Oncogene (2009) 28, 2034–2045 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE is oncogenic for human mammary carcinoma cells

J Kang1, JK Perry1, V Pandey1, GC Fielder1, B Mei2,3, PX Qian2,ZSWu4, T Zhu2, DX Liu1 and PE Lobie1,5

1The Liggins Institute, University of Auckland, Auckland, New Zealand; 2Hefei National Laboratory for Physical Sciences at Microscale and School of Life Sciences, University of Science and Technology of China, Hefei, Anhui, PR China; 3Institute of Basic Medicine, Anhui Medical University, Hefei, Anhui, PR China; 4Department of Pathology, Anhui Medical University, Hefei, Anhui, PR China and 5Department of Molecular Medicine and Pathology, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand

We report that artemin, a member of the glial cell line- All GFLs are potent (Airaksinen and derived neurotrophic factor family of ligands, is oncogenic Saarma, 2002). GDNF was identified as a trophic factor for human mammary carcinoma. Artemin is expressedin for midbrain dopaminergic neurons (Lin et al., 1993). It numerous human mammary carcinoma cell lines. Forced promotes survival of many types of neurons, including expression of artemin in mammary carcinoma cells results subpopulations of peripheral autonomic and sensory, as in increased anchorage-independent growth, increased well as central motor, dopamine and noradrenaline, colony formation in soft agar andin three-dimensional neurons (Airaksinen et al., 1999). GDNF, and Matrigel, andalso promotes a scatteredcell phenotype artemin all support the survival of peripheral sympathetic with enhancedmigration andinvasion. Moreover, forced and sensory neurons, as well as midbrain dopamine expression of artemin increases tumor size in xenograft neurons, whereas supports central nervous models and leads to highly proliferative, poorly differ- system dopamine and motor neurons, but not peripheral entiatedandinvasive tumors. Expression datain Onco- neurons (Airaksinen et al., 1999). mine indicate that high artemin expression is significantly GDNF family ligands signal through a multi- associated with residual disease after chemotherapy, component receptor complex, which includes a glyco- metastasis, relapse and death. Artemin is detect- syl-phosphatidylinositol (GPI)-anchored co-receptor a able in 65% of mammary carcinoma andits expression (termed GFRa) as a ligand-binding component and the correlates to decreased overall survival in the cohort of RET receptor as a common signaling patients. Depletion of endogenous artemin with small component (Takahashi, 2001). Binding of GFLs to a interfering RNA, or antibody inhibition of artemin, specific GFRa determines the signaling specificity of decreases the oncogenicity and invasiveness of mammary the receptor complex: GDNF preferentially binds to carcinoma cells. Artemin is therefore oncogenic for GFRa1, neurturin to GFRa2, artemin to GFRa3 and human mammary carcinoma, andtargetedtherapeutic persephin to GFRa4 (Airaksinen and Saarma, 2002). approaches to inhibit artemin function in mammary However, these binding specificities are not exclusive. carcinoma warrant consideration. For example, GDNF can also bind to GFRa2 and Oncogene (2009) 28, 2034–2045; doi:10.1038/onc.2009.66; GFRa3 and artemin to GFRa1 (Bespalov and Saarma, published online 13 April 2009 2007). Apart from the GFRa–RET signaling mechan- ism, RET-independent signaling pathways have also Keywords: artemin; GDNF; mammary; carcinoma; been described (Poteryaev et al., 1999; Trupp et al., 1999; oncogenicity Meng et al., 2001b; Paratcha et al., 2003; Popsueva et al., 2003; Sariola and Saarma, 2003; Enomoto et al., 2004). GDNF, for example, has been shown to activate cytoplasmic Src family tyrosine kinases through GFRa1 Introduction in RET-deficient cells (Poteryaev et al., 1999). Some evidence also links GFLs to cancer. Targeted Glial cell line-derived neurotrophic factor (GDNF) expression of GDNF in undifferentiated spermatogonia family ligands (GFL), a distant subgroup of the promotes malignant testicular tumors (Meng et al., transforming b superfamily, include four 2000, 2001a). GDNF has also been reported to promote members: GDNF, neurturin, artemin and persephin tumor cell invasion in pancreatic cancer cell lines (Airaksinen et al., 1999; Airaksinen and Saarma, 2002). (Okada et al., 1999; Zhu et al., 2001; Veit et al., 2004). GDNF and artemin are expressed in pancreatic cancer Correspondence: Professor PE Lobie, The Liggins Institute, University and influence neural invasion of the tumor, contributing of Auckland, 2-6 Park Avenue, Private Bag 92019, Auckland, North to cancer cell spreading along pancreatic nerves (Ito Island 1010, New Zealand. et al., 2005; Ceyhan et al., 2006, 2007). Furthermore, E-mail: [email protected] Received 9 July 2008; revised 24 February 2009; accepted 4 March 2009; mutation of the common GFL receptor RET results in published online 13 April 2009 its constitutive activation, leading to thyroid cancer, Oncogenic properties of artemin J Kang et al 2035 pheochromocytoma and multiple endocrine neoplasia tion (RT)–PCR. As observed in Figure 1b, artemin was type 2 (MEN2) (MEN2A and MEN2B) (Pasini et al., expressed in all the cancer cell lines examined. Highest 1996; Saarma and Sariola, 1999; Ichihara et al., 2004). expression was detected in DLD-1 colon, PC3 prostate We demonstrate herein that forced expression of and RL95-2 endometrial carcinoma cell lines. Moderate artemin in mammary carcinoma cells increased onco- expression was observed in the BxPC3 pancreas genicity, resulted in increased invasion and increased carcinoma cell line and in four breast cancer cell lines, tumor growth in xenograft models. Depletion or of which two (T47Dand MCF-7) are estrogen receptor functional inhibition of artemin in mammary carcinoma (ER) þ and two (BT549 and MDA-231) are ERÀ. cells concordantly reduced their oncogenic properties. Limited, but detectable, expression was observed in Artemin expression in breast cancer predicted metas- A549 lung, AGS stomach, Colo320DM colon, DU145 tasis, residual disease after chemotherapy, recurrence prostate and HepG2 liver cancer cell lines. and death. We thereby provide evidence that artemin is The expression of GFRa isoforms and RET was also important in human mammary neoplastic processes. determined by RT–PCR in MCF-7, T47Dand BT549 mammary carcinoma cells (Figure 1c). Expression of GFRa1, GFRa3 and RET was easily detected in MCF-7 Results cells, whereas expression of GFRa2 was weak and required more cycles to detect. GFRa4 was not detected Expression of artemin in human tissues and carcinoma in any of the three cell lines. The expression pattern of cell lines these receptors in T47Dand BT549 cells was quite similar We first determined the expression profile of artemin in to that in MCF-7 cells, except that T47Dexhibited a normal human tissues by the use of a commercially much lower level of GFRa1 expression and BT549 available panel of cDNAs (Figure 1a). Screening for exhibited considerably higher expression of GFRa2. mRNA expression by PCR with primers specific for human artemin revealed that the artemin is also Effect of forced expression of artemin on mammary expressed in a number of non-neural tissues, with carcinoma cell number highest expression observed in the cerebellum followed A cell model system was established to determine the by the colon. Expression was also detected in the functional consequences of artemin expression in prostate, uterus, stomach, kidney, trachea, fetal brain, mammary carcinoma cells. MCF-7 cells were stably adipose and cartilage. Artemin was not expressed or transfected with the artemin expression plasmid pIR- expressed at a low level in other remaining tissues. ESneo3-ARTN (designated as MCF7-ARTN) or with The expression of artemin in a number of human the empty vector pIRESneo3 (designated as MCF7- cancer cell lines was also examined by reverse transcrip- Vec). The forced expression of artemin in MCF7-ARTN

