The Aurora Kinase a Regulates GSK-3B in Gastric Cancer Cells

AA Dar1, A Belkhiri1 and W El-Rifai1,2,3

1Department of Surgery, Vanderbilt University Medical Center, Nashville, TN, USA; 2Department of Cancer Biology, Vanderbilt University Medical Center, Nashville, TN, USA and 3Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN, USA

Aurora kinase A (AURKA) is locatedat 20q13, a region that Introduction is frequently amplifiedin gastric cancer. In this study, we have investigatedthe role of AURKA in regulating glycogen Gastric cancers are characterized by poor response to synthase kinase (GSK)-3b and b-catenin/TCF complex in therapy with unfavorable clinical outcome that reflects gastric cancer cells. Our results demonstrate a significant an inherent protective mechanism in these tumors increase in the phosphorylation of GSK-3b at Ser 9 following against drug-induced apoptosis (Earle and Maroun, the overexpression of AURKA in AGS cells. The immuno- 1999; Jemal et al., 2006). Aurora kinase A (AURKA) is precipitation with antibodies specific for AURKA and GSK-3b located in chromosome arm 20q13, a locus that is indicated that the two proteins coexist in the same protein frequently amplified and overexpressed in gastric complex. The recombinant human AURKA protein phos- adenocarcinomas (El-Rifai et al., 2001; Varis et al., phorylatedthe GSK-3 b protein at Ser 9 in a concentration- 2001). We have recently shown frequent overexpression dependent manner, in vitro. The phosphorylation of b-catenin of AURKA in adenocarcinomas of the stomach and (Ser33/37/Thr41) by GSK-3b is known to target b-catenin esophagus (Dar et al., 2008). The AURKA gene encodes towards degradation. In line with our findings, the increase in a centrosome associated cell-cycle-regulated serine/ phospho-GSK-3b level was accompaniedby a significant threonine kinase (Giet and Prigent, 1999). Cytological decrease in b-catenin phosphorylation (Ser33/37/Thr41) and analysis revealed that overexpression of AURKA results accumulation of b-catenin protein. The knockdown in centrosome amplification, cytokinesis failure; thus, of AURKA reversedthe phosphorylation of GSK-3 b andthe production of aneuploid cells (Zhou et al., 1998). In b-catenin protein levels. The immunofluorescence analysis cancer cells, AURKA regulates several important demonstrated colocalization of AURKA and GSK-3b proteins proteins such as p53 (Katayama et al., 2004; Dar anda significant increase in the nuclear b-catenin levels in cells et al., 2008), AKT (Yang et al., 2006; Dar et al., 2008) overexpressing AURKA. The b-catenin/TCF transcription and BRCA1 (Ouchi et al., 2004). The complete spectrum activity was measuredusing the pTopFlash andits mutant of molecular signaling mechanisms and biological pFopFlash luciferase reporter vectors. Indeed, AURKA over- functions that are regulated by AURKA are currently expression ledto a significant increase in the pTopFlash not fully understood. reporter activity, whereas kinase dead AURKA mutant b-Catenin is a multifunctional protein that is both an (D274A) hadno effect. Consistent with these findings,we integral component of adheren junctions and a pivotal detected a significant mRNA up-regulation of several direct member of the Wnt signal transduction pathway. targets of the b-catenin/TCF transcription complex (cyclin Mutations and overexpression of b-catenin disrupt the D1, c-MYC, c-MYC-binding protein, CLDN1, FGF18 and function of E-cadherins. The increased level of b-catenin vascular endothelial growth factor), anda two-foldincrease in through mutations in either b-catenin or adenomatous the proliferation rate in AURKA overexpressing cells. polyposis coli (APC) has been suggested tobe an We conclude that the AURKA/GSK-3b interaction is important oncogenic step in the genesis of a number of important in regulating b-catenin, underscoring a novel malignancies and it is an early event in colorectal oncogenic potential for AURKA in gastric tumorigenesis. carcinogenesis (Morin et al., 1997; Damalas et al., Oncogene (2009) 28, 866–875; doi:10.1038/onc.2008.434; 1999b; Polakis, 1999). Cytoplasmic accumulation of published online 8 December 2008 b-catenin results in its translocation to the nucleus where it associates with the members of TCF/LEF family of Keywords: AURKA; GSK-3b; b-catenin; gastric cancer transcription factors. A large number of genes relevant for tumor formation and progression have been identified tobe transcriptionally activated by the nuclear b-catenin/TCF complex (He et al., 1998; Shtutman Correspondence: Dr W El-Rifai, Vanderbilt-Ingram Cancer Center, et al., 1999; Tetsu and McCormick, 1999). Some of them Vanderbilt University Medical Center, 1255 Light Hall, 2215 Garland are implicated in growth control and cell cycling (cyclin Avenue, Nashville, TN 37232, USA. E-mail: [email protected] D1, c-MYC, c-Jun and fra1), others are relevant for cell Received 21 February 2008; revised 10 October 2008; accepted survival (Id2, MDR1) and some are involved in 30 October 2008; published online 8 December 2008 tumor invasion and metastasis (Matrilysin, vascular Overexpression of AURKA regulates GSK-3b AA Dar et al 867 endothelial growth factor (VEGF); Crawford et al., phosphorylate b-catenin, thus leading toa morestable 1999; Tetsu and McCormick, 1999; Koh et al., 2000; b-catenin with its accumulation and subsequent nuclear Yamada et al., 2000). Although APC mutations are localization that was detected as shown in Figure 4a. infrequent in gastric carcinomas (Nakatsuru et al., Consistent with our findings of increased phospho- 1992), accumulation and nuclear localization of GSK-3b in AURKA overexpressing cells, we detected b-catenin occurs in approximately one-third of gastric a decrease in the phosphorylation level of b-catenin tumors (Clements et al., 2002). (Ser33/37/Thr41) together with an increase in its The glycogen synthase kinase (GSK)-3b is a protein level (Figure 1b). The use of AURKA small- cytosolic protein that plays a critical role in determining interfering RNA (siRNA) reversed these effects and led the levels of b-catenin in the cells. The phosphorylation to a significant reduction in the level of phospho- of b-catenin by the kinase active GSK-3b leads to GSK-3b and a subsequent decrease in the b-catenin ubiquitination and proteosomal degradation of protein level (Figure 1c). These results suggested that b-catenin. In this way, b-catenin is restricted from the AURKA can regulate the b-catenin protein levels in nucleus and prevented from interacting and activating gastric cancer cells by regulating the phosphorylation of the TCF/LEF transcription complex (Aberle et al., GSK-3b. 1997; Orford et al., 1997). However, GSK-3b is rendered inactive after being phosphorylated at Ser 9, and thus, loses its ability to phosphorylate b-catenin leading to AURKA colocalizes and interacts with GSK-3b in AGS accumulation and nuclear localization of b-catenin cells (Giles et al., 2003; Kikuchi et al., 2006). In this study, Immunofluorescence studies using AURKA and GSK- we investigated the role of AURKA in regulating GSK- 3b-specific antibodies showed that both proteins colo- 3b and the effects of this regulation on b-catenin in calize in the cytoplasm in AGS gastric cancer cells gastric cancer cells. (Figure 2a). We and others have shown cytosolic overexpression of AURKA in several malignancies such as gastric, esophageal and lung cancer (Tong et al., 2004; Dar et al., 2008; Ogawa et al., 2008). Toconfirmthe Results protein–protein interaction, two-way co-immunopreci- pitations were performed for exogenously expressed AURKA overexpression regulates GSK-3b AURKA and GSK-3b (Figure 2b) and for endo- phosphorylation and b-catenin protein levels genously expressed AURKA and GSK-3b (Figure 2c). Overexpression of AURKA in AGS and SNU1 gastric The results revealed coexistence of AURKA and GSK- cancer cells led to an up-regulation in the phosphoryla- 3b in the same protein complex. Furthermore, the in tion of GSK-3b (Ser 9; Figure 1a). The GSK-3b is an vitro kinase assay demonstrated phosphorylation of important protein kinase that regulates the phospho- GSK-3b by AURKA in a concentration-dependent rylation of b-catenin leading toits degradation. manner (Figure 2d). The trefoil factor 1 (TFF1) protein The phospho-GSK-3b is kinase inactive and cannot was used as a negative control. These results confirm

