Downregulation, Alternative Splicing and Drug Sensitization

Oncogene (2011) 30, 2874–2887 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE Anti-proliferative and pro-apoptotic actions of a novel human and mouse ovarian tumor-associated gene OTAG-12: downregulation, alternative splicing and drug sensitization

X Chen1, H Zhang1, JP Aravindakshan1, WH Gotlieb2 and MR Sairam1,3,4

1Molecular Endocrinology Laboratory, Clinical Research Institute of Montreal, Montreal, Que´bec, Canada; 2McGill University, Segal Cancer Center, Jewish General Hospital, Montreal, Que´bec, Canada; 3De´partement de Me´decine, Universite´ de Montre´al, Montreal, Que´bec, Canada and 4Department of Medicine, Division of Experimental Medicine & Department of Physiology, McGill University, Montreal, Que´bec, Canada

In studying the age dependence and chronology of ovarian apoptotic genes such as BAD, GADD45a and CIEDB, tumors in follicle stimulating hormone receptor knockout while inhibiting anti-apoptotic NAIP and Akt1 expression, mice, we identified a novel ovarian tumor associated gene- suggesting that hOTAG-12b-induced apoptosis might be 12 (OTAG-12), which is progressively downregulated and p53-independent. These results indicate that hOTAG-12b is maps to Chr. 8B3.3. OTAG-12 protein overexpression in a putative ovarian tumor suppressor gene warranting mouse ovarian and mammary tumor cells suggested further studies. powerful anti-proliferative effects. In human epithelial Oncogene (2011) 30, 2874–2887; doi:10.1038/onc.2011.11; ovarian cancers (OCs) and OC cell lines, OTAG-12 published online 21 February 2011 mRNA expression is downregulated in comparison with normal ovaries. Cloning and identification revealed that Keywords: ovarian cancer; tumor suppressor; splicing; human OTAG-12 mapping to gene-rich Chr. 19p13.12 is G2 arrest; apoptosis; drug sensitivity expressed in three spliced forms: hOTAG-12a, hOTAG- 12b and hOTAG-12c, of which b is predominant in the normal ovary. Functionally active hOTAG-12b is a simple protein with no disulfide bonds and a nuclear localization Introduction signal is present in all variants. Transfection of OTAG-12 variants in OC and tumorigenic HEK293 cells confirmed Ovarian cancer (OC) is the most aggressive cancer and nuclear localization. hOTAG-12b overexpression in OC leading cause of death from gynecological malignancies and HEK293 cells effectively suppressed cell growth, in the developed world, with an incidence of about anchorage-dependent and independent colony formation 190 000 new cases and 114 000 fatalities (Jemal et al., followed by apoptosis, whereas hOTAG-12a and hOTAG- 2009). Although OC accounts for only 4% of all cancers 12c had no such effects. Deletion mutants identified the in women, it is 3 Â more lethal than breast cancer. critical importance of carboxyl terminus for hOTAG-12b Epithelial OC (EOC), which accounts for 85% of OC in function. Doxycycline-inducible growth inhibition of post-menopausal women, is believed to originate from HEK293 cells by hOTAG-12a was associated with effects cells of the aberrant ovarian surface epithelium or on G2 cell cycle arrest and apoptosis induction. hOTAG- fimbriated region of the fallopian tube (Jarboe et al., 12b expression rendered tumorigenic cells more sensitive to 2008). Around 10% of OCs are associated with four apoptotic stimuli including etoposide—a topoisome- mutations of BRCA1 and BRCA2 (Pal et al., 2005), rase-II inhibitor. Doxycycline-induced hOTAG-12b ex- although multiple genetic and epigenetic abnormalities pression blocked xenograft tumor growth in nude mice, have been detected in different individuals. Despite whereas hOTAG-12a was ineffective. Although p53-path- improvements in cytoreductive surgery and the initially way-dependent apoptotic agents could upregulate endogen- favorable response to platinum-based chemotherapies, ous hOTAG-12b and p53 in UCI-101/107 OC cells, nearly two-thirds of the patients succumb within 5 years hOTAG-12b could also induce apoptosis in p53-null and of diagnosis (Konstantinopoulos et al., 2008). Most platinum-resistant SKOV3 OC cells and Doxycycline- patients are initially diagnosed at advanced stages and induced hOTAG-12b did not alter p53. Further study there are no reliable prognostic markers for predicting showed that hOTAG-12b increases mRNAs of pro- clinical response and guiding treatment of drug-resistant cancers. Knowledge of novel genes and molecular targets expressed in ovarian tumors is required for Correspondence: Dr MR Sairam, Molecular Endocrinology Labora- better and early detection and development of new tory, Clinical Research Institute of Montreal, 110 Pine Avenue West, therapies (Bast Jr et al., 2009). Montreal, Que´bec, Canada H2W 1R7. E-mail: [email protected] Tumor suppressor genes have key roles in cell Received 21 August 2010; revised 4 January 2011; accepted 7 January growth, proliferation, apoptosis and tumor development 2011; published online 21 February 2011 (Bruening et al., 1999; Luo et al., 2003; Bignone et al., Novel ovarian tumor suppressor gene X Chen et al 2875 2007; Bast et al., 2009). Sixteen candidate tumor Affymetrix probe, ‘1423216_a_at’, was selected, as this suppressor genes have been studied in OC (Bast Jr gene showed progressive decrease from pre-tumor stage to et al., 2009). EOC is a heterogeneous disease with ovarian tumors at 12 months (Figure 1a). BLASTN search complex pathology and more than 50% of high-grade matched this with an uncharacterized transcript: Mus serous-type EOCs contain p53 mutations and altera- musculus RIKEN cDNA 2510049I19Rik. We named the tions in the retinoblastoma (RB) pathway (Corney et al., gene as OTAG-12 to stand for ovarian tumor associated 2008; Bast Jr et al., 2009), whereas K-Ras, Pten and gene-12. Quantitative PCR (Q–PCR) confirmed OTAG-12 Wnt/b-catenin are implicated in endometrioid OC alteration in FORKO ovary (Figure 1b). Expression was subtype (Dinulescu et al., 2005; Wu et al., 2007). decreasedto30%(Po0.05) at 12 months in FORKO In studies with tissues or cells from late-stage OCs, it tumors (Figures 1a and b) consistent with microarray. has been difficult to segregate molecular changes related Maximal downregulation was found in syngenic peritoneal to disease initiation from other factors like progression ovarian epithelial tumors grown using mouse ID8 cells and drug treatment. As knowledge of early tumorigenic (Roby et al., 2000). In female mice, OTAG-12 mRNA was events and genes is urgently needed, these issues can be also expressed in several other tissues at varying levels studied systematically in models that can be experimen- (Figure 1c). tally manipulated to capture chronological changes. Thus we reported that in follitropin receptor knockout Cloning of mouse and human OTAG-12 genes and protein (FORKO) mice with sex hormone imbalances, females characteristics develop age-dependent ovarian tumors by B12 months Cloning and identification of mouse ORF confirmed (Danilovich et al., 2001; Chen et al., 2007). Tracking 100% match to hypothetical protein LOC67922, map- early and progressive ovarian tumorigenic events in this ping to murine chr. 8 B3.3. The 4046-bp mouse OTAG- model revealed the pattern of molecular changes also 12 gene has four exons. The 1921-bp cDNA codes found in EOC in women, including claudins, E-cadherin for a protein of 112 amino acids (Figure 2a). We then and platelet-derived growth factor (PDGF) ligands and performed 30/50 rapid amplification of cDNA ends PCR receptors, as well as immune system upregulation in pre- (RACE-PCR) to secure full-length human OTAG-12 of tumor stages (Chen et al., 2006; Aravindakshan et al., 1463 bp. Except for deletion of the last A (Supplemen- 2006a, b). We then hypothesized that transcriptome tary Figure 1), the sequence determined matches comparison at different stages could lead us to implicate hypothetical cDNA (NM_014077.2). This full-length novel and unknown genes in ovarian tumor develop- cDNA includes a 339-bp ORF (LOC26017) containing ment. Herein we focus on one novel gene that was ATG codon in EXON 1 within a good Kozak consensus progressively downregulated in pre-tumor and tumor- sequence (GTCATGG). A 30-UTR polyadenylation bearing mutant ovaries and was also very low in mouse site (AATAAA) lies 1068 bp downstream of the stop epithelial tumors. This novel ovarian tumor-associated codon and 19 bp upstream of a poly (dA) tail. Human gene-12 (OTAG-12) was cloned from both mouse and OTAG-12 gene spanning 6.6 kb maps to chr. 19p13.1 human ovaries and evidence is presented for its and is located near a cluster of known OC-related antiproliferative and proapoptotic effects in OC cells. genes (Figures 2b and c). The amino-acid sequence of Only one of three alternatively spliced forms of human human ovarian OTAG-12 shows high homology to OTAG-12 (hOTAG-12) exhibits potent inhibitory ef- mouse (97%) (Figure 2e and other species (Supplemen- fects on OC cells by p53-independent mechanisms. As tary Figure 2)), indicating conservation. Both mouse overexpression of hOTAG-12b sensitizes tumorigenic and human have a nuclear localization signal peptide cells to the apoptotic action of other anti-cancer agents, KRKKKKKDK; this location was confirmed by there could be potential for devising recombinant immunohistochemistry (Figure 3e). Functional motifs therapeutic modalities for mitigating tumor burden. search of OTAG-12 protein indicated potential sites for protein–DNA binding, phosphorylation and interactions with 14-3-3 (Figures 2e and f). Results Isoforms of human OTAG-12 Transcriptome analysis and discovery of novel gene When human OTAG-12 was amplified by PCR using OTAG-12 primer pairs flanking the hypothetical ORF, two or To find novel genes during ovarian tumorigenesis, we more bands were evident (Figure 2d). The major compared the transcriptome in 1-month, 8-month (pre- amplicon b corresponds to the predicted ORF of tumor) and 12-month-old (tumor-bearing) FORKO 339 bp and referred herein as hOTAG-12b (Figures 2c and wild-type littermates (Aravindakshan et al., 2010). and d). The higher amplicons correspond to isoforms a Among transcripts dysregulated X2-fold, we focused and c. Their identity as hOTAG-12a (variant 1 in data exclusively on unknown genes. Using bio-informatics data bank) and 12c (variant 3) in the normal ovary was to deduce predicted open reading frames (ORFs) for confirmed by using ORF-specific primers and sequen- hypothetical proteins, we attributed significance if the gene cing. As variants in mouse ovary are minor (Figure 2c), was conserved in humans. Second, we stipulated ovarian we pursued only the major form. Human (and mouse) expression and disease association, as in FORKOs. Third, OTAG-12b proteins are basic, with 112 amino acids we chose predicted proteins that could help test function. containing no cysteine residues or glycosylation site. Thus, one expressed sequence tag (EST) identified by However, hOTAG-12a with 107 residues, while still

