Oncogene (2002) 21, 3949 ± 3960 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc ORIGINAL PAPERS STAT3 activation is required for -6 induced transformation in tumor-promotion sensitive mouse skin epithelial cells

Cheng-Yong Yu1,6,7, Lihua Wang2,7, Alexander Khaletskiy5, William L Farrar2, Andrew Larner4, Nancy H Colburn3 and Jian Jian Li*,5

1Laboratory of Experimental Immunology, National Cancer Institute, National Institutes of Health, Frederick, Maryland, MD 21702, USA; 2Cytokine Molecular Mechanisms Section, Laboratory of Molecular Immunoregulation, National Cancer Institute, National Institutes of Health, Frederick, Maryland, MD 21702, USA; 3Laboratory of Basic Research, National Cancer Institute, National Institutes of Health, Frederick, Maryland, MD 21702, USA; 4Department of Immunology, Lerner Research Institute, The Cleveland Clinic Foundation, Cleveland, Ohio, OH 44106, USA; 5Radiation Biology, Division of Radiation Oncology, Beckman Research Institute, City of Hope National Medical Center, Duarte, California, CA 91010, USA

STAT3, a member of signal transducers and activators Oncogene (2002) 21, 3949 ± 3960. doi:10.1038/sj.onc. of transcription (STATs) originally discovered as 1205499 mediators in signaling pathways, plays an active role in oncogenesis. However, the function of Keywords: ; STAT3; transforma- STAT3 in signaling multistage carcinogenesis, especially tion response; matrix metalloproteinase; mouse skin in transformation of tumor-promotion sensitive epithelial cells cells has not been elucidated. The present study demonstrates that STAT3 is activated in interleukin-6 induced transformation in mouse skin epithelial cells. DNA binding and transcriptional activities of STAT3 Introduction were signi®cantly increased by interleukin-6. This induced anchorage-independent transformation in tumor- Carcinogenesis is a multistage process that involves promotion sensitive JB6 mouse skin P+ cells but not in multiple genetic and epigenetic events (Brown and the resistant variant P7 cells. Two forms of dominant Balmain, 1995; Balmain and Harris, 2000). These negative STAT3 (mutant of transcriptional domain, mF, events are divided into three independent stages: or DNA-binding domain, mD) were stably transfected initiation, transformation, and progression. The rate- into P+ cells. Activation of STAT3 was abolished and limiting events in multistage carcinogenesis are believed importantly, interleukin-6 induced anchorage-indepen- to occur during transformation and progression. Thus, dent growth was absent in both mutant STAT3 identifying the speci®c signaling pathways in the rate- transfectants. To determine the targeted by limiting step are important to prevent cell transforma- STAT3, three matrix metalloproteinase proteins linked tion as well as to inhibit tumor growth. Signal with carcinogenesis of epithelial cells were analysed. transducers and activators of transcription (STATs) Both basal and interleukin-6 induced expression of are a group of latent cytoplasmic transcription factors collagenase I and stromelysin I, but not gelatinase A, primarily found in cytokine-regulatory genes. As were inhibited in the mutant STAT3 transfectants. intracellular e€ectors of and Furthermore, transfection of a wild type STAT3 restored signaling pathways, STAT proteins function in the STAT3 transactivation and response to interleukin-6 control of cell proliferation, di€erentiation and apop- induced transformation in mutant STAT3 transfectants, tosis (Darnell et al., 1994; Danial et al., 1995; Yu et al., which up-regulated collagenase I and stromelysin I as 1995; Darnell, 1996, 1997; Bowman et al., 2000). As well. Together, these results provide the ®rst evidence such, binding of cytokines to their corresponding that STAT3 activation is required in the progression of receptors is able to trigger a cascade of signaling multistage carcinogenesis of mouse skin epithelial cells, pathways by activation of various protein tyrosine and matrix metalloproteinases are actively involved in kinases including Janus kinases (JAKs) (Finbloom and STAT3-mediated cell transformation. Larner, 1995; Darnell, 1997; Hirano et al., 2000). The activated JAKs by tyrosine phosphorylation then activate STATs, which are able to transactivate various *Correspondence: JJ Li, Halper South Building, H115, City of Hope cytokine-responsive genes through binding to the National Medical Center, 1500 Duarte Road, Duarte, CA 91010- gamma-activated (GAS) motifs in 3000, USA promoter regions to mediate their biological functions 6Current address: Molecular Immunology Section, Laboratory of (Zhong et al., 1994; Larner and Reich, 1996; Chow et Immunology, NEI, NIH, Bethesda, MD 20892, USA al., 2001). 