Oncogene (2009) 28, 1941–1948 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE NAC-1, a potential stem cell pluripotency factor, contributes to paclitaxel resistance in ovarian through inactivating pathway

N Jinawath1, C Vasoontara1, K-L Yap1, MM Thiaville1, K Nakayama2, T-L Wang1 andI-M Shih 1

1Departments of Pathology, Oncology, and Gynecology and Obstetrics, Johns Hopkins Medical Institutions, Baltimore, MD, USA and 2Department of Gynecology and Obstetrics, Shimane University, Izumo, Japan

Nucleus accumbens-1 (Nac1 or NAC-1) belongs to the 22 430 new cases and15 280 deathsin 2007 (Jemal et al., BTB/POZ (Pox virus and Zinc finger/Bric-a-brac Tram- 2007). Unlike other histological types of ovarian track Broad complex) transcription factor family and is a epithelial tumors, high-grade serous carcinoma is the novel that potentially participates in self-renewal most aggressive andmost rapidly growing neoplasm, and pluripotency in embryonic stem cells. In human andhas a 5-year survival rate below 30%. Patients with cancer, NAC-1 is upregulated in several types of high-grade serous carcinomas always present their neoplasms, but particularly in recurrent chemoresistant diseases at the advanced stages. These patients routinely ovarian carcinomas, suggesting a biological role for NAC- receive cytoreduction surgery followed by a combined 1 in the development of drug resistance in ovarian cancer. carboplatin/paclitaxel chemotherapy regimen. Despite We have assessed this possibility and shown a correlation initial responsiveness to the chemotherapeutic drugs, between NAC-1 expression and ex vivo paclitaxel most patients eventually develop chemoresistant tumors resistance in ovarian serous carcinoma tissues and cell andsuccumb to their diseases.Therefore, treatment lines. We found that expression of Gadd45-c-interacting failure is often attributedto primary or acquired protein 1 (Gadd45gip1), a downstream target negatively resistance to chemotherapeutic agents, which remains regulated by NAC-1, was reduced in paclitaxel-resistant a clinically formidable problem in managing most cells. Ectopic expression of NAC-1 or knockdown of cancer patients. Understanding chemoresistance at the Gadd45gip1 conferred paclitaxel resistance, whereas molecular level shouldilluminate fundamental proper- NAC-1 knockdown or ectopic expression of Gadd45gip1 ties of drug resistance and provide new therapeutic increased paclitaxel sensitivity. Furthermore, silencing targets to treat advanced neoplastic diseases. NAC-1 expression or disrupting NAC-1 homodimeriza- In an effort to elucidate the molecular etiology of tion by a dominant negative NAC-1 protein that contained drug resistance, we have earlier identified and character- only the BTB/POZ domain induced the expression of izedthe nucleus accumbens-1 (Nac1 or NAC-1) as one Gadd45c, which interacted with Gadd45gip1. Reducing of the candidate chemoresistance in ovarian Gadd45c expression by small hairpin RNAs partially cancer (Nakayama et al., 2006a, b). NAC-1 is a nuclear enhanced paclitaxel resistance. Thus, this study provides protein belonging to the BTB/POZ (Pox virus andZinc new evidence that NAC-1 upregulation and homodimer- finger/Bric-a-brac Tramtrack Broadcomplex) domain ization contribute to tumor recurrence by equipping family. NAC-1 was originally identifiedandclonedas a ovarian cancer cells with the paclitaxel-resistant pheno- novel transcript from the NAC, a unique forebrain type through negative regulation of the Gadd45 pathway. structure involvedin rewardmotivation andaddictive Oncogene (2009) 28, 1941–1948; doi:10.1038/onc.2009.37; behaviors (Koob, 1996; Cha et al., 1997; Kalivas et al., publishedonline 23 March 2009 1999; Mackler et al., 2000), andit was later shown to be involvedin the acute behavioral andneurochemical Keywords: ovarian; NAC-1; chemoresistance; paclitaxel responses to psychomotor stimulants (Mackler et al., 2008). Recently, NAC-1 has been foundto be one of the factors associatedwith pluripotency in embryonic stem (ES) cells. NAC-1 transcriptionally regulates Introduction expression of other pluripotency factors, including Nanog, Oct4, Sox2, Klf4, Sall1 andSall4, andpartici- Ovarian cancer represents the most lethal gynecological pates in constituting the protein interaction and malignant disease in the United States with an estimated transcriptional regulatory networks that are critical to establish and/or maintain ES cell pluripotency (Wang Correspondence: Dr I-M Shih, Department of Pathology, Johns et al., 2006; Kim et al., 2008; Muller et al., 2008). In Hopkins Medial Institutions, CRB-II, Room 306, 1550 Orleans Street, human cancer, NAC-1 is upregulatedin several neo- Baltimore, MD 21231, USA. plasms, including ovarian high-grade serous carcinoma, E-mail: [email protected] Received14 October 2008; revised18 February 2009; accepted19 pancreatic carcinoma, colorectal carcinoma, breast February 2009; publishedonline 23 March 2009 carcinoma, renal cell carcinoma, hepatocellular carci- NAC-1 in paclitaxel resistance N Jinawath et al 1942 noma andcervical cancer (Nakayama et al., 2006a, b; a 100 OVCAR3 TR Yeasmin et al., 2008). In ovarian high-grade serous A2780 TR carcinomas, the levels of NAC-1 expression were 80 SKOV3 TR significantly higher in recurrent post-chemotherapy 60 OVCAR3 naive samples than in primary earlier untreatedtumors A2780 naive 40 SKOV3 naive (Nakayama et al., 2006a, b; Davidson et al.,

