Background
Ovarian cancer is a heterogeneous disease of the female reproductive tract which, despite its relatively low incidence in developed countries, carries a poor prognosis. Despite advances in the detection and treatment of ovarian cancer, it remains the 5th leading cause of cancer death in women [
1]. Epithelial ovarian cancer (EOC) comprises 90% of all ovarian cancer cases [
2]. Type I EOC primarily consists of low-grade serous, mucinous, endometrioid and clear cell subtypes, and is characterized as slow growing with intact DNA repair machinery [
3]. Type II EOC, also known as high-grade serous ovarian cancer, comprises 70% of EOC cases [
4] and is characterized by rapid growth with no identified precursor lesions, and genome instability (p53 loss) [
5]. The molecular events underlying Type II EOC remain poorly understood, and despite initial response to chemotherapy, these tumors often recur with chemo-resistance. EOC is typically diagnosed at late stage, when the tumor has spread beyond the pelvic region into the peritoneal cavity, making complete surgical removal extremely difficult. Despite recent advances in surgery and adjuvant chemotherapeutics, the overall five-year survival rate for EOC remains at only 40%, in part due to common and rapid peritoneal spread of disease, indicating the need to understand the genetic and epigenetic events underlying EOC progression.
We recently identified that the CpG island-associated promoter of
ZNF300P1, a candidate long-intergenic non-coding (linc) RNA, was hypermethylated in EOC cell lines, and this was associated with loss of expression [
6]. LincRNAs are polyadenylated RNA transcripts that are transcribed by RNA polymerase II and do not encode for protein, although they do carry epigenetic signatures similar to those found in protein-coding genes [
7]. In cancer, recent studies have demonstrated a link between lincRNA expression and disease progression and outcome related to the role of lincRNAs in gene expression regulation [
7,
8].
ZNF300P1, also known as
LOC134466, is a pseudogene of the human zinc finger protein ZNF300 (sharing 89% identity), and is characterized as a non-coding transcript. Khalil and colleagues identified
ZNF300P1 as a lincRNA using a computational algorithm that eliminates transcripts with protein-coding domains and chromatin signatures that reflect transcribed genes [
9]. Our evaluation showed that
ZNF300P1 expression was also repressed in human EOC tissues compared to normal ovarian surface epithelial cells (OSE) [
6]. Quantitative methylation analysis discriminated 27 EOC tumors from 14 normal OSE samples with a high degree of accuracy (81% sensitivity, 92% specificity [
6]), suggesting its potential as an EOC biomarker. Finally, methylation-specific headloop-suppression PCR (MSH-PCR) screening of 159 high-grade EOC tumors demonstrated methylation of
ZNF300P1 in 81% of tested tumors, suggesting that it may serve a functional role in EOC [
6]. Indeed, studies have demonstrated that lincRNAs, including
ZNF300P1, are associated with Polycomb Repressive Complex (PRC2) and CoREST chromatin modifying complex proteins [
9], supporting the notion that
ZNF300P1 may be involved in gene expression regulation and thus may serve a functional role in cancer and other processes.
The function of ZNF300P1 is currently unknown. Since ZNF300P1 is frequently and specifically methylated and down-regulated in EOC, we sought to examine its potential role in regulating cell behavior in EOC. Our findings suggest that ZNF300P1 is frequently methylated in EOC and reveal a novel function for ZNF300P1 in regulating cell polarity, motility, and adhesion.
Discussion
We have previously demonstrated that methylation of the lincRNA
ZNF300P1 is a potential biomarker in EOC [
6]. Here, we investigate the novel role that
ZNF300P1 repression may play in ovarian cancer development and progression, and show that lincRNA
ZNF300P1 plays a role in regulating cell polarity, and how loss of expression may contribute to the metastatic potential of ovarian cancer cells.