7These authors have contributed equally to this work Received 15 January 2002; revised 28 February 2002; accepted 18 STAT signaling was ®rst identi®ed in v-src induced March 2002 oncogenesis (Yu et al., 1995). STAT3 mediated STAT3 and transformation response C-Y Yu et al 3950 signaling pathways are required in signaling the 6 that caused anchorage-independent transformation in transformation response induced by v-src oncoprotein, P+ but not P7 cells. Furthermore, expression of which can be blocked by a dominant negative dominant negative mutant STAT3 blocked IL-6- STAT3 (Turkson et al., 1998; Bromberg et al., 1999). induced STAT3 activity and abolished the transforma- Accumulating studies further supported the role of tion response. In addition, expression of wild type STATs in cell transformation and tumor formation STAT3 restored both STAT3 activity and anchorage (Danial et al., 1995; Besser et al., 1999; Demoulin et independent transformation. Finally, two matrix al., 2000). For instance, activation of STAT3 has been metalloproteinase genes are shown to be responsive detected in several human tumors including breast to STAT3 activation. cancer, multiple myeloma, lymphomas, and head and neck carcinoma (Weber-Nordt et al., 1996; Garcia et al., 1997; Grandis et al., 1998; Catlett-Falcone et al., Results 1999). Although di€erent STAT proteins are activated by several oncogenes (Migone et al., 1995; Garcia and IL-6 promotes JB6 P+ cell transformation Jove, 1998), STAT1a, STAT5, and STAT6 are found indirectly linked with malignant ®broblasts that are Although many cytokines are known to promote transformed by oncoproteins (Garcia et al., 1997). In tumor , it is not clear whether any single contrast, STAT3 has been proven to be an oncoprotein cytokine is able to initiate the progression of multistage itself (Bromberg, et al. 1999) and required speci®cally carcinogenesis. Using JB6 mouse skin epithelial cells, for v-src induced transformation (Yu et al., 1995; here we show that anchorage-independent transforma- Bromberg et al., 1998; Turkson and Jove, 2000). tion was induced by IL-6 (100 ng/ml) in tumor Additional evidence further suggests that STAT3 but promotion sensitive P+ cells but not in the resistant not STAT1 is required for EGF-induced cell growth variant P7 cells. About fourfold anchorage-indepen- (Grandis et al., 1998). Although STAT3 is proven to be dent transformation was induced in P+ cells compared a speci®c transcription factor in cell transformation, its to the level of control P+ cells without IL-6 signaling function as a rate-limiting event in the stimulation (Figure 1a,b). In contrast, no anchorage progression of epithelial cell transformation remains independent growth was induced in P7 cells. The to be elucidated. di€erent basal transformation levels of P+ and P7 IL-6 participates in the regulation of cell growth, cells without IL-6 exposure were the same as previously di€erentiation, and cell survival (Van Snick, 1990; Akira demonstrated (Li et al., 1996). and Kishimoto, 1992; Sui et al., 1999). IL-6 is able to function as a growth factor (Tosato et al., 1988) and its IL-6 induced DNA-binding of STAT3 in P+ cells soluble receptors are involved in signaling the growth of normal, human papillomavirus immortalized, and Next, we determined DNA-binding activity of STAT3 carcinoma derived cervical cells (Iglesias et al., 1995), by gel shift analysis with two radioactive-labeled which occurs via an autocrine mechanism (Tosato et al., oligonucleotides that bind to activated STAT1 and 1988; Kawano et al., 1989). IL-6 expression has also STAT3 (Oligo A) or to STAT6 (Oligo B) that was been con®rmed in multiple myeloma infected with included as a control. DNA binding was analysed with Kaposi's sarcoma-associated herpes virus (KSHV) the nuclear extracts of JB6 P+ and P7 cells after (Rettig et al., 1997). In addition, overexpression of the stimulation with IL-6 or EGF. Compared to control, growth-factor receptors ErbB2/neu and ErbB3 causes IL-6 exposure (100 ng/ml, 15 min) strongly induced neoplastic transformation and IL-6 increases tyrosine STAT3 but not STAT6 DNA-binding activity in P+ phosphorylation of ErbB2 and ErbB3 in prostate cancer cells (Figure 2a). STAT3 activation was also observed LNCaP cells (Qiu et al., 1998). Stimulation of cells by following exposure to IL-6 for 10 h. EGF (100 ng/ml) IL-6 induces gp130, a shared signal transduction did not induce STAT3 activation following 15 min or for a family of cytokines (Chow et al., 2001), 10 h exposure (Figure 2a,b). In addition, activation of which in turn activates protein tyrosine kinases and STATs was not detected in P7 cells following transcription factors (Akira, 1997; Qiu et al., 1998), treatment with IL-6 or EGF for 15 min or 10 h. To including STAT3 (Akira et al., 1993; Zhong et al., 1994; con®rm transcription function, STAT3 was analysed Yu et al., 1998). Any role of IL-6/STAT3 activation in using luciferase reporter controlled by the STAT3 signaling multistage carcinogenesis of epithelial cells is speci®c promoter as described in Materials and unknown. methods. Consistent with the DNA binding activity, To determine if the IL-6/STAT3 pathway acts as a a 12-fold activation of STAT3 was detected in the P+ key signaling event in tumor promotion, we studied the cells but not in P7 cells following IL-6 exposure cell transformation response in JB6 mouse epidermal (Figure 2c). cells with tumor promotion sensitive (P+) or insensi- AP-1 has been well documented in signaling cell tive (P7 ) genetic variants that have been intensively transformation especially induced by tumor promoter studied as an in vitro model for identi®cation of critical TPA (Bernstein and Colburn, 1989; Dong et al., 1994; genes during multistage carcinogenesis (Huang et al., Young et al., 1999). AP-1 also plays an active role in 1996; Dong et al., 1997). The present results demon- signaling the transformation of P+ cells by TPA strate that STAT3 but not STAT1 was activated by IL- (Bernstein and Colburn, 1989; Dong et al., 1994) and

