binding (5 ± 7 h after HGF stimulation) suggests that STAT3 may be activated as a result of de novo synthesis of signaling molecules (Schaper et al., 1997). In contrast, STAT6 activation by IL-4 and PDGF, which is critical to the mitogenic synergy between those factors observed in ®broblasts, occurs within 20 min and is not inhibited by the addition of cycloheximide (Patel et al., 1996). We did not detect

HGF-enhancement of IL-4-stimulated STAT6 activation, nor did we observe the activation of STAT6 or STAT3 in 32D/c-Met cells within 20 min of HGF treatment (Figure 8 and data not shown). Further studies are needed to determine whether HGF can stimulate de novo synthesis of STAT3 or STAT6 proteins that may, in turn, contribute to mitogenic signaling. Our results demonstrate mitogenic synergy between HGF and IL-4, two factors that are widely expressed throughout development and adulthood, each acting on a broad spectrum of target cell types, including hematopoietic cells. Investigation of the mechanism by which this synergy occurs revealed pathway convergence downstream of PI3K, which plays a central role in HGF-stimulated mitogenesis in both epithelial and hematopoietic cells, and upstream of MAP kinase activation. We present evidence that Src-related kinases contribute to the observed synergy between HGF and IL-4, but may not appear to be critical for DNA synthesis stimulated by HGF alone in myeloid cells. Independent of PI3K, we demonstrate that IL-4stimulated activation of Jak3 and in turn, STAT-6, also contribute to the observed synergy between this cytokine and HGF. These results suggest that the HGF and IL-4 signaling pathways converge at multiple levels: at least once upstream of MAP kinase activation, and further downstream at the level of cell cycle regulators under the transcriptional control of STAT6.

2209

Materials and methods Reagents, cell lines and cDNA transfections Full-length HGF was produced as described (Cioce et al., 1996). The truncated HGF isoform HGF/NK2, which contains the HGF N-terminal domain and ®rst two kringle domains, was produced as described (Stahl et al., 1997). HGF/NK2 is a potent motogen, but lacks the mitogenic and morphogenic activities of full-length HGF (Stahl et al., 1997; Chan et al., 1991; Montesano et al., 1998). Wortmannin (Sigma Chemical Company; 11.6 mM stock), LY294002 (Alexis; 20 mM stock), PD98059 (Calbiochem; 50 mM stock), and PP2 (Calbiochem; 17 mM stock) were dissolved in DMSO. The murine IL-3 dependent cell line 32D was cultured in RPMI 1640, 15% fetal bovine serum (FBS), and 5% WEHI-3B conditioned medium as a source of IL-3. 32D/c-Met cells co-expressing STAT6 or a Jak3 kinase-dead (Jak3kd) mutant were generated by transfection of 32D/cMet cells with STAT6 (Patel et al., 1996) or Jak3kd (Witthuhn et al., 1994) cDNA plasmids as described (Pierce et al., 1988). Stable cell lines were selected for expression of cMet and Stat6 by immunoblotting lysates of neomycinresistant clonal cell lines. Mitogenicity assays DNA synthesis was measured in 32D transfectants as described (Pierce et al., 1988). Brie¯y, 32D transfectants were incubated in RPMI, 15% FBS with or without growth factors at 26105 cells/ml in 24-well plates for 36 h. [3H]Thymidine (1 mCi/ml) was added for 3.5 h and cells were Oncogene