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2876 basic, has two cysteines that could alter its properties. were done in a blinded manner. The dominant form, hOTAG-12c comprising 92 amino acids also differs hOTAG-12b, in normal ovary was significantly down- from hOTAG-12b but at the N-terminus (Figure 2f). regulated in OC (Figure 3a). The relative expression of OTAG-12 in normal human ovary appears to be Expression of OTAG-12 in human ovarian tumors and hOTAG-12bc12ap12c. Whereas the expression of OC cell lines hOTAG-12a is not significantly altered in OC, hO- By designing primers (Supplementary Table 1) based TAG-12C also appears to be downregulated (see on our cloning data, we verified expression in normal Supplementary Figure 3). In five human OC cell lines, ovary and human EOCs using Q–PCR. These analyses hOTAG-12b expression was lower than in normal ovary

Figure 1 Chronology of mouse OTAG-12 expression in ovary, FORKO, various tissues and apoptotic induction of mouse ID8 and HC11-26 mouse mammary cells (Dievart et al., 1999). (a) Microarray data of OTAG-12 expression in FORKO and age-matched wild- type (WT) females at 1 month, 8 months and 12 months. (b) Q–PCR analysis of mouse OTAG-12 in ovarian samples of 12-month-old WT, age-matched FORKO, ovarian tumor and mouse epithelial tumor nodules, normalized to b-actin (n ¼ 3 mice). The tumor nodules were generated by injecting cultured mouse tumorigenic cell line ID8 into 3-month-old FORKO females and collecting the peritoneal tumors after 2–3 months. *Po0.05; **Po0.01, signiﬁcant difference. (c) Mouse OTAG-12 expression in different tissues of 4–5-month-old female mice. The expression level of wild-type ovary, set as 1 unit, is normalized by b-actin (n ¼ 3 mice). (d) Western blot of mouse OTAG-12. Cells were transiently transfected with OTAG-12 cloned into pIRES2-V5-EGFP vector. After 24 h, cells were washed by PBS and lysed by RIPA lysis buffer and 20–50 mg of lysate was used for western blotting. The mouse anti-V5 antibody (Cat. No. R960-25, Invitrogen) was used at 1:8000 dilution to detect tagged OTAG-12 protein. The 17-kDa band shown represents OTAG-12. (e) OTAG-12 overexpression inhibits colony formation. 106 cells were seeded, transfected and grown in selection media (700 mg/ml G418 for ID8 cells and 600 mg/ml G418 for HC11-26 cells) for 2–3 weeks. Colonies were stained with Coomassie blue (0.1% Coomassie blue, 30% methanol, 10% acetic acid) and photographed. (f) Colonies formed were counted and the results of three different experiments in triplicate were combined; the bar graph shows the mean±s.e. Colony numbers in OTAG-12 were compared with empty pRIS2-EGFP vector after transfection; signiﬁcant differences were observed. ***Po0.001.

Figure 2 Gene structure, chromosomal location and species conservation of mouse and human OTAG-12 proteins. (a) Mouse OTAG-12 gene, spanning 4046 bp (indicated by the solid arrow above the diagram), mapped to chromosome 8B3.3. It contains four exons (1, 2, 3, and 4) and three introns. The dashed lines indicate that the size of OTAG-12 cDNA is 1921 bp. The box indicates the coding region ORF 339 bp. (b) Human OTAG-12 is located on chromosome19 at 19p13.12, near a cluster of known ovarian cancer- related genes such as NOTCH3 (Park et al., 2006), AP-1, MUC16 (Yin et al., 2002) and hCG (Nowak-Markwitz et al., 2004) and tumor suppressor genes such as BRG1, JUNB and PRG6 (Bast Jr et al., 2009). Chromosome 19 also has three other oncogenes, PIK3R1, Cyclin 1 and AKT2 (Bast et al., 2009), that are upregulated in ovarian cancer (OC). (c) Alternative splicing of human OTAG-12 gene produces three isoforms, hOTAG-12a (107 amino acids), -b (112 amino acids) and -c (92 amino acids). (d) Real-time PCR (RT–PCR) analysis of mouse OTAG-12 ORF amplicon (arrow) and three variants of human OTAG-12 ORF (arrow head) in normal ovary. (e) Sequence conservation of OTAG-12b protein. The sequence of mouse and human OTAG12 proteins are as deduced from our current cloning work on ovarian tissue. The nuclear localization signal in OTAG12 is indicated by a line on top of the sequence, and the 3–75 region is a conserved domain called DUF1754, whose signiﬁcance is unknown. Predicted structural motifs by http:// elm.eu.org/in hOTAG-12 are indicated. Domains MAPK 1 (Johnson and Lapadat, 2002), FHA2 (Schwartz et al., 2003), 14-3-3 (Dougherty and Morrison, 2004), USP7 (Maertens et al., 2010) binding sites as well as DNA-binding sites are shown in bold. The predicted OTAG-12 proteins with accession numbers for other species are shown in Supplementary Figure 2. (f) Alignment of three variants of human OTAG-12a, b and c. In all, 73 amino acids in the N-terminal are the same between hOTAG-12a and hOTAG-12b, whereas 88 amino acids in the C-terminal are the same between hOTAG-12b and hOTAG-12c. Importantly, a potential –SÀSÀ bridge is likely to be present only in OTAG-12a. Note the differences between hOTAG-12b and 12a (see Supplementary Figure 2).