Oncogene STAT3 and transformation response C-Y Yu et al 3951 IL-6 activated STAT3 but not STAT1 To gain further insight into whether STAT1 or STAT3 is speci®cally activated in IL-6 induced transformation, we determined STAT1 and STAT3 in the DNA binding complex induced by IL-6. Supershift analysis with antibodies to STAT1 and STAT3 (Figure 4, lanes 5 and 6) show that STAT1 antibody slightly shifted the lower portion of the STAT3 complex (Figure 4, lane 5). In contrast, STAT3 antibody completely shifted the DNA complex (Figure 4, lane 6). These results suggest that STAT3 is the major component of STATs in the STAT- DNA binding complex in IL-6 induced cell transforma- tion. NF-kB has been identi®ed as a major transcription factor responsive to cytokines, and subunits of NF-kB, mainly p65, interact with other transcription factors in P+ cell transformation (Li et al., 1997). We thus included antibodies to p65 and p50 in the gel shift analysis. Supershift results excluded the presence of p65 and p50 in the STAT3 DNA binding complex, thus suggesting that interaction between p65/p50 with STAT3 is unlikely (Figure 4, lanes 7 and 8). Therefore, in P+ cells, STAT3 appears to be a unique factor in signaling a cytokine induced transformation response.

Mutant STAT3 transfectants Stable transfectants into P+ cells of two mutant forms of STAT3 gene (mutants in DNA binding mD or transcription domain mF) were accomplished with LipofectAMINE mediated transfection followed by selection with the medium containing the relevant antibiotic. To examine whether the dominant negative Figure 1 IL-1 induced transformation in P+ but not P7 cells. (a) constructs of STAT3 are able to block IL-6-induced Anchorage-independent transformation. Cell transformation was STAT3 activation, the luciferase reporter of STAT3 measured in promotion sensitive P+ and resistant P7 JB6 cells was transfected into the mutant STAT3 and control using the anchorage-independent growth assay. Cells were collected using 0.033% trypsin, and suspended in 1.5 ml 7 0.33% agar vector transfectants. Compared to the vector control EMEM over 7 ml 0.5% agar medium containing IL-6 (100 ng/ml). cells, both dominant negative STAT3mF and Anchorage independent colonies were scored after incubation at STAT3mD dramatically inhibited IL-6-induced STAT3 368C for 14 days and results were normalized to the value of control activity (Figure 5a). The inhibition levels were slightly P+ cells without IL-6 treatment. Data shown are means and s.e. from three experiments (**P50.05). (b) Growth of anchorage- lower than the basal activity in mutant transfectants independent colonies in P+ and P7 cells with or without IL-6. The without cytokine stimulation. To con®rm the expres- conditions of anchorage-independent growth in soft agar culture sion of mutant STAT3 genes, IP7 Western analysis was the same as described above. Colonies were stained after 14 showed that transgene marker HA was expressed in the days for both P+ and P7 cells cultured with or without IL-6 mutant and control vector transfectants (Figure 5b). Compared to the vector control, the endogenous STAT3 protein was absent in the nucleus of both mutant STAT3 transfectants following IL-6 exposure (Figure 5c, lanes 2 ± 5). the progression of human keratinocytes induced by human papillomavirus (Li et al., 1998, 2000). To Anchorage independent transformation was abolished in determine if AP-1 is also activated in the IL-6 induced mutant STAT3 transfectants transformation response, DNA-binding activity of AP- 1 was measured with gel shift assays using nuclear To elucidate if STAT3 transactivation is required for proteins puri®ed from P+ and P7 cells with or signaling IL-6-induced transformation, we compared without IL-6 exposure (100 ng/ml, 15 min). No sig- the transformation response in mutant STAT3 and ni®cant AP-1 activation was measured in either DNA vector control transfectants of P+ cells. Compared to binding or transcription activity (Figure 3a,b). These the vector control, transformation response was absent results suggest that the signaling pathways induced by in the mutant STAT3F (mF) and STAT3D (mD) IL-6 are di€erent from those induced by the chemical (Figure 6). The basal anchorage independent transfor- tumor promoter TPA. mation was also lowered in both mutant transfectants

Oncogene STAT3 and transformation response C-Y Yu et al 3952

Figure 2 IL-6 but not EGF activated DNA binding of STAT3. P+ and P7 cells were cultured in 4% FBS EMEM and treated in EMEM with IL-6 (100 ng/ml) or EGF (100 ng/ml) for 15 min (a)or10h(b). DNA binding activity was measured with 3 mg nuclear extracts and 32P7 labeled STAT oligonucleotides. Oligonucleotide A (Oligo A) contains one binding site for STAT3 or STAT1, and oligonucleotide B (Oligo B) contains one binding site for STAT6 used as a negative control. Electrophoresis was performed on 6% acrylamide gel and the DNA-proteins complex was visualized by exposure to X-ray ®lm. (c) STAT3 luciferase reporter transactivation. STAT3 luciferase and b-galactosidase reporters were co-transfected in P+ and P7 cells and luciferase activity was determined 36 h following IL-6 exposure (100 ng/ml for 15 min). STAT3 transactivation was normalized to the b-galactosidase reporter transactivation and the activity without IL-6 treatment. The results shown are means and s.e.m. of three experiments (**P50.05)

(22 ± 28% compared to the vector control; data not regulation of cell growth (Matrisian, 1994). Collage- shown). These results are in agreement with reported nase I and stromelysin I were directly linked with gene data that transfection of the dominant negative STAT3 expression responses in transformation of human genes blocked IL-6-induced signal transduction. We keratinocytes (Li et al., 2000). Stromelysin I also further demonstrate here that blocking IL-6/STAT3 functions as an e€ector gene in signaling multistage pathways by these mutant STAT3 is able to inhibit the carcinogenesis (Lochter et al., 1998; Sternlicht et al., transformation response in mouse skin epithelial cells 2000). To determine if these matrix metalloproteinases (Shimozaki et al., 1997; Grandis et al., 1998). are activated in STAT3-mediated epithelial cell trans- formation, we measured transcription levels of three MMP genes in mutant STAT3 and the control vector Two MMP genes targeted by STAT3 activation transfectants. Increased expression and protein levels of Activation of matrix metalloproteinases is an impor- collagenase I were observed in the IL-6 treated vector tant step in cell transformation as well as in the control P+ cells (Figure 7a,b). In contrast, mRNA

Oncogene STAT3 and transformation response C-Y Yu et al 3953

Figure 4 STAT3 but not STAT1 was detected in the IL-6 induced STAT3/DNA complex. To determine components in the DNA-binding complex induced by IL-6, nuclear proteins were analysed with radioactive-labeled oligonucleotides and supershift was performed with the antibody speci®c to STAT1 or STAT3. Nuclear proteins were ®rst incubated with indicated antibodies followed by incubation with 36104 c.p.m. 32P-labeled oligonucleo- tides. DNA binding complex was separated in 6% acrylamide gel electrophoresis. Normal antiserum (nrs), antibody to p65 or p50, or 1006 cold oligonucleotides (control, ctr) were included as controls (one of three gel shift experiments)

Restoration of transformation response by transfection of wild type STAT3 Figure 3 IL-6 did not induce activation of AP-1. (a) Gel shift To test whether the transformation response is directly analysis was performed for DNA binding activity of AP-1 in P+ linked with STAT3 activity, we transfected wild type and P-cells with or without IL-6 treatment (100 ng/ml for 15 min). STAT3 gene into mF transfectants, and the response To make AP-1 binding in P-cells detectable, 3 mg nuclear proteins from P+ cells and 6 mg protein from P-cell were incubated with of anchorage independent transformation and STAT3 32P-labeled AP-1 oligonucleotides. The result shown is a activation were analysed in the mF+wild type STAT3 representative of four independent experiments. (b) AP-1 (mF+w) transfectants following IL-6 exposure. Inter- dependent luciferase activity. AP-1 activity was measured using estingly, the anchorage independent transformation AP-1 luciferase reporter co-transfected with b-galactosidase reporters. The luciferase activity was determined 36 h following was totally restored in the mF+w STAT3 transfec- IL-6 exposure (100 ng/ml for 15 min), and the results were tants in responding to IL-6 stimulation (Figure 8a). normalized to the b-galactosidase reporter transactivation. Results Actually, the transformation level was more elevated shown are means and s.e.m. of three experiments (**P50.05) than the basal level of vector control that showed a similar response compared to wild type P+ cells (Figure 8a compared to Figure 1a). STAT3 dependent levels in both basal and induced collagenase I was luciferase expression was also restored in the mF+w blocked in mutant mF and mD transfectants (Figure transfectants (Figure 8b). Importantly, compared to 7a). IL-6 induced protein levels were also inhibited in the mF cells, protein expression levels of collagenase I both mutant transfectants, although the mutant and stromelysin I that were down-regulated in both transfectants showed an elevated basal level of mutant STAT3 transfectants (Figure 7a,b) were re- collagenase I (Figure 7b). The basal and induced expressed in the mF+w transfectants with increased stromelysin I (mRNA and proteins) was also signi®- basal and induced levels following IL-6 exposure cantly down-regulated in the mutant STAT3 transfec- (Figure 8c). tants (Figure 7a,b), and in comparison, IL-6 induced signi®cant stromelysin I expression in both mRNA and protein levels in vector control transfectants. Interest- Discussion ingly, expression of another MMP gene, gelatinase A, was not inhibited in the mutant STAT3 transfectants. This study provides the ®rst evidence that activation of These results thus indicate that speci®c matrix STAT3 is required in signaling transformation of metalloproteinases responsive to IL-6/STAT3 signaling mouse epithelial cells in di€erent stages of multistage pathways are involved in the transformation induced in carcinogenesis. Using the JB6 cell transformation mouse epithelial cells. Potential functions of the model with sensitive or resistant phenotype to several STAT3 regulated matrix metalloproteinases in signal- tumor promoters (Bernstein and Colburn, 1989; Dong ing multistage carcinogenesis are worthy to be et al., 1994; Li et al., 1997), we show IL-6 induced elucidated. DNA binding and transcriptional activity of STAT3

