Background
Cancer of cervix is the fourth most common cancer among women and the third most common cause of cancer related mortality worldwide [
1]. Developing countries contribute to nearly 80% of cervical cancer cases with an incidence to mortality ratio of nearly 50% resulting in a large number of cancer related deaths [
1] Advanced stage at presentation and resistance to standard chemoradiation are important factors leading to a high mortality and in turn may be attributed to residing cancer stem cells [CSC] which, by definition are a self-renewing population of cells within a tumor capable of initiation and maintenance of a tumor. A consensus of five defining criteria are required to prove the existence of CSCs [
2]. The expression of distinctive cell surface markers is one criterion which permits their consistent isolation, characterization and enables the exploration of alternative therapeutic strategies to target CSC [
3].
CSC have been demonstrated in several solid cancers. In cancer cervix, cancer stem-like cells were first identified to exist by Feng et al. [
4] and subsequently by others [
5‐
8]. Identification of their surface phenotype is the initial step towards their isolation and enrichment for subsequent biological studies. Unfortunately, there has been no consistency in the previous reports and a consensus marker of CSC in cervical cancer is still elusive. Thus, breast cancer-resistance protein (Brcp1) [
5], aldehyde dehydrogenase (ALDH
+) [
5,
6], Annexin II [
6], CD44+/CK17+ [
4] and CD49f [
8] have all been proposed as makers for CSC based on different approaches. Moreover, only a single study by Feng et al. [
4] is based on primary cultures from patient samples whereas the rest are based on highly passaged established cell lines such as HeLa, SiHa, CaSki and C33A [
5‐
8]. Hence our initial step was to generate short-term primary cultures from tissue biopsies of invasive cervical cancer. In this process, we were able to establish 4 low-passage cell lines, all from Indian ethnicity. Further studies on the isolation and characterization of CSC were performed on these cell lines from invasive cervical carcinoma. We applied two assays that permit enrichment of CSCs, namely, the tumorsphere assay and second, a functional assay based on aldehyde dehydrogenase [ALDH] activity. Using flow cytometry and immunofluorescence studies, we established that CD133 is a surface phenotypic marker of cancer stem cells in cervical carcinoma of both the squamous cell carcinoma and adenocarcinoma subtypes. The clinical significance of the CSCs was evaluated in biopsies from 29 cases of invasive cervical cancer.
Methods
Patients and samples
The study was approved by the Institute’s Ethics Committee (vide letter No.IC-CCRT/07/DTM-3226 dated 16.11.2009). All patients were enrolled after obtaining a voluntary written informed consent and the study was conducted as per the Helsinki declaration (2000). Tissue samples obtained from untreated patients of invasive cervical cancer comprising of 3 radical hysterectomies in FIGO stage IIA and 30 cervical biopsies from FIGO stage IIb/III were subjected to primary culture. The sample was collected directly in Dulbecco’s Modified Eagle Medium/ Nutrient Mixture F-12 Ham (DMEM/ F12, 1:1 mixture) (Invitrogen). The rest of the specimen was subjected to formalin fixation and paraffin-embedded sections were evaluated light microscopically for histopathological diagnosis.
Primary culture
The specimen collected was immediately mechanically disaggregated with a sterile surgical blade, washed 3 times with sterile phosphate buffered saline (PBS), digested by collagenase II by incubation for 1 h at 37 °C (Sigma-Aldrich Corp., St Louis, MO, USA), filtered through a 70 μm mesh to get rid of stromal fragments and obtain a cell suspension. The tumour cells were placed in 6 well plates (Corning) containing DMEM/ F12 supplemented with 10% fetal bovine serum (FBS, Sigma), penicillin (100 U/mL), and streptomycin (100 μg/mL) and placed in a humidified incubator at 37 °C containing 5% carbon dioxide (CO2) and 95% air. 17 cases showed contamination and in 16 cases successful primary cultures could be established. In 7 of these 16 cases, long-term primary cultures got established; and 4 were pursued further. The morphology of the adherent cells was observed under an indirect microscope by phase-contrast and morphology documented.