Figure 1 mRNA expression patterns of artemin in human normal tissues and carcinoma cell lines. (a) Artemin (ARTN) was amplified by PCR with specific primers in a panel of cDNAs derived from different human tissues (Primgen, Bothell, WA, USA). The tissue of origin is indicated above each lane. The b2-microglobulin (b2M) gene was used as the cDNA input control. (b) The expression of artemin in human cancer cell lines was examined by RT–PCR with the same artemin-specific primer pairs. b-actin was included as the RNA input control. (c) The expression of GFRa1–4 and RET in three breast cancer cell lines by reverse transcription (RT)–PCR as in b. The number of PCR cycles is indicated on the right. The bp sizes of amplified products are shown on the left.

Oncogene Oncogenic properties of artemin J Kang et al 2036 cells was verified at both the mRNA (Figure 2a) and 7 cells did not affect entry into S-phase, as assessed by protein (middle panel, Figure 2b) levels compared with bromodeoxyuridine (BrdU) incorporation (Figure 2e). control MCF7-Vec cells. Forced expression of artemin also resulted in increased artemin protein in conditioned media (top panel, Figure 2b). Forced expression of artemin increased cell survival We first determined whether forced expression of The effect of forced expression of artemin on the artemin affected total mammary carcinoma cell number survival of MCF-7 cells in serum-deprived conditions in monolayer culture. Forced expression of artemin in was examined by two different methods. Estimation of MCF-7 cells did not significantly increase cell number apoptosis with Hoechst 33258 dye (Figure 2f) showed a under serum-replete conditions over a period of 8 days significant reduction in apoptosis in MCF7-ARTN cells (Figure 2c). However, in serum-reduced (0.2%) media, versus MCF7-Vec cells (Po0.05). Phosphatidylserine MCF7-ARTN cells increased in number significantly externalization is a unique marker for apoptosis; faster than did MCF7-Vec cells (Figure 2d), with a 23% labeling of cells with fluorescein isothiocyanate-conju- increase in total cell number relative to MCF7-Vec after gated annexin V and propidium iodide (PI) revealed that 8 days (Po0.05). Forced expression of artemin in MCF- MCF7-ARTN cells were resistant to serum-deprivation-

MCF7-Vec MCF7-ARTN MCF7-Vec MCF7-ARTN Media - 28 kDa ARTN - 140 bp ARTN Lysate - 28 kDa β -actin - 580 bp β -actin - 43 kDa

SERUM SERUM FREE 20 15 Vector Vector 15 ARTN ARTN 10 * 10 5 5 Cell number (10,000X) Cell number (100,000X) 0 0 02468 02468 Days Days

10 20 AV AV & PI * 10 8 ** 15 8 6 10 6 3.7 4 4 1.4 5 2 2 4.4 3.5 Apoptotic cells (%) BrdU positive cells (%) Apoptotic cells (%) 0 0 0 Vector ARTN Vector ARTN Vector ARTN Figure 2 Characterization of the effects of forced expression of artemin in mammary carcinoma MCF-7 cells. Cells were stably transfected with plasmids expressing artemin (MCF7-ARTN) or the empty vector plasmid (MCF7-Vec) as a control, as indicated. (a) Total RNA was isolated and the mRNA levels of artemin (ARTN)andb-actin were detected by reverse transcription (RT)–PCR. The sizes of amplified products are shown on the right. (b) Western blot analyses of forced expression and secretion of artemin in MCF-7 cells. Soluble whole cellular extracts or concentrated conditioned media were run on an SDS–PAGE and immunoblotted using a goat anti-artemin polyclonal antibody. b-actin was used as loading control for cell lysates. The sizes of detected protein bands in kDa are shown on the right. (c and d) Total cell number assays. MCF7-ARTN (ARTN) and MCF7-Vec (Vector) cells were seeded with (c) 10% FBS (SERUM) or (d) low serum (0.2%) (SERUM FREE). (e) S-phase entry: The percentage of bromodeoxyuridine (BrdU)-positive cell nuclei relative to the total number of cell nuclei in MCF7-Vec and MCF7-ARTN cells was determined. (f and g) Apoptosis: MCF7-ARTN (ARTN) and MCF7-Vec (Vector) cells were fixed and were stained for the assessment of apoptosis either by Hoechst 33258 dye (f) or with fluorescein isothiocyanate-conjugated annexin V (AV) and propidium iodide (PI) (g). Both early (AV-positive and PI-negative; open box) and late (AV- and PI-positive; shaded box) apoptosis are presented with the percentage. *Po0.05; **Po0.01.

