Abstract
Background/Aim: Epithelial-mesenchymal transition (EMT) and cancer stem cells (CSCs) are presumed to be key conditions for malignancy. Data concerning their role in endometrial cancer (EC) are scarce. We aimed to investigate the possible link between EMT and CSCs markers in EC samples. Materials and Methods: The study encompassed 156 primary tumour samples. Using RT-qPCR, we analyzed the expression of EMT-related genes, SNAIL and SLUG, and the CSCs marker CD133. Results: SNAIL and SLUG correlated with each other (R=0.33; p=0.00003). All the studied genes were expressed in both normal and malignant endometrial tissue. Decreased SNAIL expression was found to correlate with post-menopausal status (p=0.002). Decreased SLUG expression was associated with shorter overall survival (p=0.01). Conclusion: SLUG expression could serve as a prognostic factor in EC. No correlation between the expression of EMT and CSCs markers was found, suggesting there to be no association between the EMT and CSC phenotype in endometrial cancer.
- Epithelial-mesenchymal transition
- cancer stem cells
- molecular markers
- quantitative PCR
- endometrial cancer
Endometrial cancer (EC) is the most frequent malignancy of the female genital tract in the Western world, with an estimated incidence of 10 to 20 per 100,000 women (1). Despite the high prevalence, understanding of its molecular background with regard to pathogenesis and disease progression remains unclear. Data concerning the role of cancer stem cells (CSCs) and epithelial-to-mesenchymal transition (EMT) in EC are especially scarce.
CSCs are perceived as germinative tumor cells that possess an ability to initiate and sustain tumour growth (2, 3). The presence of CSCs has been reported in many cancers, including melanoma, brain tumours, breast cancer, lung cancer, and leukaemia (4-8). Recent studies proclaim CD133 (membrane glycoprotein also known as prominin, official gene symbol PROM1) as a reliable marker for CSCs identification (9). During epithelial-to-mesenchymal transition epithelial cells acquire mobility as well as mesenchymal-like expression profile. This phenomenon has been discovered as a crucial mechanism which allows for invasion and metastasis of cancer (10-12). To undergo EMT, cancer cells down-regulate E-cadherin expression. This is partially achieved through the activation of transcription factors: SNAIL (SNAI1), and SLUG (SNAI2) (13).
Accumulating evidence suggests that EMT induces stem cell activity (14). This phenomenon was first observed by Mani et al. (15). The EMT-CSCs dependency has been confirmed in many different types of cancer (16-20). However, the data about this relationship in EC are scarce. The sole existing experimental study confirmed the relationship between EMT and CSCs in endometrial cancer cell lines, not tumour samples (21). As Tanaka et al. pointed to the important role of EMT in endometrial cancer specimens (22), we aimed to investigate the possible link between EMT and CSCs markers in EC tumor samples. We studied the relationship between the gene expression of EMT-related SNAIL and SLUG, and CSCs marker – CD133 in EC specimens. To the best of our knowledge, we are the first group to present data on gene expression level of SNAIL and SLUG in relation to CD133. Expression levels are also compared with normal endometrium, clinical data, and patients' outcome.
Materials and Methods
Patients and tissues. In the present study 156 fresh-frozen primary tumour samples retrospectively collected from a cohort of endometrial cancer patients who were operated on in the Department of Gynaecology, Gynaecological Oncology and Gynaecological Endocrinology (Medical University of Gdansk) between 2005 and 2011, were analyzed. Each patient was primarily treated by surgery, with the possible option of radiotherapy and/or chemotherapy administration. The inclusion criteria were operable endometrial cancer (stage IVB patients underwent cytoreductive surgery) confirmed by histological examination and a signed consent form. The study was accepted by the Ethics Committee of the Medical University of Gdansk.
