Oncogene (2007) 26, 1941–1948 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Expression of SOCS1 and SOCS3 is differentially regulated in breast cancer cells in response to proinflammatory and growth factor signals

MK Evans1, C-R Yu2, A Lohani1, RM Mahdi2, X Liu2, AR Trzeciak1 and CE Egwuagu2

1Laboratory of Cellular and Molecular Biology, National Institute on Aging, National Institutes of Health, Baltimore, MD, USA and 2Laboratory of Immunology, National Eye Institute, National Institutes of Health, Bethesda, MD, USA

DNA-hypermethylation of SOCS genes in breast, ova- Keywords: breast-cancer cells; SOCS; breast cancer; STAT; rian, squamous cell and hepatocellular carcinoma has led BRCA1; DNA hypermethylation; interferon to speculation that silencing of SOCS1 and SOCS3 genes might promote oncogenic transformation of epithelial tissues. To examine whether transcriptional silencing of SOCS genes is a common feature of human carcinoma, we have investigated regulation of SOCS genes expression by Introduction IFNc, IGF-1 and ionizing radiation, in a normal human mammary epithelial cell line (AG11134), two breast- Development of the mammary gland is regulated by cancer cell lines (MCF-7, HCC1937) and three prostate factors that coordinate proliferation, differentiation, cancer cell lines. Compared to normal breast cells, we maturation and apoptosis of mammary epithelial cells. observe a high level constitutive expression of SOCS2, Exquisite control of the initiation, strength and duration SOCS3, SOCS5, SOCS6, SOCS7, CIS and/or SOCS1 of signals from the diverse array of and genes in the human cancer cells. In MCF-7 and HCC1937 growth factors in mammalian breast is essential to the breast-cancer cells, transcription of SOCS1 is dramati- timely initiation of the menstrual cycle, lactation or cally up-regulated by IFNc and/or ionizing-radiation breast lobule involution (Medina, 2005). Germ-line while SOCS3 is transiently down-regulated by IFNc and mutations or epigenetic mechanisms that silence pro- IGF-1, suggesting that SOCS genes are not silenced teins that regulate cytokine activities underlie breast in these cells by the epigenetic mechanism of DNA- tumorigenesis. This has led to interest in cytokine- hypermethylation. We further showthat the kinetics of signaling mechanisms and how they can be exploited to SOCS1-mediated feedback inhibition of IFNc signaling pharmacologically inhibit aberrantly activated human is comparable to normal breast cells, indicating that cancer cells. the SOCS1 in breast-cancer cells is functional. Most growth factors/cytokines implicated in patho- We provide direct evidence that STAT3 pathways are genesis of breast cancer share the common feature of constitutively activated in MCF-7 and HCC1937 cells and activating STAT pathways (Darnell, 1997; Clevenger, may drive the aberrant persistent activation of SOCS 2004). The 6-member STAT family of transcription genes in breast-cancer cells. Our data therefore suggest factors links cytokine signals to transcriptional events that elevated expression of SOCS genes is a specific that regulate cell-cycle progression and apoptosis lesion of breast-cancer cells that may confer resistance to (Bowman et al., 2000; Darnell, 2005). The strength proinflammatory cytokines and trophic factors, by and duration of STAT signals are regulated by suppres- shutting down STAT1/STAT5 signaling that mediate sor of cytokine-signaling (SOCS) , an 8-member essential functions in the mammary gland. family of cytokine-inducible proteins that attenuate or Oncogene (2007) 26, 1941–1948. doi:10.1038/sj.onc.1209993; terminate cytokine/growth factor signals (Hilton, 1999). published online 25 September 2006 If not properly regulated, constitutive activation of STAT pathways can induce oncogenic transformation, tumor cell invasion and metastasis (Darnell, 2005; Haura et al., 2005). In fact, defects in negative-feedback regulation of STAT pathways have been identified in Correspondence: Dr MK Evans, Laboratory of Cellular and Molecular Biology, National Institute on Aging, National Institutes diverse tumors (Sutherland et al., 2004; Campbell, 2005; of Health, Baltimore, Maryland 21224, USA. Haura et al., 2005; Sultan et al., 2005). Recently, E-mail: [email protected] or hypermethylation of the SOCS1 or SOCS3 has Dr CE Egwuagu, Laboratory of Immunology, National Eye Institute, been reported in lung cancer (He et al., 2003), breast and National Institutes of Health, Bethesda, Maryland 20892, USA. E-mail: [email protected] ovarian cancer (Sutherland et al., 2004), squamous cell Received 29 April 2006; revised 9 August 2006; accepted 11 August 2006; carcinoma of the head and neck (Weber et al., 2005) and published online 25 September 2006 hepatocellular carcinoma (Niwa et al., 2005). This has SOCS genes are not silenced in breast-cancer cells MK Evans et al 1942 led to the speculation that SOCS genes are transcrip- tionally silenced during oncogenic transformation of epithelial carcinoma cells and that defective expression of SOCS proteins, with subsequent loss of feedback- regulation, may contribute to tumor progression and metastasis (Rottapel et al., 2002; He et al., 2003). In this study, we have utilized a panel of tumor cell lines to investigate whether: (i) SOCS genes are tran- scriptionally silent in breast and prostate cancer cells; (ii) whether negative-feedback responses mediated by SOCS proteins are normal or defective in breast-cancer cells. We show that SOCS genes are constitutively expressed at relatively high levels in human breast-cancer cells, indicating that SOCS promoters are not transcrip- tionally silent. Our data further indicates that elevated SOCS expression in breast-cancer cells derives from persistent activation of SOCS genes in response to constitutive activation of STAT3 pathways in these cells.

