Loss of Oncostatin M Receptor B in Metastatic Melanoma Cells

A Lacreusette1,2, J-M Nguyen3, M-C Pandolﬁno1,2, A Khammari4, BDreno 4, Y Jacques1,2, A Godard1,2,5 and F Blanchard1,2,6

1INSERM, U601, Groupe de Recherche Cytokines et Re´cepteurs, Institut de Biologie, Nantes, France; 2Universite´ de Nantes, UFR Me´decine, IFR26, Institut de Biologie, Nantes, France; 3PIMESP, Nantes, France; 4Unit of Skin Cancer, CHU de Nantes, Nantes, France and 5Laboratoire de Biochimie, CHU de Nantes, Nantes, France

Oncostatin M (OSM) is an interleukin-6 (IL-6) type feron a (IFNa) is usually given but has only a modest cytokine originally described by its capacity to inhibit success rate (Chudnovsky et al., 2005). Therefore, melanoma proliferation in vitro. Here, the mechanisms clinical trials have evaluated immunotherapeutic ap- involved in resistance to growth inhibition byOSM were proaches based on (i) vaccination, where melanoma or analysed for the first time on a large panel of metastatic dendritic cells are modified to express HLA or tumor melanoma cell lines. OSM resistance did not strictly antigens, costimulatory molecules or cytokines, and (ii) correlate with IL-6, interferon-c or tumor necrosis factor- adoptive therapy using tumor-infiltrating lymphocytes a resistance. Rather, it correlated with a specific loss of (TILs) (Nawrocki et al., 2000; Yannelli and Wroblewski, the OSM receptor-b (OSMRb) subunit, in conjunction 2004). These clinical trials gave encouraging results, with a lower level of histone acetylation in the OSMRb but many patients did not respond because of potent promoter region. Treatment of various OSM-resistant tumor escape from immune recognition (Yannelli and melanoma cells with the histone deacetylase inhibitor Wroblewski, 2004). Trichostatin A increased activityand histone acetylation A large number of cytokines have proven to be of the OSMRb promoter as well as expression of OSMRb effective in restricting melanoma progression in vivo, mRNA and protein, allowing OSM to activate the signal such as interleukin (IL)-2, IL-12, IFNa and IFNg, transducer and activator of transcription 3 (STAT3) and TNFa family members and IL-6 family members (Mule to inhibit proliferation. Other defects associated with et al., 1990; Nawrocki et al., 2000; Ozbek et al., 2001; OSM resistance were identified at the level of OSMRb Yannelli and Wroblewski, 2004). Tumor rejection and transcription or protein expression, as well as downstream prevention of metastasis were also demonstrated in mice of or parallel to STAT3 activation. Altogether, our results sarcoma models with systemic IL-6 administration suggest a role for OSM in the prevention of melanoma (Mule et al., 1990). Similarly, transfer of the IL-6 and progression and that metastatic melanoma cells could IL-6 receptor-a gene into melanoma cells led to a lower escape this growth control bythe epigenetic silencing of rate of tumor growth in mice and enhanced survival OSMRb. (Nawrocki et al., 2000; Ozbek et al., 2001). In vitro, Oncogene (2007) 26, 881–892. doi:10.1038/sj.onc.1209844; melanocytes and early-stage (radial growth phase) published online 7 August 2006 melanoma cell lines are inhibited in their growth by IFNa and IFNg, TNFa and IL-6-type cytokines, Keywords: oncostatin M; interleukin-6; STAT3; histone whereas cell lines from advanced-stage lesions are often acetylation; melanoma resistant to these cytokines, suggesting an acquired ‘multicytokine resistance’ during melanoma progression (Lu et al., 1993; Bani et al., 1996). Oncostatin M (OSM) is an IL-6-type cytokine, Introduction produced mainly by activated monocytes and lymphocytes, and first identified by its capacity to inhibit Melanoma is an aggressive malignancy with poor melanoma cell growth in vitro (Zarling et al., 1986). prognosis owing to resistance to chemotherapy. OSM is now considered as a multifunctional cytokine Although the treatment of primary melanoma essen- implicated in the activation, proliferation and/or differ- tially relies on surgery, at the stage of lymph node entiation of several cell types, such as hepatocytes, invasion (stage III), an adjuvant treatment with inter- osteoblasts or lung epithelial cells (Grant and Begley, 1999; Tanaka and Miyajima, 2003). OSM is more active Correspondence: Dr A Godard or Dr F Blanchard, INSERM, U601, than IL-6 in inhibiting the proliferation of numerous Groupe de Recherche Cytokines et Re´ cepteurs, Institut de Biologie, 9 solid tumor cell lines derived from breast or lung cancer, Quai Moncousu, 44035 Nantes cedex 1, France. hepatoma, osteosarcoma or melanoma (Grant and E-mails: [email protected]; [email protected] Begley, 1999). However, relatively few studies have 6Current address: INSERM, ERI7, 1 rue Gaston Veil, 44035 Nantes cedex 1, France. focused on the antitumor effects of OSM, and the Received 28 March 2006; revised 6 June 2006; accepted 26 June 2006; mechanisms regulating such activities are only poorly published online 7 August 2006 understood. OSM receptor expression in melanoma A Lacreusette et al 882 All IL-6-type cytokines share the signal transducing the proliferation of metastatic melanoma cell lines receptor subunit gp130 and receptor specificity is obtained from the lymph nodes of 22 patients (stage provided by additional receptor chains: IL-6Ra for III) who were enrolled in a previously described phase IL-6 and OSM receptor (OSMR)b for OSM (Heinrich II/III randomized clinical trial (Dreno et al., 2002; et al., 2003; Tanaka and Miyajima, 2003). OSM is most Labarriere et al., 2002). The WM-266-4 metastatic cell closely related to another IL-6-type cytokine, leukemia line was used as a positive control. Preliminary inhibitory factor (LIF), and both cytokines can trans- experiments indicated that a maximum growth inhibi- duce a signal via the LIF receptor composed of gp130 tion was obtained with 50 ng/ml OSM for 3 days on and LIFRa. Once recruited, these receptor complexes WM-266-4 and M140 cells (IC50 values of 1 ng/ml; data allow activation of Janus protein tyrosine kinases (Jak), not shown). Therefore, these conditions were selected and subsequently signal transducer and activator of for further analyses. As shown in Figure 1a and Table 1, transcription (STAT) (mainly STAT3) and MAPK OSM appeared to be a potent growth inhibitor, with 15 (mainly extracellular signal-regulated kinase (ERK) cell lines (65%) being inhibited by more than 15% 1/2) (Klausen et al., 2000; Heinrich et al., 2003; Tanaka (range 18–92%) and eight cell lines (35%) being and Miyajima, 2003). In melanoma or other solid resistant to OSM (less than 15% inhibition, not tumors, the key role of STAT3 in mediating the growth statistically significant). In contrast, our analysis on inhibitory effect of IL-6 or OSM has been demonstrated primary melanoma cells (seven cell lines) indicated that (Kortylewski et al., 1999), whereas activation of ERK they were all sensitive to growth inhibition by OSM 1/2 could counterbalance that of STAT3 by inducing (Heymann et al., 1995). mitogenic signals (Kim and Baumann, 1999). OSM was about twofold more active than IL-6 in In metastatic melanoma, IL-6 resistance seems to be reducing thymidine incorporation by metastatic mela- largely a consequence of an altered STAT signaling or noma cells (Table 1), and growth inhibition by OSM STAT-induced gene transcription. Thus, in various IL- was significantly correlated with growth inhibition by 6-resistant melanoma cell lines, IL-6 is still able to IL-6 (Po0.001). However, 2/8 cell lines (M134 and activate its receptor as well as STAT3, but it fails to M170) were resistant to OSM but not to IL-6 induce the cyclin-dependent kinase inhibitor p21WAF1,a (Figure 1a) and conversely, some cell lines were resistant STAT3 gene target implicated in growth inhibition by to IL-6 (less than 15% inhibition) but not to OSM (see IL-6 (Florenes et al., 1999). One study showed that the M197 or M203 cell lines in Table 1). LIF never melanoma cells derived from advanced-stage lesions are showed any effect on the growth of these cells (data not often resistant to growth inhibition by OSM and that shown). Growth inhibition by IFNg or TNFa did not OSM resistance does not always correlate with IL-6 correlate with that observed after OSM treatment resistance (Lu et al., 1993), suggesting a defect in an (Figure 1a and Table 1). In fact, all OSM-resistant cell OSM-specific pathway. We previously showed that lines were growth-inhibited by either IFNg or TNFa or responsiveness to IL-6-type cytokines can be increased both. Therefore, the mechanisms involved in OSM by histone deacetylase (HDAC) or DNA methyltrans- resistance appeared specific, suggesting defects in its ferase inhibitors (Blanchard et al., 2002; Blanchard receptor or signaling pathways. et al., 2003). Indeed, these inhibitors can induce expression of several epigenetically silenced tumor suppressor genes, resulting in cell cycle arrest and OSM resistance is associated with the loss of OSMRb differentiation (Jaenisch and Bird, 2003; Blanchard and STAT3signaling and Chipoy, 2005). Our data indicated that the Melanoma cell lines were then analysed for the early promoters for gp130 or LIFRa can be epigenetically phosphorylation/activation of STAT3 and ERK 1/2 silenced in normal or transformed cells, concomitant induced by short-term treatment with IL-6-type cyto- with a defect in receptor expression and STAT signaling kines (Figure 1b–d and Table 1). STAT3 and ERK 1/2 (Blanchard et al., 2002; Blanchard et al., 2003). were expressed in all cell lines tested (Figure 1b and data Here, we analysed 22 metastatic melanoma cell lines not shown). Activation of STAT3 and ERK 1/2 by for growth inhibition by IFNg, TNFa and IL-6 family OSM was always stronger than by IL-6 or LIF (see the members. Resistance to OSM was not always associated M140 cell line in Figure 1b as a representative example). with resistance to IL-6, IFNg or TNFa. Rather, it Growth inhibition by OSM or IL-6 was always correlated with a loss of OSMRb expression, associated associated with a corresponding activation of STAT3 with histone deacetylation in the OSMRb gene promo- (Table 1). In sharp contrast, OSM resistance was almost ter region. always associated with a loss of STAT3 activation by OSM (in seven out of eight cell lines; see the M128 and M170 cell lines in Figure 1b as representative examples), whereas IL-6 resistance was associated with a loss of Results STAT3 activation by IL-6 in half of the cell lines (six on 12). Therefore, these results suggested that in seven out OSM resistance does not strictly correlate with resistance of eight cases, OSM resistance was associated with a to other cytokines defect occurring upstream of STAT3. For the M183 cell In a first attempt to describe the responsiveness of line, resistance to OSM was presumably associated with melanomas to growth-inhibitory cytokines, we analysed defects occurring downstream of or parallel to STAT3.

