Oncogene (2009) 28, 3597–3607 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE Ect2 links the PKCi–Par6a complex to Rac1 activation and cellular transformation

V Justilien and AP Fields

Department of Cancer Biology, Mayo Clinic College of Medicine, Jacksonville, FL, USA

Protein kinase Ci (PKCi) promotes non-small cell lung domain causes transformation in mouse fibroblasts, cancer (NSCLC) by binding to Par6a and activating a whereas full-length Ect2 does not (Miki et al., 1993; Rac1-Pak-Mek1,2-Erk1,2 signaling cascade. The me- Saito et al., 2004). The relevance of this observation to chanism by which the PKCi–Par6a complex regulates human cancer is unclear. Human cancers express only Rac1 is unknown. Here we show that epithelial cell full-length Ect2, indicating that the transforming transforming sequence 2 (Ect2), a guanine nucleotide C-terminal Ect2 fragment is not relevant to human exchange factor for Rho family GTPases, is coordinately cancer (Saito et al., 2003). Paradoxically, Ect2 is amplified and overexpressed with PKCi in NSCLC overexpressed in some human tumors (Sano et al., tumors. RNA interference-mediated knockdown of Ect2 2006; Salhia et al., 2008; Zhang et al., 2008; Hirata et al., inhibits Rac1 activity and blocks transformed growth, 2009) and transient Ect2 knockdown (KD) induces invasion and tumorigenicity of NSCLC cells. Expression transitory cell cycle arrest in vitro, indicating a possible of constitutively active Rac1 (RacV12) restores transfor- role for Ect2 in tumor cell growth (Hirata et al., 2009). mation to Ect2-deficient cells. Interestingly, the role of However, the mechanism of Ect2 involvement in human Ect2 in transformation is distinct from its well-established tumorigenesis remains unresolved. role in cytokinesis. In NSCLC cells, Ect2 is mislocalized Here we report a novel genetic, biochemical and to the cytoplasm where it binds the PKCi–Par6a complex. functional link between Ect2 and the kinase Ci RNA interference-mediated knockdown of either PKCi (PKCi) oncogene. Ect2 and PKCi are coordinately or Par6a causes Ect2 to redistribute to the nucleus, overexpressed in primary non-small cell lung cancer indicating that the PKCi–Par6a complex regulates the (NSCLC) tumors as a consequence of tumor-specific co- cytoplasmic localization of Ect2. Our data indicate that amplification of the ECT2 and PRKCI . Further- Ect2 and PKCi are genetically and functionally linked in more, Ect2 is required for transformed growth and NSCLC, acting to coordinately drive tumor cell prolif- invasion of NSCLC cells in vitro and tumorigenicity eration and invasion through formation of an oncogenic in vivo. The role of Ect2 in transformation is distinct PKCi–Par6a-Ect2 complex. from its well-characterized function in cytokinesis. Ect2 Oncogene (2009) 28, 3597–3607; doi:10.1038/onc.2009.217; mislocalizes to the cytoplasm of NSCLC cells where it published online 20 July 2009 binds the oncogenic PKCi–Par6a complex and activates Rac1. Furthermore, the PKCi–Par6a complex regulates Keywords: amplification; anchorage-independent the cytoplasmic mislocalization and oncogenic function growth; invasion; Rac1; Mek-Erk signaling; cytokinesis of Ect2. Thus, Ect2 and PKCi are genetically linked by co-amplification, and functionally linked through their physical association in an oncogenic PKCi–Par6a-Ect2 complex that drives transformation through Rac1. Introduction

Epithelial cell transforming sequence 2 (Ect2) is a Results guanine nucleotide exchange factor (GEF) for Rho family GTPases that functions in cytokinesis (Tatsumoto Ect2 and PKCi are coordinately amplified et al., 1999; Kim et al., 2005; Niiya et al., 2005, 2006; and overexpressed in primary NSCLC tumors Hara et al., 2006). Ect2 is comprised of a C-terminal The gene encoding Ect2 resides on 3q26, a GEF domain, and an N-terminal regulatory domain region frequently amplified in lung squamous cell that modulates Ect2 activity and localization (Kim carcinoma (LSCC) (Brass et al., 1996; Balsara et al., et al., 2005). Ectopic expression of the Ect2 GEF 1997; Sugita et al., 2000). To assess Ect2 expression, we analysed 137 primary lung tumors (68 lung adenocarci- Correspondence: Dr AP Fields, Department of Cancer Biology, Mayo noma (LAC) and 69 LSCC) and matched normal lung Clinic College of Medicine, Griffin Cancer Research Building, Rm 212, by quantitative real-time PCR (qPCR). Ect2 mRNA 4500 San Pablo Road, Jacksonville, FL 32224, USA. E-mail: fi[email protected] was significantly higher in LAC (mean 4.6-fold increase; Received 24 March 2009; revised 23 May 2009; accepted 24 May 2009; Po0.0001) and LSCC (mean 6.8-fold increase; published online 20 July 2009 Po0.0001) tumors than matched normal lung Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3598