Oncogene Oncogenic properties of artemin J Kang et al 2037 induced apoptosis compared with control MCF7-Vec number of MCF7-ARTN cells spread from the main cells (Figure 2g). There was little difference in terms of bulk of the colony, suggesting that artemin promoted late apoptosis (annexin V-positive and PI-positive), local invasive behavior. 4.4% in MCF7-Vec cells and 3.5% in MCF7-ARTN cells. However, early apoptosis (annexin V-positive and Forced expression of artemin enhanced cell migration and PI-negative) was decreased from 3.7% for MCF7-Vec invasion cells to 1.4% for MCF7-ARTN cells, suggesting that Fluorescence microscopy to visualize filamentous actin forced expression of artemin preferentially protected revealed that MCF7-Vec cells grew in clumps with cells from early apoptosis. prominent cell-to-cell contact typical of an epithelial phenotype, whereas MCF7-ARTN cells were scattered Forced expression of artemin increased BCL2 gene and displayed little intercellular contact (Figure 3d). The expression effects of artemin on cell motility were therefore Real-time PCR (see Supplementary data 2) was evaluated, first by the use of the standard Transwell performed to determine the effect of forced expression chamber (BDFalcon, Franklin Lakes, NJ, USA) and of artemin on the relative expression levels of key subsequently by wound healing assay. Twice as many involved in cell proliferation and survival. Forced MCF7-ARTN cells migrated through Transwell inserts expression of artemin did not alter or minimally altered in 24 h than did control MCF7-Vec cells (Figure 3e). In the expression of most cell cycle-related markers, Transwell chambers coated with Matrigel, the invasive including cyclin D1 and E1, ATM, CDC25A,and capacity of MCF7-ARTN cells over 48 h was enhanced CDK 2 and 4, but did significantly decrease the 1.5-fold compared with control MCF7-Vec cells expression of cyclin-dependent kinase inhibitor-2A (Figure 3e). In the wound-healing assay (Figure 3f), (CDKN2A) (Supplementary Table S3, Supplementary MCF7-ARTN cells in monolayer culture showed data 2). CDKN2A, encoding p16(INK4a), is a tumor complete wound healing within 72 h, whereas MCF7- suppressor gene that induces G1 cell cycle arrest by Vec cells did not. Forced expression of artemin therefore inhibiting the phosphorylation of Rb protein by the enhanced the motility and invasive capacity of MCF-7 cyclin-dependent kinases, CDK4 and CDK6 (Stott cells. Concordant with this observation, real-time PCR et al., 1998). The expression of BCL2 in MCF7-ARTN showed that forced expression of artemin significantly cells was increased 8.4-fold compared with control increased the expression of a number of genes involved MCF7-Vec cells. BCL2 is a well-described cell survival in invasion and metastasis, including MMP1 (matrix gene (Zinkel et al., 2006). Another gene significantly metalloproteinase 1; by 7.6-fold), PLAU (3.8-fold) and increased by artemin was telomerase-reverse trans- SERPINE1 (3.1-fold) (Supplementary Table S3, Sup- criptase (TERT). TERT, in addition to its role in plementary data 2). immortalization, also suppresses apoptosis (Zhang et al., 2003). It is noted that g-synuclein (SNCG, also Forced expression of artemin promoted MCF-7 tumor referred as breast cancer-specific gene 1) was upregu- xenograft growth in vivo lated 1.67-fold by artemin. SNCG has been observed to To determine whether artemin enhances tumorigenicity protect tumor cells from apoptosis and to stimulate cell in vivo, we used a xenograft model of MCF-7 cells motility and metastasis (Wu et al., 2003). injected into the mammary fat pad of immunodeficient nude mice. Both MCF7-ARTN and MCF7-Vec cells Forced expression of artemin promoted anchorage- formed palpable and measurable tumors, but tumors independent growth formed by MCF7-ARTN were larger than those formed The effect of forced expression of artemin on anchorage- by control MCF7-Vec cells (Figure 4a). Statistically independent growth was examined in suspension culture significant enhancement of tumor volume was achieved and by colony formation in soft agar. MCF7-ARTN on week 2 (P ¼ 0.023), and tumors formed by MCF7- cells increased in number faster than did control MCF7- ARTN cells were more than double the size of tumors Vec cells in suspension culture (Figure 3a), with a 42% formed by MCF7-Vec cells after 6.5 weeks. increase in total cell number relative to MCF7-Vec after MCF-7 is a rather well-differentiated tumor cell line, 10 days (Po0.01). In soft agar, the number of colonies as reflected by its estrogen-dependent growth. Histology formed by MCF7-ARTN cells was increased twofold with hematoxylin and eosin staining showed that tumors (Po0.001) compared with MCF7-Vec cells (Figure 3b). formed by MCF7-Vec cells were well differentiated (left MCF7-ARTN cells and control MCF7-Vec cells were panel, Figure 4b), whereas those formed by MCF7- also cultured in three-dimensional Matrigel. Concor- ARTN cells were poorly differentiated (right panel, dant with colony formation in soft agar, an increase in Figure 4b). Unlike control MCF7-Vec cells, nuclei of colony number was observed for MCF7-ARTN cells which were regularly shaped and uniform with a clear compared with MCF7-Vec cells (data not shown). cytoplasmic boundary, MCF7-ARTN cells were small Moreover, the colonies formed by MCF7-ARTN cells spindle-shaped pleomorphic cells with irregular hyper- in Matrigel were larger and the cells in those colonies chromatic nuclei, indistinct cytoplasmic borders and were more motile and invasive compared with colonies scanty cytoplasm. formed by MCF7-Vec cells (Figure 3c). MCF7-Vec To quantify proliferation and apoptosis within the cells produced circumscribed colonies, whereas a large tumor, BrdU incorporation and TUNEL (terminal

Oncogene Oncogenic properties of artemin J Kang et al 2038 SOFT AGAR Vector 20 1400 *** ARTN 1200 15 ** 1000 800 10 600 5 400 Colony number 200 Cell number (10,000X) 0 0 0246810 Vector ARTN Days

Day 6 Day 12 Day 24 Vector ARTN MCF7-ARTN MCF7-Vec

0h 48h 72h Vector ARTN 500 ** ** 400 Vector 300

200

Cell number 100

0 ARTN Migration Invasion (24h) (48h)

Figure 3 Forced expression of artemin significantly promotes anchorage-independent growth and enhances cell migration and invasion. (a) Suspension culture. MCF7-ARTN (ARTN) and MCF7-Vec (Vector) cells were cultured in plates precoated with poly- HEMA. At the indicated times, viable cells were counted using a hemocytometer. (b) Soft agar assay: Cells were seeded in 0.35% agarose and colonies formed were counted after incubation for 14 days. (c) Representative photomicrographic pictures of colonies formed by the cells growing in Matrigel. (d) Filamentous actin architecture of MCF7-ARTN and MCF7-Vec cells revealed by FITC- phalloidin. (e) Migration and invasion of MCF7-Vec and MCF7-ARTN cells were determined by use of Transwell inserts as described in Materials and methods. (f) Wound healing assay: ‘Scratch’ wounds were created in vitro by scraping the cell monolayer with a sterile pipette tip. Photographs were taken at the time indicated. Bar, 100 mminc and f; and 10 mmind.**Po0.01; ***Po0.001.

doxynucleotidyl transferase-mediated dUTP nick end) Forced expression of artemin enhanced anchorage- labeling were carried out on tumor sections. Unlike independent growth, cell migration and invasion in T47D monolayer culture in vitro (Figure 2e), the MCF7- and BT549 cells ARTN tumors exhibited a higher percentage of BrdU- We also stably forced the expression of artemin in ER þ labeled nuclei compared with MCF7-Vec tumors T47Dand ER À BT549 cells (Figure 5a). Both T47D- (Figure 4c). TUNEL labeling to assess apoptosis also ARTN and BT549-ARTN cells with forced expression showed that artemin enhanced tumor cell survival, as of artemin exhibited significant increases in colony there were significantly fewer TUNEL-positive cells in formation in soft agar and in migration and invasion the MCF7-ARTN tumors compared with the MCF7- in Transwell assays compared with their respective Vec tumors (Figure 4d). control T47D-Vec and BT549-Vec cells (Figures 5b–d).