Figure 1 Aurora kinase A (AURKA) regulates the phosphorylation of glycogen synthase kinase (GSK)-3b.(a) AGS and SNU1 cells were transfected with Flag-tagged AURKA and control pcDNA vectors using Lipofectamine 2000 (Invitrogen) following manufacturer’s instructions. The overexpression of AURKA in AGS and SNU1 cells up-regulated phosphorylation levels of GSK- 3b (Ser 9). (b) AURKA overexpression in AGS and SNU1 cells resulted in a decrease in the phospho-b-catenin (Ser33/37Thr41) with up-regulation of its protein level. (c) Knockdown of endogenous AURKA by a speciﬁc siRNA oligonucleotide reduced the protein levels of phospho-GSK-3b (Ser 9) and b-catenin, thus validating the results of AURKA overexpression (a and b).

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 868

Figure 2 Aurora kinase A (AURKA) colocalizes and interacts with glycogen synthase kinase (GSK)-3b.(a) Immunoﬂuorescence assay demonstrates colocalization of AURKA (green) and GSK-3b (red) in AGS cells. (b) Co-immunoprecipitation (CO-IP) of GSK- 3b and AURKA. AGS cells were cotransfected with Flag-AURKA and HA-GSK3b plasmids for 48 h. For AURKA and GSK-3b lysates were subjected to IP with Flag and HA antibodies followed by immunoblotting with HA and Flag antibodies (Cell signaling) to detect AURKA and GSK-3b, respectively. The results indicate their presence in the same protein complex. (c) The CO-IP studies were performed for endogenous AURKA and GSK-3b and demonstrated similar ﬁndings. (d) In vitro kinase assay was performed using recombinant AURKA and GSK-3b proteins followed by immunoblotting and detection with phospho-GSK-3b Ser9 antibody. GSK- 3b concentration was kept constant with an increasing concentration of AURKA as indicated antibody. A concentration dependent increase in the phosphorylation GSK-3b was observed. The trefoil factor 1 (TFF1) protein was used as a negative control.

that AURKA can directly bind to and phosphorylate cells, we detected a decrease in the phosphorylation GSK-3b. The inactivation of GSK-3b by phosphoryla- level of b-catenin (Ser33/37/Thr41) together with an tion can be mediated by the PI3-kinase/AKT survival increase in its protein level (Figure 1b). Accumulation of signaling pathway (Pap and Cooper, 1998; Mitsiades b-catenin is a major step that promotes its nuclear et al., 2004; Sourbier et al., 2006). We have used AKT- localization and activates its oncogenic properties. specific inhibitor in the presence and absence of The immunofluorescence assay on AGS cells stably AURKA and demonstrated its efficacy in abrogating overexpressing AURKA demonstrated a remarkable the phospho-AKT signals (Figure 3b). At the same time, increase in the nuclear b-catenin, as compared to control the AGS cells overexpressing AURKA had approxi- cells (Figure 4a). Several reports have indicated that the mately five-fold higher signal for phospho-GSK-3b (S9) oncogenic functions of b-catenin are mediated by its than the control cells in the presence of AKT inhibitor interaction with members of the TCF/LEF family of (Figure 3b). These findings together with the transcription factors in the nucleus (Behrens et al., 1996; co-immunoprecipitation results (Figures 2b and c) and Miller et al., 1999). We utilized the pTopFlash/pFop- the in vitro recombinant kinase assay (Figure 2d) Flash luciferase vectors as a measure of b-catenin/TCF support the conclusion that AURKA directly can complex transcriptional activity. AURKA overexpres- regulate the GSK-3b phosphorylation. However, the sion led to a five-fold induction of the luciferase activity extent of GSK-3b phosphorylation in the presence of of pTopFlash luciferase vector but, as expected, had no AKT inhibitor was less than in its absence. Therefore, it effect on the pFopFlash luciferase vector (Figure 4b). is possible that AURKA regulates GSK-3b phospho- Conversely, expressing kinase dead AURKA mutant rylation through direct interaction as well as through an (D274A) did not induce luciferase activity when AURKA/AKT axis-dependent mechanism. transfected along with pTopFlash or pFopFlash vectors; thus, indicating that the induced luciferase activity in cells transfected with pTopFlash vector is because of The interaction between AURKA and GSK-3b leads to overexpression of kinase active AURKA. As expected, accumulation and nuclear translocation of b-catenin with the b-catenin transfection (positive control) led to activation of the b-catenin/TCF complex induction of pTopFlash, but not pFopFlash luciferase The aforementioned results indicated that AURKA activity. Taken together, our results as shown in Figures binds toand regulates GSK-3 b phosphorylation in 1–3, indicate that AURKA interaction with GSK-3b gastric cancer cells. Consistent with our findings of regulates the b-catenin/TCF transcription activity. We increased phospho-GSK-3b in AURKA overexpressing have confirmed this finding by measuring the transcript