Oncogene STAT3 and transformation response C-Y Yu et al 3954

Figure 5 Inhibition of STAT3 transactivation by mutant STAT3 transfection. (a) Two mutant forms of STAT3 gene inhibited IL- 6 induced STAT3 transactivation in P+ cells. Mutant STAT3 transfectants (mF and mD) and empty vector (Vector control) transfectants were grown in 24-well cell culture plates and were co-transfected with 0.3 mg STAT3-luciferase reporter DNA with 0.2 mg b-galactosidase reporters. Luciferase activity was measured with an equal volume of luciferase substrate, and results were normalized to b-galactosidase activity and expressed as the activity of cells without IL-6 treatment (data represent the means and s.e.m. of four experiments, (**P50.05). (b) Expression of transgene marker in STAT3 transfectants. HA marker was used to detect the mutant STAT3 expression by IP Western. Lane 1=P7 cells without transfection (-), lane 2=P+ cells without transfection (7), lane 3=mutant F transfectants (mF), lane 4=mutant D transfectants (mD), and lane 5=vector control transfectants (v). (c) Inhibition of STAT3 protein in the nucleus of mutant STAT3 transfectants. Nuclear proteins were puri®ed as described in Materials and methods and IP Western blotting was performed with 120 mg nuclear protein and the antibody to wild type STAT3. Lane 1=parental P+ cells without transfection, lanes 2 and 3=mutant STAT3F (F) and mutant STAT3D (D) transfectants respectively, lanes 4 and 5=vector control transfectants with or without IL-6 exposure (Non-speci®c bands are indicated as a loading control)

and anchorage independent transformation in tumor independent transformation induced in tumor promo- promotion sensitive P+ but not resistant P7 cells. In tion sensitive cells but not the resistant variant addition, mutant STAT3 genes blocked STAT3 indicates a -mediated STAT3 activation required activation and the transformation response induced only for cells at the tumor promotion ready-stage. This by IL-6 in P+ cells. The requirement of STAT3 was is an important ®nding for the approach of identifying further implicated by transfection of wild type STAT3 speci®c molecular targets at tumor promotion stage. that restored STAT3 activity and the transformation Several cytokine ligands such as IFN-a, IL-2, IL-6, IL- response in the mutant STAT3 transfectants. Two 7, IL-12 and IL-15, have been implicated in STAT3 matrix metalloproteinases, collagenase I and stromely- activation with a pattern of cell type dependence sin I, appeared to be actively regulated in STAT3 (Sadowski et al., 1993; Johnston et al., 1995; Stancato induced epithelial cell transformation. et al., 1996; Takeda et al., 1998). STAT3 is activated in The present results strongly suggest a di€erent human tumor samples and tumor cell lines, especially signaling pattern of ligand-mediated cell transforma- in oncogene-induced cell transformation (Bowman et tion between tumor promotion sensitive and resistant al., 2000). STAT3 activation is proven to be a key step mouse epithelial cells. The result of anchorage- for Src oncoprotein-mediated transformation, and this

Oncogene STAT3 and transformation response C-Y Yu et al 3955 pathway is independent of v-ras induced transforma- what function of STAT3 may have in initiating the tion (Bromberg et al., 1998). Expression of STAT3 has progression of carcinogenesis in cells with di€erent been shown to enhance the transforming potential of v- promotion stages. The present results suggest that src while three dominant negative STAT3 mutants ligand activated STAT3 participates in the progression inhibit the v-src induced transformation (Bromberg et of carcinogenesis in skin epithelial cells at tumor al., 1998; Bromberg, 1999). However, it is not clear promotion stage. Our results con®rm the varied response to the ligand-induced STAT3 activation between the tumor promotion sensitive P+ and resistant P7 cells. Therefore, in agreement with the reported results of anticancer treatment by blocking STAT3 activation (Catlett-Falcone et al., 1999; Bow- man et al., 2000), the ligand-mediated STAT3 activa- tion is worthy to be tested further as a key target event to block the progression of epithelial cell transforma- tion. The ligand-activated STAT3 is a necessary step for the transformation induced by inducible receptor and insulin-like growth factor I receptor (Zong et al., 1998). However, experiments with EGF receptor-speci®c inhibitor have indicated that activation of STAT3 is not dependent on ligand- receptor activation, although EGF is shown to further increase STAT3 activation (Garcia et al., 1997). These results suggest that activation of STAT3 can be ligand-dependent and independent, and both pathways Figure 6 Anchorage-independent transformation was inhibited can be activated during oncogenic transformation that in transfectants of two forms of mutant STAT3. P+ transfectants of mutant STAT3 genes as described in Materials and methods activates speci®c tyrosine kinases (Garcia et al., 1997). (mF or mD) and control vector, were cultured in complete Our present results show that IL-6 but not EGF medium. Anchorage-independent transformation was determined induced STAT3 activation in P+ but not P7 skin in soft agar medium containing 100 ng/ml of IL-6. Colonies epithelial cells (Figure 2), suggesting that IL-6 is a formed 14 days after being cultured in soft agar were scored and data was normalized to the value of vector control without IL-6. speci®c ligand required for STAT3 activation neces- The results represent the means and s.e.m. of three experiments sary for transformation of mouse skin epithelial cells. (**P50.05) The IL-6 induced transformation was, however, low