Oncogene Oncogenic properties of artemin J Kang et al 2039

500 *** Oncomine cancer microarray database to analyse the MCF7-Vec 450 expression profile of artemin in a variety of human MCF7-ARTN *** cancers. Please refer to Supplementary data 3 for 400 ** *** )

*** tabulation and discussion of the findings from this 3 350 *** database. *** *** 300 250 Immunohistochemical analysis of artemin in breast cancer *** 200 and correlation of artemin expression to overall survival

* Immunohistochemical analysis showed that artemin 150 protein expression in breast cancer tissue was signifi- Volume of tumor (mm 100 cantly increased compared with normal tissue 50 (Po0.001). In total, 65.4% (104 out of 159) of breast cancer samples exhibited positive artemin immunoreac- 0 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 tivity compared with only 21.4% (3 out of 14) for the Weeks after injection corresponding normal breast tissue (Supplementary Table S7, Supplementary data 4). As observed in Figures 6a–c, immunoreactivity was observed predomi- MCF7-Vec MCF7-ARTN nantly in the cytoplasm of tumor cells. Normal epithelial cells of mammary duct and acini and smooth muscle cells of blood vessels sometimes exhibited variable intensities of artemin immunoreactivity. Artemin expression was observed to be significantly correlated to higher tumor stages (P ¼ 0.032) and was also highly correlated to HER2/neu (c-erbB-2) positivity (P ¼ 0.001) (Supplementary Table S8, Supple- BrdU TUNEL mentary data 4). However, no association of artemin 25 40 ** ** expression was observed with age of the patient and 20 30 tumor size, nor grade, ER or progesterone receptor 15 status of the tumor. Significant correlation of increased 20 artemin expression to decreased overall survival of 10 patients was observed (P ¼ 0.021, Figure 6d). An 10 5 analysis of artemin expression in matched stage and Positive cells (%) Positive cells (%) 0 0 HER2 status in patients are presented in Supplementary Vector ARTN Vector ARTN Table S9, Supplementary data 4. Only in HER2-positive Figure 4 Effect of forced expression of artemin on in vivo growth patients, was artemin expression not significantly of MCF-7 xenografts in immunodeficient mice. MCF7-ARTN and correlated to the outcome. MCF7-Vec cells in Matrigel were transplanted into the mammary fat pad of immunodeficient mice. (a) Tumor volume in relation to the day of surgery is shown. (b) Histological staining of Depletion of endogenous artemin by siRNAsignificantly representative tumors with hematoxylin and eosin (H&E). (c and d) increased apoptotic cell death Cell proliferation was assessed by BrdU incorporation (c) at the Next, we depleted MCF-7 cells of endogenous artemin end of the experiment, and apoptosis was measured by TUNEL with small interfering RNA (siRNA) and examined the labeling (d). Bar, 10 mm. *P 0.05; **P 0.01; ***P 0.001. o o o effects on cellular behavior. To determine the depletion efficacy of the siRNA, artemin cDNA was inserted The effects were more pronounced in BT549 cells than within the 30UTR of an enhanced green fluorescent in T47Dcells. T47D-ARTN cells formed 35% more protein (EGFP) expression vector (pEGFP-ARTN) and colonies than did control T47D-Vec cells in soft the efficiency of siRNA in targeting the artemin agar, whereas BT549-ARTN cells formed 64% more transcript could be determined by examination of the colonies than did control BT549-Vec cells (Figure 5b). expression level of EGFP protein by fluorescence Enhancement of migration and invasion by forced microscopy. One of the artemin-specific siRNA con- expression of artemin amounted to 37 and 74%, structs effectively reduced the fluorescence of EGFP in respectively, in T47Dcells and to 272 and 169%, cells transiently transfected with pEGFP-ARTN com- respectively, in BT549 cells (Figures 5c and d). BT549 pared with the siRNA-negative control plasmid (data cells with forced expression of artemin exhibited a more not shown). The artemin-specific siRNA did not reduce scattered, and mesenchymal morphology compared with EGFP expression in pEGFP-transfected cells compared the control cells (Figure 5e). with the siRNA-negative control (data not shown). Two stable MCF-7 cell lines, MCF7-siRNA and MCF7-CK, Correlation of artemin expression with cancer type and were established by stable transfection of the artemin- clinical outcome specific siRNA plasmid (pSilencer-ARTN) and the To determine the clinical relevance of artemin expres- negative control plasmid (pSilencer-CK), respectively. sion in human mammary carcinoma, we first used the The artemin expression level in these two cell lines was

Oncogene Oncogenic properties of artemin J Kang et al 2040

Vector ARTN 600 *** ** RT-PCR: T47D-Vec T47D-ARTN BT549-Vec BT549-ARTN 500

ARTN - 140 bp 400 β -actin - 580 bp 300 WB: 200 ARTN - 28 kDa Colony number 100 β -actin - 43 kDa 0 T47D BT549 T47D BT549

T47D-Vec BT549-Vec BT549-Vec T47D-ARTN BT549-ARTN 700 * 3500 *** 600 3000 500 2500 ** 400 2000 * BT549-ARTN 300 1500 Cell number 200 Cell number 1000 100 500 0 0 Migration Invasion Migration Invasion

Figure 5 Forced expression of artemin significantly enhances anchorage-independent growth, cell migration and invasion in T47Dand BT549 cells. T47Dand BT549 cells were stably transfected with plasmids expressing artemin ( T47D-ARTN and BT549-ARTN)ortheempty vector plasmid (T47D-Vec and BT549-Vec) as controls, respectively. The expression of artemin (ARTN) in these cell lines was confirmed by reverse transcription (RT)–PCR and Western blot (a). The sizes of amplified RT–PCR products in bp or molecular weight in kDa of detected protein bands are shown on the right. Soft agar (b), migration and invasion (c and d) assays were carried out as in Figure 3. The morphology of BT549-ARTN and BT549-Vec cells is also shown (e). Bar, 10 mm. *Po0.05; **Po0.01; ***Po0.001.