Results

SOCS genes are transcriptionally active and constitutively Figure 1 SOCS genes are transcriptionally active and constitu- tively expressed in breast and prostate cancer cells. (a) Ribonu- expressed in breast-cancer cells clease protection assay (RPA). (b) Real-time RT–PCR analysis of We utilized two breast-cancer cell lines (MCF-7 and normal human prostate cells (RWPE-1) or malignant prostate HCC1937) and a normal mammary epithelial cell line cancer cell lines (LNCaP, PC-3, DU-145). (AG11134) to examine whether transcriptional silencing of SOCS genes occurs frequently in human breast carcinoma cells. As a positive control, we also analysed a B-lymphoblastoid cell line (HCC1937BL) that consti- SOCS genes are constitutionally expressed, but also that tutively expresses high levels of most SOCS family genes the cancer cells are capable of inducing transcription of (unpublished data). The RPA data presented in Figure 1a SOCS genes in context of classical feedback response to reveals that most SOCS genes are constitutively expressed growth-factor/cytokine signaling. To this end, MCF-7, at high levels in the breast-cancer cells relative to the HCC1937 and AG11134 cells were starved for 12 h, normal breast cells. Although SOCS1, SOCS2 and stimulated with IFNg and then analyzed for induction SOCS3 genes are known to be hypermethylated in of SOCS expression. We show that IFNg activates MCF-7 cells (He et al., 2003; Sutherland et al., 2004), transcription of SOCS1 and CIS in these cells. On the this result clearly reveals that these genes are not other hand, SOCS3 expression is induced in AG11134 transcriptionally silent in MCF-7 cells. In addition, but downregulated in the two cancer cell lines at the SOCS2, SOCS3, SOCS5, SOCS6 and SOCS7 genes 120-min time point (Figure 2). However, moderate induc- are expressed in HCC1937 cells at levels comparable to tion of SOCS3 is also observed in HCC1937 at 30-min. MCF-7 (Figure 1a). We also analysed SOCS expression The complex pattern of SOCS5 or SOCS6 expression in a normal prostate cell line (RWPE-1) and three observed is consistent with the fact that these SOCS prostate cancer cell lines (LNCaP, PC-3 and DU-145) members are not classic feedback regulators of cytokine- and as shown in Figure 1b, SOCS1, SOCS3, SOCS5 and signaling and the mechanism of transcriptional regula- CIS genes are transcriptionally active in these cell lines, tion of both genes is poorly understood. These results albeit to varying extent. Both the PC-3 and DU-145 clearly suggest that SOCS promoters of the breast- express levels of SOCS1, SOCS3, SOCS5, CIS genes that cancer cells are responsive to cytokine-induced tran- are significantly lower than the normal RWPE prostate scriptional regulation as indicated by upregulation of cells. On the other hand, LNCaP cells express SOCS1, SOCS1 and CIS genes and downregulation of SOCS3 SOCS3, SOCS5 and CIS at levels comparable to RWPE, gene. underscoring the fact that this carcinoma cell line exhibits anormalpatternofSOCS genes expression. These results Duration of STAT signals is inversely correlated with SOCS indicate that genes are not silenced in all breast or SOCS levels in breast cells prostate cancer cells. To investigate whether SOCS proteins in breast-cancer cells are capable of mediating feedback-inhibition of Interferon-g activates transcription of SOCS genes in cytokine-signaling, we stimulated MCF-7, HCC1937, MCF-7 or HCC1937 cells HCC1937BL and AG11134 cells with IFNg for varying To demonstrate that SOCS genes are not silenced in amounts of time and determined the kinetics of STAT1 breast-cancer cells, it is not only important to show that activation and subsequent termination of the STAT1