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2877 (Figure 3b). hOTAG-12b expression in surgically Effects of overexpression of OTAG-12 on OC, mammary removed normal human ovarian cortex used as a and tumorigenic HEK293 cells surrogate for OSE was similar to that in normal ovary. To explore the possible involvement of the novel gene in Expression in HEK293 cells was slightly higher. Perusal cell growth, we ﬁrst transfected mouse pIRES2-OTAG- of public microarray expression data using Affymetrix 12-V5-GFP into mouse ID8 and HC11-26 mammary probes (http://symatlas.gnf.org/SymAtlas/ http://biogps.gnf. cancer cells (Dievart et al., 1999). Transient transfection org/#goto ¼ genereport&id ¼ 26017) indicates OTAG-12 revealed the corresponding protein in the cell lysate expression in a variety of human tissues. (Figure 1d); however, within 5 days cell proliferation

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2878

Figure 3 Expression and growth inhibition in different cell lines, localization of isoforms of human OTAG-12. (a) Q–PCR analysis of expression level of hOTAG-12b in human ovarian samples (normal ovaries 13, benign ovarian tumor 2 and grade 3C serous ovarian cancer 11). The OTAG-12b expression level of normal ovary is signiﬁcantly higher than that of human ovarian cancer samples. Normalized by GAPDH. *Po0.05. (b) Q–PCR analysis of expression level of hOTAG-12b in different OC cells compared with normal human ovary and cortex. Total RNA extracted from tissue or cultured OC cells and HEK 293 cells were used. Normal human ovary expression set as 1 for comparison. hOVCÀ represents (fresh frozen) surgically removed human normal ovarian cortex and is used as a surrogate for human ovarian epithelial cells. UCI 101 and UCI 107 OC cells were obtained from Dr Carpenter (Fuchtner et al., 1993; Gamboa et al., 1995) and others from ATCC. OTAG-12b expression level of all OC cells is lower than hOVC and normal ovary. (c) Overexpression of hOTAG-12b reduces anchorage-dependent colony formation (relative number of colonies; control vector taken as 100) in four cell lines. At 24 h after transient transfection, cells were trypsinized, washed, 1:20 diluted and grown in six-well plates in selection media containing G418. Cells were grown for 2–3 weeks; colonies were stained with Coomassie blue (0.1% Coomassie blue, 30% methanol, 10% acetic acid) and counted. At least three experiments were performed in quadruplicate for each cell line. ***Po0.001. (d) Soft agar assays. We mixed 5 Â 104cells at 37 1C with 0.4% agarose in complete medium and poured the mixture over a layer of 0.8% agarose in complete medium in six-well plates. The top layer was allowed to gel at room temperature for at least 15 min, and the plates were then incubated at 37 1C for 2–3 weeks with or without G418. At the end, the plates were stained by adding p-iodonitrotetrazolium violet to each well and incubated at 37 1C for 6 h, after which the colonies were photographed with a Zeiss microscope and counted by using a software associated microscope. Colony numbers in hOTAG-12b were compared with hOTAG-12a and empty pIRES2 vector transfection; signiﬁcant differences were observed for UCI107. The soft agar result is consistent with the result of anchorage-dependent cell growth. The image is representative of at least three separate experiments. (e) Localization of hOTAG-12a and hOTAG-12b by using V5 antibody in human OC cell line UCI107. Both hOTAG-12a and hOTAG-12b are predominantly localized in the nucleus. (f) Nuclear fractions (40 mg) of transiently transfected UCI101/107 cells with hOTAG-12a, hOTAG-12b or vector each were analyzed by western blot with V5 antibody. The arrow shows apparent molecular weight.

was inhibited (Figures 1e and f). Mouse OTAG-12 nucleus (Figure 3e) of UCI107 OC cells. By western also inhibited anchor-independent cell growth (not blotting, the hOTAG-12b and -12a proteins were also shown). found only in the nuclear fraction (Figure 3f). We extended the studies to all three human OTAG-12 variants on three OC and tumorigenic HEK293 cells. Importance of the C-terminal sequence of hOTAG-12b for Only hOTAG-12b but not the isoforms -12a and -12c function (not shown) inhibited growth and induced apoptosis in As hOTAG-12b protein differs from hOTAG-12a in the the cells (Figure 3c). Assessment of anchorage-depen- carboxyl terminus, we reasoned that domains in this dent cell growth or independent colony formation region could be important for biological function. showed hOTAG-12b alone to exert potent cell suppres- To verify this, we constructed deletion mutants and sion (Figure 3d). tested them as shown in Figure 4. Deletion of the UP domain produced signiﬁcant but incomplete suppres- Subcellular localization of OTAG-12 proteins sion. However, further deletion of the F14 domain as in Immunohistochemical probing using V5-tag showed hOTAG-12bDC21 abolished cell-suppressing activity predominant hOTAG-12b and 12a localization in the (Figures 4C and D). Equivalent protein expression of

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2879

Figure 4 Construction, expression and functional comparison of wild-type and deletion mutant OTAG-12b proteins. (A) Schematic presentation of hOTAG-12a (107 residues), hOTAG-12b wild-type (112 amino acids) and its deletion mutants. hOTAG-12b DC10: deletion mutant lacking the last 10 amino acids (UP domain) of hOTAG-12b; hOTAG-12b DC21: deletion mutant lacking the last 21 amino acids (UP and F14 domains) of hOTAG-12b; hOTAG-12b DC37: deletion mutant lacking the last 37 amino acids of hOTAG-12b; NLS: nuclear localization sequence; F14: predicted binding sites of FHA-2 and 14-3-3; UP: predicted binding site of USP-7 (ubiquitin-specific peptidase 7). (B) Immunohistochemical staining of transiently transfected tumorigenic HEK293 cells by different constructs using V5 antibody (top). a: hOATG-12a; b: hOTAG-12b wild type; c: hOTAG- 12b DC10; d: hOTAG-12b DC21; e: hOTAG-12b DC37; f: vector. Immunoblot profiles of expression of hOTAG-12a, and hOTAG-12b wild-type and mutants in transfected cells probed with V5 tag antibody (bottom). Expression level of hOTAG- 12bWT and mutants hOTAG-12b DC10 and hOTAG-12b DC21 is similar, but hOTAG-12b DC37 expression appears to be low. (C) Effect of hOTAG-12a, and hOTAG-12bWT and its deletion mutants on HEK293 cells. Cells were transfected with same amount of plasmids and after 24 h diluted 1:20 into the selection medium containing 700 mg/ml G418 and incubated for 8 days. Cells were then stained by Coomassie Blue. a: vector; b: hOATG-12a; c: hOTAG-12b wild-type; d: hOTAG-12b DC37; e: hOTAG-12b DC21; f: hOTAG-12b DC10. (D) Quantification of the cell staining of hOTAG-12a, and hOTAG-12b wild-type and mutants. Each bar indicates±s.e.m. The results show that hOTAG-12b wild-type could induce cell death, mutant hOTAG-12b DC10 could suppress the cell growth, but hOTAG-12a, hOTAG-12b DC21 and hOTAG-12b DC37 have no effect, suggesting that F14 domain of hOTAG-12b is critical for inducing cell death. At least three expriments were performed, **Po0.01; ***Po0.001. all mutants except DC37 (slightly reduced) was con- shown in Figure 5A, hOTAG-12b expression under firmed by western Blot and IHC using the V5 antibody DOX induction varied in different clones; fewer cells (Figure 4B). survived in the highest inducible hOTAG-12b subclone- c33. Clone 32 with lowest expression retained better Inducible expression of hOTAG-12 in HEK293 cells growth, with clone c25 being intermediate. In hOTAG- As hOTAG-12b’s effects in transfected cells were 12b-c33, clone induction with 1.6 mg/ml DOX showed dramatic, we were unable to establish stable cell lines potent suppression with minimal toxicity (Figures 5Ba–f). constitutively expressing hOTAG-12b in all tested cells. However, cell growth continued upon DOX induction Therefore, we tried to establish doxycycline (DOX)- of hOTAG-12a or vector control. The level of hOTAG- inducible stable cell lines expressing OTAG-12s using 12 could be controlled by varying DOX (Figure 5C); OC cells and HEK293 cells. Under identical conditions, hOTAG-12a and hOTAG-12b were inducible to similar we succeeded only in securing stable clones of the more levels within 24 h, reaching a steady state by 2À3 days easily transfectable HEK293 cell line but not OC cells. (Figure 5D). Hence, all studies on DOX-induced expression could To verify that hOTAG-12b was not toxic to cells, we first henceforth be conducted using stable HEK293 cells. As induced hOTAG-12b expression for a day or two; when cell