determined by both RT–PCR and Western blot, inverse relationship between artemin expression and showing that endogenous expression of artemin was induction of apoptosis. reduced in MCF7-siRNA cells at both the mRNA and protein levels compared with the control MCF7-CK cells (Figure 7a). Depletion of endogenous artemin by siRNAsignificantly Depletion of endogenous artemin by siRNA did not inhibited anchorage-independent growth in MCF-7 cells affect MCF-7 cell number in serum-replete conditions, siRNA-mediated depletion of artemin in MCF-7 cells but significantly inhibited (by 20%) growth in serum- inhibited their anchorage-independent growth. The reduced (0.2%) conditions (Figures 7b and c). As number of colonies formed by MCF7-siRNA cells in expected, depletion of endogenous artemin by siRNA soft agar was significantly reduced (by 28.5%, Po0.001) significantly increased apoptotic cell death induced by compared with that of control MCF7-CK cells serum deprivation. On the basis of nuclear morphology (Figure 8a). Polyclonal antibody inhibition of artemin after Hoechst 33258 visualization, 13% of MCF7-CK also significantly reduced (35%, Po0.001) soft agar cells versus 20% of MCF7-siRNA cells were identified colony formation in MCF-7 cells compared with control as apoptotic after serum deprivation for 24 h IgG (data not shown). Similar results were also obtained (Figure 7d). Similar changes were observed when with a second polyclonal antibody generated in chicken apoptosis was assessed by annexin-V labeling. The (data not shown). When grown in three-dimensional proportion of apoptotic cells in MCF7-siRNA cells Matrigel, a decrease in colony number was also was twice that in MCF7-CK cells (Figure 7e). Of these observed for MCF7-siRNA cells compared with apoptotic cells, the ratio of early apoptosis to late MCF7-CK cells (data not shown). The morphology of apoptosis was approximately 1 to 2.47 (1.9 versus 4.7%) colonies formed by MCF7-siRNA cells in Matrigel for MCF7-siRNA cells and 1 to 3.71 (0.7 versus 2.6%) differed from that of MCF7-CK cells (Figure 8b). for MCF7-CK cells, providing further support for an Artemin-depleted MCF7-siRNA cells formed small and

Oncogene Oncogenic properties of artemin J Kang et al 2041 NORMAL INVASIVE DUCTAL CARCINOMA

INVASIVE DUCTAL CARCINOMA 1.0

0.8

0.6 p = 0.021* 0.4 Artemin positive

Overall survival 0.2 Artemin negative 0.0 0 102030405060 Months Figure 6 Increased expression of artemin protein in breast cancer and correlation of artemin expression with clinical outcome. (a–c) Immunohistochemical localization of artemin in normal breast tissue and in two cases of invasive ductal carcinoma. (d) Kaplan–Meier analysis of the correlation of artemin protein expression determined by immunohistochemistry to overall survival of breast cancer patients (for details, see Supplementary data 4). Bar, 10 mm. *Po0.05. shrunken colonies, whereas the control MCF7-CK cells artemin signaling pathways may therefore constitute a formed larger and more voluminous colonies. targeted therapeutic approach to ameliorate human mammary carcinoma. Targeting of artemin itself may be Depletion of endogenous artemin by siRNAand antibody preferable for a number of reasons. GFLs are promis- impaired cell migration and invasion cuous both in their binding to the four described GFRa We assessed whether depletion of endogenous artemin isoforms (Bespalov and Saarma, 2007) and in the would impair the motility and/or invasiveness of MCF-7 signaling receptors they use (RET, NCAM, Syndecan et al. cells. The numbers of cells that migrated through the and Met) (Takahashi, 2001; Paratcha , 2003; non-coated (migration) or Matrigel-coated (invasion) Sariola and Saarma, 2003). For example, GFRa1and porous filters of Transwell inserts were both significantly RET were co-expressed in 18% of mammary carcino- et al. reduced in artemin-depleted MCF7-siRNA cells com- mas studied (Esseghir , 2007), whereas 42% of pared with control MCF7-CK cells (Figure 8c). Arte- mammary carcinomas were GFRa1-positive and RET- negative. Thus, specific targeting of RET alone would min-depleted MCF7-siRNA cells also displayed a significant reduction in wound-induced migration, as not be efficacious in GFL-dependent mammary carci- wounds created in a monolayer of MCF7-siRNA cells noma. Similarly, specific targeting of the individual closed less rapidly than did wounds in monolayers of GFRa isoforms may not be efficacious in abrogating MCF7-CK cells (Figure 8d). Consistent with the effect GFL signaling in mammary carcinoma, as other GFRa isoforms may compensate for the inhibition allowing of siRNA, a rabbit polyclonal antibody raised against artemin inhibited MCF-7 cell invasion in a dose- escape and development of resistance. For example, expression of GFRa3 is increased in tamoxifen-resistant dependent manner (Figure 8e), with 51.3 and 74.6% reductions observed at antibody concentrations of 200 mammary carcinoma cells treated with gefitinib, which targets the EGF receptor (Hutcheson et al., 2006). and 400 mg/ml, respectively. Further increases in anti- body concentration did not enhance this effect. Similar Furthermore, it is not yet clear whether the GFRa isoforms are the only receptors/binding for results were also obtained with a second polyclonal antibody generated in chicken (data not shown). GFLs; GFLs may also bind to NCAM, albeit with low affinity (Sariola and Saarma, 2003), or require other cell surface proteins, such as heparin sulfate glycosamino- glycans (Sariola and Saarma, 2003), for signaling. Other Discussion proteins, such as GAS1 (Schueler-Furman et al., 2006), also share homology with GFRa and may function as Herein, we have identified artemin as a functionally and GFL-receptors/binding proteins. Further, at least from clinically validated molecule involved in human mam- data published to date, GFRa expression is not mary neoplastic progression. Functional inhibition of correlated with patient survival, although expression is

Oncogene Oncogenic properties of artemin J Kang et al 2042

SERUM 18 CK MCF7-CK MCF7-siRNA 16 RT-PCR: siRNA 14 ARTN - 140 bp 12 10 β 8 -actin - 580 bp 6 WB: 4

ARTN - 28 kDa Cell number (X100,000) 2 0 02468 β-actin - 43 kDa Days

SERUM FREE 25 AV & PI AV 16 ** CK 10 14 20 siRNA ** 12 * 8 15 10 6 1.9 8 10 6 4 0.7 4 Apoptotic cells (%) 5 2 4.7