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1943

Figure 2 Transcription of SOCS genes is induced by IFNg in MCF-7 and HCC1937 breast cancer cells. Real-time RT–PCR quantification of relative abundance of SOCS mRNA transcripts in IFNg-stimulated AG11134, MCF-7 or HCC1937 cells.

signal. STAT1 activation, as indicated by phosphoryla- fold-increase noted for AG11134 and HCC1937 (40-fold) tion of STAT1 (pSTAT1), is not detected in unstarved derives from their low endogenous SOCS1 levels. and unstimulated cells (Figure 3a), suggesting that Similar to data presented (Figure 2), expression of STAT1 pathway is not constitutively activated in these SOCS1 and SOCS3 is upregulated by IFNg in AG11134 cells. On the other hand, relatively high levels of while SOCS3 transcription is repressed in the breast- pSTAT1 are detected in MCF-7, HCC1937 and cancer cells (Figure 3b). These results reveal a funda- AG11134 following 30min stimulation by IFN g mental difference in the response of SOCS3 promoter to (Figure 3a). However, after 120min pSTAT1 is con- IFNg-induced STAT-signaling in normal breast and siderably lower in MCF-7 cells while the STAT1-signal breast-cancer cells and suggest that duration of STAT1- is still intense in AG11134 that has the lowest levels of signals is inversely correlated with steady-state SOCS endogenous SOCS expression. In contrast, after 30min mRNA levels in the cells. only a modest level of pSTAT1 is detected in HCC1937BL with the highest endogenous SOCS expression and duration of STAT1 signal is short-lived Growth factor signals downregulate transcription of (Figure 3a). However, after 3 h pSTAT1 is no longer SOCS genes in breast-cancer cells detectable in any of the stimulated cells (data not In view of the high endogenous levels of SOCS shown), suggesting a complete termination of the STAT expression in breast-cancer cells (Figure 1), it was of signal. To examine whether termination of STAT1 interest to examine whether activation of STAT path- signal correlates temporally with induction of SOCS ways by growth factors or their feedback regulation is transcription, we quantified SOCS1 and SOCS3 expres- defective in these cells. Whole cell protein extracts, RNA sion during the course of stimulation with IFNg.As and nuclear extracts from MCF-7 or HCC1937 cells noted in Figure 3b, there is a 7.5-fold increase in SOCS1 stimulated with IGF-1 were therefore analysed for expression in AG11134 and a threefold increase in STAT activation or SOCS by Western MCF-7 cells in response to IFNg. The lower fold- blotting, EMSA or real-time polymerase chain reaction increases noted for HCC1937BL and MCF-7 reflects (PCR). Detection of pSTAT1 and pSTAT3 in stimu- the fact that absolute level of SOCS1 transcripts is lated MCF-7 and HCC1937 (Figure 4a), suggests that higher in these cells (Figure 1a) and calculation of fold- high basal level of SOCS expression does not prevent changes is based on comparison to the basal level of activation of these pathways in breast-cancer cells. We SOCS expression. Conversely, the relatively higher provide evidence that STAT3 pathway is constitutively