Figure 1 Epithelial cell transforming sequence 2 (Ect2) and protein kinase Ci (PKCi) are coordinately amplified and over- expressed in primary non-small cell lung cancer (NSCLC) tumors. Ect2 is overexpressed in primary lung adenocarcinoma (LAC) and squamous cell carcinoma (LSCC) (a). Sixty-eight LAC cases, 69 Figure 2 Epithelial cell transforming sequence 2 (Ect2) is over- LSCC cases and matched normal control lung tissues were expressed and mislocalized to the cytoplasm in primary non-small analysed by quantitative real-time PCR for Ect2 mRNA abun- cell lung cancer tumors. Immunohistochemical localization of Ect2 dance. Ect2 mRNA is significantly higher in LAC and LSCC tumor in normal lung epithelium (a), lung adenocarcinoma (LAC) (b) and samples than in matched normal lung (*Po0.0001 and Po0.0001, lung squamous cell carcinoma (LSCC) (c) was carried out as respectively). (b) ECT2 gene amplification drives Ect2 expression in described in Materials and methods. Ect2 localizes predominantly LSCC. LSCC tumors were analysed for ECT2 gene copy number to the nucleus of normal lung epithelial cells (a). Ect2 is as described in Materials and methods. Tumors harboring ECT2 overexpressed in both LAC (b) and LSCC (c) tumors, and is gene amplification express higher Ect2 mRNA than tumors mislocalized to the cytoplasm of tumor cells. Ect2 staining without ECT2 amplification (*Po 0.001). (c) ECT2 and PRKCI specificity was confirmed using an excess of Ect2 peptide in the are co-amplified in LSCC. ECT2 and PRKCI gene copy number antibody solution (a–c, Ect2 þ peptide). exhibit a significant correlation in NSCLC tumors (n ¼ 94; R2 ¼ 0.76; *Po0.00001). (d) Ect2 and PKCi mRNA are coordi- nately overexpressed in LSCC tumors harboring gene amplification (n ¼ 67; R2 ¼ 0.52; *Po0.00001). amplification (Figure 1d; R2 of 0.58, Po0.00001). Thus, ECT2 and PRKCI are co-amplified and coordinately overexpressed in LSCC tumors. (Figure 1a). Elevated Ect2 mRNA (defined as >2 s.d. Immunohistochemical analysis reveals that Ect2 is above the mean Ect2 mRNA value of matched normal restricted to the nucleus of normal lung epithelial cells lung) was observed in 113/137 (82%) of NSCLC tumors (Figure 2a, left panel). In contrast, LAC and LSCC and was equally prevalent in LAC and LSCC. tumors display increased Ect2 staining in the cytoplasm We next assessed ECT2 copy number. In all, 67 of (Figures 2b and c, left panels). Staining specificity was 94 (71%) LSCC tumors exhibited ECT2 amplification confirmed using a 10-fold excess of recombinant Ect2 (defined as a gain of one or more ECT2 alleles). Analysis peptide in the antibody solution, which significantly of 40 LAC tumors revealed no increase in ECT2 copy inhibited Ect2 staining (Figure 2a–c, right panels). number (data not shown), consistent with the restricted Analysis of 68 primary NSCLC tumors (36 LSCC and distribution of 3q26 amplification to LSCC tumors 32 LAC) revealed elevated cytoplasmic Ect2 staining in (Brass et al., 1996; Balsara et al., 1997; Sugita et al., 57/68 or B84% of tumors, showing that Ect2 protein is 2000). LSCC tumors harboring ECT2 amplification frequently overexpressed in NSCLC tumors. showed significantly higher Ect2 mRNA than tumors lacking amplification (Po0.001) (Figure 1b). PRKCI is Ect2 is required for anchorage-independent growth adjacent to ECT2 on chromosome 3q26 and is a target and invasion of NSCLC cells of frequent tumor-specific gene amplification in LSCC We next assessed whether Ect2 has a role in the tumors (Regala et al., 2005b). ECT2 copy number transformed phenotype of NSCLC cells. H1703 LSCC showed a statistically significant, positive correlation cells were analysed because they harbor B4-fold with PRKCI copy number (Figure 1c, R2 of 0.76, increase in ECT2 copy number and express abundant Po0.00001). In addition, PKCi and Ect2 mRNA Ect2 protein. Cells were stably transduced with one of expression correlate in LSCC harboring 3q26 gene three lentiviral Ect2 RNA interference (RNAi) con-

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3599

Figure 3 Epithelial cell transforming sequence 2 (Ect2) is important for anchorage-independent growth and invasion of non-small cell lung cancer cells. H1703 cells were infected with one of three recombinant lentiviruses containing RNA interference (RNAi)-targeting Ect2 (Ect2-RNAi #1–3) or a non-target (NT) sequence. Cell populations stably transduced with each RNAi were isolated and analysed as described in Materials and methods. (a) Cells expressing Ect2-RNAi or NT-RNAi constructs were assessed for Ect2 mRNA abundance by quantitative real-time PCR and Ect2 protein by immunoblot analysis. Cells expressing Ect2-RNAi or NT-RNAi were analysed for anchorage-independent growth in soft agar (b) and cellular invasion through Matrigel-coated chambers (c). Ect2-RNAi (construct #3) and NT-RNAi cells were transfected with a plasmid expressing GFP-Ect2 ( þ Ect2) or a control empty plasmid, and assessed for GFP-Ect2 and endogenous Ect2 expression by immunoblot analysis (d). Cells from d were assessed for anchorage- independent growth in soft agar (e) and invasion (f). Data in panels a–c, e and f are expressed as % NT control and represent the mean±s.e.m.; n ¼ 3. *denotes a statistically significant difference from NT; **denotes a statistically significant difference from Ect2 knockdown cells in the absence of GFP-Ect2, Po0.05).

structs or a non-target (NT) RNAi control lentivirus, were injected subcutaneously into the flanks of athymic and cell populations were assessed for Ect2 mRNA nude mice and tumor growth was monitored over 4 abundance and protein expression (Figure 3a). Each of weeks. A549/Ect2 KD cells exhibited impaired tumor the three Ect2 RNAi constructs induced a reduction growth when compared with A549/NT cells (Figure 4a). in Ect2 mRNA and protein when compared with NT Immunohistochemical analysis confirmed reduced cells (Figure 3a). Functionally, Ect2 KD cells exhibited Ect2 expression in A549/Ect2 KD tumors (Figure 4b). impaired anchorage-independent growth (Figure 3b) A549/Ect2 KD tumors exhibited a significant 2.3-fold and invasion (Figure 3c) commensurate with the level reduction in labeling index (defined as BrdU-labeled of Ect2 KD. To confirm that this cellular phenotype nuclei/total nuclei) compared with A549/NT tumors is due to Ect2 KD, we expressed RNAi-resistant, (Figure 4c). In contrast, A549/NT and A549/Ect2 KD GFP-tagged Ect2 in NT and Ect2 KD cells tumors exhibited no significant difference in TUNEL- (Figure 3d). Exogenous Ect2 significantly restored positive nuclei (0.19±0.05% in A549/NT tumors versus anchorage-independent growth (Figure 3e) and invasion 0.19±0.09% in A549/Ect2 KD tumors; P>0.97) or (Figure 3f) in Ect2 KD cells but had no effect on NT CD31 staining (Figure 4d), indicating that Ect2 KD cells. We also analysed a panel of four cell lines primarily affects tumor cell proliferation. representing major subtypes of NSCLC, LAC (A549 and H358), large-cell lung carcinoma (H460) and LSCC (A427) (Supplementary Figure 2). Each cell line Ect2 KD does not induce cytokinesis defects exhibited a significant inhibition of anchorage-indepen- in NSCLC cells dent growth and invasion after Ect2 KD indicating that A major physiologic function of Ect2 is the regulation of Ect2 is important for transformation in many NSCLC cytokinesis (Tatsumoto et al., 1999; Niiya et al., 2005; cell lines. Hara et al., 2006). Ect2 KD in non-transformed cells leads to a cytokinesis defect characterized by decreased cell growth and accumulation of multinucleated cells Ect2 is important for NSCLC tumorigenicity in vivo (Tatsumoto et al., 1999). To determine whether Ect2 We next assessed the role of Ect2 in NSCLC tumor- KD in NSCLC cells affects cytokinesis, A549, H1703 igenicity in vivo. A549 cells were chosen because they and Madin-Darby canine kidney (MDCK) cells were grow as subcutaneous tumors with well-established stably transduced with either Ect2 or NT RNAi. The growth kinetics. A549/Ect2 KD and A549/NT cells target sequence of Ect2 RNAi construct #3 is conserved