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 869

Figure 3 Aurora kinase A (AURKA), AKT and glycogen synthase kinase (GSK)-3b expression in gastric cancer cells. (a) Western blot analysis of AURKA, AKT, p-AKT, GSK-3b, p-GSK-3b, and b-catenin in AGS, MKN45, MKN75, and SNU16 gastric cancer cell lines. The levels of p-GSK-3b (S9) and b-catenin correlated with the expression level of AURKA. The MKN45 cells demonstrated high levels of AURKA and p-GSK-3b and b-catenin, despite a relatively low level of p-AKT (S473). (b) Western blot analysis of AGS cells following treatment with dimethyl sulfoxide (DMSO) as a control (left panel) or AKT inhibitor VIII (right panel). AGS cells overexpressing AURKA showed an increase in the phosphorylation of AKT (S473) and GSK-3b (S9) as compared to pcDNA3 empty vector control cells. The treatment of AGS cells with AKT inhibitor VIII (1 mM) for 24 h led to a signiﬁcant down-regulation of phospho-AKT (S473) in AURKA-expressing AGS cells (right panel). However, GSK-3b phosphorylation (S9) was ﬁve-fold higher in AURKA overexpressing cells than in pcDNA3 empty vector control cells, in presence of AKT inhibitor VIII.

level of several direct targets of the b-catenin/TCF an inherent protective mechanism in these tumors complex. The quantitative real-time RT-PCR (qRT- against drug-induced apoptosis (Earle and Maroun, PCR) indicated up-regulation of cyclin D1, c-MYC, c- 1999; Jemal et al., 2006). The critical molecular MYC-binding protein, CLDN1, FGF-18 and VEGF in alterations and signaling pathways that drive gastric AURKA overexpressing cells as compared to control tumorigenesis remain largely uncharacterized. In this (Figure 5a). Using cyclin D1 as an example of these study, we have investigated the role of AURKA in this targets, we further confirmed the induction of its devastating disease. AURKA is located at chromosome promoter in AURKA overexpressing cells (Figure 5b) 20q and is frequently amplified and overexpressed in and demonstrated the increase in its protein level in gastric cancer (El-Rifai et al., 2001; Varis et al., 2001; AGS and SNU1 cells (Figure 5c). Cyclin D1 is Kamada et al., 2004; Dar et al., 2008). We and others important in cell cycle and proliferation. Indeed, AGS have shown AURKA overexpression in a wide range of cells stably expressing AURKA showed >2.5-fold primary tumors including adenocarcinomas of the increase in their proliferation rate as compared to breast, colorectal, pancreatic, ovarian, esophageal, control vector cells, as determined by 5-Bromo- gastric and bladder cancers (Bischoff et al., 1998; Zhou 20-deoxyuridine (BrdU) incorporation assay (Figure 6a). et al., 1998; Tanaka et al., 1999; Kamada et al., 2004; The cell-cycle analysis demonstrated a doubling in the S- Giet et al., 2005; Hu et al., 2005). phase in AGS (from 10 to 18%) and RIE (from 6.5 to In this study, we have demonstrated that AURKA 16%) cells overexpressing AURKA as compared to interacts with GSK-3b and directly binds toand control vector-expressing cells (Figures 6b and c). Taken phosphorylates GSK-3b at Ser 9. The GSK-3b has been together, these results indicate that the AURKA/GSK-3b previously identified as a key downstream target of the interaction results in a functional signaling outcome that is PI3-kinase/AKT survival signaling pathway (Pap and manifested by an increase in the b-catenin protein, Cooper, 1998; Mitsiades et al., 2004; Sourbier et al., 2006). activation of the b-catenin/TCF transcription complex We have used AKT-specific inhibitor in the presence and and increase in the cell proliferation rate. Other oncogenic absence of AURKA and our results demonstrated a high consequences of the AURKA/GSK-3b interaction and level of phospho-GSK-3b (S9) in the presence of AKT nuclear accumulation and activation of b-catenin remain inhibitor in AURKA overexpressing cells, supporting our tobe elucidated in further studies. notion that AURKA can directly regulate GSK-3b.Itis worth mentioning that we and others have shown that AURKA can phosphorylate and activate AKT (Yang Discussion et al., 2006; Dar et al., 2008). Therefore, it is possible that AURKA regulates GSK-3b phosphorylation through Gastric cancers are characterized by poor response to direct interaction and/or through AURKA/AKT axis- therapy with unfavorable clinical outcome that reflects dependent mechanism.