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1944 Expression of SOCS1 gene is regulated by IRF-1 in BRCA1-deficient breast-cancer cells Transcriptional activation of SOCS1 by IFNg requires IRF-1 (IFN regulatory factor 1) and STAT1 (Saito et al., 2000). Recently, STAT1 was shown to collaborate with BRCA1 in activating genes that mediate growth inhibitory and pro-apoptotic effects of IFNg and among these are p21WAF1 and IRF-1 (Ouchi et al., 2000). We, therefore, investigated whether similar to SOCS1,if expression of other IFNg/STAT1-inducible genes is also defective in BRCA1-deficient HCC1937 cells. We show that the amount of IRF-1 or p21WAF1 protein detected in HCC1937 is low compared to MCF-7, HCC1937BL or AG11134 cells (Figure 5a), suggesting that the low endogenous SOCS1 expression in BRCA1-deficient HCC1937 cells (Figure 1a) may derive in part from a global defects in expression of BRCA1/STAT1 regu- lated genes. To directly examine whether the IRF-1 and p21WAF1 genes are transcriptionally silent, we stimulated HCC1937 cells with IFNg or g-irradiation, two agents that are known to induce transcription of these genes. As shown in Figure 5b, IFNg induces significant increase in IRF-1 expression and peak of IRF-1 protein Figure 3 Duration of STAT signals is inversely correlated with expression correlates temporarily with maximum induc- SOCS levels in breast cancer and normal mammary epithelial cells. tion of SOCS1 expression in HCC1937 cells (Figure 2). (a) Western blot analysis of IFNg-stimulated cells. Antibodies Thus, although BRCA1 is required for optimal expres- used for immunoblotting are indicated to the right. (b) Real-time sion of IRF-1 and SOCS1, the requirement of BRCA1 RT–PCR analysis of the IFNg-stimulated cells. Data are shown as can be overcome by increasing the intensity of STAT1- fold increases in SOCS mRNA in stimulated compared to non- stimulated cells. signal in HCC1937 cells. Interestingly, whereas expres- sion of IRF-1 and p21WAF1 genes is upregulated by g-radiation (Figure 5c), the p21WAF1 promoter is not responsive to activation by IFNg (Figure 5b). None- activated (Figure 4b, left panel) and activated STAT3 theless, the observed effects of g-irradiation is consistent homodimers interacting with GAS motifs of STAT3- with reports showing that IRF-1 and p53 converge regulated genes in MCF-7 or HCC1937 cells increase functionally to upregulate p21WAF1 in response to following IGF-1 stimulation (Figure 4b, right panel). genotoxic stress (Tanaka et al., 1996). As SOCS1 and We show that the SIE GAS probe interacts with p21WAF1 genes are regulated by IRF-1 (Coccia et al., activated STAT1 and STAT3 in IFNg-stimulated cells 1999), these results suggest that the low basal expression (Figure 4c) and super-shift assays reveal that the IGF1- of SOCS1 and p21WAF1 genes derive indirectly from induced DNA-binding activity is pSTAT3 (Figure 4d). defects in IRF-1 expression in HCC1937 cells. The data Surprisingly, pSTAT3 but not pSTAT1 is able to bind also underscore the fact that the IRF-1, SOCS1, p21WAF1 DNA despite the fact that both proteins are activated in promoter can be active or silent depending on the MCF-7 and HCC1937 cells following IGF-1 stimulation physiological state of the cell but are clearly not (see Figure 4a and d). This suggests that IGF-1 signals ‘transcriptionally silent’ in HCC1937 cells. may interfere with DNA-binding by pSTAT1, thereby inhibiting activation of STAT1-dependent anti-proli- ferative pathways. Of particular interest is our finding that stimulation of MCF-7 or HCC1937 by IGF-1 Discussion induces a transient downregulation of SOCS1 and SOCS3 gene expression (Figure 4d and e) that is Recent reports of aberrant DNA-hypermethylation of followed by a gradual increase to their steady-state SOCS genes in a number of breast-cancer cell lines have levels (data not shown). Although the functional led to the suggestion that SOCS genes are silenced implication of this is unknown, similar observations during transformation of breast epithelium and may have been made in lymphocytes where the initiation of therefore contribute to development of breast cancer lymphocyte proliferation and differentiation is coupled (He et al., 2003; Sutherland et al., 2004). The data to a downregulation of SOCS3 (Yu et al., 2004). presented here does not support that claim. Compared Interestingly, the downregulation of SOCS1 following to non-malignant mammary epithelial cells, expression stimulation with IGF-1 is in stark contrast to the of SOCS2, SOCS3, SOCS5, SOCS6, SOCS7 and/or response to IFNg. Whereas IGF-1 represses transcrip- SOCS1 genes is elevated in MCF-7 and HCC1937, two tion of SOCS1 (Figure 4e), the SOCS1 gene is trans- cell lines that serve as prototypic breast-cancer cell activated by IFNg (Figure 3). types. Both SOCS1 and SOCS3 promoters are shown to