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2880

Figure 5 Doxycycline (DOX)-inducible expression of OTAG-12 in stable clones of HEK293 cells. HEK293/TR/OTAG-12a, HEK293/TR/OTAG-12b and HEK293/TR parental cell subclones are mentioned here and in later figures as hOTAG-12a, hOTAG- 12b and Vector in the DOX-inducible system. (A) Expression level of hOTAG-12b after DOX treatment for 24 h in different hOATG- 12b subclones c32, c25 and c33. The bottom dishes show the result of inhibition of anchorage-dependent growth of these subclones with 1.6 mg/ml DOX treatment for 5 days. (B) Growth inhibition by OTAG-12b depends on DOX concentration. aÀf are dishes seeded with OTAG-12b-inducible cells, a, no DOX; b, 0.2 mg/ml DOX; c, 0.4 mg/ml DOX; d, 0.8 mg/ml DOX; e, 1.6 mg/ml DOX; f, 3.2 mg/ml DOX; g, hOTAG-12a cells with 1.6 mg/ml DOX; h, vector cells with 1.6 mg/ml DOX; and k, vector cells without DOX treatment. Cells were stained by Coomassie blue after 6 days. (C) hOTAG-12 induction is under the control of DOX. Cells with variant hOTAG-12a and hOTAG-12b clones were cultured at the indicated concentration of DOX for 24 h. (D) The expression level of hOTAG-12a and b depends on the time of the induction. The representative clones were cultured in the presence of 1.6 mg/ml DOX for the time indicated. In (C) and (D), total RNA was subjected to RT and Q–PCR using respective (specific) primers. (E) hOTAG-12b-dependent growth suppression is reversible. Stable clones were left untreated (filled squares) or treated with DOX (open squares and filled triangles) for 1 or 2 days. Cells were then washed and cultured in the absence of DOX for another 8 days. The number of live cells was determined by MTT at each time point. Each point represents the mean±s.e.m. from three separate experiments. (F) The level of hOTAG-12b at the time indicated was determined by Q–PCR. Three separate experiments were done.

growth was greatly suppressed by hOTAG-12b, we washed of cells. As illustrated in Figure 6b, about 30% increase out DOX from the medium and reseeded the remaining in the G2 population was observed only when hOTAG- cells at the original density. In cells limited to 1-day DOX 12b was induced by DOX, suggesting that hOTAG-12b exposure, hOTAG-12b mRNA started to decrease on day 2 exerted its effects on cell cycle progression by inducing and was similar in amount to the untreated cells at day 3 G2 arrest. (Figure 5F). Correlating with hOTAG-12b level, cells pre-exposed to hOTAG-12b for 1 day resumed growth hOTAG-12b induces cell apoptosis relatively quickly (Figure 5E); cells exposed to the 2-day To investigate whether hOTAG-12b could regulate effects of hOTAG-12b induction resumed growth after a lag apoptosis and cause cell death, we performed several period. Thus, growth inhibition is dependent upon hOTAG- tests. Annexin-V APC detection (Figures 7a and b) 12b expression level and induction effects are reversible. revealed that apoptosis was significantly increased in cells expressing exogenous hOTAG-12b 3 days after DOX treatment (P 0.01) compared with controls. OTAG-12 suppresses cell growth o Apoptosis was evident as early as 2 days (not shown). The effects of hOTAG-12a and -b on cell prolifera- We found no significant apoptosis in cells expressing tion were further examined by MTT assay. Cells with hOTAG-12a or vector. Hematoxylin and Hoechst 33342 vector, hOTAG-12a or hOTAG-12b were seeded at the staining (Figure 7c) showed morphological features same density and cultured in the absence or presence of of hOTAG-12b-triggered cell death as early as 66 h. 1.6 mg/ml DOX for 5 days. Compared with vector and In contrast, most cells were viable in the absence of DOX. hOTAG-12a, the growth rate of hOTAG-12b-inducible The increase of dead cells correlated with the decrease stable cells was significantly slower (Figure 6a). After of live cells, suggesting that cell death was the major 4–5 days, very few hOTAG-12b-expressing cells survived. mechanism for growth suppression. Assessing the DNA ladder, we found typical changes on day 3 of DOX hOTAG-12b influences the cell cycle treatment in hOTAG-12b cells but not in hOTAG-12a and Addressing the mechanism of growth suppression, we vector (Figure 6d). Anchorage-independent assays con- examined cell cycle progression using an inducible clone firmed the above results. No colony was detected in

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2881 hOTAG-12b overexpression inhibits tumor growth in nude mice We then examined the effects of hOTAG-12 on in vivo tumor growth in xenografted nude mice. Stable and DOX controlled vector, hOTAG-12a and hOTAG-12b-c25/ c33 lines were injected into nude mice (Figures 7g and h). Control mice without DOX developed visible and solid tumors, indicating vigorous growth in 4 weeks (Figure 7g); DOX treatment from day À1selectively blocked tumors at sites where hOTAG-12b-expressing cells were injected. The blockade was more effective at sites where cells with a higher level of hOTAG-12b (OTAG-12b-H) were injected as compared with sites with a lower level of hOTAG-12b-L cells (Figure 7h). Overall these results indicate that tumor growth in vivo could be controlled by regulated expression of hOTAG-12b. Histology of the excised tumors conﬁrmed the inhibitory effects of hOTAG-12b (see Supplementary Figure 4).

Potential apoptotic pathways affected by hOTAG-12b in OC cells OC cells were treated with apoptotic drugs including etoposide, puromycin, carboplatin, taxol and hygromycin-B. The first three use a p53-dependent mechanism, whereas the latter two trigger p53-independent apoptosis. Following treatment, the OTAG-12b increase in UCI 107/101 OC cells paralleled p53 changes induced by etoposide, puromycin and carboplatin, whereas both taxol and hygromycin-B did not upregulate hOTAG- 12b or p53 (Figure 8a). This suggested that upregulation of endogenous expression of hOTAG-12b and p53 induced by etoposide, puromycin or carboplatin could be contributing to their apoptotic effects in OC cells. In p53-null SKOV3 OC cells resistant to various agents, transfection of hOTAG-12b but not hOTAG-12a caused cell loss (Figure 8b), suggesting that overexpression of hOTAG-12b alone in the absence of wild-type p53 is sufficient to trigger apoptosis. To explore the downstream apoptotic events activated by hOTAG-12b, we verified several genes (Supplemen- tary Table 2) in cells with DOX-controlled hOTAG-12b. In this study, pro-apoptotic genes such as GADD45a, CIDEB and BAD increased, whereas anti-apoptotic NAIP and proliferation promoter Akt1 decreased Figure 6 Expression of OTAG-12b inhibits cell growth and induces (Figure 8c), suggesting cumulative effects: increasing G2 arrest. (a) OTAG-12b suppresses HEK293 cell growth. Inducible pro-apoptotic and decreasing anti-apoptotic genes. hOTAG-12a, hOTAG-12b subclone and vector cell lines were left Notably, GADD45a, CIDEB and BAD increased as untreated or treated with either DOX for 5 days. The number of live cells was determined by MTT at each time point. Each point early as after 12–24 h of hOTAG-12b exposure, whereas represents mean±s.e.m. from three separate experiments. (b)G2 most cell death occurred after 2–3 days. GADD45a arrest of inducible hOTAG-12b cells. Inducible hOTAG-12a, upregulation that induces G2 arrest (Chen et al., 2001) hOTAG-12b subclone and vector-containing cell line were grown was maximum at 72 h. p53-mRNA and target genes with or without 1.6 mg/ml DOX for 48 h. Cells were stained with including p21 and BAX were not significantly altered by propidium iodide (PI) and analyzed by FACS. Modified Krishan buffer consists of 0.1% sodium citrate 0.3% NP-40, with 50 mg/ml PI hOTAG-12b (not shown). These indicate that hOTAG- and 20 mg/ml RNase-A (Sigma, St Louis, MO, USA). Results of 12b-induced G2 arrest (Figure 6) and apoptosis could be three different experiments showed a significant proportion of cells independent of p53. Western blotting using a phospho- arrested at the G2 phase when expressing hOTAG-12b compared p53 antibody also showed only slight increase during with cells expressing hOTAG-12a or the null vector cells. apoptosis induced by hOTAG-12b as compared with the effects seen with etoposide (Figure 8d). A probable HEK293 hOTAG-12b under DOX induction. In control scenario depicting hOTAG-12b’s effects on cell cycle cells, containing only repressor plasmid or hOTAG-12a, arrest and apoptosis is depicted in Supplementary DOX had no effect (Figures 7e and f). Figure 5.