Oncogene OSM receptor expression in melanoma A Lacreusette et al 883 (Chudnovsky et al., 2005), this modification correlated well with an activating mutation in B-Raf (Table 1 and data not shown, P ¼ 0.02), an upstream activator of the mitogen-induced extracellular kinase/ERK pathway. This constitutively high level of phosphorylated ERK 1/2 did not allow for further activation by IL-6-type cytokines and was observed in either OSM-resistant (M170 and M128 cells in Figure 1b) or sensitive cell lines (see the M200 cell line which is B-Raf-mutated but OSM-sensitive; Figure 1d and Table 1). Because defects in OSM-resistant cell lines could be present at the receptor level, we next analysed the relative mRNA expression levels for various IL-6-type receptor subunits by quantitative reverse transcription–polymer- ase chain reaction (RT–PCR). Individual expression levels are shown in Supplementary Figure 1a and b and the mean expression levels by cytokine-sensitive or resistant cell lines are shown in Figure 2a and b. A major observation was the strongly reduced expression of OSMRb mRNA in OSM-resistant cell lines (39-folds, Po10À4). The correlation between OSMRb mRNA expression and growth inhibition by OSM was categorical (cutoff 15% of growth inhibition) and not linear (Supplementary Table 1a). This defect appeared specific, as gp130 transcripts were expressed at similar levels, regardless of whether the cell lines were resistant or sensitive to OSM or IL-6. Only one OSM-resistant cell line (M120) expressed OSMRb mRNA at a level close to the M140 reference cell line (relative expression 0.9). OSM-resistant cell lines also expressed a lower amount of LIFRa mRNA (9.4-fold reduction), but this was not significant (P ¼ 0.07). In fact, five out of 14 OSM- sensitive cell lines did not express detectable LIFRa mRNA (Supplementary Figure 1a), indicating that the LIFRa subunit was not necessarily used by OSM to mediate growth inhibitory signals. In IL-6-resistant cell lines, the mean IL-6Ra mRNA expression was slightly decreased (1.9-folds) albeit not significantly (P ¼ 0.17). In order to assess whether OSMRb mRNA expression by cultured cell lines reflected the in vivo situation, immunohistochemical analyses of OSMRb were performed in biopsies from primary melanomas and Figure 1 Cytokine resistance of metastatic melanoma cell lines. corresponding invaded lymph nodes from which the (a) The anti-proliferative activity of OSM and IL-6 (top panel) or cell lines were established. We observed that the IFNg and TNFa (bottom panel) at 50 ng/ml each was measured by expression of OSMRb by tumor cells in lymph nodes [3H]-thymidine incorporation. Four OSM-resistant cell lines are presented in comparison with the M140-sensitive cell line. Data are was lost in five out of 14 cases (Supplementary Table 2 the means7s.d. of triplicate cultures and are representative of three and see also Figure 2c: M117 corresponding to an independent experiments. (b) and (d) The phosphorylation of OSMRbhigh and M170 corresponding to an OSMRblow STAT3 and ERK1/2 in response to OSM, IL-6 (50 ng/ml each) and cell line). These cases all corresponded to OSMRb- LIF (100 ng/ml) was analysed by Western blotting in the M140 and defective cell lines (Supplementary Table 2). In contrast, M200 sensitive cell line versus two OSM-resistant cell lines (M170, M128). Equal protein loading was monitored by total STAT3 and the four primary melanoma biopsies tested expressed ERK evaluation in the samples. Cytokine responsiveness among OSMRb similarly (Supplementary Table 2), suggesting cell lines was quantitatively compared by the use of the internal that OSMRb mRNA expression in melanoma cell lines reference OSM-treated WM266.4 cell line (WM). (c) Digital correlated well with OSMRb protein expression in densitometry of the patterns in (b) was expressed as a percentage of OSM-treated WM266.4. lymph node biopsies, but not in primary tumors.