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3600 between human and canine Ect2 resulting in efficient data indicate that the role of Ect2 in transformation is Ect2 KD in all three cell lines (Figure 5a). As expected, distinct from its role in cytokinesis. MDCK/Ect2 KD cells exhibited an increased popula- tion-doubling time (PDT) compared with MDCK/NT cells (Figure 5b). In contrast, neither A549/Ect2 KD nor Ect2 regulates Rac1 activity in NSCLC cells H1703/Ect2 KD cells showed a significant change in As Ect2 can regulate the activity of the Rho GTPases PDT compared with their NT cells (Figure 5b). Rac1, Cdc42 and RhoA in vitro (Miki et al., 1993), we Furthermore, MDCK/Ect2 KD cell cultures exhibited assessed Rho GTPase activity in H1703/Ect2 KD and an accumulation of large, multinucleated cells, which H1703/NT cells. Ect2 KD cells exhibited decreased was not observed in A549/Ect2 KD or H1703/Ect2 KD Rac1 activity, but no change in Cdc42 or RhoA activity cell cultures (Figure 5c). Quantitative analysis revealed a compared with NT cells (Figure 6a). Quantitative 12-fold increase in multinucleated cells in MDCK/Ect2 analysis revealed a significant B50% inhibition in KD cultures but no significant change in A549/Ect2 KD Rac1 activity in Ect2 KD cells (Figure 6b). Further- or H1703/Ect2 KD cultures when compared with their more, expression of a constitutively active Rac1 allele respective NT cell cultures (Figure 5d). We also (RacV12) in H1703/Ect2 KD and H1703/NT cells observed no accumulation of multinucleated cells in (Figure 6c) restored anchorage-independent growth A549/Ect2 KD tumors in vivo (1.3±0.10% in A549/NT and invasion in Ect2 KD cells, although having no cells versus 1.3±0.15% in A549/Ect2 KD cells; effect in NT cells (Figures 6d and e). Similar results were P ¼ 0.98, n ¼ >500 cells counted per tumor). These observed in A549 cells (data not shown).

Figure 4 Epithelial cell transforming sequence 2 (Ect2) is required for non-small cell lung cancer tumorigenicity in vivo. A549/non- target (NT) and A549/Ect2 knockdown (KD) cells were injected subcutaneously into the flanks of nude mice and tumor volumes were determined over a 4-week period as described in Materials and methods. (a) A549/Ect2 KD tumors grow significantly slower than A549/NT cell tumors. *indicates a statistically significant difference between A549/NT and A549/Ect2 KD tumor volumes at the indicated time points (Po0.05). (b) A549/NT and A549/Ect2 KD tumors stained with hematoxylin and eosin or Ect2 by immunohistochemistry. Ect2 expression is diminished in A549/Ect2 KD tumors. (c) BrdU staining of A549/NT and A549/Ect2 KD tumors (inset). BrdU-positive nuclei were counted as described in Materials and methods. A549/Ect2 KD tumors exhibit a statistically significant decreased in BrdU labeling index when compared with A549/NT tumors (*Po0.05). (d) A549/NT and A549/Ect2 KD tumors exhibited no differences in TUNEL or CD31 staining.

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3601

Figure 5 Epithelial cell transforming sequence 2 (Ect2) knockdown (KD) does not induce cytokinesis defects in non-small cell lung cancer cells. (a) RNA interference (RNAi)-mediated knockdown of Ect2 expression in A549, H1703 and Madin-Darby canine kidney (MDCK) cells. A549/Ect2 KD, A549/non-target (NT), H1703/Ect2 KD, H1703/NT, MDCK/Ect2 KD and MDCK/NT cells were analysed for Ect2 mRNA and Ect2 protein. (b) Ect2 KD causes an increase in population-doubling time (PDT) in MDCK cells, but not A549 or H1703 cells. PDTs were determined as described in Materials and methods. Results are expressed as mean doubling time in hours ±s.e.m.; n ¼ 3; *Po0.05. (c) Ect2 KD induces accumulation of multinucleated cells in MDCK but not A549 or H1703 cells. Cells were stained with phalloidin and 40,6 diamidino-2-phenylindole and analysed by confocal immunofluorescence microscopy as described in Materials and methods. Arrows indicate multinucleated cells in MDCK/Ect2 KD cultures. (d) Quantitative analysis of multinucleated cells. At least 1000 cells were counted from each cell line in each of three independent experiments. Data are expressed as % multinucleated cells±s.e.m. *indicates a statistically significant difference from MDCK/NT cells; Po0.02, n ¼ 3.

Rac1 drives tumor cell proliferation by activating a Thus, Ect2 KD inhibits the Rac1-Pak-Mek1,2-Erk1,2 Pak-Mek1,2-Erk1,2 proliferative signaling pathway proliferative signaling pathway in vivo. (Regala et al., 2005a). Therefore, we assessed the status of this pathway in A549/NT and A549/Ect2 KD tumors in vivo. Immunoblot analysis confirmed that A549/Ect2 Ect2 associates with the PKCi–Par6a complex KD tumors express less Ect2 than A549/NT tumors in NSCLC cells (Figure 7a). Quantitative analysis reveals a significant Both Ect2 and the PKCi–Par6a complex regulate Rac1 B5-fold decrease in Ect2 protein in A549/Ect2 KD signaling in NSCLC cells. Therefore, we assessed tumors (Figure 7b). A549/Ect2 KD tumors exhibit whether Ect2 binds the PKCi–Par6a complex (Figure 8). lower phospho-298 Mek1,2 than A549/NT tumors with As commercially available Par6 antibodies do not no change in total Mek1,2 (Figure 7a) resulting in a efficiently detect Par6a, we generated MDCK and decreased ratio of phospho-298-Mek1,2 to total Mek1,2 H1703 cells expressing Flag-tagged wild-type human (Figure 7c). A similar decrease in phospho-Erk1,2 Par6a (wtPar6) or empty plasmid (Vector). Immuno- (without a change in total Erk1,2) is observed in blot analysis confirmed expression of endogenous A549/Ect2 KD tumors (Figure 7a), resulting in a Ect2, PKCi and Flag-Par6a in MDCK/wtPar6 cells decreased ratio of phospho-Erk1,2 to total Erk1,2 (Figure 8a, lysate). When PKCi was immunoprecipi- (Figure 7d). No evidence for PARP cleavage, a tated from MDCK/Vector and MDCK/wtPar6 cells, biochemical marker of apoptosis, was observed in either Ect2 was detected as a faint band after a long exposure; A549 Ect2 KD or NT tumors (Figure 7a), consistent however, Flag-Par6a efficiently co-precipitated with with our tumor terminal deoxynucleotidyltransferase- PKCi in MDCK/wtPar6 cells as expected (Figure 8a, mediated dUTP nick-end labeling (TUNEL) staining. PKCi IP). Similarly, when Flag-Par6a was immuno-

