Background
Ovarian cancer is a major gynaecological malignancy worldwide with 125,000 deaths reported each year [
1]. The development of ascites and peritoneal metastases is a major clinical issue in the prognosis and management of ovarian cancer. A significant proportion of ovarian cancer cells within the peritoneal ascites exist as multicellular aggregates or spheroids which have the capacity to invade nearby organs [
2]. The pathology of peritoneal-based metastasis includes the attachment of shed primary ovarian tumor cells onto the mesothelial-lined spaces of the peritoneum in the form of spheroids resulting in multiple tumor masses necessary for secondary growth. Current treatment strategies for advanced-stage ovarian cancer patients results in initial remission in up to 80 % of patients [
3]. However, following a short remission period (usually 16–22 months), recurrence occurs in almost all patients ultimately resulting in patient mortality. This high rate of recurrence is largely due to the ability of tumor cells to evade the cytotoxic effects of chemotherapy associated with intrinsic or acquired chemoresistance, a property commonly associated with CSCs [
4,
5].
The concept of CSCs supports the existence of a sub-population of tumor cells which drive tumor growth and progression, while also sustaining the cytotoxic pressure imposed by therapy to promote the re-growth of therapy-resistant tumors [
6,
7]. In this scenario, it can be postulated that the development of an effective therapy for recurrent ovarian tumors will depend on the identification of tumor specific CSCs, as well as the pathways/regulators controlling their survival and sustenance.
Oct4 (Oct3/4 or POU5F1) is a member of the POU-domain family of transcription factors and has been shown to play an important role in the maintenance of self-renewal and pluripotency in embryonic stem cells (ESCs). It is commonly expressed in unfertilized oocytes, the inner cell mass (ICM) of a blastocyst, germ cells, embryonic carcinoma cells and embryonic germ cells [
8]. Up regulation of Oct4 expression has been shown to sustain an undifferentiated pluripotent stem cell state, while a loss of Oct4 expression results in the induction of differentiation in stem cells, producing a heterogeneous population of highly specialized daughter cells [
8]. Additionally, Oct4 has consistently been shown to be an integral factor necessary for the reprogramming of somatic cells into induced pluripotent stem cells (iPSCs). Although a cocktail of transcription factors are typically involved in this process (eg Oct4, Sox2, Klf4 and c-Myc), reprogramming efficiency is reduced if Oct4 is not present, thus indicating an absolute requirement for Oct4 in maintaining a stem cell-like state [
9]. Importantly however, Oct4 is highly expressed in many tumor types, suggesting that the reprogramming of somatic cells as well as tumor development and progression may share common cellular mechanisms [
10].
The Oct4 gene encodes for three isoforms, generated by alternative splicing of genes, known as Oct4A, Oct4B and Oct4B1 [
11,
12]. At the nucleotide level, both Oct4A and Oct4B share exons 2–5. However, exon 1 is missing in Oct4B and is replaced by exon 2a [
11,
12]. These differences appear to have significant biological implications on isoform function with Oct4A specifically expressed in the nucleus of ESCs, human somatic stem cells, tumor stem cells and in some adult stem cells [
11,
12]. Oct4B on the other hand, is localised to the cytoplasm and expressed at low levels in human somatic stem cells, tumor cells, adult tissues as well as pluripotent stem cells. For investigations in stem cell biology, it is therefore crucial that the Oct4A isoform is specifically targeted.
The interaction between CSCs and the neighbouring microenvironment forms a ‘niche’ which is critical for sustaining the stemness of cancer cells [
12]. Integrins are heterodimeric transmembrane receptors composed of a combination of different α and β subunits. They are essential in sensing the microenvironment and triggering cellular responses by bridging physical connections between the interior and exterior environments of cells [
13]. This allows the flow of bi-directional signals that control basic cellular functions such as adhesion, migration, proliferation, and survival as well as differentiation [
13]. In the context of CSCs, integrin receptors have been shown to promote a more malignant phenotype for tumor promotion and drug resistance [
14,
15]. These receptors are highly expressed in stem cell niches and contribute to diverse CSC functions [
14‐
16]. In this study using cancer cell lines, we demonstrate a direct link between the expression of α2, α5 and β1 integrin subunits with Oct4A expression in ovarian cancer and discuss the implications of these findings in relation to CSCs and progression of ovarian cancer in patients.
Methods
Patient samples
Tissue collection
Primary high grade serous epithelial ovarian tumors and normal ovarian tissues were obtained from patients requiring surgical resection at The Royal Women’s Hospital, Melbourne, Australia. The histopathological diagnosis, tumor grades and stages were determined by anatomical pathologists at the Royal Women’s Hospital as part of clinical diagnosis. Patients who were treated with chemotherapy prior to surgery were excluded from specimen collection. Tissues were paraffin embedded or snap frozen at the time of collection and stored at −80 °C until processed.