LincRNAs are gaining increasing importance both as biomarkers in cancer (HOTAIR: breast [
8], liver [
16], pancreas [
17]; MALAT-1: NSCLC [
18]; PCAT-1: prostate [
7]), and as regulators of complex and diverse biological functions. Guttman
et al. developed a screen to identify lincRNAs using histone methylation markers H3K4 and H3K36 to demarcate transcribed RNAs located outside of protein coding regions [
19]. Khalil and colleagues applied this approach in human cells, and identified
ZNF300P1 as a lincRNA and used RIP-Chip to show its association with Polycomb Repressive Complex (PRC2) and CoREST chromatin modifying complex proteins, indicating the potential for
ZNF300P1 to be involved in modification of chromatin structure [
9]. LincRNAs have also been proposed to alter transcriptional networks using four modes of action [
20]. These are: 1) as decoys to titrate DNA-binding proteins (or miRNAs [
21]); 2) as scaffolds to bring proteins together within a complex; 3) as guides to recruit proteins, such as chromatin modification complexes to DNA; or 4) as enhancers to bring distal portions of the genome into close proximity through looping. The transcriptional analyses undertaken within this study, using siRNA against
ZNF300P1, indicate that loss of
ZNF300P1 expression, as seen resulting from hypermethylation in ovarian cancer samples, results in both up- and down- regulated gene expression (Additional file
3: Table S1). While the precise mechanisms by which these genes are transcriptionally altered are outside the scope of this study, it is possible that
ZNF300P1 suppression may alter transcriptional networks via modification of chromatin, guiding the repressive complex to genes/promoters to regulate their transcription.
However, a number of recent studies have also proposed that a sub-population of lincRNAs arise from pseudogenes [
21,
22], and act as endogenous competitors, altering the distribution of miRNA molecules on their targets [
21].
ZNF300P1, also known as
LOC134466, is a pseudogene of the human zinc finger protein ZNF300. Little is known of the function of this nucleus-restricted transcription factor [
23]. Demonstrated to be up-regulated in cancer biopsies [
24], ZNF300 binds the sequence C(t/a)GGGGG(g/c)G, found in the promoter regions of genes including IL-2, IL-2Rb, CD44, p53, TNF-a, and TRAF2, which play crucial roles in various tumorigenic and inflammatory processes. Furthermore, ZNF300 induces the NF-kB pathway, in turn inducing IL-6 and IL-8, potentially exacerbating inflammation and promoting tumor metastasis [
24]. One potential means by which lincRNAs arising from pseudogenes have been proposed to function is by acting as a decoy for miRNAs targeting the protein-coding gene. Given that ZNF300 is up-regulated in cancer, and promotes inflammation and metastasis, the repression of
ZNF300P1 associated with methylation in ovarian cancer make it unlikely to be functioning as an miRNA decoy. Furthermore, our analysis of ZNF300 expression upon
ZNF300P1 silencing in HOSE17.1 cells showed no significant changes (Additional file
2: Figure S2A), indicating that it is unlikely that
ZNF300P1 acts as an endogenous decoy for miRNAs targeting
ZNF300.
Notably, we have confirmed nuclear enrichment of
ZNF300P1 (Figure
2A) [
23], and have shown that loss of
ZNF300P1 expression in ovarian cells results in the perturbation of several key pathways involved in cell cycle, cell movement, and cell-to-cell signaling and interaction (Table
1 and Additional file
2: Figure S2C). These results suggest that DNA hypermethylation of the promoter CpG island of
ZNF300P1 may play a key role in the malignant progression of ovarian cancer. While repression of the lincRNA clearly affected the ability of cells to proliferate and form colonies, it had no discernible effect on the cell cycle, indicating that
ZNF300P1 does not function as a classic tumor suppressor. Interestingly,
ZNF300P1 was recently identified as methylated in small cell lung cancer-derived cell lines [
25] potentially indicating that loss of expression of this transcript may be common to several cancer types.
The major phenotypic outcomes we identified to be associated with
ZNF300P1 silencing was decreased polarity and the resultant loss of migratory persistence (Figures
2 and
3), both key in cancer development and the aberrant spread of cancer cells [
26]. These phenotypes, together with the enhanced ability of HOSE17.1 cells lacking
ZNF300P1 to adhere to mouse peritoneum, may provide an advantage for cells to more efficiently colonize the peritoneal tissue, thus reducing the efficacy of surgical treatment in EOC.