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 870

Figure 4 Nuclear accumulation of b-catenin and activation of the b-catenin/TCF complex. (a) Immunoﬂuorescence (IF) assay for endogenous b-catenin indicates its nuclear accumulation in AGS cells stably overexpressing Aurora kinase A (AURKA), as compared to control vector-expressing cells. In total, 200 cells were counted from AURKA and control vector for b-catenin IF. AURKA (70%) overexpressing cells showed nuclear b-catenin staining (bar graph). Western blot analysis demonstrates overexpression of AURKA in AGS cells. (b) The luciferase reporter assay using pTopFlash and its mutant pFopFlash vectors was utilized to study the b-catenin TCF/LEF transcriptional activity. The pTopFlash reporter vector has six TCF-binding sites, although pFopFlash has mutated TCF- binding sites. AGS cells were cotransfected with different expression vectors as indicated. Overexpressing AURKA along with pTopFlash vector resulted in the induction of luciferase activity, indicating the transcription activity of the b-catenin/TCF complex. The kinase dead AURKA mutant (D274A) had no effect on pTopFlash activity. The wild-type b-catenin-expressing vector (pSWTBCAT) was used as a positive control. The overexpression of AURKA or its kinase dead mutant (D274A) with pFopFlash vectors had no effect on the luciferase activity. Luciferase activity was measured using a Dual luciferase reporter assay kit (Promega) following manufacturer’s instructions. Results were averaged from three independent experiments and expressed as mean values (±standard deviation (s.d.)).

GSK-3b is a major protein kinase that regulates the decrease of phosphorylated GSK-3b (Ser 9) and phosphorylation of b-catenin in the cytoplasm, targeting b-catenin protein levels. it todegradation(Kikuchi et al., 2006; Giles et al., Up-regulation of b-catenin and its nuclear accumula- 2007). The inactivation of GSK-3b by phosphorylation tion has been reported to play a major role in gastric has been shown to reduce the ubiquitination of cancer (Morin et al., 1997; Washington et al., 1998; b-catenin, resulting in nuclear accumulation and tran- Clements et al., 2002). Accumulation of the b-catenin scriptional activity of b-catenin (Giles et al., 2003; protein is attributed to either mutation of APC or Kikuchi et al., 2006). Indeed, we detected a decrease in b-catenin itself. One of the most widely observed the phosphorylation level of b-catenin (Ser33/37/Thr41) mechanisms for regulating b-catenin in cancer cells together with a significant up-regulation in the total involves mutations in APC gene that is an important b-catenin protein levels following overexpression of oncogenic step in the genesis of a number of malig- AURKA in AGS and SNU1 cells. These results nancies and it is an early event in colorectal carcinogen- suggested that down-regulation in the phosphorylation esis (Morin et al., 1997; Damalas et al., 1999a; Polakis, of b-catenin (Ser 33/37/Thr41) has ultimately led toan 1999). Mutations in b-catenin have been identified accumulation of b-catenin in AURKA overexpressing in many human tumors, including hepatocellular cells. We have validated our results using siRNA carcinoma (Terris et al., 1999), colorectal cancer (Sparks knockdown of AURKA and demonstrated a significant et al., 1998), ovarian cancer (Gamallo et al., 1999) and

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 871

Figure 5 Up-regulation of the b-catenin downstream targets. (a) Quantitative real-time PCR demonstrates up-regulation of several direct downstream transcriptional targets of b-catenin/TCF complex in AGS cells overexpressing Aurora kinase A (AURKA) as compared to empty vector control. The mRNA expression levels of cyclin D1, c-MYC, c-MYC-binding protein, FGF18, CLDN1 and vascular endothelial growth factor (VEGF) were normalized to hypoxanthine phosphorybosyl transferase 1 (HPRT1) mRNA expression. For comparison, the expression in empty vector control cells was set to 1. (b) The luciferase assay using the cyclin-D1-luc reporter indicated induction (4.75-fold) of cyclin D1 transcriptional activity in AURKA overexpressing cells. (c) Immunoblot analysis indicated up-regulation in the total protein level of cyclin D1 in AGS and SNU1 cells upon overexpressing AURKA. Flag antibody was used to detect exogenously expressed AURKA.

prostate cancer (Voeller et al., 1998). Although APC kinase dead AURKA mutant (D274A) did not induce mutations are rare in gastric carcinomas (Nakatsuru this activity, confirming the role of the kinase et al., 1992), accumulation and nuclear localization of active AURKA. Furthermore, we demonstrated the b-catenin occur in approximately one-third of gastric mRNA up-regulation of several direct transcription tumors (Clements et al., 2002). Therefore, the AURKA/ targets of the b-catenin/TCF complex; cyclin D1, cMYC, GSK-3b axis may explain some of these findings. cMYC-binding protein, CLDN1, FGF18 and VEGF (A Several reports have indicated that the oncogenic list of targets of b-catenin/TCF is available at http://www. functions of b-catenin are mediated by the nuclear stanford.edu/Brnusse/pathways/targets.html, with speci- fraction because of its interaction with the members of fic references for each of them). Taken together, these the TCF/LEF family of transcription factors. The results indicate that overexpression of AURKA leads to b-catenin/TCF transcription complex regulates the an increase in b-catenin protein level with induction of its expression of several genes that are involved in human nuclear translocation and activation of the b-catenin/TCF carcinogenesis such as cyclin D1 and c-MYC (He et al., transcriptional activity. 1998; Pennica et al., 1998; Crawford et al., 1999; Cyclin D1 is a transcription target of the b-catenin/ Shtutman et al., 1999; Hojilla et al., 2007). Indeed, our TCF complex that is important in cell proliferation, results demonstrate up-regulation and nuclear localiza- tumor progression and metastasis (Zhang et al., 1993; tion of b-catenin. The AURKA overexpression in- Wang et al., 2004). We have demonstrated induction of creased the pTopFlash luciferase activity whereas the its promoter activity following AURKA overexpression.