2007). Furthermore, intense NAC-1 immunoreactivity Cell number 20

in primary ovarian serous carcinomas was associated (% of untreated cells) 0 with early recurrence, suggesting that NAC-1 is 0.0 0.5 2000 4000 6000 8000 10000 potentially involvedin the pathogenesis of drug Paclitaxel (nM) resistance. b At the molecular level, NAC-1 form homo- OVCAR3 TR dimers through the BTB/POZ domain, and this process OVCAR3 naive is essential for tumor-cell growth andsurvival in tumors with NAC-1 overexpression (Nakayama et al., A2780 TR 2006a, b). NAC-1 has been shown to interact A2780 naive with nuclear proteins, including CoRest and histone deacetylases, HDAC3 and HDAC4, in which the SKOV3 TR binding with deacetylases is mediated by a unique SKOV3 naive BEN domain on NAC-1 protein (Korutla et al., 2005, 2007; Abhiman et al., 2008). Ectopic expression of full- 0 10 20 30 40 50 60 length NAC-1 proteins is sufficient to enhance tumor- Relative NAC-1 mRNA expression igenecity of ovarian surface epithelial cells andof Figure 1 Upregulation of NAC-1 transcript in paclitaxel-resistant NIH3T3 cells in athymic nu/nu mice. We have pre- ovarian cancer cell lines, including SKOV3, OVCAR3 and A2780. (a) Dose-dependent decrease in cell number in the presence of viously identified growth arrest and DNA-damage- paclitaxel. Paclitaxel-resistant cells (TR) are less sensitive to inducible Gadd45-g interacting protein (Gadd45gip1) paclitaxel than are the parental (naive) controls. (b) Paclitaxel- as one of the genes negatively regulatedby NAC-1, and resistant cell lines were analysedfor NAC-1 mRNA expression have shown that NAC-1 contributes to tumor growth using quantitative real-time reverse transcription PCR. NAC-1 expression levels were normalizedto the GAPDH expression level. andsurvival, at least in part, by inhibiting Gadd45gip1 NAC-1 is upregulatedin all paclitaxel-resistant cell lines. GAPDH, expression (Nakayama et al., 2007). The above findings glyceraldehyde-3-phosphate dehydrogenase; NAC-1, nucleus strongly suggest that NAC-1 is a tumor recurrence- accumbens-1. associatedgene in high-gradeovarian serous carcino- mas. In this study, by correlating NAC-1 expression and drug-resistance status, we determined that NAC-1 Table 1 Correlation of NAC-1 immunoreactivity andextreme drug resistance (EDR) status in ovarian cancer tissues causally contributes to chemoresistance for paclitaxel andcarboplatin andshowedthat the Gadd45pathway Taxol-EDR Taxol-LDR is involvedin NAC-1-mediatedchemoresistance in Positive staining 51 41 cancer cells. Negative staining 7 14 Fisher’s exact test P-value ¼ 0.05

Carbo-EDR Carbo-LDR

Results Positive staining 44 48 Negative staining 9 12 Correlation of NAC-1 expression and paclitaxel Fisher’s exact test P-value ¼ 0.43 resistance in ovarian carcinoma Three ovarian cancer cell lines, SKOV3, OVCAR3 and Abbreviations: EDR, extreme drug resistance; LDR, low drug A2780, were usedto generate paclitaxel-resistant cells. resistance. Dose-dependent growth inhibition effects by paclitaxel are shown in Figure 1a. The IC50s of paclitaxel at 96 h for which the ex vivo drug-response status was available. were o0.1 nM in all the three parental naive cell lines, We showeda correlation between the intensity of NAC-1 but were more than 5000 nM in all the three paclitaxel- immunoreactivity andan extreme paclitaxel-resistance resistant cell lines. Quantitative real-time reverse tran- status (P ¼ 0.05) (Table 1). In contrast, there was no scription PCR was usedto determineNAC-1 transcript significant correlation (P ¼ 0.43) between carboplatin levels in the resistant andparental naive cells from each resistance andNAC-1 immunointensity. Therefore, we cell line. As shown in Figure 1b, NAC-1 mRNA was focusedon the mechanism underlying how NAC-1 highly expressedin paclitaxel-resistant cells as compared participates in paclitaxel resistance in ovarian cancer. with parental cells in all three different cell lines (Student’s t test P-value o0.001). To extrapolate the above findings to clinical specimens, we performed NAC-1 overexpression confers paclitaxel resistance immunohistochemistry using an anti-NAC-1 antibody To assess whether or not NAC-1 expression causally on a total of 113 high-grade ovarian serous carcinomas, contributes to paclitaxel resistance, we usedboth

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1943 ab1.2 125 Vector 1.0 C1 100 C2 0.8 C3 75 0.6

50 0.4

Cell number 0.2

(% of untreated) 25 Relative expression ratio expression Relative 0.0 0 10 100 1000 Paclitaxel (nM) trol shRNA shNAC1-1 shNAC1-3 Con