Methods
Cell line, tissue and OSE collection and processing
Ten standard cancer cell lines derived from various subtypes of EOC: serous (SKOV3, OVCAR3, IGROV, OV90, COLO316, A2780 and CaOV3), endometrioid (TOV112D), clear cell (TOV21G) and mucinous (EFO27), as well as two human immortalized OSE cell lines (HOSE 6.3 and HOSE 17.1), were obtained and cultured as described previously [
6,
27]. Fresh-frozen tumor (FFT) samples were obtained with written informed consent from women undergoing debulking surgery for EOC at the Royal Hospital for Women (RHW, Sydney, Australia), snap frozen in liquid nitrogen and stored at −80°C. Clinico-pathological characteristics of patient samples are presented in Additional file
6: Table S2. Pathologically normal OSE cell samples were obtained with written informed consent, by brushing the ovary during surgery for non-ovarian gynecological malignancies followed by expansion of epithelial cells in culture. Cultures were evaluated for purity by staining for high molecular weight cytokeratin to exclude stromal contamination and maintained in culture as previously described [
28]. Cell pellets from passage 3 or earlier were processed for DNA extraction. Experimental procedures were approved by the Human Research Ethics Committee of the Sydney South East Area Hospital Service Northern Section (00/115).
Nucleic acid extraction and processing
Total RNA was extracted with Trizol (Invitrogen) or RNeasy mini kit (Qiagen). Genomic DNA was extracted from tumor and OSE with QiaAMP mini kits (Qiagen), from FFPE tissue with Gentra Puregene DNA isolation kit (Qiagen) and from cell lines with the Stratagene DNA extraction Kit (Agilent). 1–2 μg genomic DNA isolated from human blood (unmethylated control, Roche Applied Sciences),
in vitro methylated DNA (Chemicon International) and RNase-treated cell and tumor DNA was bisulphite-converted either using the Epitect kit (Qiagen) or as previously described [
29,
30].
Sequenom massARRAY quantitative methylation analysis
Sequenom methylation PCR assays were designed according to [
31] to interrogate methylation levels at CpG islands immediately upstream (
DCTN4,
MST150 and
ZNF300) and downstream (
GPX3 and
TNIP1) of
ZNF300P1. Primers (Additional file
7: Table S3) were optimized for bisulphite-converted DNA specificity, and tested for bias using a thermal gradient on mixes of 50:50 methylated: unmethylated template. Triplicate PCR reactions were pooled, and applied to spectrochips according to manufacturer’s instructions for MALDI-TOF analysis (Sequenom). Results were analyzed using epityper software and RseqMeth [
32] CpG methylation levels were averaged across the amplicon to determine average methylation.
Methylation-specific headloop-suppression PCR assay (MSH-PCR)
Methylation specific headloop-suppression assays (MSH-PCR) were performed as previously described [
6]. Briefly, triplicate MSH-PCRs were performed on bisulphite-converted DNA from FFPE tissue samples. By normalizing the signal to
SFN, which is methylated in all samples [
33], or bisulfite-converted genomic DNA (rDNA), relative methylation was calculated for all samples.
Knockdown conditions and transfection of siRNA
Colloidal suspensions of 25 nM
ZNF300P1 (Ambion #n263234) or non-targeting control (NT-C; Ambion #4390483) siRNA in Lipofectamine 2000 and opti-MEM (Invitrogen) were prepared for transfecting HOSE17.1 cells according to Lipofectamine 2000 transfection protocol. Peak knockdown of ~50% was observed in HOSE17.1 cells by qPCR (Primer sequences, see Additional file
7: Table S3) at 48–72 hours post transfection (Additional file
1: Figure S1C).
Transcript profiling
Biological triplicate RNA samples from 72 hours post transfection with 25 nM siRNA were submitted to the Ramaciotti Centre (University of New South Wales) for transcription, labeling and application to Affymetrix Human Gene 1.0ST mRNA transcript-profiling arrays.
Data analysis
Array data were evaluated using the LIMMA package either in GenePattern (Broad Institute, Cambridge MA, USA) or in the R environment [
34]. Briefly, array CEL files were normalized and background-corrected by the RMA method. Differential expression between probe-sets in triplicate control or
ZNF300P1 siRNA arrays was calculated using Bayesian linear models with stringency cut-offs of 1.3 fold up or down following knockdown, with an unadjusted p-value of <0.01. Evaluation of gene list properties was performed using Functional annotation (GO terms and KEGG pathways) and gene networks in Ingenuity Pathway Analysis
™ software.
