Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

HOXA9 promotes homotypic and heterotypic cell interactions that facilitate ovarian cancer dissemination via its induction of P-cadherin

HOXA9 promotes homotypic and heterotypic cell interactions that facilitate ovarian cancer... Background: Epithelial ovarian cancer (EOC) is a lethal disease that frequently involves the peritoneal cavity. Dissemination of EOC is a multi-step process in which exfoliated tumor cells survive in the peritoneal fluid as multi-cellular aggregates and then form invasive implants on peritoneal surfaces. The mechanisms that control this process are poorly understood. We previously identified that high expression of the developmental patterning gene HOXA9 is associated with poor survival in EOC patients. In this study, we investigated the significance and mechanisms of HOXA9 in controlling aggregation and implantation of floating EOC cells. Methods: HOXA9 was inhibited by shRNAs or expressed in EOC cells that were propagated in suspension cultures and in the peritoneal cavity of mice. Cell death was assayed by flow cytometry and ELISA. Cell aggregation, attachment and migration were evaluated by microscopy, transwell chamber assays and histopathologic analysis. DNA-binding of HOXA9 and its effect on expression of the cell adhesion molecule P-cadherin were assayed by chromatin immunoprecipitation, quantitative RT-PCR and Western blot. HOXA9 and P-cadherin expression was evaluated in publicly available datasets of EOC clinical specimens. Results: We identified that HOXA9 promotes aggregation and inhibits anoikis in floating EOC cells in vitro and in xenograft models. HOXA9 also stimulated the ability of EOC cells to attach to peritoneal cells and to migrate. HOXA9bound thepromoterofthe CDH3 gene that encodes P-cadherin, induced CDH3 expression in EOC cells, and was associated with increased CDH3 expression in clinical specimens of EOC. Inhibiting P-cadherin in EOC cells that expressed HOXA9 abrogated the stimulatory effects of HOXA9 on cell aggregation, implantation and migration. Conversely, these stimulatory effects of HOXA9 were restored when P-cadherin was reconstituted in EOC cells in which HOXA9 was inhibited. Conclusion: These findings indicate that HOXA9 contributes to poor outcomes in EOC in part by promoting intraperitoneal dissemination via its induction of P-cadherin. Keywords: Ovarian cancer, Homeobox gene, Metastasis, Cell adhesion, P-cadherin Background lymphatic routes, EOC cells typically spread by shed- More than 60% of women with epithelial ovarian can- ding into the peritoneal fluid which transports tumor cer (EOC) are diagnosed with advanced-stage disease cells, either as multi-cellular aggregates or as single that has disseminated throughout the peritoneal cavity cells, throughout the peritoneal cavity [2-4]. Subse- [1]. Despite advances in tumor debulking surgery and quently, EOC cells attach to the mesothelium-lined chemotherapy, patients with advanced-stage EOC have peritoneal surfaces, such as the cavity wall, diaphragm a 5-year survival rate of only 30% [1]. Whereas many other and omentum, where they form invasive implants [2-4]. types of solid tumors metastasize via hematogenous or This ‘seeding’ of the peritoneal cavity with tumor cells is often associated with ascites formation. Although it is unclear whether EOC cells detach from the primary * Correspondence: hnaora@mdanderson.org tumoras clustersor assingle cells that then assemble Department of Molecular & Cellular Oncology, University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA © 2014 Ko and Naora; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ko and Naora Molecular Cancer 2014, 13:170 Page 2 of 13 http://www.molecular-cancer.com/content/13/1/170 into clusters, multi-cellular aggregates of floating EOC shRNA-expressing SKOV3ip cell lines [10]. These EOC cells are increasingly regarded as ‘seeds of metastasis’ cell lines stably expressed GFP, enabling their detection that are able to escape anoikis and implant on to among host cells in ascites. Floating EOC cells in asci- peritoneal surfaces [3-5]. However, the tissue-specific tes of control xenograft models (that expressed empty mechanisms that facilitate the aggregation and implant- vector or non-targeting shRNA) were present as large ation of floating EOC cells and drive the unique clinical compact aggregates [Figure 1A]. In contrast, ascites behavior of this disease are poorly understood. collected from HOXA9-knockdown models (shA9-A, Homeobox genes encode transcription factors, com- shA9-B) contained smaller aggregates or single EOC monly termed homeoproteins, that play essential roles in cells [Figure 1A]. The aggregation of floating EOC cells controlling developmental patterning and are expressed is thought to enable these cells to escape anoikis [3,4]. in a tightly regulated temporal and tissue-specific man- Cell death was evaluated within the population of ner [6-8]. Many homeoproteins are aberrantly expressed GFP-expressing ascitic EOC cells by flow cytometric in a variety of malignancies, but their functional signifi- analysis of 7-amino actinomycin D (7AAD) staining. cance in tumor progression is poorly understood as only As showninFigure1B, theproportionofascitic EOC few bona fide target genes have been identified [6,7]. cells that exhibited cell death was substantially higher The homeobox gene HOXA9 is normally expressed in the HOXA9-knockdown models than in control during differentiation of the Müllerian ducts into the models. These observations raise the possibility that female reproductive tract [9]. We have identified that HOXA9 promotes aggregation of floating EOC cells high HOXA9 expression is strongly associated with and inhibits anoikis. poor overall survival of EOC patients [10]. Studies of mouse xenograft models revealed that expression of HOXA9 also increases EOC cell implantation and HOXA9 in EOC cells promotes growth of solid peri- invasiveness toneal implants by inducing normal peritoneal fibro- Floating EOC cells that are transported by the peritoneal blasts and mesenchymal stem cells to acquire features fluid frequently implant on the diaphragm, peritoneal of cancer-associated fibroblasts that in turn supported cavity wall, omentum and mesentery [2-4]. We evalu- tumor growth and angiogenesis [10]. This stimulatory ated sections of tissues collected from mice that were effect of HOXA9 on solid tumor growth was attributed inoculated with equivalent numbers of control and to its activation of the gene encoding transforming HOXA9-knockdown SKOV3ip cells. Whereas numer- growth factor-β2(TGF-β2) that acted in a paracrine ous mesenteric implants were detected in the control manner on stromal cells [10]. Because EOC cells in groups, fewer implants were detected in the HOXA9- solid tumors and in ascites have different biological knockdown groups [Figure 1C]. In EOC patients, implants behaviors and exist in different microenvironments, we tend to only invade the superficial bowel serosa and investigated the possibility that HOXA9 mediates other not the deeper layers [3]. Some superficial bowel serosa types of effects in free-floating EOC cells. In this study, invasion was observed in control mice, but was markedly we identified that HOXA9 promotes the assembly of reduced in the HOXA9-knockdown groups (P <0.05) floating EOC cells into multi-cellular aggregates and [Additional file 1: Figure S1A]. Implants on the dia- inhibits anoikis, and also stimulates tumor-peritoneum phragm extensively invaded adjacent muscle in the control interactions and tumor cell migration. These stimula- groups, whereas invasive depth of diaphragmatic implants tory effects of HOXA9 were found to be largely attrib- was significantly reduced in the HOXA9-knockdown utable to its induction of the cell adhesion molecule groups (P < 0.01) [Figure 1D; Additional file 1: Figure S1B]. P-cadherin that is encoded by the CDH3 gene, a tran- Similarly, implants on the peritoneal cavity wall were scriptional target of HOXA9. significantly less invasive in the HOXA9-knockdown groups than in the control groups (P < 0.01) [Additional Results file 1: Figure S1C]. These findings indicate that HOXA9 HOXA9 promotes aggregation and survival of floating not only promotes aggregation and survival of floating EOC cells in i.p. xenograft models EOC cells, but also increases the peritoneal implant- We previously identified that expression of HOXA9 in ation and invasiveness of these cells. EOC cells promotes growth of solid tumor xenografts, but does not stimulate proliferation of EOC cells in vitro HOXA9 promotes aggregation and inhibits anoikis of EOC [10]. Because the biological behavior of ascitic tumor cells in vitro cells markedly differs from that of solid tumors, we The development of tumor implants at distal sites investigated the effect of HOXA9 on floating EOC cells depends on the survival of free-floating EOC cells that in ascites. I.p. xenograft models were generated from escape anoikis by forming multi-cellular aggregates [3,4]. previously established HOXA9+ control and HOXA9 We investigated the possibility that HOXA9 promotes Ko and Naora Molecular Cancer 2014, 13:170 Page 3 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 1 Knockdown of HOXA9 inhibits aggregation and survival of ascitic EOC cells and decreases EOC cell implantation and invasiveness in i.p. xenograft models. Female nude mice (n = 5 per group) were inoculated i.p. with equivalent numbers (2 × 10 )of GFP-expressing HOXA9+ control (empty vector, non-targeting) and HOXA9-knockdown (shA9-A, shA9-B) SKOV3ip cells. Mice were sacrificed at 3 weeks thereafter. (A,B) Ascitic cells were collected and stained with Hoechst dye to visualize nuclei and with 7AAD to assay cell death. (A) Representative examples of Hoechst staining (shown in blue) of ascitic cells where GFP-expressing tumor cells (shown in green) were visualized by fluorescence microscopy. Bar, 50 μm. (B) Flow cytometric analysis of 7AAD staining within the gated population of GFP + ascitic tumor cells. (C,D) Representative examples of HE-stained sections of (C) bowel tissues with mesenteric implants (bar, 1 mm) and (D) diaphragmatic implants (bar, 100 μm). aggregation and survival of floating EOC cells in in vitro aggregates of loosely clustered cells [Figure 2A]. Cell assays, independently of implantation and of effects of death was observed in HOXA9-knockdown cells as host cells. Cells of control and HOXA9-knockdown detected by staining with 7AAD [Figure 2B]. In contrast, SKOV3ip lines were incubated as suspension cultures control cells were mostly viable [Figure 2B]. Evaluation of in plates coated with poly(2-hydroxyethyl methacrylate) active caspase 3 levels indicated higher levels of apoptotic (polyHEMA), an inert polymer that prevent cells cell death in HOXA9-knockdown cells than in control from adhering to substratum. Whereas control cells cells [Figure 2C]. The increased level of cell death in formed large aggregates, suspension cultures of HOXA9- HOXA9-knockdown cells was additionally confirmed by knockdown cells comprised of single cells and small assaying mono- and oligo- nucleosomes in cell lysates Ko and Naora Molecular Cancer 2014, 13:170 Page 4 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 2 Knockdown of HOXA9 inhibits aggregation of floating EOC cells, increases anoikis, decreases EOC-mesothelial cell interactions, and reduces EOC cell migration in vitro. (A-D) Cells of control and HOXA9-knockdown SKOV3ip lines were incubated as suspension cultures in polyHEMA-coated plates for 3 days and evaluated for changes in cell morphology and cell death. (A) Cell morphology viewed by phase-contrast light microscopy. Bar, 100 μm. The extent of cell death was evaluated by flow cytometric analysis of (B) 7AAD staining and (C) staining of active caspase-3. In (D), cell death was evaluated by assaying mono- and oligo- nucleosomes in cell lysates by ELISA. Shown are mean ± sd values of three independent experiments. Significance of differences was evaluated by Student t-test. (E,F) Equivalent numbers of control and HOXA9-knockdown SKOV3ip cells that stably expressed GFP were seeded onto confluent monolayers of normal human omental mesothelial cells. At 1 h thereafter, attached tumor cells were viewed by fluorescence microscopy and counted in three random 100× microscopic fields per assay. (E) Representative examples of SKOV3ip cells (shown in green) attached to mesothelial monolayers. Bar, 100 μm. (F) Mean ± sd values of three independent attachment assays. (G) Equivalent numbers of control and HOXA9-knockdown SKOV3ip cells were seeded in transwell chambers. At 6 h thereafter, migrating cells were counted in three random 100x microscopic fields per assay. Shown are mean ± sd values of three independent migration assays. Ko and Naora Molecular Cancer 2014, 13:170 Page 5 of 13 http://www.molecular-cancer.com/content/13/1/170 [Figure 2D]. Together, these observations indicate that levels of other classical cadherins [Figure 3A]. SKO- HOXA9 promotes the assembly of floating EOC cells V3ip is an aggressive subclone that originally derived into multi-cellular aggregates and inhibits anoikis. from the parental SKOV3 cell line (SKOV3-Par) [15]. In contrast to SKOV3ip cells, SKOV3-Par cells did not HOXA9 stimulates interactions between EOC cells and endogenously express HOXA9 or P-cadherin [Figure 3A]. peritoneal mesothelial cells and promotes EOC cell Stable expression of HOXA9 in SKOV3-Par cells induced migration in vitro P-cadherin expression at the protein and mRNA levels In subsequent experiments, we performed short-term [Figures 3A,B]. The Cancer Genome Atlas (TCGA) in vitro attachment assays to evaluate the effect of dataset is the largest compilation of gene expression HOXA9 on interactions between EOC cells and peri- data of clinical specimens of EOC. To determine whether toneal mesothelial cells independently of its effects on P-cadherin expression is elevated in clinical specimens aggregation and survival of EOC cells. Equivalent num- of EOC that highly express HOXA9, we stratified EOC bers of control and HOXA9-knockdown SKOV3ip cases in the TCGA dataset (n = 567 cases) into quartile cells were seeded as single