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2882

Figure 7 hOTAG-12b induces apoptosis. (a) Results of Annexin V-APC analysis. Inducible OTAG-12a, TR/OTAG-12b subclone and vector cell line were grown with or without 1.6 mg/ml DOX for 72 h and stained with Annexin V-APC before analysis using FACS. Results of three different experiments in triplicate were combined. (b) The percentage of apoptotic cells, in the right (UR þ LR) quadrant, is significantly increased in hOTAG-12b-expressing cells (39.1 versus 12.1% in No DOX), whereas that of live cells decreased to 40.8% from 81.5% (No DOX) (LL). *Po0.05; **Po0.01; ***Po0.001. UL, upper left; UR, upper right, LL, lower left, LR, lower right. (c) Morphological examination of inducible OTAG-12b cells in the presence or absence of 1.6 mg/ml DOX. At 66 h (hematoxylin staining, top layer), there are some cells in the dividing phase (arrows) in the untreated plate, whereas there are numerous apoptotic cells (arrow head) after induction by DOX. For Hoechst 33342 staining, fluorescence was observed under a fluorescence microscope. At 3 days (10 mM Hoechst 33342 staining, bottom layer) after DOX treatment, the fragment of nuclei or brightly condensed chromatin of apoptosis is shown. The bottom right shows inducible hOTAG-12b cells treated by 100 mM etoposide for 24 h as positive apoptotic control. (d) Result of DNA ladder. Assays measured internucleosomal DNA cleavage in stable clones of hOTAG-12a, hOTAG-12b, and vector cells with or without DOX induction at different time points. Lanes 1, 6 and 11 represent no DOX treatment; lanes 2, 7 and 12 represent DOX treatment for 1 day; lanes 3, 8 and 13 represent DOX treatment for 2 days; lanes 4, 9 and 14 represent DOX treatment for 3 days; lanes 5, 10 and 15 represent DOX treatment for 4 days. Ethidium bromide stain of agarose gels with genomic DNA extracted from cells. Molecular weight markers (right first lane) were also included. Very clear DNA ladder formation was found on day 3 of DOX treatment in hOTAG-12b cells, whereas there were no such effects in others. (e) Anchorage-independent assays. 5 Â 104 cells were seeded for soft agar assays. DOX was used for protein induction at 1.6 mg/ml for HEK293/TR/OTAG-12a, HEK293/ TR/OTAG-12b subclone and HEK293/TR parental cell line. Dishes were incubated for 10–14 days, colonies stained with 0.005% crystal violet and counted. (f) Results of three different experiments in triplicate were combined; the bar graph shows mean±s.e.m. of three experiments. The colony count was normalized to uninduced cells (no DOX). A significant difference was observed only for cells expressing hOTAG-12b- no colonies (***Po0.001). (g) Suppression of tumorigenesis. 1 Â 107 cells were injected subcutaneously into both flanks of 5–6-week-old female nude mice (n ¼ 6–10). After 4 weeks, mice were killed and tumor mass was removed and weighed. Note that tumor size at the site injected with higher hOTAG-12b-expressing clone H is smaller than the site receiving hOTAG-12b-L cells. (h) Tumorigenesis was suppressed by DOX-induced hOTAG-12b but not hOTAG-12a or vector DNA. Tumorigenesis was completely blocked in four out of six injection sites of hOTAG-12b cells. **Po0.01.

OTAG-12b regulates cell sensitivity to apoptotic stimuli caused by etoposide and carboplatin, hOTAG-12b’s We also investigated whether hOTAG-12b induction sensitizing effects were more evident (Figures 9a and b). would lead to increased sensitivity to apoptosis by More cells died if they were pre-exposed to DOX- known anti-cancer agents. In drug-induced apoptosis induced hOTAG-12b, such that a lower concentration

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2883

Figure 8 hOTAG-12b induces apoptosis by p53-independent pathway. (a) p53-inducing compounds also increase the expression of hOTAG-12b during apoptosis. UCI107 cells treated by drugs for 24 h (hygromycin-B 800 mg/ml; taxol 3 mM; puromycin 2 mg/ml; etoposide 20 mM; carboplatin 500 mM), RNA was extracted for Q–PCR. (b) SKOV3 (p53-null and platinum-resistant OC cell line) apoptosis induced by hOTAG-12b transfection. SKOV3 cells were transfected with OTAG-12a, OTAG-12b and pIRES2 vector using Lipofectamine 2000, then cultured in 1000 mg/ml G418 for 2–3 weeks. (c) hOTAG-12b induced apoptosis by upregulation of pro- apoptosis gene BAD, CIDEB, and GADD45a and downregulation of pro-survival gene Akt1 and NAIP. Each of the bars represents the mean±s.e.m. from three separate experiments; *Po0.05; **Po0.01 compared with control. (d) Lack of signiﬁcant phosphorylation of p53 upon DOX induction of OTAG-12b in HEK293 cells. Phospho-p53 was assessed in cell lysates at different times using a speciﬁc antibody (Ser15, Cell Signaling Technology Inc., Danvers, MA, USA). In the control HEK293 cells that were treated only by etoposide (ET), robust phosphorylation of p53 is evident after 6 h (left of d).

of etoposide (10 mM) that was ineffective by itself now gives rise to the clinical marker CA 125 for late OC, caused profound inhibition (Figure 9b). hOTAG-12b NOTCH 3, hCG b and other tumor suppressor genes also sensitized the effect of hygromycin or taxol that (see Figure 2b), as well as human disease genes such cause apoptosis via p53-independent mechanisms; at the as LDL-receptor and Apo-E, is noteworthy. Of three concentration tested, these drugs alone did not produce variants that were cloned, only hOTAG-12b, the dramatic effects on cells. predominant isoform in normal ovary, is active in cell suppression in model systems that include OC cells and xenografts (Figure 1e, Figures 3 and 7). Present evidence Discussion indicates the existence of only the hOTAG-12b form in mouse. The high structural homology between With the aim of identifying novel genes that might be human OTAG-12b and mouse OTAG-12 proteins of value in understanding the etiology of OC that is deduced from our data shows conserved function as described as a ‘silent killer’, we utilized clues emerging both proteins were effective in suppressing OC or from FORKO mice that develop ovarian tumors. Our HEK293 cells. studies reported herein conﬁrm that the novel OTAG-12 The variants hOTAG-12a and -12c, which are shorter gene downregulated in FORKO tumors as well as (107 and 92 residues, respectively) and less abundant mouse epithelial tumors (Figure 1) is also downregu- in normal ovary, are inactive towards cell suppression. lated in human OC and established OC cells (Figure 3). The nuclear location of hOTAG-12b and 12a proteins This is the ﬁrst report to identify and characterize (Figures 3e and f) is consistent with the presence human OTAG-12 and its isoforms in human ovary and of a nuclear localization signal in both proteins. As OC, and to demonstrate its potent anti-proliferative hOTAG-12b and hOTAG-12a differ only in the last 34 effects on OC cells. Existence of this gene based on amino acids, this domain in hOTAG-12b is apparently hypothetical bioinformatics predictions had suggested a important for function as supported by our C-terminal putative protein indicated as Fam32A with unknown deletion studies (Figure 4). The F14 and UP domains in function (NCBI (National Center for Biotechnology hOTAG-12b could be important for interaction with Information)). The location of hOTAG-12 on gene-rich functional partners within the nucleus to exert its human chr. 19 (Grimwood et al., 2004) in proximity inhibitory action. Several experimental manipulations to many genes implicated in OC, such as MUC 16 that used in this study provide clear evidence that increasing

