Oncogene (2009) 28, 2466–2475 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE C delta 1 is a novel 3p22.3 tumor suppressor involved in cytoskeleton organization, with its epigenetic silencing correlated with high-stage gastric cancer

X-T Hu1, F-B Zhang1, Y-C Fan2, X-S Shu2, AHY Wong2, W Zhou3, Q-L Shi1, H-M Tang1,LFu4, X-Y Guan4, SY Rha5, Q Tao2 and C He1,3

1Biomedical Research Center, Sir Run Run Shaw Hospital, Zhejiang University and Key Laboratory of Biotherapy of Zhejiang Province, Hangzhou, China; 2Cancer Epigenetics Laboratory, State Key Laboratory in Oncology in South China, Department of Clinical Oncology, Sir YK Pao Center for Cancer, Hong Kong Cancer Institute and Li Ka Shing Institute of Health Sciences, Chinese University of Hong Kong, Hong Kong; 3Department of Colorectal Surgery, Sir Run Run Shaw Hospital, Zhejiang University, Hangzhou, China; 4Department of Clinical Oncology, University of Hong Kong, Hong Kong and 5Yonsei Cancer Center, Yonsei University College of Medicine, Seoul, Korea

Located at the important tumor suppressor , 3p22, Introduction PLCD1 encodes an that mediates regulatory signaling of energymetabolism, calcium homeostasis and Gastric cancer is the second most frequent malignancy intracellular movements. We identified PLCD1 as a among all tumors worldwide(Parkin et al., 1993), the downregulated in aerodigestive carcinomas through leading cause of cancer mortality in many developing expression profiling and epigenetic characterization. We nations including China (Fuchs and Mayer, 1995). The found that PLCD1 was expressed in all normal adult main cause of recurrence after curative resection of tissues but low or silenced in 84% (16/19) gastric cancer advanced tumors is peritoneal metastasis, which is also cell lines, well correlated with its CpG island (CGI) the most common reason for non-curative surgery methylation status. Methylation was further detected in (Maehara et al., 2000; Kaminishi, 2005). Its molecular 62% (61/98) gastric primarytumors, but none of normal pathogenesis is believed to be a multi-step process gastric mucosa tissues. PLCD1 methylation was signifi- involving the deregulation of multiple cancer cantlycorrelated with tumor high stage. Detailed including the inactivation of tumor suppressor genes methylation analysis of 37 CpG sites at the PLCD1 (TSGs) (Panani, 2008). CGI bybisulfite genomic sequencing confirmed its 3p21–22 is an important chromosomal region, fre- methylation. PLCD1 silencing could be reversed by quently deleted in multiple tumors including gastric pharmacological demethylation with 5-aza-20-deoxycyti- cancer (Hesson et al., 2007). It is believed that multiple dine, indicating a direct epigenetic silencing. Ectopic candidate TSGs are located in this region. We have been expression of PLCD1 in silenced gastric tumor cells searching for candidate TSGs at this locus in aero- dramaticallyinhibited their clonogenicityand migration, digestive tumors through massive expression profiling possiblythrough downregulating MMP7 expression and and epigenetic characterization, and identified several hampering the reorganization of cytoskeleton through interesting targets (Ying and Tao, manuscript in cofilin inactivation byphosphorylation. Thus, epigenetic preparation), including BLU (Qiu et al., 2004) and the inactivation of PLCD1 is common and tumor-specific in two closely located 3p22.3 genes, DLEC1 (Qiu et al., gastric cancer, and PLCD1 acts as a functional tumor 2008) and PLCD1. Recently, PLCD1 was also identified suppressor involved in gastric carcinogenesis. as a novel TSG in esophageal squamous cell carcinoma Oncogene (2009) 28, 2466–2475; doi:10.1038/onc.2009.92; with its downregulation associated with loss of published online 18 May 2009 heterozygosity and promoter methylation, playing an important suppressive role in esophageal squamous Keywords: PhospholipaseC delta1; tumor suppressor cell carcinoma development and progression (Fu et al., gene; methylation; 3p22; gastric cancer 2007). (PLC) is one of the key in the phosphoinositide metabolism system and hydrolyses phosphatidylinositol 4,5-bisphosphate(PI(4,5)P2) to Correspondence: Professor C He, Biomedical Research Center, Sir generate two second messengers, inositol 1,4,5-trispho- Run Run Shaw Hospital, Zhejiang University, Hangzhou, Zhejiang sphate and diacylglycerol. Diacylglycerol mediates the 310016, China. activation of protein kinase C and inositol 1,4,5- E-mail: [email protected] or Professor Q Tao, Rm 315, Cancer trisphosphate triggers the release of Ca2 þ from intra- Center, PWH, Chinese University of Hong Kong, Shatin, Hong Kong. cellular stores. Phosphoinositide metabolism is an E-mail: [email protected] Received 29 January 2009; revised 6 March 2009; accepted 12 March important intracellular signaling system involved in a 2009; published online 18 May 2009 variety of cellular functions, including secretion of PLCD1 methylation in gastric cancer X-T Hu et al 2467 hormones, transduction of neurotransmitters, growth We further investigated PLCD1 expression in normal factor signaling, membrane trafficking and regulation of and gastric cancer tissues by immunohistochemistry. the cytoskeleton (Janetopoulos and Devreotes, 2006; PLCD1 was highly expressed in the cytoplasm of Santarius et al., 2006; Carvou et al., 2007). To date, 13 epithelial and some stromal cells in the normal gastric PLC isozymes have been identified and categorized into mucosa (Figure2). In contrast, PLCD1 expression was six classes, b, g, d, e, z and Z-type, on the basis of obviously decreased in primary gastric carcinomas. Of structure and regulatory activation mechanisms (Rhee, 60 tumor cases examined, 44 (73%) had no or weak 2001; Fukami, 2002; Nakahara et al., 2005). Among the expression compared with the normal gastric mucosa PLC family members, PLCD is considered to be the (Supplementary Table 2). We further studied whether basic isoform and composed of three isozymes, PLCD1, there is any correlation between PLCD1 methylation D3 and D4 (Essen et al., 1996; Irino et al., 2004). status and its expression. Among 38 gastric carcinoma In this study, we fully characterized the epigenetic samples with PLCD1 methylation detected by MSP, alteration and functions of PLCD1 in gastric cancer. only five(13%) showed PLCD1 expression, whereas Our results showed that PLCD1 acts as a functional 11/22 (50%) cases without methylation exhibited TSG in this tumor, and frequent methylation of PLCD1 expression (P ¼ 0.005, Fisher’s exact test) (Supplemen- is significantly correlated with the high-stage gastric tary Table2). cancer.