Apoptotic cells (%) 2.6

Cell number (X10,000) 2 0 0 0 CK siRNA 02468 CK siRNA Days Figure 7 Depletion of endogenous artemin by siRNA in MCF-7 cells. (a) Reverse transcription (RT)–PCR and Western blot to show efficiency of siRNA-mediated depletion of artemin (ARTN) mRNA and protein. The sizes of amplified RT–PCR products in bp or molecular weight in kDa of detected protein bands are shown on the right. (b and c) Total cell number assays for MCF7-siRNA (siRNA) and MCF7-CK (CK) cells were carried out as described earlier with (b) 10% FBS (SERUM) or (c) low serum (0.2%) (SERUM FREE). (d and e) Apoptosis assays. Cells were fixed and were stained for the assessment of apoptosis either by Hoechst 33258 dye (d) or with fluorescein isothiocyanate-conjugated annexin V (AV) and propidium iodide (PI) (e). Both early (AV-positive and PI-negative; open box) and late (AV- and PI-positive; shaded box) apoptosis are presented with the percentage. *Po0.05; **Po0.01.

correlated with lymphovascular invasion and lymph the tumor microenvironment plays a major role in node metastasis of the carcinoma (Esseghir et al., 2007) . dictating tumor behavior and progression, as well as In contrast, artemin expression is highly and signifi- response to therapy (Lubaroff et al., 1980; Chung, 1993; cantly associated with multiple outcome variables in Pettaway et al., 1996; Singh et al., 2007). Indeed, mammary carcinoma including survival (see Supple- mammary stromal fibroblasts have been shown both mentary Tables S4 and S5, Supplementary data 3). in vitro and in vivo to modulate mammary carcinoma cell Although GDNF may exert similar functional effects in proliferation (Brouty-Boye and Raux, 1993; Sadlonova mammary carcinoma cells (Esseghir et al., 2007), it may et al., 2005). Therefore, the ARTN stimulated increase not be as clinically relevant as artemin because it in BrdU-positive cells in vivo, but not in vitro,is (currently) lacks the multiple highly significant correla- indicative that artemin also possesses stromal or tions (Supplementary Table S6, Supplementary data 3) paracrine interactions in vivo. Such in vitro versus with disease and outcome observed for artemin. Of the in vivo differences have been observed for other proteins in the GFL-GFRa axis, artemin may therefore oncogenic molecules (Anderson et al., 2001; Duxbury be the preferred therapeutic target. We have provided et al., 2004). In this regard, it is interesting that evidence that targeting of artemin with either specific syndecan, a GFL binding protein (Sariola and Saarma, siRNA or antibody will reduce the oncogenic character- 2003), shed by stromal fibrobalsts, stimulates human istics of mammary carcinoma cells. Consideration mammary carcinoma cell proliferation (Su et al., 2007). should therefore be given to the development of novel We note some discrepancies in the correlation of therapeutics targeting artemin for use in artemin- artemin mRNA expression to clinicopathological para- expressing and -dependent tumors. meters in the Oncomine database compared with the Unlike monolayer culture in vitro MCF7-ARTN immunohistopathological data for artemin generated xenografts exhibited a higher percentage of BrdU- herein. This discrepancy may simply be indicative of labeled nuclei compared with those derived from differences inherent in measurement of artemin mRNA MCF7-Vec cells. It has been widely recognized that versus artemin protein and compounded by the fact that

Oncogene Oncogenic properties of artemin J Kang et al 2043 SOFT AGAR Day 7 Day 13 Day 18 800 *** 600

400

200 CK Colony number 0 CK siRNA

CK siRNA 400 ** 300

200 siRNA

100 Cell Number 0 Migration Invasion 0h 48h 72h 400 * INVASION 300

** CK 200 n/s

Cell number 100

0 Antibody 0 200 400 600 Con. IgG 600 400 200 0 siRNA (µg/ml)

Figure 8 Depletion of endogenous artemin by siRNA significantly decreased anchorage-independent growth and impaired cell migration and invasion. (a) Soft agar assay: MCF7-siRNA (siRNA) and MCF7-CK (CK) cells were seeded in 0.35% agarose and colonies formed after incubation for 14 days were counted. (b) Representative photomicrographic pictures of colonies formed by the cells growing in Matrigel. (c) Migration and invasion of MCF7-siRNA and MCF7-CK cells were determined by use of Transwell inserts as described in Materials and methods. (d) Wound healing assays: ‘Scratch’ wounds were created in vitro by scraping the cell monolayer with a sterile pipette tip. Photographs were taken at the time indicated. (e) Inhibition of invasion of MCF-7 cells by rabbit polyclonal antiserum to artemin. Bar, 100 mm. *Po0.05; **Po0.01; ***Po0.001. artemin is a secreted protein. Further analyses of the with disease outcome, artemin is thus identified as a function and regulation of artemin in mammary potential therapeutic target in oncology. carcinoma cells may also be useful to clarify potentially discrepant associations. For example, we observe Materials andmethods prominent regulation of artemin by HER2 (manuscript in preparation) supportive of the positive correlation Plasmid constructs observed between HER2 positivity and artemin protein The coding sequence for human artemin transcript variant 4 expression in the cohort presented herein. In any case, (GenBank accession number NM_057090) was cloned into higher expression of both artemin mRNA (Oncomine the mammalian expression vector pIRESneo3 (Invitrogen, data) and artemin protein (reported herein) predicted a Carlsbad, CA, USA), designated as pIRESneo3-ARTN. To significantly worse survival outcome. However, detailed visualize the depletion efficiency of siRNA, we inserted the 0 correlation of artemin mRNA and protein expression to artemin cDNA into the 3 UTR of an EGFP expression vector specific clinicopathological parameters requires further pEGFP-C1, designated as pEGFP-ARTN. The cDNA frag- ment coding for the mature artemin peptide of 113 amino acids analysis. was cloned into the pGEX-4T1 vector (Amersham Biosciences, In summary, we have shown the oncogenic properties Piscataway, NJ, USA) in frame with the N-terminal glu- of artemin in mammary carcinoma cells in vitro as well tathione-S-transferase to generate pGEX-4T1-ARTN plasmid as in xenograft models in vivo. Together with the highly for the expression of glutathione-S-transferase fusion proteins significant clinical association of artemin expression in E. coli.