cell suspensions onto con- sub-groups according to the levels of HOXA9 tran- fluent monolayers of normal human omental mesothe- scripts of tumors. As shown in Figure 3C, levels of lial cells. Attachment of SKOV3ip cells to mesothelial CDH3 transcripts were found to be significantly higher cells was assayed at 1 h after seeding during which in HOXA9-High tumors (upper quartile sub-group) than time no significant change in cell aggregation or sur- in HOXA9-Low tumors (lower quartile sub-group) (P = vival occurred. The numbers of HOXA9-knockdown 0.011). Two putative consensus HOXA9-binding sites were SKOV3ip cells that bound to mesothelial cells were identified in the CDH3 promoter at 1.8 kb (site A) and significantly lower than the numbers of bound control at 2.1 kb (site B) upstream of the transcription start site SKOV3ip cells (P < 0.001) [Figures 2E,F]. To evaluate [Figure 3D]. Binding of endogenous HOXA9 to both sites the effect of HOXA9 on the migratory potential of was detected by chromatin immunoprecipitation (IP) assays EOC cells independently of its effects on cell aggrega- in control SKOV3ip cells, but no binding was detected in tion and mesothelial attachment, we assayed migration HOXA9-knockdown SKOV3ip cells [Figure 3D]. These of EOC cells that were seeded as single cell suspen- findings indicate that HOXA9 induces P-cadherin ex- sions into transwell chambers. As compared to control pression and that the CDH3 gene is a transcriptional SKOV3ip cells, HOXA9-knockdown SKOV3ip cells target of HOXA9. exhibited significantly reduced migration (P < 0.001) [Figure 2G]. These findings indicate that HOXA9 P-cadherin inhibition abrogates the stimulatory effects of expression in EOC cells stimulates EOC-mesothelial HOXA9 on EOC cell aggregation, implantation and cell interactions and EOC cell migration, and that migration these stimulatory effects are independent of its effect To determine whether HOXA9 promotes EOC cell on EOC cell aggregation and survival. aggregation, implantation and migration by inducing P-cadherin expression, we evaluated the effects of inhi- HOXA9 binds the CDH3 promoter and induces P-cadherin biting P-cadherin in EOC cells that expressed HOXA9. expression P-cadherin levels that were induced by enforced HOXA9 The classical cadherin P-cadherin has been reported to expression in SKOV3-Par cells were knocked-down by be induced during early stages of EOC progression and shRNAs targeting two different sites within the CDH3 to promote EOC cell migration [11-13]. We recently gene (shPCAD-A, shPCAD-B) [Figure 4A]. Quantifica- identified that P-cadherin also facilitates aggregation of tion of this knockdown is shown in Additional file 2: floating EOC cells, prevents anoikis and mediates the Figure S2B. Knockdown of P-cadherin in HOXA9+ attachment of EOC cells to peritoneal mesothelial SKOV3-Par cells reduced cell aggregation [Figure 4B]. cells [14]. Because HOXA9 is a transcription factor, Cell death in HOXA9+ SKOV3-Par cells was signifi- we investigated the possibility that HOXA9 induces cantly increased when P-cadherin was knocked-down P-cadherin expression. P-cadherin levels were mark- (P < 0.001) [Figures 4C,D]. To confirm our findings, edly reduced when HOXA9 was knocked-down in we evaluated the effects of inhibiting P-cadherin by a SKOV3ip cells [Figure 3A]. Quantification of this pro- neutralizing antibody (Ab). Treatment of suspension tein down-regulation is shown in Additional file 2: cultures of HOXA9+ SKOV3-Par cells with P-cadherin Figure S2A. The down-regulation of P-cadherin was Ab inhibited cell aggregation and increased cell death, confirmed by quantitative reverse transcription PCR as compared to cells that were treated with control IgG (qRT-PCR) analysis of transcript levels of CDH3,the [Additional file 3: Figure S3A,B]. In addition, knock- gene encoding P-cadherin [Figure 3B]. Knockdown down of P-cadherin in HOXA9+ SKOV3-Par cells of HOXA9 in SKOV3ip cells did not down-regulate significantly reduced the ability of these tumor cells to Ko and Naora Molecular Cancer 2014, 13:170 Page 6 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 3 HOXA9 induces P-cadherin expression in EOC cells. (A) Western blot analysis of HOXA9 and cadherin levels in control and HOXA9-knockdown SKOV3ip lines and in vector-control and HOXA9-transfected SKOV3-Par lines. The 34 kD band corresponding to HOXA9 is indicated by an arrow. (B) qRT-PCR analysis of relative CDH3 mRNA levels in SKOV3ip and SKOV3-Par lines. Significance of differences in CDH3 mRNA levels was evaluated by Student t-test. (C) EOC cases from the TCGA Project (n = 567) were stratified according to HOXA9 expression in tumors, where HOXA9 mRNA levels were defined as High (≥ upper quartile) and Low (≤ lower quartile). Significance of differences in CDH3 mRNA levels (log2 scale) between upper and lower quartile sub-groups was evaluated by Mann–Whitney U-test. (D) Schematic representation of the human CDH3 promoter showing locations of the two HOXA9-binding sites (site A: 5′-TCATTTAAAAC-3′ and site B: 5′-TAATTTATTTAATAC-3′). Binding of endogenous HOXA9 in control SKOV3ip cells to these sites was detected by chromatin IP. Negative controls included IP using cells expressing HOXA9 shRNA (shA9-B) and IP with IgG. GAPDH was amplified as an irrelevant gene control. attach to mesothelial cells (P < 0.01) and to migrate (P < 0.001) [Figure 4D]. Together, these observations indicate that the stimulatory effects of HOXA9 on aggregation and survival of floating EOC cells, meso- thelial attachment and EOC cell migration are largely mediated by its induction of P-cadherin expression. Stimulatory effects of HOXA9 on EOC cell aggregation, implantation and migration are recapitulated by P-cadherin in vitro and in vivo In converse experiments, we evaluated whether the stimulatory effects of HOXA9 on EOC cell aggrega- tion, implantation and migration can be restored when P-cadherin is reconstituted in EOC cells in which HOXA9 is inhibited. CDH3 cDNA was stably expressed in HOXA9-knockdown SKOV3ip cells (shA9-B + PCAD) to restore the level of P-cadherin that was similar to the endogenous P-cadherin level in HOXA9+ control (non-targeting) SKOV3ip cells [Figure 5A, Additional file 2: Figure S2C]. Reconstitution of P-cadherin in HOXA9-knockdown SKOV3ip cells markedly increased cell aggregation and significantly decreased cell death in suspension cultures (P < 0.001) [Figures 5B-D]. Recon- stitution of P-cadherin also increased the ability of HOXA9-knockdown SKOV3ip cells to attach to meso- thelial monolayers and to migrate in in vitro assays [Figure 5D]. To confirm the findings of our in vitro studies, we firstly evaluated anoikis in floating EOC cells in short- term in vivo assays. Based on our prior experience using i.p. xenograft models derived from SKOV3ip cells [10,14], the earliest time-point at which solid peritoneal implants can be detected in mice is approxi- mately 10 days following tumor cell inoculation. To assay anoikis in floating EOC cells prior to implant- ation, groups of mice were inoculated with equivalent Ko and Naora Molecular Cancer 2014, 13:170 Page 7 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 4 Inhibition of P-cadherin in HOXA9-overexpressing EOC cells abrogates the stimulatory effects of HOXA9 on cell aggregation, survival, implantation and migration. (A) Western blot analysis of P-cadherin levels in vector-control and HOXA9-transfected SKOV3-Par lines, and in HOXA9-transfected SKOV3-Par lines that stably expressed non-targeting shRNA and shRNAs targeting two different sites within the CDH3 gene (shPCAD-A, shPCAD-B). (B,C) Cells of SKOV3-Par lines were incubated as suspension cultures in polyHEMA-coated plates for 3 days. (B) Cell morphology viewed by phase-contrast light microscopy. Bar, 50 μm. (C) Evaluation of cell death by flow cytometric analysis of 7AAD staining. (D) Evaluation of cell death by ELISA in suspension cultures of SKOV3-Par cells, attachment of SKOV3-Par cells to confluent mesothelial cell monolayers, and migration of SKOV3-Par cells were assayed as described in Figures 2D, E and G, respectively. Shown in D are mean + sd values of three independent assays. Figure 5 Reconstitution of P-cadherin in HOXA9-knockdown EOC cells restores the stimulatory effects of HOXA9 on cell survival, implantation and migration in vitro. (A) Western blot analysis of P-cadherin levels in HOXA9+ control SKOV3ip cells (non-targeting), HOXA9-knockdown SKOV3ip cells (shA9-B) and HOXA9-knockdown SKOV3ip cells that stably expressed P-cadherin (shA9-B + PCAD). (B) Morphology of SKOV3ip cells following incubation as suspension cultures in polyHEMA-coated plates for 3 days. Bar, 100 μm. (C) Evaluation of cell death in suspension cultures by flow cytometric analysis of 7AAD staining. (D) Evaluation of cell death by ELISA in suspension cultures of SKOV3ip cells, attachment of SKOV3ip cells to confluent mesothelial cell monolayers, and migration of SKOV3ip cells. Shown in D are mean + sd values of three independent assays. Ko and Naora Molecular Cancer 2014, 13:170 Page 8 of 13 http://www.molecular-cancer.com/content/13/1/170 numbers of non-targeting, shA9-B and shA9-B + PCAD subsequently developed on the peritoneal cavity wall in SKOV3ip cells. At 3 days thereafter, GFP-expressing EOC the shA9-B + PCAD group were significantly higher than cells in the peritoneal fluid were evaluated for cell death. the numbers of implants in the shA9-B group (P <0.01) The proportion of floating EOC cells that exhibited cell and almost equivalent to the numbers of implants in the death in the shA9-B + PCAD model was markedly lower control group [Figure 6B]. Similar results were obtained than that in the shA9-B model and similar to that in the when we evaluated the numbers of mesenteric implants control model [Figure 6A]. The numbers of implants that in these groups of mice [Figure 6C]. Tumor implants that Figure 6 Reconstitution of P-cadherin in HOXA9-knockdown EOC cells restores the stimulatory effects of HOXA9 on cell survival, implantation and migration in vivo. Female nude mice were inoculated i.p. with equivalent numbers (2 × 10 ) of cells of control (non-targeting) and HOXA9-knockdown (shA9-B) SKOV3ip lines and HOXA9-knockdown SKOV3ip cells that stably expressed P-cadherin (shA9-B + PCAD). In (A),mice were sacrificed at 3 days following tumor cell inoculation. Floating cells in the peritoneal cavity were collected and stained with 7AAD. Cell death within the gated population of GFP+ tumor cells was evaluated by flow cytometric analysis of 7AAD staining. In (B-D), mice were analyzed at 3 weeks following tumor cell inoculation. (B) Representative examples of GFP-expressing tumor implants (shown in green) attached to the peritoneal cavity wall as viewed by fluorescence microscopy. Bar, 1 mm. Numbers of implants on the peritoneal cavity wall were counted in five random 25 mm fields per mouse. (C) Representative examples of HE-stained sections of bowel tissues with mesenteric implants. Bar, 1 mm. Numbers of mesenteric implants were counted in three random 1 cm fields per mouse. (D) Representative examples of HE-stained sections of diaphragmatic implants. Bar 100 μm. Ko and Naora Molecular Cancer 2014, 13:170 Page 9 of 13 http://www.molecular-cancer.com/content/13/1/170 derived from shA9-B + PCAD cells and formed on the abolished the stimulatory effect of HOXA9 on aggre- diaphragm, bowel serosa and cavity wall were found to gation, whereas aggregation was almost completely be as invasive as control tumors and to exhibit greater restored when P-cadherin was reconstituted in EOC invasive depth than implants that derived from shA9-B cells in which HOXA9 was inhibited. HOXA9 had little cells and formed at the same sites [Figure 6D; Additional effect on levels of other classical cadherins, but it is file 1: Figures S1A-C]. Together, our findings indicate not possible to exclude the possibility that HOXA9 that HOXA9 promotes the aggregation and survival of might also promote tumor cell aggregation by inducing floating EOC cells and also the peritoneal attachment expression of other types of adhesion molecules. The and invasiveness of EOC cells, and that these stimula- classical cadherins have well-characterized functions in tory effects of HOXA9 are largely due to its induction facilitating homotypic cell adhesion [26,27]. The sig- of P-cadherin expression. nificance of P-cadherin in mediating aggregation of floating EOC cells is supported by our recent study in Discussion which we demonstrated that inhibiting P-cadherin Substantial evidence indicates that aberrations in sig- reduces aggregation of ascitic EOC cells in xenograft naling pathways that control normal developmental models [14]. High expression of P-cadherin has been patterning play pivotal roles in driving tumorigenesis found to be associated with poor overall survival of [16-18]. Homeoproteins comprise another important EOC patients [12]. A study by Patel and colleagues class of patterning regulators and have been increas- identified that P-cadherin is the predominant type of ingly found to be aberrantly expressed in a variety of cadherin expressed in tumor cells in the peritoneal cancers [6,7]. Several homeoproteins promote tumor fluid of EOC patients [11]. Sivertsen and colleagues cell proliferation by deregulating transcription of genes reported that P-cadherin is frequently expressed in that control cell cycle progression and autocrine effusion specimens of EOC patients (51 of 53 cases), growth-stimulatory pathways [19-21]. There is also but also found that E-cadherin and N-cadherin are evidence that some homeoproteins promote metastasis. expressed at similar frequency [28]. There is discordance For example, SIX1 induces epithelial-to-mesenchymal in the literature regarding the expression of E-cadherin transition (EMT) in breast cancer cells by activating and N-cadherin in effusion specimens of EOC patients. transcription of the gene encoding the TGF-β type I Two studies have reported that levels of E-cadherin and receptor [22] and promotes metastasis of rhabdomyo- N-cadherin are decreased in effusions as compared to sarcoma by inducing expression of the cytoskeletal pro- solid tumors [11,29], whereas a third study identified tein