A high basal level of ERK 1/2 activation could be The OSMRb promoter is silenced by histone detected in 54% of the cell lines (see the M170, M128 deacetylation in OSM resistant cells and M200 cell lines in Figure 1b and d as represent- Thus far our results suggested that a defect in OSMRb ative examples). In agreement with previous reports expression was the major molecular event leading to

Oncogene OSM receptor expression in melanoma A Lacreusette et al 884 Table 1 Sensitivity of melanoma cell lines to antiproliferative cytokines Melanoma cell lines Growth inhibition (%)a STAT3phosphorylation b V600E mutated B-Rafc

OSM IL-6 IFNg TNFa OSM IL-6

OSM-sensitive cell lines WM266.4 9271377143766273+++++ M200 8971427578717872+++++ M140 86714072 À2723274++++ À M197 677212726673 À5672++ À + M88 62724771974776+++++ M182 6272607464739671+ + À M147 607337735874 À671+++ + M136 567620713 56772974+++++ M110 547145723876271++++ND M203 3975157747727471+ À + M187 367267949727174+++ + M117 32721972374 À1376+++ À M153 197287459731376+ + À M119 1875107874723274+++ + M138 1875074673 À4072+++ +

OSM-resistant cell lines M134 877377553751977 À ++ À M102 875107856728971 ÀÀ + M210 5750752571573 ÀÀ + M196 37887731713 4777 ÀÀ + M170 2733071 À3771371 À ++ M120 176 À67852751272 À ++ À M128 07397126757872 À ++ À M183 À2276 À18718 58727714 ++ À ND

Abbreviations: IFN, interferon; IL, interleukin; ND, not determined; OSM, oncostatin M; TNF, tumor necrosis factor; STAT, signal transducer and activator of transcription. aCells were treated for 3 days with the indicated cytokine and cell growth was determined by [3H] thymidine incorporation. Growth inhibition values (to control untreated cultures, means7s.d.) are representative of three independent experiments. A threshold of 15% was used to discriminate between OSM-sensitive and -resistant cell lines. bCells were treated for 15 min with cytokines. STAT3 Tyr705 phosphorylation was determined by Western blotting and quantiﬁed by densitometry using the OSM-treated WM266.4 cell line as the 100% internal reference: – (p5%), + (5–50%) and ++ (X50%). cV600E mutation within B-Raf was detected by RT–PCR.

OSM resistance in stage III melanoma cell lines. Because We also tested a longer promoter construct (À1229 to we previously showed that gp130 and LIFRa expression þ 229; hOSMRb Pro1) containing 536 additional bases can be regulated by epigenetic mechanisms in various preceding the CpG island (Figure 3a). This segment was cells (Blanchard et al., 2002; Blanchard et al., 2003), we inactive in all cell lines tested (Figure 3a and data not asked whether the OSMRb gene could be similarly shown), suggesting the presence of inhibitory elements controlled in melanoma. upstream of the CpG island, the precise location and To address this issue, we first cloned and character- function of which are currently unknown. ized the human OSMRb promoter (Figure 3). This In the six OSM-resistant cell lines that expressed a low promoter was found to be immediately 50 of the human level of OSMRb mRNA, chromatin immunoprecipita- OSMRb cDNA (Mosley et al., 1996), and to contain a tions (ChIPs) demonstrated a low level of acetylated CpG island and a potential transcription start site histone H3 and H4 associated with the promoter region (Figure 3a and b). Indeed, the À693 to þ 229 segment of OSMRb (Figure 4a, top panel). In comparison to (hOSMRb Pro2), designed to contain the majority of three OSM-sensitive cell lines with high levels of the CpG island (Figure 3a), was active in transfection OSMRb mRNA, the levels of acetylated histone H3 experiments when inserted upstream of the luciferase and H4 were reduced 2.4- and 3.5-fold, respectively. gene (Figure 3a and c). In comparison to OSM-sensitive Importantly, TSA induced OSMRb-associated acetyla- cell lines with a high level of OSMRb mRNA (M140, tion of histones in five out of six OSMRblow cell lines to a M200 and M117), activity of this promoter was lower in level comparable to or even higher than in the three out of four OSMRblow cell lines that could be OSMRbhigh cells (Figure 4a). In correlation with the efficiently transfected (M128, M102 and M170, increased histone acetylation and promoter activity, Figure 3c). Moreover, trichostatin A (TSA), a broad TSA also induced OSMRb mRNA expression in four HDAC inhibitor, induced promoter activity in these out of six OSMRblow cell lines, although this level OSM-resistant cells, although this activity remained low remained low in the case of the M196 cell line in M170 cells. In M210 OSM-resistant cells, activity of (Figure 4a, bottom panel). In the OSMRbhigh cell lines, the OSMRb promoter was as high as in OSM-sensitive TSA did not significantly modify the level of histone cells and TSA did not induce this activity (Figure 3c). acetylation, in correlation with a stable expression of

Oncogene OSM receptor expression in melanoma A Lacreusette et al 885 increased gp130 cell surface expression in M128 cells two-fold (Figure 4b). Altogether, these results indicated that HDACs were implicated in the repression of the OSMRb gene in various OSM-resistant cells. The functionality of OSMRs induced by TSA treatment was next evaluated by STAT3 activation and by growth inhibition. In four OSM-resistant cell lines (M102, M128, M196 and M134), TSA treatment restored a prominent phosphorylation of STAT3 (Figure 5a). In three of them, TSA also restored growth inhibition by OSM, although this was observed at a high TSA concentration in the case of M196 cells (Figure 5b). This growth inhibition by OSM was observed despite or in addition to the growth inhibition observed after TSA treatment (data not shown). In TSA-treated M196 cells, we also observed activation of STAT3 by LIF (Figure 5a), in correlation with an increased expression of LIFRa mRNA (data not shown). In contrast to its effect on OSM or LIF responses, TSA reduced STAT3 activation by IL-6 in the M128 and M170 cell lines (Figure 5a). This differential effect of TSA on cytokine responses was further supported by the observation that among the 13 melanoma cell lines treated with TSA, seven showed an increased OSMRb mRNA expression, six an increased LIFRa, three an increased gp130 and four a decreased IL-6Ra mRNA expression (data not shown). By Western blot, no modiﬁcation in STAT3 protein expression were ever observed after TSA treatment (data not shown), suggesting that modiﬁca- tion of cytokine responsiveness by TSA was largely due to increased or decreased receptor expression.