60000 V V Taxol C1 C1 Taxol c 125 Control shRNA C2 C2 Taxol shNAC1-1 C3 C3 Taxol 100 shNAC1-3 40000 75

50 20000 Cell number

(% of untreated) 25

Relative fluorescent intensity Relative 0 0 0 1 2 3 4 100 1000 10000 Days after Paclitaxel treatment Paclitaxel (nM) Figure 2 NAC-1 expression is sufficient andessential to inducepaclitaxel (taxol) resistance. ( a) Upper panel: three NAC-1-expressing NIH3T3 clones (C1, C2 andC3) show an increasedresistance to paclitaxel as comparedwith the vector-only control. Lower panel: the effect of paclitaxel on cell growth in both control andtaxol resistant NIH3T3 clones. ( b) The efficiency of two NAC-1 shRNAs (À1 and À3) in reducing NAC-1 expression was determined by quantitative real-time reverse transcription PCR in SKOV3 cells. (c) NAC-1 shRNA significantly sensitizedpaclitaxel-resistant SKOV3 cells to paclitaxel treatment. The effect is more pronouncedat higher drug concentrations. NAC-1, nucleus accumbens-1; shRNA, small hairpin RNA. forward(overexpression) andreverse (gene knockdown) Gadd45gip1 is associated with paclitaxel resistance approaches. First, NIH3T3 cells were transfectedwith a in ovarian cancer cells NAC-1 expression vector andstable NAC-1-expressing Nucleus accumbens-1 (NAC-1) is thought to be a clones were selected. As shown in Figure 2a, three transcription regulator andhas been shown to nega- independent NAC-1-expressing clones were more resis- tively control Gadd45gip1 expression (Nakayama et al., tant to paclitaxel-induced growth suppression than a 2007). Thus, it is likely that Gadd45gip1 may contribute, control clone transfectedwith the vector alone. The at least in part, to the paclitaxel-resistance phenotype. IC50s of paclitaxel at 96 h were 785, 755 and692 n M in To test this hypothesis, we first comparedthe mRNA NAC-1-expressing clones in contrast to 59 nM in the levels of Gadd45gip1 between paclitaxel-resistant and control clone (Figure 2a, upper panel). Growth curve parental naive ovarian cancer cell lines, andfoundthat analysis showedan increase in proliferation rate of the Gadd45gip1 transcript level was significantly re- NAC-1-expressing clones as comparedwith the control duced in paclitaxel-resistant cells as compared with their cells, in the presence of paclitaxel (Figure 2a, lower naive counterparts in all three cell lines (Figure 3a). panel). Next, we appliedsmall hairpin RNAs (shRNAs) Next, we knocked down Gadd45gip1 expression using targeting NAC-1 in SKOV3 cells that expressedhigh shRNA to determine whether reducing Gadd45gip1 levels of NAC-1 protein. The efficiency of shRNAs in level enhancedpaclitaxel resistance. Figure 3b shows reducing NAC-1 protein expression is shown in that the mRNA levels of Gadd45gip1 were significantly Figure 2b. Similar to our previous finding (Nakayama reduced 48 h after treating paclitaxel-resistant SKOV3 et al., 2006a, b), NAC-1 knockdown resulted in growth cells with two Gadd45gip1 shRNAs. Growth curve suppression in parental SKOV3 cells, but not in analysis showed that Gadd45gip1 shRNA transfection paclitaxel-resistant SKOV3 cells (Supplementary did not significantly alter the proliferation rate in either Figure 1A). The sensitivity to paclitaxel was then paclitaxel-resistant or control SKOV3 cells (Suppleme- determined in control and NAC-1 shRNA-treated ntary Figure 1B). However, Gadd45gip1 shRNAs paclitaxel-resistant SKOV3 cells. We foundthat decreased paclitaxel sensitivity in paclitaxel-resistant NAC-1 knockdown increased paclitaxel sensitivity, SKOV3 cells. The IC50s of paclitaxel (measuredafter especially at higher concentrations of paclitaxel. The 96 h) were 3108 nM in the Gadd45gip1-knockdown IC50s of paclitaxel at 96 h were 2839 and3275 n M in paclitaxel-resistant cells, and1805 n M in the control shNAC1-1 andshNAC1-3-treatedgroups, resepectively, paclitaxel-resistant cells (Figure 3c). Likewise, Gadd45- and12 087 n M in the control shRNA-treatedcells gip1 shRNAs also decreased paclitaxel sensitivity in (Figure 2c). naive SKOV3 cells (data not shown). Furthermore, we

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1944

a b 1.2 1.0 OVCAR3 TR 0.8 OVCAR3 naive 0.6 a A2780 TR 0.4 A2780 naive SKOV3-TRSKOV3-TR-Gadd45gip1 0.2 expression ratio expression 25 kDa - Anti-V5 SKOV3 TR Gad45gip1 relative 0.0 SKOV3 naive GAPDH 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1 shRNA Gadd45gip1 relative b 125 expression ratio Control shRNA SKOV3-TR SKOV3-TR-Gadd45gip1 Gadd45gip 100 c 125 TR-shControl 75 100 TR-shGadd45gip1 50 75 Cell number

50 (% of untreated) 25

Cell number 25 0 (% of untreated) 100 1000 10000 0 10 100 1000 10000 Paclitaxel (nM) Paclitaxel (nM) Figure 4 Gadd45gip1 overexpression reduced paclitaxel-resistant property in paclitaxel chemoresistant cells. Paclitaxel-resistant Figure 3 The role of Gadd45gip1 in NAC-1-mediated paclitaxel SKOV3 (TR) cells were infected with Gadd45gip1-V5-expressing resistance. (a) The level of Gadd45gip1 mRNA as determined by retrovirus or control virus containing only the pWZL vector. quantitative real-time reverse transcription (RT)–PCR was de- (a) Western blot shows Gadd45gip1-V5 expression in Gadd45gip1- creasedin paclitaxel-resistant OVCAR3, A2780 andSKOV3 cells V5 retrovirus-infectedSKOV3-TR cells. ( b) At 48 hours after as comparedwith the parental naı ¨ ve counterparts. (b) The infection, cells were incubatedwith a series of concentrations of efficiency of Gadd45gip1 shRNA in reducing its mRNA level is paclitaxel andthe IC50 was measured4 dayslater. Ectopic shown by quantitative real-time RT–PCR in paclitaxel-resistant Gadd45gip1 expression increased the paclitaxel sensitivity in SKOV3 cells. The level of Gadd45gip1 expression was normalized SKOV3-TR cells. Gadd45gip1, Gadd45-g-interacting protein 1. to that of GAPDH. (c) Reducing Gadd45gip1 using shRNA in cells increases paclitaxel resistance as comparedwith control shRNA treatment. Gadd45gip1, Gadd45-g-interacting protein 1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; shRNA, small hairpin RNA; TR, paclitaxel-resistant cells. N130wt N130mut