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3602

Figure 6 Rac1 is an effector of epithelial cell transforming sequence 2 (Ect2)-mediated transformation in non-small cell lung cancer cells. (a) Ect2 knockdown (KD) inhibits Rac1, but not Cdc42 or RhoA activity. Representative immunoblots of active and total Rac1, Cdc42 and RhoA in H1703 non-target (NT) and Ect2 KD cells. (b) Quantitative analysis of Rac1, Cdc42 and RhoA activity in H1703 NT and Ect2 KD cells. Data are expressed as % NT control ±s.e.m. (*Po0.05, n ¼ 3). (c) Expression of constitutively active Rac1 (RacV12) in H1703/NT and H1703/Ect2 KD cells. (d) Reconstitution of anchorage-independent soft agar growth in H1703/Ect2 KD cells by a RacV12 (n ¼ 5). (e) Reconstitution of cellular invasion in H1703/Ect2 KD cells by RacV12 (n ¼ 4). Data in d and e are expressed as % NT control ±s.e.m. *indicates a statistically significant difference from NT control (Po0.05) and **indicates a statistically significant difference from Ect2 KD in the absence of RacV12 (Po0.05).

precipitated from MDCK/wtPar6 cells, abundant preferentially binds to the PKCi–Par6a complex when PKCi, but very little Ect2, was detected in the resultant compared with PKCi or Par6a alone. Similar results precipitates (Figure 8a, Flag IP). In contrast, when were obtained in A549 cells (data not shown). PKCi was immunoprecipitated from H1703/wtPar6 cells, both endogenous Ect2 and Flag-Par6a were easily detected in the resultant precipitates (Figure 8b, PKCi Cytoplasmic localization of Ect2 in NSCLC cells IP). Likewise, both Ect2 and PKCi efficiently co- is regulated by PKCi and Par6a precipitated with Flag-Par6a in H1703/wtPar6 cells As Ect2 is mislocalized to the cytoplasm of NSCLC (Figure 8b, Flag IP). Thus, a significant amount of tumors (Figure 2), we examined Ect2 localization by Ect2 associates with PKCi–Par6a in NSCLC cells, confocal immunofluorescence microscopy (Figure 9). whereas only a small amount of Ect2 is found in Ect2 is almost exclusively nuclear in MDCK cells association with PKCi–Par6a in MDCK cells. (Figure 9a). In contrast, Ect2 localizes to both the To further determine whether Ect2 binds the PKCi– nucleus and cytoplasm of H1703/NT cells (Figure 9b). Par6a complex, H1703/PKCi KD cells were transfected Interestingly, both H1703/PKCi KD and H1703/Par6a with either Flag-tagged wild-type PKCi, a PKCi-D63A KD cells showed significantly less cytoplasmic Ect2 mutant, which does not bind Par6a (Frederick et al., staining than H1703/NT cells (Figures 9c and d), 2008), or control plasmid (Vector) (Figure 8c). Immuno- indicating that PKCi and Par6a regulate cytoplasmic blot analysis confirmed expression of Ect2, Flag-wild- Ect2 localization. QPCR confirmed efficient PKCi and type PKCi and Flag-PKCi-D63A in these cells Par6a KD respectively in these cells (Supplementary (Figure 8c, lysate). Immunoprecipitation of Flag-PKCi Figure 3). A similar change in Ect2 localization from revealed that Ect2 associates with wtPKCi, but not cytoplasm to nucleus was observed in A549/PKCi KD PKCi-D63A (Figure 8c, Flag IP). A complementary cells (Supplementary Figure 4). experiment was conducted in which H1703/Par6a KD The effects of PKCi KD and Par6a KD on Ect2 cells were transfected with either wild-type Par6a localization were confirmed by cellular fractionation or a Par6a-K19A mutant, which does not bind PKCi (Figure 10). Ect2 is predominantly nuclear in MDCK (Frederick et al., 2008) (Figure 8d). Immunoblot cells (Figure 10a). In contrast, H1703/NT cells express analysis confirmed the expression of Ect2, PKCi and equal amounts of Ect2 in nuclear and cytoplasmic the appropriate Flag-Par6a alleles in these cells fractions, consistent with our immunofluorescence (Figure 8d, Lysate). Both PKCi and Ect2 co-immuno- results (Figure 10b). H1703/PKCi KD and H1703/ precipitated with wild-type Par6a, but neither PKCi nor Par6a KD cells exhibit an increase in nuclear Ect2 Ect2 immunoprecipitated with Flag-Par6a-K19A and a concomitant decrease in the cytoplasmic Ect2 (Figure 8d, Flag IP). These results indicate that Ect2 (Figure 10b). Total Ect2 in H1703/PKCi KD and

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3603

Figure 7 Epithelial cell transforming sequence 2 (Ect2) activates the Pak-Mek1,2-Erk1,2 signaling pathway in vivo.(a) Immunoblot analysis of A549/non-target (NT) and A549/Ect2 knockdown (KD) tumors for Ect2, phospho-298-Mek1,2, total Mek1,2, phospho- Erk1,2, total Erk1,2, actin, and intact and cleaved PARP. Lysates from Hela cells treated with taxol for 48 h were used as positive control for PARP cleavage. (b–d) Quantitative analysis demonstrates a statistically significant decrease in: (b) Ect2 protein expression (n ¼ 4/group; *Po0.0001), (c) phospho-298-Mek1,2/total Mek1,2 ratio (n ¼ 4/group; *Po0.004), and (d) phospho-Erk1,2/total Erk1,2 ratio (n ¼ 4/group; *Po0.03) (d) in A549/Ect2 KD tumors when compared with A549/NT tumors.

