Background
Endometrial cancer is the most common cancer of the female reproductive organs in the United States [
1‐
5]. The continual remodeling of the endometrial lining at menses strongly argues for the presence of a stem/progenitor cell population with regenerative capabilities. This is further supported by studies of the benign endometrium in primate models and clonogenicity assays of human derived uterine cells [
6‐
10]. Mouse studies utilizing pulse-chase experiments to demonstrate evidence of label retaining cells in the uterus [
11,
12] provide additional functional evidence to support this concept. Thus, it has been proposed that an aberrant stem/progenitor cell or a cell that regains some stem-like properties can contribute to pre-malignant endometrial hyperplasia and/or endometrial cancer [
6,
7,
13].
Several investigators have identified putative stem/progenitor cells in solid tumors and cancer cell lines within the side population (SP), which is distinguished by differential efflux of Hoechst 33342 dye via verapamil-sensitive multidrug resistance transporters [
14‐
16]. Our previous work [
13] identified a tumorigenic SP within a human endometrial cancer cell line that displayed increased chemoresistance and quiescence
in vitro relative to its non-SP counterpart. Hubbard and colleagues [
7] demonstrated endometrial cancer cells with clonogenic, self-renewing, differentiating and tumorigenic properties further supporting the hypothesis that a cancer stem cell population may be responsible for seeding tumors or metastatic lesions.
Tumor initiating cells have been identified in leukemia [
17] and in a variety of solid tumors [
18‐
23] based on differential expression of one or more cell surface markers, suggesting that tumor initiating cell heterogeneity exists for each specific tumor type. The CD133 (human Prominin-1, AC133) cell surface antigen was originally identified in hematopoetic stem cells [
24,
25] and shown to be expressed on primitive cells of neural, endothelial and epithelial lineages. Several investigators have identified CD133 as a potential tumor initiating cell marker in solid tumors of the brain [
18], prostate [
19], colon [
20] and more recently the ovary [
21,
22] and endometrium [
23]. CD133
+ cells have been associated with an increase in
in vivo tumor initiation [
18,
26], asymmetric cell division and increased resistance to chemotherapeutic drugs [
26], as compared to CD133
- cells. Additionally, SP fractions have been reported to be enriched for CD133
+ cells [
27]. In the ovary, CD133
+ cells have been associated with the presence of primary disease rather than with normal ovaries or metastatic omental lesions [
22] and sorted ovarian CD133
+ tumors cells form more aggressive tumor xenografts compared to their CD133
- progeny [
26]. Similarly, CD133
+ cells isolated from endometrioid adenocarcinomas were resistant to cisplatin- and paclitaxel-induced cytotoxicity [
23]. When serially transplanted into NOD/SCID mice, CD133
+ cells were capable of initiating tumor formation that resembled the phenotype of the original tumor [
23]. Together these data support the hypothesis that CD133 is expressed by human endometrial cancers and may serve as a marker of more tumorigenic cells.
Recent studies have indicated that CD133 expression and antigenic potential [
28] may be regulated in part by histone modification [
26], DNA methylation [
29,
30] and/or glycosylation [
28,
31]. Regulation of CD133 expression by DNA methylation-dependent mechanisms has been observed in glioblastoma [
24,
32], colorectal cancer [
29,
30] and ovarian cancer [
26]. Our objectives were to confirm the tumorigenic potential of CD133
+ cells in immunocompromised mice, assess whether CD133 levels increased in serially transplanted tumors concurrently with their accelerated tumor formation rate and determine whether methylation status was associated with changes in the levels of CD133.
Our results confirm that endometrial tumors contain CD133+ cells, which can generate new tumors following injection in NOD/SCID mice. Cell fractions enriched for CD133+ cells gave rise to tumors at a faster rate than CD133- cell populations at fewer cell numbers injected. Interestingly, the level of CD133+ cells, as determined by immunofluorescence, in tumor explants appeared to be enriched with sequential serial transplantation of the tumor cells although this apparent increase in CD133 levels was not consistently detected by flow cytometry. Despite the fact that the percentage of CD133+ cells varies widely in established endometrial cancer cell lines, the level of mRNA encoding CD133 was elevated following treatment with the demethylating agent, 5-aza-2'-deoxycitidine (5-aza-dc), suggesting that methylation status may be important in the regulation of CD133 expression or epitope presentation. This concept was further supported by evidence that the CD133 promoter is hypomethylated in primary endometrial cancer tissue compared to benign endometrium. Collectively, these data support the hypothesis that CD133 may serve as a marker to assess potential tumorigenicity of endometrial cancer cells and that its expression levels are controlled in part through epigenetic regulation.
Methods
Human primary endometrial epithelial cell isolation
All primary human uterine tissues were collected in accordance with the policies of the Massachusetts General Hospital (MGH) Institutional Review Board. A subset of samples was collected utilizing the MGH GYN tissue repository after obtaining informed consent and a second subset was collected as anonymized discarded tissue. As per IRB protocol, the samples (n = 12) were not linked to clinical information. The histological subtype and grade of each sample was retrospectively assessed by an MGH pathologist (LRZ). The details of the tumor samples used in this study are listed in Table
1.
Table 1
Pathological evaluation of the human endometrial tumor samples used
T1 | Grade 1 Endometrioid |
T2 | Grade 3 Endometrioid |
T3 | Grade 3 Endometrioid |
T4 | Grade 1 Endometrioid |
T5 | Grade 1 Endometrioid |
T6 | Grade 1 Endometrioid |
T7 | Grade 2-3 Endometrioid |
T8 | Grade 3 Endometrioid |
T9 | Grade 1 Endometrioid |
Endometrial carcinoma tissues were minced to yield 2 mm3 pieces and incubated with agitation in HBSS (Cambrex Corp., East Rutherford, NJ)/2% FBS (HyClone, Logan, UT)/1 mM EDTA (Sigma-Aldrich, St. Louis, MO) containing 1 mg/ml collagenase Type II (Sigma) and 0.025% DNase I (Sigma) at 37°C for 1 hour. Following this incubation, the supernatant was removed and discarded. The remaining digested tissue was washed with Dulbecco's phosphate buffered saline (PBS) (Cambrex) and resuspended in DMEM medium (Mediatech Inc., Herndon, VA) containing 2% FBS, added L-glutamine (100 U/ml), penicillin (1%), streptomycin (1%) and 2.5 μg/ml amphotericin B (Sigma), and incubated in a T75 flask at 37°C, 5% CO2 in a humidified chamber for 1 hour. The flask was rotated at 20-minute intervals during this incubation period to maximize binding of stromal cells to the vessel walls. Endometrial epithelial cells, which make up the bulk of the cells in the non-adherent cell population, were then removed from the flask. The number of non-viable cells, as determined by Trypan blue staining (Mediatech Inc.), was assessed and these cells were eliminated from the suspension using the Dead Cell Removal Kit (Miltenyi Biotec Inc., Auburn, CA) as necessary.
In vivo endometrial tumorigenesis assay
All experiments utilizing mouse models were reviewed and approved by the MGH Institutional Animal Care and Use Committee and were performed in strict accordance with the NIH Guide for the Care and Use of Laboratory Animals. Six to twelve week old female NOD/SCID mice (strain NOD.CB17-Prkdcscid/J, Jackson Laboratory, Bar Harbor, ME) were used for all injections of human primary endometrial tumor epithelial cells. Defined numbers of isolated primary endometrial epithelial cells were suspended in 1:1 PBS/Matrigel® (BD Biosciences, San Jose, CA) and subcutaneously (s.c.) injected into the right dorsal side of NOD/SCID mice. Control animals were simultaneously injected with 1:1 PBS/Matrigel® only. Tumor development was assessed bi-weekly.
Tumor initiating capacity of transplanted endometrial cells
Xenograft processing
Mice bearing tumors generated following injection of primary endometrial tumor epithelial cells were euthanized by CO2 inhalation. The generated tumors were isolated aseptically, minced to yield 2 mm3 pieces and incubated with agitation at 37°C for 30 minutes in HBSS/2% FBS, 1 mM EDTA containing 1 mg/ml collagenase Type II, 0.025% DNase I and 2.5 μg/ml amphotericin B. Cells were filtered through a 100 μm mesh filter (BD Biosciences) and washed 3 × 5 minutes in HBSS/2% FBS/1 mM EDTA. Pelleted cells were resuspended in ACK lysis buffer (Cambrex) for 1 minute at room temperature to lyse red blood cells. The remaining cells were washed in HBSS, resuspended in HBSS/2% FBS/1 mM EDTA and layered over Ficoll-Paque™ PLUS (GE Healthcare Bio-Sciences Corp., Piscataway, NJ). Non-viable cells were eliminated from the suspension using a Dead Cell Removal Kit (Miltenyi Biotec Inc.). H-2Kd+ mouse cells were removed using a FITC conjugated antibody (BD Biosciences) and MACS® LD separation columns (Miltenyi Biotec Inc) as per manufacturers' recommendations. H-2Kd- cells were suspended in 1:1 PBS/Matrigel® and injected s.c. into the right dorsal side of NOD/SCID mice. Tumor development was assessed bi-weekly.
CD133 isolation via magnetic beads
Isolated H-2Kd- cells were separated into CD133+ and CD133- fractions using CD133 microbeads (Miltenyi Biotec Inc.) and MACS® LD separation columns as per manufacturers' recommendations. To acquire a pure CD133- population, CD133- cells were passed twice through separate LD columns. Defined numbers of CD133+ and CD133- cells were injected into NOD/SCID mice as previously described.
Flow cytometry
CD133 profiling
To examine expression of CD133, single human endometrial cells from primary or transplanted tumors were isolated as outlined previously [
33]. Following incubation with FcR blocking reagent (Miltenyi Biotec Inc.) to reduce unwanted binding of antibody to Fc receptor-expressing cells, tumor cells were resuspended in PBS/2% FBS/1 mM EDTA and stained with 0.5 μg anti-CD133 (phycoerythrin (PE)-conjugated; Miltenyi Biotec Inc.). Respective IgG isotype antibodies were included as negative controls. Non-viable cells were excluded using the LIVE/DEAD Fixable Dead Cell Stain kit (Invitrogen, Carlsbad, CA) as per manufacturer's recommendations. For primary tumors, CD31
+ and CD45
+ cells were excluded using FITC-conjugated CD31 and CD45 antibodies (Miltenyi Biotec Inc.) Similarly, for xenograft tumors, H-2K
d+ cells were eliminated using a FITC-conjugated H-2K
d antibody. After washing in PBS/2% FBS/1 mM EDTA, cells were fixed by incubation in 4% paraformaldehyde for 60 minutes and analyzed using a LSRII (BD Biosciences) within 24 hours. Data were analyzed using FlowJo version 8.2 software.
Cell sorting
Single cell suspensions derived from endometrial xenograft tumors were stained with anti-CD133 as described. CD133+ and CD133- cell populations were separated using a FACSAria flow cytometer, with post-sort analysis performed to confirm population purity. Sorted cell populations were serially diluted in 1:1 PBS:Matrigel® and injected s.c. into female NOD/SCID mice.
Oligonucleotide array CGH (aCGH)
Array CGH was performed to determine if there were any DNA copy number changes in cells derived from serially transplanted endometrial tumors and from tumors generated from CD133
+ and CD133
- injected cells using Agilent Human 105K oligonucleotide microarrays as per manufacturer's instructions
http://www.home.agilent.com/agilent/home.jspx. Genomic coordinates for this array are based on the NCBI build 36, March 2006 freeze of the assembled human genome (UCSC hg18), available through the UCSC Genome Browser. This array provides an average spatial resolution of 21.7 kb.
Genomic DNA was isolated from primary tumors using standard protocols. For array hybridizations, 5 micrograms each of tumor and normal DNA were digested with Dpn II for 3 hours at 37°C and purified with QIAquick PCR purification columns. One microgram each of purified tumor and normal DNA was labeled with Cy3-dCTP and Cy5-dCTP, respectively, using Bioprime labeling kit (Invitrogen) in accordance with the manufacturer's instructions. Unincorporated nucleotides were removed using Sephadex G-50 columns. Labeled tumor and reference samples were precipitated with 50 micrograms of human Cot-1 DNA and resuspended in 250 microliters of hybridization buffer provided in the Agilent oligonucleotide array CGH kit. Prior to hybridization, probe mixtures were denatured for 5 minutes at 95°C and incubated at 37°C for 30 minutes. Samples were then hybridized onto the oligonucleotide array in the Agilent SureHyb microarray hybridization chamber and hybridization was performed for 42 hours at 65°C. The arrays were disassembled and washed as recommended by the manufacturer. Slides were dried and scanned with an Axon 4000B microarray scanner using GenePix Pro 4.0. Microarray images were analyzed and data points generated using the Feature Extraction software (version 9.1, Agilent Technologies) with linear normalization (protocol-v4_91). Data were subsequently imported into CGH Analytics software (version 3.4.40, Agilent Technologies). Detection of gains and losses were based on the z-score algorithm (threshold 2.5) and visual inspection of the log2 ratios. Log2 ratios ≥0.4 in at least five consecutive probes were considered a reliable copy number alteration. Probes with log2 ratios greater than 2 were considered highly amplified.
CD133 immunofluorescence
Immunofluorescence was carried out on 6 micron sections of formalin-fixed, paraffin embedded (FFPE) biopsies from primary human endometrial and xenograft tumors. Antigen retrieval was carried out using 10 mM Citrate (pH 6.0). After blocking, sections were incubated with CD133 antibody (K-18; Santa Cruz) or Goat Negative Control at 1:50 dilution overnight at 4°C. Donkey anti-goat Alexa Fluor 568 was used as secondary antibody. DAPI was used to stain nuclei.
Methylation analysis of CD133 promoter
Human endometrial cancer cell lines were treated with either vehicle or 5-aza-2'-deoxycytidine for 72 hours. CD133 expression was evaluated by RT-PCR and flow cytometry.
Expression of CD133 mRNA in benign and malignant samples
FFPE tissue specimens were obtained from the MGH Pathology archives with IRB approval. Histological review was performed by a pathologist on hematoxylin and eosin stained slides to confirm pathology and designate areas of tumor on relevant slides. Tissue was macrodissected from serial 5 micron slides and total nucleic acids were extracted using a custom fully-automated platform based on the FormaPure System (Beckman Coulter Genomics, Danvers, MA) and Beckman Coulter Biomek NXP workstation.
RT-PCR analysis
Single stranded cDNA was prepared with the Superscript First-Strand System (Invitrogen). Mock reactions were prepared under the same conditions but lacked reverse transcriptase. The cDNAs were amplified by PCR using the following primer sets: CD133-3: 5'-AGCTTCTCTGGATTTTGCTCA-3' (forward) and 5'-CACAGAAAGACATCAACAGCAG-3' (reverse); CD133-5: 5'-CAGAAGGCATATGAATCCAAAA-3' (forward) and 5'-CTGTCGCTGGTGCATTTCT-3' (reverse); β-actin: 5'-CTTCCAGCCTTCCTTCCTG-3' (forward) and 5'-TTGGCGTACAGGTCTTTGC-3' (reverse). The PCR products were analyzed by 1.5% agarose gel elctrophoresis.
Bisulfite-treated genomic DNA sequencing
Genomic DNA was isolated from normal endometrium and endometrial tumor using the high pure PCR template preparation kit (Roche). Bisulfite treatment was carried out using EZ DNA methylation kit (Zymo Research). Mapping of methylated cytosines was carried out by bisulfite-treated genomic DNA sequencing. The CD133 promoter region that encompasses the CpG island was divided into three regions defined by the PCR primers used to amplify the bisulfite-treated DNA. Ten individual clones were analyzed per region and tissue sample. The percentage of CpG methylation within each region was compared between benign and malignant tissue. All malignant tissues analyzed (n = 3, T7-T9) were endometrioid endometrial adenocarcinoma. Benign endometrial tissue (n = 3) was derived from pre-menopausal women with no evidence of malignant disease who were undergoing hysterectomy as a result of extensive uterine fibroids.
Statistical analysis
Student's t test was used for statistical comparisons where appropriate. A p value of < 0.05 for the t test was considered to be statistically significant.
Discussion
To date, the complement of surface markers utilized for the isolation of tumor initiating cells (also referred to as cancer stem or cancer initiating cells) from solid tumors have varied in breast (CD44
+, CD24
-/low, EpCAM
+, Lineage
-[
36]), brain (CD133
+[
18]), pancreas (CD44
+/CD24
+/EpCAM
+[
37]), colon (CD133
+[
20,
38]) and prostate (CD44
+/α
2β
1hi/CD133
+[
19]) cancer. More recently, gynecological cancer initiating cells have been isolated based on either differential expression of a number of surface antigens (CD44
+, CD117
+ or CD133) or differential exclusion of Hoechst dye [
13,
16,
33,
34,
39,
40]. In studies to date, CD133 appears to be one of the most consistent markers of gynecological cancer initiating cells. While the link to tumorigenicity has been met with some criticism [
28], differential CD133 expression has proven to be an effective tool in the isolation of sub-populations of ovarian and endometrial cancer cells with increased tumor forming capacity [
22,
23,
33]. Consistent with previous reports, we detected CD133 expressing cell populations in tumors obtained from patients diagnosed with endometrioid endometrial adenocarcinoma. The CD133
+ cells had increased tumor forming capacity relative to their CD133
- counterparts. Interestingly, we determined that CD133 expression increased in serially transplanted tumor xenografts when assessed by immunofluorescence although this increase was less consistently evident in flow cytometric analyses. The differences in the results obtained from these analyses may reflect technique-related variation in antigen presentation/epitope exposure as described by others [
28]. The inconsistencies in the flow cytometric analyses likely result in part from variability in sample processing and antibody preparation. Our aCGH analyses determined that the increase in CD133 expression across the serial transplants did not result from changes in gene copy number. We did, however, detect a relative decrease in the level of methylation at the CD133 chromosomal locus (data not shown). This observation led us to more thoroughly investigate potential epigenetic regulation of CD133 expression.
The CD133 gene was previously reported to have 5 distinct promoter regions (P1-P5) that are activated in a tissue-specific manner [
41]. Promoter regions P1-P3 are located within a CpG island and P1 and P2 are inactive when methylated, suggesting epigenetic regulation of these regions. We first assessed the level of CD133 mRNA in the human endometrial cancer cell lines Hec 1A, Hec 1B, AN3CA and Ishikawa and found it to be highly variable. Treatment with a demethylating agent resulted in an increase in the relative expression of CD133 mRNA in three of the four cell lines, suggesting that CD133 expression in these cell lines is regulated in part by methylation. This hypothesis was further supported by flow cytometric analyses of the same cell lines which determined that CD133 expression was increased relative to vehicle treated controls following treatment with the demethylating agent.
We next investigated the methylation status of the P1 and P2 CD133 promoter regions in primary human endometrial tumors. Our analyses indicated that the P1 promoter region was significantly demethylated in endometrial tumors compared to benign endometrium (Figure
6A). We found no obvious differences in P2 suggesting that this promoter region is not utilized in primary human endometrial tumors. This is in contrast to ovarian tumors, for example, where significant differences in P2 promoter methylation have been reported [
26].
We hypothesized that the in vivo level of CD133 expression should be elevated in malignant tissue relative to the level in benign tissue. Indeed, we detected a higher level of CD133 protein expression in primary human endometrial tumors compared to expression in benign proliferative and secretory endometrium. Moreover, the relative level of CD133 mRNA was much greater in malignant samples when compared to the levels in benign endometrial samples. Additionally, analysis of the relative methylation in 3 different regions of the CD133 promoter revealed significant hypomethylation of one region in malignant endometrial tumor tissue. Interestingly, we analyzed the methylation status of a primary tumor sample and compared it to its corresponding serial transplants and found the level of CD133 promoter methylation relative to that detected in the primary human endometrial tumor appeared to be reduced over the course of serial transplantation. As noted, the number of cells required to generate tumor significantly decreased with subsequent serial transplantation with a concomitant decrease in the time to onset of tumor formation. Our findings suggest that this may result from increased CD133 expression due to progressive promoter hypomethylation. Future studies designed to determine the transcription factors, co-activators or co-repressors that directly or indirectly function through the methylated CD133 promoter regions to regulate CD133 expression may be of importance.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AMF, LZ, RF and BRR conceived the study and participated in its design. AMF, LZ and MDC conducted the study. VAT and SEB were responsible for consenting patients and initial tissue processing. PAS processed tissue and isolated nucleotides. DRB isolated nucleotides from paraffin blocks and edited the manuscript. GM conducted aCGH. LZ served as the pathologist in this study. AMF, RF and BRR were primarily responsible for writing the manuscript. All authors read and approved the final manuscript.