Figure 7 Expression of collagenase I and stromelysin I but not gelatinase A was down-regulated in mutant STAT3 transfectants. To determine the down stream genes in IL-6/STAT3 induced cell transformation, expression of matrix metalloproteinases collagenase I, stromelysin I, and gelatinase A were measured by RT ± PCR (a) and Western blot (b) in the mutant STAT3 (mF) and (mD) and vector control (V) transfectants with or without IL-6 treatment. GAPDH primers were internal controls for RT ± PCR. The RT ± PCR result shown is a representative of ®ve experiments and Western blot is one of two experiments

Oncogene STAT3 and transformation response C-Y Yu et al 3956

Figure 8 Restoration of transformation response and matrix metalloproteinases by transfection of wild type STAT3 gene. (a) Transformation response. Stable transfectant of a wild type STAT3 gene was derived from mF STAT3 transfectants and anchorage- independent transformation was determined in soft agar with or without IL-6. The results represent the means and s.e.m. of three experiments (**P50.01). (b) Luciferase reporter transcription of STAT3 in mF+wild type STAT3 (mF+w). The luciferase reporters of STAT3 and b-galactosidase reporter were co-transfected and luciferase activation was measured 24 h following IL-6 stimulation (represents the means and s.e.m. of three experiments, *P50.05). (c) Re-expression of STAT3 responsive matrix metalloproteinases. Protein levels of collagenase I and stromelysin I were measured by Western blotting in the parental mutant mF and mF+w transfectants with or without IL-6 treatment. The result shown is a representative of three experiments

compared with the e€ect of phorbol ester and TNF-a the rate-limiting event during multistage carcinogen- (Benedetti et al., 1990; Dong et al., 1994; Li et al., esis. 1996). In addition, IL-6 at lower concentrations did Activation of the endogenous IL-6 gene may be an not show transformation activity (Benedetti et al., essential step in multistage carcinogenesis of skin 1990). Therefore, a di€erent signaling pathway is epithelial cells. Although IL-6 is regulated by a required for STAT3 mediated transformation com- feedback control of JAK protein kinase (Lutticken et pared to that induced by phorbol ester. IL-6 induced al., 1994) and STAT3 (Akira, 1997), expression of IL-6 transformation did not require AP-1 transactivation is controlled by a variety of transcription factors necessary for cells transformed by TPA or by human including AP-1 and NF-kB (Dokter et al., 1996; Xia papilloma virus (Li et al., 1996, 2000), which supports et al., 1997). Ionizing radiation has been reported to the concept of a unique STAT3 signaling pathway. induce IL-6 expression via NF-kB activation (Beetz et However, similar to these tumor promoters (Bernstein al., 2000). Our previous results suggest that radiation and Colburn, 1989; Li, 1996; Dong et al., 1997), induces both AP-1 and NF-kB(Liet al., 2001), and transformation seen only in the sensitive P+ but not overexpression of manganese-containing superoxide resistant P7 cells suggests that ligand-activated dismutase (MnSOD) inhibits expression of the en- STAT3 may share a common signaling network with dogenous IL-6 via down-regulation of AP-1 and NF- other tumor promoters. Elucidating the shared kB(Liet al., 1998). Therefore, inhibition of tumor signaling pathway is important in an attempt to block growth (Li et al., 1995) and blockage of cell

Oncogene STAT3 and transformation response C-Y Yu et al 3957 transformation (Amstad et al., 1991) achieved by subset of matrix metalloproteinases are involved in the overexpressing the anticancer antioxidant enzyme signaling network causing tumor promotion in mouse MnSOD could be explained by the down-regulation skin epithelial cells. of IL-6/STAT3 pathways. Activation of STAT3 has been shown to target Cyclin D1, c-, and Bcl-x and blocking constitutive STAT3 Materials and methods expression results in growth inhibition and of STAT3-positive tumor cells in vitro and in vivo (Turkson Reagents and Jove, 2000). The present results indicate that matrix IL-6 and EGF were purchased from Pepro Tech, Inc. (Rocky metalloproteinases are the down-stream targets of Hill, NJ, USA). LipofectAMINE was from Life Technologies STAT3-mediated transformation in mouse skin epithe- Co. (Grand Island, NY, USA). [a-32P]ATP was from lial cells. Matrix metalloproteinases are a group of zinc- Amersham Life Science Co. (Arlington Heights, IL, USA). dependent endopeptidases that degrade most extracel- Luciferase assay substrate was from Promega (Madison, WI, lular matrix (Matrisian, 1994; Massova et al., 1998; USA). G418 sulfate was purchased from Life Technologies Nagase and Woessner, 1999). Matrix metalloproteinases Co. (Gaithersburg, MD, USA) and hygromycin B was from Calbiochem-Novabiochem Co. (Center Court, San Diego, have been indicated in the JAK-STAT pathways CA, USA). Antibodies to STAT1 and STAT3 were made in (Korzus et al., 1997) and v-src activation of the Lerner Research Institute, The Cleveland Clinic Foundation. collagenase I promoter is mediated via PEA3 and STAT Antibodies to p50, p65, HA, collagenase I (MMP-13), regulation (Vincenti et al., 1998). In addition, we have stromelysin I (MMP-3), and Gelatinase A (MMP-2) were reported that collagenase I and stromelysin I participate obtained from Santa Cruz Biotechnology Co. (Santa Cruz, in the transformation of human keratinocytes induced CA, USA). by human papillomavirus and v-fos genes (Li et al., 2000). Results have been reported that strongly support Plasmids the requirement of stromelysin I in signaling carcinogen- esis (Sternlicht et al., 2000). Most importantly, activa- Wild type HA-STAT3, mutant HA-STAT3F, HA-STAT3D, and control vectors were kindly provided by T Hirano tion of stromelysin is involved in signaling early stages of (Shimozaki et al., 1997). The STAT luciferase reporter tumor promotion and the generation of a malignant construct was made with the reporter vector pGL2-basic tumor phenotype (Lochter et al., 1998). Three matrix (Promega, Madison, WI, USA) containing three-repeats of metalloproteinases involved in epithelial cell transforma- the SIE GAS motif that eciently binds to STAT3 (Yu et al., tion and malignant tumor growth (Massova et al., 1998; 1998, 1999). AP-1 luciferase reporters were constructed as Nagase and Woessner, 1999) were detected in the described previously (Li et al., 1997). CMV-neo-selecting STAT3-mediated mouse skin transformation. Both plasmid (pBKCMV) and hygromycin-selecting plasmid basal and induced expression of collagenase I and (pCEP4) were obtained from Invitrogen Co. (San Diego, stromelysin I was blocked in the mutant F and D CA, USA). CMV-b-galactosidase plasmid was purchased STAT3 transfectants (Figure 7a,b). In addition, col- from Clontech Laboratories, Inc. (Palo Alto, CA, USA). lagenase I and stromelysin I down-regulated in both mutant transfectants were re-expressed at the protein Anchorage-independent transformation level when the STAT3 activity was restored by IL-6-induced cell transformation was analysed in tumor transfection of the wild type STAT3 into the mutant promotion sensitive P+ and resistant P7 cells using the STAT3 transfectant (Figure 8). It is important to note described anchorage-independent growth that has been that transformation response was also restored. These shown to be irreversible and to correlate with tumorigenicity results raise the possibility that the STAT3-mediated (Bernstein and Colburn, 1989). Brie¯y, exponentially growing activation of matrix metalloproteinase plays a key role cells were washed three times with PBS and trypsinized with in the transformation of epithelial cells. However, we did 0.033% trypsin and 0.13 mM EDTA. Cells were diluted and not ®nd di€erences in AP-1 activation between P+ and suspended in 1.5 ml 0.33% agar EMEM over 7 ml 0.5% agar medium containing IL-6 (100 ng/ml). The cultures were P7 cells, which excludes the possibility that modulation maintained at 368C and 1.0 ml fresh medium containing of matrix metalloproteinases is mediated by AP-1. 100 ng/ml IL-6 was added every 3 days. Anchorage Therefore, activation of both STAT3 and matrix independent colonies were scored after incubation at 368C metalloproteinases appears to be induced by a speci®c for 14 days. The anchorage independent growth in wild type pathway required for the transformation of skin cells and the eciency of inhibition in the mutant STAT3 epithelial cells. transfectants were compared to the level of wild type P+ In conclusion, IL-6 transformed tumor promotion cells with or without IL-6 treatment. sensitive mouse skin epidermal cells but not the resistant variant. STAT3 but not STAT1 was specially Reporter transfection and luciferase assay activated by IL-6, and expression of mutant STAT3 Transfection of luciferase reporters into JB6 mouse epidermal genes inhibited STAT3 activation and the anchorage cells followed the published procedure (Li, 1996; Li et al., independent transformation. In addition, reconstitution 1997). Both P+ and P7 cells were plated in 24-well culture of STAT3 by transfection of a wild type STAT3 plates and cultured in 4% EMEM at 368C for 18 h (70 ± 80% restored the transformation response and induced con¯uence) before transfection. To measure transcription expression of two matrix metalloproteinase genes factor activities, 16104 cells were co-transfected by Lipofec- inhibited by mutant STAT3. Therefore, STAT3 and a tAMINE using 0.3 mg luciferase reporter DNA and 0.2 mg b-

