Objective
We sought to determine a microRNA (miRNA) profile of surgically staged endometrial cancers.
Study Design
RNA was extracted from archival primary endometrial cancers, and an miRNA profile was established using a microarray and confirmed with real-time polymerase chain reaction. Targets of differentially expressed miRNAs were explored using real-time polymerase chain reaction and Western blot in endometrial cell lines.
Results
Endometrial cancer has an miRNA profile distinct from normal endometrium, even in patients with stage IA grade 1 tumors. This miRNA cancer profile was able to correctly assign a specimen as a malignancy with a sensitivity of 92%. Overexpressed miRNAs were predicted to target PTEN, and transfection of cell lines with these miRNAs led to down-regulation of PTEN expression. In advanced disease, an miRNA pattern distinct from early-stage disease was seen, and overexpression of mir-199c predicted improved cancer survival in this population.
Conclusion
Endometrial cancer has a distinct miRNA profile, and miRNAs can be used as a predictive biomarker.
Endometrial cancer is the most common malignancy of the female genital tract, with an estimated 42,160 cases expected in 2009 and 7780 deaths due to the disease. Surgery is typically curative, with the majority of women presenting with early-stage disease. Nonendometrioid (type II) endometrial cancers, however, tend to present at more advanced stages, are more aggressive, and are associated with different molecular fingerprints.
There is emerging research about the role of microRNAs (miRNAs) in a variety of pathologic conditions, including both solid and hematologic malignancies. miRNAs are short, 22- to 25-nucleotide noncoding sequences of RNA. These sequences control gene expression either by translational repression or degradation of the messenger RNA transcript. These miRNAs may be up-regulated or down-regulated in various tumor types, thus displaying their roles as potential tumor suppressors or oncogenes. The target of the miRNA is the 3’UTR (untranslated region) of the messenger RNA. Early studies with Caenorhabditis elegans showed that a great number of these sequences are highly conserved across all species, demonstrating the important roles that miRNAs play in cellular differentiation, proliferation, and cell cycle control. Extensive work has thus far been accomplished with leukemias, lymphomas, and a variety of solid tumors. High-risk subtypes of leukemia are able to be identified by miRNA profiles, and targeted, personalized treatment can be accomplished. The miRNA fingerprint of gynecologic malignancies is in its infancy; to date, limited data existed describing miRNAs in endometrial cancer. Here, we present a comprehensive analysis of the miRNA profile of surgically staged early and advanced endometrial cancers and investigate targets associated with differentially expressed miRNAs and their value as predictive biomarkers.
Materials and Methods
Patients
Following approval from the Cancer Institutional Review Board of The Ohio State University College of Medicine, Columbus, OH, all patients with primary stage I endometrioid endometrial cancers and stages III and IV endometrioid and nonendometrioid endometrial cancers surgically managed by the Division of Gynecologic Oncology, Department of Obstetrics and Gynecology at The Ohio State University Medical Center, Columbus, OH, from January 1997–July 2003, were identified from a clinical database. Patients who underwent comprehensive surgical staging (including extrafascial or radical hysterectomy, bilateral salpingo-oophorectomy, and bilateral pelvic and paraaortic lymphadenectomy) were selected for analysis. In general, patients with uterine-confined disease who are comprehensively staged do not undergo any adjuvant therapy. Data regarding patient demographics, operative findings, tumor pathology, and survival were recorded from the source data.
miRNA microarray
The 2-mm archived paraffin-embedded tissue cores were obtained from primary surgical specimens. Although microdissection was not used, cases were selected that had adequate material for macroscopic dissection to assure >75% neoplastic cellularity. The cores were deparaffinized in 1 mL of xylene and heated at 50°C for 3 minutes. RNA extraction was performed using specified instructions with the RecoverAll kit (Ambion, Austin, TX). RNA concentrations were measured using the NanoDrop-1000 (NanoDrop Technologies Inc, Wilmington, DE) and were generally between 10 ng–5 μg of RNA. Due to the duration of time between primary surgery and RNA extraction, this RNA was often degraded but still usable for miRNA analysis. In brief, 5 μg of total RNA were reverse transcribed using biotin-labeled random octamer oligonucleotide primer. The biotin-labeled complementary DNA targets were hybridized to the miRNA chip (OSUCCC [Ohio State University Comprehensive Cancer Center] chip v.3) that contains the oligo probes generated and derived from 326 human and 249 mouse miRNA genes printed in duplicate. After hybridization, the chips were processed for signal detection and amplification using streptavidin Alexa-647 (Invitrogen, Carlsbad, CA) conjugate. The processed array slides were scanned on a microarray scanner (Axon 4000B; Axon Instruments, Sunnyvale, CA). The microarray data were extracted by GenePix Pro software (Axon Instruments) and exported as a *.gpr data file. For the analysis of the stages III and IV cancers, 20 unselected stage I endometrial cancers and all of the advanced-stage cancers were screened using the OSUCCC chip v.4 array (which expands the number of miRNAs from v.3).