ezrin [23]. Both normal patterning and tumor that E-cadherin levels were higher in effusions than in progression involve dynamic changes in homotypic and matching primary solid tumors [30]. heterotypic cellular interactions, but surprisingly only Although it is widely thought that aggregation enables few genes that control cell adhesion have been identi- floating tumor cells to escape anoikis, the function of fied as homeoprotein targets [24,25]. We previously cadherins in anoikis appears to be cell-specific. Inhib- identified that high HOXA9 expression is strongly asso- ition of E-cadherin increases anoikis sensitivity in Ewing ciated with poor overall survival in EOC patients and sarcoma cells [31], but increases anoikis resistance in promotes tumor progression in EOC xenograft models mammary epithelial cells [32]. Our recent studies sup- [10]. The findings of this present study indicate that the port the notion that P-cadherin primarily inhibits anoikis CDH3 gene which encodes P-cadherin is a transcrip- in EOC cells by promoting cell aggregation. Inhibition of tional target of HOXA9 and that HOXA9 promotes P-cadherin increased cell death in suspension cultures both homotypic and heterotypic cell interactions that but not in adherent cultures of EOC cells, and increased facilitate intraperitoneal dissemination of EOC through cell death in ascitic EOC cells but not in solid tumor xenografts [14]. The possibility that P-cadherin might its induction of P-cadherin. The propensity for intraperitoneal ‘seeding’ is a hall- promote cell survival independently of cell aggregation mark of EOC and depends on adaptation by EOC cells cannot be excluded. Cell adhesion is facilitated by in- teractions between the extracellular domains of cad- to two dynamic changes in their microenvironment. The first change occurs when EOC cells are exfoliated and herin molecules on adjacent cells [26]. Intracellular transported by the flow of the peritoneal fluid. It is signal transduction triggered by P-cadherin involves activation of Rac1 and Cdc42 Rho GTPases [33], but it widely thought that aggregation enables floating EOC cells to escape anoikis [3,4], but the mechanisms that has been found that Rac1 and Cdc42 activation alone facilitate this aggregation are poorly understood. The does not fully account for the ability of P-cadherin to suppress anoikis [14]. There is evidence that EMT present study indicates that HOXA9 promotes aggrega- tion of floating EOC cells by inducing expression of confers anoikis resistance in some cell types [32]. How- P-cadherin. Inhibiting P-cadherin in HOXA9+ EOC cells ever, we have found that P-cadherin does not alter Ko and Naora Molecular Cancer 2014, 13:170 Page 10 of 13 http://www.molecular-cancer.com/content/13/1/170 expression of EMT-promoting transcription factors in function. Cadherin-mediated cell adhesion is essential EOC cells [14]. We also cannot exclude the possibility for maintaining the structure and function of repro- that HOXA9 inhibit anoikis by a mechanism that is ductive tissues [43], but mice with targeted disruption independent of P-cadherin. Because TGF-β signaling of the Cdh3 or Hoxa9 genes are fertile [44,45]. The induces EMT [34] and HOXA9 induces TGF-β2ex- mechanisms that induce HOXA9 expression in EOC pression in EOC cells [10], HOXA9 might inhibit anoikis arealsopoorlyunderstood. Onepossiblemechanism is by promoting EMT. However, we found that HOXA9 deregulated promoter methylation. HOXA9 promoter neither represses E-cadherin expression [Figure 3A] nor methylation has been detected at significantly lower alters expression of EMT-promoting transcription factors frequency in advanced-stage EOC than in early-stage in EOC cells, including those that are induced by TGF-β tumors [46]. Another possible mechanism is histone signaling [10]. Autocrine TGF-β signaling appears to be modification. Polycomb group (PcG) proteins form impaired in EOC cells that express HOXA9 [10] and this complexes that dynamically alter chromatin structure impairment could stem from the ability of HOXA9 to by modifying specific residues in histone tails and PcG- block TGF-β-induced Smad-dependent transcription [35]. mediated repression is a principal mechanism by which Collectively, these findings strongly suggest that the ability HOX gene expression is tightly regulated during devel- of HOXA9 to inhibit anoikis in EOC cells is largely attrib- opment [47]. AKT signaling has been reported to sup- utable to its induction of P-cadherin and the facilitation of press the histone methyltransferase activity of the PcG EOC cell aggregation by P-cadherin. protein EZH2 [48], and is frequently activated in high- The second important rate-limiting step in the pro- grade, advanced-stage EOCs [49]. It is therefore pos- gression of EOC is the implantation of floating tumor sible that AKT activation in EOC derepresses HOXA9 cells on to peritoneal surfaces. Interactions of EOC expression by inhibiting EZH2 activity. Interestingly, cells with peritoneal mesothelial cells and/or the sub- HOXA9 is not endogenously expressed in the parental mesothelial extracellular matrix are mediated by vari- SKOV3 cell line but in the aggressive subclone SKOV3ip ous cell surface molecules including CD44 and several [Figure 3A]. The expression of HOXA9 in SKOV3ip cells integrins [36,37]. We previously identified that HOXA10, might be attributable to SKOV3ip cells having higher a homeoprotein that is expressed in a subset of EOCs, can ERBB2 levels than the parental SKOV3 cells [15]. promote interactions between EOC cells and peritoneal mesothelial cells by inducing expression of αvβ3 integrin Conclusion [38]. The notion that P-cadherin facilitates tumor- The present study supports increasing evidence of the peritoneum interactions is supported by findings that functional significance of homeoproteins in dictating P-cadherin is frequently expressed in malignant periton- the clinical behavior of tumors through controlling eal effusions of EOC patients and is the most predominant transcription of distinct sets of target genes. Because type of cadherin in normal peritoneal tissues [11,28,39]. homeoproteins are transcription factors and share The findings of the present study indicate that the ability tracts of homology, it is challenging to therapeutically of HOXA9 to promote interactions between EOC cells target these proteins. However, elucidating the mecha- and peritoneal mesothelial cells is primarily mediated nisms of homeoproteins and their downstream effec- by its induction of P-cadherin. Inhibiting P-cadherin in tors in tumors can yield important insights into more HOXA9+ EOC cells abrogated the stimulatory effect of effective therapeutic strategies. Inhibiting P-cadherin is HOXA9 on mesothelial cell attachment, and reconstitu- particularly promising as a human monoclonal Ab to tion of P-cadherin in HOXA9-knockdown cells restored P-cadherin has been developed and is undergoing clin- attachment capability. Our work also demonstrates that ical trials [50]. HOXA9 promotes migratory potential of EOC cells via its induction of P-cadherin. The function of P-cadherin in cell migration varies depending on the type of cell. Methods It has been reported that P-cadherin inhibits migration Abs and plasmids of melanoma and oral squamous cell carcinoma cells Abs to HOXA9 were purchased from Millipore (for [40,41], but promotes migration of pancreatic and blad- Western blot) and Santa Cruz Biotechnology (for chro- der cancer cells [33,42]. Our findings support prior matin IP). Abs to P-cadherin were purchased from BD Bio- reports that P-cadherin stimulates migratory potential sciences (for Western blot) and Abcam (for neutralization). of EOC cells [13,14]. Other Abs were as follows: E-cadherin (Invitrogen), Because the precise mechanism of HOXA9 in con- N-cadherin, active caspase-3 (BD Biosciences), actin, trolling female reproductive development is not secondary Abs (Sigma-Aldrich). HOXA9 cDNA was known, it is unclear how its induction of P-cadherin in provided by Corey Largman (Veterans Affairs Medical EOC might be related to its normal developmental Center, San Francisco). CDH3 cDNA and pGFP-V-RS Ko and Naora Molecular Cancer 2014, 13:170 Page 11 of 13 http://www.molecular-cancer.com/content/13/1/170 plasmids containing HOXA9, CDH3 and non-targeting described [38]. Where indicated, EOC cells were incu- shRNAs were purchased from OriGene Technologies. bated with neutralizing P-cadherin Ab or with control IgG at a final concentration of 10 μg/ml. At 3 days there- Cell culture and transfection after, cell morphology was viewed by phase-contrast light SKOV3ip cell lines that stably express non-targeting microscopy. Cells were evaluated for cell death by staining and HOXA9 shRNAs have been previously described with 7AAD (Sigma-Aldrich) and with Ab to active caspase- [10]. The parental SKOV3 cell line was purchased from 3. Staining was detected by flow cytometry (FACS Calibur, American Type Culture Collection. Cell lines were BD Biosciences). Cell death was also evaluated by assaying authenticated by short tandem repeat analysis performed mono- and oligo- nucleosomes in cell lysates by using the by the MD Anderson Cancer Center Characterized Cell Cell Death Detection ELISA kit (Roche). Three independ- Line Core Facility. SKOV3ip and SKOV3 cells were ent experiments were performed for each assay. cultured in McCoys 5A medium (Invitrogen) supple- mented with 10% FBS and penicillin-streptomycin. Cells Cell attachment assays were transfected by using Lipofectamine 2000 reagent Single cell suspensions of GFP-expressing EOC cells were (Invitrogen) and selected by the addition of puromycin seeded in 96-well plates (1.5 × 10 per well) containing (0.5 μg/ml). Primary cultures of normal human omental confluent monolayers of normal omental mesothelial cells mesothelial cells have been previously described [51] as previously described [38]. At 1 hr after seeding of EOC and were provided by Ernst Lengyel (University of Chicago, cells, wells were washed with culture medium to remove Chicago, IL). unattached EOC cells. Attached EOC cells were viewed by fluorescence microscopy using a fluorescein filter. Three Western blot and qRT-PCR analyses independent experiments were performed in which Cell lysates were prepared by using M-PER buffer attached cells were counted in three random 100× (Pierce Biotechnology), separated by SDS-PAGE and microscopic fields in each experiment. transferred to polyvinylidene difluoride membranes (GE Healthcare). CDH3 transcripts were analyzed by using Cell migration assays SYBR®Green qPCR Master Mix (SABiosciences) and the EOC cells were seeded as single cell suspensions in the following primers: forward: 5′-CAGGTGCTGAACATC upper chamber of 24-well transwell chambers (BD ACGGACA-3′, reverse: 5′-CTTCAGGGACAAGACCA Biosciences) (5 × 10 cells per well). At 6 h thereafter, CTGTG-3′. RPL32 transcript levels were used as migrating cells were stained with Giemsa solution. controls for normalization and were detected by using Three independent experiments were performed in the following primers: forward: 5′-ACAAAGCACA which migrating cells were counted in three random TGCTGCCCAGTG-3′ reverse: 5′-TTCCACGATGG 100x microscopic fields in each experiment. CTTTGCGGTTC-3′. Mouse i.p. xenograft studies Chromatin IP Four-week-old female nude mice (purchased from the Na- Chromatin IP assays were performed by using the tional Cancer Institute, Frederick, MD) were inoculated i. EZ-ChIP Assay kit (Millipore). Sheared chromatin was p. with 2 × 10 cells of GFP-expressing EOC lines (n = 5 incubated overnight with 1 μg of HOXA9 Ab. DNA was mice per group). Mice were euthanized by CO asphyxi- purified from precipitated complexes. Fragments of the ation at time points indicated in the figure legends. human CDH3 promoter were amplified by using the fol- GFP-expressing tumor implants were visualized under a lowing primers: for Site A (150 bp), forward: 5′-CCCCCA Leica MZML III stereomicroscope equipped with a mer- CCCTCGCAACGCAAGCAA-3′, reverse: 5′-TGGGATT cury lamp power supply and GFP filter set. Formalin- ACAGGCGTGAGAAACG-3′; for Site B (104 bp), fixed, paraffin-embedded tissue sections were stained with forward: 5′-GCTGTGTGCCAGATATGATGTTTAG-3′, hematoxylin-eosin (HE) and analyzed under a light micro- reverse: 5′-CTGTCAAATGGGGCTGTTATGATC-3′.A scope. Invasive depth of implants was measured using a 166 bp fragment of the GAPDH gene was amplified as an stage micrometer slide in five random microscopic fields irrelevant control by using the following primers: forward: of tissue sections of each mouse at the magnifications 5′-TACTAGCGGTTTTACGGGCG-3′,reverse: 5′-TCG indicated in the figure legends. Floating cells were col- AACAGGAGGAGCAGAGAGCGA-3′. lected from peritoneal fluid, stained with Hoechst dye and viewed by fluorescence microscopy. Floating cells were Cell death assays also stained with 7AAD and cell clusters gently disaggre- EOC cells were seeded in 96-well plates (1 × 10 cells per gated by passing through 35 μm nylon mesh. Immediately well) that were coated with polyHEMA (Sigma-Aldrich) thereafter, 7AAD staining was analyzed by flow cytometry to block cell attachment to substratum as previously within the gated population of GFP + tumor cells. Ko and Naora Molecular Cancer 2014, 13:170 Page 12 of 13 http://www.molecular-cancer.com/content/13/1/170 Bioinformatic analysis cadherin; PcG: Polycomb group; PolyHEMA: Poly (2-hydroxyethyl methacrylate); shRNAs: Short hairpin RNAs; TGF: Transforming growth factor. Gene expression data of EOC cases from the TCGA pro- ject (n = 567) were downloaded from the TCGA data Competing interest portal site (http://tcga-data.nci.nih.gov/tcga/). Where there The authors have declared that no conflict of interest exists. were multiple probe sets for an individual gene, the mean value for the given gene for each case was used. Patients Authors’ contribution were stratified according to the level of HOXA9 expression SYK and HN designed and performed experiments and wrote the in tumors, where HOXA9 transcript levels were defined as manuscript. Both authors read and approved the final manuscript. High (≥ upper quartile) and Low (≤ lower quartile) as pre- viously described [10]. Acknowledgements This work was supported by Cancer & Prevention Research Institute of Texas grant RP120390 (to H. Naora) and U.S. National Institutes of Health grant Statistical analysis CA141078 (to H. Naora). We thank Nicolas Barengo for technical assistance and members of the Naora laboratory for helpful discussions. Statistical analysis was performed by using STATISTICA6 software (StatSoft Inc.). Statistical significance of data of Received: 22 March 2014 Accepted: 4 July 2014 in vitro and in vivo assays was assessed by unpaired two- Published: 14 July 2014 tailed Student’s t-test. Data represent mean ± s.d. Signifi- cance of differences in gene expression between groups of References 1. Siegel R, Naishadham D, Jemal A: Cancer statistics, 2013. CA Cancer J Clin patients was assessed by Mann–Whitney U-test. P values 2013, 63:11–30. of < 0.05 were considered significant. 