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 872

Figure 6 Aurora kinase A (AURKA) overexpression enhances cell proliferation of gastric cancer cells. (a) AGS cells stably expressing AURKA showed enhancement in the cell proliferation as evidenced by the 5-Bromo-20-deoxyuridine (BrdU) incorporation assay. BrdU assay was performed by using the BrdU labeling and detection kit I (Roche Applied Science), following the manufacturer’s protocol. Quantiﬁcation of BrdU-positive-stained cells indicated X2.5-fold increase in the cell proliferation of AGS cells expressing AURKA compared to control vector-expressing cells (bar graph). The results represent the average of three independent stable clones. Western blot demonstrates the overexpression of AGS cells stably overexpressing AURKA. (b) Cell cycle analysis of AGS cells overexpressing AURKA indicated an almost doubling of the S-phase (10–18%) as compared to control vector-expressing cells. Transfection of AGS cells with b-catenin-expressing vector (pSWTBCAT) was used as a positive control. (c) Primary gastrointestinal epithelial RIE-1 cells transduced with AURKA retroviruses showed more proliferation than cells transduced with control retroviruses as determined by an increase from 6.5 to 16% in the S-phase of cell cycle.

The AGS cells stably expressing AURKA showed more medium together with 10% fetal bovine serum (Invitrogen Life than two-fold increase in their proliferation rate as Technologies, Carlsbad, CA, USA) at 37 1C in an atmosphere compared to control vector cells and demonstrated an containing 5% CO2. AGS, SNU1 and SNU16 cells were almost doubling in the S-phase of cells overexpressing obtained from American Tissue Culture Collection (ATCC, AURKA as compared to control vector-expressing cells. Manassas, VA, USA). MKN45 cells were obtained from the German Collection of Microorganisms and Cell Cultures The increase in b-catenin level promotes neoplastic (Braunschweig Germany). The MKN75 cells were a gift from conversion and plays distinct roles in uncontrolled Dr Barbara Schneider and the RIE-1 cells were obtained from progression into the cell cycle through the G1 phase Dr Robert J Coffey, Jr (Vanderbilt University, Nashville, TN, (Sherr and Roberts, 1999; Shtutman et al., 1999; Tetsu USA). The expression plasmid for AURKA was generated by and McCormick, 1999). Our results suggest that PCR amplification of the full-length-coding sequence of AURKA/GSK-3b axis, by regulating b-catenin, may AURKA and cloned in frame into pcDNA3.1 (Invitrogen play a critical role in this process in gastric cancer cells. Life Technologies). A synthetic Flag-tag sequence was added In summary, our results demonstrate that at the N terminus of AURKA. Cloning of AURKA was AURKA regulates the GSK-3b phosphorylation and confirmed by sequencing and restriction enzyme digestion. activity leading to accumulation and activation of the pSWTBCAT plasmid (wild-type b-catenin cloned into p-shuttle backbone vector from Clontech, Mount View, CA, b-catenin/TCF complex in gastric cancers. The USA) was a kind gift from Dr Andy Lowy (University of AURKA/GSK-3b axis underscores a novel angle for California, San Diego). Retroviral expression constructs, the role of AURKA in cancer where other biological pBabe-puro, containing either the full length of AURKA- and cell signaling consequences of this axis remain to be coding sequence (pBabe-AURKA) or a kinase-dead AURKA elucidated in further studies. mutant (D274A) and HA-GSK-3b pcDNA were purchased from Addgene (Cambridge, MA, USA). Validated siRNA oligonucleotides specific for AURKA and negative control siRNA were purchased from Ambion (Austin, TX, USA). Materials andmethods Transient transfections were performed by using Fugene-6 (Roche Applied Science, Indianapolis, IN, USA), and Cell culture, vectors, small-interfering RNA and transfection Lipofectamine (Invitrogen Life Technologies, Carlsbad, CA, AGS, SNU1, SNU16, MKN45, MKN75, and RIE-1 cells were USA) following the manufacturer’s protocols. Recombinant cultured in F-12 (HAM) or Dulbecco’s modified Eagle’s active GSK-3b and AURKA proteins were purchased from