Statistical analysis
Microarray images were analyzed using GenePix Pro 6.0 software. Average values of the replicate spots of each miRNA were background subtracted, normalized, and further analyzed. Normalization was performed using a global median. The miRNAs identified as having statistically significant differential expression were measured as present in at least as many samples as the smallest class in the dataset (25%). Absent calls were thresholded to 4.5 (log2 scale) before statistical analysis, representing the average minimum intensity level detectable in the system. More than 95% of blank probes (negative controls) fall below the threshold value of 4.5. MiRNAs differentially expressed at a 2-fold level were considered to be significant.
miRNA signatures were determined by the class prediction (biometric research branch) method. Six methods of prediction were used: compound covariate predictor, diagonal linear discriminate analysis, nearest neighbor (using k = 1 and 3), nearest centroid, and support vector machines. The compound covariate predictor and support vector machines are only implemented for the case when the phenotype variable contains only 2 class labels, whereas the diagonal linear discriminate analysis, k-nearest neighbor, and nearest centroid may be used even when the phenotype variable contains >2 class labels. The class prediction analysis was also performed on paired samples. The criterion for inclusion of a gene in the predictor was a P value less than a specified threshold value. For the 2-classes prediction problem, a specified limit on the univariable misclassification rate was used instead of the parametric P value. The output contains the result of the permutation test on the cross-validated misclassification rate, and a listing of genes that comprise the predictor, with parametric P values for each gene and the coefficient of variation-support percent (percent of times when the gene was used in the predictor for a leave-one-out cross-validation procedure).
The ability to classify a specimen into a class was assessed by calculating the sensitivity (the probability for a sample to be predicted to be in its correct class) and positive predictive value (the probability that a sample predicted to be in a class is actually in that class). To determine the association of specific miRNAs with survival, Kaplan-Meier survival analysis was performed. The 2 populations were compared with the log-rank test, and P < .05 was considered to be a statistically significant difference in survival.
DNA mismatch repair
DNA mismatch repair (MMR) status was determined using a tissue microarray stained with antibodies against the MMR gene protein products MLH1, MSH2, MSH6, and PMS2.
Targeting of PTEN by mir-200c
Targets of specific miRNAs were investigated using bioinformatics technology. “Confirmed” targets are those targets that have been experimentally confirmed and reported, while “predicted” targets are those based on sequence similarity without experimental confirmation. Confirmed targets were examined on http://diana.cslab.ece.ntua.gr . Predicted targets were examined with http://microrna.sanger.ac.uk/sequences . RL95-2 and KLE endometrial cancer cell lines (kindly provided by Paul J. Goodfellow, PhD, Washington University School of Medicine, St. Louis, MO) were grown in culture and harvested. RL95-2 cells were grown in Dulbecco’s modified Eagle Medium/10% fetal bovine serum/antibiotic. KLE cells were grown in Ham’s F12/10% FBS/antibiotic. The cells were grown to 90% confluence, at which time they were washed with phosphate-buffered saline (PBS), trypsinized, and centrifuged. Cell pellets were frozen at –80°C until RNA or protein extraction was performed.
For transfections, cell lines were grown to 90% confluence, at which time they were trypsinized and resuspended in 30 mL of PBS. A cell counter was used to determine viable cells/mL. Approximately 2 × 10 6 cells were plated and grown in media devoid of antibiotic. Twenty-four hours later the transfections of the cells with the specific miRNAs were performed using lipofectamine (Ambion) according to the manufacturer’s instructions. Cells were harvested at 24, 48, and 72 hours and the protein extracted. Western blotting was performed utilizing antibodies to both PTEN and a standard housekeeping gene.