2. Naora H, Montell DJ: Ovarian cancer metastasis: integrating insights from disparate model organisms. Nat Rev Cancer 2005, 5:355–366. 3. Lengyel E: Ovarian cancer development and metastasis. Am J Pathol 2010, Additional files 177:1053–1064. 4. Sodek KL, Murphy KJ, Brown TJ, Ringuette MJ: Cell-cell and cell-matrix dynamics in intraperitoneal cancer metastasis. Cancer Metastasis Rev 2012, Additional file 1: Figure S1. Invasiveness of EOC cells in i.p. xenograft 31:397–414. models. Female nude mice (n=5 per group) were inoculated i.p. with 5. Burleson KM, Casey RC, Skubitz KM, Pambuccian SE, Oegema TR Jr, Skubitz equivalent numbers of cells (2 × 10 ) of SKOV3ip lines and sacrificed at 3 AP: Ovarian carcinoma ascites spheroids adhere to extracellular matrix weeks thereafter. Invasive depth of implants on the bowel, diaphragm and components and mesothelial cell monolayers. Gynecol Oncol 2004, peritoneal cavity wall was measured in five random microscopic fields of 93:170–181. HE-stained tissue sections of each of these sites in each mouse. An average 6. Abate-Shen C: Deregulated homeobox gene expression in cancer: cause invasive depth was calculated for each site of each mouse. (A) Depth of or consequence? Nat Rev Cancer 2002, 2:777–785. superficial bowel serosa invasion (evaluated at 200× magnification) and 7. Samuel S, Naora H: Homeobox gene expression in cancer: insights from representative examples of HE-stained sections. Bar, 50 μm. (B) Depth of developmental regulation and deregulation. Eur J Cancer 2005, invasion of diaphragmatic implants into adjacent muscle (evaluated at 41:2428–2437. 100x magnification). Representative examples of HE-stained sections of 8. Pearson JC, Lemons D, McGinnis W: Modulating Hox gene functions diaphragmatic implants are shown in Figure 1D. (C) Invasive depth of during animal body patterning. Nat Rev Genet 2005, 6:893–904. implants on the peritoneal cavity wall (evaluated at 100× magnification) 9. Taylor HS, Vanden Heuvel GB, Igarashi P: A conserved Hox axis in the and representative examples of HE-stained sections. Bar, 100 μm. mouse and human female reproductive system: late establishment and Additional file 2: Figure S2. Quantification of protein levels. Protein persistent adult expression of the Hoxa cluster genes. Biol Reprod 1997, levels were evaluated by measuring intensity of bands on Western blots 57:1338–1345. shown in Figures 3A, 4A and 5A using the TINA 20 program (Raytest). 10. Ko SY, Barengo N, Ladanyi A, Lee JS, Marini F, Lengyel E, Naora H: (A) Levels of an individual protein in SKOV3ip lines transfected with HOXA9 promotes ovarian cancer growth by stimulating cancer-associated non-targeting and HOXA9 shRNAs are expressed relative to its level in fibroblasts. J Clin Invest 2012, 122:3603–3617. the empty vector control SKOV3ip line. Levels of an individual protein in 11. Patel IS, Madan P, Getsios S, Bertrand MA, MacCalman CD: Cadherin the HOXA9-transfected SKOV3-Par line are expressed relative to its level switching in ovarian cancer progression. Int J Cancer 2003, 106:172–177. in the empty vector control SKOV3-Par line. (B) Levels of P-cadherin in 12. Quattrocchi L, Green AR, Martin S, Durrant L, Deen S: The cadherin switch HOXA9-transfected SKOV3-Par lines that were co-transfected with no in ovarian high-grade serous carcinoma is associated with disease shRNA, non-targeting shRNA or CDH3 shRNAs are expressed relative to progression. Virchows Arch 2011, 459:21–29. its level in the empty vector control SKOV3-Par line. (C) Levels of P- 13. Cheung LW, Leung PC, Wong AS: Cadherin switching and activation of cadherin in HOXA9-knockdown SKOV3ip cells and HOXA9-knockdown p120 catenin signaling are mediators of gonadotropin-releasing hormone SKOV3ip cells that stably expressed P-cadherin are expressed relative to to promote tumor cell migration and invasion in ovarian cancer. Oncogene its level in SKOV3ip cells expressing non-targeting shRNA. 2010, 29:2427–2440. Additional file 3: Figure S3. Effects of P-cadherin Ab on 14. Usui A, Ko SY, Barengo N, Naora H: P-cadherin promotes ovarian cancer HOXA9-overexpressing EOC cells. Cells of vector-control and dissemination through tumor cell aggregation and tumor-peritoneum HOXA9-transfected SKOV3-Par lines were incubated as suspension tnteractions. Mol Cancer Res 2014, 12:504–513. cultures in polyHEMA-coated plates for 3 days with the addition of 15. Yu D, Wolf JK, Scanlon M, Price JE, Hung MC: Enhanced c-erbB-2/neu neutralizing P-cadherin Ab or control IgG. (A) Cell morphology expression in human ovarian cancer cells correlates with more severe viewed by phase-contrast microscopy. Bar 50 μm. (B) Cell death was malignancy that can be suppressed by E1A. Cancer Res 1993, 53:891–898. evaluatedby assayingmono- andoligo-nucleosomes in cell lysates 16. Pasca di Magliano M, Hebrok M: Hedgehog signalling in cancer formation by ELISA. Shown are mean + sd values of three independent experiments. and maintenance. Nat Rev Cancer 2003, 3:903–911. 17. Klaus A, Birchmeier W: Wnt signalling and its impact on development and cancer. Nat Rev Cancer 2008, 8:387–398. Abbrevations 18. Ranganathan P, Weaver KL, Capobianco AJ: Notch signalling in solid 7AAD: 7-amino actinomycin D; Ab: Antibody; EOC: Epithelial ovarian cancer; tumours: a little bit of everything but not all the time. Nat Rev Cancer ECAD: E-cadherin; HE: Hematoxylin-eosin; NCAD: N-cadherin; PCAD: P- 2011, 11:338–351. Ko and Naora Molecular Cancer 2014, 13:170 Page 13 of 13 http://www.molecular-cancer.com/content/13/1/170 19. Care A, Silvani A, Meccia E, Mattia G, Stoppacciaro A, Parmiani G, Peschle C, 41. Bauer K, Dowejko A, Bosserhoff AK, Reichert TE, Bauer RJ: P-cadherin Colombo MP: HOXB7 constitutively activates basic fibroblast growth induces an epithelial-like phenotype in oral squamous cell carcinoma factor in melanomas. Mol Cell Biol 1996, 16:4842–4851. by GSK-3beta-mediated Snail phosphorylation. Carcinogenesis 2009, 20. Coletta RD, Christensen K, Reichenberger KJ, Lamb J, Micomonaco D, 30:1781–1788. Huang L, Wolf DM, Muller-Tidow C, Golub TR, Kawakami K, Ford HL: 42. Mandeville JA, Silva Neto B, Vanni AJ, Smith GL, Rieger-Christ KM, Zeheb R, The Six1 homeoprotein stimulates tumorigenesis by reactivation of Loda M, Libertino JA, Summerhayes IC: P-cadherin as a prognostic cyclin A1. Proc Natl Acad Sci U S A 2004, 101:6478–6483. indicator and a modulator of migratory behaviour in bladder 21. Trinh BQ, Barengo N, Naora H: Homeodomain protein DLX4 counteracts carcinoma cells. BJU Int 2008, 102:1707–1714. key transcriptional control mechanisms of the TGF-beta cytostatic 43. Rowlands TM, Symonds JM, Farookhi R, Blaschuk OW: Cadherins: crucial program and blocks the antiproliferative effect of TGF-beta. Oncogene regulators of structure and function in reproductive tissues. Rev Reprod 2011, 30:2718–2729. 2000, 5:53–61. 22. Micalizzi DS, Wang CA, Farabaugh SM, Schiemann WP, Ford HL: 44. Lawrence HJ, Helgason CD, Sauvageau G, Fong S, Izon DJ, Humphries RK, Homeoprotein Six1 increases TGF-beta type I receptor and converts Largman C: Mice bearing a targeted interruption of the homeobox gene TGF-beta signaling from suppressive to supportive for tumor growth. HOXA9 have defects in myeloid, erythroid, and lymphoid hematopoiesis. Cancer Res 2010, 70:10371–10380. Blood 1997, 89:1922–1930. 45. Radice GL, Ferreira-Cornwell MC, Robinson SD, Rayburn H, Chodosh LA, 23. Yu Y, Davicioni E, Triche TJ, Merlino G: The homeoprotein six1 Takeichi M, Hynes RO: Precocious mammary gland development in P- transcriptionally activates multiple protumorigenic genes but requires cadherin-deficient mice. J Cell Biol 1997, 139:1025–1032. ezrin to promote metastasis. Cancer Res 2006, 66:1982–1989. 46. Wu Q, Lothe RA, Ahlquist T, Silins I, Trope CG, Micci F, Nesland JM, Suo Z, 24. Daftary GS, Troy PJ, Bagot CN, Young SL, Taylor HS: Direct regulation of Lind GE: DNA methylation profiling of ovarian carcinomas and their beta3-integrin subunit gene expression by HOXA10 in endometrial cells. in vitro models identifies HOXA9, HOXB5, SCGB3A1, and CRABP1 as Mol Endocrinol 2002, 16:571–579. novel targets. Mol Cancer 2007, 6:45. 25. Zhu R, Wong KF, Lee NP, Lee KF, Luk JM: HNF1alpha and CDX2 47. Soshnikova N, Duboule D: Epigenetic regulation of vertebrate Hox genes: transcriptional factors bind to cadherin-17 (CDH17) gene promoter and a dynamic equilibrium. Epigenetics 2009, 4:537–540. modulate its expression in hepatocellular carcinoma. J Cell Biochem 2010, 48. Cha TL, Zhou BP, Xia W, Wu Y, Yang CC, Chen CT, Ping B, Otte AP, Hung 111:618–626. MC: Akt-mediated phosphorylation of EZH2 suppresses methylation of 26. Gumbiner BM: Regulation of cadherin-mediated adhesion in lysine 27 in histone H3. Science 2005, 310:306–310. morphogenesis. Nat Rev Mol Cell Biol 2005, 6:622–634. 49. Yuan ZQ, Sun M, Feldman RI, Wang G, Ma X, Jiang C, Coppola D, Nicosia SV, 27. Halbleib JM, Nelson WJ: Cadherins in development: cell adhesion, sorting, Cheng JQ: Frequent activation of AKT2 and induction of apoptosis by and tissue morphogenesis. Genes Dev 2006, 20:3199–3214. inhibition of phosphoinositide-3-OH kinase/Akt pathway in human 28. Sivertsen S, Berner A, Michael CW, Bedrossian C, Davidson B: Cadherin ovarian cancer. Oncogene 2000, 19:2324–2330. expression in ovarian carcinoma and malignant mesothelioma cell 50. Zhang CC, Yan Z, Zhang Q, Kuszpit K, Zasadny K, Qiu M, Painter CL, Wong effusions. Acta Cytol 2006, 50:603–607. A, Kraynov E, Arango ME, Mehta PP, Popoff I, Casperson GF, Los G, Bender S, 29. Veatch AL, Carson LF, Ramakrishnan S: Differential expression of the Anderes K, Christensen JG, VanArsdale T: PF-03732010: a fully human cell-cell adhesion molecule E-cadherin in ascites and solid human monoclonal antibody against P-cadherin with antitumor and ovarian tumor cells. Int J Cancer 1994, 58:393–399. antimetastatic activity. Clin Cancer Res 2010, 16:5177–5188. 30. Davidson B, Berner A, Nesland JM, Risberg B, Berner HS, Trope CG, 51. Kenny HA, Krausz T, Yamada SD, Lengyel E: Use of a novel 3D culture model Kristensen GB, Bryne M, Florenes AV: E-cadherin and alpha-, beta-, and to elucidate the role of mesothelial cells, fibroblasts and extra-cellular gamma-catenin protein expression is up-regulated in ovarian carcinoma matrices on adhesion and invasion of ovarian cancer cells to the cells in serous effusions. J Pathol 2000, 192:460–469. omentum. Int J Cancer 2007, 121:1463–1472. 31. Kang HG, Jenabi JM, Zhang J, Keshelava N, Shimada H, May WA, Ng T, Reynolds CP, Triche TJ, Sorensen PH: E-cadherin cell-cell adhesion in doi:10.1186/1476-4598-13-170 ewing tumor cells mediates suppression of anoikis through activation of Cite this article as: Ko and Naora: HOXA9 promotes homotypic and the ErbB4 tyrosine kinase. Cancer Res 2007, 67:3094–3105. heterotypic cell interactions that facilitate ovarian cancer dissemination 32. Kumar S, Park SH, Cieply B, Schupp J, Killiam E, Zhang F, Rimm DL, Frisch via its induction of P-cadherin. Molecular Cancer 2014 13:170. SM: A pathway for the control of anoikis sensitivity by E-cadherin and epithelial-to-mesenchymal transition. Mol Cell Biol 2011, 31:4036–4051. 33. Taniuchi K, Nakagawa H, Hosokawa M, Nakamura T, Eguchi H, Ohigashi H, Ishikawa O, Katagiri T, Nakamura Y: Overexpressed P-cadherin/CDH3 promotes motility of pancreatic cancer cells by interacting with p120ctn and activating rho-family GTPases. Cancer Res 2005, 65:3092–3099. 34. Thiery JP, Acloque H, Huang RY, Nieto MA: Epithelial-mesenchymal transitions in development and disease. Cell 2009, 139:871–890. 35. Shi X, Bai S, Li L, Cao X: Hoxa-9 represses transforming growth factor-beta-induced osteopontin gene transcription. J Biol Chem 2001, 276:850–855. 36. Lessan K, Aguiar DJ, Oegema T, Siebenson L, Skubitz AP: CD44 and beta1 integrin mediate ovarian carcinoma cell adhesion to peritoneal mesothelial cells. Am J Pathol 1999, 154:1525–1537. Submit your next manuscript to BioMed Central 37. Heyman L, Kellouche S, Fernandes J, Dutoit S, Poulain L, Carreiras F: and take full advantage of: Vitronectin and its receptors partly mediate adhesion of ovarian cancer cells to peritoneal mesothelium in vitro. Tumour Biol 2008, 29:231–244. • Convenient online submission 38. Ko SY, Lengyel E, Naora H: The Mullerian HOXA10 gene promotes growth • Thorough peer review of ovarian surface epithelial cells by stimulating epithelial-stromal interactions. Mol Cell Endocrinol 2010, 317:112–119. • No space constraints or color figure charges 39. Chen GT, Tai CT, Yeh LS, Yang TC, Tsai HD: Identification of the cadherin • Immediate publication on acceptance subtypes present in the human peritoneum and endometriotic lesions: potential role for P-cadherin in the development of endometriosis. • Inclusion in PubMed, CAS, Scopus and Google Scholar Mol Reprod Dev 2002, 62:289–294. • Research which is freely available for redistribution 40. Van Marck V, Stove C, Van Den Bossche K, Stove V, Paredes J, Vander Haeghen Y, Bracke M: P-cadherin promotes cell-cell adhesion and Submit your manuscript at counteracts invasion in human melanoma. Cancer Res 2005, 65:8774–8783. www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Molecular Cancer Springer Journals

HOXA9 promotes homotypic and heterotypic cell interactions that facilitate ovarian cancer dissemination via its induction of P-cadherin

Molecular Cancer , Volume 13 (1) – Jul 14, 2014

Loading next page...