PLCD1 inhibited the growth of gastric tumor cells in vitro Results To investigate the effects of ectopic PLCD1 expression on thegrowth of gastric tumor cells, PLCD1 expression Frequent downregulation and methylation of PLCD1 vector was transfected into gastric tumor cells with in gastric cancer cell lines complete methylation and silencing (MKN-45 and First, we examined PLCD1 expression in a panel of SGC-7901). The colony formation efficiencies of each human normal adult tissues by semi-quantitative RT– transfected cell line were evaluated by soft agar PLCD1 PCR. Our results showed that was expressed (Figure3a) and monolayercultureassays (Figure3b). in all normal tissues including digestive tissues like Though, ectopic expression of PLCD1 did not affect cell stomach, though at different expression levels proliferation and cell-cycle arrest (data not shown), it (Figure1a). significantly suppressed the colony-formation efficiency We then determined PLCD1 expression in gastric of PLCD1-expressing tumor cells as compared with the cancer cell lines. Results indicated that PLCD1 expres- vector control-transfected cells (P 0.05, Figures 3a and sion was silenced or dramatically reduced in most cell o b), indicating that PLCD1 indeed possessed growth PLCD1 lines (Figure 1b). On the basis of the CpG island inhibitory activities in gastric tumor cells. (CGI) sequence, we designed methylation-specific PCR (MSP) primers to analyse its methylation status. The PLCD1 CGI was methylated in all cell lines with PLCD1 suppressed the migration of gastric tumor cells silenced or reduced PLCD1 expression (Figure 1b). To evaluate the effect of PLCD1 expression on the Moreover, PLCD1 expression was significantly induced migration of gastric tumor cells, wound-healing assay after Aza treatment in all cancer cell lines, representative was performed. The PLCD1-transfected cells took much results were shown in Figure 1b. We further examined longer time (>48 h) to close a scratch wound than the the detailed methylation profiles of PLCD1 CGI by control vector-transfected cells (Figure 3c). bisulfite genomic sequencing (BGS) analysis of 37 CpG We further examined the migration of PLCD1- sites, including those CpG sites analysed by MSP. transfected cells by a 24-transwell system. As shown in Densely methylated CpG sites were detected in three cell Figure3d, PLCD1 expression led to significant sup- lines with no PLCD1 expression (Figure 1c). Mean- pression of gastric tumor cell migration after a 24 h PLCD1 while, both MSP and BGS showed that the CGI incubation. was dramatically demethylated after Aza treatment, showing a direct link between CGI methylation and PLCD1 silencing. Effect of PLCD1 on gastric tumor cell morphology and cytoskeleton Correlation of PLCD1 methylation with clinico- Theorganization of theactin cytoskeletonwas pathological features of gastric cancer patients examined by rhodamine-conjugated phalloidin staining Weanalysed PLCD1 methylation status in 98 primary of F-actin. PLCD1-transfected cells had symmetrical gastric cancer cases. PLCD1 methylation was detected round morphology with fewer thin fibers and formed in 62% (61/98) of tumors, whereas none of the normal nuclear F-actin rings, whereas the vector-transfected gastric mucosa tissues showed methylation. BGS analy- cells had asymmetrical polarized morphology and much sis confirmed the MSP results in some cases (Figures 1c thicker central stress fibers. Moreover, actin protrusion and d). PLCD1 methylation was found to be signifi- in the direction of movement was significantly reduced cantly correlated with high-stage tumors but not any at the leading edge in PLCD1-transfected cells (Po0.05) other clinico-pathological parameter (Table 1). (Figure4a).