Transcript localization
Nuclear and cytoplasmic RNA was isolated using Ambion’s PARIS kit, according to the manufacturer’s instructions. qRT-PCR was performed using MALAT1 and GAPDH as nuclear- and cytoplasmic-enriched controls, respectively. ZNF300 and ZNF300P1 RNA levels were determined in HOSE17.1 cells transfected with siRNAs either against ZNF300P1, or a control sequence.
Expression quantitative real-time PCR (qPCR)
qPCR assays for 18 s rRNA (cat# Hs99999901_s1), GAPDH (cat# Hs99999905_m1),ZNF300P1 (cat# Hs00859547_m1), ZNF300 (cat# Hs04177113_m1) and MALAT1 (cat# Hs00273907_s1) were purchased from Applied Biosystems and utilized for measuring gene expression. SYBR green qPCR assays were designed in PrimerExpressTM (Life Technologies; sequences in Additional file
7: Table S3)
Population growth curves
HOSE17.1 cells treated with ZNF300P1 or NT-C siRNA were seeded at 2.5x103 cells/cm2 (population growth assays) or 600 cells/cm2 (colony formation). At indicated times, cells were fixed and stained using Diff Quick (Lab Aids). For quantitation, 10% acetic acid was used to destain the plates, and absorption measured at 595 nm.
Wound healing: polarity assay and live cell tracking
Transfected cells were plated either onto glass coverslips (polarity assays), or into 6/12 well plates (live cell tracking), to reach confluence at 48 hrs post transfection. Using a 20 μl pipette tip, a wound was made in each monolayer, cells rinsed twice with medium, and incubated for 6 hrs (polarity assay) at 37°C, or into the heated chamber of a Zeiss Inverted microscope (live cell tracking). For polarity assays, cells were fixed using 4% PFA, and Golgi were stained using an anti-gm130 antibody (BD Biosciences, cat# 610823, 1/100 dilution), followed by alexa fluor-conjugated secondary, and counter-stained using phalloidin and DAPI. Slides were imaged using AxioVision (version 4.7) software, and Golgi orientation (relative to nucleus and wound space) determined using ImageJ software. For time-lapse photography, cells were imaged at 10 minute intervals over 24 hours by a Zeiss Axiovert 200 M inverted microscope and Image J software was used to track cell nuclei.
Ex vivo peritoneal adherence assay
Ovarian cancer cell peritoneal adhesion was determined using an
ex vivo assay, modified from previous studies [
35]. Briefly, the peritoneal tissue was excised from euthanized 10–12 wk female Balb/c mice, divided along the midline into two pieces and placed into serum-free media. Syto9-labeled cells (1x10
5) under each transfection condition were added to 96-well plates and peritoneal tissue laid over the wells, mesothelial surface down. The tissue was then covered by a glass coverslip and the inverted plate was incubated for 3 hrs at 37°C. The peritoneal tissue was then washed with serum-free medium, and attached cells observed and imaged using a Leica MZ16FA fluorescent dissection microscope, attached to a Leica DFC420C camera. Image J was used to count 6–9 fields per well.
Acknowledgements
The Ovarian Cancer Research Group acknowledges support from the Gynaecological Oncology (GO) Fund of the Royal Hospital for Women Foundation, Sydney, Australia. SJC and GS acknowledge support from the National Health & Medical Research Council of Australia, National Breast Cancer Foundation, Cancer Council NSW, Cancer Australia, the Australian Cancer Research Foundation, The Petre Foundation and the RT Hall Trust. The authors wish to thank Fatima Valdes-Mora and Andrew Burgess for internal review of this manuscript. Thanks to Nicola Armstrong and Elena Zotenko for statistical input, and Gillian Lehrbach, Liz Caldon and Nikki Alling for technical assistance. The authors have no financial interest to disclose.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BG and KM-J designed and performed the experiments; VL provided technical support; MG developed the normal cell lines; JS provided pathology expertise; NFH provided clinical samples and expertise; RLF was Head of the Cancer Program and originated the clinical biobank; SJC and GS conceived the idea and supervised the project. All authors have read and approved the final manuscript.