OSM responsiveness can be regulated by additional mechanisms The results obtained with some OSM-resistant cell lines revealed additional regulatory mechanisms that did not necessarily depend on histone deacetylation of the OSMRb gene. Indeed, the M210 OSMRblow cell line was not sensitive to TSA treatment in terms of increased OSMRb promoter activity (Figure 3c), histone acetylation, OSMRb expression (Figure 4a) or STAT3 activation (data not shown). The whole OSMRb gene was not Figure 2 OSM resistance correlates with the loss of OSMRb deleted in these cells, because PCR amplification of the expression. The relative mRNA expression levels were evaluated by OSMRb promoter region gave positive results quantitative RT–PCR for gp130, OSMRb and LIFRa in OSM- (Figure 4a). Therefore, silencing of the OSMRb gene sensitive versus-resistant cell lines (A) and for gp130 and IL-6Ra in was more likely the result of other genetic and/or IL-6-sensitive versus -resistant cell lines (B). See Materials and methods for the detailed description of quantification. Data for epigenetic mechanisms resistant to TSA treatment. each receptor subunit are the means7s.d. *Po10À4.(C) Primary A second case is the M170 cell line that responded melanoma biopsies (a, b) or melanoma-invaded lymph nodes (c, d) well to the TSA treatment by an increase of histone were subjected to immunohistochemical staining for OSMRb. acetylation (Figure 4a). However, OSMRb promoter Nuclei were counterstained with hematoxylin. Original magnifica- tion, Â 20. activity remained low (Figure 3c), expression of OSMRb mRNA remained unchanged (Figure 4a), and there was neither activation of STAT3 (Figure 5a) nor growth suppression by OSM (data not shown). Therefore, in OSMRb mRNA (Figure 4a). In the M128 cell line addition to a TSA-sensitive OSMRb gene silencing, a treated with TSA, the appearance of the OSMRb specific defect in OSMRb gene transcription was protein at the cell surface was observed at a level detected in these cells. comparable to that observed in the WM266.4 cell line The behavior of the M134 cell line was surprising used as a positive control (Figure 4b). TSA also because TSA treatment increased histone acetylation,

Oncogene OSM receptor expression in melanoma A Lacreusette et al 886

Figure 3 Identiﬁcation and activity of the human OSMRb promoter. (a) Schematic map of the 50 ﬂanking region of hOSMRb cDNA (square exon 1), showing the CpG island localization and putative transcription start site (arrow). The hOSMRb promoter segments (Pro1 and 2) were cloned upstream of the luciferase gene and their activity was measured in the OSM-sensitive M140 cell line. Data are the means7s.d. of three transfection experiments and are expressed as fold increase over pGL3-Basic activity. (b) The partial nucleotide sequence of the CpG island is presented with potential transcription factor binding sites and the putative transcription start site (arrow). The known cDNA is underlined. (c) Activity of the hOSMRb Pro2 promoter was measured by Luciferase assay in the indicated OSM-sensitive (OSMRbhigh) and OSM-resistant (OSMRblow) cell lines pretreated or not with 500 nM TSA.

OSMRb mRNA expression (Figure 4a) and OSM- M120 cells expressed OSMRb mRNA at a relatively mediated STAT3 activation (Figure 5a) but did not high level (0.9 relative to the M140 cells) but no allow any growth inhibition by OSM (Figure 5b). These corresponding OSMRb protein at the cell surface results indicated that in addition to the OSMRb defect, (Figure 4b) or by immunochemistry (Supplementary there was a defect downstream of or parallel to STAT3 Table 2) and were therefore resistant in terms of STAT3 in the M134 cell line. activation or growth inhibition by OSM (Figure 5c and

Oncogene OSM receptor expression in melanoma A Lacreusette et al 887

Figure 4 The OSMRb promoter is silenced by histone deacetylation in OSM-resistant cell lines. (a, top panel) The specific association of acetylated histone H3 and H4 on the hOSMRb promoter was evaluated by quantitative ChIP. Various OSM-sensitive (OSMRbhigh) and OSM-resistant (OSMRblow) cell lines, pretreated or not with 500 nM TSA, were compared with the input DNA as an internal reference. Data are the means7s.d. of at least two experiments. (a, bottom panel) The relative OSMRb mRNA expression levels were evaluated by quantitative RT–PCR in the same cell lines. Data are expressed relative to the OSMRb mRNA expression level in the M140 cell line. ND: not determined. (b) Gp130 and OSMRb cell surface expression was determined on melanoma cells by flow cytometry. The OSM-sensitive WM266.4 cell line served as a positive control, the M120 and M128 cell lines are resistant to OSM and the M128 cell line is an example of re-induction of OSM receptor subunits by 500 nM TSA treatment (24 h). Values shown as insets correspond to the mean of fluorescence intensity ratios between the anti-gp130 or OSMRb and the control antibodies.

Table 1). After TSA treatment, we observed an induced These results strongly suggested that TSA allowed the activation of STAT3 by OSM and LIF (Figure 5Ca), recruitment of the gp130-LIFRa complex by OSM, but without growth inhibition (data not shown), and despite the fact that we could not detect the LIFRa by neutralizing antibodies against the LIFRa totally pre- RT–PCR in these cells, even after TSA treatment. In an vented STAT3 activation by OSM or LIF (Figure 5Ca). attempt to force the cell surface expression of OSMRb,

Oncogene OSM receptor expression in melanoma A Lacreusette et al 888

Figure 5 TSA treatment of OSM-resistant melanoma cell lines increases their response to OSM. (A) The phosphorylation of STAT3 in response to OSM, IL-6 (50 ng/ml each) and LIF (100 ng/ml) was analysed by Western blotting in OSM-sensitive (M140) or -resistant cell lines pretreated or not with 500 nM TSA. Cytokine responsiveness among cell lines was compared by the use of the internal reference OSM-treated WM-266-4 cell line (WM). (B) The antiproliferative activity of OSM at 50 ng/ml with increasing doses of TSA was measured by [3H]-thymidine incorporation in the same resistant cell lines. Data are the means7s.d. of triplicate cultures, are representative of three independent experiments and are expressed as percentage of OSM-untreated cells (TSA alone). *Po0.05. (Ca) The TSA-induced phosphorylation of STAT3 in response to OSM or LIF was analysed by Western blotting in the M120 cell line in the presence or absence of blocking anti-LIFRa antibodies. (Cb) The phosphorylation of STAT3 in response to OSM, IL-6 and LIF was analysed by Western blotting in M120 cells stably transfected with empty or hOSMRb expression vector. (Cc) The effect of OSM, IL-6 and LIF on proliferation of OSMRb overexpressing M120 cells was measured by [3H]-thymidine incorporation. Data are the means7s.d. of triplicate cultures and are expressed as percentage of untreated cells.