Oncogene Oncogenic properties of artemin J Kang et al 2044 To generate siRNA oligonucleotides targeting all four immunostaining were carried out as described (Pandey et al., variants of artemin, the DNA sequence 50-AACTGGCCTG 2008). TACTCACTCAT-30 was selected to construct an siRNA expression plasmid using the pSilencer 2.1-U6 hygro vector Immunohistochemical staining of artemin in breast cancer 0 (Ambion, Austin, TX, USA) according to the manufacturer s patients protocol. The resultant vector was designated as pSilencer- Tissue samples were collected from 159 female breast cancer ARTN. The negative control siRNA plasmid (pSilencer-CK) patients from the First Affiliated Hospital of Anhui encodes an siRNA, which has no significant sequence Medical University (Hefei, PR China) between 2001 and similarity to human gene sequences (Ambion). 2002. Methodology for localization of ARTN and clinical and pathological information concerning these samples can be Cell culture found in Supplementary data 4. All cell lines used in this study were obtained from the ATCC (American Type Culture Collection) and cultured in condi- Production of recombinant protein and generation of antibody tions as recommended. MCF-7, T47Dand BT549 cells were Recombinant artemin protein was expressed and purified in stably transfected with pIRESneo3-ARTN or the empty E. coli using glutathione-S-transferase tag from pGEX-4T1- pIRESneo3 vector plasmids to establish stable cell lines ARTN plasmid as described (Masure et al., 1999) and was MCF7-ARTN, T47D-ARTN and BT549-ARTN with forced used to generate a rabbit polyclonal antibody. expression of artemin and their control cell lines MCF7-Vec, T47D-Vec and BT549-Vec, respectively. Similarly, MCF-7 Statistics cells were stably transfected with pSilencer-ARTN and All experiments were repeated three to five times. All pSilencer-CK plasmids to obtain MCF7-siRNA and MCF7- numerical data are expressed as mean±s.e.m. (standard error CK cells, respectively. of the mean) from a representative experiment performed in triplicate, and statistical analyses were assessed by Student’s PCR, RT–PCR and real-time PCR t test using Microsoft Excel XP. PCR, RT–PCR and real-time PCR were performed as described earlier (Pandey et al., 2008) (see Supplementary data 1 and 2). Abbreviations

Western blot analysis BrdU, bromodeoxyuridine; EGFP, enhanced green fluorescent Western blot analysis was carried out as described earlier (Liu protein; ER, estrogen receptor; GDNF, glial –cell line-derived and Lobie, 2007) using the following antibodies: goat anti- neurotrophic factor; GFL, GDNF family ligand; GFRa, artemin polyclonal antibody (R&DSystems, Minneapolis, GDNF family receptor a; PR, progesterone receptor; RT, MN, USA) and a monoclonal antibody against b-actin reverse transcription; siRNA, small interfering RNA; TU- (Sigma, St Louis, MO, USA). NEL, terminal doxynucleotidyl transferase-mediated dUTP nick end labeling. Cell death, proliferation and oncogenicity assays Apoptotic analysis was carried out as described earlier (Zhu et al., 2005). Other biological assays such as total cell number, Conflict of interest MTT assay, BrdU incorporation, anchorage-independent growth (soft agar and suspension culture), three-dimensional DXL and PEL are named on PCT application PCT/NZ2008/ morphogenesis, in vitro cell motility (migration, invasion and 000152. wound healing) assays were carried out as described (Pandey et al., 2008). Acknowledgements

Tumor xenografts in nude mice We thank Alan Beedle for critical reading of the manuscript. Tumor xenograft growth in vivo was essentially as described This work was funded by the Breast Cancer Research Trust earlier (Zhu et al., 2005). MCF7-ARTN and MCF7-Vec cells (NZ), the Foundation for Research, Science and Technology were injected subcutaneously into the mammary fat pad of of New Zealand, the Hundred-Talent Scheme of Chinese 3- to 4-week-old BALBc nu/nu mice (Shanghai Slaccas Co., Academy of Sciences, the National Natural Science Founda- Shanghai, China) (n ¼ 8 for each). Histological analysis with tion of China (2007CB914801 & 2007CB914503) and the hematoxylin and eosin (H&E), and BrdU and TUNEL National Basic Research Program of China (30571030).

References

Airaksinen MS, Saarma M. (2002). The GDNF family: signalling, Bespalov MM, Saarma M. (2007). GDNF family receptor complexes biological functions and therapeutic value. Nat Rev Neurosci 3: 383–394. are emerging drug targets. Trends Pharmacol Sci 28: 68–74. Airaksinen MS, Titievsky A, Saarma M. (1999). GDNF family Brouty-Boye D, Raux H. (1993). Differential influence of stromal neurotrophic factor signaling: four masters, one servant? Mol Cell fibroblasts from different breast tissues on human breast tumour cell Neurosci 13: 313–325. growth in nude mice. Anticancer Res 13: 1087–1090. Anderson J, Ramsay A, Gould S, Pritchard-Jones K. (2001). PAX3- Ceyhan GO, Bergmann F, Kadihasanoglu M, Erkan M, Park W, Hinz FKHR induces morphological change and enhances cellular U et al. (2007). The neurotrophic factor artemin influences the proliferation and invasion in rhabdomyosarcoma. Am J Pathol extent of neural damage and growth in chronic pancreatitis. Gut 56: 159: 1089–1096. 534–544.