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2468

Figure 1 PLCD1 expression and its CpG island (CGI) methylation in gastric cancer cell lines and primary tumors. (a) Broad expression of PLCD1 in normal adult tissue panel by semi-quantitative RT–PCR, with GAPDH as a control. (b) PLCD1 was greatly reduced or silenced in most cell lines as a result of its CGI methylation. Pharmacological demethylation with Aza restored PLCD1 expression in methylated and silenced cell lines. (c) Sequence of the PLCD1 CGI with locations of the 37 CpG sites analysed and primers used. Methylation-specific PCR (MSP) and bisulfite genomic sequencing (BGS) regions are also shown. Right panel, vertical lines indicate individual CpG sites. Cloned BGS-PCR products were sequenced and each clone was shown as an individual row, representing a single allele of the CGI. Filled circle, methylated; open circle, unmethylated. (d) Representative analysis of PLCD1 methylation in normal tissues (N) and primary tumors (T) by MSP. U, unmethylated; M, methylated.

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2469 Effect of PLCD1 on the expression of MMP7 PLCD1 inhibited peritoneal metastasis in mice and cytoskeletal reorganization proteins We did not observe any effect of PLCD1 by in vivo To investigate the molecular mechanism of PLCD1 in subcutaneous tumor model (data not shown). We then suppressing the motility of gastric tumor cells, we investigated whether PLCD1 transfection had a sup- examined the effect of PLCD1 on the expression and pressive effect on peritoneal metastasis by in vivo model. modification of a panel of MMPs and several key All five mice inoculated with vector-transfected MKN- cytoskeletal reorganization regulators, including Cdc42, 45 cells developed peritoneal metastasis with large Rac1/2/3, activeRhoA, phospho-Rac1/cdc42, cofilin volumes of bloody ascetic fluid (Figure 5a). Among and phospho-cofilin. We firstly checked intracellular them, one died after 3 weeks (Table 2). Peritoneal MMPs levels by semi-quantitative RT–PCR, among them only MMP7 was apparently decreased in PLCD1- transfected cells. MMP7 also decreased at the protein level in PLCD1-transfected cells (Figure 4b). Mean- while, total cofilin decreased and phospho-cofilin increased in PLCD1-transfected cells. But unexpectedly, the RhoA activity was detected to be slightly elevated in PLCD1-transfected cells compared with that of vector- transfected cells. There was no difference of Cdc42, Rac1/2/3 and phospho-Rac1/cdc42 expression between the PLCD1-transfected and vector-transfected cells. (Figure4c).

Table 1 Association between PLCD1 methylation and clinico- pathological parameters of gastric cancer patients Clinico-pathological n ¼ 98 PLCD1 methylation status P-value parameters Unmethylated Methylated (n ¼ 37) (n ¼ 61)

Sex Male76 25 51 Female 22 12 10 0.082

Age X60.52 46 17 29 o60.52 52 20 32 1.000

Size o30 mm 13 4 9 X30 mm 85 33 52 0.761

Differentiation Well and 35 15 20 moderate Poor 63 22 41 0.516

Depth of invasion T1–T3 10 6 4 T4 88 31 57 0.171

Clinical stage I–III 85 36 49 IV 13 1 12 0.016*

Lymph node metastasis Absent 75 29 46 Present 23 8 15 0.809

Liver metastasis Absent 92 36 56 Present 6 1 5 0.404 Figure 2 Representative immunohistochemical staining for Peritoneal dissemination PLCD1 in normal and carcinoma tissues of the stomach. Absent 94 37 57 (a) PLCD1 expression was detected in normal gastric tissue. Present 4 0 4 0.294 (b) Positive PLCD1 staining in unmethylated gastric cancer case T8. (c) Negative PLCD1 staining in cancer cells of a methylated *Statistically significant (Po0.05). caseT4. Original magnification: Â 200 in (a), Â 400 in (b and c).

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2470

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2471 dissemination was recognized from the innumerable decreased in mice inoculated with PLCD1-transfected whitish nodules visualized in the abdominal cavity, cells (Figure 5b and Supplementary Figure 1). mesenterium, omentum, parietal peritoneum and dia- phragm. But for miceinoculatedwith PLCD1-trans- fected MKN-45 cells, only 2/5 developed peritoneal Discussion metastasis with smaller tumor nodules and small volumes of bloody ascetic fluid. Furthermore, the PLCD1 is an enzyme that mediates regulatory signaling number of tumor nodules (>5 mm) significantly of energy metabolism pathways, calcium homeostasis

Figure 4 Effect of PLCD1 on gastric tumor cell morphology, cytoskeleton rearrangement and the expression of MMP7 and other cytoskeleton reorganization proteins. (a) Gastric tumor cell morphology and cytoskeleton. Arrowheads indicate nuclear F-actin rings in PLCD1-transfected cells and arrows indicate the thick polarized stress fiber protrusion at the leading edge of vector-transfected cells. Photos were taken under  400 magnification. The lower panel shows percentage of cells with actin protrusion formation. Results are means±s.d. of three independent experiments. (b) PLCD1 expression was confirmed by western blot analysis in PLCD1-transfected cells. MMP7 mRNA and protein were decreased in PLCD1-transfected cells. (c) Expression of several key cytoskeleton reorganization regulators were compared between PLCD1- and vector-transfected cells. Active RhoA bound to GTP (RhoA-GTP) in cells was measured by a pull-down assay and detected by immunoblot with an anti-RhoA antibody. Total RhoA was detected by direct immunoblot with anti-RhoA antibody.