 
/lp/springer-journals/hoxa9-promotes-homotypic-and-heterotypic-cell-interactions-that-ZR0OuVQnXT

References (51)

Publisher
Springer Journals
Copyright
Copyright © 2014 by Ko and Naora; licensee BioMed Central Ltd.
Subject
Biomedicine; Cancer Research; Oncology
eISSN
1476-4598
DOI
10.1186/1476-4598-13-170
pmid
25023983
Publisher site
See Article on Publisher Site

Abstract

Background: Epithelial ovarian cancer (EOC) is a lethal disease that frequently involves the peritoneal cavity. Dissemination of EOC is a multi-step process in which exfoliated tumor cells survive in the peritoneal fluid as multi-cellular aggregates and then form invasive implants on peritoneal surfaces. The mechanisms that control this process are poorly understood. We previously identified that high expression of the developmental patterning gene HOXA9 is associated with poor survival in EOC patients. In this study, we investigated the significance and mechanisms of HOXA9 in controlling aggregation and implantation of floating EOC cells. Methods: HOXA9 was inhibited by shRNAs or expressed in EOC cells that were propagated in suspension cultures and in the peritoneal cavity of mice. Cell death was assayed by flow cytometry and ELISA. Cell aggregation, attachment and migration were evaluated by microscopy, transwell chamber assays and histopathologic analysis. DNA-binding of HOXA9 and its effect on expression of the cell adhesion molecule P-cadherin were assayed by chromatin immunoprecipitation, quantitative RT-PCR and Western blot. HOXA9 and P-cadherin expression was evaluated in publicly available datasets of EOC clinical specimens. Results: We identified that HOXA9 promotes aggregation and inhibits anoikis in floating EOC cells in vitro and in xenograft models. HOXA9 also stimulated the ability of EOC cells to attach to peritoneal cells and to migrate. HOXA9bound thepromoterofthe CDH3 gene that encodes P-cadherin, induced CDH3 expression in EOC cells, and was associated with increased CDH3 expression in clinical specimens of EOC. Inhibiting P-cadherin in EOC cells that expressed HOXA9 abrogated the stimulatory effects of HOXA9 on cell aggregation, implantation and migration. Conversely, these stimulatory effects of HOXA9 were restored when P-cadherin was reconstituted in EOC cells in which HOXA9 was inhibited. Conclusion: These findings indicate that HOXA9 contributes to poor outcomes in EOC in part by promoting intraperitoneal dissemination via its induction of P-cadherin. Keywords: Ovarian cancer, Homeobox gene, Metastasis, Cell adhesion, P-cadherin Background lymphatic routes, EOC cells typically spread by shed- More than 60% of women with epithelial ovarian can- ding into the peritoneal fluid which transports tumor cer (EOC) are diagnosed with advanced-stage disease cells, either as multi-cellular aggregates or as single that has disseminated throughout the peritoneal cavity cells, throughout the peritoneal cavity [2-4]. Subse- [1]. Despite advances in tumor debulking surgery and quently, EOC cells attach to the mesothelium-lined chemotherapy, patients with advanced-stage EOC have peritoneal surfaces, such as the cavity wall, diaphragm a 5-year survival rate of only 30% [1]. Whereas many other and omentum, where they form invasive implants [2-4]. types of solid tumors metastasize via hematogenous or This ‘seeding’ of the peritoneal cavity with tumor cells is often associated with ascites formation. Although it is unclear whether EOC cells detach from the primary * Correspondence: hnaora@mdanderson.org tumoras clustersor assingle cells that then assemble Department of Molecular & Cellular Oncology, University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA © 2014 Ko and Naora; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ko and Naora Molecular Cancer 2014, 13:170 Page 2 of 13 http://www.molecular-cancer.com/content/13/1/170 into clusters, multi-cellular aggregates of floating EOC shRNA-expressing SKOV3ip cell lines [10]. These EOC cells are increasingly regarded as ‘seeds of metastasis’ cell lines stably expressed GFP, enabling their detection that are able to escape anoikis and implant on to among host cells in ascites. Floating EOC cells in asci- peritoneal surfaces [3-5]. However, the tissue-specific tes of control xenograft models (that expressed empty mechanisms that facilitate the aggregation and implant- vector or non-targeting shRNA) were present as large ation of floating EOC cells and drive the unique clinical compact aggregates [Figure 1A]. In contrast, ascites behavior of this disease are poorly understood. collected from HOXA9-knockdown models (shA9-A, Homeobox genes encode transcription factors, com- shA9-B) contained smaller aggregates or single EOC monly termed homeoproteins, that play essential roles in cells [Figure 1A]. The aggregation of floating EOC cells controlling developmental patterning and are expressed is thought to enable these cells to escape anoikis [3,4]. in a tightly regulated temporal and tissue-specific man- Cell death was evaluated within the population of ner [6-8]. Many homeoproteins are aberrantly expressed GFP-expressing ascitic EOC cells by flow cytometric in a variety of malignancies, but their functional signifi- analysis of 7-amino actinomycin D (7AAD) staining. cance in tumor progression is poorly understood as only As showninFigure1B, theproportionofascitic EOC few bona fide target genes have been identified [6,7]. cells that exhibited cell death was substantially higher The homeobox gene HOXA9 is normally expressed in the HOXA9-knockdown models than in control during differentiation of the Müllerian ducts into the models. These observations raise the possibility that female reproductive tract [9]. We have identified that HOXA9 promotes aggregation of floating EOC cells high HOXA9 expression is strongly associated with and inhibits anoikis. poor overall survival of EOC patients [10]. Studies of mouse xenograft models revealed that expression of HOXA9 also increases EOC cell implantation and HOXA9 in EOC cells promotes growth of solid peri- invasiveness toneal implants by inducing normal peritoneal fibro- Floating EOC cells that are transported by the peritoneal blasts and mesenchymal stem cells to acquire features fluid frequently implant on the diaphragm, peritoneal of cancer-associated fibroblasts that in turn supported cavity wall, omentum and mesentery [2-4]. We evalu- tumor growth and angiogenesis [10]. This stimulatory ated sections of tissues collected from mice that were effect of HOXA9 on solid tumor growth was attributed inoculated with equivalent numbers of control and to its activation of the gene encoding transforming HOXA9-knockdown SKOV3ip cells. Whereas numer- growth factor-β2(TGF-β2) that acted in a paracrine ous mesenteric implants were detected in the control manner on stromal cells [10]. Because EOC cells in groups, fewer implants were detected in the HOXA9- solid tumors and in ascites have different biological knockdown groups [Figure 1C]. In EOC patients, implants behaviors and exist in different microenvironments, we tend to only invade the superficial bowel serosa and investigated the possibility that HOXA9 mediates other not the deeper layers [3]. Some superficial bowel serosa types of effects in free-floating EOC cells. In this study, invasion was observed in control mice, but was markedly we identified that HOXA9 promotes the assembly of reduced in the HOXA9-knockdown groups (P <0.05) floating EOC cells into multi-cellular aggregates and [Additional file 1: Figure S1A]. Implants on the dia- inhibits anoikis, and also stimulates tumor-peritoneum phragm extensively invaded adjacent muscle in the control interactions and tumor cell migration. These stimula- groups, whereas invasive depth of diaphragmatic implants tory effects of HOXA9 were found to be largely attrib- was significantly reduced in the HOXA9-knockdown utable to its induction of the cell adhesion molecule groups (P < 0.01) [Figure 1D; Additional file 1: Figure S1B]. P-cadherin that is encoded by the CDH3 gene, a tran- Similarly, implants on the peritoneal cavity wall were scriptional target of HOXA9. significantly less invasive in the HOXA9-knockdown groups than in the control groups (P < 0.01) [Additional Results file 1: Figure S1C]. These findings indicate that HOXA9 HOXA9 promotes aggregation and survival of floating not only promotes aggregation and survival of floating EOC cells in i.p. xenograft models EOC cells, but also increases the peritoneal implant- We previously identified that expression of HOXA9 in ation and invasiveness of these cells. EOC cells promotes growth of solid tumor xenografts, but does not stimulate proliferation of EOC cells in vitro HOXA9 promotes aggregation and inhibits anoikis of EOC [10]. Because the biological behavior of ascitic tumor cells in vitro cells markedly differs from that of solid tumors, we The development of tumor implants at distal sites investigated the effect of HOXA9 on floating EOC cells depends on the survival of free-floating EOC cells that in ascites. I.p. xenograft models were generated from escape anoikis by forming multi-cellular aggregates [3,4]. previously established HOXA9+ control and HOXA9 We investigated the possibility that HOXA9 promotes Ko and Naora Molecular Cancer 2014, 13:170 Page 3 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 1 Knockdown of HOXA9 inhibits aggregation and survival of ascitic EOC cells and decreases EOC cell implantation and invasiveness in i.p. xenograft models. Female nude mice (n = 5 per group) were inoculated i.p. with equivalent numbers (2 × 10 )of GFP-expressing HOXA9+ control (empty vector, non-targeting) and HOXA9-knockdown (shA9-A, shA9-B) SKOV3ip cells. Mice were sacrificed at 3 weeks thereafter. (A,B) Ascitic cells were collected and stained with Hoechst dye to visualize nuclei and with 7AAD to assay cell death. (A) Representative examples of Hoechst staining (shown in blue) of ascitic cells where GFP-expressing tumor cells (shown in green) were visualized by fluorescence microscopy. Bar, 50 μm. (B) Flow cytometric analysis of 7AAD staining within the gated population of GFP + ascitic tumor cells. (C,D) Representative examples of HE-stained sections of (C) bowel tissues with mesenteric implants (bar, 1 mm) and (D) diaphragmatic implants (bar, 100 μm). aggregation and survival of floating EOC cells in in vitro aggregates of loosely clustered cells [Figure 2A]. Cell assays, independently of implantation and of effects of death was observed in HOXA9-knockdown cells as host cells. Cells of control and HOXA9-knockdown detected by staining with 7AAD [Figure 2B]. In contrast, SKOV3ip lines were incubated as suspension cultures control cells were mostly viable [Figure 2B]. Evaluation of in plates coated with poly(2-hydroxyethyl methacrylate) active caspase 3 levels indicated higher levels of apoptotic (polyHEMA), an inert polymer that prevent cells cell death in HOXA9-knockdown cells than in control from adhering to substratum. Whereas control cells cells [Figure 2C]. The increased level of cell death in formed large aggregates, suspension cultures of HOXA9- HOXA9-knockdown cells was additionally confirmed by knockdown cells comprised of single cells and small assaying mono- and oligo- nucleosomes in cell lysates Ko and Naora Molecular Cancer 2014, 13:170 Page 4 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 2 Knockdown of HOXA9 inhibits aggregation of floating EOC cells, increases anoikis, decreases EOC-mesothelial cell interactions, and reduces EOC cell migration in vitro. (A-D) Cells of control and HOXA9-knockdown SKOV3ip lines were incubated as suspension cultures in polyHEMA-coated plates for 3 days and evaluated for changes in cell morphology and cell death. (A) Cell morphology viewed by phase-contrast light microscopy. Bar, 100 μm. The extent of cell death was evaluated by flow cytometric analysis of (B) 7AAD staining and (C) staining of active caspase-3. In (D), cell death was evaluated by assaying mono- and oligo- nucleosomes in cell lysates by ELISA. Shown are mean ± sd values of three independent experiments. Significance of differences was evaluated by Student t-test. (E,F) Equivalent numbers of control and HOXA9-knockdown SKOV3ip cells that stably expressed GFP were seeded onto confluent monolayers of normal human omental mesothelial cells. At 1 h thereafter, attached tumor cells were viewed by fluorescence microscopy and counted in three random 100× microscopic fields per assay. (E) Representative examples of SKOV3ip cells (shown in green) attached to mesothelial monolayers. Bar, 100 μm. (F) Mean ± sd values of three independent attachment assays. (G) Equivalent numbers of control and HOXA9-knockdown SKOV3ip cells were seeded in transwell chambers. At 6 h thereafter, migrating cells were counted in three random 100x microscopic fields per assay. Shown are mean ± sd values of three independent migration assays. Ko and Naora Molecular Cancer 2014, 13:170 Page 5 of 13 http://www.molecular-cancer.com/content/13/1/170 [Figure 2D]. Together, these observations indicate that levels of other classical cadherins [Figure 3A]. SKO- HOXA9 promotes the assembly of floating EOC cells V3ip is an aggressive subclone that originally derived into multi-cellular aggregates and inhibits anoikis. from the parental SKOV3 cell line (SKOV3-Par) [15]. In contrast to SKOV3ip cells, SKOV3-Par cells did not HOXA9 stimulates interactions between EOC cells and endogenously express HOXA9 or P-cadherin [Figure 3A]. peritoneal mesothelial cells and promotes EOC cell Stable expression of HOXA9 in SKOV3-Par cells induced migration in vitro P-cadherin expression at the protein and mRNA levels In subsequent experiments, we performed short-term [Figures 3A,B]. The Cancer Genome Atlas (TCGA) in vitro attachment assays to evaluate the effect of dataset is the largest compilation of gene expression HOXA9 on interactions between EOC cells and peri- data of clinical specimens of EOC. To determine whether toneal mesothelial cells independently of its effects on P-cadherin expression is elevated in clinical specimens aggregation and survival of EOC cells. Equivalent num- of EOC that highly express HOXA9, we stratified EOC bers of control