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2884 Owing to its potent cell-suppressing effects, we were unable to secure stable clones of OC cells or HEK293- expressing hOTAG-12b or mouse-OTAG-12, whereas this difficulty was not encountered for hOTAG-12a (not shown). Others have reported a similar inability for overexpression of wild-type gene and for establishing cell lines for tumor suppressors such as OVCA1 (Bruening et al., 1999) and RPS6KA2 (Bignone et al., 2007) in OC cells. This led us to generate DOX- controlled strategies that resulted in securing a workable system with HEK293 cells and providing an experimental tool to manipulate hOTAG-12b expression and explore mechanistic issues. Using DOX-controlled expression, we reveal that elevated hOTAG-12b effects can be reversed following termination of gene induction. Other evidence includes the fact that a related protein such as hOTAG-12a differing only in the carboxyl-terminus and providing a natural control is not detrimental to cells. Our C-terminal deletion studies of hOTAG-12 (Figure 4) also lend additional support to the fact that apoptotic effects are unique to the whole structure and are of physiological/pathological significance. Addressing the molecular nature of cellular inhibition, we have found that hOTAG-12b stimulates expression of several pro-apoptotic genes such as BAD (Willis and Adams, 2005; Boisvert-Adamo and Aplin, 2008), CIDEB (Lugovskoy et al., 1999; Chen et al., 2000) and GADD45a (Thyss et al., 2005; Al-Romaih et al., 2008), and inhibits expression of anti-apoptotic genes such as NAIP (Liston et al., 2003; O’Riordan et al., 2008) and Akt1 (Zhu et al., 2008). hOTAG-12b, similar to p53, induces a set of overlapping genes such as GADD45a and distinct genes such as Akt1, BAD, CIDEB and NAIP. Although p53-inducing compounds in OC cells (Figure 8a) could stimulate the expression of endogenous hOTAG-12b, transfection of hOTAG-12b by itself does not stimulate p53 or its direct target genes such as p21 and BAX. These results suggest that hOTAG-12b induces apoptosis via a p53-independent pathway (a result confirmed by overexpression of hOTAG-12b in p53-null cell line-SKOV3 cells); how- Figure 9 hOTAG-12 regulates the sensitivity of cells to apoptotic ever, we could not exclude hOTAG-12b inducing stimuli. The stable clones of hOTAG-12a, hOTAG-12b and vector apoptosis through p53 downstream signaling. Our were left untreated or treated with DOX for 24 h (removed after that) and then subjected to the following treatments for another current working model for hOTAG-12b’s effects de- 20 h: (a) Etoposide- (topoisomerase II inhibitor) as an example at picted in the scheme (Supplementary Figure 5) includes the concentration indicated. Cells were stained by trypan blue. the cumulative effect of opposing pathways—activation (b) Etoposide-ET, Carboplatin-CB, Hygromycin-HM and Taxol-TA of cell cycle arrest and apoptosis and inhibition of stimulation. A total of 7000 cells were seeded in 96-well plates with medium change after 1 day for treatment with or without cellular proliferation. Although the target gene(s) DOX for 24 h. Following DOX removal, cells were exposed to the responsible for the primary effect of expression of compounds. In each case, the number of cells was determined by hOTAG-12b remains unclear, stimulated expression of MTT assay. Each bar represents mean±s.e.m. from three separate GADD45a as early as 12 h after DOX induction experiments; ***Po0.001 compared with DOX treatment only. (Figure 8c) is noteworthy. GADD45a is also regulated by a p53-independent pathway (Chen et al., 2001; Zhu et al., 2008). In addition, GADD45a, a known transcriptional target of BRCA1 (Harkin et al., 1999), is the level of OTAG-12b protein suppresses cell growth responsible at least in part for the regulation of a G2 cell by inhibiting cell proliferation, inducing G2 cell cycle cycle checkpoint (Wagner et al., 2008). arrest, and also by promoting apoptosis. Such functions The in vivo effects of hOTAG-12b on xenografts in have also been reported for tumor suppressor genes such nude mice (Figures 7g and h) suggest good potential for as IRF-5 (Barnes et al., 2003). tumor abrogation. In addition, data emerging from our

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2885 study on combinatorial effects of hOTAG-12b and anti- View, CA, USA), with V5 and GFP tag. pIRES2-expressing cancer agents in enhancing cellular sensitivity are OTAG-12 or empty vector as control was transfected using interesting for practical considerations (Figure 9). This Lipofectamine 2000 (Invitrogen). Carboxyl deletion mutants assumes importance because drug resistance is the main of hOTAG-12b were generated by PCR using the primers clinical issue confronting OC. Although 75% of patients indicated in Supplementary Table 1 and cloned into the same with advanced disease respond to initial (carboplatin- vector for cell transfections. For anchorage-dependent assay, at 24 h post-transfection, containing) chemotherapy, the majority will relapse and cells were trypsinized, washed, diluted 1/10 and seeded onto 25% of patients are intrinsically resistant, that is, will six-well plates in media containing G418 (400–1200 mg/ml) for not respond to chemotherapy at all (Bast Jr et al., 2009). selection. Cells were grown for 2–3 weeks; colonies were One approach to overcoming resistance in EOC is stained with 0.1% Coomassie blue and counted. At least to reactivate pro-apoptosis pathways. Chemoresistant three experiments were performed in quadruplicate for each OC cells become sensitive to cisplatin-induced apop- cell line. tosis following PTEN overexpression (Yan et al., 2006). Thus, discovery of a novel cell suppressor like hOTAG- DOX-controlled expression system 12b is important because induction of tumor cell Inducible stable cell lines overexpressing the gene of interest senescence may represent a therapeutic approach for under the control of Tet-repressor were generated using cancer. This would require strategies to upregulate T-REX Tetracycline-Regulated Mammalian Expression Sys- hOTAG-12b gene expression or protein delivery. As tem (provided by Dr Jolicoeur). The rtTA retrovirus for hOTAG-12b is the simplest protein among all currently expression of tetracycline repressor protein (TetR) has puromycin selection. Each gene was cloned in LITMUS 29/ known ovarian tumor suppressors (Bast Jr et al., 2009), TO (G418 resistance). HEK293 and UCI101/107 cells were the prospects for bio-engineering appear favorable. transfected simultaneously or sequentially with repressor and To further elucidate the role of OTAG-12 in OC, expression vectors using Lipofectamine 2000. Transfected cells generating variant-speciﬁc antibodies could help future were selected in the presence of 2 mg/ml puromycin and 600– studies for analyzing the protein in tumors including 1000 mg/ml G418 for 2–3 weeks. Individual clones were heritable cases. Securing OTAG-12-null mice will help expanded and tested for expression of transfected gene in the address issues in development and function. Understand- presence of DOX. ing transcriptional control and silencing of hOTAG-12b in tumors and the biochemical basis of its function in Proliferation assay-MTT inhibiting cell growth as well as cellular localization or In total, 2000 cells were plated onto a 96-well plate in alteration in cancers would also be important. Dulbecco’s modiﬁed Eagle’s medium–10% fetal bovine serum and left to adhere. After 24 h, 1.6 mg/ml DOX was added. Proliferation was measured by MTT assay (Vybrant MTT Materials and methods Cell Proliferation Assay Kit, Invitrogen). Absorbance was measured at 570 nm. At least three experiments were Cell lines and culture conditions performed in quadruplicate. The following cells used in this study—HC11-26 mammary tumor cells (Dr Jolicoeur), ID8 mouse ovarian epithelial cells Soft agar assays (Dr Roby), OC cells UCI101, UCI107 cells (Dr Carpenter), To assess anchorage-independent growth, we mixed the cells at SKOV3, OVCAR3 and OVCAR4 (ATCC), HEK293 (ATCC 37 1C with 0.4% agarose in complete medium and poured the CRL-1573, transformed with adenovirus-5 DNA)—were mixture over a layer of 0.8% agarose in six-well plates. After grown in the medium according to the provider’s recommen- 15 min at room temperature, plates were incubated at 37 1Cfor dations. 2–3weeks.Theywerethenstainedbyp-iodonitrotetrazolium violet and incubated at 37 1C for 6 h, after which colonies RNA isolation and cDNA preparation and RT–PCR were photographed and counted using microscope-associated and Q–PCR software. RNA was isolated from human ovarian samples (collected by Gotlieb lab) as well as cell lines, using TRIzol Reagent Assessment of cell viability (Invitrogen, Carlsbad, CA, USA) according to the manufac- Cell viability was determined by examining the apoptotic turer’s instructions. Total RNA (2 mg) was reverse transcribed morphology of nuclear staining with hematoxylin or Hoechst using SuperScript II Reverse Transcriptase (Invitrogen, 33342 (Sigma) as described (Allen et al., 2001). Carlsbad, CA, USA), and Oligo-dT. Reverse transcriptase– PCR (RT–PCR) and Q–PCR were performed as described Fragmentation by DNA ladder (Aravindakshan et al., 2010). DNA ladder formation in DOX-inducible clones was determined as described by Wu et al. (2005). DNA fragments were Gene cloning—RACE separated on a 1.5% agarose gel, visualized with ethidium To obtain full-length 50and 30 ends of cDNA using the bromide and photographed. sequence from expressed sequence tags, RACE was performed using GeneRacer Kit (Invitrogen). After cloning RACE Fluorescence-activated cell sorting apoptosis assay products, both DNA strands were sequenced. The induction of apoptosis by OTAG-12 was assessed by Annexin V-APC assay according to the manufacturer’s Transfection and colony formation assays instructions (Annexin V-APC Apoptosis Detection Kit, A cDNA fragment containing each OTAG-12 ORF was EBioscience, San Diego, CA, USA). Vector, hOTAG-12a cloned into vector pIRES2 (Clontech Laboratories, Mountain and hOTAG-12b HEK293-inducible stable clone cells were