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1945

Figure 4 Growth factor-induced effects on STAT3 pathways in normal and breast cancer cells correlate temporally with downregulation of SOCS1 and SOCS3 genes transcription. (a) Western blot analysis of whole-cell extracts (40 mg/lane) from IGF-1- stimulated MCF-7 or HCC1937 cells. (b)EMSA(b) and supershift (c). C1, C2 or C3, indicate retarded DNA-protein binding complex and super-shifted bands are indicated by arrowheads. aSTAT3*, indicates Abs from another commercial source. Details are published as supporting information on the Oncogene website (d) Real-time RT–PCR analysis of SOCS expression in stimulated or unstimulated MCF-7, HCC1937 or AG11134 cells. (e) Real-time RT–PCR data are shown as fold increases in SOCS1 or SOCS3 mRNA in stimulated compared to non-stimulated cells.

be highly responsive to regulation by cytokine or growth factor signals, even in breast-cancer cells with hyper- methylated CpG islands in SOCS1, SOCS2 and SOCS3 promoters (He et al., 2003; Sutherland et al., 2004). In addition, we have shown that the mechanism of SOCS- mediated feedback inhibition of cytokine signaling is not defective in MCF-7 and HCC1937 and comparable to normal breast cells. Our results showing overexpression of SOCS genes in breast-cancer cells is clearly at variance with another report suggesting that SOCS1 and SOCS2 genes are transcriptionally silent in breast- cancer cells (He et al., 2003; Sutherland et al., 2004). However, our data is in line with clinical findings of elevated SOCS1, SOCS2, SOCS3, CIS expression within in situ ductal carcinomas, infiltrating ductal carcinomas, human tumor tissues and in 10additional breast-cancer cell lines: In fact, CIS is elevated in all/ most breast-cancer lines and may drive proliferation by activation of ERK (Raccurt et al., 2003). Figure 5 Expression of SOCS1 and STAT1-induced growth The most compelling evidence in support of the regulatory proteins, IRF-1 and p21WAF1 is defective in BRCA1- notion that SOCS genes are silenced by DNA-hyper- deficient breast cancer cells. Western blot analysis of extracts methylation in breast-cancer cells is based on experi- from freshly isolated cell lines (a) or IFNg-stimulated or unstimulated HCC1937 cells (b). (c) Real-time RT–PCR quantification ments in which SOCS1 expression was re-activated in of relative abundance of IRF-1 or p21WAF1 mRNA transcripts in breast-cancer cell lines following treatment with 5-aza- g-irradiated AG11134 or HCC1937 cells. 20-deoxycytidine (Sutherland et al., 2004). However,