Oncogene STAT3 and transformation response C-Y Yu et al 3958 galactosidase reporters, followed by cytokine treatment. Cells 50 ml loading bu€er, heated at 708C for 10 min, size were washed with PBS and lysed with 200 ml luciferase assay separated in 12.5% acrylamide SDS ± PAGE, and transferred cell lysis bu€er (Promega, Madison, WI, USA). Luciferase to PVDF protein blot membrane (Millipure Co., Bedford, activity was read on a luminometer using 100 ml cell lysate MA, USA). The protein membrane was then blocked at and an equal volume of luciferase substrate (Moonlight 2010, room temperature for 2 h in blocking solution (Pierce Co., Analytical Luminescence Laboratory, San Diego, CA, USA). Rockford, IL, USA), washed with 0.01% Tween PBS, and incubated overnight at 48C at 1 : 300 dilution of antibody. The membrane was visualized with the ECL Western blot Nuclear protein analysis detection solution (Amersham Co., Arlington Heights, IL, Nuclear proteins were prepared following published methods USA). (Yu et al., 1996). JB6 cells (56107 cells) collected after trypsinization were washed three times with PBS and RT ± PCR resuspended in 1 ml cell lysis bu€er. The lysis mixture was then centrifuged at 48C for 1 min and the pellets were washed Total RNA was extracted by using TRIzol Reagent (Life with 0.3*0.5 ml washing bu€er once by centrifuging at 48C Technologies, Inc., Gaithersburg, MD, USA). After con- for 1 min. The cell nuclei were then resuspended in 50 ± ®rmation of integrity on an agarose gel, total RNA was 100 ml nuclear protein extraction bu€er, centrifuged at 48C digested with RNase-free DNase I for 20 min, extracted using for 5 min, and supernatants were saved as nuclear proteins. phenol-chloroform, and then precipitated with 2.5 volume The SIE GAS-like motif containing oligonucleotide: 5'- ethanol. cDNA was synthesized in a reaction volume of 20 ml AGCTTGTCGACATTTCCCGTAAATCGTCGG-3' and Ce bu€er containing 0.1 mg total RNA, 100 pmol of random GAS-like motif containing oligonucleotide: 5'-AGTCAA- hexadeoxynucleotides (Promega Co., Madison, WI, USA) GACCTTTTCCCAAGAAATCTATC-3' were used to and 200 units moloney murine leukemia virus reverse detect the STAT/DNA-binding activity (Yu et al., 1996, transcriptase (MMLV-RT, Gibco/BRL, Gaithersburg, MD, 1998). Oligonucleotides containing the motif for control USA). For PCR reactions, 5 ml cDNA products were mixed transcription factor AP-1: 5'-AGCATGAGTCAGACACCT- in reaction bu€er (Promega) containing 2.0 units Taq DNA CTGGC-3' were purchased from Promega. Gel shift analysis polymerase (Promega) and 0.1 mM of following primers for was performed with 1 ± 3 mg nuclear proteins in 20 ml DNA MMP genes designed with Genbank database: Collagenase I binding bu€er. The reaction mixture was incubated on ice for (MMP-13; Genbank #: NM_008607; 349 bp fragment): 10 min and then further incubated for 20 min at room AGGCCTTCAGAAAAGCCTTC (forward); GAAATGGC- temperature after adding 56104 c.p.m. 32P-labeled oligonu- TTTTGCCAGTGT (backward); Stromelysin I (MMP-3; cleotide probe. The DNA-protein complexes were electro- Genbank #: NM_010809; 407 bp fragment): TCAGTAC- phoresed for 2 h at room temperature in nondenaturing CTTCCCAGGTTCG (forward), AGCCTTGGCTGAGT- acrylamide gel. For supershift assays, nuclear extracts were GGTAGA (backward); Gelatinase A (MMP-2; Genbank #: incubated with antibodies at room temperature for 20 min 008610; 406 bp fragment): GCACTCTGGAGCGAGGA- and then reacted with radioactive-labeled DNA oligonucleo- TAC, CACATCCTTCACCTGGTGTG (backward). Frag- tide for another 20 min. The supershift complexes were ments were enhanced by 25 ± 35 cycles. detected by non-denaturing acrylamide gel.