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2886 incubated with or without 1.6 mg/ml DOX for 1, 2 and 3 days. tumor growth in animals (n ¼ 10) was checked weekly. In all, Detached cells were labeled with Annexin V-APC. PI was 2 mg/ml DOX was added to the drinking water 1 day before added to stain necrotic, dead and membrane-compromised injection and changed twice a week. After 4 weeks, mice were cells. Cells were analyzed by ﬂuorescence-activated cell sorting killed, and tumor mass at each location was measured. All of 10 000 events. A log FL1 (X-axis) versus log FL3 (Y-axis) procedures were carried out with approval by the Institutional dot plot was generated. Animal Care and Use Committee.

Cell cycle analysis Cell sensitization Cell cycle analysis was done according to the modified Krishan UCI101 and UCI107 cells were grown 106/well in six-well procedure. Briefly, 1 Â 106 DOX-inducible stable cells were plates and treated with or without the compounds for 24 h. washed and resuspended in 1 ml of modified Krishan staining HEK293-inducible stable cells were treated by DOX for 24 h buffer (see Figure 6b). Cells were analyzed by FACS (Becton and washed with Hank’s balanced salt solution, then incubated Dickenson, Cowley, UK) using MODFIT LT for MACPPC with or without compounds. software. A total of 10 000 single events were recorded each time, and the percentage of cells in G1, G2 and S phases was Statistics calculated. The experiments were repeated at least three times Student’s t test was done to assess significance (P 0.05). Data for each variable. o are shown as mean±s.e.m. Total protein extraction and subcellular fractionation Cell lysates for western blotting were prepared as described (Chen et al., 2006). Cytoplasmic and nuclear fractions were Conflict of interest prepared according to Dai et al. (2009). The authors declare no conflict of interest. Immunohistochemistry Subcellular localization in transfected cells was performed using the V5 antibody and standard methods as employed in Acknowledgements our laboratory (Aravindakshan et al., 2006b). We are grateful to Dr P Carpenter (UC Irvine) for providing Tumorigenicity assay ovarian cancer cell lines UCI101 and UCI107, Dr Roby KF Cells grown in log-phase and harvested by trypsinization (Univ Kansas) for ID8 cells, Drs Jolicoeur for HC11-26 cells were washed twice with PBS and resuspended in Dulbecco’s and Seidah at IRCM for the modified HEK293 cell lines. We modified Eagle’s medium to a final concentration of 3.3 Â 107/ appreciate the useful suggestions provided by Drs Zhongjun ml. The cells (1 Â 107 cells in 300 ml PBS/site) were injected Dong and Hua Lin during this investigation. We also subcutaneously into both flanks of 5- to 6-week-old female acknowledge the help of Drs Marie-Claude Beauchamp and nude mice (nu/fox1, Harlan Laboratories, Indianapolis, IN, Amber Yasmeen in this study. This investigation was USA). A total of 10 sites were tested for each cell line and supported in part by the NCIC and the CIHR.

References

Al-Romaih K, Sadikovic B, Yoshimoto M, Wang Y, Zielenska M, Bignone PA, Lee KY, Liu Y, Emilion G, Finch J, Soosay AE et al. Squire JA. (2008). Decitabine-induced demethylation of 50 CpG (2007). RPS6KA2, a putative tumour suppressor gene at 6q27 in island in GADD45A leads to apoptosis in osteosarcoma cells. sporadic epithelial ovarian cancer. Oncogene 26: 683–700. Neoplasia 10: 471–480. Boisvert-Adamo K, Aplin AE. (2008). Mutant B-RAF mediates Allen S, Sotos J, Sylte MJ, Czuprynski CJ. (2001). Use of Hoechst resistance to anoikis via Bad and Bim. Oncogene 27: 3301–3312. 33342 staining to detect apoptotic changes in bovine mononuclear Bruening W, Prowse AH, Schultz DC, Holgado-Madruga M, Wong phagocytes infected with Mycobacterium avium subsp. paratuber- A, Godwin AK. (1999). Expression of OVCA1, a candidate tumor culosis. Clin Diagn Lab Immunol 8: 460–464. suppressor, is reduced in tumors and inhibits growth of ovarian Aravindakshan J, Chen X, Sairam MR. (2006a). Differential expres- cancer cells. Cancer Res 59: 4973–4983. sion of claudin family proteins in mouse ovarian serous papillary Chen F, Zhang Z, Leonard SS, Shi X. (2001). Contrasting roles of NF- epithelial adenoma in aging FSH receptor-deﬁcient mutants. kappaB and JNK in arsenite-induced p53-independent expression of Neoplasia 8: 984–994. GADD45alpha. Oncogene 20: 3585–3589. Aravindakshan J, Chen XL, Sairam MR. (2006b). Age-dependent Chen X, Aravindakshan J, Yang Y, Sairam MR. (2007). Early bimodal GDNF regulation during ovarian tumorigenesis in alterations in ovarian surface epithelial cells and induction of follitropin receptor mutant mice. Biochem Biophys Res Commun ovarian epithelial tumors triggered by loss of FSH receptor. 351: 507–513. Neoplasia 9: 521–531. Aravindakshan J, Chen XL, Sairam MR. (2010). Chronology Chen X, Aravindakshan J, Yang Y, Tiwari-Pandey R, Sairam MR. and complexities of ovarian tumorigenesis in FORKO mice: (2006). Aberrant expression of PDGF ligands and receptors in the age-dependent gene alterations and progressive dysregulation of tumor prone ovary of follitropin receptor knockout (FORKO) Major Histocompatibility Complex (MHC) Class I and II proﬁles. mouse. Carcinogenesis 27: 903–915. Mol Cell Endocrinol 329: 37–46. Chen Z, Guo K, Toh SY, Zhou Z, Li P. (2000). Mitochondria Barnes BJ, Kellum MJ, Pinder KE, Frisancho JA, Pitha PM. (2003). localization and dimerization are required for CIDE-B to induce Interferon regulatory factor 5, a novel mediator of cell cycle arrest apoptosis. J Biol Chem 275: 22619–22622. and cell death. Cancer Res 63: 6424–6431. Corney DC, Flesken-Nikitin A, Choi J, Nikitin AY. (2008). Bast Jr RC, Hennessy B, Mills GB. (2009). The biology of ovarian cancer: Role of p53 and Rb in ovarian cancer. Adv Exp Med Biol 622: new opportunities for translation. Nat Rev Cancer 9: 415–428. 99–117.