Figure 3 PLCD1 inhibited the growth and migration of gastric tumor cells. (a) Soft agar colony formation and quantitativeanalysis. The numbers of colonies in each vector-transfected control were set to 100%, whereas PLCD1-transfected cells were presented as mean±s.d. Three independent experiments in triplicate wells were performed. Asterisk indicates statistically significant difference (*Po0.05). (b) Colony formation assay by monolayer culture and quantitative analysis. The numbers of colonies in each vector- transfected control were set to 100%, whereas PLCD1-transfected cells were presented as mean±s.d. Three independent experiments in triplicate wells were carried out. Asterisk indicates statistically significant difference (*Po0.05). (c) Wound-healing assay. Photos were taken every 24 h (original magnification: Â 100). (d) Migration assay by the24-transwellsystemand quantitativeanalysis. The pictures were taken 24 h after seeding (original magnification: Â 100). The numbers of migrated cells were counted in five representative high power fields per transwell. The numbers of colonies in each vector-transfected control were set to 100%, whereas PLCD1-transfected cells were presented as mean±s.d. Three independent experiments were carried out in triplicate. Asterisk indicates statistically significant difference (*Po0.05).

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2472 thekeyfactors controlling actin organization is PI(4,5)P2 of the plasma membrane. The effect of PI(4,5)P2 on actin dynamics is because of its interaction with a number of actin-binding proteins that govern all aspects of the actin skeleton architecture. Interestingly, a PLCD1-interacting protein, GTPase-activating protein, p122 (p122RhoGAP) isolated from rat, is localized in focal adhesion and may inhibit tumor cell migration through enhancing the PI(4,5)P2-hydrolyzing activity of PLCD1 (Homma and Emori, 1995; Sekimata et al., 1999; Kawai et al., 2004). Later, deleted in liver cancer 1 (DLC1) gene, the human homolog of p122RhoGAP was identified as a TSG in several types of cancers through modulating cytoskeleton reorganization (Durkin et al., 2007). This prompted us to study whether PLCD1 plays a tumor suppressive role in gastric cancer through inhibiting F-actin reorganization. Our results showed that PLCD1 is associated with an inhibitory effect on correct polarization, actin polymerization and stress fiber formation. A common and early requirement for different forms of cell motility is actin polymerization, which drives the formation of cell protrusions that are used to adhere to the extracellular matrix, define the direction of migra- tion and initiate cell crawling. Animal cells respond to signaling at the plasma membrane by remodeling their actin cytoskeleton (Ridley and Hall, 1992). As the burst of actin polymerization in response to signaling is Figure 5 Representative macroscopic results of the in vivo limited by the availability of actin monomers, a robust peritoneal metastasis model. (a) Abdominal distension because of protrusive response also requires the coincident disas- bloody ascites was evident (arrow indicated) in peritoneal sembly of the existing actin network. This is catalysed by metastasis mice. (b) Quantitativeanalysis of thenumbersof tumor nodules (>5 mm). Asterisk indicates statistically significant the conserved actin-binding protein, cofilin (Carlier difference (*Po0.05). et al., 1997; Bamburg, 1999; Hotulainen et al., 2005). It has been proposed that cofilin aids directional motility Table 2 Tumor nodules, presence or absence of bloody ascites and by severing actin filaments close to the cell front to survival status in the peritoneal dissemination model create new barbed ends to promote new local actin Groups Total Average Bloody Survival polymerization (Mouneimne et al., 2004). Multiple number numbers ascites mechanisms have been identified that regulate cofilin, of nodule including cofilin inactivation through phosphorylation (>5 mm) and polyphosphoinositide interaction, the effects of Miceinoculatedwith 530±42.4 2/5 All pH and the synergistic or competitive interactions of PLCD1-transfected alive cofilin with other actin-binding proteins. It is also MKN45 cells hypothesized that a pool of cofilin is associated with Mice inoculated with vector- 5 119.4±69.4a 5/5 One and inhibited by PIP2 (phosphatidylinositol 4,5-bipho- transfected MKN45 cells died sphate), and that it can be released and activated on aThe average numbers of peritoneal nodule (>5 mm) were significantly PLC-mediated PIP2 hydrolysis. Though it remains to be higher in the mice inoculated with vector-transfected MKN45 cells determined whether PIP2-mediated cofilin release occurs compared with the mice inoculated with PLCD1-transfected MKN45 in gastric cells, our study indicated that cofilin phos- cells (Po0.05). phorylation was affected by changes of the PLCD1 expression level. and intracellular movements (Ochocka and Pawelczyk, Activation of Rho GTPases has been shown to 2003). Recently, it was characterized as a novel TSG for stimulate the formation of stress fibers and focal esophageal squamous cell carcinoma (Fu et al., 2007). adhesions (Ridley and Hall, 1992). An unexpected Our current study showed that PLCD1 is also frequently finding in this work was that a slightly elevated RhoA silenced by epigenetic alteration in a tumor-specific activity was detected in PLCD1-transfected cells but not manner, and PLCD1 methylation is significantly corre- control cells with thicker central stress fibers. Although lated with high-stage gastric cancer. Moreover, PLCD1 some RhoA activity is required for migration, high- functions as a TSG suppressing the clonogenicity and RhoA activity actually inhibits movement (Takaishi motility/migration/metastasis of gastric cancer. et al., 1994; Nobes and Hall, 1999). RhoA activity could An early event of cell migration is characterized as the functionally antagonizetheformation of protrusions rapid reorganization of the actin cytoskeleton. Among through excessive contractility. Moreover, our results