miRNA Northern blots
For Northern blot analysis, 5 μg of total RNA was separated on a 15% denaturing polyacrylamide gel. The gels were washed and stained with ethidium bromide for 5 minutes. The presence of RNA was confirmed using ultraviolet illumination. Gels were then washed and the RNA was electrotransferred to Hybond-N+ membranes (Amersham Biosciences, Buckinghamshire, UK) overnight. Oligonucleotide probes complementary to mir-200c were labeled with γ-adenosine triphosphate. Hybridization to the membrane was performed at 42°C overnight and the membrane was then washed in 2X saline-sodium phosphate-ethylene diaminetetraacetic acid/0.1% sodium dodecyl sulfate (SDS) and 0.5X saline-sodium phosphate-ethylene diaminetetraacetic acid/0.1% SDS for 15 minutes each. The membrane was then exposed to a phosphor-screen (GE Healthcare Life Sciences, Piscataway, NJ) overnight. Imaging was performed on the Typhoon 9410 variable mode imager (GE Healthcare Life Sciences).
Quantitative real-time polymerase chain reaction
The single tube TaqMan miRNA assays were used to detect and quantify mature miRNAs on real-time (RT) polymerase chain reaction (PCR) instruments (Applied Biosystems, Foster City, CA). All reagents, primers, and probes were obtained from Applied Biosystems. RNU44 was used to normalize all RNA samples. Reverse transcriptase reactions and RT-PCR were performed according to the manufacturer’s protocols, except 7.5 and 10 μL were used for the RT and PCR reactions, respectively. RNA concentrations were determined with a NanoDrop. A total of 1 ng RNA per sample was used for the assays. All RT reactions, including no-template controls and RT minus controls, were run in a GeneAmp PCR 9700 Thermocycler (Applied Biosystems). Gene expression levels were quantified using the ABI Prism 7900HT sequence detection system (Applied Biosystems). Comparative RT-PCR was performed in triplicate, including no-template controls. Relative expression was calculated using the comparative C t method.
Gene expression
All reagents, primers, and probes were obtained from Applied Biosystems. Glyceraldehyde 3-phosphate dehydrogenase was used to normalize all RNA samples. Reverse transcriptase reactions and RT-PCR were performed according to the manufacturer’s protocols. RNA concentrations were determined with a NanoDrop. A total of 150 ng RNA per sample was used for the assays. All RT reactions, including no-template controls and RT minus controls, were run in a GeneAmp PCR 9700 Thermocycler. Gene expression levels were quantified using the ABI Prism 7900HT sequence detection system. Comparative RT-PCR was performed in triplicate, including no-template controls. Relative expression was calculated using the comparative C t method.
Western blot
Protein was extracted from the endometrial cancer cell lines using the cell lysis buffer according to the manufacturer’s specified instructions. In all, 50 μg of total protein was mixed with electrophoresis sample buffer (125 mmol/L Tris-hydrochloric acid, 2% SDS, 5% glycerol, .003% bromophenol blue, and 1% beta-mercaptoethanol). The samples were heated to 100°C for 10 minutes. The samples were loaded into a gel and 1 L of running buffer was added to the apparatus. The gel electrophoresed at 100 V until the protein band was approximately 1 cm from the bottom of the gel. The gel was subsequently transferred to the filter using a blotting apparatus. The filter was incubated with mouse monoclonal antibodies for PTEN overnight (1:1000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The filter was subsequently washed with Tween/PBS for 25, 5, and 5 minutes. The secondary antibody was diluted per manufacturer’s instructions in Tween/PBS/1 g nonfat dry milk. The filter was incubated for 1.5 hours at room temperature. The antibody solution was decanted and the filter was washed with Tween/PBS for 25, 5, and 5 minutes. The washing solution was decanted, the development solution (electrochemiluminescence) was added, and the tray was agitated for 1 minute. The filter was dried between 2 pieces of 3-mm paper and exposed to x-ray film (Denville Scientific, Metuchen, NJ) for approximately 2 minutes and developed.
Results
Subjects
Of the 478 patients available for study in the divisional endometrial cancer database, 141 were selected who were surgically staged and had adequate archival material for RNA extraction. Of these patients, 121 had stage I disease (surgical stage IA, IB, and IC disease comprising 37%, 43%, and 20%, respectively, of the overall stage I group) ( Table 1 ); all had endometrioid histology. Of the 20 patients with advanced disease, 10 (50%) had stage III and 10 (50%) had stage IV disease. In the group of patients with advanced disease, 7 were endometrioid and 13 serous adenocarcinomas ( Table 1 ). Twenty unmatched endometrial samples (10 premenopausal unselected for menstrual day and 10 postmenopausal) served as the unaffected control population.