overexpressed Gadd45gip1 in paclitaxel-resistant DAPI NAC1 DAPI NAC1 SKOV3 cells using the retrovirus delivery system. Western blot analysis showed Gadd45gip1 protein expression in Gadd45gip1 virus-transduced cells, but not in pWZL (vector control) virus-transduced cells N130wt merged N130mut merged (Figure 4a). At 48 h after viral infection, cells were incubatedwith serial concentrations of paclitaxel and the IC50 was measuredfour dayslater. We foundthat ectopic Gadd45gip1 expression significantly increased the paclitaxel sensitivity in paclitaxel-resistant SKOV3 Figure 5 N130wt, but not N130mut, colocalizes with NAC-1. Immunofluorescence staining shows that N130wt (redfluorescence) cells. The IC50 s of paclitaxel were 2320 nM in Gadd45- colocalizes with NAC-1 (green fluorescence), whereas N130mut does not. gip1-overexpressedpaclitaxel-resistant cells and 35 512 nM in the vector control paclitaxel-resistant cells (Figure 4b). 2006a, b). We also generatedN130 mut by random mutagenesis, andthe sequence of N130 mut comparing with the N130wt is shown in Supplementary Figure 2A. The role of Gadd45g in the development of paclitaxel We establishedboth N130 wt- andN130 mut-inducible resistance SKOV3 cell lines using the Tet-off system. We found As Gadd45gip1 interacts with all three Gadd45 iso- that unlike N130wt protein, N130mut protein was not forms, including Gadd45a, Gadd45b and Gadd45g,we colocalized with endogenous NAC-1 (Figure 5) and did determined whether Gadd45 protein(s) was also regu- not exert a growth inhibitory effect (Supplementary latedby NAC-1. For these experiments we usedSKOV3 Figure 2B). Next, we assessedgene expression in three cells, which were engineeredwith an N130 Tet-off Gadd45 isoforms after N130wt or N130mut induction on inducible system. N130 is a truncated NAC-1 protein the basis of the assumption that NAC-1 signaling could (1–130 amino acids) containing only the BTB/POZ also affect the expression levels of Gadd45gip1 binding domain. It has been shown that N130 proteins partners for an efficient regulation of the function in the competitively interact with full-length NAC-1 mole- Gadd45/Gadd45gip1 complex. As shown in Figure 6a, cules, preventing them from forming NAC-1 homo- Gadd45g expression increased24 h after N130 induc- dimers, and induce cell death (Nakayama et al., tion. We also foundthat Gadd45 a mRNA levels

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1945 10.0 Gadd45γ -N130 wt the majority of them eventually develop recurrent γ 7.5 Gadd45 -N130 mt chemoresistant tumors, which account for the high mortality in ovarian cancer patients. In an earlier report, 5.0 we showedthat NAC-1 upregulation in ovarian serous 2.5 expression carcinomas was significantly associatedwith early tumor Relative mRAN Relative 0.0 recurrence after cytoreduction therapy and paclitaxel– 0 6 12 18 24 36 48 carboplatin combinedchemotherapy (Nakayama et al., Hours after N130 induction 2006a, b). In this study, we determined the biological 1.5 6 roles of NAC-1 in developing drug resistance in ovarian SKOV3 γ 5 SKOV3 cancer. Our results provide evidence that NAC-1 Hela Hela 1.0 4 upregulation contributes to tumor recurrence by equip- 3 ping ovarian cancer cells with the paclitaxel-resistant 0.5 2 phenotype through negative regulation of the Gadd45 expression expression 1 pathway. Relative NAC-1 Relative Relative Gadd45 Relative 0.0 0 The results of this study have implications regarding the pathogenesis of ovarian cancer. Our finding that NAC-1 participates in the development of paclitaxel shNAC-1 shNAC-1 shNAC-1 shNAC-1 resistance in ovarian cancer is consistent with a recent shRNAcontrol shRNAcontrol shRNAcontrol shRNAcontrol report (Ishibashi et al., 2008), andfurther supports the Figure 6 The effects of N130 induction on the expression of view that NAC-1 is a paclitaxel-resistance-associated Gadd45 family members. (a) Quantitative real-time reverse gene. It has been shown that one of the stemness transcription PCR shows that Gadd45g is upregulated24 h after properties of cancer-initiating (stem-like) cells is the N130wt induction in SKOV3 cells, whereas induction of N130mut does not alter Gadd45g expression. (b) NAC-1 shRNA treatment development of resistance to chemotherapeutic agents. (left panel) resulted in a significant upregulation of Gadd45g For example, ovarian cancer-initiating cells are highly (right panel) in both SKOV3 andHeLa cell lines that overexpress resistant to cisplatin or paclitaxel treatment (Zhang NAC-1. shRNA, small hairpin RNA. et al., 2008). Acquisition of the stem cell-like phenotype is often associatedwith upregulation of several ES cell increased in the first 24 h then decreased afterwards, markers, including Notch-1, Nanog, nestin and Oct-4, whereas Gadd45b expression did not show any sig- as comparedwith parental tumor cells. Recently, new nificant change after N130 induction (data not shown). evidence has shown a role of NAC-1 as an important In contrast, induction of N130mut did not have any effect transcriptional regulator in maintaining ES cell pluri- on the expression of all the Gadd45 isoforms. The potency. An extended regulatory network necessary for effect of N130mut on Gadd45g is shown in Figure 6a. preserving the pluripotent state of ES cells has been Next, we transfectedSKOV3 andHeLa cells with NAC- recently reported(Wang et al., 2006; Kim et al., 2008), 1 shRNA andobservedthat the copy number of andNAC-1 was shown to be a primary Nanog- Gadd45g transcript increasedafter NAC-1 knockdown interacting protein that is part of the protein regulatory (Figure 6b). In SKOV3 cells, a smaller reduction in complex responsible for maintaining pluripotency NAC-1 RNA causeda mild(but significant) increase in (Wang et al., 2006). NAC-1 protein complex has been GADD45-g expression, whereas in HeLa cells, a larger shown to transcriptionally regulate approximately 800 reduction in NAC-1 led to a pronounced increase in genes in ES cells (Kim et al., 2008). Interestingly, NAC-1 GADD45-g expression. Thus, the findings based on itself was determined to regulate transcription of the both NAC-1 knockdown and N130mut approaches transcription factors, Nanog, Oct4 andSox2, which indicated that NAC-1 expression and homodimerization seemedto be essential for development andmaintenance were essential to suppress Gadd45g expression. We also of the pluripotent state of ES cells (Boyer et al., 2005). determined whether Gadd45g was causally involvedin Given that NAC-1 likely participates in transcriptional developing paclitaxel resistance on the basis of IC50 regulation of these genes in the maintenance of cell curve analysis after gene knockdown. We found that pluripotency, it is plausible that NAC-1 contributes to both paclitaxel-resistant and naive cells in the Gadd45g drug resistance by enhancing stem-cell-like features of knockdown group became more resistant to paclitaxel- ovarian cancer cells. induced growth suppression than the control-treated Demonstration of NAC-1-mediated drug resistance group. The IC50s of paclitaxel at 96 h were B12 000 nM by the Gadd45 pathway is of great interest, because the in Gadd45g-knockdown paclitaxel-resistant cells and Gadd45 pathway plays a major role in arrest B6000 nM in the control paclitaxel-resistant cells andapoptosis under cellular stress, such as in the (Supplementary Figure 3). presence of chemotherapeutic agents (Zerbini and Libermann, 2005a, b). In this report andour earlier study, we have shown that NAC-1 proteins negatively regulate the expression of Gadd45gip1 (Crif1) (Nakayama Discussion et al., 2007) and Gadd45g. Gadd45gip1 (Crif1) has been shown to interact with all Gadd45 isoforms, and Most ovarian cancer patients are initially responsive to importantly, the interaction between Gadd45gip1 and carboplatin–paclitaxel combination chemotherapy, but Gadd45 members enhances the functions of the Gadd45