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1946 treatment of cancer cells with the methylase inhibitor, in the mammary gland (Le Provost et al., 2005). 5-aza-20-deoxycytidine, has been shown to inhibit Although SOCS3 is elevated in MCF-7 and HCC1937 growth of cancer cells by activating interferon/STAT1- (Figure 1a), it does not exert efficient feedback-regula- signaling pathways (Missiaglia et al., 2005). Thus, tion on STAT3-signaling as evidenced by constitutive enhanced expression of SOCS1 or 5-aza-20-deoxycyti- activation of STAT3 in these breast-cancer cells dine-induced growth suppression of breast-cancer cells (Figure 4b). Interestingly, stimulation of MCF-7 or noted in the study by Sutherland et al. may well have HCC1937 by IGF-1 activates both STAT1, as well as, derived from activation of STAT1, a transcription STAT3 and whereas activated STAT3 is capable of factor that is the primary inducer of SOCS1 (Hilton, binding DNA, DNA-binding by activated STAT1 is 1999). Discerning the exact mechanism responsible for inhibited (Figure 4d). A similar finding has been made in re-activating SOCS1 expression in the 5-aza-20-deox- a T-cell lymphoma (Brender et al., 2005). In that study, ycytidine-treated cells is further confounded by the fact malignant cells expressed high levels of SOCS3 but were that transcription factors that regulate the SOCS1 gene unable to inhibit and terminate STAT3-signaling. may also be silenced by DNA-hypermethylation in Instead, SOCS3 functioned as a tumor promoter by breast-cancer cells. For example, the BRCA1 gene that inhibiting IFNa/STAT1-mediated growth inhibition has been implicated in breast tumorigenesis (Welcsh without affecting cell growth or apoptosis (Brender et al., 2002) is also a target of DNA methylases (Wei et al., 2005). Our results suggest that elevated SOCS3 et al., 2005) and hypermethylation of the BRCA1 gene may also function as a tumor promoter in breast cancer occurs in 7–31% of sporadic breast and ovarian cancers and inability of SOCS3 to terminate STAT3-signaling (Catteau and Morris, 2002). It is therefore, difficult to in MCF-7 and HCC1937 cells is consistent with the rule out that the re-activation of SOCS1 expression by association of constitutive STAT3 activation with 5-aza-20-deoxycytidine results from demethylation of primary tumors of high-risk breast cancer patients other silenced genes that constitute the CpG island (Gritsko et al., 2006). methylator phenotype in breast cancer (Issa, 2004). Although DNA-hypermethylation is widely recog- The etiology of breast cancer is still not well under- nized as a common feature of human neoplasia (Issa, stood. Although overexpression or silencing of SOCS 2004), this epigenetic mechanism also occurs as part of genes may be involved, the most common genetic lesion normal aging process and most often, CpG islands that associated with development of breast cancer is germ- are methylated in cancer cells are also methylated in line mutation or epigenetic silencing of the BRCA1 gene normal cells during aging (Toyota et al., 1999). Our data (Cui et al., 1998; Aprelikova et al., 2001). BRCA1 is a suggest that DNA-hypermethylation of SOCS genes potent transcriptional activator with roles in DNA observed in breast carcinoma may be similar to the damage repair, cell cycle checkpoint control and hypermethylation of SOCS3, P16, p21WAF1 or PTEN apoptosis (Andrews et al., 2002; Welcsh et al., 2002) gene in advanced ovarian cancer (Teodoridis et al., and has recently been shown to collaborate with STAT1 2005). In that study, the hypermethylation had no in transcriptional activation of growth-inhibitory and effects on the biological activities of these proteins. pro-apoptotic genes (Ouchi et al., 2000). As transcrip- Analysis of SOCS expression in BRCA1-deficient tion of the SOCS1 gene is regulated by IRF-1 (Saito HCC1937 is particularly instructive as it reveals that et al., 2000) and expression of IRF family of transcrip- transcriptional-silencing of SOCS1 gene derives from tion factors require synergism between STAT1 and defects in the transcription of a transcription factor BRCA1 (Andrews et al., 2002), we used the BRCA1- upstream of SOCS1 activation pathway and not due to deficient HCC1937 to investigate whether expression of DNA-hypermethylation. On the other hand, our data SOCS1 gene requires BRCA1. Analysis of HCC1937 suggests that increased expression of SOCS genes is a cells reveals that IRF-1 expression is defective specific lesion of breast-cancer cells and may derive from (Figure 5a) and may account for low SOCS1 expression constitutive activation of STAT3 pathways (Figure 4b). in this breast-cancer cell line (Figure 1a). However, Additional studies are needed to firmly establish that treatment of HCC1937 with IFNg significantly enhances elevated endogenous SOCS expression in breast cancer IRF-1 and SOCS1 expression (Figures 2, 3b and 5b), derives from aberrant host/tumor response to autocrine/ indicating that optimal expression of the IRF-1 and paracrine stimulation of STAT3 pathways. However, its SOCS1 genes requires BRCA1 as a co-activator of consequence appears to be in conferring resistance to STAT1. This requirement can however, be overcome by proinflammatory cytokines by shutting down STAT1 increasing the intensity of STAT1-signaling in HCC1937 signaling pathway that mediate essential functions in cells. In fact, expression of p21WAF1, another IFNg- immune-surveillance. inducible gene, is also defective in HCC1937 cells (Figure 5a), suggesting that the low SOCS1 expression in HCC1937 maybe due to global defects in BRCA1/ STAT1-dependent transcription. Materials and methods In AG11134 cells, SOCS3 mediates negative-feedback Cells and culture conditions regulation of IGF-1 signaling and following termination The breast-cancer cell lines (MCF-7, HCC1937), prostate of growth signal SOCS3 expression returns to its basal cancer cell lines (LNCaP, PC-3 and DU-145), non-malignant level, consistent with its role in inhibiting JAK-kinases prostate cell line (RWPE-1) and BRCA1À/ þ lymphoblastoid and promoting apoptosis of differentiated epithelial cells cell line (HCC1937BL) were obtained from American Type

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1947 Culture Collection (Manassas, VA, USA). The normal human incubation at 371C, media was aspirated and cells were mammary epithelial cells AG11134 were from the Aging Cell g-irradiated at 10Gy in a Gammacell 40 Exactor 137Cs Repository (National Institute of General Medical Sciences g-radiation source (Nordion, Ontario, Canada). Subsequently, Coriell) (details relating to culture conditions for the cell lines the cells were incubated in their respective media at 371C for is published as supporting information on the Oncogene 2 h. Unirradiated cells were used as controls. website).