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2473 are consistent with earlier reports that RhoA may act 5-aza-20-deoxycytidine treatment upstream of PLCD1 pathway as PLCD1 expression did Cells were seeded at a density of 1 Â 106 cells/ml. After not affect RhoA activation significantly (Homma and overnight culture, cells were treated with Aza (Sigma-Aldrich, Emori, 1995; Kawai et al., 2004). St Louis, MO, USA) at a final concentration of 10 mM for On theotherhand, themovement of cells into a tightly 3 days with changing of Aza-containing medium every 24 h, then harvested for DNA and RNA extraction. woven extracellular matrix may require an active proteo- lytic system, which can cleave a path for cell migration (Singer et al., 1999). Among the extracellular matrix- Semi-quantitative RT–PCR Semi-quantitative RT–PCR for the human multiple tissue degrading proteases, a great deal of attention has been panel, containing normal adult tissues (Stratagene, La Jolla, paid to theMMP family. Though PLC activation was CA, USA or Millipore Chemicon, Billerica, MA, USA) was onceindicatedtoberequiredforMMP13proteinperformed as described before (Ying et al., 2006). The expression, little is known about the possible role of GAPDH mRNA sequence was also amplified as an internal PLCD1 in MMP7 protein expression. MMP7 is a matrix- control. Primers used in this study were listed in Supplemen- degrading enzyme that is mainly produced from cancer tary Table1. cells, and involved in the invasion and metastasis of cancer. MMP7 wasidentifiedastheonlydifferentiallyBisulfite treatment and promoter methylation analysis overexpressed gene in gastric carcinoma cells, in compar- Bisulfite modification of DNA, MSP and BGS were carried ison to normal cells, among MMP family in a DNA out as described earlier (Murray et al., 2004; Ying et al., 2006). microarray analysis (Mori et al.,2002).Liuet al. (2002) MSP and BGS primers were listed in Supplementary Table 1. reported that a high frequency of MMP7 protein at the invasive front in gastric cancer was related to poor PLCD1-expressing plasmid and cell transfection prognosis. There are also several reports that peritoneal Gastric cancer cell lines, MKN-45 and SGC-7901 disseminationand lymph nodemetastasis areassociated were transfected with control or PLCD1-expressing plasmid with MMP7 in gastric cancer (Honda et al., 1996; (Fu et al., 2007), respectively, using FuGene 6 (Roche Nagashima et al.,1997;Adachiet al., 1998; Senota Molecular Systems, Alameda, CA, USA). Stable PLCD1- expressing clones (PLCD1-45 and PLCD1-7901) were selected et al., 1998; Yonemura et al., 2000; Yonemura et al.,2001; for further study. Kitoh et al., 2004). Thus, MMP7 is considered to have a direct rolein theprogression of gastric cancer.This study showed that MMP7 is a candidate gene directly or Colony formation assays For colony formation assay by monolayer culture, cells (2Â 105/ indirectly regulated by PLCD1 in gastric cancer. well) were plated in a 12-well plate and transfected with PLCD1- In conclusion, our results showed that PLCD1 is expression plasmid or the empty vector (2mg each) using frequently disrupted by epigenetic mechanism in gastric FuGene 6 (Roche, Indianapolis, IN, USA). Cells were collected tumors in a tumor-specific manner. We also showed the andplatedina10cmdish48hpost-transfection,andselectedfor PLCD1 could suppress the migration of gastric cancer 16–20 days with G418 (0.4 mg/ml). Surviving colonies (X50 cells, possibly through downregulating MMP7 expres- cells/colony) were counted after Giemsa staining. sion and hampering the reorganization of cytoskeleton For thesoft agar assay, B500 stable transfected cells were through cofilin inactivation by phosphorylation. A suspended in medium containing 0.3% low-melt agarose, better understanding of the molecular functions of seeded into a six-well plate that was overlaid with 0.5% low- 1 PLCD1 in tumor cell migration and cytoskeletal melt agarose, and allowed to grow for 2 weeks at 37 Cin5% CO2. The colonies containing more than 50 cells were counted regulation would provide novel therapeutic strategies under a microscope. Three wells were analysed for each to block metastatic progression, reduce dissemination of experiment. tumor cells and improve cancer patient survival. Wound-healing assay Cell motility was assessed using a scratch wound assay. Stably Materials and methods transfected cells were cultured in 35 mm dishes until confluent. The cell layers were carefully wounded using sterile tips and Cell lines, tumor samples and normal control tissues washed twice with fresh medium. After incubation for 24 and A panel of gastric cancer cell lines (AGS, SGC-7901, KatoIII, 48 h, the cells were photographed under a phase contrast MKN28, MKN45, BGC-823, NCI87, SNU1, SNU16, SNU719, microscope. The experiments were performed in triplicate. YCC1, YCC2, YCC3, YCC6, YCC7, YCC9, YCC10, YCC11 and YCC16) were studied. A total of 98 patients with gastric Cell migration assay cancer who underwent surgery from Feb 2004 to June 2006 at Cells were trypsinized and resuspended in Dulbecco’s modied theSir Run Run Shaw Hospital (Hangzhou, Zhejiang, China) Eagle’s medium containing 1% fetal bovine serum at a density were investigated in this study. There were 76 male and 22 of 1 Â 106 cells/ml. 100 ml of the cell suspension was added into female patients ranging in age from 24 to 79 years. The mean age the upper chamber of a transwell (Corning, Corning, NY, of patients was 60.52±15.28 years. Patients who received USA) consisted of inserts containing 8-mm pore-size PET preoperative chemotherapy were excluded from this study. membrane. Dulbecco’s modied Eagle’s medium (600 ml) con- Among these cases, 60 paraffin tumor blocks were available taining 10% fetal bovine serum was placed in the lower for immunohistochemistry. Eight normal gastric mucosa biopsy chamber. After a 24 h incubation at 37 1C, cells remained in the samples were used as normal controls. This study was approved upper chamber was removed carefully by cotton swab and the and monitoredby theethicscommitteeoftheSir Run Run Shaw membrane was cut off by an operating knife. The side facing Hospital, Zhejiang University. lower chamber was stained with 0.05% crystal violet and