we stably transfected the OSMRb cDNA in these cells. appears more active than IL-6 in reducing melanoma OSMRb mRNA and cell surface protein were increased growth in vitro, whereas LIF is totally inactive. in OSMRb-transfected M120 cells (data not shown), The key role of OSM in melanoma growth was allowing STAT3 activation by OSM but not by LIF conﬁrmed by the high percentage (35%) of metastatic (Figure 5Cb). However, OSM was again unable to cell lines that were resistant to OSM-mediated growth inhibit the growth of the OSMRb overexpressing cells inhibition with a concomitant decrease in OSMRb (Figure 5Cc). Together, these results suggested the mRNA expression (29%) and STAT3 activation presence of two defects in M120 cells: one at the post- (30%). Similarly, melanoma-invaded lymph node transcriptional level for OSMRb expression and another biopsies corresponding to the OSMRblow cell lines downstream of or parallel to STAT3 activation. revealed a lower expression of OSMRb. In contrast, all the primary melanoma cell lines tested so far were sensitive to OSM and high expression of OSMRb was found in primary melanoma biopsies, suggesting that Discussion melanoma progression toward a metastatic state is accompanied by silencing of the OSMRb gene. These In this study, we provide evidence that OSM is an results indicate that speciﬁc mechanisms target the important regulator of melanoma growth. This cytokine OSMRb receptor and not the pathways used by other is very active in inhibiting melanoma cell proliferation, growth inhibitory cytokines such as IL-6, IFNg or by a mechanism which presumably depends on the TNFa. They also help to explain why the phenotype successful recruitment and activation of the OSMRb/ of ‘multicytokine resistance’ was not observed in our gp130 complex and subsequently of STAT3. OSM studies.

Oncogene OSM receptor expression in melanoma A Lacreusette et al 889 In previous studies conducted in numerous normal or stream of or parallel to STAT3 (M183 cell line). The transformed cells, such as hepatoma cells, gp130 or inefficient OSMRb protein expression in M120 cells OSMRb downregulations were rarely observed, in could rely on Jaks deficiency or low availability as contrast to the LIFRa gene that was frequently described previously (Radtke et al., 2002). Other defects epigenetically silenced (Blanchard et al., 2002, 2003). such as point mutations in the OSMRb gene could also In stage III melanoma, both LIFRa and OSMRb were alter its expression in certain melanomas. To our found frequently downregulated, but only the OSMRb knowledge, there is no report describing mutation in recruitment correlated with OSM-mediated growth this gene and only one study described mutations in the inhibition. TSA treatment re-induced OSMRb promoter LIFRa gene that alter mRNA stability, resulting in the acetylation (in 5/8 resistant cell lines) and mRNA levels absence of the LIFRa protein and in the impairment of (in 4/8 resistant cell lines), resulting in the recovery of STAT3 activation by LIF (Dagoneau et al., 2004). OSM sensitivity. Thus, we identified epigenetic mod- Additional experiments are needed to discriminate ifications in the OSMRb promoter, i.e. histone deace- between these possibilities. tylation, as a major regulatory mechanism responsible Defects downstream of STAT3 are presumably at the for OSMRb silencing and OSM resistance. The OSMRb level of key cell cycle regulatory proteins. We observed promoter structure thus appears similar to the one that OSM treatment of sensitive melanomas effectively already described for the gp130 and LIFRa genes blocks the cells in the G1 phase of the cell cycle, but (Blanchard et al., 2002). The LIFRa and OSMRb genes there was no induction of either p21WAF1 or p27KIP1 (data are located within 500 kbp on chromosome 5 (in 5p13), not shown) as described elsewhere (Florenes et al., 1999; suggesting a common ancestral gene, whereas the gp130 Kortylewski et al., 1999; Klausen et al., 2000; Chipoy gene is located near them in 5q11. Their TATA-less et al., 2004). Therefore, a more global analysis of gene promoter regions are all composed of a CpG island expression is necessary to identify the genes implicated associated with differential histone acetylation sensitive in OSM-mediated growth inhibition and potentially to HDAC inhibition (Blanchard et al., 2002, 2003). deregulated in certain resistant melanoma cells. Addi- However, the OSMRb promoter is the only one to be tionally, STAT3 has been described to be acetylated and negatively regulated by a region immediately upstream its transcriptional activity is inhibited by HDACs (Yuan of the CpG island. It remains to be determined which et al., 2005). Therefore, it is possible that HDACs inhibit molecular events induce the epigenetic loss of LIFRa OSM signaling not only by reducing OSMRb transcrip- and/or OSMRb in some tumors like melanoma or tion but also by preventing expression of STAT3 target hepatoma. This loss could be related to enhanced genes. Defects parallel to STAT3 could rely on other expression of tissue-specific HDACs and/or associated OSM-activated signal transduction pathways such as regulatory factors (Blanchard and Chipoy, 2005). STAT1, STAT5, ERK 1/2, AKT, p38MAPK, JNK or Analysis of IFN or IL-6 resistance in melanoma PKCd (Grant and Begley, 1999; Heinrich et al., 2003; revealed major defects at the level of STAT activation or Tanaka and Miyajima, 2003; Chipoy et al., 2004; Boing downstream at the level of induction of gene transcrip- et al., 2006). Despite the large incidence and the key role tion (Wong et al., 1997; Florenes et al., 1999; Pansky of activating mutations in B-Raf upstream of ERK 1/2 et al., 2000; Bohm et al., 2001). Our results confirm that in the etiology of melanoma (Chudnovsky et al., 2005), IL-6 resistance is often independent of IL-6Ra loss, and these modifications do not prevent growth inhibition 50% of these resistant cell lines can still activate STAT3 by OSM. in response to IL-6. Interestingly, IL-6Ra expression Accumulating evidences have shown that constitu- and STAT3 activation by IL-6 were usually decreased tively active STAT3 contributes to the development of by TSA (Figure 5a). In fact, among the genes that can be various cancers, including melanoma, by inducing regulated by HDAC inhibitors (2–9% of the genome), tumor cell survival, proliferation, angiogenesis, meta- equal fractions are activated or repressed (Chambers stasis and immune evasion (Bromberg and Darnell, et al., 2003; Blanchard and Chipoy, 2005). Inhibition 2000; Niu et al., 2002; Kortylewski et al., 2005). Because of gene expression by TSA is prominent for pro- STAT3 was not found constitutively active in our panel inflammatory mediators such as TNFa, IL-6, IL-12 or of melanoma cell lines, it can be anticipated that the nitric oxide, and implies inhibition of the transcription mode of STAT3 activation, for example, transient factor nuclear factor-kappa B(NF- kB) (Blanchard and activation following OSM treatment or prolonged Chipoy, 2005). It is therefore tempting to speculate that stimulation resulting in constitutively active STAT3, both IL-6 and IL-6Ra are mainly controlled by NF-kB will dictate the outcome on tumor progression. Alter- and thus inhibited by TSA, whereas LIFRa, OSMRb natively, the contradictory effects of STAT3 on tumor and possibly gp130 are mostly sensitive to inhibition by growth could depend on other co-stimulated cellular HDACs and are thus activated by TSA. factors that determine the target genes of STAT3. In addition to the major HDAC-dependent loss of Understanding of this important phenomenon is cer- OSMRb, additional mechanisms are also implicated in tainly crucial in the course of cancer treatment and OSM resistance. Indeed, we detected abnormalities in prognosis. several stages of OSMRb production and activation, In conclusion, our results suggest that OSM is an such as TSA-resistant gene silencing (M210 cell line), important factor in the control of melanoma growth. transcriptional inactivation (M170 cell line), inefficient Silencing and loss of OSMRb are major events protein expression (M120 cell line) and defects down- associated with OSM resistance in metastatic melanoma