Oncogene Oncogenic properties of artemin J Kang et al 2045 Ceyhan GO, Giese NA, Erkan M, Kerscher AG, Wente MN, Giese T Pasini B, Ceccherini I, Romeo G. (1996). RET mutations in human et al. (2006). The neurotrophic factor artemin promotes pancreatic disease. Trends Genet 12: 138–144. cancer invasion. Ann Surg 244: 274–281. Pettaway CA, Pathak S, Greene G, Ramirez E, Wilson MR, Killion JJ Chung LW. (1993). Implications of stromal-epithelial interaction in et al. (1996). Selection of highly metastatic variants of different human prostate cancer growth, progression and differentiation. human prostatic carcinomas using orthotopic implantation in nude Semin Cancer Biol 4: 183–192. mice. Clin Cancer Res 2: 1627–1636. Duxbury MS, Matros E, Ito H, Zinner MJ, Ashley SW, Whang EE. Popsueva A, Poteryaev D, Arighi E, Meng X, ngers-Loustau A, (2004). Systemic siRNA-mediated gene silencing: a new approach to Kaplan D et al. (2003). GDNF promotes tubulogenesis of GFRa1- targeted therapy of cancer. Ann Surg 240: 667–674. expressing MDCK cells by Src-mediated phosphorylation of Met Enomoto H, Hughes I, Golden J, Baloh RH, Yonemura S, . J Cell Biol 161: 119–129. Heuckeroth RO et al. (2004). GFRa1 expression in cells lacking Poteryaev D, Titievsky A, Sun YF, Thomas-Crusells J, Lindahl M, RET is dispensable for organogenesis and nerve regeneration. Billaud M et al. (1999). GDNF triggers a novel ret-independent Src Neuron 44: 623–636. kinase family-coupled signaling via a GPI-linked GDNF receptor Esseghir S, Todd SK, Hunt T, Poulsom R, Plaza-Menacho I, Reis- a1. FEBS Lett 463: 63–66. Filho JS et al. (2007). A role for glial cell derived neurotrophic Saarma M, Sariola H. (1999). Other neurotrophic factors: glial cell factor induced expression by inflammatory cytokines and RET/ line-derived neurotrophic factor (GDNF). Microsc Res Tech 45: GFRa1 receptor up-regulation in breast cancer. Cancer Res 67: 292–302. 11732–11741. Sadlonova A, Novak Z, Johnson MR, Bowe DB, Gault SR, Page GP Hutcheson IR, Knowlden JM, Jones HE, Burmi RS, McClelland RA, et al. (2005). Breast fibroblasts modulate epithelial cell proliferation Barrow D et al. (2006). Inductive mechanisms limiting response to in three-dimensional in vitro co-culture. Breast Cancer Res 7: anti-epidermal therapy. Endocr Relat Cancer R46–R59. 13(Suppl 1): S89–S97. Sariola H, Saarma M. (2003). Novel functions and signalling pathways Ichihara M, Murakumo Y, Takahashi M. (2004). RET and for GDNF. J Cell Sci 116: 3855–3862. neuroendocrine tumors. Cancer Lett 204: 197–211. Schueler-Furman O, Glick E, Segovia J, Linial M. (2006). Is GAS1 a Ito Y, Okada Y, Sato M, Sawai H, Funahashi H, Murase T et al. co-receptor for the GDNF family of ligands? Trends Pharmacol Sci (2005). Expression of glial cell line-derived neurotrophic factor 27: 72–77. family members and their receptors in pancreatic cancers. Surgery Singh S, Sadanandam A, Singh RK. (2007). Chemokines in tumor 138: 788–794. angiogenesis and metastasis. Cancer Metastasis Rev 26: 453–467. Lin LF, Doherty DH, Lile JD, Bektesh S, Collins F. (1993). GDNF: a Stott FJ, Bates S, James MC, McConnell BB, Starborg M, Brookes S glial cell line-derived neurotrophic factor for midbrain dopaminer- et al. (1998). The alternative product from the human CDKN2A gic neurons. Science 260: 1130–1132. locus, p14(ARF), participates in a regulatory feedback loop with Liu DX, Lobie PE. (2007). Transcriptional activation of p53 by Pitx1. p53 and MDM2. EMBO J 17: 5001–5014. Cell Death Differ 14: 1893–1907. Su G, Blaine SA, Qiao D, Friedl A. (2007). Shedding of syndecan-1 by Lubaroff DM, Canfield L, Reynolds CW. (1980). The Dunning stromal fibroblasts stimulates human breast cancer cell proliferation tumors. Prog Clin Biol Res 37: 243–263. via FGF2 activation. J Biol Chem 282: 14906–14915. Masure S, Geerts H, Cik M, Hoefnagel E, Van Den KG, Tuytelaars A Takahashi M. (2001). The GDNF/RET signaling pathway and human et al. (1999). Enovin, a member of the glial cell-line-derived diseases. Cytokine Growth Factor Rev 12: 361–373. neurotrophic factor (GDNF) family with growth promoting activity Trupp M, Scott R, Whittemore SR, Ibanez CF. (1999). Ret-dependent on neuronal cells. Existence and tissue-specific expression of and -independent mechanisms of glial cell line-derived different splice variants. Eur J Biochem 266: 892–902. neurotrophic factor signaling in neuronal cells. J Biol Chem 274: Meng X, de Rooij DG, Westerdahl K, Saarma M, Sariola H. (2001a). 20885–20894. Promotion of seminomatous tumors by targeted overexpression of Veit C, Genze F, Menke A, Hoeffert S, Gress TM, Gierschik P et al. glial cell line-derived neurotrophic factor in mouse testis. Cancer Res (2004). Activation of phosphatidylinositol 3-kinase and extracellular 61: 3267–3271. signal-regulated kinase is required for glial cell line-derived Meng X, Lindahl M, Hyvonen ME, Parvinen M, de Rooij DG, Hess neurotrophic factor-induced migration and invasion of pancreatic MW et al. (2000). Regulation of cell fate decision of undifferentiated carcinoma cells. Cancer Res 64: 5291–5300. spermatogonia by GDNF. Science 287: 1489–1493. Wu K, Weng Z, Tao Q, Lin G, Wu X, Qian H et al. (2003). Stage- Meng X, Pata I, Pedrono E, Popsueva A, de Rooij DG, Janne M et al. specific expression of breast cancer-specific gene g-synuclein. Cancer (2001b). Transient disruption of spermatogenesis by deregulated Epidemiol Biomarkers Prev 12: 920–925. expression of neurturin in testis. Mol Cell Endocrinol 184: 33–39. Zhang P, Chan SL, Fu W, Mendoza M, Mattson MP. (2003). TERT Okada Y, Takeyama H, Sato M, Morikawa M, Sobue K, Asai K et al. suppresses apoptotis at a premitochondrial step by a mechanism (1999). Experimental implication of celiac ganglionotropic invasion requiring reverse transcriptase activity and 14-3-3 protein-binding of pancreatic-cancer cells bearing c-ret proto-oncogene with ability. FASEB J 17: 767–769. reference to glial-cell-line-derived neurotrophic factor (GDNF). Zhu T, Starling-Emerald B, Zhang X, Lee KO, Gluckman PD, Int J Cancer 81: 67–73. Mertani HC et al. (2005). Oncogenic transformation of human Pandey V, Perry JK, Mohankumar KM, Kong XJ, Liu SM, Wu ZS mammary epithelial cells by autocrine human growth hormone. et al. (2008). Autocrine human growth hormone stimulates Cancer Res 65: 317–324. oncogenicity of endometrial carcinoma cells. Endocrinology 149: Zhu ZW, Friess H, Wang L, Bogardus T, Korc M, Kleeff J et al. 3909–3919. (2001). exerts differential effects on the growth Paratcha G, Ledda F, Ibanez CF. (2003). The neural cell adhesion of human pancreatic cancer cells. Clin Cancer Res 7: 105–112. molecule NCAM is an alternative signaling receptor for GDNF Zinkel S, Gross A, Yang E. (2006). BCL2 family in DNA damage and family ligands. Cell 113: 867–879. cell cycle control. Cell Death Differ 13: 1351–1359.

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

Oncogene