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2474 attached cells were counted under a light microscope. The antibody; anti-phospho-Rac1/cdc42 rabbit monoclonal anti- experiment was performed three times. body; anti-cofilin rabbit monoclonal antibody; anti-phospho- cofilin rabbit monoclonal antibody (above antibodies were all Cell proliferation assay from Cell Signaling, Beverly, MA, USA used at 1:1000); anti- Stable clones were seeded in 96-well plates. The colorimetric MTT MMP7 mouse monoclonal antibody (ThermoFisher Scientific, (Sigma-Aldrich) assay was used to measure cell numbers at Fremont, CA, USA); anti-GAPDH mouse monoclonal antibody various time points. The experiments were performed in triplicate. and anti-PLCD1 polyclonal antibody (1:1000, Santa Cruz Biotechnologies). Detection was carried out using the ECL kit (Pierce Chemical Co., Rockford, IL, USA) and the blots were Cell-cycle analysis developed using a Fujifilm Las-4000 Imaging System. A total of 1–2 Â 106 cells were cultured in Dulbecco’s modied Eagle’s medium containing 10% fetal bovine serum. Serum was withdrawn from culture medium when cells were 70% RhoA activation assay confluent. After 72 h, 10% fetal bovine serum was added in the To assess RhoA activation, the amount of RhoA-GTP bound medium for an additional 12 h. Cells were fixed in 70% to the Rhotekin RBD was determined using the Rho ethanol, stained with propidium iodide and DNA content was Activation Assay Kit (UpstateBiotechnology,LakePlacid, analysed by FACSCalibur Flow Cytometer (BD Biosystems, NY, USA) according to theinstruction of themanufacturer. Heidelberg, Germany). In vivo subcutaneous tumor model Immunohistochemistry All of the in vivo experimental protocols were approved by the The ChemMate EnVision Detection Kit (DAKO, Carpinteria, animal carecommitteeofSir Run Run Shaw Hospital, 6 CA, USA) was used for immunohistochemistry according to Zhejiang University. Viable MKN-45 cells (2 Â 10 cells/ company’s recommended procedure. Briefly, after being 0.2 ml/mouse) were injected subcutaneously into right and left deparaffinized and hydrated, the paraffin-embedded sections flank of 5-week-old female BALB/c nude mice (eight mice per were placed in 0.01 M sodium citratebuffer(pH 6.0), and group). Tumor volume was assessed every 2 days for 4 weeks. subjected to pressure cooker treatment for 2 min at full Tumor volumewas calculatedby thefollowing formula: (short 2 pressure with a domestic pressure cooker. After cooling to diameter) Â (long diameter)/2. room temperature, the slides were rinsed with Tris buffered saline(0.05 M Tris/0.15 M NaCl, pH 7.6). The endogenous In vivo peritoneal metastasis model peroxidase activity was blocked by incubating the sections with MKN-45 cells were injected into the peritoneal cavity of mice 3% hydrogen peroxide. The primary antibody used in this (2 Â 106 cells/1 ml/mouse, 5 mice per group). Four weeks later, study is a rabbit polyclonal antibody against thehuman all mice were killed. We determined the body weight, the PLCD1 protein (clone H-140, Santa Cruz Biotechnology, CA, number of tumor nodules, and the presence or absence of any USA). The sections were incubated with the primary antibody bloody ascites as described before (Sako et al., 2004). (1:400 dilution) overnight at 4 1C. Then, the ChemMate EnVision/HRP, Rabbit/Mouse (ENV) reagent was applied to Statistical analysis the sections, followed by application of ChemMate DAB þ The results are expressed as values of mean±s.d. Statistical Chromogen included in the kit. The slides were lightly analysis was performed in SPSS 11.0 for Windows (SPSS Inc., counterstained with hematoxylin. Chicago, IL, USA). The two-tailed Chi-square and Fisher’s exact tests were used to analyze the association of the PLCD1 Immunofluorescence CGI methylation with the different clinico-pathological para- Cells were cultured on glass coverslips and fixed in 3.7% meters and the PLCD1 protein expression. For all tests, paraformaldehyde in phosphate buffered saline (pH 7.4) for Po0.05 was considered to be of statistical significance. 10 min, permeabilized in 0.1% Triton X-100 in phosphate buffered saline for 4 min, blocked with 1% bovine serum albumin/phosphate buffered saline for 1 h, and then incubated Conflict of interest at room temperature for 1 h with rhodamine-conjugated phalloidin (Invitrogen, Carlsbad, CA, USA) at 1:100 in The authors declare no conflict of interest. blocking solution. Nuclei were counterstained with DAPI. Images were acquired with an Olympus BX51 microscope. Acknowledgements