2210

collected on a ®lter using an automatic cell harvester (Skatron). [3H]-Thymidine incorporation was determined by scintillation counting. Detection of apoptosis and cell cycle analysis 32D cell transfectants were incubated at 26105 cells/ml in DMEM, 15% FBS with growth factor(s) for 36 h. Cells were washed twice in ice cold phosphate buered saline (PBS), and ®xed by the addition of cold ethanol (80% in water). One hour before FACS analysis, cells were pelleted and resuspended in 0.5 ml propidium iodide (PI) staining solution (0.05 mg/ml RNase A, 4.5% PI (w/v) in PBS). Cells were analysed using a FacScan and FacStation software (Becton Dickinson). Calculations of cell populations were determined using ModFit LT Cell Cycle Analysis software (Verity Software House). Immunoblot analysis Cells were extracted in lysis buer and cleared by centrifugation as described (Day et al., 1999). Equal amounts of protein were immunoprecipitated for 2 h at 48C with antiserum (1 : 100) and GammaBind Plus Sepharose (1 : 10) (Pharmacia Biotech). Proteins were separated by SDS ± PAGE, transferred to Immobilon-P (Millipore), blocked in PBS+5% BSA at 378C for 2 h, and probed with antibodies against human c-Met, ERK-1/2 (c-16; Santa Cruz Biotechnology), phosphotyrosine (anti-PY; 4G10; Upstate Biotechnology), phospho-p44/42 MAP kinase (New England Biolabs), Shc (Transduction Laboratories) or STAT6 (Santa Cruz Biotechnology). Immunoreactive bands were visualized using enhanced chemiluminescence (Amersham Life Science).

MAP kinase assays 32D transfectants were starved for 3 h in DMEM, treated with growth factors, and extracted in buered non-ionic detergent at 48C as described (Day et al., 1999). Lysates were immunoprecipitated with anti-ERK-1/2 and GammaBind Plus Sepharose for 2 h at 48C. The sepharose beads were processed for kinase reactions as described (Day et al., 1999). 32 P-phosphorylated MBP peptide was determined by liquid scintillation counting. Phosphatidylinositol 3-kinase assays Phosphatidylinositol 3-kinase (PI3K) assays were performed as described (Day et al., 1999; Wang et al., 1992). Cells were starved 4 h in DMEM prior to treatment with growth factors. Cells were then washed, lysed, and lysates were immunoprecipitated 4 h at 48C with GammaBind Plus Sepharose and anti-PY. Immunoprecipitates were washed three times and processed as described (Day et al., 1999). Kinase reaction products were subjected to thin layer chromatography, and visualized by autoradiography.

Acknowledgments We thank Drs Jacalyn Pierce, Jerey Rubin, William Taylor, William LaRochelle, and Yuichiro Suzuki for helpful discussions. We also thank Nelson Ellmore for technical assistance.

References Adams DH, Harvath L, Bottaro DP, Interrante R, Catalano G, Tanaka Y, Strain A, Hubscher SG and Shaw S. (1994). Proc. Natl. Acad. Sci. USA, 91, 7144 ± 7148. Alessi DR, Cuenda A, Cohen P, Dudley DT and Saltiel AR. (1995). J. Biol. Chem., 270, 27489 ± 27494. Balkovetz DF and Lipschutz JH. (1999). Int. Rev. Cytol., 186, 225 ± 260. Bardelli A, Maina F, Gout I, Fry MJ, Water®eld MD, Comoglio PM and Ponzetto C. (1992). Oncogene, 7, 1973 ± 1978. Birchmeier C and Gherardi E. (1998). Trends Cell. Biol., 8, 404 ± 410. Birchmeier W, Brinkmann V, Niemann C, Meiners S, DiCesare S, Naundorf H and Sachs M. (1997). Ciba Found Symp., 212, 230 ± 240. Boccaccio C, Ando M, Tamagnone L, Bardelli A, Michieli P, Battistini C and Comoglio PM. (1998). Nature, 391, 285 ± 287. Bon®ni L, Migliaccio E, Pelicci G, Lanfrancone L and Pelicci PG. (1996). TIBS, 21, 257 ± 261. Borset M, Seidel C, Hjorth-Hansen H, Waage A and Sundan A. (1999). Leuk. Lymphoma, 32, 249 ± 256. Chan AML, Rubin JS, Bottaro DP, Hirsch®eld DW, Chedid M and Aaronson SA. (1991). Science, 254, 1382 ± 1385. Cioce V, Czaky KG, Chan AML, Bottaro DP, Taylor WG, Jensen R, Aaronson SA and Rubin JS. (1996). J. Biol. Chem., 271, 13110 ± 13115. Cosman D. (1993). Cytokine, 5, 95 ± 106. Cowley S, Paterson H, Kemp P and Marshall CJ. (1994). Cell, 77, 841 ± 852. Oncogene