The samples were collected from the core of the tumour by surgical excision prior to any systemic treatment and were immediately frozen and stored in −80°C. They were kept on ice during transport. The tumour samples included all stages of endometrial carcinoma, from non-invasive stage IA to metastatic IVB cancers, as distinguished by FIGO in 2009 (International Federation of Gynecology and Obstetrics) (23). The patients' characteristics are summarised in Table I. The median age was 63 (range: 30-87) years. Patients with a body-mass index higher than 30 were classified as obese (24). The progesterone receptor (PR), oestrogen receptor-α (ERα), and Human epidermal growth factor receptor-2 (HER2) status was determined immunohistochemically (IHC) using Allred score (positivity cut-off: ≥3) for PgR and ERα and the HercepTest, with the score of 3 classified as HER2-positive. A survival analysis was performed for 94 patients. After a median follow-up of 54.5 months (range: 0-96), 16 patients (17.0%) had died. The last follow-up data were collected in May 2013. The study was performed in accordance with the REMARK criteria (25).
Controls. Normal endometrium from patients treated for ailments other than EC was collected as control samples. Control samples were accessed histopathologically; no hyperplasia was observed. For standard curve generation synthetic templates, 60-mer single-stranded oligonucleotides composed of the first (5’) and last (3’) 30 nucleotides of the amplicon sequence, were used.
RNA isolation. Prior to nucleic acid isolation, tissue specimens (25 mg per sample) were homogenized (1 min, 6,000 rpm) using a MagNALyser (F. Hoffmann-La Roche Ltd, Basel, Switzerland). RNA was isolated with AllPrep DNA/RNA Mini Kit (Qiagen, Hamburg GmbH, Hamburg, Germany), according to the manufacturer's instructions. After the isolation, the RNA concentration and integrity were determined using the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). Median RNA concentration was 628.8 ng/μl (range: 20.5-1968.6), median RIN (RNA Integrity Number) was: 7.9 (range: 2.2-9.6). The RNA samples were stored in −80°C. RNA was subsequently reverse-transcribed to cDNA with the Transcriptor First Strand cDNA Synthesis Kit (F. Hoffmann-La Roche Ltd, Basel, Switzerland) according to the manufacturer's instructions, using random hexamer primers. The total amount of RNA per reaction was 1000 ng.
geNorm pilot study. A geNorm pilot experiment was performed to select proper, stably-expressed reference genes for data normalization, using a set of 10 candidate reference genes on 10 representative endometrial samples (both tumors and controls). Data analysis was performed using Biogazelle's qbase+ software (Biogazelle NV, Zwijnaarde, Belgium) which is built upon a state-of-the-art and proven quantification model (26). To determine the best reference genes, the geNorm module in qbase+ was used. geNorm is the most popular algorithm to find stable reference genes (27). Three reference genes were chosen.
Genes and primers. SYBR Green I-based PrimePCR assays for the three genes of interest (SNAIL, SLUG, CD133), and reference targets were designed using validated pipeline primerXL for design and in silico verification of high quality assays. All assays are wet-lab validated and evaluated with the gold standard method of standard curves based on a 6-point 10-fold serial dilution series using synthetic templates to create a broad detection range. The serial dilution was applied to determine assay efficiency, sensitivity and linearity. The performance and functionality of the assays was further evaluated on gDNA, cDNA and non-template controls. Subsequently, next-generation sequencing of the amplicons was performed to verify the specificity. Amplicon sequences are presented in Table II. Used PrimePCR assays met the highest quality and specificity available. All the analyses were performed by Biogazelle (Biogazelle NV, Zwijnaarde, Belgium) and were fully MIQE (Minimum Information for publication of Quantitative real-time PCR Experiments) compliant (28). Quality control on the post-qPCR data was performed using Biogazelles's qbase+ software (Biogazelle NV, Zwijnaarde, Belgium).