H1703/Par6a KD cells was comparable with H1703/NT and protein has been reported in human gliomas, lung, cells, indicating that PKCi KD and Par6a KD primarily esophageal and pancreatic cancers (Takai et al., 1995; affect Ect2 localization, and not expression. The purity Sano et al., 2006; Zhang et al., 2008; Hirata et al., 2009). of cellular fractions was confirmed using nuclear (lamins Here, we show that ECT2 and PRKCI are co-amplified A/C) and cytoplasmic (Mek1,2) markers (Figure 10a as part of the 3q26 amplicon, leading to coordinate and b). As expected, PKCi is predominantly cytoplas- overexpression of Ect2 and PKCi in LSCC tumors. Our mic in MDCK and H1703 cells and was efficiently data provide the first evidence that Ect2 and PKCi are knocked down in H1703/PKCi KD cells (Figures 10a genetically linked in human cancer, and reveal that Ect2 and b). Quantitative analysis of three independent is a relevant target of the 3q26 amplicon. experiments confirmed that Ect2 is largely nuclear in We also show that Ect2 is functionally important in MDCK cells but is distributed equally between the anchorage-independent growth and invasion of NSCLC nucleus and cytoplasm of H1703/NT cells, and that cells in vitro and tumorigenicity in vivo. Our data are H1703/PKCi KD and H1703/Par6a KD cells exhibit a consistent with several recent reports implicating Ect2 in significant redistribution of Ect2 from the cytoplasm to the growth and invasion of cancer cells in vitro (Salhia the nucleus (Figure 10c). et al., 2008; Hirata et al., 2009). Our results provide important new mechanistic insight into how Ect2 participates in transformation and extend previous findings to the in vivo setting. Ect2 KD blocks Discussion tumorigenicity by inhibiting tumor cell proliferation, without affecting tumor apoptosis or vascularization. Rho family GTPases are regulated by GEFs. As Rac1 Furthermore, we show that Ect2 selectively regulates activity is high in NSCLC cells as a result of the PKCi– Rac1 activity in NSCLC cells, and Rac1 is a critical Par6a complex (Regala et al., 2005a; Frederick et al., downstream effector of Ect2-dependent transformation 2008), we hypothesized that a Rac1-GEF may bind the in vitro. Finally, Ect2 drives tumor cell proliferation by PKCi–Par6a complex to regulate Rac1 activity in activating a Rac1-Pak-Mek1,2-Erk1,2 signaling axis NSCLC tumors. The Ect2 GEF is an attractive in vivo. Further studies will be required to dissect the candidate for such a role as the Ect2 gene, ECT2,is signaling pathway(s) involved in Ect2-, Rac1-dependent located at chromosome 3q26, and elevated Ect2 mRNA invasion.

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3604

Figure 8 Epithelial cell transforming sequence 2 (Ect2) associates with the protein kinase Ci (PKCi)–Par6a complex in non-small cell Figure 9 Epithelial cell transforming sequence 2 (Ect2) is lung cancer cells. (a) Madin-Darby canine kidney (MDCK) cells mislocalized to the cytoplasm of non-small cell lung cancer cells. were stably transfected with wild-type human Flag-tagged Par6a Madin-Darby canine kidney (MDCK) cells, H1703/non-target (wtPar6) or control empty vector (Vector). Total cell lysates (NT), H1703/protein kinase Ci (PKCi) knockdown (KD) and (Lysate) confirm expression of endogenous Ect2, PKCi and Flag- H1703/Par6a KD cells were fixed, stained with 40,6 diamidino-2- Par6a. Immunoprecipitation of PKCi co-precipitates Flag-Par6a phenylindole (DAPI) and Ect2 antibody and analysed by confocal and a small amount of endogenous Ect2 from MDCK/wtPar6 cells immunofluorescence microscopy as described in Materials and (PKCi IP). Immunoprecipitation of Flag-Par6a co-precipitates methods. (a) Ect2 localizes almost exclusively to the nucleus of PKCi and a small amount of Ect2 from MDCK/wtPar6 cells (Flag MDCK cells. (b) Ect2 localizes to both the nucleus and cytoplasm IP). (b) H1703/pBabe (Vector) and H1703/wtPar6 (wtPar6) cells of H1703/NT cells. In contrast, Ect2 is largely localized to the were subjected to immunoprecipitation analysis as described in (a). nucleus of H1703/PKCi KD (c) and H1703/Par6a KD (d) cells. Endogenous Ect2 co-precipitates with both PKCi and Par6a in these cells. (c) H1703/PKCi knockdown (KD) cells were trans- fected with wtPKCi, a PKCi-D69A mutant or empty plasmid as described in Materials and methods. Total cell lysates confirm defect (or change in PDT) in any of the five NSCLC expression of Ect2 and Flag-PKCi alleles (Lysates). Immunopre- cipitation using Flag antibody co-precipitates Ect2 with wtPKCi cell lines tested argues for the existence of an Ect2- but not PKCi-D63A (Flag IP). (d) H1703/Par6a KD cells were independent cytokinesis mechanism in NSCLC cells. transfected with Flag-tagged wtPar6a, Par6a-K19A or empty Our finding that Ect2 KD in NSCLC cells does not control plasmid (Vector). Cell lysates confirm expression of Ect2, inhibit RhoA, the well-characterized target of Ect2 in PKCi and Flag-Par6a alleles (Lysates). Immunoprecipitation using Flag antibody co-precipitates Ect2 and PKCi with wtPar6a but not cytokinesis (Tatsumoto et al., 1999), further suggests the Par6a-K19A. existence of such an alternative cytokinesis mechanism. In either case, our analysis reveals a novel Ect2 function in transformation that is distinct from cytokinesis. Ect2 plays a physiologic role in cytokinesis, suggesting Ect2 localizes to the nucleus of MDCK and normal a possible mechanism by which Ect2 KD could block lung epithelial cells. In contrast, Ect2 is overexpressed NSCLC cell proliferation. Surprisingly however, and mislocalized to the cytoplasm of primary NSCLC although Ect2 KD leads to a severe cytokinesis defect tumor cells and NSCLC cell lines. The cytoplasmic in MDCK cells, no such defect occurs in NSCLC cells mislocalization of Ect2 is reminiscent of the previous in vitro, or A549 tumors in vivo. Our data differ from a observations correlating cytoplasmic mislocalization of recent report that transient Ect2 KD in A549 cells leads Ect2 fragments with transformation. When expressed in to a transitory inhibition of cell cycle (Hirata et al., mouse fibroblasts, the C-terminal Ect2 GEF domain 2009). This discrepancy may result from the use of localized to the cytoplasm, exhibited constitutive GEF transient Ect2 KD by short interfering RNA as opposed activity and caused cellular transformation (Miki et al., to stable Ect2 KD by lentiviral short hairpin RNA. Our 1993; Saito et al., 2004; Solski et al., 2004). In contrast, stably transduced Ect2 KD cells may express sufficient full-length Ect2 localized to the nucleus and exhibited no Ect2 to remain above a critical threshold required for transforming activity (Saito et al., 2004; Solski et al., cytokinesis. Alternatively, NSCLC cells may use a 2004). recently described alternative cytokinesis mechanism Our data show that full-length, cytoplasmic Ect2 that makes fibrosarcoma cells resistant to Ect2 KD- binds the oncogenic PKCi–Par6a complex in NSCLC induced cytokinesis defects (Kanada et al., 2008). Our cells. Ect2 binds more efficiently to the PKCi–Par6a finding that Ect2 KD does not induce a cytokinesis complex than PKCi or Par6a alone, indicating that both