and HOXA9-knockdown SKOV3ip cases in the TCGA dataset (n = 567 cases) into quartile cells were seeded as single cell suspensions onto con- sub-groups according to the levels of HOXA9 tran- fluent monolayers of normal human omental mesothe- scripts of tumors. As shown in Figure 3C, levels of lial cells. Attachment of SKOV3ip cells to mesothelial CDH3 transcripts were found to be significantly higher cells was assayed at 1 h after seeding during which in HOXA9-High tumors (upper quartile sub-group) than time no significant change in cell aggregation or sur- in HOXA9-Low tumors (lower quartile sub-group) (P = vival occurred. The numbers of HOXA9-knockdown 0.011). Two putative consensus HOXA9-binding sites were SKOV3ip cells that bound to mesothelial cells were identified in the CDH3 promoter at 1.8 kb (site A) and significantly lower than the numbers of bound control at 2.1 kb (site B) upstream of the transcription start site SKOV3ip cells (P < 0.001) [Figures 2E,F]. To evaluate [Figure 3D]. Binding of endogenous HOXA9 to both sites the effect of HOXA9 on the migratory potential of was detected by chromatin immunoprecipitation (IP) assays EOC cells independently of its effects on cell aggrega- in control SKOV3ip cells, but no binding was detected in tion and mesothelial attachment, we assayed migration HOXA9-knockdown SKOV3ip cells [Figure 3D]. These of EOC cells that were seeded as single cell suspen- findings indicate that HOXA9 induces P-cadherin ex- sions into transwell chambers. As compared to control pression and that the CDH3 gene is a transcriptional SKOV3ip cells, HOXA9-knockdown SKOV3ip cells target of HOXA9. exhibited significantly reduced migration (P < 0.001) [Figure 2G]. These findings indicate that HOXA9 P-cadherin inhibition abrogates the stimulatory effects of expression in EOC cells stimulates EOC-mesothelial HOXA9 on EOC cell aggregation, implantation and cell interactions and EOC cell migration, and that migration these stimulatory effects are independent of its effect To determine whether HOXA9 promotes EOC cell on EOC cell aggregation and survival. aggregation, implantation and migration by inducing P-cadherin expression, we evaluated the effects of inhi- HOXA9 binds the CDH3 promoter and induces P-cadherin biting P-cadherin in EOC cells that expressed HOXA9. expression P-cadherin levels that were induced by enforced HOXA9 The classical cadherin P-cadherin has been reported to expression in SKOV3-Par cells were knocked-down by be induced during early stages of EOC progression and shRNAs targeting two different sites within the CDH3 to promote EOC cell migration [11-13]. We recently gene (shPCAD-A, shPCAD-B) [Figure 4A]. Quantifica- identified that P-cadherin also facilitates aggregation of tion of this knockdown is shown in Additional file 2: floating EOC cells, prevents anoikis and mediates the Figure S2B. Knockdown of P-cadherin in HOXA9+ attachment of EOC cells to peritoneal mesothelial SKOV3-Par cells reduced cell aggregation [Figure 4B]. cells [14]. Because HOXA9 is a transcription factor, Cell death in HOXA9+ SKOV3-Par cells was signifi- we investigated the possibility that HOXA9 induces cantly increased when P-cadherin was knocked-down P-cadherin expression. P-cadherin levels were mark- (P < 0.001) [Figures 4C,D]. To confirm our findings, edly reduced when HOXA9 was knocked-down in we evaluated the effects of inhibiting P-cadherin by a SKOV3ip cells [Figure 3A]. Quantification of this pro- neutralizing antibody (Ab). Treatment of suspension tein down-regulation is shown in Additional file 2: cultures of HOXA9+ SKOV3-Par cells with P-cadherin Figure S2A. The down-regulation of P-cadherin was Ab inhibited cell aggregation and increased cell death, confirmed by quantitative reverse transcription PCR as compared to cells that were treated with control IgG (qRT-PCR) analysis of transcript levels of CDH3,the [Additional file 3: Figure S3A,B]. In addition, knock- gene encoding P-cadherin [Figure 3B]. Knockdown down of P-cadherin in HOXA9+ SKOV3-Par cells of HOXA9 in SKOV3ip cells did not down-regulate significantly reduced the ability of these tumor cells to Ko and Naora Molecular Cancer 2014, 13:170 Page 6 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 3 HOXA9 induces P-cadherin expression in EOC cells. (A) Western blot analysis of HOXA9 and cadherin levels in control and HOXA9-knockdown SKOV3ip lines and in vector-control and HOXA9-transfected SKOV3-Par lines. The 34 kD band corresponding to HOXA9 is indicated by an arrow. (B) qRT-PCR analysis of relative CDH3 mRNA levels in SKOV3ip and SKOV3-Par lines. Significance of differences in CDH3 mRNA levels was evaluated by Student t-test. (C) EOC cases from the TCGA Project (n = 567) were stratified according to HOXA9 expression in tumors, where HOXA9 mRNA levels were defined as High (≥ upper quartile) and Low (≤ lower quartile). Significance of differences in CDH3 mRNA levels (log2 scale) between upper and lower quartile sub-groups was evaluated by Mann–Whitney U-test. (D) Schematic representation of the human CDH3 promoter showing locations of the two HOXA9-binding sites (site A: 5′-TCATTTAAAAC-3′ and site B: 5′-TAATTTATTTAATAC-3′). Binding of endogenous HOXA9 in control SKOV3ip cells to these sites was detected by chromatin IP. Negative controls included IP using cells expressing HOXA9 shRNA (shA9-B) and IP with IgG. GAPDH was amplified as an irrelevant gene control. attach to mesothelial cells (P < 0.01) and to migrate (P < 0.001) [Figure 4D]. Together, these observations indicate that the stimulatory effects of HOXA9 on aggregation and survival of floating EOC cells, meso- thelial attachment and EOC cell migration are largely mediated by its induction of P-cadherin expression. Stimulatory effects of HOXA9 on EOC cell aggregation, implantation and migration are recapitulated by P-cadherin in vitro and in vivo In converse experiments, we evaluated whether the stimulatory effects of HOXA9 on EOC cell aggrega- tion, implantation and migration can be restored when P-cadherin is reconstituted in EOC cells in which HOXA9 is inhibited. CDH3 cDNA was stably expressed in HOXA9-knockdown SKOV3ip cells (shA9-B + PCAD) to restore the level of P-cadherin that was similar to the endogenous P-cadherin level in HOXA9+ control (non-targeting) SKOV3ip cells [Figure 5A, Additional file 2: Figure S2C]. Reconstitution of P-cadherin in HOXA9-knockdown SKOV3ip cells markedly increased cell aggregation and significantly decreased cell death in suspension cultures (P < 0.001) [Figures 5B-D]. Recon- stitution of P-cadherin also increased the ability of HOXA9-knockdown SKOV3ip cells to attach to meso- thelial monolayers and to migrate in in vitro assays [Figure 5D]. To confirm the findings of our in vitro studies, we firstly evaluated anoikis in floating EOC cells in short- term in vivo assays. Based on our prior experience using i.p. xenograft models derived from SKOV3ip cells [10,14], the earliest time-point at which solid peritoneal implants can be detected in mice is approxi- mately 10 days following tumor cell inoculation. To assay anoikis in floating EOC cells prior to implant- ation, groups of mice were inoculated with equivalent Ko and Naora Molecular Cancer 2014, 13:170 Page 7 of 13 http://www.molecular-cancer.com/content/13/1/170 Figure 4 Inhibition of P-cadherin in HOXA9-overexpressing EOC cells abrogates the stimulatory effects of HOXA9 on cell aggregation, survival, implantation and migration. (A) Western blot analysis of P-cadherin levels in vector-control and HOXA9-transfected SKOV3-Par lines, and in HOXA9-transfected SKOV3-Par lines that stably expressed non-targeting shRNA and shRNAs targeting two different sites within the CDH3 gene (shPCAD-A, shPCAD-B). (B,C) Cells of SKOV3-Par lines were incubated as suspension cultures in polyHEMA-coated plates for 3 days. (B) Cell morphology viewed by phase-contrast light microscopy. Bar, 50 μm. (C) Evaluation of cell death by flow cytometric analysis of 7AAD staining. (D) Evaluation of cell death by ELISA in suspension cultures of SKOV3-Par cells, attachment of SKOV3-Par cells to confluent mesothelial cell monolayers, and migration of SKOV3-Par cells were assayed as described in Figures 2D, E and G, respectively. Shown in D are mean + sd values of three independent assays. Figure 5 Reconstitution of P-cadherin in HOXA9-knockdown EOC cells restores the stimulatory effects of HOXA9 on cell survival, implantation and migration in vitro. (A) Western blot analysis of P-cadherin levels in HOXA9+ control SKOV3ip cells (non-targeting), HOXA9-knockdown SKOV3ip cells (shA9-B) and HOXA9-knockdown SKOV3ip cells that stably expressed P-cadherin (shA9-B + PCAD). (B) Morphology of SKOV3ip cells following incubation as suspension cultures in polyHEMA-coated plates for 3 days. Bar, 100 μm. (C) Evaluation of cell death in suspension cultures by flow cytometric analysis of 7AAD staining. (D) Evaluation of cell death by ELISA in suspension cultures of SKOV3ip cells, attachment of SKOV3ip cells to confluent mesothelial cell monolayers, and migration of SKOV3ip cells. Shown in D are mean + sd values of three independent assays. Ko and Naora Molecular Cancer 2014, 13:170 Page 8 of 13 http://www.molecular-cancer.com/content/13/1/170 numbers of non-targeting, shA9-B and shA9-B + PCAD subsequently developed on the peritoneal cavity wall in SKOV3ip cells. At 3 days thereafter, GFP-expressing EOC the shA9-B + PCAD group were significantly higher than cells in the peritoneal fluid were evaluated for cell death. the numbers of implants in the shA9-B group (P <0.01) The proportion of floating EOC cells that exhibited cell and almost equivalent to the numbers of implants in the death in the shA9-B + PCAD model was markedly lower control group [Figure 6B]. Similar results were obtained than that in the shA9-B model and similar to that in the when we evaluated the numbers of mesenteric implants control model [Figure 6A]. The numbers of implants that in these groups of mice [Figure 6C]. Tumor implants that Figure 6 Reconstitution of P-cadherin in HOXA9-knockdown EOC cells restores the stimulatory effects of HOXA9 on cell survival, implantation and migration in vivo. Female nude mice were inoculated i.p. with equivalent numbers (2 × 10 ) of cells of control (non-targeting) and HOXA9-knockdown (shA9-B) SKOV3ip lines and HOXA9-knockdown SKOV3ip cells that stably expressed P-cadherin (shA9-B + PCAD). In (A),mice were sacrificed at 3 days following tumor cell inoculation. Floating cells in the peritoneal cavity were collected and stained with 7AAD. Cell death within the gated population of GFP+ tumor cells was evaluated by flow cytometric analysis of 7AAD staining. In (B-D), mice were analyzed at 3 weeks following tumor cell inoculation. (B) Representative examples of GFP-expressing tumor implants (shown in green) attached to the peritoneal cavity wall as viewed by fluorescence microscopy. Bar, 1 mm. Numbers of implants on the peritoneal cavity wall were counted in five random 25 mm fields per mouse. (C) Representative examples of HE-stained sections of bowel tissues with mesenteric implants. Bar, 1 mm. Numbers of mesenteric implants were counted in three random 1 cm fields per mouse. (D) Representative examples of HE-stained sections of diaphragmatic implants. Bar 100 μm. Ko and Naora Molecular Cancer 2014, 13:170 Page 9 of 13 http://www.molecular-cancer.com/content/13/1/170 derived from shA9-B + PCAD cells and formed on the abolished the stimulatory effect of HOXA9 on aggre- diaphragm, bowel serosa and cavity wall were found to gation, whereas aggregation was almost completely be as invasive as control tumors and to exhibit greater restored when P-cadherin was reconstituted in EOC invasive depth than implants that derived from shA9-B cells in which HOXA9 was inhibited. HOXA9 had little cells and formed at the same sites [Figure 6D; Additional effect on levels of other classical cadherins, but it is file 1: Figures S1A-C]. Together, our findings indicate not possible to exclude the possibility that HOXA9 that HOXA9 promotes the aggregation and survival of might also promote tumor cell aggregation by inducing floating EOC cells and also the peritoneal attachment expression of other types of adhesion molecules. The and invasiveness of EOC cells, and that these stimula- classical cadherins have well-characterized functions in tory effects of HOXA9 are largely due to its induction facilitating homotypic cell adhesion [26,27]. The sig- of P-cadherin expression. nificance of P-cadherin in mediating aggregation of floating EOC cells is supported by our recent study in Discussion which we demonstrated that inhibiting P-cadherin Substantial evidence indicates that aberrations in sig- reduces aggregation of ascitic EOC cells in xenograft naling pathways that control normal developmental models [14]. High expression of P-cadherin has been patterning play pivotal roles in driving tumorigenesis found to be associated with poor overall survival of [16-18]. Homeoproteins comprise another important EOC patients [12]. A study by Patel and colleagues class of patterning regulators and have been increas- identified that P-cadherin is the predominant type of ingly found to be aberrantly expressed in a variety of cadherin expressed in tumor cells in the peritoneal cancers [6,7]. Several homeoproteins promote tumor fluid of EOC patients [11]. Sivertsen and colleagues cell proliferation by deregulating transcription of genes reported that P-cadherin is frequently expressed in that control cell cycle progression and autocrine effusion specimens of EOC patients (51 of 53 cases), growth-stimulatory pathways [19-21]. There is also but also found that E-cadherin and N-cadherin are evidence that some homeoproteins promote metastasis. expressed at similar frequency [28]. There is discordance For example, SIX1 induces epithelial-to-mesenchymal in the literature regarding the expression of E-cadherin transition (EMT) in breast cancer cells by activating and N-cadherin in effusion specimens of EOC patients. transcription of the gene encoding the TGF-β type I Two studies have reported that levels of E-cadherin and receptor [22] and promotes metastasis of rhabdomyo- N-cadherin are decreased in effusions as compared to sarcoma by inducing expression of the cytoskeletal pro- solid tumors [11,29], whereas a third study identified tein ezrin [23]. Both normal patterning and tumor that E-cadherin levels were higher in effusions than in progression involve dynamic changes in homotypic and matching primary solid tumors [30]. heterotypic cellular interactions, but surprisingly only Although it is widely thought that aggregation enables few genes that control cell adhesion have been identi- floating tumor cells to escape anoikis, the function of fied as homeoprotein targets [24,25]. We previously cadherins in anoikis appears to be cell-specific. Inhib- identified that high HOXA9 expression is strongly asso- ition of E-cadherin increases anoikis sensitivity in Ewing ciated with poor overall survival in EOC patients and sarcoma cells [31], but increases anoikis resistance in promotes tumor progression in EOC xenograft models mammary epithelial cells [32]. Our recent studies sup- [10]. The findings of this present study indicate that the port the notion that P-cadherin primarily inhibits anoikis CDH3 gene which encodes P-cadherin is a transcrip- in EOC cells by promoting cell aggregation. Inhibition of tional target of HOXA9 and that HOXA9 promotes P-cadherin increased cell death in suspension cultures both homotypic and heterotypic cell interactions that but not in adherent cultures of EOC cells, and increased facilitate intraperitoneal dissemination of EOC through cell death in ascitic EOC cells but not in solid tumor xenografts [14]. The possibility that P-cadherin might its induction of P-cadherin. The propensity for intraperitoneal ‘seeding’ is a hall- promote cell survival independently of cell aggregation mark of EOC and depends on adaptation by EOC cells cannot be excluded. Cell adhesion is facilitated by in- teractions between the extracellular domains of cad- to two dynamic changes in their microenvironment. The first change occurs when EOC cells are exfoliated and herin molecules on adjacent cells [26]. Intracellular transported by the flow of the peritoneal fluid. It is signal transduction triggered by P-cadherin involves activation of Rac1 and Cdc42 Rho GTPases [33], but it widely thought that aggregation enables floating EOC cells to escape anoikis [3,4], but the mechanisms that has been found that Rac1 and Cdc42 activation alone facilitate this aggregation are poorly understood. The does not fully account for the ability of P-cadherin to suppress anoikis [14]. There is evidence that EMT present study indicates that HOXA9 promotes aggrega- tion of floating EOC cells by inducing expression of confers anoikis resistance in some cell types [32]. How- P-cadherin. Inhibiting P-cadherin in HOXA9+ EOC cells ever, we have found that P-cadherin does not alter Ko and Naora Molecular Cancer 2014, 13:170 Page 10 of 13 http://www.molecular-cancer.com/content/13/1/170 expression of EMT-promoting transcription factors in function. Cadherin-mediated cell adhesion is essential EOC cells [14]. We also cannot exclude the possibility for maintaining the structure and function of repro- that HOXA9 inhibit anoikis by a mechanism that is ductive tissues [43], but mice with targeted disruption independent of P-cadherin. Because TGF-β signaling of the Cdh3 or Hoxa9 genes are fertile [44,45]. The induces EMT [34] and HOXA9 induces TGF-β2ex- mechanisms that induce HOXA9 expression in EOC pression in EOC cells [10], HOXA9 might inhibit anoikis arealsopoorlyunderstood. Onepossiblemechanism is by promoting EMT. However, we found that HOXA9 deregulated promoter methylation. HOXA9 promoter neither represses E-cadherin expression [Figure 3A] nor methylation has been detected at significantly lower alters expression of EMT-promoting transcription factors frequency in advanced-stage EOC than in early-stage in EOC cells, including those that are induced by TGF-β tumors [46]. Another possible mechanism is histone signaling [10]. Autocrine TGF-β signaling appears to be modification. Polycomb group (PcG) proteins form impaired in EOC cells that express HOXA9 [10] and this complexes that dynamically alter chromatin structure impairment could stem from the ability of HOXA9 to by modifying specific residues in histone tails and PcG- block TGF-β-induced Smad-dependent transcription [35]. mediated repression is a principal mechanism by which Collectively, these findings strongly suggest that the ability HOX gene expression is tightly regulated during devel- of HOXA9 to inhibit anoikis in EOC cells is largely attrib- opment [47]. AKT signaling has been reported to sup- utable to its induction of P-cadherin and the facilitation of press the histone methyltransferase activity of the PcG EOC cell aggregation by P-cadherin. protein EZH2 [48], and is frequently activated in high- The second important rate-limiting step in the pro- grade, advanced-stage EOCs [49]. It is therefore pos- gression of EOC is the implantation of floating tumor sible that AKT activation in EOC derepresses HOXA9 cells on to peritoneal surfaces. Interactions of EOC expression by inhibiting EZH2 activity. Interestingly, cells with peritoneal mesothelial cells and/or the sub- HOXA9 is not endogenously expressed in the parental mesothelial extracellular matrix are mediated by vari- SKOV3 cell line but in the aggressive subclone SKOV3ip ous cell surface molecules including CD44 and several [Figure 3A]. The expression of HOXA9 in SKOV3ip cells integrins [36,37]. We previously identified that HOXA10, might be attributable to SKOV3ip cells having higher a homeoprotein that is expressed in a subset of EOCs, can ERBB2 levels than the parental SKOV3 cells [15]. promote interactions between EOC cells and peritoneal mesothelial cells by inducing expression of αvβ3 integrin Conclusion [38]. The notion that P-cadherin facilitates tumor- The present study supports increasing evidence of the peritoneum interactions is supported by findings that functional significance of homeoproteins in dictating P-cadherin is frequently expressed in malignant periton- the clinical behavior of tumors through controlling eal effusions of EOC patients and is the most predominant transcription of distinct sets of target genes. Because type of cadherin in normal peritoneal tissues [11,28,39]. homeoproteins are transcription factors and share The findings of the present study indicate that the ability tracts of homology, it is challenging to therapeutically of HOXA9 to promote interactions between EOC cells target these proteins. However, elucidating the mecha- and peritoneal mesothelial cells is primarily mediated nisms of homeoproteins and their downstream effec- by its induction of P-cadherin. Inhibiting P-cadherin in tors in tumors can yield important insights into more HOXA9+ EOC cells abrogated the stimulatory effect of effective therapeutic strategies. Inhibiting P-cadherin is HOXA9 on mesothelial cell attachment, and reconstitu- particularly promising as a human monoclonal Ab to tion of P-cadherin in HOXA9-knockdown cells restored P-cadherin has been developed and is undergoing clin- attachment capability. Our work also demonstrates that ical trials [50]. HOXA9 promotes migratory potential of EOC cells via its induction of P-cadherin. The function of P-cadherin in cell migration varies depending on the type of cell. Methods It has been reported that P-cadherin inhibits migration Abs and plasmids of melanoma and oral squamous cell carcinoma cells Abs to HOXA9 were purchased from Millipore (for [40,41], but promotes migration of pancreatic and blad- Western blot) and Santa Cruz Biotechnology (for chro- der cancer cells [33,42]. Our findings support prior matin IP). Abs to P-cadherin were purchased from BD Bio- reports that P-cadherin stimulates migratory potential sciences (for Western blot) and Abcam (for neutralization). of EOC cells [13,14]. Other Abs were as follows: E-cadherin (Invitrogen), Because the precise mechanism of HOXA9 in con- N-cadherin, active caspase-3 (BD Biosciences), actin, trolling female reproductive development is not secondary Abs (Sigma-Aldrich). HOXA9 cDNA was known, it is unclear how its induction of P-cadherin in provided by Corey Largman (Veterans Affairs Medical EOC might be related to its normal developmental Center, San Francisco). CDH3 cDNA and pGFP-V-RS Ko and Naora Molecular Cancer 2014, 13:170 Page 11 of 13 http://www.molecular-cancer.com/content/13/1/170 plasmids containing HOXA9, CDH3 and non-targeting described [38]. Where indicated, EOC cells were incu- shRNAs were purchased from OriGene Technologies. bated with neutralizing P-cadherin Ab or with control IgG at a final concentration of 10 μg/ml. At 3 days there- Cell culture and transfection after, cell morphology was viewed by phase-contrast light SKOV3ip cell lines that stably express non-targeting microscopy. Cells were evaluated for cell death by staining and HOXA9 shRNAs have been previously described with 7AAD (Sigma-Aldrich) and with Ab to active caspase- [10]. The parental SKOV3 cell line was purchased from 3. Staining was detected by flow cytometry (FACS Calibur, American Type Culture Collection. Cell lines were BD Biosciences). Cell death was also evaluated by assaying authenticated by short tandem repeat analysis performed mono- and oligo- nucleosomes in cell lysates by using the by the MD Anderson Cancer Center Characterized Cell Cell Death Detection ELISA kit (Roche). Three independ- Line Core Facility. SKOV3ip and SKOV3 cells were ent experiments were performed for each assay. cultured in McCoys 5A medium (Invitrogen) supple- mented with 10% FBS and penicillin-streptomycin. Cells Cell attachment assays were transfected by using Lipofectamine 2000 reagent Single cell suspensions of GFP-expressing EOC cells were (Invitrogen) and selected by the addition of puromycin seeded in 96-well plates (1.5 × 10 per well) containing (0.5 μg/ml). Primary cultures of normal human omental confluent monolayers of normal omental mesothelial cells mesothelial cells have been previously described [51] as previously described [38]. At 1 hr after seeding of EOC and were provided by Ernst Lengyel (University of Chicago, cells, wells were washed with culture medium to remove Chicago, IL). unattached EOC cells. Attached EOC cells were viewed by fluorescence microscopy using a fluorescein filter. Three Western blot and qRT-PCR analyses independent experiments were performed in which Cell lysates were prepared by using M-PER buffer attached cells were counted in three random 100× (Pierce Biotechnology), separated by SDS-PAGE and microscopic fields in each experiment. transferred to polyvinylidene difluoride membranes (GE Healthcare). CDH3 transcripts were analyzed by using Cell migration assays SYBR®Green qPCR Master Mix (SABiosciences) and the EOC cells were seeded as single cell suspensions in the following primers: forward: 5′-CAGGTGCTGAACATC upper chamber of 24-well transwell chambers (BD ACGGACA-3′, reverse: 5′-CTTCAGGGACAAGACCA Biosciences) (5 × 10 cells per well). At 6 h thereafter, CTGTG-3′. RPL32 transcript levels were used as migrating cells were stained with Giemsa solution. controls for normalization and were detected by using Three independent experiments were performed in the following primers: forward: 5′-ACAAAGCACA which migrating cells were counted in three random TGCTGCCCAGTG-3′ reverse: 5′-TTCCACGATGG 100x microscopic fields in each experiment. CTTTGCGGTTC-3′. Mouse i.p. xenograft studies Chromatin IP Four-week-old female nude mice (purchased from the Na- Chromatin IP assays were performed by using the tional Cancer Institute, Frederick, MD) were inoculated i. EZ-ChIP Assay kit (Millipore). Sheared chromatin was p. with 2 × 10 cells of GFP-expressing EOC lines (n = 5 incubated overnight with 1 μg of HOXA9 Ab. DNA was mice per group). Mice were euthanized by CO asphyxi- purified from precipitated complexes. Fragments of the ation at time points indicated in the figure legends. human CDH3 promoter were amplified by using the fol- GFP-expressing tumor implants were visualized under a lowing primers: for Site A (150 bp), forward: 5′-CCCCCA Leica MZML III stereomicroscope equipped with a mer- CCCTCGCAACGCAAGCAA-3′, reverse: 5′-TGGGATT cury lamp power supply and GFP filter set. Formalin- ACAGGCGTGAGAAACG-3′; for Site B (104 bp), fixed, paraffin-embedded tissue sections were stained with forward: 5′-GCTGTGTGCCAGATATGATGTTTAG-3′, hematoxylin-eosin (HE) and analyzed under a light micro- reverse: 5′-CTGTCAAATGGGGCTGTTATGATC-3′.A scope. Invasive depth of implants was measured using a 166 bp fragment of the GAPDH gene was amplified as an stage micrometer slide in five random microscopic fields irrelevant control by using the following primers: forward: of tissue sections of each mouse at the magnifications 5′-TACTAGCGGTTTTACGGGCG-3′,reverse: 5′-TCG indicated in the figure legends. Floating cells were col- AACAGGAGGAGCAGAGAGCGA-3′. lected from peritoneal fluid, stained with Hoechst dye and viewed by fluorescence microscopy. Floating cells were Cell death assays also stained with 7AAD and cell clusters gently disaggre- EOC cells were seeded in 96-well plates (1 × 10 cells per gated by passing through 35 μm nylon mesh. Immediately well) that were coated with polyHEMA (Sigma-Aldrich) thereafter, 7AAD staining was analyzed by flow cytometry to block cell attachment to substratum as previously within the gated population of GFP + tumor cells. Ko and Naora Molecular Cancer 2014, 13:170 Page 12 of 13 http://www.molecular-cancer.com/content/13/1/170 Bioinformatic analysis cadherin; PcG: Polycomb group; PolyHEMA: Poly (2-hydroxyethyl methacrylate); shRNAs: Short hairpin RNAs; TGF: Transforming growth factor. Gene expression data of EOC cases from the TCGA pro- ject (n = 567) were downloaded from the TCGA data Competing interest portal site (http://tcga-data.nci.nih.gov/tcga/). Where there The authors have declared that no conflict of interest exists. were multiple probe sets for an individual gene, the mean value for the given gene for each case was used. Patients Authors’ contribution were stratified according to the level of HOXA9 expression SYK and HN designed and performed experiments and wrote the in tumors, where HOXA9 transcript levels were defined as manuscript. Both authors read and approved the final manuscript. High (≥ upper quartile) and Low (≤ lower quartile) as pre- viously described [10]. Acknowledgements This work was supported by Cancer & Prevention Research Institute of Texas grant RP120390 (to H. Naora) and U.S. National Institutes of Health grant Statistical analysis CA141078 (to H. Naora). We thank Nicolas Barengo for technical assistance and members of the Naora laboratory for helpful discussions. Statistical analysis was performed by using STATISTICA6 software (StatSoft Inc.). Statistical significance of data of Received: 22 March 2014 Accepted: 4 July 2014 in vitro and in vivo assays was assessed by unpaired two- Published: 14 July 2014 tailed Student’s t-test. Data represent mean ± s.d. Signifi- cance of differences in gene expression between groups of References 1. Siegel R, Naishadham D, Jemal A: Cancer statistics, 2013. CA Cancer J Clin patients was assessed by Mann–Whitney U-test. P values 2013, 63:11–30. of < 0.05 were considered significant. 2. Naora H, Montell DJ: Ovarian cancer metastasis: integrating insights from disparate model organisms. Nat Rev Cancer 2005, 5:355–366. 3. Lengyel E: Ovarian cancer development and metastasis. Am J Pathol 2010, Additional files 177:1053–1064. 4. Sodek KL, Murphy KJ, Brown TJ, Ringuette MJ: Cell-cell and cell-matrix dynamics in intraperitoneal cancer metastasis. Cancer Metastasis Rev 2012, Additional file 1: Figure S1. Invasiveness of EOC cells in i.p. xenograft 31:397–414. models. Female nude mice (n=5 per group) were inoculated i.p. with 5. Burleson KM, Casey RC, Skubitz KM, Pambuccian SE, Oegema TR Jr, Skubitz equivalent numbers of cells (2 × 10 ) of SKOV3ip lines and sacrificed at 3 AP: Ovarian carcinoma ascites spheroids adhere to extracellular matrix weeks thereafter. Invasive depth of implants on the bowel, diaphragm and components and mesothelial cell monolayers. Gynecol Oncol 2004, peritoneal cavity wall was measured in five random microscopic fields of 93:170–181. HE-stained tissue sections of each of these sites in each mouse. An average 6. Abate-Shen C: Deregulated homeobox gene expression in cancer: cause invasive depth was calculated for each site of each mouse. (A) Depth of or consequence? Nat Rev Cancer 2002, 2:777–785. superficial bowel serosa invasion (evaluated at 200× magnification) and 7. Samuel S, Naora H: Homeobox gene expression in cancer: insights from representative examples of HE-stained sections. Bar, 50 μm. (B) Depth of developmental regulation and deregulation. Eur J Cancer 2005, invasion of diaphragmatic implants into adjacent muscle (evaluated at 41:2428–2437. 100x magnification). Representative examples of HE-stained sections of 8. Pearson JC, Lemons D, McGinnis W: Modulating Hox gene functions diaphragmatic implants are shown in Figure 1D. (C) Invasive depth of during animal body patterning. Nat Rev Genet 2005, 6:893–904. implants on the peritoneal cavity wall (evaluated at 100× magnification) 9. Taylor HS, Vanden Heuvel GB, Igarashi P: A conserved Hox axis in the and representative examples of HE-stained sections. Bar, 100 μm. mouse and human female reproductive system: late establishment and Additional file 2: Figure S2. Quantification of protein levels. Protein persistent adult expression of the Hoxa cluster genes. Biol Reprod 1997, levels were evaluated by measuring intensity of bands on Western blots 57:1338–1345. shown in Figures 3A, 4A and 5A using the TINA 20 program (Raytest). 10. Ko SY, Barengo N, Ladanyi A, Lee JS, Marini F, Lengyel E, Naora H: (A) Levels of an individual protein in SKOV3ip lines transfected with HOXA9 promotes ovarian cancer growth by stimulating cancer-associated non-targeting and HOXA9 shRNAs are expressed relative to its level in fibroblasts. J Clin Invest 2012, 122:3603–3617. the empty vector control SKOV3ip line. Levels of an individual protein in 11. Patel IS, Madan P, Getsios S, Bertrand MA, MacCalman CD: Cadherin the HOXA9-transfected SKOV3-Par line are expressed relative to its level switching in ovarian cancer progression. Int J Cancer 2003, 106:172–177. in the empty vector control SKOV3-Par line. (B) Levels of P-cadherin in 12. Quattrocchi L, Green AR, Martin S, Durrant L, Deen S: The cadherin switch HOXA9-transfected SKOV3-Par lines that were co-transfected with no in ovarian high-grade serous carcinoma is associated with disease shRNA, non-targeting shRNA or CDH3 shRNAs are expressed relative to progression. Virchows Arch 2011, 459:21–29. its level in the empty vector control SKOV3-Par line. (C) Levels of P- 13. Cheung LW, Leung PC, Wong AS: Cadherin switching and activation of cadherin in HOXA9-knockdown SKOV3ip cells and HOXA9-knockdown p120 catenin signaling are mediators of gonadotropin-releasing hormone SKOV3ip cells that stably expressed P-cadherin are expressed relative to to promote tumor cell migration and invasion in ovarian cancer. Oncogene its level in SKOV3ip cells expressing non-targeting shRNA. 2010, 29:2427–2440. Additional file 3: Figure S3. Effects of P-cadherin Ab on 14. Usui A, Ko SY, Barengo N, Naora H: P-cadherin promotes ovarian cancer HOXA9-overexpressing EOC cells. Cells of vector-control and dissemination through tumor cell aggregation and tumor-peritoneum HOXA9-transfected SKOV3-Par lines were incubated as suspension tnteractions. Mol Cancer Res 2014, 12:504–513. cultures in polyHEMA-coated plates for 3 days with the addition of 15. Yu D, Wolf JK, Scanlon M, Price JE, Hung MC: Enhanced c-erbB-2/neu neutralizing P-cadherin Ab or control IgG. (A) Cell morphology expression in human ovarian cancer cells correlates with more severe viewed by phase-contrast microscopy. Bar 50 μm. (B) Cell death was malignancy that can be suppressed by E1A. Cancer Res 1993, 53:891–898. evaluatedby assayingmono- andoligo-nucleosomes in cell lysates 16. Pasca di Magliano M, Hebrok M: Hedgehog signalling in cancer formation by ELISA. Shown are mean + sd values of three independent experiments. and maintenance. Nat Rev Cancer 2003, 3:903–911. 17. Klaus A, Birchmeier W: Wnt signalling and its impact on development and cancer. Nat Rev Cancer 2008, 8:387–398. Abbrevations 18. Ranganathan P, Weaver KL, Capobianco AJ: Notch signalling in solid 7AAD: 7-amino actinomycin D; Ab: Antibody; EOC: Epithelial ovarian cancer; tumours: a little bit of everything but not all the time. Nat Rev Cancer ECAD: E-cadherin; HE: Hematoxylin-eosin; NCAD: N-cadherin; PCAD: P- 2011, 11:338–351. Ko and Naora Molecular Cancer 2014, 13:170 Page 13 of 13 http://www.molecular-cancer.com/content/13/1/170 19. Care A, Silvani A, Meccia E, Mattia G, Stoppacciaro A, Parmiani G, Peschle C, 41. Bauer K, Dowejko A, Bosserhoff AK, Reichert TE, Bauer RJ: P-cadherin Colombo MP: HOXB7 constitutively activates basic fibroblast growth induces an epithelial-like phenotype in oral squamous cell carcinoma factor in melanomas. Mol Cell Biol 1996, 16:4842–4851. by GSK-3beta-mediated Snail phosphorylation. Carcinogenesis 2009, 20. Coletta RD, Christensen K, Reichenberger KJ, Lamb J, Micomonaco D, 30:1781–1788. Huang L, Wolf DM, Muller-Tidow C, Golub TR, Kawakami K, Ford HL: 42. Mandeville JA, Silva Neto B, Vanni AJ, Smith GL, Rieger-Christ KM, Zeheb R, The Six1 homeoprotein stimulates tumorigenesis by reactivation of Loda M, Libertino JA, Summerhayes IC: P-cadherin as a prognostic cyclin A1. Proc Natl Acad Sci U S A 2004, 101:6478–6483. indicator and a modulator of migratory behaviour in bladder 21. Trinh BQ, Barengo N, Naora H: Homeodomain protein DLX4 counteracts carcinoma cells. BJU Int 2008, 102:1707–1714. key transcriptional control mechanisms of the TGF-beta cytostatic 43. Rowlands TM, Symonds JM, Farookhi R, Blaschuk OW: Cadherins: crucial program and blocks the antiproliferative effect of TGF-beta. Oncogene regulators of structure and function in reproductive tissues. Rev Reprod 2011, 30:2718–2729. 2000, 5:53–61. 22. Micalizzi DS, Wang CA, Farabaugh SM, Schiemann WP, Ford HL: 44. Lawrence HJ, Helgason CD, Sauvageau G, Fong S, Izon DJ, Humphries RK, Homeoprotein Six1 increases TGF-beta type I receptor and converts Largman C: Mice bearing a targeted interruption of the homeobox gene TGF-beta signaling from suppressive to supportive for tumor growth. HOXA9 have defects in myeloid, erythroid, and lymphoid hematopoiesis. Cancer Res 2010, 70:10371–10380. Blood 1997, 89:1922–1930. 45. Radice GL, Ferreira-Cornwell MC, Robinson SD, Rayburn H, Chodosh LA, 23. Yu Y, Davicioni E, Triche TJ, Merlino G: The homeoprotein six1 Takeichi M, Hynes RO: Precocious mammary gland development in P- transcriptionally activates multiple protumorigenic genes but requires cadherin-deficient mice. J Cell Biol 1997, 139:1025–1032. ezrin to promote metastasis. Cancer Res 2006, 66:1982–1989. 46. Wu Q, Lothe RA, Ahlquist T, Silins I, Trope CG, Micci F, Nesland JM, Suo Z, 24. Daftary GS, Troy PJ, Bagot CN, Young SL, Taylor HS: Direct regulation of Lind GE: DNA methylation profiling of ovarian carcinomas and their beta3-integrin subunit gene expression by HOXA10 in endometrial cells. in vitro models identifies HOXA9, HOXB5, SCGB3A1, and CRABP1 as Mol Endocrinol 2002, 16:571–579. novel targets. Mol Cancer 2007, 6:45. 25. Zhu R, Wong KF, Lee NP, Lee KF, Luk JM: HNF1alpha and CDX2 47. Soshnikova N, Duboule D: Epigenetic regulation of vertebrate Hox genes: transcriptional factors bind to cadherin-17 (CDH17) gene promoter and a dynamic equilibrium. Epigenetics 2009, 4:537–540. modulate its expression in hepatocellular carcinoma. J Cell Biochem 2010, 48. Cha TL, Zhou BP, Xia W, Wu Y, Yang CC, Chen CT, Ping B, Otte AP, Hung 111:618–626. MC: Akt-mediated phosphorylation of EZH2 suppresses methylation of 26. Gumbiner BM: Regulation of cadherin-mediated adhesion in lysine 27 in histone H3. Science 2005, 310:306–310. morphogenesis. Nat Rev Mol Cell Biol 2005, 6:622–634. 49. Yuan ZQ, Sun M, Feldman RI, Wang G, Ma X, Jiang C, Coppola D, Nicosia SV, 27. Halbleib JM, Nelson WJ: Cadherins in development: cell adhesion, sorting, Cheng JQ: Frequent activation of AKT2 and induction of apoptosis by and tissue morphogenesis. Genes Dev 2006, 20:3199–3214. inhibition of phosphoinositide-3-OH kinase/Akt pathway in human 28. Sivertsen S, Berner A, Michael CW, Bedrossian C, Davidson B: Cadherin ovarian cancer. Oncogene 2000, 19:2324–2330. expression in ovarian carcinoma and malignant mesothelioma cell 50. Zhang CC, Yan Z, Zhang Q, Kuszpit K, Zasadny K, Qiu M, Painter CL, Wong effusions. Acta Cytol 2006, 50:603–607. A, Kraynov E, Arango ME, Mehta PP, Popoff I, Casperson GF, Los G, Bender S, 29. Veatch AL, Carson LF, Ramakrishnan S: Differential expression of the Anderes K, Christensen JG, VanArsdale T: PF-03732010: a fully human cell-cell adhesion molecule E-cadherin in ascites and solid human monoclonal antibody against P-cadherin with antitumor and ovarian tumor cells. Int J Cancer 1994, 58:393–399. antimetastatic activity. Clin Cancer Res 2010, 16:5177–5188. 30. Davidson B, Berner A, Nesland JM, Risberg B, Berner HS, Trope CG, 51. Kenny HA, Krausz T, Yamada SD, Lengyel E: Use of a novel 3D culture model Kristensen GB, Bryne M, Florenes AV: E-cadherin and alpha-, beta-, and to elucidate the role of mesothelial cells, fibroblasts and extra-cellular gamma-catenin protein expression is up-regulated in ovarian carcinoma matrices on adhesion and invasion of ovarian cancer cells to the cells in serous effusions. J Pathol 2000, 192:460–469. omentum. Int J Cancer 2007, 121:1463–1472. 31. Kang HG, Jenabi JM, Zhang J, Keshelava N, Shimada H, May WA, Ng T, Reynolds CP, Triche TJ, Sorensen PH: E-cadherin cell-cell adhesion in doi:10.1186/1476-4598-13-170 ewing tumor cells mediates suppression of anoikis through activation of Cite this article as: Ko and Naora: HOXA9 promotes homotypic and the ErbB4 tyrosine kinase. Cancer Res 2007, 67:3094–3105. heterotypic cell interactions that facilitate ovarian cancer dissemination 32. Kumar S, Park SH, Cieply B, Schupp J, Killiam E, Zhang F, Rimm DL, Frisch via its induction of P-cadherin. Molecular Cancer 2014 13:170. SM: A pathway for the control of anoikis sensitivity by E-cadherin and epithelial-to-mesenchymal transition. Mol Cell Biol 2011, 31:4036–4051. 33. Taniuchi K, Nakagawa H, Hosokawa M, Nakamura T, Eguchi H, Ohigashi H, Ishikawa O, Katagiri T, Nakamura Y: Overexpressed P-cadherin/CDH3 promotes motility of pancreatic cancer cells by interacting with p120ctn and activating rho-family GTPases. Cancer Res 2005, 65:3092–3099. 34. Thiery JP, Acloque H, Huang RY, Nieto MA: Epithelial-mesenchymal transitions in development and disease. Cell 2009, 139:871–890. 35. Shi X, Bai S, Li L, Cao X: Hoxa-9 represses transforming growth factor-beta-induced osteopontin gene transcription. J Biol Chem 2001, 276:850–855. 36. Lessan K, Aguiar DJ, Oegema T, Siebenson L, Skubitz AP: CD44 and beta1 integrin mediate ovarian carcinoma cell adhesion to peritoneal mesothelial cells. Am J Pathol 1999, 154:1525–1537. Submit your next manuscript to BioMed Central 37. Heyman L, Kellouche S, Fernandes J, Dutoit S, Poulain L, Carreiras F: and take full advantage of: Vitronectin and its receptors partly mediate adhesion of ovarian cancer cells to peritoneal mesothelium in vitro. Tumour Biol 2008, 29:231–244. • Convenient online submission 38. Ko SY, Lengyel E, Naora H: The Mullerian HOXA10 gene promotes growth • Thorough peer review of ovarian surface epithelial cells by stimulating epithelial-stromal interactions. Mol Cell Endocrinol 2010, 317:112–119. • No space constraints or color figure charges 39. Chen GT, Tai CT, Yeh LS, Yang TC, Tsai HD: Identification of the cadherin • Immediate publication on acceptance subtypes present in the human peritoneum and endometriotic lesions: potential role for P-cadherin in the development of endometriosis. • Inclusion in PubMed, CAS, Scopus and Google Scholar Mol Reprod Dev 2002, 62:289–294. • Research which is freely available for redistribution 40. Van Marck V, Stove C, Van Den Bossche K, Stove V, Paredes J, Vander Haeghen Y, Bracke M: P-cadherin promotes cell-cell adhesion and Submit your manuscript at counteracts invasion in human melanoma. Cancer Res 2005, 65:8774–8783. www.biomedcentral.com/submit

Journal

Molecular CancerSpringer Journals

Published: Jul 14, 2014

There are no references for this article.