qPCR-based gene expression analysis. For gene expression profiling of the genes of interest and the reference genes, validated standard operating procedures for verification of the quality of the cDNA, qPCR amplification, and data analysis were performed. All measurements were performed using CFX384 instrument (Bio-Rad Laboratories, Inc., Hercules, CA, USA) in 384-well plates in a reaction volume of 5 μl. Reaction mixture included 10 ng cDNA, 2× SsoAdvanced™ SYBR® Green Supermix (Bio-Rad Laboratories, Inc., Hercules, CA, USA) and forward and reverse primers at a final concentration of 250 nM. Negative control (no reverse transcriptase) was always included. Each PCR reaction was performed in duplicate. Sample maximization method was implemented. For 25% of the samples (randomly selected), cDNA quality was assessed by means of three different gene expression qPCR assays (high expressed, medium expressed and low expressed gene). Gene expression analysis was performed using Biogazelle's qbase+ software (Biogazelle NV, Zwijnaarde, Belgium).
Statistical analysis. STATISTICA software (Statsoft, Inc., Tulsa, OK, USA, version 10) was used for all calculations. The tests that were used and their applications were as follows: testing normality of the data set — Shapiro-Wilk test; comparison of the expression in the normal endometrium versus tumour samples – box-and-whisker diagram with Mann-Whitney test used for p-value determination; correlations between continuous expression quantities – Spearman's correlation; correlations between continuous relative expression and clinicopathological data of the patients – Mann-Whitney test. The Kaplan-Meier estimator was employed for survival analysis, and the results were verified with the F Cox test. The end-point for the study was overall survival (OS). OS was defined as the time from sample collection to death or censoring. Censoring was defined as loss of follow-up or alive at the end of follow-up. Statistical significance was assumed when p≤0.05. Cox proportional hazards regression analysis was used to identify the independent predictors of OS. Univariate predictors significant with a value of p≤0.10 were entered into a step-wise multivariate model to identify those with independent prognostic information. For Kaplan-Meier curve generation and univariate analysis, tumors were dichotomized based on a median expression value used as a cut-off.
Results
General. Out of 160 RNA samples (including 4 control samples), 160 (100%) had amplifiable cDNA.
Assay validation. Serial dilution was applied to determine assay efficiency, sensitivity, and linearity (Table III). The three most stably-expressed reference genes (ERCC6, HMBS, UBE4A) were retained from geNorm analysis for further gene expression normalization. High reference target stability was observed, with average geNorm M≤0.5. No indications for failed cDNA synthesis were observed. The results confirmed that the experiment was performed according to the internal quality standards: replicate variability fell within the set limit of 0.5 cycles for 87.6% of sample-target combinations. Through analyzing the samples, medium-to-low reference target stability was observed for the selected reference genes (average geNorm M value between 0.5 and 1), which is typically seen in a heterogeneous set of samples.
mRNA expression within clinical and control samples. The studied genes were expressed in both normal and malignant endometrial tissues (Figure 1a, 1b and 1c). A trend was observed towards higher mRNA CD133 expression in EC samples compared to normal endometrium samples (p=0.06). Expression of SNAIL and SLUG was reduced in EC samples (p=0.01 and p=0.002, respectively).
Gene expression correlation and mRNA expression in clinical samples. Expression levels of CD133, SNAIL and SLUG were compared with each other (Table IV). Expression levels of CD133, SNAIL and SLUG were compared with clinical and pathological data (Table V).
Survival analysis. No prognostic impact was observed for SNAIL and CD133 (Figure 2a and 2b). Decreased SLUG expression correlated with shorter OS (p=0.01; Figure 2c). In univariate analysis, factors predicting for shortened OS included type II histology, higher stage and grade, myometrial invasion, presence of metastases, positive PgR status and SLUG overexpression. Step-wise multivariate analysis showed higher FIGO stage, and decreased SLUG gene expression to be associated with decreased OS (Table VI).