Establishment of the STAT mutant cell lines Abbreviations JB6 P+ cells with 60% con¯uence in 100 mm dishes were AP-1,activatorprotein1;EGF,epithelialgrowthfactor; transfected using LipofectAMINE with 30 mg mutant STAT3 GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IL- plasmids HA-STAT3F and HA-STAT3D together with 2 mg 6, Interleukin-6; , Jak; PKC, protein kinase C; hygromycin-selecting plasmid pCEP4. Selection for 10 ± 14 MMPs, matrix metalloproteinases; Manganese-containing days with 10 ± 20 mg/ml hygromycin produced no colonies in superoxide dismutase, MnSOD; NF-kB, nuclear factor the control P+ cells without transfection while producing kappaB; STATs, signal transducers and activators of 20*30 clones in the HA-mutant and HA-vector controls. transcription; TNF, tumor necrosis factor; TPA, 12-O- The selected cells were pooled and cultured for at least ®ve tetradecanoylphorbol-13-acetate passages with 10 mg/ml hygromycin, and experiments were performed with transfected cells free of hygromycin. Response to IL-6-induced transformation was tested by anchorage-independent growth in soft agar. To establish Acknowledgments stable transfectants of wild type STAT3 cell lines, HA- We thank Dr JE Darnell at Rockefeller University for STAT3 vector was co-transfected with pcDNA3 plasmids critical comments and suggestions. We are grateful to Dr T into HA-STAT3F (MutantF) transfectants and stable Hirano at NIH for providing the wild type HA-STAT3, transfectants were obtained by G418 selection. Experiments mutated HA-STAT3F and HA-STAT3D, and control were performed in medium without selecting drugs. vectors; Drs J Shively and T LeBon at City of Hope for insightful reading and editorial help; and Mrs V Boore for assistance in preparing the manuscript. By acceptance of Immunoprecipitation and Western blotting this article, the publisher or recipient acknowledges the For immunoprecipitations, whole cell lysate (120 mg total right of the US Government to retain a non-exclusive, protein) was precleaned with 50 ml agarose beads conjugated royalty-free license in and to any copyright covering the with nonspeci®c rabbit IgG (Sigma Co.) and reacted with article. The contents of this publication do not necessarily 5 mg/ml STAT3 antibodies. Reactions were gently agitated re¯ect the views or policies of the Department of Health overnight at 48C and bound complexes were pelleted by and Human Services, nor does mention of trade names, centrifuge at 48C for 10 min followed by washing three times commercial products, or organizations imply endorsement with 0.01% Tween PBS. The bound proteins were mixed with by the US Government.