Electron microscopy
Approximately 5 × 105 cells from RSBS-14 and RSBS-43 cell lines were centrifuged gently and fixed in 3% glutaraldehyde followed by fixation in 1% Osmium tetroxide, processed and embedded in Taab-812 embedding medium. 60 nm thick sections on Nickel grids were stained with uranyl acetate and lead citrate and examined using JEOL Transmission Electron Microscope, JEM-1400Plus (JEOL, Tokyo, Japan) equipped with XR81M-B Camera (Advanced Microscopy Techniques Corp, Woburn, MA, United States).
Karyotyping
Karyotyping was performed on metaphase spreads of RSBS-9, RSBS-14, RSBS-23 and RSBS-43 cells at passage 32 by standard G-banding.
Tumorsphere assay
Cultured cells were also placed under anchorage-independent/stem-cell conditions in Ultra Low Attachment plates (Corning, USA) in serum-free DMEM/F12 supplemented with 5 μg/mL insulin (Sigma), 20 ng/mL human recombinant epidermal growth factor (EGF; Invitrogen), 10 ng/mL basic fibroblast growth factor (bFGF; Invitrogen), and 0.4% bovine serum albumin (BSA; Sigma). Floating aggregates or tumorspheres were observed microscopically and their numbers per well documented. Efficiency of tumorsphere formation ability was calculated based on manual counting of number of tumorspheres after plating equal numbers of sorted cells wherever required.
ALDEFLUOR assay
Adherent cells derived from samples were used for ALDEFLUOR assay using ALDEFLUOR kit (Stem cell Technologies) as per manufacturer’s instructions detailed in Additional file
1. Based on their ALDH activity, cells were sorted as ALDH
high and ALDH
low cells (Additional file
1).
Detection of putative markers of cancer stem cells by flow cytometry and FACS
At least 1 × 10
6 cells derived from tumorspheres and corresponding adherent cells were screened for surface markers by flow Cytometry using fluorescent (FITC, PE, APC, Per-CP) labelled antibodies against CD117, CD90, CD44, CD49f and CD71. Isolation of the CD133
+ cells was performed using a FACS Aria flow cytometer (BD Biosciences, USA) and analyzed further (Additional file
1).
Direct immunofluorescence
Direct immunofluorescence was carried out on adherent cells and tumorspheres with FITC/PE labelled antibodies against CD44, CD49f and CD133 and on corresponding frozen tissue sample of cervical cancer (Additional file
1). Nuclei were counter stained with DAPI. Florescence microscopy was carried out with EVOS FL Auto cell imaging system (Invitrogen, Thermo Fisher Scientific, USA).
Screening of patient samples of invasive cervical carcinoma for cancer stem cell percentage and patients after radiotherapy
Flow cytometric immunophenotyping (FCI) was performed on the biopsies of patients with untreated invasive cervical cancer (
n = 22) and in 6 patients who had been treated with chemoradiation therapy 46 Gy/23 fraction external beam radiation administered concurrently with 40 mg/m
2 cisplatin weekly dose followed by intracavitary brachytherapy and who had subsequently relapsed. The levels of CD133, CD49f and CD44 was evaluated in each case after gating for CD45 negative cells (Additional file
1).
Human papillomavirus (HPV) screening
All cell lines developed were evaluated for HPV DNA type by PCR with type specific primers against HPV-16, − 18, − 31 and − 45 (Additional file
1: Table S1).
Assay of growth characteristics
MTT assay
Cells from all the 4 cell lines generated and which were at the same passage 35 (P35) were used to plot the growth curve by using the MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay detailed in Additional file
1. The growth curves were plotted and the population doubling time of these 4 cell lines were calculated during the exponential growth phase of the cells.
Cell cycle analysis
Cell cycle analysis was performed after labelling the cellular deoxyribonucleic acid (DNA) with propidium iodide (PI) (Invitrogen, Carlsbad, CA, USA). The cells were then subjected to cell cycle analysis by FACS (BD FACS Aria II). Each experiment was analysed in duplicate and four independent experiments were performed (see Additional file
1).
Variable nucleotide tandem repeat [VNTR] assay
Highly polymorphic VNTR loci D1S80 and IGH (VNTR at the immunoglobin heavy chain enhancer HSS 1.2) were used to prove and validate that the cervical cancer cell lines were derived from primary cervical cancer biopsies. Amplification of VNTR loci D1S80 (range of amplification product 387 to 723 bps) was achieved using the primers (Additional file
1).
Chemosensitivity assay
Cell lines RSBS-9, RSBS-14, RSBS-23 and RSBS-43 were treated with different doses of cisplatin (10, 20, 30, and 40 μM) for variable duration (24, 48, 72 h). Cytotoxicity was determined by MTT assay All experiments were performed twice in triplicates to determine the IC50 values.
Quantitative real time PCR (qRT-PCR) for stemness/EMT/markers
Total RNA from the cells was isolated using the RNA isolation kit (Agilent technologies, USA) per the manufacturer’s instructions. 1 μg RNA was reverse-transcribed into cDNA using specific primers and per the manufacturer’s instructions (Thermo Fisher Scientific, USA). 1 μL of cDNA was used as template for qRT-PCR for
OCT-4, SNAIL, SLUG, VIMENTIN, ABCG2, N-CDHERIN, NANOG, TIWST and
E-CADHERIN using gene specific primers (Additional file
1: Table S2), and normalized to β-ACTIN housekeeping gene transcript.
Discussion
In this study, we established four cervical carcinoma derived cell lines from Indian subjects designated RSBS-9, RSBS -14, RSBS -23 representing squamous cell carcinoma and RSBS -43 representing adenocarcinoma, the two major types of invasive carcinoma of the uterine cervix. Each of these cell lines has unique modal chromosome numbers and have been proved to be derived from the parental tissue biopsy specimen as evidenced by identical band patterns in the VNTR assay. The overall success rate for establishing cell lines was 21.2% which compares favourably with a 26% success rate in a previous report on cell lines derived from invasive cervical cancer; however, the cells were grown on an irradiated layer of 3 T3 fibroblasts to achieve successful growth in the initial passages which was not required in the present study [
9]. On the other hand, in gastric carcinoma, a very low success rate of just 3% from primary gastric tumors was reported although it was 25% from metastatic sites [
10]. Differences in the histological characteristics of a tumour, especially the tumor microenvironment and associated features such as desmoplasia and inflammation could explain the differences in the success rate between the two sites. The role of prior tumor xenografts in athymic nude mouse models in improving success rates has been highlighted by a report on the development of colon cancer cell lines [
11].
Out of the 6 commonly used cell lines in cervical cancer research, CaSki and HT-3 cell line are derived from metastatic sites of cancer the small intestine and lymph node respectively, whereas SiHa, HeLa, C-4 II and C-33A are derived from primary sites. The CaSki cell line was developed following surgery and irradiation. All the 4 cell lines we have developed are from primary cervical cancer who had not received any form of prior therapy.
All cell lines reported here grew as adherent monolayers in standard conditions and as tumorspheres under ‘stemness’ conditions successfully and hence these cell lines could serve as useful models for study of cancer stem cell biology as well. The morphology of the cell lines was in keeping with the parent tissue histology as evidenced in the cell blocks prepared from the cell lines. This was further corroborated by ultrastructure studies which revealed hemidesmosomes and keratin filaments in the squamous cell carcinoma derived cell lines in comparison to a prominent nucleolus, abundant rough endoplasmic reticulum and small surface microvilli in the adenocarcinoma derived cell line.
All the four cell lines were HPV16 positive which is the most frequent type of HPV associated with cervical carcinoma, both worldwide as well as in India [
12‐
15]. In the meta-analysis of by Bhatla et al. [
13], in invasive cervical carcinoma, HPV-16 was the most common type seen in 68.4% followed by HPV-18 in 14.7% cases; even in adenocarcinoma histotype, HPV-16 is more frequent (34.5%). Multiple infections from more than one type of HPV was seen in only 14% cases. Several studies have demonstrated the requirement of HPV oncogenic expression for proliferation, growth or survival of primary cervical cancer cells in culture [
16‐
18]. Among the commonly used cervical cancer derived cell lines, CaSki and SiHa are positive for HPV-16, HeLa, C-41 are HPV-18 positive whereas C-33A and HT-3 test negative for HPV sequences.