Western blot analysis This study was supported by a grant from the Chinese Western blot analysis was performed as described earlier University of Hong Kong, a grant from National Natural (Ying et al., 2008). Primary antibodies were used as follows: Science Foundation of China (Grant no. 30770920/ anti-Cdc42 rabbit monoclonal antibody; anti-Rac1/2/3 C03030202) and a grant from Natural Science Foundation of rabbit monoclonal antibody; anti-RhoA rabbit monoclonal Zhejiang Province (Grant no. Y206090).

References

Adachi Y, Itoh F, Yamamoto H, Matsuno K, Arimura Y, Kusano M Bamburg JR. (1999). Proteins of the ADF/cofilin family: essential et al. (1998). Matrix metalloproteinase matrilysin (MMP-7) regulators of actin dynamics. Annu Rev Cell Dev Biol 15: 185–230. participates in the progression of human gastric and esophageal Carlier MF, Laurent V, Santolini J, Melki R, Didry D, Xia GX et al. cancers. Int J Oncol 13: 1031–1035. (1997). Actin depolymerizing factor (ADF/cofilin) enhances the rate

Oncogene PLCD1 methylation in gastric cancer X-T Hu et al 2475 of filament turnover: implication in actin-based motility. J Cell Biol endothelial cells adjacent to matrilysin-producing tumors. Int J 136: 1307–1322. Cancer 72: 441–445. Carvou N, Norden AG, Unwin RJ, Cockcroft S. (2007). Signalling Nakahara M, Shimozawa M, Nakamura Y, Irino Y, Morita M, Kudo through phospholipase C interferes with clathrin-mediated endocy- Y et al. (2005). A novel phospholipase C, PLC(eta)2, is a neuron- tosis. Cell Signal 19: 42–51. specific isozyme. J Biol Chem 280: 29128–29134. Durkin ME, Yuan BZ, Zhou X, Zimonjic DB, Lowy DR, Nobes CD, Hall A. (1999). Rho GTPases control polarity, protrusion, Thorgeirsson SS et al. (2007). DLC-1: a Rho GTPase-activating and adhesion during cell movement. J Cell Biol 144: 1235–1244. protein and tumour suppressor. J Cell Mol Med 11: Ochocka AM, Pawelczyk T. (2003). Isozymes delta of phosphoinosi- 1185–1207. tide-specific phospholipase C and their role in signal transduction in Essen LO, Perisic O, Cheung R, Katan M, Williams RL. (1996). thecell. Acta Biochim Pol 50: 1097–1110. Crystal structure of a mammalian phosphoinositide-specific phos- Panani AD. (2008). Cytogenetic and molecular aspects of gastric pholipaseC delta. Nature 380: 595–602. cancer: clinical implications. Cancer Lett 266: 99–115. Fu L, Qin YR, XieD, Hu L, Kwong DL, Srivastava G et al. (2007). Parkin DM, Pisani P, Ferlay J. (1993). Estimates of the worldwide Characterization of a novel tumor-suppressor gene PLC delta 1 incidence of eighteen major cancers in 1985. Int J Cancer 54: at 3p22 in esophageal squamous cell carcinoma. Cancer Res 67: 594–606. 10720–10726. Qiu GH, Tan LK, Loh KS, Lim CY, Srivastava G, Tsai ST et al. Fuchs CS, Mayer RJ. (1995). Gastric carcinoma. N Engl J Med 333: (2004). The candidate tumor suppressor gene BLU, located at the 32–41. commonly deleted region 3p21.3, is an E2F-regulated, Fukami K. (2002). Structure, regulation, and function of phospholi- stress-responsive gene and inactivated by both epigenetic and paseC isozymes. J Biochem 131: 293–299. genetic mechanisms in nasopharyngeal carcinoma. Oncogene 23: Hesson LB, Cooper WN, Latif F. (2007). Evaluation of the 3p21.3 4793–4806. tumour-suppressor gene cluster. Oncogene 26: 7283–7301. Qiu GH, Salto-Tellez M, Ross JA, Yeo W, Cui Y, Wheelhouse N et al. Homma Y, Emori Y. (1995). A dual functional signal mediator (2008). The tumor suppressor gene DLEC1 is frequently silenced by showing RhoGAP and phospholipase C-delta stimulating activities. DNA methylation in hepatocellular carcinoma and induces G1 EMBO J 14: 286–291. arrest in cell cycle. J Hepatol 48: 433–441. Honda M, Mori M, Ueo H, Sugimachi K, Akiyoshi T. (1996). Rhee SG.. (2001). Regulation of phosphoinositide-specific phospholi- Matrix metalloproteinase-7 expression in gastric carcinoma. Gut 39: paseC. Annu Rev Biochem 70: 281–312. 