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 873 BioVision (Mountain View, CA, USA) and Cell Sciences 41C). Supernatants were collected into new tubes and 50 ml (Canton, MA, USA), respectively. AKT inhibitor VIII was of protein G-Agarose were added to samples and incubated at purchased from Calbiochem (Gibbstown, NJ, USA) and AGS 41C for 10 min. Samples were then centrifuged for 15 min and cells overexpressing AURKA or control vector were treated divided into three tubes. Specific and nonspecific antibodies with 1 mM of AKT inhibitor VIII for 24 h. were added to their respective tubes containing protein samples and incubated on ice for 15 min. Briefly, 50 mlof Luciferase assays protein G-Agarose beads were added to each tube and The TCF-responsive luciferase construct pTopFlash, its incubated on a rotating platform at 41C overnight. Samples mutant pFopFlash (Upstate Biotechnology, Waltham, MA, were centrifuged for 1 min and washed 8 Â with RIPA buffer. USA) and cyclin D1-Luc (Tetsu and McCormick, 1999) were Briefly, 20 mlof2Â protein-loading buffer were added to each used to study the transcriptional activity. The pTopFlash sample and boiled for 5 min. Samples were resolved by SDS/ vector has six TCF-binding sites, each three in the opposite PAGE and subjected towestern blotting. direction. The pFopFlash, a negative control, has mutant TCF-binding sites. The luciferase activity was determined Quantitative real-time RT-PCR using a dual-luciferase reporter assay kit (Promega, Madison, Total RNA was isolated using the RNeasy mini kit WI, USA). The relative luciferase activity was determined after (Qiagen, Valencia, CA, USA), and single-stranded cDNA 48 h of transfection and normalized to protein concentration. was subsequently synthesized using the Advantage RT-for- Results are average of three independent experiments and PCR Kit (Clontech, Palo Alto, CA, USA). Gene-specific expressed as mean values (±standard deviation (s.d.)). primers for cyclin D1, c-MYC, c-MYC-binding protein, CLDN1, FGF18, VEGF and HPRT1 were designed using the Fluorescence-activated cell sorting online software, Primer 3 (http://frodo.wi.mit.edu/cgi-bin/ Cell-cycle analysis was performed using flow cytometry. Cells primer3/primer3_www.cgi). The forward and the reverse were synchronized by culture in the presence of 1% fetal bovine primers were designed tospan twodifferent exons.All primers serum for 24 h, followed by 10% fetal bovine serum. Cells were were purchased from GeneLink (Hawthorne, NY, USA) and trypsinized, washed twice with 1 Â ice-cold phosphate buffer their sequences are available from the authors upon request. saline (PBS) and then resuspended in 0.2 ml ice-cold PBS. Cells qRT-PCR was performed using an iCycler (Bio-Rad Labora- were then fixed in 1 ml ice-cold 70% ethanol and incubated for tories) with a threshold cycle number determined with the use 1h at41C. Cells were centrifuged and resuspended in 1 ml PBS of iCycler software version 3.0. Reactions were performed in and treated with propidium iodide (50 mg/ml) and RNase (1 mg/ml) triplicate and threshold cycle numbers were averaged. for 30 min at 371C and analysed by BD LSR III flow cytometer The results were normalized to HPRT1 as described earlier (BD Biosciences, San Jose, CA, USA). Data were analysed by BD (El-Rifai et al., 2002). The fold change in all samples was FACS Diva software. calculated according to the formula 2(RtÀEt)/2(RnÀEn),as described previously (El-Rifai et al., 2002). Western blot analysis Cell lysates were prepared in 1 Â PBS buffer containing 1 Â In vitro kinase assay Halt protease inhibitor cocktail and 1 Â Halt phosphatase The in vitro kinase assay was performed using active inhibitor cocktail (Pierce, Rockford, IL, USA), centrifuged at recombinant AURKA (Cell sciences, Canton, MA, USA) 3500 r.p.m. for 10 min at 41C. Protein concentration was and recombinant GSK-3b (BioVision, Mountain View, CA, measured using a Bio-Rad protein assay (Bio-Rad Labora- USA) following the manufacturer’s protocol. Briefly, reaction tories, Hercules, CA, USA). Proteins (10–15 mg) from each was carried out in 20 ml of assay buffer (50 mM HEPES, sample were subjected to SDS/polyacrylamide gel electrophor- pH7.4 þ 3mM Mgcl2 þ 3mM Mncl2 þ 1mM DTT þ 3 mM Na- esis (PAGE) and transferred onto nitrocellulose membrane. orthovanadate þ 0.5 mM ATP) containing a constant concen- Target proteins were detected by using specific antibodies tration of 0.2 mg of GSK-3b and increasing concentrations against AURKA (TransGenic Inc., Japan); b-actin, b-catenin, from 0.1 to 0.8 mg of human recombinant AURKA. The TFF1 phospho-b-catenin (Ser33/37/Thr41), GSK-3b, phospho- protein (Abnova, Taipei City, Taiwan) was used as a negative GSK3b (Ser 9), cyclin D1, AKT, pAKT(S473), HA-tag and control as it is not a phospho protein and is not known to Flag (Cell Signaling Technology, Boston, MA, USA). interact with AURKA or GSK-3b. Reaction mixtures were incubated at 301C for 30 min. Briefly, 6 mlof2Â protein- 5-Bromo-20-deoxyuridine labeling and detection loading buffer was added to each sample and boiled for 5 min. 5-Bromo-20-deoxyuridine labeling was performed using the Samples were resolved by SDS/PAGE and subjected to BrdU labeling and detection kit I (Roche Applied Science), western blotting using specific antibodies for AURKA and following the manufacturer’s protocol. In brief, cells were pGSK-3b Ser 9. grown in an eight-well chamber and BrdU-labeling medium was added tocells and incubated at 37 1C for 3 h. Cells were Immunofluorescence assay washed once with 1 Â PBS followed by ethanol fixation for AGS cells (3 Â 103 per well) were plated in eight-well plate 20 min at À201C. After washing cells, anti-BrdU reagent was chamber for 24 h. Cells were washed with 1 Â PBS and fixed added for 30 min at 371C. For fluorescence detection, anti- with acetone/methanol (1:1) mixture for 15 min at À201C. mouse immunoglobulin fluorescein was added and incubated Cells were then washed twice with 1 Â PBS, followed by for 30 min. Cells were visualized under an Olympus BX51 blocking with 10% normal goat serum blocking solution fluorescence microscope (Olympus Co., Tokyo, Japan). (Zymed Laboratories, Carlsbad, CA, USA) for 20 min at room temperature in a humidified chamber. Cells were hybridized Immunoprecipitation with the specific primary antibodies against AURKA, Cells (5 Â 106) were centrifuged for 5 min at 1500 r.p.m. After b-catenin or GSK-3b (Cell Signaling, Danvers, MA, USA) discarding the supernatant, cell pellet was resuspended in diluted in 1 Â PBS (1:400) for 2 h at room temperature in a RIPA buffer. Samples were sonicated 3 Â for 10 s and humidified chamber. Cells were washed three times in 1 Â PBS centrifuged for 20 min at maximum speed (13 000 r.p.m., and hybridized with fluorescently conjugated secondary

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 874 antibody (1:1000; Jackson Immunoresearch, West Grove, PA, Acknowledgements USA) for 45 min at room temperature in a humidified chamber. AURKA and b-catenin were detected by fluorescein isothio- This study was supported by a pilot fund from the National cyanate green fluorescence and GSK-3b by tetramethylrhoda- Cancer Institute GI SPORE grant CA95103 and the mine isothiocyanate (red fluorescence). Cells were washed in Vanderbilt-Ingram Cancer Center support funds. The contents 1 Â PBS, mounted with Vectashield/DAPI (Vector Labora- of this work are solely the responsibility of the authors tories, Burlingame, CA, USA) and visualized by an Olympus and do not necessarily represent the official views of BX51 fluorescence microscope (Olympus Co., Tokyo, Japan). the National Cancer Institute or Vanderbilt University At least 200 cells were counted from each experiment. Medical Center.