Oncogene OSM receptor expression in melanoma A Lacreusette et al 890 and they could lead to a growth advantage during hexamers and SuperScript II reverse transcriptase (Invitrogen). particular inflammatory stages or anticancer immune Relative quantification of OSMRb, gp130, LIFRa and IL-6Ra reactions. HDAC inhibitors have been used in phase expression was carried out using Brilliant SYBR Green QPCR I–II clinical trials for the treatment of various hemato- Master Mix (Stratagene Europe, Amsterdam, The Nether- logical or solid malignancies such as melanoma (Sandor lands) with exon–exon junction-spanning primers (Supple- mentary Table 1) and an Mx4000 multiplex quantitative PCR et al., 2002; Marks et al., 2004). They have shown instrument (Stratagene). Average threshold cycle (C ) values antitumor activities against melanoma in vitro and in T from duplicate PCR reactions were normalized to average CT vivo and represent a promising therapeutic approach for values for three housekeeping genes (b-actin, b-2-microglobulin patients with malignant melanoma (Yoshida et al., 2001; and RPLPO) from the same cDNA preparations. The stability Boyle et al., 2005). Their capacity to restore OSMRb of these housekeeping genes was approved by the BestKeeper expression and growth inhibition by OSM could help to software (Pfaffl et al., 2004). Thus, the relative expression ratio restrain the progression of this aggressive malignancy. of a target gene was calculated based on the PCR efficiency (E) and the CT deviation between a given cell line (x) and a reference cell line (calibrator), expressed in comparison with the geometric mean of the housekeeping genes (Pfaffl, 2001): Materials and methods DCTtarget ðcalibratorÀxÞ Ratio ¼ðEtargetÞ =geo mean Cell lines and reagents DCThousekeepingðcalibratorÀxÞ Metastatic melanoma cell lines were established from tumor- ððEhousekeepingÞ Þ invaded lymph node fragments as described previously (Gervois et al., 1990; Dreno et al., 2002; Labarriere et al., Several RNA extracts were independently analysed and 2002). Melanoma cell lines were established before injection of reproducibility was confirmed with a correlation coefficient autologous TILs plus IL-2 or IL-2 only. The metastatic r ¼ 0.99. melanoma WM-266-4 cell line was purchased from the The V600E mutation in B-Raf was detected as described American Type Culture Collection (LGC Promochem, Mol- previously (Jarry et al., 2004). sheim, France). All cells were maintained in RPMI 1640 or Dulbecco’s modified Eagle’s medium (Sigma, Lyon, France) Cloning of the human OSMRb cDNA and promoter supplemented with 10% fetal calf serum and 2 mML- The hOSMRb cDNA, cloned into the pDC409 vector (a kind glutamine. All recombinant human cytokines were from gift from Dr H Baumann, RPCI, Buffalo, USA), was R&D Systems (Lille, France). TSA was from Calbiochem subcloned into the EcoRV and NotI sites of the pcDNA3.1/ (La Jolla, CA, USA). Hygro ( þ ) vector (Invitrogen). The 50 flanking sequence of the hOSMRb cDNA (NM_003999) was obtained using the BLAT alignment tool Cell proliferation assay from UCSC Genome Browser (http://genome.ucsc.edu/). The antiproliferative activity of OSM (50 ng/ml), IL-6 (50 ng/ Putative transcription start site was predicted using the Human ml), LIF (100 ng/ml), IFNg (50 ng/ml) and TNFa (50 ng/ml) 3 Core-promoter Finder (http://rulai.cshl.org/), whereas tran- was measured by [ H]-thymidine incorporation after 3 days of scription factor binding sites were predicted using the et al treatment as described previously (Heymann ., 1995; MatInspector of Genomatix (http://www.genomatix.de/). Blanchard et al., 2002). For sensitization of resistant cells to Two overlapping segments were amplified by PCR using OSM, increasing doses of TSA were incubated with OSM genomic DNA from the M88 melanoma cell line. These (50 ng/ml) and [3H]-thymidine incorporation was measured segments contained the nucleotides from position À1229 to after 48 h of treatment. þ 229 (hOSMRb Pro1) and –693 to þ 229 (hOSMRb Pro2), relative to the transcription start site. They were then cloned Western blotting into the pCR-Blunt II-TOPO vector (Invitrogen) and sub- After 4 h of serum deprivation and 15 min of cytokine cloned into the KpnI and XhoI sites of the luciferase reporter stimulation, control or 24 h TSA-pretreated cell monolayers plasmid pGL3-Basic (Promega, Charbonnieres, France). All were lysed in RIPA buffer. For the blockade of LIFRa, cells constructs were confirmed by sequence analysis. were preincubated with neutralizing anti-LIFRa antibodies (Blanchard et al., 1997) for 90 min before cytokine stimulation. Luciferase reporter assay Equal amounts of protein lysates were separated on 10% Melanoma cell lines were transiently transfected using sodium dodecyl sulfate–polyacrylamide gels and transferred to FuGENE6 (Roche Diagnostics, Meylan France) or Lipofec- polyvinylidine difluoride membranes (Immobilon-P, Millipore, tAMINE (Invitrogen) with empty or hOSMRb promoter- Bedford, MA, USA). Western blotting was performed using driven pGL3-Basic which encodes Firefly luciferase as a antibodies against phospho-STAT3 (Tyr705), phospho-ERK reporter, together with a 50:1 ratio of Renilla luciferase vector 1/2 (Thr202/Tyr204) (Cell Signaling, Beverly, MA, USA), (pRL-CMV; Promega) as an internal control. Luciferase STAT3 (BD Bioscience, Le Pont de Claix, France) or ERK 1/2 activity was measured 48 h following transfection and 24 h (Cell Signaling). Digital densitometry of the patterns was after TSA treatment of OSMRblow cells, with the Dual- analysed with the NIH Image software. To enable the Luciferase kit (Promega) and a single-sample luminometer quantitative comparison of cytokine responsiveness among (Berthold Technologies, Bad Wildbad, Germany). Whenever cell lines, an OSM-treated WM266.4 cell lysate was used as an possible, each transfection was reproduced three times and internal reference (100% level). human OSMRb promoter activity was calculated as a ratio to Renilla luciferase activity. Real-time quantitative RT–PCR Total RNA was extracted from control or 24 h TSA-treated Quantitative ChIP assay cells by Trizol reagent (Invitrogen, Cergy Pontoise, France). ChIP analysis was performed according to a modified protocol Retrotranscription was performed using 2 mg of RNA, random of Upstate Biotechnology (Dundee, UK) and essentially as

Oncogene OSM receptor expression in melanoma A Lacreusette et al 891 described previously (Blanchard et al., 2002). Control or 24 h Immunohistochemistry TSA-treated melanoma cell lines were fixed, lysed and Five micrometer sections of formalin-fixed, paraffin-embedded sonicated to shear DNA. After 1 h of preclearing with 50 ml primary melanoma biopsies or frozen melanoma-invaded of protein G-sepharose (Invitrogen), samples were incubated lymph nodes were incubated for 30 min with a goat antihuman overnight at 41C with 5 mg of anti-acetylated histone H3, anti- OSMRb antibody (Santa Cruz Biotechnology, Heidelberg, acetylated histone H4 (Upstate Biotechnology) or irrelevant Germany) diluted 1/25 in phosphate-buffered saline–bovine rabbit IgG (Sigma) and 2 mg of salmon sperm DNA (Sigma). serum albumin 0.1% or with diluent alone as a negative Immune complexes were recovered with 100 ml of salmon control. The rinsed sections were then incubated for a further sperm DNA-saturated protein G-sepharose. After elution and 30 min with a 1/200 dilution of biotinylated rabbit antigoat purification, the ChIP-captured DNA was then screened for antibody (DakoCytomation, Trappes, France). Immunostain- hOSMRb promoter fragments by real-time PCR analysis using ing for OSMRb was revealed using streptavidine-HRP and specific primers (sense 50-TTGAAGATTGGATGGCATTT-30 diaminobenzidine (DakoCytomation), and nuclei were coun- and antisense 50-AAAACGTGTTTGAACCGAA-30) located terstained with hematoxylin. All sections were examined by in the hOSMRb Pro2 segment described above. Serial three investigators, one of whom (BD) was blinded to the dilutions of input DNA generated a standard curve used to molecular and clinical data. determine the relative amount of precipitated DNA. Dupli- cates for both standards and the samples were simultaneously amplified and at least two independent immunoprecipitations Statistical analyses were analysed. Data are expressed as means7s.d. and frequencies. Contin- uous data were compared using the Wilcoxon test. Categorical Flow cytometric analysis data were compared using the Fisher exact test. Correlations Expression of receptor subunits at the cell surface was between continuous data were accessed using the Spearman monitored using an antihuman gp130 antibody (B-R3, mouse rank test. A value of Po0.05 was considered as significant. IgG2a, Diaclone Research, Besançon, France) or an antihuman OSMRb antibody (P1, rabbit serum, Amgen Inc., Thousand Oaks, USA) with mouse IgG (BD Biosciences) or Acknowledgements rabbit pre-immune serum as isotype controls, respectively. These primary antibodies were revealed by anti-mouse- We thank Dr M Denis (Inserm U539, Nantes, France) for phycoerythrin and anti-rabbit-Alexa488, respectively. Cell- performing the B-Raf mutation analysis. This work was associated fluorescence was acquired by a FACScalibur supported by Association pour la Recherche sur le Cancer fluorocytometer (BD Biosciences) and analysed using Cell- (F Blanchard), the Ligue contre le Cancer and the Institut Quest software. National de la Sante´ et de la Recherche Me´ dicale.

References

Bani MR, Rak J, Adachi D, Wiltshire R, Trent JM, Kerbel RS Boyle GM, Martyn AC, Parsons PG. (2005). Histone et al. (1996). Multiple features of advanced melanoma deacetylase inhibitors and malignant melanoma. Pig Cell recapitulated in tumorigenic variants of early stage (radial Res 18: 160–166. growth phase) human melanoma cell lines: evidence for a Bromberg J, Darnell Jr JE. (2000). The role of STATs in dominant phenotype. Cancer Res 56: 3075–3086. transcriptional control and their impact on cellular function. Blanchard F, Chipoy C. (2005). Histone deacetylase inhibitors: Oncogene 19: 2468–2473. new drugs for the treatment of inflammatory diseases? Drug Chambers AE, Banerjee S, Chaplin T, Dunne J, Debernardi S, Discov Today 10: 197–204. Joel SP et al. (2003). Histone acetylation-mediated regula- Blanchard F, Kinzie E, Wang Y, Duplomb L, Godard A, Held tion of genes in leukaemic cells. Eur J Cancer 39: 1165–1175. WA et al. (2002). FR901228, an inhibitor of histone Chipoy C, Berreur M, Couillaud S, Pradal G, Vallette F, deacetylases, increases the cellular responsiveness to IL-6 Colombeix C et al. (2004). Downregulation of osteoblast type cytokines by enhancing the expression of receptor markers and induction of the glial fibrillary acidic protein by proteins. Oncogene 21: 6264–6277. oncostatin M in osteosarcoma cells require PKCdelta and Blanchard F, Pitard V, Taupin JL, Raher S, Hallet MM, STAT3. J Bone Miner Res 19: 1850–1861. Moreau JF et al. (1997). Epitope-function relationships of Chudnovsky Y, Khavari PA, Adams AE. (2005). Melanoma human leukemia inhibitory factor receptors using a novel set genetics and the development of rational therapeutics. J Clin of anti-gp190mAB. Int Immunol 9: 1775–1784. Invest 115: 813–824. Blanchard F, Tracy E, Smith J, Chattopadhyay S, Wang Y, Dagoneau N, Scheffer D, Huber C, Al-Gazali LI, Di Rocco Held WA et al. (2003). DNA methylation controls the M, Godard A et al. (2004). Null leukemia inhibitory factor responsiveness of hepatoma cells to leukemia inhibitory receptor (LIFR) mutations in Stuve–Wiedemann/Schwartz– factor. Hepatology 38: 1516–1528. Jampel type 2 syndrome. Am J Hum Genet 74: 298–305. Bohm M, Schulte U, Funk JO, Raghunath M, Behrmann I, Dreno B, Nguyen JM, Khammari A, Pandolfino MC, Tessier Kortylewski M et al. (2001). Interleukin-6-resistant melano- MH, Bercegeay S et al. (2002). Randomized trial of adoptive ma cells exhibit reduced activation of STAT3 and lack of transfer of melanoma tumor-infiltrating lymphocytes as inhibition of cyclin E-associated kinase activity. J Invest adjuvant therapy for stage III melanoma. Cancer Immunol Dermatol 117: 132–140. Immunother 51: 539–546. Boing I, Stross C, Radtke S, Lippok BE, Heinrich PC, Florenes VA, Lu C, Bhattacharya N, Rak J, Sheehan C, Hermanns HM. (2006). Oncostatin M-induced activation of Slingerland JM et al. (1999). Interleukin-6 dependent stress-activated MAP kinases depends on tyrosine 861 in the induction of the cyclin dependent kinase inhibitor OSM receptor and requires Jak1 but not Src kinases. Cell p21WAF1/CIP1 is lost during progression of human Signal 18: 50–61. malignant melanoma. Oncogene 18: 1023–1032.

Oncogene OSM receptor expression in melanoma A Lacreusette et al 892 Gervois N, Heuze F, Diez E, Jotereau F. (1990). Selective Mule JJ, McIntosh JK, Jablons DM, Rosenberg SA. (1990). expansion of a specific anti-tumor CD8+ cytotoxic T Antitumor activity of recombinant interleukin 6 in mice. lymphocyte clone in the bulk culture of tumor-infiltrating J Exp Med 171: 629–636. lymphocytes from a melanoma patient: cytotoxic activity Nawrocki S, Murawa P, Malicki J, Kapcinska M, Gryska K, and T cell receptor gene rearrangements. Eur J Immunol 20: Izycki D et al. (2000). Genetically modified tumour vaccines 825–831. (GMTV) in melanoma clinical trials. Immunol Lett 74: Grant SL, Begley CG. (1999). The oncostatin M signalling 81–86. pathway: reversing the neoplastic phenotype? Mol Med Niu G, Bowman T, Huang M, Shivers S, Reintgen D, Daud A Today 5: 406–412. et al. (2002). Roles of activated Src and Stat3 signaling in Heinrich PC, Behrmann I, Haan S, Hermanns HM, Muller- melanoma tumor cell growth. Oncogene 21: 7001–7010. Newen G, Schaper F. (2003). Principles of interleukin (IL)-6- Ozbek S, Peters M, Breuhahn K, Mann A, Blessing M, Fischer type cytokine signalling and its regulation. Biochem J 374: M et al. (2001). The designer cytokine hyper-IL-6 mediates 1–20. growth inhibition and GM-CSF-dependent rejection of B16 Heymann D, Blanchard F, Raher S, De Groote D, melanoma cells. Oncogene 20: 972–979. Godard A. (1995). Modulation of LIF expression in human Pansky A, Hildebrand P, Fasler-Kan E, Baselgia L, Ketterer S, melanoma cells by oncostatin M. Immunol Lett 46: Beglinger C et al. (2000). Defective Jak-STAT signal 245–251. transduction pathway in melanoma cells resistant to growth Jaenisch R, Bird A. (2003). Epigenetic regulation of gene inhibition by interferon-alpha. Int J Cancer 85: 720–725. expression: how the genome integrates intrinsic and envir- Pfaffl MW. (2001). A new mathematical model for relative onmental signals. Nat Genet 33(Suppl): 245–254. quantification in real-time RT-PCR. Nucleic Acids Res 29: Jarry A, Masson D, Cassagnau E, Parois S, Laboisse C, Denis e45. MG. (2004). Real-time allele-specific amplification for Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP. (2004). sensitive detection of the BRAF mutation V600E. Mol Cell Determination of stable housekeeping genes, differentially Probes 18: 349–352. regulated target genes and sample integrity: BestKeeper – Kim H, Baumann H. (1999). Dual signaling role of the Excel-based tool using pair-wise correlations. Biotechnol protein tyrosine phosphatase SHP-2 in regulating Lett 26: 509–515. expression of acute-phase plasma proteins by interleukin-6 Radtke S, Hermanns HM, Haan C, Schmitz-Van De Leur H, cytokine receptors in hepatic cells. Mol Cell Biol 19: Gascan H, Heinrich PC et al. (2002). Novel role of Janus 5326–5338. kinase 1 in the regulation of oncostatin M receptor surface Klausen P, Pedersen L, Jurlander J, Baumann H. (2000). expression. J Biol Chem 277: 11297–11305. Oncostatin M and interleukin 6 inhibit cell cycle progression Sandor V, Bakke S, Robey RW, Kang MH, Blagosklonny by prevention of p27kip1 degradation in HepG2 cells. MV, Bender J et al. (2002). Phase I trial of the histone Oncogene 19: 3675–3683. deacetylase inhibitor, depsipeptide (FR901228, NSC Kortylewski M, Heinrich PC, Mackiewicz A, Schniertshauer 630176), in patients with refractory neoplasms. Clin Cancer U, Klingmuller U, Nakajima K et al. (1999). Interleukin-6 Res 8: 718–728. and oncostatin M-induced growth inhibition of human A375 Tanaka M, Miyajima A. (2003). Oncostatin M, a multi- melanoma cells is STAT-dependent and involves upregula- functional cytokine. Rev Physiol Biochem Pharmacol 149: tion of the cyclin-dependent kinase inhibitor p27/Kip1. 39–52. Oncogene 18: 3742–3753. Wong LH, Krauer KG, Hatzinisiriou I, Estcourt MJ, Hersey Kortylewski M, Jove R, Yu H. (2005). Targeting STAT3 P, Tam ND et al. (1997). Interferon-resistant human affects melanoma on multiple fronts. Cancer Metastasis Rev melanoma cells are deficient in ISGF3 components, STAT1, 24: 315–327. STAT2, and p48-ISGF3gamma. J Biol Chem 272: Labarriere N, Pandolfino MC, Gervois N, Khammari A, 28779–28785. Tessier MH, Dreno B et al. (2002). Therapeutic efficacy of Yannelli JR, Wroblewski JM. (2004). On the road to a tumor melanoma-reactive TIL injected in stage III melanoma cell vaccine: 20 years of cellular immunotherapy. Vaccine 23: patients. Cancer Immunol Immunother 51: 532–538. 97–113. Lu C, Rak JW, Kobayashi H, Kerbel RS. (1993). Increased Yoshida M, Furumai R, Nishiyama M, Komatsu Y, Nishino resistance to oncostatin M-induced growth inhibition of N, Horinouchi S. (2001). Histone deacetylase as a new target human melanoma cell lines derived from advanced-stage for cancer chemotherapy. Cancer Chemother Pharmacol lesions. Cancer Res 53: 2708–2711. 48(Suppl 1): S20–26. Marks PA, Richon VM, Miller T, Kelly WK. (2004). Histone Yuan ZL, Guan YJ, Chatterjee D, Chin YE. (2005). Stat3 deacetylase inhibitors. Adv Cancer Res 91: 137–168. dimerization regulated by reversible acetylation of a single Mosley B, De Imus C, Friend D, Boiani N, Thoma B, Park LS lysine residue. Science 307: 269–273. et al. (1996). Dual oncostatin M (OSM) receptors. Cloning Zarling JM, Shoyab M, Marquardt H, Hanson MB, Lioubin and characterization of an alternative signaling subunit MN, Todaro GJ. (1986). Oncostatin M: a growth regulator conferring OSM-specific receptor activation. J Biol Chem produced by differentiated histiocytic lymphoma cells. Proc 271: 32635–32643. Natl Acad Sci USA 83: 9739–9743.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene