Identification of new molecules and biomarkers involved in the initiation and progression of ovarian cancer

Resumo

GLOBAL SUMMARY & DISCUSSION Cancer is responsible for approximately 13% of the world¿s mortality and this percentage is continually rising [391]. The World Cancer Research Foundation (WCRF), the international organization that leads a global network of cancer charities dedicated to the prevention and control of cancer, has warned that in the last decade the number of cancer cases have increased by 20%. Research is needed to achieve earlier detection and more efficient treatment and prevention strategies. It is well established that the major causes of failure in cancer treatment are invasiveness and metastasization Therefore, the identification of molecules and/or signaling pathways involved in these processes is crucial to the development of an effective therapy for those types of cancers like OC, for which early-stage detection is still a barrier [392]. Most cases of OC are still diagnosed in advanced stages (>75% of cases are stage III/IV), when the disease has already metastasized. Early diagnosis is difficult to achieve due to the lack of specificity of early disease symptoms. In advanced cases the current standard of care combines radical surgery and platinum-paclitaxel chemotherapy given every three weeks for 6 cycles. Surgery is rarely able to render patients free of disease and postoperative chemotherapy is required in most cases [393, 394]. First-line chemotherapy has yielded response rates over 80% in late stage disease patients; however, most of those patients will eventually relapse and die. Consequently, advanced stage disease in OC is still associated with poorer survival outcomes [2]. Currently, two main aspects of OC are being investigated: 1) the identification of early events in ovarian carcinogenesis that could yield markers for early detection (before metastases develop) and markers for monitoring disease progression, and 2) the development of tools to overcome chemoresistance. The success of targeted therapies in other cancers has encouraged research in this area for OC. Despite key advances in surgical and chemotherapeutic strategies, these approaches have led to small improvements in outcome [395]: In fact, the five-year survival rate for OC has increased by only 8% in the last 30 years (OC Research Foundation, 2011). Aberrant expression and modification of members of the ETS transcription factor family has been observed in numerous cancers. ETS transcription factors have been reported to regulate growth-, apoptosis-, angiogenesis-, invasion- and metastasis-related genes in tumor cells [231, 234, 396, 397]. This is largely due to their ability to activate different kind of molecules such as proteases, integrins and adhesion molecules [293]. Specifically, an association between the expression of some ETS members (PEA3 and ETS1) and the expression of some integrin subunits, MMP, TIMP-2 and angiogenic genes, which play a central role to both tumor invasion and angiogenesis, has been found in OC [298, 398, 399] and has also been associated with poor survival [295, 296, 298]. ETV5 transcription factor is a member of the PEA3 subfamily of ETS transcription factors and one of the main molecules studied in this thesis. FGF and ERK-MAPK signaling pathways have been associated with the expression of this molecule [242, 342]. Mice models with targeted disruption of ETV5 have demonstrated its role in the development of the germ cells. Firstly, problems in the spermatogenesis process affecting the spermatogonial stem cell self-renewal have been observed in male mice with disruption of ETV5 [343]. Secondly, the knock down of ETV5 in female mice models has shown complex ovarian defects in tissue architecture that are manifested as reduced developmental competence of oocytes after fertilization as well as mating and ovulation defects [347]. Our group has previously characterized the up-regulation of the ETV5 gene in EC, which has shown a specific and significant increase in those tumor stages associated with myometrial invasion [356, 359]. In addition, we have demonstrated that the overexpression of ETV5 promotes cell migration and invasion through MMP2 activation in the HEC1-A endometrial cancer cell line [330]. Following the research that studied the involvement of ETV5 in EC, we decided to examine the role of ETV5 transcription factor in EOC. Our first approach was to examine if ETV5 was up-regulated in OC samples. We found ETV5 up-regulated in ovarian tumor samples compared to ovarian tissue controls. Next, we examined the biological effects of ETV5 in OC by knock down and up-regulation of ETV5 expression in two OC cell lines (OV90 and TOV112). The inhibition of ETV5 clearly modulated cell proliferation in serum-deprived conditions, suggesting its role in ovarian tumor progression. Significantly, the in vitro modulation of ETV5 levels in two OC cell lines was able to regulate E-cadherin expression. The inhibition of ETV5 down-regulated E-cadherin in OV90 cells, while the overexpression of ETV5 up-regulated E-cadherin expression in TOV112 cells. We found a clinical correlation between E-cadherin and ETV5 expression in human ovarian tumor samples, suggesting that ETV5 may contribute to the up-regulation of E-cadherin expression during ovarian tumor development. We found that the regulation of E-cadherin levels through changes in ETV5 expression were associated with changes in the expression of integrins ¿5 and ß1. The ectopic expression of ETV5 in the TOV112 ovarian cancer cell line induced E-cadherin and ß1 integrin expression and enhanced cell survival when cells were grown in a spheroid model. Similar results were reported by Casey and colleagues [205], who found that ß1 integrins regulated the formation and adhesion of ovarian spheroids in vitro. We hypothesize that the up-regulation of ETV5 would act as an antiapoptotic factor in ovarian spheroids, enhancing cell dissemination when the cells are shed from the primary tumor through the modulation of cell to cell adhesion proteins. Integrins are also known to be the main mediators of cell to matrix adhesion. We found that ETV5 could modulate adhesion to the extracellular matrices fibronectin and collagen type I, which are matrix components secreted by mesothelial cells lining the peritoneal wall. ¬¬Our results suggest that the up-regulation of ETV5 detected in ovarian tumors may enhance the cell attachment of ovarian cells to the mesothelial monolayer during OC cell dissemination. Other studies based on OC spheroids have demonstrated that integrins can also mediate the disaggregation of OC spheroids on ECM through the activation of metalloproteases [400]. Our findings suggest that the overexpression of ETV5 detected in ovarian cancer cells may contribute to ovarian tumor progression through the ability of ETV5 to enhance ovarian cancer cell proliferation in a tumor microenviroment with a lack of cell nutrients. In addition, the up-regulation of ETV5 would also play a role in ovarian cancer cell dissemination and metastasis into the peritoneal cavity by protecting OC cells from apoptosis and by increasing the adhesion of ovarian cancer cells to the peritoneal wall through the regulation of cell adhesion molecules. Our results on ETV5 and ovarian cancer cell progression and dissemination, together with our previous findings involving ETV5 in EC migration and invasion [330], highlight the role of ETV5 as a key regulator of the invasive phenotype in both tumor types. Future research will identify how ETV5 regulates and alters the expression of ETV5 target genes promoting tumor cell invasion and dissemination. In order to understand the molecular pathways associated with ETV5 involved in ovarian cancer cell progression and dissemination, in the second part, we analyzed the expression profile of ETV5 down-regulation in ovarian cancer cells by gene expression microarray technology. Analysis of the deregulated genes in OV90i4 cells pointed to the TGFß signaling and cell cycle pathways as the main networks altered in OV90 cells with ETV5 down-regulation. It is well known that TGFß plays a dual role as tumor suppressor and pro-oncogenic factor [401-404]. In advanced cancer stages, it promotes tumor invasion, angiogenesis and metastasis [405]. In OC, TGFß enhances OC metastatic potential through the induction of an Epithelial to Mesenchymal Transition (EMT) [406, 407]. In addition, ETS transcription factors have already been described to mediate TGFß induced EMT [408-410]. Also, ETS1 oncogenic activity has been shown to be mediated by the TGF¿ mitogen [408]. Our previous work showed that the down-regulation of ETV5 induces EMT through the induction of Zeb1, which delays the rate of cell proliferation [411]. Analysis of the TGFß pathway in OV90 controls and in OV90i4 cells showed an interaction between the TGFß pathway and the ETV5 transcription factor. This confirms that the analysis of gene interactions using the IPA software is a useful approach for identifying key pathways. In addition, in our analysis of Network 2 cell cycle regulation was found to be altered in OV90i4 cells with cyclin B1 indicated as the core molecule. ETS transcription factors have been shown to activate the expression of cyclin D1 and D3 [412, 413]. We confirmed the down-regulation of cyclin B1 in OV90i4 cells, suggesting that the observed decrease in cell proliferation was due to an inhibition of cell cycle progression. Several studies have described the overexpression of FOXM1 in various human malignancies, including prostate, breast, lung, ovary, colon, pancreas, stomach, bladder, liver and kidney [414] cancers, suggesting that it may play a role in the development and progression of human cancers. We found that FOXM1 is regulated by ETV5 in OC cells through the direct binding of ETV5 to its promoter region. Our group has previously described the role of ETV5 in the protection against oxidative stress [357], and we suggest that this effect may be partly due to the regulation of FOXM1. Increased levels of ROS are observed in OV90 cells with ETV5 down-regulation, which indicates that these two factors are associated. We demonstrated that in OV90 cells, exogenous H2O2 was able to induce an increase in FOXM1 protein levels concomitant to an increase in ETV5 protein levels. Interestingly, in tumor cells ROS is an important factor that directly regulates the expression of ETS1, the founder member of the ETS family of transcription factors [415, 416]. Similarly, H2O2 is able to transcriptionally up-regulate ETS1 in OC cells [417]. There is also evidence that ETS2, a closely related family member of ETS1, is transcriptionally up-regulated by H2O2 treatment in NIH3T3 fibroblasts and that cells with defects to their antioxidant defense systems display endogenously increased levels of ETS2 [418]. The concomitant regulation of FOXM1 through the direct binding of ETV5 to its promoter suggests that ETV5 may partly protect cells from oxidative stress through the up-regulation of FOXM1. It has been previously shown that FOXM1 protects cells from oxidative stress, regulating the intracellular levels of ROS through the up-regulation of antioxidant genes [419]. To better understand the putative role of FOXM1 in OC, we analyzed the expression of FOXM1 in a gene expression microarray experiment previously performed in our lab. ETV5, FOXM1 and its target genes [420] were up-regulated in ovarian tumors. FOXM1 overexpression was further confirmed in a new set of ovarian tumor samples by RT-Q-PCR. We found a significant association between levels of FOXM1 mRNA expression and tumor grade (Student¿s ttest, p<0.05). In breast cancer patients, it has been shown that a higher expression of FOXM1 is associated with poor prognosis [421]. Moreover, FOXM1 expression could also serve as an independent predictor of poor survival in gastric cancer [422]. Our results indicated that in OC increased levels of FOXM1 expression were also associated with a more aggressive phenotype, suggesting a role of this molecule in the progression of OC. In addition, we found that the expression of ETV5 and FOXM1 were clinically correlated in human ovarian tumor samples. This suggests that ETV5 may contribute to the up-regulation of FOXM1 during ovarian tumor development. The overexpression of FOXM1 is has been previously reported in various malignancies [421-427]. However, to our knowledge this is the first study that describes FOXM1 overexpression in OC (reviewed in [428]). FOXM1 plays a key role in cell cycle progression at the S and G2/M phases (mitotic division) [420, 429], and it has been shown to regulate the maintenance of chromosomal segregation and genomic stability [420]. FOXM1 has been shown to correlated with oncogenic stress [419] and to counteract oxidative stress-induced premature senescence [430]. A recent report has also hypothesized that FOXM1 may induce cancer initiation through stem/progenitor cell expansion [431]. FOXM1 can also promote drug resistance to herceptin, paclitaxel [432] and cisplatin [433] in breast cancer cells. In conclusion, the analysis of the genes and signaling pathways under the control of ETV5 in OV90 cells has unraveled new signaling pathways that interact with ETV5. In addition, it has confirmed the protective role of this transcription factor against oxidative stress during oncogenesis, as well as having identified FOXM1 as a target gene overexpressed in OC. Future research will identify the biological role of FOXM1 in OC tumorigenesis. More recently, two studies have highlighted the critical importance of this transcription factor in the pathophysiology of the serous ovarian carcinomas [434] and in the mediation of the metastatic potential of OC cells through the activation of the ERK signaling pathway [435]. Interestingly, FOXM1 has also been described as an important mediator of cisplatin [436] and paclitaxel [432] resistance (current chemotherapeutic agents in the OC treatment), and as a master regulator of tumor metastasis by inducing EMT [437]. Collectively, all these data suggest that over-expression of FOXM1 may confer metastatic and chemoresistant capabilities to OC cells The impairment of metastatic potential of cancer cells by FOXM1 inhibitors highlights the potential role of these inhibitors in the treatment of advanced ovarian tumors. Findings about the crucial role of FOXM1 in the development and progression of many human cancers and the possible targeted therapies against this transcription factor have been recently summarized by Wang and colleagues [438]. Although the transition from early- to advanced-stage OC is a critical determinant of survival, little is known about the molecular bases of OC metastasis. Over the past decade, gene expression profiling has permitted comprehensive genetic profiling of cancer by allowing the simultaneous study of tens of thousands of genes. As a result, multigene signatures that can classify histological tumor subtypes predict clinical end points and provide insights into the mechanisms of cell growth and drug resistance in cancer have been discovered. In the third task of this thesis, we aimed at identifying new molecules involved in the progression of OC and also new prognostic biomarkers [362, 439-441] by comparing the gene expression of five human paired samples: ovarian primary tumors, ascites and metastatic peritoneal implants. Previous studies using microarray data had already generated a substantial number of potential new markers, but only some of them had been validated in small patient cohorts. These markers include prostasin, ostopontin, HE4, kallikrein and brain creatine kinase. The results of these studies support the concept that novel tumor markers might be identified by gene profiling [442]. Ongoing efforts to reanalyze microarray data after combining several existing datasets might overcome some limitations of single studies (use of different controls, processing of samples, small sample size and other). Microarray techniques have also provided mechanistic insights into the biology of OC progression. The results of our microarray analysis reveal minor differences in the gene expression profiles between primary tumors and their matched metastasis, confirming the results shown by Adib and colleagues [362] which support the idea that these tumors have already acquired the metastatic phenotype prior to their dissemination. However, the differences in gene expression profile found in ascites may be due to the particular characteristics of these cells. Tumor cells in ascites are detached from the primary tumors into the peritoneal cavity, where they aggregate as spheroids in order to survive under anchorage independent conditions. These spheroids can then attach to the extraovarian mesothelial wall and subsequently invade, establishing tumors at secondary sites. It is already known that several genes, such as adhesion and proliferation molecules, change their expression to adapt to these particular anchorage independent conditions [364]. In contrast, the three dimensional characteristics of primary tumors and metastasis, as well as the microenvironment where they are placed, are very similar [362]. The analysis of the differentially expressed genes revealed that among the biological processes most significantly altered in the ascites versus the primary tumor group were cell death, cell to cell signaling and interaction and tissue development. Of the proteins involved in this network, p53 was identified as the core molecule. The tumor suppressor p53 plays a central role in coordinating responses, including cell cycle arrest, DNA repair, senescence or apoptosis and stresses induced by a wide array of stimuli. As a major apoptosis regulator, p53 plays a critical role in anoikis and metastasis. p53-dependent anoikis has been demonstrated in many cell types, including epithelial cells [365]. Accordingly, we have found that inhibition of cell death is one of the main adaptive changes that tumor cells acquire when they detach from the primary tumor and must survive under anchorage independent conditions. For our biomarker discovery strategy, we analyzed the genes up-regulated between metastasis and primary tumors. We selected six up-regulated genes (FABP4, MXRAS, INHBA, PXDN, MUC16 and GREM1), based on their p-values and differential fold change. Based on an extensive literature search and antibody availability, we decided to focus our studies on the following genes: FABP4, INHBA, GREM1, SFRP2, and ADAM12. MUC16 corresponds to the tumor associated antigen CA-125, a widely known biomarker for OC detection and progression [29]. This finding supported our biomarker discovery strategy. Interestingly, FOXM1 expression also appeared highly up-regulated in metastasis versus primary tumors, thus confirming our previous work. Experiments to validate protein levels were done by WB and IHC. Protein extraction from the same set of matched primary tumor, ascites and metastasis samples validated the up-regulation of FABP4, INHBA and GREM1 genes at the protein level. Finally, an independent set of paired primary tumor and metastasis tissue slides were used to validate the expression levels of four genes (FABP4, INHBA, GREM1, SFRP2). In summary, our results suggest that FABP4 could be a good biomarker for neoangiogenesis associated with metastatic growth. FABP4 expression analysis in primary tumors associated with clinical outcome will determine the potential use of this protein as a prognostic biomarker, as well as clarify its role in ovarian tumorigenesis. We also found a moderate increase in INHBA expression in the metastatic samples. Similarly, the analyses of more samples will determine the putative role of this protein as a new prognostic biomarker. Previous microarray-based studies carried out by Adip and colleagues [362], in which four normal ovarian samples plus six paired primary and secondary samples from the same individual were analyzed, FABP4 was also found overexpressed in metastasis. In another study done by Lancaster and colleagues [440] , overexpression of FABP4 was also detected after screening different sites of 47 omental metastasis from 20 OC patients. In a different work done by Bignotti and colleagues [441], in which data from multiple cancers were analyzed using a computational method, a set of genes that indicates high-stage cancer was identified and INHBA expression was also detected as overexpressed in metastasis. Finally, in a recent study done by Kim and colleagues [384], the results showed that INHBA was overexpressed in metastasis in an analysis of 14 primary and 17 omental metastasis unpaired samples. In this last study, the prominent presence of INHBA in the signature of all cancers strongly suggested a biological mechanism centered on activin A induced TGFß signaling. Top-ranked genes associated with carcinoma stage in four ovarian and colorectal cancers were also included: INHBA, GREM1 and ADAM12. Although some microarray studies had been performed on the molecular bases of OC development, no previous research had simultaneously analyzed primary tumor, ascites and peritoneal implants of the same patients. The up-regulation of molecules such us FABP4, INHBA and GREM1 is shared with previous studies, and contributes to highlight the possible importance of these proteins in OC development. Further studies are required to determine whether some of these identified molecules have a prognostic significance in OC patients. Also, this future research should consider testing new antibodies against the same molecules. Moreover, other genes differentially expressed in our metastasis versus primary tumor analysis should be investigated as potential candidates for ovarian biomarkers. In addition the molecules up-regulated in AvsT and MvsA need further consideration. Finally, in the fourth part of this thesis we studied CTC from OC. Our aim was to screen and evaluate novel surface markers for OC CTCs detection and isolation. The search of biomarkers for CTC isolation was based on a set of differentially expressed genes derived from a comparative microarray analysis of EOC tissue and peripheral blood, performed in the course of the 6th EC framework project ¿OVCAD ¿ Diagnosis of a Silent Killer¿. With the use of some recently described markers [388], we established a method for detecting and isolating specific OC CTCs using magnetic Dynabeads and antibodies binding to extracellular expressed gene products of the above mentioned gene set. In total, eight antibodies against eight surface proteins and four cell lines were firstly tested by flow cytometry to examine the expression of the cell surface proteins. Next, isolation of OC cells using magnetic beads was tested. Different magnetic beads antibody-based methods were used to measure the efficiency of the selected cell surface markers. Efficacy of the bead-mediated detection of cells was visually checked using a microscope, which also permitted single cell picking, a useful method to further characterize the cells. Biotinylation of the primary antibodies in combination with streptavidin coupled magnetic beads yielded more positive cells and more beads per cell. At this point, the sensitivity of the bead mediated detection of tumor cells spiked into control blood samples needs to be verified in future experiments. Similarly, the specificity of the method needs to be tested using non-spiked control blood samples. The sensitivity and specificity of the bead mediated detection of tumor cells using the novel panel of antibodies needs to be compared to the detection of tumor cells using known cell surface markers. All these experiments will be continued in Dr. R. Zeillinger laboratory (Medical University of Vienna). Although CTCs are not considered the most common way of OC dissemination, the idea that hematogeneous spread of OC could play an important role in OC progression has been explored in several studies [156-160]. The results of this research have shown that CTCs can predict Progression Free Survival (PFS) and Overall Survival (OS) in patients with relapsed/recurrent advanced OC [156, 216, 219, 220]. Low percentages of CTCs (12%-18.7%) have been shown in previous studies [161]. Moreover, the presence of CTCs defined by anti-epithelial antibody markers, such us EpCAM, is controversial due to the high variation observed between tumors and leukocytes in its gene expression [443]. Similarly, the analysis of hMAM in blood showed high variation in breast cancer patients [444]. Consequently, the identification of new markers for the detection of CTCs should be considered a priority. In conclusion, the data presented in this thesis aimed at improving our understanding of the process of OC dissemination. It is widely accepted that spreading of malignant tumors is a process in which a large number of molecules participate. Understanding the molecular basis of ovarian carcinoma metastatic spread has the potential to refine the management of these tumors and to develop novel, more specific and more effective treatment strategies. Moreover, the screening and selection of candidates for further serological study will benefit the diagnosis and prognosis of OC patients. A combination of biomarkers rather than a single biomarker such as CA-125, is more likely to achieve a signature specific for epithelial ovarian carcinoma. More importantly, this research aspires to contribute to the concerted effort to improve the diagnosis, prognosis and treatment of OC patients.