References

Aberle H, Bauer A, Stappert J, Kispert A, Kemler R. (1997). Beta- He TC, Sparks AB, RagoC, Hermeking H, Zawel L, da CostaLT catenin is a target for the ubiquitin-proteasome pathway. EMBO J et al. (1998). Identification of c-MYC as a target of the APC 16: 3797–3804. pathway. Science 281: 1509–1512. Behrens J, von Kries JP, Kuhl M, Bruhn L, Wedlich D, Grosschedl R Hojilla CV, Kim I, Kassiri Z, Fata JE, Fang H, Khokha R. (2007). et al. (1996). Functional interaction of beta-catenin with the Metalloproteinase axes increase beta-catenin signaling in primary transcription factor LEF-1. Nature 382: 638–642. mouse mammary epithelial cells lacking TIMP3. J Cell Sci 120: Bischoff JR, Anderson L, Zhu Y, Mossie K, Ng L, Souza B et al. 1050–1060. (1998). A homologue of Drosophila aurora kinase is oncogenic Hu W, Kavanagh JJ, Deaver M, Johnston DA, Freedman RS, and amplified in human colorectal cancers. EMBO J 17: Verschraegen CF et al. (2005). Frequent overexpression of STK15/ 3052–3065. Aurora-A/BTAK and chromosomal instability in tumorigenic cell Clements WM, Wang J, Sarnaik A, Kim OJ, MacDonald J, cultures derived from human ovarian cancer. Oncol Res 15: 49–57. Fenoglio-Preiser C et al. (2002). Beta-catenin mutation is a frequent Jemal A, Siegel R, Ward E, Murray T, Xu J, Smigal C et al. (2006). cause of Wnt pathway activation in gastric cancer. Cancer Res 62: Cancer statistics, 2006. CA Cancer J Clin 56: 106–130. 3503–3506. Kamada K, Yamada Y, HiraoT, FujimotoH,Takahama Y, UenoM Crawford HC, Fingleton BM, Rudolph-Owen LA, Goss KJ, et al. (2004). Amplification/overexpression of Aurora-A in human Rubinfeld B, Polakis P et al. (1999). The metalloproteinase gastric carcinoma: potential role in differentiated type gastric matrilysin is a target of beta-catenin transactivation in intestinal carcinogenesis. Oncol Rep 12: 593–599. tumors. Oncogene 18: 2883–2891. Katayama H, Sasai K, Kawai H, Yuan ZM, Bondaruk J, Suzuki F Damalas A, Ben-Ze’ev A, Simcha I, Shtutman M, Leal JF, Zhurinsky et al. (2004). Phosphorylation by aurora kinase A induces Mdm2- J et al. (1999a). Excess beta-catenin promotes accumulation of mediated destabilization and inhibition of p53. Nat Genet 36: 55–62. transcriptionally active p53. EMBO J 18: 3054–3063. Kikuchi A, Kishida S, Yamamoto H. (2006). Regulation of Wnt Damalas A, Ben-Ze’ev A, Simcha I, Shtutman M, Leal JF, Zhurinsky signaling by protein–protein interaction and post-translational J et al. (1999b). Excess beta-catenin promotes accumulation of modifications. Exp Mol Med 38: 1–10. transcriptionally active p53. EMBO J 18: 3054–3063. Koh TJ, Bulitta CJ, Fleming JV, Dockray GJ, Varro A, Wang TC. Dar AA, Zaika A, Piazuelo MB, Correa P, Koyama T, Belkhiri A (2000). Gastrin is a target of the beta-catenin/TCF-4 growth- et al. (2008). Frequent overexpression of Aurora kinase A in signaling pathway in a model of intestinal polyposis. J Clin Invest upper gastrointestinal adenocarcinomas correlates with potent 106: 533–539. antiapoptotic functions. Cancer 112: 1688–1698. Miller JR, Hocking AM, Brown JD, Moon RT. (1999). Mechanism Earle CC, Maroun JA. (1999). Adjuvant chemotherapy after and function of signal transduction by the Wnt/beta-catenin and curative resection for gastric cancer in non-Asian patients: Wnt/Ca2+ pathways. Oncogene 18: 7860–7872. revisiting a meta-analysis of randomised trials. Eur J Cancer 35: Mitsiades CS, Mitsiades N, Koutsilieris M. (2004). The Akt pathway: 1059–1064. molecular targets for anti-cancer drug development. Curr Cancer El-Rifai W, Frierson HJ, Moskaluk C, Harper J, Petroni G, Bissonette Drug Targets 4: 235–256. E et al. (2001). Genetic differences between adenocarcinomas arising Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, Vogelstein B in Barrett’s esophagus and gastric mucosa. Gastroenterology 121: et al. (1997). Activation of beta-catenin-Tcf signaling in colon cancer 592–598. by mutations in beta-catenin or APC. Science 275: 1787–1790. El-Rifai W, Moskaluk CA, Abdrabbo MK, Harper J, Yoshida C, Nakatsuru S, Yanagisawa A, Ichii S, Tahara E, KatoY, Nakamura Y Riggins GJ et al. (2002). Gastric cancers overexpress S100A calcium- et al. (1992). Somatic mutation of the APC gene in gastric cancer: binding proteins. Cancer Res 62: 6823–6826. frequent mutations in very well differentiated adenocarcinoma and Gamallo C, Palacios J, Moreno G, Calvo de Mora J, Suarez A, signet-ring cell carcinoma. Hum Mol Genet 1: 559–563. Armas A. (1999). beta-catenin expression pattern in stage I and II Ogawa E, Takenaka K, Katakura H, Adachi M, Otake Y, Toda Y ovarian carcinomas: relationship with beta-catenin gene mutations, et al. (2008). Perimembrane Aurora-A expression is a significant clinicopathological features, and clinical outcome. Am J Pathol 155: prognostic factor in correlation with proliferative activity in non- 527–536. small-cell lung cancer (NSCLC). Ann Surg Oncol 15: 547–554. Giet R, Petretti C, Prigent C. (2005). Aurora kinases, aneuploidy Orford K, Crockett C, Jensen JP, Weissman AM, Byers SW. (1997). and cancer, a coincidence or a real link? Trends Cell Biol 15: Serine phosphorylation-regulated ubiquitination and degradation of 241–250. beta-catenin. J Biol Chem 272: 24735–24738. Giet R, Prigent C. (1999). Aurora/Ipl1p-related kinases, a new Ouchi M, Fujiuchi N, Sasai K, Katayama H, Minamishima YA, oncogenic family of mitotic serine-threonine kinases. J Cell Sci Ongusaha PP et al. (2004). BRCA1 phosphorylation by Aurora-A in 112(Part 21): 3591–3601. the regulation of G2 to M transition. J Biol Chem 279: 19643–19648. Giles FJ, Cortes J, Jones D, Bergstrom D, Kantarjian H, Freedman SJ. Pap M, Cooper GM. (1998). Role of glycogen synthase kinase-3 in the (2007). MK-0457, a novel kinase inhibitor, is active in patients with phosphatidylinositol 3-Kinase/Akt cell survival pathway. J Biol chronic myeloid leukemia or acute lymphocytic leukemia with the Chem 273: 19929–19932. T315I BCR-ABL mutation. Blood 109: 500–502. Pennica D, Swanson TA, Welsh JW, Roy MA, Lawrence DA, Lee J Giles RH, van Es JH, Clevers H. (2003). Caught up in a Wnt storm: et al. (1998). WISP genes are members of the connective tissue Wnt signaling in cancer. Biochim Biophys Acta 1653: 1–24. growth factor family that are up-regulated in wnt-1-transformed