Discussion
EMT and CSCs are presumed to be key conditions for epithelial malignancy (2, 11). Thus, the mRNA expression of the genes responsible for the aforementioned phenomena was assumed to characterize between the endometrium lacking EMT and CSCs and endometrial cancer. We found CD133, SNAIL and SLUG mRNA expression to vary greatly even within the control samples. Similar observations with regard to CD133 mRNA levels were reported in human endometrial cancer cell lines, Hec1A, Hec1B, AN3CA and Ishikawa (29). Many EMT markers were also reported to be highly expressed in normal tissues (30). In our study, the levels of CD133 expression in control and EC samples did not differ significantly and the expression of SLUG and SNAIL was even more reduced in EC samples compared to normal endometrium. Observations concerning CD133 might be partially explained as the expression of this protein was reported in the endometrial glandular cells (31). An analogous study revealed CD133 to be commonly expressed not only in cholangiocarcinoma but also in many kinds of cells in normal liver (32). It was reported that CD133 exhibits a variable expression in endometrial primary tumours, ranging from 1.3% to 62.6% (33). Low SNAIL expression in EC samples when compared to control endometrium stays in contrast to the results of Montserrat et al. where SNAIL mRNA expression was reported to be higher in EC samples (34). This analysis, however, included a relatively small set of EC samples (42) and was based on only one control sample which might lead to confounding results, given the high variability observed in controls. Our observations are shared by the results of qPCR analysis in breast cancer cell lines that also showed decreased SNAIL mRNA in relation to normal mammary cells (35). Finally, reduced SLUG expression was also detected in invasive breast cancer when compared with normal breast epithelium (30).
Even though SNAIL and SLUG mRNA expression correlated with each other, we failed to find a link between these genes and CD133. Studies performed by Na et al. did indeed report such an association in hepatocellular carcinoma (36), however, they were performed on cell lines, not tumours. Several similar results were reported (36-38), but they also concerned individual cells, not actual patients' samples. Storci et al. managed to show a correlation between CD133 protein expression and high SLUG mRNA levels in breast cancer tumours, yet the study included only 21 specimens (39).
The strongest correlation was found between post-menopausal status and decreasing SNAIL expression (40). Results obtained by Scherbakov et al. (41) revealed the inverse relationship between ERα and SNAIL. The authors speculated there might be a negative feedback between ERα and SNAIL which would explain the decrease of SNAIL expression in post-menopausal patients. Similarly to our work, studies of Martin et al. (42) on SNAIL mRNA expression in breast cancer were often inconclusive but they also rather pointed to SNAIL being reduced in more advanced disease.
In our study, SLUG expression correlated conversely with advanced age and HER2-positivity. Similar results concerning HER2 status were reported in breast cancer (39). These authors also failed to find any association of SLUG expression with other clinicopathological data.
SLUG mRNA expression was significantly reduced in patients with poor prognosis. Similar results in EC, but concerning SNAIL mRNA expression, have been reported (42). Numerous studies confirmed CD133 overexpression to be the indicator of poor outcome in cancer (43-48) but so far there is no conclusive evidence showing such prognostic value in EC. Also, search for prognostic significance of EMT in head and neck squamous cell carcinoma yielded no definitive conclusions (14). On the other hand, high SNAIL expression was associated with poor patient prognosis in prostate cancer (49), bladder cancer (50), or hilarcholangiocarcinoma (51). In EC, EMT status, defined as reduced E-cadherin and increased SNAIL expression, was also reported to have prognostic value but there are no survival analyses focusing on SNAIL and SLUG expression independently (22). Our data showing SLUG expression as the indicator of good prognosis are in compliance with the work on EMT markers in diffuse large B-cell lymphoma (52). Also, similar results, regarding TWIST expression, were obtained by Montserrat et al. (53). The authors suggested that this down-regulation might in fact facilitate tumour invasiveness, a possibility worthy of further exploration.
We were unable to confirm that CD133, SNAIL and SLUG expression are unambiguously implicated in cancer tumorigenicity. Nevertheless, studies supporting those facts were often based on cell lines or animal models (29, 54, 55), not tumour samples collected from patients.