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1946 complex (Chung et al., 2003). Expression of Gadd45a et al., 2004; Materna et al., 2007; Bourguignon et al., and Gadd45g activates MEKK4/MTK1, which then 2008). activates MKK4. MKK4 activation, in turn, leads to In summary, the findings from this study provide a p38/JNK (janus ) activation by phosphorylation, possible molecular mechanism by which NAC-1 over- resulting in /growth inhibition (Verheij et al., expression contributes to paclitaxel resistance. Our 1996; De Smaele et al., 2001; Zerbini andLibermann, results show the essential role of NAC-1 homodimeriza- 2005a, b). It has also been shown that Gadd45a and tion in conferring paclitaxel resistance andalso highlight Gadd45g are down-regulated as a consequence of NF- the critical role of the Gadd45 pathway in mediating this kB pathway activation through upregulation of c-myc. phenotype. It wouldbe of great interest to determine Suppression of both Gadd45 members is essential for how NAC-1 proteins collaborate with other molecules cancer cells to survive (De Smaele et al., 2001; Tang andpathways in renderingstem cell anddrug-resistance et al., 2001; Zerbini et al., 2004). Thus, survival signals phenotypes in cancer cells. Furthermore, as NAC-1 may from at least two different oncogenic pathways, that is, serve as an upstream transcriptional regulator for NAC-1 andNF- kB, converge on one molecular ‘hub’, several signaling pathways, it wouldbe worthwhile to which is Gadd45/Gadd45gip1. Inhibition of NF-kB has explore the potential of developing NAC-1-targeted been shown to sensitize ovarian cancer cells to cisplatin therapy as a means to enhance paclitaxel sensitivity in andpaclitaxel-inducedapoptosis (Mabuchi et al., 2004; cancer patients. It might be imaginedthat NAC-1- Liu et al., 2006). Thus, turning off the Gadd45 pathway targetedtherapy couldbe usedeither as a stand-alone seems to be critical for cancer cell survival andto sensitizing agent for chemotherapeutics, or usedin facilitate tumor growth under cellular stress. In addition combination with other inhibitors of known drug- to the establishedNF- kB-Gadd45 molecular circuit, resistance pathways. downregulation of Gadd45gip1 by NAC-1 overexpres- sion represents another molecular switch to suppress Gadd45 death signals. It is likely that in the evolution of Materials and methods cancer cell species, tumor cells acquire multiple strate- gies to inactivate Gadd45 cell suppressor signaling. In Cell lines and culture conditions fact, overexpression of NF-kB (Mabuchi et al., 2004; Cancer cell lines, including 293 T, OVCAR3, SKOV3, A2780 Liu et al., 2006), gene amplification andtranscriptional andHeLa cells, were purchasedfrom ATCC (Rockville, MD, upregulation of c-myc (Baker et al., 1990; Abeysinghe USA). All cell lines usedin this studywere culturedin et al., 1999; Dimova et al., 2006) and Gadd45 RPMI1640 medium supplemented with 5% fetal bovine (Qiu et al., 2004; Ying et al., 2005; serum. To generate chemoresistant ovarian cancer cells, Zerbini andLibermann, 2005a, b) have been reported SKOV3, A2780 andOVCAR3 cells were selectedby contin- uous treatment with different concentrations of paclitaxel in human , including ovarian cancer cells. (ranging from 0.2 mM–2.0 mM) or carboplatin (0.2 mg/ml–4.0 mg/ Furthermore, our recent study also showed that homo- ml). Resistant colonies of each cell line were pooledand zygous deletion and downregulation of MKK-4 were subsequently usedfor experiments. Chemoresistant cell lines detected in ovarian carcinomas (Nakayama et al., were maintainedin selective mediumcontaining the highest 2006a, b). The above data together with the results concentrations of paclitaxel (0.25 mM for SKOV3, 0.5 mM for reportedin this studysuggest that cancer cells likely A2780 andOVCAR3) or carboplatin (2.0 mg/ml for SKOV3, employ at least a dual molecular system to switch off 1.0 mg/ml for A2780 and0.5 mg/ml for OVCAR3) that Gadd45-mediated growth suppression and apoptosis, permittedgrowth. that is, the NF-kB/c-myc/Gadd45 pathway and the NAC-1/Gadd45gip1 pathway. Retrovirus, shRNA transfection and generation of N130mut The biological roles of NAC-1 in promoting paclitaxel A retrovirus carrying the Gadd45gip1-V5 expression vector resistance imply that NAC-1 might be usedas a was produced by cloning the entire coding sequence into the surrogate stand-alone marker or in combination with pWZL-Hygro retroviral vector. High-titer retroviral stocks other markers to predict paclitaxel resistance. Such were generatedusing the Phoenix ecotropic packaging cell line. shRNA plasmids that target NAC-1 (pRS-shNAC1) and information may aidin individualizedselection of Gadd45gip1 (pRS-shGadd45gip1), as well as the shRNA effective therapies in ovarian cancer patients. For plasmidthat targets GFP (green fluorescent protein) (pRS- example, patients with NAC-1-overexpressing tumors shGFP) were obtainedfrom Origene (Rockville, MD, USA). might be treatedwith chemotherapeutic agents other The shRNA plasmids targeting Gadd45g andthe control than paclitaxel in future clinical trials. Furthermore, plasmid(pLKO.1-puro vector-only) were obtainedfrom NAC-1 may also serve as a molecular target to enhance Sigma (St Louis, MO, USA). The shRNA sequences that paclitaxel sensitivity in ovarian cancer patients. The target different genes are listed in Supplementary Table 1. results from this study suggest that several strategies can We have earlier generatedN130-induciblecells that express be appliedfor NAC-1-targetedtherapy, including the first 130 amino-acidsegment of NAC-1 (N130), which disruption of NAC-1 homodimerization, silencing corresponds to the BTB/POZ domain. A mutant of N130, N130mut, was generated in this study by random mutagenesis NAC-1 or forcedexpression of Gadd45gip1.Although using the JBS dNTP-mutagenesis kit (Jena Bioscience, Jena, the above represent our favoredview, it shouldbe noted Germany). To completely abolish the interaction between that multiple genes andpathways are likely involvedin N130 andNAC-1, we introducedmissensemutations that the development of paclitaxel resistance in cancer cells changed30 amino acidsin the N130 fragment. The mutant (Duan et al., 2003, 2006a, b; Yusuf et al., 2003; Mabuchi N130 sequence was then clonedinto pcDNA4-Xp and

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1947 subclonedinto the pTRE-induciblevector. The cloning has been validated in earlier reports (Nakayama et al., 2006b). primers for N130 were: 50-tgaccagaattcgccatggcccagacactgcag-30 An EnVision þ System peroxidase kit (DAKO, Carpinteria, (forward) and 50-tggtcactcgagagtcgcagctcggggagctca-30 (reverse). CA, USA) was usedfor staining, following the manufacturer’s To determine whether N130mut interactedwith NAC-1, we protocol. Immunohistochemical staining was carriedout on cotransfected293T cells with NAC-1 taggedwith V5 and tissue microarrays containing 113 anonymous high-grade N130mut taggedwith Xpress. Double immunofluorescence was ovarian serous carcinoma tissues for which ex vivo chemo- performed as described earlier using V5 and Xpress antibodies response status was available. The definition of extreme drug- (Nakayama et al., 2006a, b). resistance status was that tumor cell growth was greater than one standard deviation above the median, whereas the low Drug sensitivity and cell viability assays drug resistance was that cell growth was less than the median For drug sensitivity assays, cells in both experimental and growth (Holloway et al., 2002; Loizzi et al., 2003). Immuno- control groups were seeded in 384-well plates at a density of intensity for NAC-1 was scoredusing a four-tier system from 0 750 cells/well. After overnight culture, the cells were treated (undetectable), 1 þ (weakly positive), 2 þ (moderately posi- with a series of concentrations of paclitaxel. Four days after tive) to 3 þ (intensely positive). This criteria was usedin our transfection (that is, 3 days after drug treatment), viable cell earlier study (Nakayama et al., 2006b). numbers were measuredon the basis of the ability of living Western blot analysis was conducted using a standard cells to convert a redox dye (resazurin) into a fluorescent end protocol. The antibodies used in this study included a mouse product (resorufin) using CellTiter-Blue reagent (Promega, NAC-1 monoclonal antibody at a 1:200 dilution, an anti-V5 Madison, WI, USA). Data read by a fluorescence microplate antibody (Invitrogen, Cat no. R960-25) at a 1:1000 dilution reader (Fluostar from BMG, Durham, NC, USA) were and a rabbit anti-GAPDH (glyceraldehyde-3-phosphate dehy- determined from three replicates, and were expressed as a drogenase) polyclonal antibody (Sigma, Cat no. G9545) at a percent of the group without paclitaxel treatment. IC50 was 1:4000 dilution. Similar amounts of total protein from each defined as the concentration that resulted in a 50% decrease in lysate were loaded and separated on 4–12% Tris–Glycine–SDS the number of cells, as reflectedby fluorescence intensity, polyacrylamide gels (Novex, San Diego, CA, USA) and comparedwith that of the control cultures in the absence of electroblottedto Millipore Immobilon-P polyvinylidene the drug. difluoride membranes (Millipore, Billerica, MA, USA). Western blots were developed by chemiluminescence (Pierce, Rockford, IL, USA). Quantitative real-time reverse transcription PCR An RNA extraction kit (Qiagen, Germantown, MD, USA) was usedto extract total RNA, andthe Superscript II First-strand cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA) was used Conflict of interest to generate cDNA. Real-time PCR was performedon a Bio-Rad iCyclers (MyIQ, IQ4), anddataanalysis was performedusing The authors declare no conflict of interest. the Bio-RadIQ5 v2 software (Hercules, CA, USA). The PCR primers usedin this studyare listedin Supplementary Table 1. The mean Ct of the gene of interest was calculatedfrom replicate Acknowledgements measurements andnormalizedwith the mean Ct of a control gene, APP (b-amyloidprecursor gene), for which the expression This work is in memory of Ms Sean Patrick, the founder of is relatively constant among specimens. HERA women’s cancer foundation who has fought coura- geously with recurrent ovarian cancer. The authors appreciate Immunohistochemistry and western blot the valuable comments from members in our laboratory. This A mouse monoclonal anti-NAC-1 antibody was obtained from work is supportedby NIH RO1CA103937 (IMS), NIH Novus (Littleton, CO, USA) andwas usedin immunohisto- RO1CA129080 (IMS) andthe HERA women’s cancer chemistry at a dilution of 1:200. The specificity of this antibody foundation OSB1 (NJ, USA).

References

Abeysinghe H.R, Cedrone E, Tyan T, Xu J, Wang N. (1999). Cha XY, Pierce RC, Kalivas PW, Mackler SA. (1997). NAC-1, a rat Amplification of C-MYC as the origin of the homogeneous staining mRNA, is increasedin the nucleus accumbens three region in ovarian carcinoma detected by micro-FISH. Cancer Genet weeks after chronic cocaine self-administration. J Neurosci 17: Cytogenet 114: 136–143. 6864–6871. Abhiman S, Iyer LM, AravindL. (2008). BEN: a novel domain in Chung HK, Yi YW, Jung NC, Kim D, Suh JM, Kim H et al. (2003). chromatin factors andDNA viral proteins. Bioinformatics 24: 458–461. CR6-interacting factor 1 interacts with Gadd45 family proteins and Baker VV, Borst MP, Dixon D, Hatch KD, Shingleton HM, Miller D. modulates the cell cycle. J Biol Chem 278: 28079–28088. (1990). c-myc amplification in ovarian cancer. Gynecol Oncol 38: Davidson B, Berner A, Trope CG, Wang TL, Shih Ie M. (2007). 340–342. Expression andclinical role of the bric-a-brac tramtrack broad Bourguignon LY, Peyrollier K, Xia W, GiladE.. (2008). Hyaluronan- complex/poxvirus andzinc protein NAC-1 in ovarian carcinoma CD44 interaction activates stem cell marker Nanog, effusions. Hum Pathol 38: 1030–1036. Stat-3-mediated MDR1 , and ankyrin-regulated De Smaele E, Zazzeroni F, Papa S, Nguyen DU, Jin R., Jones J et al. multidrug efflux in breast and ovarian tumor cells. J Biol Chem 283: (2001). Induction of gadd45beta by NF-kappaB downregulates pro- 17635–17651. apoptotic JNK signalling. Nature 414: 308–313. Boyer LA, Lee TI, Cole MF, Johnstone SE, Levine SS, Zucker JP et al. Dimova I, Raitcheva S, Dimitrov R, Doganov N, Toncheva D. (2006). (2005). Core transcriptional regulatory circuitry in human embryo- Correlations between c-myc gene copy-number andclinicopatho- nic stem cells. Cell 122: 947–956. logical parameters of ovarian tumours. Eur J Cancer 42: 674–679.

Oncogene NAC-1 in paclitaxel resistance N Jinawath et al 1948 Duan Z, Duan Y, Lamendola DE, Yusuf RZ, Naeem R, Penson RT expression is predictive for clinical outcome in ovarian carcinoma et al. (2003). Overexpression of MAGE/GAGE genes in paclitaxel/ patients. Virchows Arch 450: 187–194. doxorubicin-resistant human cancer cell lines. Clin Cancer Res 9: Muller FJ, Laurent LC, Kostka D, Ulitsky I, Williams R, Lu C et al. 2778–2785. (2008). Regulatory networks define phenotypic classes of human Duan Z, Foster R, Bell DA, Mahoney J, Wolak K, Vaidya A et al. stem cell lines. Nature 455: 401–405. (2006a). Signal transducers and activators of transcription 3 Nakayama K, Nakayama N, Davidson B, Katabuchi H, Kurman RJ, pathway activation in drug-resistant ovarian cancer. Clin Cancer Velculescu VE et al. (2006a). Homozygous deletion of MKK4 in Res 12: 5055–5063. ovarian serous carcinoma. Cancer Biol Ther 5: 630–634. Duan Z, Foster R, Brakora KA, Yusuf RZ, Seiden MV. (2006b). Nakayama K, Nakayama N, Davidson B, Sheu JJ, Jinawath N, GBP1 overexpression is associatedwith a paclitaxel resistance Santillan A et al. (2006b). A BTB/POZ protein, NAC-1, is relatedto phenotype. Cancer Chemother Pharmacol 57: 25–33. tumor recurrence andis essential for tumor growth andsurvival. Holloway RW, Mehta RS, Finkler NJ, Li KT, McLaren CE, Parker Proc Natl Acad Sci USA 103: 18739–18744. RJ et al. (2002). Association between in vitro platinum resistance in Nakayama K, Nakayama N, Wang T-L, Shih I-M. (2007). NAC-1 the EDR assay andclinical outcomes for ovarian cancer patients. controls cell growth andsurvival by repressing transcription Gynecol Oncol 87: 8–16. of Gadd45GIP1, a candidate tumor suppressor. Cancer Res 67: Ishibashi M, Nakayama K, Yeasmin S, Katagiri A, Iida K, Nakayama 8058–8064. N et al. (2008). A BTB/POZ gene, NAC-1, a tumor recurrence- Qiu W, Zhou B, Zou H, Liu X, Chu PG, Lopez R et al. (2004). associatedgene, as a potential target for Taxol resistance in ovarian Hypermethylation of growth arrest DNA damage-inducible gene 45 cancer. Clin Cancer Res 14: 3149–3155. beta promoter in human . Am J Pathol Jemal A, Siegel R, WardE, Murray T, Xu J, Thun MJ. (2007). Cancer 165: 1689–1699. statistics, 2007. CA Cancer J Clin 57: 43–66. Shih Ie M, Sheu JJ, Santillan A, Nakayama K, Yen MJ, Bristow RE Kalivas PW, Duffy P, Mackler SA. (1999). Interruptedexpression of et al. (2005). Amplification of a chromatin remodeling gene, Rsf-1/ NAC-1 augments the behavioral responses to cocaine. Synapse 33: HBXAP, in ovarian carcinoma. Proc Natl Acad Sci USA 102: 153–159. 14004–14009. Kim J, Chu J, Shen X, Wang J, Orkin SH. (2008). An extended Tang G, Minemoto Y, Dibling B, Purcell NH, Li Z, Karin M et al. transcriptional network for pluripotency of embryonic stem cells. (2001). Inhibition of JNK activation through NF-kappaB target Cell 132: 1049–1061. genes. Nature 414: 313–317. Koob G.F. (1996). Drug addiction: the yin and yang of hedonic Verheij M, Bose R, Lin XH, Yao B, Jarvis WD, Grant S et al. (1996). homeostasis. 16: 893–896. Requirement for ceramide-initiated SAPK/JNK signalling in stress- Korutla L, Degnan R, Wang P, Mackler SA. (2007). NAC1, a cocaine- induced apoptosis. Nature 380: 75–79. regulatedPOZ/BTB protein interacts with CoREST. J Neurochem Wang J, Rao S, Chu J, Shen X, Levasseur DN, Theunissen TW et al. 101: 611–618. (2006). A protein interaction network for pluripotency of embryonic Korutla L, Wang PJ, Mackler SA. (2005). The POZ/BTB protein stem cells. Nature 444: 364–368. NAC1 interacts with two different histone deacetylases in neuronal- Yeasmin S, Nakayama K, Ishibashi M, Katagiri A, Iida K, Purwana like cultures. J Neurochem 94: 786–793. IN et al. (2008). Expression of the bric-a-brac tramtrack broad Liu GH, Wang SR, Wang B, Kong BH. (2006). Inhibition of nuclear complex protein NAC-1 in cervical carcinomas seems to correlate factor-kappaB by an antioxidant enhances paclitaxel sensitivity in with poorer prognosis. Clin Cancer Res 14: 1686–1691. ovarian carcinoma cell line. Int J Gynecol Cancer 16: 1777–1782. Ying J, Srivastava G, Hsieh WS, Gao Z, Murray P, Liao SK et al. Loizzi V, Chan JK, Osann K, Cappuccini F, DiSaia PJ, Berman ML. (2005). The stress-responsive gene GADD45G is a functional tumor (2003). Survival outcomes in patients with recurrent ovarian cancer suppressor, with its response to environmental stresses frequently who were treatedwith chemoresistance assay-guidedchemotherapy. disrupted epigenetically in multiple tumors. Clin Cancer Res 11: Am J Obstet Gynecol 189: 1301–1307. 6442–6449. Mabuchi S, Ohmichi M, Nishio Y, Hayasaka T, Kimura A, Ohta T Yusuf RZ, Duan Z, Lamendola DE, Penson RT, Seiden MV. (2003). et al. (2004). Inhibition of NFkappaB increases the efficacy of Paclitaxel resistance: molecular mechanisms andpharmacologic cisplatin in in vitro and in vivo ovarian cancer models. J Biol Chem manipulation. Curr Cancer Drug Targets 3: 1–19. 279: 23477–23485. Zerbini LF, Libermann TA. (2005a). GADD45 deregulation in cancer: Mackler S, Pacchioni A, Degnan R, Homan Y, Conti AC., Kalivas P frequently methylatedtumor suppressors andpotential therapeutic et al. (2008). Requirement for the POZ/BTB protein NAC1 in acute targets. Clin Cancer Res 11: 6409–6413. but not chronic psychomotor stimulant response. Behav Brain Res Zerbini LF, Libermann TA. (2005b). Life anddeathin cancer. 187: 48–55. GADD45 alpha andgamma are critical regulators of NF-kappaB Mackler SA, Korutla L, Cha XY, Koebbe MJ, Fournier KM, Bowers mediated escape from programmed cell death. Cell Cycle 4: 18–20. MS et al. (2000). NAC-1 is a brain POZ/BTB protein that can Zerbini LF, Wang Y, Czibere A, Correa RG, Cho JY, Ijiri K et al. prevent cocaine-induced sensitization in the rat. J Neurosci 20: (2004). NF-kappa B-mediated repression of growth arrest- and 6210–6217. DNA-damage-inducible proteins 45alpha and gamma is essential Mao TL, Hsu CY, Yen MJ, Gilks B, Sheu JJ, Gabrielson E et al. for cancer cell survival. Proc Natl Acad Sci USA 101: 13618–13623. (2006). Expression of Rsf-1, a chromatin-remodeling gene, in Zhang S, Balch C, Chan MW, Lai HC, Matei D, Schilder JM ovarian andbreast carcinoma. Hum Pathol 37: 1169–1175. et al. (2008). Identification and characterization of ovarian Materna V, Surowiak P, Kaplenko I, Spaczynski M, Duan Z, Zabel M cancer-initiating cells from primary human tumors. Cancer Res et al. (2007). Taxol-resistance-associatedgene-3 (TRAG-3/CSAG2) 68: 4311–4320.

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

Oncogene