Oncogene Overexpression of AURKA regulates GSK-3b AA Dar et al 875 cells and aberrantly expressed in human colon tumors. Proc Natl Varis A, Puolakkainen P, Savolainen H, Kokkola A, Salo J, Nieminen O Acad Sci USA 95: 14717–14722. et al. (2001). DNA copy number profiling in esophageal Barrett Polakis P. (1999). The oncogenic activation of beta-catenin. Curr Opin adenocarcinoma: comparison with gastric adenocarcinoma and eso- Genet Dev 9: 15–21. phageal squamous cell carcinoma. Cancer Genet Cytogenet 127: 53–58. Sherr CJ, Roberts JM. (1999). CDK inhibitors: positive and negative Voeller HJ, Truica CI, Gelmann EP. (1998). Beta-catenin mutations in regulators of G1-phase progression. Genes Dev 13: 1501–1512. human prostate cancer. Cancer Res 58: 2520–2523. Shtutman M, Zhurinsky J, Simcha I, Albanese C, D’AmicoM, Pestell Wang C, Li Z, Fu M, Bouras T, Pestell RG. (2004). Signal R et al. (1999). The cyclin D1 gene is a target of the beta-catenin/ transduction mediated by cyclin D1: from mitogens to cell LEF-1 pathway. Proc Natl Acad Sci USA 96: 5522–5527. proliferation: a molecular target with therapeutic potential. Cancer Sourbier C, Lindner V, Lang H, Agouni A, Schordan E, Danilin S Treat Res 119: 217–237. et al. (2006). The phosphoinositide 3-kinase/Akt pathway: a Washington K, Chiappori A, Hamilton K, Shyr Y, Blanke C, Johnson new target in human renal cell carcinoma therapy. Cancer Res 66: D et al. (1998). Expression of beta-catenin, alpha-catenin, and 5130–5142. E-cadherin in Barrett’s esophagus and esophageal adenocarcinomas Sparks AB, Morin PJ, Vogelstein B, Kinzler KW. (1998). Mutational (In Process Citation). Mod Pathol 11: 805–813. analysis of the APC/beta-catenin/Tcf pathway in colorectal cancer. Yamada T, Takaoka AS, Naishiro Y, Hayashi R, Maruyama K, Cancer Res 58: 1130–1134. Maesawa C et al. (2000). Transactivation of the multidrug resistance Tanaka T, Kimura M, Matsunaga K, Fukada D, Mori H, Okano Y. 1 gene by T-cell factor 4/beta-catenin complex in early colorectal (1999). Centrosomal kinase AIK1 is overexpressed in invasive ductal carcinogenesis. Cancer Res 60: 4761–4766. carcinoma of the breast. Cancer Res 59: 2041–2044. Yang H, He L, Kruk P, Nicosia SV, Cheng JQ. (2006). Aurora-A Terris B, Pineau P, Bregeaud L, Valla D, Belghiti J, Tiollais P et al. induces cell survival and chemoresistance by activation of Akt (1999). Close correlation between beta-catenin gene alterations and through a p53-dependent manner in ovarian cancer cells. Int J nuclear accumulation of the protein in human hepatocellular Cancer 119: 2304–2312. carcinomas. Oncogene 18: 6583–6588. Zhang YJ, Jiang W, Chen CJ, Lee CS, Kahn SM, Santella RM et al. Tetsu O, McCormick F. (1999). Beta-catenin regulates expression of (1993). Amplification and overexpression of cyclin D1 in cyclin D1 in colon carcinoma cells. Nature 398: 422–426. human hepatocellular carcinoma. Biochem Biophys Res Commun Tong T, Zhong Y, Kong J, Dong L, Song Y, Fu M et al. (2004). 196: 1010–1016. Overexpression of Aurora-A contributes to malignant development Zhou H, Kuang J, Zhong L, Kuo WL, Gray JW, Sahin A et al. (1998). of human esophageal squamous cell carcinoma. Clin Cancer Res 10: Tumour amplified kinase STK15/BTAK induces centrosome ampli- 7304–7310. fication, aneuploidy and transformation. Nat Genet 20: 189–193.

Oncogene