Cell lines
Four established human epithelial ovarian cancer cell lines SKOV3, OVCAR5, OVCA433, and HEY were used in this study. The growth conditions of these cell lines have been described previously [
17]. The human ovarian surface epithelial cell line (IOSE398) transfected with the SV-40 antigen was obtained from Dr Nelly Auersperg, University of British Columbia, Canada [
18]. The development of the vector control, Oct4A KD1 and Oct4A KD2 cell lines and their growth conditions have been described previously [
19]. Cells were routinely checked for mycoplasma infection.
Antibodies
Mouse monoclonal anti-human Oct4A and Sox2 were obtained from R&D Systems (Minneapolis, Minnesota, USA) and Cell Signalling Technology (Danvers, Massachusetts, USA) respectively. Rabbit polyclonal anti-human GAPDH was obtained from IMGENIX (CA, USA). Mouse anti-human integrin α2 (CD49b), anti-human α5 integrin (CD49e) and anti-human β1 (CD29) were obtained from Millipore (Billerica, Massachusetts, USA). Goat F(ab')2 anti-mouse IgG was purchased from Southern Biotech (Birmingham, AL, USA). Rabbit polyclonal anti-human cytokeratin 7 (CK-7), anti-human Glut-1, anti-human CD34 and anti-human CD31 were obtained from Ventana (Tucson, USA). The DAPI nucleic acid stain and Alexa Fluor® 488 goat anti-mouse IgG were obtained from Life Technologies (Carlsbad, CA, USA). Ventana antibodies used for the immunohistochemical staining of tumor xenografts were obtained from Roche (Basel, Switzerland) as described previously [
19‐
21].
Immunofluorescence analysis
For primary tissue analysis, paraffin embedded tissue samples were sectioned at 5 μm and deparaffinised by xylene and graded ethanol wash. Slides were blocked for 10 min in CAS-Block™ Histochemical Reagent (Invitrogen Corporation). For non-adherent sphere populations, 100–200 μL of sphere containing media was added per chamber well containing 200 μL appropriate fresh growth media and cultured on 8 well μ-Slides (ibidi, Martinsried, Germany) for 24 h to allow for adhesion to plastic before being fixed with 4 % paraformaldehyde. Monolayer cell lines were seeded at 5 × 103 cells per well onto the 8-well Nunc™ Lab-Tek™ Chamber Slide™ System (Thermo Scientific) and cultured as monolayer in complete RPMI-1640 growth media before being fixed with 4 % paraformaldehyde. Samples were probed overnight at 4 °C with either Oct4A (1:200), integrin β1 (1:200) or integrin α5 (1:200) primary antibodies, detected with Alexa Fluor® 488 Goat-Anti-Mouse antibody (1:200) and counterstained with 4’,6-diamidino-2-phenylindole (DAPI) (1:10,000). Fluorescence imaging was visualized and captured using an Olympus CellR fluorescence microscope and associated software (Olympus Corporation, Tokyo, Japan). Semi-quantitative analysis to assess fluorescence intensity of the antibody of interest was performed using the inbuilt CellR software. Results are expressed as a fold change of the protein of interest compared to DAPI for each analysis.
RNA extraction and real-time PCR
Quantitative real-time PCR was performed as described previously [
19]. Relative quantification of gene expression was normalized to 18S and calibrated to the appropriate control sample using the SYBR Green-based comparative CT method (2
-ΔΔCt). The primer set of Oct4A and β1 integrin are described in Table
1. The probe for 18S has been described previously [
22].
Table 1
Primer sequences of oligos used in quantitative Real-Time PCR
Oct4A | F | CTC CTG GAG GGC CAG GAAT C |
R | CCA CAT CGG CCTG TGT ATA T |
Integrin β1 | F | ATC CCA GAG GCT CCA AAG AT |
R | CTA AAT GGG CTG GTG CAG TT |
Western blotting
Cell lysates were extracted using the NU-PER nuclear and cytoplasmic extraction kit (Thermo Scientific, Waltham, MA, USA) as per manufacturer’s instructions. SDS-PAGE and Western blot was performed on the cell lysates as described previously [
19].
The sphere forming ability of cells and subsequent sphere adhesion ability was determined as described previously [
19]. Cellular aggregates with a diameter greater than 200 μm were classified as spheres.
Flow cytometric analysis
Flow cytometry was used to assess the expression of cell surface makers as described previously [
23]. Briefly, cells were grown as monolayer cultures, harvested and 10
6 cells incubated with primary antibody (1:100) for 30 mins at 4 °C. Cells were washed with 1X PBS, stained with secondary Goat F(ab’)2 anti-mouse IgG antibody conjugated with phycoerythin for 30 mins at 4 °C and resuspended in 200 μL 1XPBS prior to flow cytometry analysis. All data was analysed using Cell Quest software (Becton-Dickinson, Bedford, MA, USA) and expressed as background IgG staining subtracted from the IgG staining of the antibody of interest.
Adhesion assay
Cell adhesion assays were used to assess the ability of cells to adhere to extracellular matrix proteins. Briefly, 5 × 104 cells were seeded in complete growth media on culture plates pre-coated with 10 μg/mL collagen, Type 1 (Sigma-Aldrich) or 10 μg/mL fibronectin (Sigma-Aldrich) with sterile 1X PBS used as a diluent. Cells were incubated for 90 mins at 37 °C in a humidified atmosphere in the presence of 5 % CO2. The growth media was removed and cells were washed vigorously with 1X PBS using an orbital rocker on full speed twice for 5 mins to remove non-adhering cells. Cells were fixed with 4 % paraformaldehyde before being stained for 10 mins with 5 % Crystal Violet (Sigma-Aldrich) diluted in 0.2 % ethanol. Following crystal violet staining, cells were gently rinsed with 1X PBS and plates allowed to dry at room temperature before performing a dry reading at OD550nm with the SpectraMax190 Absorbance Microplate Reader and SoftMax® Pro Computer Software (Molecular Devices). Adhesion was calculated by subtracting the OD550nm reading of the negative control from the OD550nm reading of coated wells.
Gelatin zymography
This was performed as described previously [
24]. Briefly, complete growth medium from cells grown as sub-confluent monolayer cultures was discarded and replaced by serum free medium in a humidified atmosphere at 37 °C in the presence of 5 % CO
2. After 48 h, the serum free medium was collected and concentrated using 10 kDa Amicon Ultra-4 spin columns (Merck-Millipore, Billerica, MA, USA). Samples were resolved on 10 % (v/v) Tris–HCl acrylamide gels containing 0.1 % (w/v) gelatin, washed and stained with 0.2 % Coomassie blue. The gel was de-stained and areas void of blue stain indicative of areas of enzyme activity. Semi-quantitative densitometric analysis was performed on all gels to determine the extent of enzymatic digestion using Image Quant software (GE Healthcare) and expressed as the intensity of Pro-MMP9 or Pro-MMP2 bands of interest.
Animal studies
Animal experiments were performed on Balb/c nude mice as described previously [
19‐
21,
25,
26].
Immunohistochemistry of mouse tumors
Immunohistochemistry analysis of mouse tumors was performed as described previously [
19‐
21,
25,
26].
Statistical analysis
All results are presented as the mean ± standard error of the mean (SEM) of three independent experiments unless otherwise indicated. Statistical significance was measured compared to the vector control using one way-ANOVA and Dunnett’s Multiple Comparison test unless otherwise indicated. For primary tissue analysis, Student’s t-test was used to compare normal and high grade tissue samples. A probability level of <0.05 was adopted throughout to determine statistical significance.
Discussion
It is well recognized that the interaction between tumor cells and the ECM in the rapidly evolving tumor microenvironment is essential for the generation of regulatory signals that ultimately determine the fate of tumor cells and influence the evolution of a malignant phenotype [
35]. The integrin family of cell surface receptors is an important component of the ECM which sense micro environmental changes and trigger a range of cellular responses by forming a physical link between the inside and outside of tumor cells. Moreover, they allow for the bidirectional regulation of signals necessary to promote tumor progression [
14]. However, it is not yet clear if the effects that integrins have in modulating tumor cell behavior are also related to their associated role with CSCs in the tumor microenvironment. In this study we demonstrate a direct association between CSC specific Oct4A transcription factor expression and the β1 family of integrins primarily the α2 and α5 subunits.
In a previous study, we demonstrated significantly enhanced expression of the stem cell specific Oct4A isoform which increased according to histological grades of serous ovarian tumors compared to normal ovarian epithelium by immunohistochemical methods [
19]. In this study we further validate these results in high-grade serous ovarian tumor samples using immunofluorescence and Western blot techniques. We demonstrate that the expression of Oct4A is enhanced in high-grade serous samples compared to normal ovaries and localized to the nucleus of tumor cells. These results support the hypothesis that Oct4A has a specific role in ovarian cancer progression.