444–448. Ridley AJ, Hall A. (1992). The small GTP-binding protein Rho Hotulainen P, Paunola E, Vartiainen MK, Lappalainen P. (2005). regulates the assembly of focal adhesions and actin stress fibers in Actin-depolymerizing factor and cofilin-1 play overlapping roles in response to growth factors. Cell 70: 389–399. promoting rapid F-actin depolymerization in mammalian nonmus- Sako A, Kitayama J, Koyama H, Ueno H, Uchida H, Hamada H et al. clecells. Mol Biol Cell 16: 649–664. (2004). Transduction of soluble Flt-1 gene to peritoneal mesothelial Irino Y, Cho H, Nakamura Y, Nakahara M, Furutani M, Suh PG cells can effectively suppress peritoneal metastasis of gastric cancer. et al. (2004). Phospholipase C delta-type consists of three isozymes: Cancer Res 64: 3624–3628. bovine PLCdelta2 is a homologue of human/mouse PLCdelta4. Santarius M, Lee CH, Anderson RA. (2006). Supervised membrane Biochem Biophys Res Commun 320: 537–543. swimming: small G-protein lifeguards regulate PIPK signalling and Janetopoulos C, Devreotes P. (2006). Phosphoinositide signaling plays monitor intracellular PtdIns(4,5)P2 pools. Biochem J 398: 1–13. a key role in cytokinesis. J Cell Biol 174: 485–490. Sekimata M, Kabuyama Y, Emori Y, Homma Y. (1999). Morpho- Kaminishi M. (2005). Diversity of gastric carcinogenesis. Oncology logical changes and detachment of adherent cells induced by p122, a 69(Suppl 1): 1–8. GTPase-activating protein for Rho. J Biol Chem 274: 17757–17762. Kawai K, Yamaga M, IwamaeY, Kiyota M, Kamata H, Hirata H Senota A, Itoh F, Yamamoto H, Adachi Y, Hinoda Y, Imai K. (1998). et al. (2004). A PLCdelta1-binding protein, p122RhoGAP, is Relation of matrilysin messenger RNA expression with invasive localized in focal adhesions. Biochem Soc Trans 32: 1107–1109. activity in human gastric cancer. Clin Exp Metastasis 16: 313–321. Kitoh T, Yanai H, Saitoh Y, Nakamura Y, Matsubara Y, Kitoh H Singer CF, Marbaix E, Lemoine P, Courtoy PJ, Eeckhout Y. (1999). et al. (2004). Increased expression of matrix metalloproteinase-7 in Local cytokines induce differential expression of matrix metallo- invasive early gastric cancer. J Gastroenterol 39: 434–440. proteinases but not their tissue inhibitors in human endometrial Liu XP, Kawauchi S, Oga A, Tsushimi K, Tsushimi M, Furuya T et al. fibroblasts. Eur J Biochem 259: 40–45. (2002). Prognostic significance of matrix metalloproteinase-7 Takaishi K, Sasaki T, Kato M, Yamochi W, Kuroda S et al. (1994). (MMP-7) expression at the invasive front in gastric carcinoma. Involvement of Rho p21 small GTP-binding protein and its Jpn J Cancer Res 93: 291–295. regulator in the HGF-induced cell motility. Oncogene 9: 273–279. Maehara Y, Hasuda S, Koga T, Tokunaga E, Kakeji Y, Sugimachi K. Ying J, Li H, Seng TJ, Langford C, Srivastava G, Tsao SW et al. (2000). Postoperative outcome and sites of recurrence in (2006). Functional epigenetics identifies a protocadherin PCDH10 patients following curative resection of gastric cancer. Br J Surg as a candidate tumor suppressor for nasopharyngeal, esophageal 87: 353–357. and multiple other carcinomas with frequent methylation. Oncogene Mouneimne G, Soon L, DesMarais V, Sidani M, Song X, Yip SC et al. 25: 1070–1080. (2004). Phospholipase C and cofilin are required for carcinoma Ying J, Li H, Yu J, Ng KM, Poon FF, Wong SC et al. (2008). WNT5A cell directionality in response to EGF stimulation. J Cell Biol 166: exhibits tumor-suppressive activity through antagonizing the Wnt/ 697–708. beta-catenin signaling, and is frequently methylated in colorectal Mori M, Mimori K, Yoshikawa Y, Shibuta K, Utsunomiya T, cancer. Clin Cancer Res 14: 55–61. Sadanaga N et al. (2002). Analysis of the gene-expression profile Yonemura Y, Endou Y, Fujita H, Fushida S, Bandou E, Taniguchi K regarding the progression of human gastric carcinoma. Surgery 131: et al. (2000). Role of MMP-7 in the formation of peritoneal S39–S47. dissemination in gastric cancer. Gastric Cancer 3: 63–70. Murray PG, Qiu GH, Fu L, Waites ER, Srivastava G, Heys D et al. Yonemura Y, Fujimura T, Ninomiya I, Kim BS, Bandou E, Sawa T (2004). Frequent epigenetic inactivation of the RASSF1A tumor et al. (2001). Prediction of peritoneal micrometastasis by peritoneal suppressor gene in Hodgkin’s lymphoma. Oncogene 23: 1326–1331. lavaged cytology and reverse transcriptase-polymerase chain reac- Nagashima Y, Hasegawa S, Koshikawa N, Taki A, Ichikawa Y, tion for matrix metalloproteinase-7 mRNA. Clin Cancer Res 7: Kitamura H et al. (1997). Expression of matrilysin in vascular 1647–1653.

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

Oncogene