Oncogene Novel ovarian tumor suppressor gene X Chen et al 2887 Dai Y, Liu M, Tang W, Li Y, Lian J, Lawrence TS et al. (2009). A N-terminal extension that inhibits growth of ovarian and breast Smac-mimetic sensitizes prostate cancer cells to TRAIL-induced cancers. Oncogene 22: 2897–2909. apoptosis via modulating both IAPs and NF-kB. BMC Cancer 9: Maertens GN, El Messaoudi-Aubert S, Elderkin S, Hiom K, Peters G. 392–406. (2010). Ubiquitin-specific proteases 7 and 11 modulate Polycomb Danilovich N, Roy I, Sairam MR. (2001). Ovarian pathology and high regulation of the INK4a tumour suppressor. EMBO J 29: incidence of sex cord tumors in follitropin receptor knockout 2553–2565. (FORKO) mice. Endocrinology 142: 3673–3684. Nowak-Markwitz E, Jankowska A, Andrusiewicz M, Szczerba A. Dievart A, Beaulieu N, Jolicoeur P. (1999). Involvement of Notch1 (2004). Expression of beta-human chorionic gonadotropin in in the development of mouse mammary tumors. Oncogene 18: ovarian cancer tissue. Eur J Gynaecol Oncol 25: 465–469. 5973–5981. O’Riordan MX, Bauler LD, Scott FL, Duckett CS. (2008). Inhibitor of Dinulescu DM, Ince TA, Quade BJ, Shafer SA, Crowley D, Jacks T. apoptosis proteins in eukaryotic evolution and development: a (2005). Role of K-ras and Pten in the development of mouse model of thematic conservation. Dev Cell 15: 497–508. models of endometriosis and endometrioid ovarian cancer. Nat Med Pal T, Permuth-Wey J, Betts JA, Krischer JP, Fiorica J, Arango H 11: 63–70. et al. (2005). BRCA1 and BRCA2 mutations account for a large Dougherty MK, Morrison DK. (2004). Unlocking the code of 14-3-3. proportion of ovarian carcinoma cases. Cancer 104: 2807–2816. J Cell Sci 117: 1875–1884. Park JT, Li M, Nakayama K, Mao TL, Davidson B, Zhang Z et al. Fuchtner C, Emma DA, Manetta A, Gamboa G, Bernstein R, Liao (2006). Notch3 gene amplification in ovarian cancer. Cancer Res 66: SY. (1993). Characterization of a human ovarian carcinoma cell 6312–6318. line: UCI 101. Gynecol Oncol 48: 203–209. Roby KF, Taylor CC, Sweetwood JP, Cheng Y, Pace JL, Tawfik O Gamboa G, Carpenter PM, Podnos YD, Dorion G, Iravani L, et al. (2000). Development of a syngeneic mouse model for events Bolton D et al. (1995). Characterization and development of UCI related to ovarian cancer. Carcinogenesis 21: 585–591. 107, a primary human ovarian carcinoma cell line. Gynecol Oncol Schwartz MF, Lee SJ, Duong JK, Eminaga S, Stern DF. (2003). FHA 58: 336–343. domain-mediated DNA checkpoint regulation of Rad53. Cell Cycle Grimwood J, Gordon LA, Olsen A, Terry A, Schmutz J, Lamerdin J 2: 384–396. et al. (2004). The DNA sequence and biology of human chromo- Thyss R, Virolle V, Imbert V, Peyron JF, Aberdam D, Virolle T. (2005). some 19. Nature 428: 529–535. NF-kappaB/Egr-1/Gadd45 are sequentially activated upon UVB Harkin DP, Bean JM, Miklos D, Song YH, Truong VB, Englert C irradiation to mediate epidermal cell death. EMBO J 24: 128–137. et al. (1999). Induction of GADD45 and JNK/SAPK-dependent Wagner MW, Li LS, Morales JC, Galindo CL, Garner HR, Bornmann apoptosis following inducible expression of BRCA1. Cell 97: WG et al. (2008). Role of c-Abl kinase in DNA mismatch repair- 575–586. dependent G2 cell cycle checkpoint arrest responses. J Biol Chem Jarboe EA, Folkins AK, Drapkin R, Ince TA, Agoston ES, Crum CP. 283: 21382–21393. (2008). Tubal and ovarian pathways to pelvic epithelial cancer: a Willis SN, Adams JM. (2005). Life in the balance: how BH3-only pathological perspective. Histopathology 53: 127–138. proteins induce apoptosis. Curr Opin Cell Biol 17: 617–625. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. (2009). Cancer Wu R, Hendrix-Lucas N, Kuick R, Zhai Y, Schwartz DR, Akyol A statistics, 2009. CA Cancer J Clin 59: 225–249. et al. (2007). Mouse model of human ovarian endometrioid Johnson GL, Lapadat R. (2002). Mitogen-activated protein kinase adenocarcinoma based on somatic defects in the Wnt/beta-catenin pathways mediated by ERK, JNK, and p38 protein kinases. Science and PI3K/Pten signaling pathways. Cancer Cell 11: 321–333. 298: 1911–1912. Wu Z, Tandon R, Ziembicki J, Nagano J, Hujer KM, Miller RT et al. Konstantinopoulos PA, Spentzos D, Cannistra SA. (2008). Gene- (2005). Role of ceramide in Ca2+-sensing receptor-induced expression profiling in epithelial ovarian cancer. Nat Clin Pract apoptosis. J Lipid Res 46: 1396–1404. Oncol 5: 577–587. Yan X, Fraser M, Qiu Q, Tsang BK. (2006). Over-expression of PTEN Liston P, Fong WG, Korneluk RG. (2003). The inhibitors of sensitizes human ovarian cancer cells to cisplatin-induced apoptosis apoptosis: there is more to life than Bcl2. Oncogene 22: in a p53-dependent manner. Gynecol Oncol 102: 348–355. 8568–8580. Yin BW, Dnistrian A, Lloyd KO. (2002). Ovarian cancer antigen Lugovskoy AA, Zhou P, Chou JJ, McCarty JS, Li P, Wagner G. CA125 is encoded by the MUC16 mucin gene. Int J Cancer 98: (1999). Solution structure of the CIDE-N domain of CIDE-B and a 737–740. model for CIDE-N/CIDE-N interactions in the DNA fragmenta- Zhu QS, Ren W, Korchin B, Lahat G, Dicker A, Lu Y et al. (2008). tion pathway of apoptosis. Cell 99: 747–755. Soft tissue sarcoma cells are highly sensitive to AKT blockade: a Luo RZ, Fang X, Marquez R, Liu SY, Mills GB, Liao WS et al. role for p53-independent up-regulation of GADD45 alpha. Cancer (2003). ARHI is a Ras-related small G-protein with a novel Res 68: 2895–2903.

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

Oncogene