One of the limitations of the study was the short follow-up times and missing survival data for 62 patients. However, the possible explanation for numerous inconclusive results may lie in tumor heterogeneity. The samples used for RNA isolation might not reflect the heterogeneity across a tumor, as they were excised manually from a single tumor region and only small tissue fragments were used. As EMT actively occurs at multiple sites and CSCs are distributed in a heterogeneous manner within the tumor, it is possible that the excised fragments were not representative for the studied phenomena. CD133-positive cells comprise only 5.7% to 27.4% of primary tumour in EC (29). The phenotype of EMT is also difficult to capture as these cells constitute minority in primary tumours (56). Furthermore, studying primary tumours-alone might not be sufficient. It has recently been reported that gene expression of SNAIL and SLUG can be easier captured in lymph node metastasis compared to matching primary tumors and provides more prognostic information (57). Metastatic deposits contain selected, more aggressive clones of tumor cells that more closely determine clinical course of disease. Therefore, analysis of prognostic markers in such a subpopulation of cancer cells might provide with more conclusive results. Unfortunately, this kind of material was unavailable for EC samples. Perhaps a micro-dissection technique of formalin-fixed paraffin-embedded EC samples might help to further overcome the problem tumor heterogeneity.
It is possible that the RT-qPCR approach is simply not able to detect EMT-related gene expression within primary tumor bulk, where tumor cells with EMT characteristics might be diluted within the rest of tumor cells. Nevertheless, tremendous effort has been put to ensure the highest possible quality of the obtained results. Quality control tests showed excellent reaction performance, specificity of the designed primers, high cDNA/RNA integrity and quality, and thus no sample loss. The developed protocol allows for highly coherent transcriptomic measurements.
The crucial role of EMT and CSCs in driving malignancy is widely described in various cancers (32, 36, 42, 48, 58, 59). Also the link between those two phenomena leading to development of CSCs due to EMT activation has been extensively investigated (14, 15, 20, 30, 60, 61). However, data concerning EMT in relation to CSC in EC are scarce. To the best of our knowledge, we are the first to present data on gene expression levels of the EMT-related markers SNAIL and SLUG in relation to the CSC marker CD133. Unfortunately, mRNA expression levels of SNAIL, SLUG and CD133 do not provide with valuable information on EMT and CSCs in EC. We found no link between the expression of these markers, which might suggest there is no link between EMT and CSC phenotype in EC cancer or additional markers should be examined. Further studies, including larger groups of patients and probably more markers are needed to elucidate whether EMT and CSCs markers carry clinically-significant information in EC.
Acknowledgements
We would like to acknowledge Agata Łuczak for her excellent technical support. This work was supported by a grant from the National Science Centre (5715/B/P01/2010/38) and a grant from the Foundation for Polish Science Parent-Bridge Programme co-financed by the European Union within the European Regional Development Fund (DPS-424-5053/11).This research was co-financed by the European Commission in the frame of the European Social Fund, by the European Social Fund, the State Budget, and the Pomorskie Voivodeship Budget according to the Operational Programme Human Capital, Priority VIII, Action 8.2, Under-action 8.2.2: ‘Regional Innovative Strategy’ within the system project of the Pomorskie Voivodeship “InnoDoktorant – Scholarships for PhD students, Vth edition”. The publication was also financed by the European Social Fund as a part of the project “Educators for the elite - integrated training program for PhD students, post-docs and professors as academic teachers at University of Gdansk” within the framework of the Human Capital Operational Programme, Action 4.1.1, Improving the quality of educational offer of tertiary education institutions. The sponsors had no involvement in the study design, collection, analysis, and interpretation of the data, writing the manuscript, or the decision to submit the manuscript for publication.
Footnotes
-
Conflicts of Interest
The Authors declare that they have no conflicts of interest.
- Received September 24, 2013.
- Revision received October 25, 2013.
- Accepted October 29, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved