Characterization of MicroRNA-200 pathway in ovarian cancer and serous intraepithelial carcinoma of fallopian tube

Ovarian cancer cell lines DOV13, CAOV3, OVCA420, OVCA429, OVCA432, OVCA680, and OVCA810 were established in our laboratory from ovarian carcinomas obtained from different patients. SKOV3 was obtained from American Tissue Culture Collection (ATCC). RMG-1 was purchased from Japanese Collection of Research Bioresources (JCRB Cell Bank). All ovarian cancer cell lines except RMG-1 and OCA810 were grown in medium 199 (Sigma-Aldrich, St. Louis, MO) and MCDB 105 (Sigma-Aldrich, St. Louis, MO) (1:1) supplemented with 10% fetal calf serum (FCS). RMG-1 and OVCA810 were cultured with DMEM-F12 HAM medium (Life Technologies, Grand Island, NY) plus 10% FCS. Ovarian surface epithelial (OSE) and fallopian tube epithelial (FTE) primary cultures were derived from ovarian and distal fallopian tube scrapes from patients who underwent prophylactic salpingo-oophorectomies. Ovarian and fallopian tube specimens were collected with an IRB approved protocol at Brigham and Women’s Hospital in Boston.

A Leica CM3050S cryostat was used to cut 7 μm-thick sections from frozen tissues embedded in OCT compound. The sections were placed on a polyethylene terephthalate (PET) UV-absorbing membrane-slide (Leica Microsystems), fixed using absolute methanol, and then stained with Hematoxylin to reveal tissue morphologies. A Leica AS LMD laser microdissection system (Leica Microsystems) was used to procure pure cell populations from the tissue sections and the RNA was extracted using an RNAqueous kit (Life Technologies, Carlsbad, CA). TRIzol reagent (Life Technologies, Carlsbad, CA) was used to extract RNA from primary cultures. TaqMan MicroRNA Reverse Transcription Kit and TaqMan MicroRNA assay kits (Applied Biosystems, Foster City, CA) were used for qRT-PCR determination of miRNA levels, performed on a 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA). RNU6B RNA was used as the internal control to normalize sample input. For target gene and effector gene determination, High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) was used for cDNA synthesis. The synthesized cDNAs were adjusted in 1× TaqMan PreAmp Master Mix (Applied Biosystems, Foster City, CA), mixed with a pool of TaqMan Gene Expression Assays for target and effector genes, and a 14-cycle multiplexed pre-amplification PCR reaction was performed. The pre-amplified cDNAs and individual TaqMan Gene Expression Assays were loaded to individual wells of a BioMark 48.48 Dynamic Array (Fluidigm Corporation, South San Francisco, CA), which allowed simultaneous run of 48 qPCRs for 48 samples on a BioMark System (Fluidigm Corporation, South San Francisco, CA). Data collection and analysis was performed automatically by the Fluidigm instrument. Gene expression levels were determined using 2 ^-ΔΔCT method [ 29]. The data were exported to TIGR MEV (MultiExperiment Viewer) V4.0 ( http://​www.​tm4.​org) for heat map construction.

Normal OSE and FTE cells were seeded at 2 × 10 ⁵/well in a 6-well culture plate (BD Biosciences, San Jose, CA). Ambion Pre-miR miRNA precursors for miR-200 family members and miR-205, as well as the nontarget negative control precursor, were purchased from Life Technologies (Carlsbad, CA). Transfection of the precursors into the cells was performed using the siPORT NeoFX transfection agent (Life Technologies, Carlsbad, CA) according to the manufacturer manual. The cells were harvested after 2 days and RNA was extracted for qRT-PCR.

FTE cells were seeded at 1 × 10 ⁴/well in an 8-well chamberslide (BD Biosciences, San Jose, CA). Cells were transfected with the miRNA precursors on the second day, using siPORT NeoFX transfection agent as described above. After 4 days of incubation, cells were fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO) and permeabilized with PBS containing 0.5% Triton X-100 (Sigma-Aldrich, St. Louis, MO). After blocking with 10% FBS, mouse monoclonal antibodies for CA125 (Thermo Fisher Scientific, Waltham, MA) and EpCAM (Dako North America, Inc. Carpinteria, CA) were added and incubated at room temperature for 2 h, then with Alexa Fluor 647 conjugated secondary antibodies (Invitrogen, Carlsbad, CA). A mounting medium with DAPI (Vector Laboratories, Burlingame, CA) was used for counterstaining. Microscopic images were captured by a Leica DM IRE2 fluorescence microscope (Leica Microsystems, Bannockburn, IL) and analyzed by the OpenLab Cell Imaging System software (Leica Microsystems, Bannockburn, IL). Lysates of two OSE and two FTE primary cultures and OVCA420 cell line were analyzed using standard Western blotting with CA125 antibody (Abcam, Cambridge, MA) and chemiluminescent signals were revealed using WesternBright ECL kit (Advansta, Menlo Park, CA).

Our previous gene expression profiling study of ovarian 3D cellular structures growing sequentially in Matrigel and then collagen I matrices which mimic inclusion cysts and invading tumors has identified concurrent elevated expression of miR-200 family members and effector protein CDH1, and decreased expression of transforming growth factor β (TGFβ) pathway genes in the cancer 3D cultures compared to the 3D cultures derived from normal OSE cells [ 26]. Suppression of CDH1 expression in ovarian cancer cells disrupted inclusion cyst formation and collective movement demonstrated in the cancer 3D cultures [ 26]. Identification of miR-200 targets is the key to understand the function of these miRNAs in ovarian cancer. E-cadherin transcriptional repressors ZEB1 and ZEB2 are two of the targets of miR-200 family and miR-205 [ 27]. Burk et al. have also identified TGFβ as another potential target of the miR-200 family [ 28]. Aigner et al. have shown that the transcription factor ZEB1 regulates besides E-cadherin other effector genes that are key determinants of epithelial cell phenotype [ 30]. TGFβ, the other miR-200 target, suppresses epithelial cell polarity by activating TGFβ receptors to phosphorylate the polarity complex protein Par6, which results in ubiquitination and degradation of Rho GTPase family member RhoA and the loss of tight junctions [ 31]. The members of the Par polarity complex interact with Lethal giant larvae (Lgl), a member of the basolateral Scribble complex and control apicobasal polarity, cell proliferation, migration and epithelial structure during tumor development [ 32]. Hence, the net results of miR-200 overexpression will be down-regulation of the target genes ZEB1, ZEB2, and TGFβ, with the corresponding up-regulation of multiple epithelial markers and more differentiated phenotype (MET) of the cells.

To evaluate expression of the miR-200 family and the associated pathway in ovarian cancer, we performed a large-scale quantitative real-time reverse transcription-polymerase chain reaction (qRT-PCRs) study with RNA from a panel of enriched ovarian cells from either primary cultures or laser-capture microdissection. These included 15 normal OSE preparations derived from five primary scrapes and microdissection from 10 archived frozen normal ovarian tissues, and epithelial tumor cells microdissected from three benign tumors, six borderline tumors, 14 nonserous malignant tumors (six mucinous, one clear cell, and seven endometrioid), 26 serous malignant tumors, 11 ovarian cancer cell lines, and six stromal preparations microdissected from ovarian tumor tissues. In this study, we also explored the potential involvement of miR-200 pathway in the hypothesized fallopian tube precursor lesions of high-grade serous tumors and hence have also collected four primary cultures of normal fallopian tube epithelial (FTE) cells derived from distal fallopian tube scrapes, epithelial tumor cells and stromal cells microdissected from three cases of frozen STIC tissues. qRT-PCRs were performed to determine the expression levels of the five members of the miR-200 family and miR-205, another miRNA involved in MET [ 27]; the target genes ZEB1, ZEB2, TGFβ1, and TGFβ2; and a panel of potential effector genes downstream of the ZEB1 transcriptional repressor selected from the article by Aigner et al. [ 30]. These included cadherins, claudins, LLGL2 and tumor epithelial marker EpCAM. We have also included in the effector gene list two well-known ovarian tumor markers, Muc16 (CA125) and HE4.

The heat maps in Fig. 1 serve as a visual presentation of the qRT-PCR results for miRNA and target genes. Unlike miR-205, the miR-200 family members showed in general upregulated expression in ovarian tumors and cancer cell lines when compared with normal OSE cells (Fig. 1a). Non-parametric Kruskal Wallis analyses of the qRT-PCR data showed that there were significant differences among groups in the miRNA expression (Table 1). Multiple comparisons using Mann-Whitney U test showed that miR-200 family members were all significantly overexpressed in the borderline (BOT), serous (SMT) and non-serous malignant tumors (NSMT), and cancer cell lines (CCL) compared with normal OSE cells. MiR-141 was the only RNA that was also significantly overexpressed in benign tumors (BNT) ( P = 0.030). In our analysis, miR-205 was overexpressed only in SMT and CCL when compared with normal OSE cells. Albeit not significant in multiple comparisons, miR-200 expression appeared to be more downregulated in the ovarian stroma (OV_ST) than in OSE cells, in consistence with the notion that miR-200 are for epithelial differentiation. For the fallopian tube cells, there was an apparent increased expression of miR-200 in the STIC tumor cells than normal FTE cells (Fig. 1b), although multiple comparisons did not show significance because of the small sample size (Table 1).

For the target genes, ZEB1 was significantly underexpressed only in CCL ( P = 0.048) and overexpressed in Ov_ST ( P = 0.048) when compared with normal OSE cells (Fig. 1 and Table 2). On the other hand, ZEB2 was broadly underexpressed in BOT ( P = 0.048), NSMT ( P = 0.002) and CCL ( P = 0.001), but was overexpressed in Ov_ST ( P = 0.012) when compared with normal OSE cells. TGFβ1 was significantly underexpressed in BOT ( P = 0.036), NSMT ( P = 0.001) and SMT ( P = 0.006); TGFβ2 was underexpressed in NSMT ( P = 0.006) and SMT ( P = 0.012) when compared with normal OSE cells. Similar to the miRNA results, the target genes were not significantly different between normal FTE cells and STIC and FT stroma (FT_ST) on multiple comparisons, even though overall P-values were significant for ZEB1, ZEB2 and TGFβ1.

We also performed correlation analyses to determine the expression relationship within the miRNA group and between miRNA and target genes in all samples (Additional file 1: Table S1). MiR-200 family members showed very high expression correlation with each other (Pearson correlation coefficients r range between 0.666 and 0.889). MiR-205 showed modest correlations with miR-200b ( r = 0.52) and miR-200c ( r = 0.491). MiR-200b, miR-200c and miR-205 showed significant negative correlation with the expression of target genes ZEB1 and ZEB2, but not with that of TGFβ1 and TGFβ2, suggesting that the miRNA regulated MET primarily through ZEB1 and ZEB2. However, the expression of ZEB1 showed strong correlations with ZEB2, TGFβ1 and TGFβ2. There was also significant correlation between TGFβ1 and TGFβ2 expression (Additional file 1: Table S1).

For the 21 effector genes that we measured (Fig. 2 and Table 3), all five categories (BNT, BOT, NSMT, SMT, and CCL) of ovarian tumor cells showed significantly elevated levels of CDH1, PPL, EVA1, TSPAN1, SH3YL1, and TMEM30B expression than normal OSE cells. In addition, CRB3, LLGL2, CLDN7, SFN, EpCAM, PATJ, SCEL, PKP3, MAL2 and MUC1 were elevated in BOT, NSMT, SMT, and CCL. In contrast, the expression of CDH11 was negatively associated with these four tumor cell categories, consistent with the finding that this cadherin was associated with TGFβ production and EMT [ 33]. It was noted that the stromal cells extracted from tumor samples also expressed significant levels of CDH3, LLGL2, EVA1, EpCAM, TSPAN1, SCEL, and SH3YL1, suggesting that the tumor stroma also exhibited changes during the tumorigenesis process. Correlation analyses (Additional file 1: Table S2) showed that the expression of all miR-200 family was closely correlated. The expression of CRB3, CLDN7, EpCAM, and PKP3 was strongly associated with the expression of all miR-200 family members, whereas the expression of OCLN and MAL2 was highly correlated with miR-200a and miR-200b only. A core of epithelial effector genes including CDH1, CRB3, LLGL2, CLDN7, PPL, EVA1, EpCAM, TSPAN1, and MUC1 had very high correlation coefficients ( r ≥ 0.6) with each other and apparently were co-regulated in the samples.

It is also of interest that the two clinical ovarian cancer markers, MUC16 and HE4, were not highly expressed in all tumor cell categories. HE4 was significantly elevated only in NSMT and SMT, and MUC16 was significantly elevated only in BOT, SMT, and CCL, but not in NSMT (Table 3). More significantly, mucinous malignant tumors (marked by a box in Fig. 2) showed a markedly reduced level of MUC16 expression, which was significantly different from the expression in endometrioid ( P = 0.024) and high-grade serous ( P = 0.012) malignant tumors (Additional file 1: Table S3). Tumor type-dependent expression differences of HE4 [ 34] and CA125 [ 35] have been described previously. There was a high correlation ( r = 0.646) in the expression between these two biomarkers. HE4 also had additional expression correlations with CLDN7 ( r = 0.619) and MUC1 ( r = 0.605, Additional file 1: Table S2).

For the fallopian tube cells, there were no significant differences in effector gene expression despite significant overall P-values (Table 3). Based on the fact that the heat maps for the expression of miRNAs, target genes and effector genes in fallopian tube cells were similar to the heat maps for the ovarian cells (Figs. 1 and 2), it is very likely that the lack of significant differences between STIC tumor cells and normal FTE cells was due to small sample sizes.

To explore further the relationships between microRNA, target genes, and effector genes, individuals and combinations of synthetic miR-200 precursors were transfected into normal OSE and FTE cells and the expression of target genes and effector genes was analyzed from the transfected cells using qRT-PCR. The heat maps in Fig. 3 and two-way ANOVA analysis in Additional file 1: Tables S4 and S5 show the responses of target genes and effector genes upon introduction of miR-200 and miR-205 precursor molecules. First, miR-205 was found not very effective in inducing changes in the expression of target genes ( P = 0.935 and 0.986) and effector genes ( P = 0.941 and 0.763) in OSE and FTE cells, respectively, suggesting that miR-205 does not play any important role in the miR-200 pathway in OSE and FTE cells. The miR-200 family precursor molecules, however, were effective in suppressing expression of all target genes in OSE cells and ZEB1 and ZEB2 in FTE (Additional file 1: Table S4), and induce the expression of effector genes in OSE and FTE cells (Additional file 1: Table S5). This is consistent with the correlation analysis results (Additional file 1: Table S1), which showed that the expression of miR-200 was more correlated with ZEB1 and ZEB2 expression than with TGFβ1 and TGFβ2 expression. The miR-200 family members did not contribute equally to affect the expression of target genes and effector genes. While the introduction of miR-200 precursors into both normal OSE and FTE cells stimulated the expression of cadherins CDH1 and CDH3, and effector genes like CRB3, EpCAM, TSPAN1, and PKP3 that had high correlation coefficients with miR-200 expression, there were striking differences of responses in some effector gene expression from these two normal cell types (Additional file 1: Table S5). CLDN7, the expression of which was highly correlated with miR-200 in correlation analyses (Additional file 1: Table S2), was only induced in FTE cells (Additional file 1: Table S5). Another interesting finding was the expression of MUC16, which in correlation analysis was not related to any miR-200 or any other epithelial marker expression except HE4, was found to be induced only in FTE cells (Fig. 3 and Additional file 1: Table S5). In contrast, HE4 expression was not induced at all by miR-200 introduction in both normal cell types.

We repeated the miRNA precursor transfections with different preparations of normal OSE and FTE cells and the results (Fig. 4a) confirmed the specific MUC16 mRNA expression mediated by miR-200 precursor molecules in FTE cells and not in OSE cells. Furthermore, immunostaining of the normal FTE cells confirmed the increased expression of EpCAM and CA125 proteins in the FTE cells after transfection with miR-200 precursors (Fig. 4b). A Western blot analysis performed with lysates derived from two OSE and two FTE primary cultures (Additional file 1: Fig. S1) showed that there was negligible basal expression of CA125 in both OSE and FTE cells.