Day RM, Cioce V, Breckenridge D, Castagnino P and Bottaro DP. (1999). Oncogene, 18, 3399 ± 3406. Delaney B, Koh WS, Yang KH, Strom SC and Kaminski NE. (1993). Life Sci., 53, PL89 ± PL93. Gandino L, Longati P, Medico E, Prat M and Comoglio PM. (1994). J. Biol. Chem., 269, 1815 ± 1820. Gohda E, Nakamura S, Yamamoto I and Minowada J. (1995). Leuk. Lymphoma, 19, 197 ± 205. Graziani A, Gramaglia D, dalla Zonca P, Cantley LC and Comoglio PM. (1991). J. Biol. Chem., 266, 22087 ± 22090. Hanke JH, Gardner JP, Dow RL, Changelian PS, Brissette WH, Weringer EJ, Pollok BA and Connelly PA. (1996). J. Biol. Chem., 271, 695 ± 701. Hassuneh MR, Nagarkatti PS and Nagarkatti M. (1997). Blood, 89, 610 ± 620. Hoon DSB, Okun BE, Banez M, Irie FR and Morton DL. (1991). Cancer Res., 51, 5687 ± 5693. Iguchi T, Sogo S, Hisha H, Taketani S, Adachi Y, Miyazaki R, Ogata H, Masuda S, Sasaki R, Ito M, Fukuhara S and Ikehara S. (1999). Stem Cells, 17, 82 ± 91. Ikehara S. (1996). Leuk. Lymphoma, 20, 297 ± 303. Ikizawa K, Kajiwara K, Koshio T and Yanagihara Y. (1994). J. Allergy Clin. Immunol., 94, 620 ± 624. Kaplan MH, Daniel C, Schindler U and Grusby KJ. (1998). Mol. Cell. Biol., 18, 1996. Keegan AD, Nelms K, Wang L-M, Pierce JH and Paul WE. (1994). Immunol. Today, 15, 423 ± 432. Kmiecik TE, Keller JR, Rosen E and Vande Woude GF. (1992). Blood, 80, 2454 ± 2457.


Machide M, Kamitori K, Nakamura Y and Kohsaka S. (1998). J. Neurochem., 71, 592 ± 602. Mansour SJ, Matten WT, Hermann AS, Candia JM, Rong S, Fukasawa K, Vande Woude GF and Ahn NG. (1994). Science, 265, 966 ± 970. Michalopoulos GK and DeFrances MC. (1997). Science, 276, 60 ± 66. Montesano R, Soriano JV, Malinda KM, Ponce LM, Kleinman HK, Ba®co A, Bottaro DP and Aaronson SA. (1998). Cell. Growth Di., 9, 355 ± 365. Morisaki T, Yuzuki DH, Lin RT, Foshag LJ, Morton DL and Hoon DSB. (1992). Cancer Res., 52, 6059 ± 6065. Morisaki T, Uchiyama A, Yuzuki DH, Essner R, Morton DL and Hoon DSB. (1994). Cancer Res., 54, 1113 ± 1118. Patel BKR, Wang L-M, Lee C-C, Taylor WG, Pierce JH and LaRochelle WJ. (1996). J. Biol. Chem., 271, 22175 ± 22182. Paul WE. (1991). Blood, 77, 1859 ± 1870. Pierce JH, Ruggiero M, Fleming TP, Di Fiore PP, Greenberger JS, Varticovski L, Schlessinger J, Rovera G and Aaronson SA. (1988). Science, 239, 628 ± 631. Postlethwaite AE, Holness MA, Katai H and Raghow R. (1992). J. Clin. Invest., 90, 1479 ± 1485. Postlethwaite AE and Seyer JM. (1991). J. Clin. Invest., 87, 2147 ± 2152. Rahimi N, Tremblay E and Elliott B. (1996). J. Biol. Chem., 271, 24850 ± 24855. Rubin JS, Bottaro DP and Aaronson SA. (1993). Biochim. Biophys. Acta., 1155, 357 ± 371. Sakata H, Stahl SJ, Taylor WG, Rosenberg JM, Sakaguchi K, Wing®eld PT and Rubin JS. (1997). J. Biol. Chem., 272, 9457 ± 9463. Schwall RH, Chang LY, Kahn DW, Hillan KJ, Bauer KD and Zioncheck TF. (1996). J. Cell. Biol., 133, 709 ± 718. Schaper F, Siewert E, Gomez-Lechon MJ, Gatsios P, Sachs M, Birchmeier W, Heinrich PC and Castell J. (1997). FEBS Lett., 405, 99 ± 103. Soon L, Flechner L, Gutkind JS, Wang LH, Baserga R, Pierce JH and Li W. (1999). Mol. Cell. Biol., 38, 16 ± 28. Stahl SJ, Wing®eld PT, Kaufman JD, Pannell LK, Cioce V, Sakata H, Taylor WG, Rubin JS and Bottaro DP. (1997). Biochem. J., 326, 763 ± 772.

Streit AC and Stern CD. (1997). Ciba. Found. Symp., 212, 155 ± 165. Sunitha I, Shen R, McKillop IH, Lee JH, Resau J and Avigan M. (1999). Exp. Cell. Res., 250, 86 ± 98. Torigoe T, O'Connor R, Santoli D and Reed JC. (1992). Blood, 80, 617 ± 624. Uchiyama A, Essner R, Doi F, Nguyen T, Ramming KP, Nakamura T, Morton DL and Hoon DSB. (1996). J. Cell. Biochem., 62, 443 ± 453. Vande Woude GF, Jeers M, Cortner J, Alvord G, Tsarfaty I and Resau J. (1997). Ciba. Found. Symp., 212, 119 ± 130. van Dijk M, Muriana FJ, van Der Hoeven PC, de Widt J, Schaap D, Moolenaar WH and van Blitterswijk WJ. (1997). Biochem. J., 323, 693 ± 699. Wang L-M, Keegan AD, Paul WE, Heidaran MA, Gutkind JS and Pierce JH. (1992). EMBO J., 11, 4899 ± 4908. Wang L-M, Myers Jr MG, Sun X-J, Aaronson SA, White MF and Pierce JH. (1993). Science, 261, 1591 ± 1594. Weimar IS, de Jong D, Muller EJ, Nakamura T, van Gorp JM, de Gast GC and Gerritsen WR. (1997). Blood, 89, 990 ± 1000. Weimar IS, Miranda N, Muller EJ, Hekman A, Kerst JM, de Gast GC and Gerritsen WR. (1998). Exp. Hematol., 9, 885 ± 894. Witthuhn BA, Silvennoinen O, Miura O, Lai KS, Cwik C, Liu ET and Ihle JN. (1994). Nature, 370, 153 ± 157. Yanagita K, Matsumoto K, Sekiguchi K, Ishibashi H, Niho Y and Nakamura T. (1993). J. Biol. Chem., 268, 21212 ± 21217. Yashiro M, Chung Y-S, Inoue T, Nichimura S, Matsuoka T, Fukihara T and Sowa M. (1996). Int. J. Cancer, 67, 289 ± 293. Yu CZ, Hisha H, Li Y, Lian Z, Nishino T, Toki J, Adachi Y, Inaba M, Fan TX, Jin T, Iguchi T, Sogo S, Hosaka N, Song TH, Xing J and Ikehara S. (1998). Stem Cells, 16, 66 ± 77. Zamorano J, Wang HY, Wang L-M, Pierce JH and Keegan AD. (1996). J. Immunol., 157, 4926 ± 4934. Zarnegar R and Michalopoulos GK. (1995). J. Cell. Biol., 129, 1177 ± 1180.

2211

Oncogene