Quantitative real-time PCR analysis Western blot analysis and electrophoretic mobility shift assay Real-time 50-nuclease fluorogenic RT–PCR analysis was Western blotting and electrophoretic mobility shift assay performed on an ABI 7700 (PE Biosystems, Foster City, CA, (EMSA) were performed as previously described (Yu et al., USA) using DNase-treated RNA, SuperScript II Reverse 2004). Brief description of the procedures and sources of Transcriptase (RT) (Invitrogen Life Technologies, Gaithers- antibodies and probes used are published as supporting information on the Oncogene website. burg, MD, USA), and oligo(dT)12–16 as previously described (Yu et al., 2004) (primers, probes and details of the quantitative real-time PCR procedure are published as Statistical analysis supporting information on the Oncogene website). The Student’s t-test was performed on the data presented. P-values p0.05 are considered statistically significant and Ribonuclease protection assay indicated by an asterisk (*) in Figures 1b and 5c. Ribonuclease protection assay (RPA) was performed with RNA (10 mg), [a-32P]-UTP radiolabeled RNA probes and human SOCS RPA (BD Biosciences, San Diego, CA, Acknowledgements USA) as recommended by manufacturer. This research was funded by the Intramural Research Exposure of HCC1937 cells to ionizing radiation Programs of the NIH, NIA and NEI. Authors thank Cells were grown to approximately 70% confluency and re- Dr Simon Nyaga for technical assistance and critical reading plated at a density of 3 Â 106 cells per dish in media containing of manuscript and Dr Patrice Morin for critical reading of 0.5% albumin but without growth factors. After 12 h the manuscript.

References

Andrews HN, Mullan PB, McWilliams S, Sebelova S, Gritsko T, Williams A, Turkson J, Kaneko S, Bowman T, Quinn JE, Gilmore PM et al. (2002). BRCA1 regulates the Huang M et al. (2006). Persistent activation of interferon gamma-mediated apoptotic response. J Biol Chem signaling induces survivin gene expression and confers 277: 26225–26232. resistance to apoptosis in human breast cancer cells. Clin Aprelikova O, Pace AJ, Fang B, Koller BH, Liu ET. (2001). Cancer Res 12: 11–19. BRCA1 is a selective co-activator of 14-3-3 sigma gene- Haura EB, Turkson J, Jove R. (2005). Mechanisms of disease: transcription in mouse embryonic-stem cells. J Biol Chem Insights into the emerging role of signal transducers and 276: 25647–25650. activators of transcription in cancer. Nat Clin Pract Oncol 2: Bowman T, Garcia R, Turkson J, Jove R. (2000). STATs in 315–324. oncogenesis. Oncogene 19: 2474–2488. He B, You L, Uematsu K, Zang K, Xu Z, Lee AY et al. (2003). Brender C, Lovato P, Sommer VH, Woetmann A, Mathiesen SOCS-3 is frequently silenced by hypermethylation and AM, Geisler C et al. (2005). Constitutive SOCS-3 expression suppresses cell growth in human lung cancer. Proc Natl Acad protects T-cell lymphoma against growth-inhibition by Sci USA 100: 14133–14138. IFNalpha. Leukemia 19: 209–213. Hilton DJ. (1999). Negative regulators of cytokine signal Campbell IL. (2005). Cytokine-mediated inflammation, transduction. Cell Mol Life Sci 55: 1568–1577. tumorigenesis, and disease-associated JAK/STAT/SOCS Issa JP. (2004). CpG island methylator phenotype in cancer. signaling circuits in the CNS. Brain Res Brain Res Rev 48: Nat Rev Cancer 4: 988–993. 166–177. Le Provost F, Miyoshi K, Vilotte JL, Bierie B, Robinson GW, Catteau A, Morris JR. (2002). BRCA1 methylation: a Hennighausen L. (2005). SOCS3 promotes apoptosis of significant role in tumour development? Semin Cancer Biol mammary differentiated cells. Biochem Biophys Res Commun 12: 359–371. 338: 1696–1701. Clevenger CV. (2004). Roles and regulation of stat family Medina D. (2005). Mammary developmental fate and breast transcription factors in human breast cancer. Am J Pathol cancer risk. Endocr Relat Cancer 12: 483–495. 165: 1449–1460. Missiaglia E, Donadelli M, Palmieri M, Crnogorac-Jurcevic T, Coccia EM, Del-Russo N, Stellacci E, Orsatti R, Benedetti E, Scarpa A, Lemoine NR. (2005). Growth delay of human Marziali G et al. (1999). Activation and repression of the pancreatic-cancer-cells by methylase inhibitor 5-aza-20- 2-5A synthetase and p21 gene promoters by IRF-1 and deoxycytidine-treatment is associated with activation of the IRF-2. Oncogene 18: 2129–2137. interferon signalling pathway. Oncogene 24: 199–211. Cui JQ, Shao N, Chai Y, Wang H, Reddy ES, Rao VN. (1998). Niwa Y, Kanda H, Shikauchi Y, Saiura A, Matsubara K, BRCA1 splice variants BRCA1a and BRCA1b associate Kitagawa T et al. (2005). Methylation-silencing of SOCS-3 with CBP co-activator. Oncol Rep 5: 591–595. promotes cell growth and migration by enhancing JAK/ Darnell Jr JE. (1997). STATs and gene regulation. Science 277: STAT and FAK-signalings in human hepatocellular carci- 1630–1635. noma. Oncogene 24: 6406–6417. Darnell JE. (2005). Validating Stat3 in cancer therapy. Nat Ouchi T, Lee SW, Ouchi M, Aaronson SA, Horvath CM. Med 11: 595–596. (2000). Collaboration of signal transducer and activator of