The cell lines developed by us showed variable chemosensitivity to cisplatin with RSBS-23 and RSBS-43 relatively more chemosensitive and RSBS-9 and RSBS-14 more chemoresistant. In a previously published report, HeLa cell line is reported to be more chemosensitive as compared to SiHa which is cisplatin resistant [
19]. Thus the variable chemosensitivity of these cell lines reflects better tumor biology and variation in the chemotherapeutic response and hence are useful in screening of therapeutic agents which are of potential value in invasive cervical cancer.
The next part of the study was aimed at isolation and characterization of the cancer stem cells using these low-passage cell lines. We employed two well-known assays which enrich for cancer stem cells. These included the ALDEFLUOR functional assay which enriches for ALDH
high cells [
20], and the tumorsphere assay which too enriches for CSC [
21]. The tumorsphere assay is considered as a surrogate assay for cancer stem cells and correlates well with in vivo tumorigenicity assay [
4]. Feng et al. reported CD44 as a putative CSC marker based on its expression in sphere-derived cells, however, comparison of its level in non-tumorsphere cells was not performed. In the current study, CD44 was not upregulated in spheres; rather, the levels were lower than in adherent cells upon flow cytometry, further corroborated by direct immunofluorescence studies on adherent versus tumorspheres. Moreover, tissue sections of invasive cervical cancer expressed high levels of CD44 (by flow Cytometry and direct IF) making it highly unlikely to be a cancer stem cell marker. The two promising markers were CD49f and CD133. Lopez et al. used HeLa, SiHa, Ca Ski and C-41 cell lines to derive sphere-forming cells and showed CD49f expression at high levels (> 90% cells) and CD133 (1.4–61%) at lower levels and so concluded that CD49f is a putative marker [
8]. We observed a 2–3-fold upregulation of CD49f in sphere-derived cells as compared to adherent cells; however, by direct immunofluorescence and flow cytometry, CD49f levels were high in the adherent cells and in tissue specimen and hence less likely to be a CSC marker. On the other hand, CD133 showed a 10–13-fold difference between adherent cells versus sphere-derived cells with low levels in the adherent cells upon flow cytometry and further confirmed by direct immunofluorescence studies. The CD133
+ sorted cells also showed a greater ability to form spheres as compared to the CD133
− bulk cells providing further evidence in support of it being a marker of CSC. A limitation of this study is the lack of corroborative in vivo tumorigenicity studies in immunocompromised mice; however, the tumorsphere assay is considered as a surrogate assay [
6] which was employed extensively in this study.
Screening of untreated cervical carcinoma tissue biopsies revealed median CD133 levels were 10.4% and this satisfies one of the defining criteria of stem cells that they should constitute a minority of the tumor cell population. We recruited 6 cases of carcinoma cervix who had relapsed following chemoradiation therapy to provide proof-of-principle for the involvement of cancer stem cells in radiation-resistance. Supporting this hypothesis, these relapsed cases showed upregulation of CD133 and down-regulation of CD44. Thus, our study provides in-vivo evidence for the role of CSC in radiation-resistance corroborating the in vitro evidence demonstrated previously in cell lines wherein sphere-derived cells were shown to be more resistant to radiation than their adherent counterparts [
8].
Our observations that CD133+ is a phenotypic marker of cancer stem cells in carcinoma cervix are supported by a recent study by Tyagi et al. who have further shown the role of the HPV E6 oncogene in maintenance of the stemness characteristics of cancer stem cells in carcinoma cervix [
22]. An overall but variable upregulation of the stemness and EMT related transcripts in the CD133
+ CSC was observed confirming previous reports [
4,
8]. Out of the 4 cell lines, RSBS-23 showed marked vimentin expression with concomitant low levels of E-cadherin implicating epithelial-mesenchymal transition occurring even in the basal state. The heterogeneity of the expression of stemness and EMT markers among the CD133
+ cancer stem cells in these low-passage cell lines better reflects the in vivo picture rather than studies on highly passaged cell lines.