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3605 molecular mechanisms that regulate the formation, dynamics and activity of the oncogenic PKCi–Par6a– Ect2 complex in NSCLC cells. In conclusion, our data reveal a novel genetic, biochemical and functional link between PKCi and Ect2 in NSCLC. PKCi and Ect2 are coordinately amplified and overexpressed in NSCLC as part of the 3q26 amplicon. Ect2 has a requisite role in the anchorage-independent growth and invasion of NSCLC cells in vitro and tumorigenicity in vivo. Ect2 is mislocalized to the cytoplasm of NSCLC cells as a consequence of its interaction with the oncogenic PKCi– Par6a complex, and the PKCi–Par6a–Ect2 complex drives transformed growth through activation of a Rac1-Pak-Mek1,2-Erk1,2 signaling axis. Our data are the first to identify Ect2 as a relevant target of tumor- specific 3q26 amplification and to elucidate a molecular mechanism by which Ect2 participates in the trans- formed phenotype of human tumor cells. Furthermore, our data provide new mechanistic insight into previous observations regarding the cytoplasmic localization of Ect2 fragments and transforming potential. Finally, our data reveal a novel paradigm in which two oncogenes within a single amplicon are genetically, biochemically Figure 10 PKCi and Par6a regulate the cytoplasmic distribution of epithelial cell transforming sequence 2 (Ect2) in non-small cell and functionally linked to cellular transformation. Ect2 lung cancer cells. Madin-Darby canine kidney (MDCK) cells (a), and PKCi may be similarly linked in other tumors that H1703/non-target (NT), H1703/protein kinase Ci (PKCi) knock- harbor 3q26 amplification, such as SCC of the head and down (KD) and H1703/Par6a KD (b) cells were fractionated into neck (Snaddon et al., 2001), esophagus (Imoto et al., cytoplasmic and nuclear fractions. Total cell lysates, nuclear and cytoplasmic fractions from an equivalent cell number were 2001), cervix (Sugita et al., 2000) and ovary (Sonoda subjected to immunoblot analysis for Ect2, PKCi, lamins A/C et al., 1997 and Sugita et al., 2000). and Mek1. Lamins A/C served as a marker of nuclei and Mek1 as a marker of cytoplasm. (c) Quantitative analysis of nuclear and cytoplasmic Ect2 in MDCK, H1703/NT, H1703/PKCi KD and H1703/Par6a KD cells. Immunoblots from three independent Materials and methods fractionation experiments were analysed by densitometry for Ect2. Results are plotted as % total Ect2 in the nucleus (N) and Antibody reagents and cell lines cytoplasm (C). *, **,#and ## indicate a statistically significant Antibodies used are: Ect2, RhoA and CD31/Pecam-1 (Santa difference in the amount of nuclear (*and #) and cytoplasmic Cruz Biotechnology, Santa Cruz, CA, USA); PKCi, Rac1 and (**and ##) Ect2 in the indicated cell line when compared with Cdc42 (BD Transduction Labs, San Jose, CA, USA); b-actin, H1703/NT cells (*Po0.02), (**Po0.03), (#Po0.0005) and ## PARP/cleaved PARP, MEK1/2, p298-MEK1/2, ERK, ( Po0.01); n ¼ 3. pThr202/Tyr204ERK and Lamin A/C (Cell Signaling, Dan- vers, MA, USA); FLAG epitope (Sigma-Aldrich, St Louis, MO, USA); BrdU (DAKO, Carpenteria, CA, USA). Phalloi- din-Alexa Fluor 594 and anti-Rabbit Alexa Fluor 594 were PKCi and Par6a contribute to Ect2 binding. PKCi or from Invitrogen (Carlsbad, CA, USA), and TUNEL staining Par6a KD blocks NSCLC cell transformation (Freder- kit from Promega BioSciences (San Luis Obispo, CA, USA). ick et al., 2008) and leads to redistribution of Ect2 to the A427, A549, H358, H460, H1703 and MDCK cells were from nucleus. These data argue that the PKCi–Par6a com- the American Type Culture Collection (Manassas, VA, USA). plex regulates the cytoplasmic localization and onco- genic potential of Ect2. Our data are interesting in light Human primary NSCLC sample analysis of previous reports that Ect2 can bind the atypical PKC- Total RNA was isolated from primary NSCLC tumors and Par6-Par3 polarity complex at cell–cell junctions in matched normal lung using RNAqueous 4PCR (Ambion, MDCK cells (Liu et al., 2004, 2006). Our data confirm Austin, TX, USA). Ect2 and PKCi mRNA abundance was that a very small amount of Ect2 can be found in determined by Taqman qPCR on an Applied Biosystems association with PKCi and Par6a in MDCK cells; 7900HT sequence analyzer (Applied Biosystems, Foster City, furthermore, we detect very faint Ect2 staining at cell– CA, USA) using 18S for normalization. Primer/probe sets are cell junctions in MDCK cells (data not shown). In listed in Supplementary Figure 1a. Genomic DNA was NSCLC cells, which have lost cellular polarity, Ect2 isolated from primary NSCLC tumors and matched normal lung by phenol/chloroform extraction and ECT2 and PRKCI exhibits punctate staining throughout the cytoplasm, gene copy number was determined by qPCR using the and efficiently binds to the oncogenic PKCi–Par6a RNASEP1 gene for normalization. Primer/probe sets are complex to activate Rac1 and drive transformation. listed in Supplementary Figure 1a. Statistical analyses were Further studies will be required to determine the basis carried out using the two-tailed Student’s t-test and R2 using for the punctate Ect2 staining pattern, and the Pearson correlation analysis.

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3606 Tissue microarrays of primary NSCLC and matched normal IN, USA) as previously described (Regala et al., 2005a). lung were analysed for Ect2 protein by immunohistochemistry Tumor volume, BrdU labeling, immunohistochemical and using the Envision Plus Dual Labeled Polymer Kit (DAKO). immunoblot analyses were carried out as previously described Specificity was determined by incubating primary Ect2 anti- (Regala et al., 2005a). Ect2, Mek1,2, p298-Mek1,2, Erk1,2, p- body with a 10-fold excess of Ect2 peptide. Images were Erk1,2, PARP/cleaved PARP and b-actin antibodies were used captured and analysed using an Aperio ScanScope scanner and at 1:1000. Images were captured and analysed using Aperio ImageScope software (Aperio Technologies, Vista, CA, USA). ScanScope scanner and software (Aperio Technologies).

Lentiviral RNAi constructs, cell transduction and immunoblot Confocal immunofluorescence microscopy analysis Cells grown on glass chamber slides (Nalge Nunc Interna- Lentiviral RNAi against human Ect2, PKCi and Par6a were tional, Naperville, IL, USA) were fixed in 4% paraformalde- obtained from Sigma-Aldrich Mission short hairpin RNA hyde for 30 min, permeabilized in 0.2% (v/v) Triton X-100, library and packaged into recombinant lentiviruses as pre- incubated in phosphate-buffered saline containing 3% bovine viously described (Frederick et al., 2008). A non-target serum albumin for 30 min and incubated with phalloidin lentiviral RNAi (NT-RNAi) that does not recognize any antibody conjugated to Alexa Fluor 594 (1:500) for 2 h at human or canine genes was used as a negative control. RNAi room temperature. Slides were mounted in medium containing target sequences are listed in Supplementary Figure 1b. Stably 1.5 mg/ml 40,6 diamidino-2-phenylindole (DAPI) for confocal transduced cell populations were generated as previously fluorescence microscopy (Vectashield: Vector Laboratories, described (Frederick et al., 2008). Burlingame, CA, USA). Ten representative fields were photo- Epithelial cell transforming sequence 2, PKCi and Par6a graphed at  40, and >1000 cells from each cell type were RNAi constructs were analysed for efficiency of target gene analysed for multiple nuclei. Experiments were independently KD by qPCR (Supplementary Figure 1a). NT, Ect2 and PKCi repeated three times. KD cells were subjected to immunoblot analysis as previously Cells grown on glass chamber slides (Nalge Nunc Interna- described (Regala et al., 2005a). tional), were fixed, permeabilized and incubated overnight at 4 1C with Ect2 antibody (1:500), followed by incubation with Plasmids, transfections and immunoprecipitations anti-rabbit secondary antibody conjugated to Alexa Fluor 594 Cells were transfected with pEGFP-C1 containing RNAi- (1:500) for 2 h at room temperature. Cells were mounted in resistant, full-length human Ect2 or empty pEGFP-C1 plasmid medium containing DAPI and subjected to confocal fluores- (Clonotech, Mountain View, CA, USA) using Lipofectamine cence microscopy (Vectashield: Vector Laboratories). Cells 2000 (Invitrogen). The Ect2 cDNA was rendered RNAi- were photographed with a Zeiss LSM 510 meta confocal resistant by introducing silent mutations that disrupt the Ect2- microscope and imaging software (Carl Zeiss, Inc., Maple RNAi target site as previously described (Frederick et al., Grove, MN, USA), using a EC Plan-Neofluar  40 oil 2008). In some experiments, cells were stably transfected with immersion objective (1.3 numerical aperture). DAPI was LZRS containing Myc-tagged RacV12 or empty LZRS retro- detected with a Diode 405–430 laser at 405 nm using a viral vector as previously described (Frederick et al., 2008). 420 nm long-pass filter. Ect2 was detected with a HeNe laser Cells stably transfected with Flag-tagged full-length, RNAi- at 594 nm using a 615 nm long-pass filter. resistant wild-type human Par6a, K19A Par6a mutant (Par6a- K19A), wild-type PKCi, D63A PKCi mutant (PKCi-D63A) or empty vector were subjected to immunoprecipitation and Subcellular fractionation immunoblot analysis as previously described (Frederick et al., Cells were fractionated using NE-PER Nuclear and Cytoplas- 2008). mic Extraction Reagents (Pierce Biotechnology, Rockford, IL, USA). Total, nuclear and cytoplasmic fractions were subjected Soft agar growth, cellular invasion and PDT analysis to immunoblot analysis using antibodies to Ect2, Lamin A/C, Anchorage-independent growth and invasion were assayed as Mek1/2 and PKCi (all at 1:1000), HRP-labeled secondary previously described (Regala et al., 2005a; Frederick et al., antibodies and ECL detection. Ect2 levels were determined by 2008). Matrigel-coated plates were obtained from BD Bios- densitometry and expressed as % total cellular Ect2. The ciences (San Jose, CA, USA). PDT was determined during experiments were independently conducted three times. logarithmic growth phase in six-well tissue culture plates (Corning Costar, Corning, NY, USA) (2  104 cells per well) using the formula PDT ¼ ln2t/ln(N/No) (N ¼ final cell num- ber, No ¼ initial cell number and t ¼ time between No and N). Conflict of interest Experiments were independently carried out at least three times. The authors declare no conflict of interest.

Rac1, Cdc42 and RhoA activity assays Acknowledgements Rho GTPase activity was determined by affinity pulldown as previously described (Zhang et al., 2004; Frederick et al., We acknowledge Dr Roderick P Regala for assistance with the 2008). Densitometry was performed using Kodak Molecular ectopic tumor studies, Dr Andras Khoor and Capella Weems imaging software (Carestream Molecular Imaging, New for analysis of primary NSCLC tumors, Pam Kreinest and Haven, CT, USA). Results were normalized to total Rho Brandy Edenfield for immunohistochemistry, Dr Lee Jamieson GTPase. and Alyssa Kunz for technical assistance, and Drs E Aubrey Thompson and Nicole R Murray for critical review of the Tumorigenicity in nude mice paper. This work was supported in part by grants from the A549 NT and Ect2 KD cells were grown subcutaneously in National Institutes of Health (CA081436), The V Foundation athymic nude mice (Harlan-Sprague-Dawley, Indianapolis, for Cancer Research and The Mayo Foundation to APF.

Oncogene Ect2 and PKCi are genetically and functionally linked in NSCLC V Justilien and AP Fields 3607 References

Balsara BR, Sonoda G, du Manoir S, Siegfried JM, Gabrielson E, Regala RP, Weems C, Jamieson L, Khoor A, Edell ES, Lohse CM Testa JR. (1997). Comparative genomic hybridization analysis et al. (2005b). Atypical protein kinase C iota is an oncogene in detects frequent, often high-level, overrepresentation of DNA human non-small cell lung cancer. Cancer Res 65: 8905–8911. sequences at 3q, 5p, 7p, and 8q in human non-small cell lung Saito S, Liu XF, Kamijo K, Raziuddin R, Tatsumoto T, Okamoto I carcinomas. Cancer Res 57: 2116–2120. et al. (2004). Deregulation and mislocalization of the cytokinesis Brass N, Ukena I, Remberger K, Mack U, Sybrecht GW, Meese EU. regulator ECT2 activate the Rho signaling pathways leading to (1996). DNA amplification on chromosome 3q26.1-q26.3 in malignant transformation. J Biol Chem 279: 7169–7179. squamous cell carcinoma of the lung detected by reverse chromo- Saito S, Tatsumoto T, Lorenzi MV, Chedid M, Kapoor V, some painting. Eur J Cancer 32A: 1205–1208. Sakata H et al. (2003). Rho exchange factor ECT2 is induced Frederick LA, Matthews JA, Jamieson L, Justilien V, Thompson EA, by growth factors and regulates cytokinesis through the Radisky DC et al. (2008). Matrix metalloproteinase-10 is a critical N-terminal cell cycle regulator-related domains. J Cell Biochem effector of protein kinase Ciota-Par6alpha-mediated lung cancer. 90: 819–836. Oncogene. Salhia B, Tran NL, Chan A, Wolf A, Nakada M, Rutka F et al. Hara T, Abe M, Inoue H, Yu LR, Veenstra TD, Kang YH et al. (2008). The guanine nucleotide exchange factors trio, Ect2, and (2006). Cytokinesis regulator ECT2 changes its conformation Vav3 mediate the invasive behavior of glioblastoma. Am J Pathol through phosphorylation at Thr-341 in G2/M phase. Oncogene 25: 173: 1828–1838. 566–578. Sano M, Genkai N, Yajima N, Tsuchiya N, Homma J, Tanaka R et al. Hirata D, Yamabuki T, Miki D, Ito T, Tsuchiya E, Fujita M et al. (2006). Expression level of ECT2 proto-oncogene correlates with (2009). Involvement of epithelial cell transforming sequence-2 prognosis in glioma patients. Oncol Rep 16: 1093–1098. oncoantigen in lung and esophageal cancer progression. Clin Cancer Snaddon J, Parkinson EK, Craft JA, Bartholomew C, Fulton R. Res 15: 256–266. (2001). Detection of functional PTEN lipid phosphatase protein Imoto I, Pimkhaokham A, Fukuda Y, Yang ZQ, Shimada Y, Nomura and enzyme activity in squamous cell carcinomas of the head and N et al. (2001). SNO is a probable target for gene amplification at neck, despite loss of heterozygosity at this locus. Br J Cancer 84: 3q26 in squamous-cell carcinomas of the esophagus. Biochem 1630–1634. Biophys Res Commun 286: 559–565. Solski PA, Wilder RS, Rossman KL, Sondek J, Cox AD, Campbell SL Kanada M, Nagasaki A, Uyeda TQ. (2008). Novel functions of Ect2 in et al. (2004). Requirement for C-terminal sequences in regulation of polar Lamellipodia formation and polarity maintenance during Ect2 guanine nucleotide exchange specificity and transformation. ‘contractile ring-Independent’ cytokinesis in adherent cells. Mol Biol J Biol Chem 279: 25226–25233. Cell 19: 8–16. Sonoda G, Palazzo J, du Manoir S, Godwin AK, Feder M, Yakushiji Kim JE, Billadeau DD, Chen J. (2005). The tandem BRCT domains of M et al. (1997). Comparative genomic hybridization detects Ect2 are required for both negative and positive regulation of Ect2 frequent overrepresentation of chromosomal material from 3q26, in cytokinesis. J Biol Chem 280: 5733–5739. 8q24, and 20q13 in human ovarian carcinomas. Genes Liu XF, Ishida H, Raziuddin R, Miki T. (2004). Nucleotide exchange Cancer 20: 320–328. factor ECT2 interacts with the polarity protein complex Par6/Par3/ Sugita M, Tanaka N, Davidson S, Sekiya S, Varella-Garcia M, West J protein kinase Czeta (PKCzeta) and regulates PKCzeta activity. et al. (2000). Molecular definition of a small amplification domain Mol Cell Biol 24: 6665–6675. within 3q26 in tumors of cervix, ovary, and lung. Cancer Genet Liu XF, Ohno S, Miki T. (2006). Nucleotide exchange factor ECT2 Cytogenet 117: 9–18. regulates epithelial cell polarity. Cell Signal 18: 1604–1615. Takai S, Long JE, Yamada K, Miki T. (1995). Chromosomal locali- Miki T, Smith CL, Long JE, Eva A, Fleming TP. (1993). Oncogene zation of the human ECT2 proto-oncogene to 3q26.1–>q26.2 by ect2 is related to regulators of small GTP-binding . Nature somatic cell analysis and fluorescence in situ hybridization. 362: 462–465. Genomics 27: 220–222. Niiya F, Tatsumoto T, Lee KS, Miki T. (2006). Phosphorylation of the Tatsumoto T, Xie X, Blumenthal R, Okamoto I, Miki T. (1999). cytokinesis regulator ECT2 at G2/M phase stimulates association of Human ECT2 is an exchange factor for Rho GTPases, phosphory- the mitotic kinase Plk1 and accumulation of GTP-bound RhoA. lated in G2/M phases, and involved in cytokinesis. J Cell Biol 147: Oncogene 25: 827–837. 921–928. Niiya F, Xie X, Lee KS, Inoue H, Miki T. (2005). Inhibition of cyclin- Zhang J, Anastasiadis PZ, Liu Y, Thompson EA, Fields AP. (2004). dependent kinase 1 induces cytokinesis without chromosome Protein kinase C bII induces cell invasion through a Ras/MEK-, segregation in an ECT2 and MgcRacGAP-dependent manner. PKCiota/RAC 1-dependent signaling pathway. J Biol Chem 279: J Biol Chem 280: 36502–36509. 22118–22123. Regala RP, Weems C, Jamieson L, Copland JA, Thompson EA, Fields Zhang ML, Lu S, Zhou L, Zheng SS. (2008). Correlation AP. (2005a). Atypical protein kinase Ciota plays a critical role in between ECT2 and methylation change of ECT2 human lung cancer cell growth and tumorigenicity. J Biol Chem 280: promoter region in pancreatic cancer. Hepatobiliary Pancreat 31109–31115. Dis Int 7: 533–538.

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

Oncogene