Oncogene SOCS genes are not silenced in breast-cancer cells MK Evans et al 1948 transcription-1 (STAT1) and BRCA1 in differential-regula- Teodoridis JM, Hall J, Marsh S, Kannall HD, Smyth C, Curto J tion of IFN-gamma target genes. Proc Natl Acad Sci USA et al. (2005). CpG-island methylation of DNA damage 97: 5208–5213. response genes in advanced ovarian cancer. Cancer Res 65: Raccurt M, Tam SP, Lau P, Mertani HC, Lambert A, Garcia- 8961–8967. Caballero T et al. (2003). Suppressor of cytokine-signalling Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, gene expression is elevated in breast-carcinoma. Br J Cancer Baylin SB, Issa JP. (1999). CpG-island methylator 89: 524–532. phenotype in colorectal cancer. Proc Natl Acad Sci USA Rottapel R, Ilangumaran S, Neale C, La-Rose J, Ho JM, 96: 8681–8686. Nguyen MH et al. (2002). The tumor suppressor activity of Weber A, Hengge UR, Bardenheuer W, Tischoff I, Sommerer F, SOCS-1. Oncogene 21: 4351–4362. Markwarth A et al. (2005). SOCS-3 is frequently methy- Saito H, Morita Y, Fujimoto M, Narazaki M, Naka T, lated in head and neck squamous-cell carcinoma and its Kishimoto T. (2000). IFN regulatory factor-1-mediated transcrip- precursor lesions and causes growth inhibition. Oncogene 24: tional activation of mouse STAT-induced STAT inhibitor-1 6699–6708. gene promoter by IFN-gamma. J Immunol 164: 5833–5843. Wei M, Grushko TA, Dignam J, Hagos F, Nanda R, Sveen L Sultan AS, Xie J, LeBaron MJ, Ealley EL, Nevalainen MT, et al. (2005). BRCA1 promoter methylation in sporadic Rui H. (2005). Stat5 promotes homotypic adhesion and breast cancer is associated with reduced BRCA1 copy- inhibits invasive characteristics of human breast-cancer cells. number and -17 aneusomy. Cancer Res 65: Oncogene 24: 746–760. 10692–10699. Sutherland KD, Lindeman GJ, Choong DY, Wittlin S, Welcsh PL, Lee MK, Gonzalez-Hernandez RM, Black DJ, Brentzell L, Phillips W et al. (2004). Differential hyper- Mahadevappa M, Swisher EM et al. (2002). BRCA1 methylation of SOCS genes in ovarian and breast-carcino- transcriptionally regulates genes involved in breast tumori- mas. Oncogene 23: 7726–7733. genesis. Proc Natl Acad Sci USA 99: 7560–7565. Tanaka N, Ishihara M, Lamphier MS, Nozawa H, Yu CR, Mahdi RM, Ebong S, Vistica BP, Chen J, Guo Y et al. Matsuyama T, Mak TW et al. (1996). Cooperation of the (2004). Cell proliferation and STAT6 pathways are nega- tumour suppressors IRF-1 and p53 in response to DNA tively-regulated in T-cells by STAT1 and suppressors of damage. Nature 382: 816–818. cytokine-signaling. J Immunol 173: 737–746.

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

Oncogene