We have previously demonstrated that suppression of Oct4A is capable of inhibiting sphere forming abilities in the HEY ovarian cancer cell line [
19]. To expand on this observation, the expression of Oct4A in a range of ovarian cancer cell lines was seen to be directly correlated to the anchorage independent sphere forming abilities of each cell line. Sphere formation has been identified as a property of CSCs and the phenomenon has been reported in leukemia and in solid tumors of breast, colon and brain [
27]. However, the molecular mechanisms of how sphere-forming cells retain their stem-like characteristics remain unknown. Recent studies, have demonstrated that mouse ESCs are capable of maintaining long-term stemness within a 3D scaffold by manipulating integrin signaling [
36,
37]. These studies reported greatly increased expression of known stem cell markers Oct4 and Nanog was associated with simultaneous activation of Akt1 and Smad1/5/8 by α5β1, αvβ5, α6β1 and α9β1 integrins within the 3D scaffold. This maintained the self-renewal capacity of ESCs in the absence of leukemia inhibitory factor (LIF) signaling [
37]. These studies suggest that a ‘stem cell niche-specific integrin signaling mechanism’ within a 3D microenvironment can sustain the survival of ESCs without signals received from growth factors like LIF which are absolutely essential for the survival of ESCs.
Many of the same integrins which support ESC fate are also markers of CSCs in different cancers [
15]. These include α6 integrin enriching for CSCs in breast [
38], prostate [
39], squamous cell carcinoma [
40] and colorectal cancer [
41]. Integrin β3 is critical for stemness in breast [
42,
43], pancreas [
44] and lung cancer [
44], while β1 integrin is necessary for the stemness characteristics in glioblastomas [
45]. These studies are in agreement with our findings which demonstrate that integrins β1, α2 and α5 were down regulated by the knockdown of stem cell-specific Oct4A expression in HEY cell line. These observations were made in cells maintained as monolayer as well as sphere cultures and were consistent with the loss of adhesion of these cells on collagen and fibronectin. However, it should be noted, that integrins can also influence CSC niches independent of their capacity to interact with ECM [
15].
We have previously shown that enhanced expression of α2β1 integrin in ovarian cancer spheres facilitates sphere disaggregation, pro-MMP-2/9 expression and MMP-2/9 activation [
31]. In the current study, this is consistent with a decrease in α2 and β1 integrin expressions in Oct4A knockdown HEY cells along with reduced pro-MMP-2 secretion. This data correlates to the loss of sphere disaggregating and migratory ability previously shown in Oct4 KD cells [
19]. These observations in our Oct4A knockdown model are consistent with observations in pancreatic cell line models, where loss of Oct4 also resulted in reduced MMP2 expression and subsequent reduced tumor cell invasive ability [
46]. Interestingly, MMP2 is known to actively degrade fibronectin and collagen I [
32,
33,
47], thus potentially linking these results to the decreased ability of Oct4A KD cells to adhere to fibronectin and collagen I. In the pancreatic cancer cell line model, diminution of Oct4 was associated with a decrease in MMP9 expression [
48]. In our study, no such relationship in Oct4A KD cells was observed. Interestingly however, the Oct4B isoform has been identified to regulate both MMP2 and MMP9 expression in cervical cancer, suggesting the lack of MMP9 suppression may be due to the fact that Oct4B expression was not specifically suppressed in the current study [
49].
We also demonstrate that the suppression of Oct4A in HEY cells resulted in a significant reduction in the tumor growth, weight and size in Balb/c nude mouse model. In addition, tumor xenografts derived from Oct4A KD cells displayed relatively lower abundance of markers associated with ovarian cancer (CK7), cancer metabolism (Glut-1) and angiogenesis (CD31 and CD34). This indicates that the suppression of Oct4A not only reduced the tumor initiating ability of cells in vivo but also resulted in a reduction in angiogenic potential which consequently may have resulted in slowed or abrogated tumor growth. These results are consistent with the positive correlation between the expression of Oct4 and vasculogenic mimicry formation and poor prognosis in breast cancer patients [
50]. In addition, Oct4 has been shown to promote glioblastoma progression through vascular endothelial growth factor production [
51]. Importantly, integrins are well documented to play significant roles in mediating tumor vascularity and angiogenesis [
52]. Specifically, integrin α5β1 has been identified to play a role in tumor angiogenesis in in vivo mouse models [
53,
54]. Collectively, this tie in with the reduced expression of β1 and α5 subunits demonstrated in Oct4A KD cells and overall reduced tumorigenic and angiogenic profiles in this study.
Abbreviations
ANOVA, analysis of variance; CK7, cytokeratin 7; CSCs, cancer stem cells; DAPI, 4’,6-diamidino-2-phenylindole; ECM, extracellular matrix; ESC, embryonic stem cells; IP, intraperitoneal; iPSCs, induced pluripotent stem cells; LIF, leukaemia inhibitory factor; MMP, matrix metalloproteinase; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis
Acknowledgements
The authors wish to acknowledge the help of Ms Bronwyn Christensen, Anatomical Pathology, The Royal Children’s Hospital, Parkville, Australia for assisting with the mouse xenograft immunohistochemistry analysis.