Oncogene STAT3 and transformation response C-Y Yu et al 3959 References

Akira S. (1997). Int. J. Biochem. Cell Biol., 29, 1401 ± 1418. Johnston JA, Bacon CM, Finbloom DS, Rees RC, Kaplan Akira S and Kishimoto T. (1992). Immunol. Rev., 127, 25 ± D, Shibuya K, Ortaldo JR, Gupta S, Chen YQ, Giri JD 50. and O'Shea JJ. (1995). Proc. Natl. Acad. Sci. USA, 92, Akira S, Taga TK and Ishimoto T. (1993). Adv. Immunol., 8705 ± 8709. 54, 1±78. Kawano M, Kuramoto A, Hirano T and Kishimoto T. Amstad P, Peskin A, Shah G, Mirault ME, Moret R, (1989). Cancer Surv., 8, 905 ± 919. Zbinden I and Cerutti P. (1991). Biochemistry, 30, 9305 ± Korzus E, Nagase H, Rydell R and Travis J. (1997). J. Biol. 9313. Chem., 272, 1188 ± 1196. Balmain A and Harris CC. (2000). Carcinogenesis, 21, 371 ± Larner A and Reich NC. (1996). Biotherapy, 8, 175 ± 181. 377. Li JJ, Dong Z, Dawson MI and Colburn NH. (1996). Cancer Beetz A, Peter RU, Oppel T, Ka€enberger W, Rupec RA, Res., 56, 483 ± 489. Meyer M, van Beuningen D, Kind P and Messer G. (2000). Li JJ, Cao Y, Young MR and Colburn NH. (2000). Mol. Int. J. Radiat. Biol., 76, 1443 ± 1453. Carcinog., 29, 159 ± 169. Benedetti FD, Colburn NH, Oppenheim JJ and Faltynek Li JJ, Oberley LW, St Clair DK, Ridnour LA and Oberley CR. (1990). Mol. Cell. Biol. Cytokines, 1, 275 ± 280. TD. (1995). Oncogene, 10, 1989 ± 2000. Bernstein LR and Colburn NH. (1989). Science, 244, 566 ± Li JJ, Rhim JS, Schlegel R, Vousden KH and Colburn NH. 569. (1998). Oncogene, 16, 2711 ± 2721. Besser D, Bromberg JF, Darnell Jr JE and Hanafusa H. Li JJ, Westergaard C, Ghosh P and Colburn NH. (1997). (1999). Mol. Cell. Biol., 19, 1401 ± 1409. Cancer Res., 57, 3569 ± 3576. Bowman T, Garcia R, Turkson J and Jove R. (2000). LiZ,XiaL,LeeLM,KhaletskiyA,WangJ,WongJYCand Oncogene, 19, 2474 ± 2488. Li JJ. (2001). Rad. Res., 155, 543 ± 553. Bromberg JF, Wrzeszynska MH, Devgan G, Zhao Y, Pestell Lochter A, Sternlicht MD, Werb Z and Bissell MJ. (1998). RG, Albanese C and Darnell Jr JE. (1999). Cell, 98, 295 ± Ann. N.Y. Acad. Sci., 857, 180 ± 193. 303. Lutticken C, Wegenka, UM, Yuan J, Buschmann J, Bromberg JF, Horvath CM, Besser D, Lathem WW and Schindler C, Ziemiecki A, Harpur AG, Wilks AF, Darnell Jr JE. (1998). Mol. Cell. Biol., 18, 2553 ± 2558. Yasukawa K, Taga T, Kishimoto T, Barbieri G, Pellegrini Brown K and Balmain A. (1995). Cancer Metastasis Rev., 14, S, Sendtner M, Heinrich PC and Horn F. (1994). Science, 113 ± 124. 263, 89 ± 92. Catlett-Falcone R, Dalton WS and Jove R. (1999). Curr. Massova I, Kotra LP, Fridman R and Mobashery S. (1998). Opin. Oncol., 11, 490 ± 496. FASEB J., 12, 1075 ± 1095. Chow D, He X, Snow AL, Rose-John S and Garcia KC. Matrisian LM. (1994). Ann. N.Y. Acad. Sci., 732, 42 ± 50. (2001). Science, 291, 2150 ± 2155. Migone TS, Lin JX, Cereseto A, Mulloy JC, O'Shea JJ, Danial NN, Pernis A and Rothman PB. (1995). Science, 269, Franchini G and Leonard WJ. (1995). Science, 269, 79 ± 1875 ± 1877. 81. Darnell Jr JE. (1996). Recent Prog. Horm. Res., 51, 391 ± 403. Nagase H and Woessner Jr JF. (1999). J. Biol. Chem., 274, Darnell Jr JE. (1997). Science, 277, 1630 ± 1635. 21491 ± 21494. Darnell Jr JE, Kerr IM and Stark GR. (1994). Science, 264, Rettig MB, Ma HJ, Vescio RA, Pold M, Schiller G, Belson 1415 ± 1421. D, Savage A, Nishikubo C, Wu C, Fraser J, Said JW and Demoulin JB, Uyttenhove C, Lejeune D, Mui A, Groner B Berenson JR. (1997). Science, 276, 1851 ± 1854. and Renauld JC. (2000). Cancer Res., 60, 3971 ± 3977. Sadowski HB, Shuai K, Darnell Jr JE and Gilman MZ. Dokter WHA, Koopmans SB and Vellenga E. (1996). (1993). Science, 261, 1739 ± 1744. Leukemia, 10, 1308 ± 1316. Shimozaki K, Nakajima K, Hirano T and Nagata S. (1997). Dong Z, Birrer MJ, Watts RG, Matrisian LM and Colburn J. Biol. Chem., 272, 25184 ± 25189. NH. (1994). Proc. Natl. Acad. Sci. USA, 91, 609 ± 613. Stancato LF, David M, Carter-Su C, Larner AC and Pratt Dong Z, Crawford HC, Lavrosky V, Taub D, Watts R, WB. (1996). J. Biol. Chem., 271, 4134 ± 4137. Matrisian LM and Colburn NH. (1997). Mol. Carcinog., Sternlicht MD, Bissell MJ and Werb Z. (2000). Oncogene, 19, 19, 204 ± 212. 1102 ± 1113. Finbloom DS and Larner AC. (1995). Cell Signal, 7, 739 ± SiuX,TsujiK,EbiharaY,TanakaR,MuraokaK,Yoshida 745. M, Yamada K, Yasukawa K, Taga T, Kishimoto T and Garcia R and Jove R. (1998). J. Biomed. Sci., 5, 79 ± 85. Nakahata T. (1999). Blood, 93, 2525 ± 2532. Garcia R, Yu CL, Hudnall A, Catlett R, Nelson KL, Takeda K, Kaisho T, Yoshida N, Takeda J, Kishimoto T and Smithgall T, Fujita DJ, Ethier SP and Jove R. (1997). Cell Akira S. (1998). J. Immunol., 161, 4652 ± 4660. Growth Di€er., 8, 1267 ± 1276. Tosato G, Seamon KB, Goldman ND, Sehgal PB, May LT, Grandis JR, Drenning SD, Chakraborty A, Zhou MY, Zeng Washington GC, Jones KD and Pike SE. (1988). Science, Q, Pitt AS and Tweardy DJ. (1998). J. Clin. Invest., 102, 239, 502 ± 504. 1385 ± 1392. Turkson J, Bowman T, Garcia R, Caldenhoven E, De Groot Hirano T, Ishihara K and Hibi M. (2000). Oncogene, 19, RP and Jove R. (1998). Mol. Cell. Biol., 18, 2545 ± 2552. 2548 ± 2556. Turkson J and Jove R. (2000). Oncogene, 19, 6613 ± 6626. Huang C, Ma WY and Dong Z. (1996). Mol. Cell. Biol., 16, Van Snick J. (1990). Annu.Rev.Immunol.,8, 253 ± 278. 6427 ± 6435. Vincenti MP, Schroen DJ, Coon CI and Brinckerho€ CE. Iglesias M, Plowman GD and Woodworth CD. (1995). Am. (1998). Mol. Carcinog., 21, 194 ± 204. J. Pathol., 146, 944 ± 952.

Oncogene STAT3 and transformation response C-Y Yu et al 3960 Weber-Nordt RM, Riley JK, Greenlund AC, Moore KW, Yu CR, Lin JX, Fink DW, Akira S, Bloom ET and Darnell JE and Schreiber RD. (1996). J. Biol. Chem., 271, Yamauchi A. (1996). J. Immunol., 157, 126 ± 137. 27954 ± 27961. Yu CR, Ortaldo JR, Curiel RE, Young HA, Anderson SK Xia C, Cheshire JK, Patel H and Woo P. (1997). Int. J. and Gosselin P. (1999). J. Immunol., 162, 2785 ± 2790. Biochem. Cell. Biol., 29, 1525 ± 1539. Yu CR, Young HA and Ortaldo JR. (1998). J. Leukoc. Biol., Young MR, Li JJ, Rincon M, Flavell RA, Sathyanarayana 64, 245 ± 258. BK, Hunziker R and Colburn NH. (1999). Proc. Natl. Zhong Z, Wen Z and Darnell Jr JE. (1994). Science, 264, 95 ± Acad. Sci. USA, 96, 9827 ± 9832. 98. YuCL,MeyerDJ,CampbellGS,LarnerAC,Carter-SuC, Zong CS, Zeng L, Jiang Y, Sadowski HB and Wang LH. Schwartz J and Jove R. (1995). Science, 269, 81 ± 83. (1998). J. Biol. Chem., 273, 28065 ± 28072.

Oncogene