MicroRNAs overexpressed in ovarian ALDH1-positive cells are associated with chemoresistance

The human ovarian carcinoma cell line SKOV3 was obtained from the American Type Culture Collection (Manassas, VA, USA). Paclitaxel (PTX)-resistant cell lines (SKpac) were produced from the parent cell line (SKOV3) by continuous exposure of a stepwise, escalating concentration of PTX from an IC₅₀ of 10% to 1000% over a period of 12 months. Seven different sublines (SKpac-8, -10, -11, -12, -13, -16, and −17) were generated. SKpac cells were 365.5-fold more resistant to PTX (IC₅₀ = 10.21 μM, 7.59 μM, 6.77 μM, 6.57 μM, 8.69 μM, 8.38 μM, and 5.32 μM, respectively) than that of the SKOV3 cell line (IC₅₀ = 22 nM). All cell lines were maintained in 5A medium (Gibco/Invitrogen, Carlsbad, CA, USA) with 10% fetal bovine serum (FBS) (Invitrogen), 100 U/mL penicillin, and 50 μg/mL streptomycin in a humidified atmosphere containing 5% CO₂ at 37°C. The primary tumor cells were obtained at the time of surgery from patients who had undergone oophorectomy for ovarian serous carcinoma. Tumors were mechanically dissected into single-cell suspensions and incubated at 37°C for 1 hour in Ca²⁺/Mg²⁺-free PBS containing 50 U/ml collagenase A (Roche, Pleasanton, CA, USA). The enzymatic reaction was blocked by adding RPMI medium (Gibco/Invitrogen) containing 10% FBS. Cells were reacted with BerEP4-coated magnetic Dynal beads (Invitrogen) for 30 min to select epithelial cells and then cultured with RPMI medium containing 10% FBS, 1% penicillin-streptomycin, and 10 μg/ml of endothelial cell growth factor (Invitrogen).

Thirty-four ovarian carcinomas were obtained at the time of surgery from patients who had undergone oophorectomies for ovarian epithelial tumors at the CHA Bundang Medical Center. The samples were immediately frozen in liquid nitrogen and stored at −80°C. Tumor cells comprised nearly 80% of frozen section tissue.

Clinical and pathological data were retrieved from clinical databases including the original pathology reports with histological grade and tumor stage. A histopathological examination of these samples was performed in the Department of Pathology according to the criteria of the International Federation of Gynecology and Obstetrics grading system and the World Health Organization classification. The patient’s samples were devided into two groups according to the responsiveness to the first-line chemotherapy. Based on the NCCN guidelines, intrinsically chemoresistant tumors were defined as those with persistent or recurrent disease within 6 months after the initiation of first-line taxol-platinum-based combination chemotherapy. Chemosensitive tumors were classified as those with a complete response to chemotherapy and a platinum-free interval of >6 months. This study was approved by the Ethical Committee of the CHA Bundang Medical Center, and informed consent was obtained from each patient prior to surgery. The clinicopathological data are summarized in Table 1.

The ovarian cancer cell line (SKOV3), chemoresistant sublines (SKpac-12, 16, and 17), and primary ovarian tumor cell suspensions were counted and ALDH enzymatic activity was detected using the ALDEFLUOR assay kit (StemCell Technologies Vancouver, BC, Canada) as described by the manufacturer. Half of the cell/substrate mixture in each sample was treated with 50 mM diethylaminobenzaldehyde (DEAB), and the cells were incubated for 40 minutes. Gating was established using propidium iodide (PI) exclusion for viability, and ALDEFLUOR/DEAB treated cells were used to define negative gates. FACS was performed with 1 × 10⁶ cells using the BD FACS Aria III (Becton Dickinson, Franklin Lakes, NJ, USA) under low pressure in the absence of UV light. FACS data were analyzed using BD FACS Diva software V6.1.3.

Semiquantitative RT-PCR was performed to assess the amount of stemness-related gene mRNA in ALDH1(+) cells compared with that in ALDH1(−) cells. Total RNA was extracted from ALDH1(+) or ALDH1(−) cells acquired by FACs sorting using Trizol reagent (Invitrogen). RNA (1 μg) was reverse-transcribed in a total volume of 20 μl using the SuperScript First-strand Synthesis system (Invitrogen). The primers for these genes are listed in Table 2. The PCR products were loaded on a 2% agarose gel, the ethidium bromide stained gel image was digitalized using the Molecular Imager Gel Doc™ XR System (Bio-Rad, Hercules, CA, USA), and calculated by densitometry. GAPDH was used as the reference gene to normalize mRNA amounts between samples. Results are presented as relative mRNA expression of each gene to that of GAPDH mRNA.

The FACS-sorted ALDH1 (+) or ALDH1(−) cells were plated in wells of an ultra-low attachment surface 6-well culture plate (Corning, Acton, MA, USA) at a density of 3,000 cells/ml in serum-free DMEM/F12 medium (Invitrogen) supplemented with 20 ng/ml epidermal growth factor (Invitrogen), 10 ng/ml basic fibroblast growth factor (Sigma-Aldrich, St Louis, MO, USA), 0.4% bovine serum albumin (Sigma-Aldrich), and 5 μg/mL insulin (Sigma-Aldrich). Spheroid formation (each with 50–100 cells/sphere) was assessed 12 days after seeding.

miRNA expression profiles of the FACS-isolated ALDH1(+) or ALDH1(−) cells from four PTX-resistant SKpac cell lines (SKpac-16-#1,-16 #2, -17#1, and −17#2) were determined with an Affymetrix GeneChip® miRNA array to select candidate miRNAs associated with ALDH1. The samples were prepared according to the manufacturer’s instructions. Total miRNA was isolated using TriZol reagent (Invitrogen), and RNA quality was assessed with an Agilent 2100 Bioanalyzer using the RNA 6000 Nano Chip (Agilent Technologies, Amstelveen, The Netherlands). One μg of total RNA was used as input for the labeling reaction, and labeled miRNA was hybridized to the array for 16 hours at 48°C and 60 rpm as described in the protocol. After hybridization, the chips were stained and washed in a Genechip Fluidics Station 450 (Affymetrix) and scanned with a Genechip Array scanner 3000 7G (Affymetrix).

We performed real-time qRT-PCR for ALDH1 and six selected miRNAs that showed significantly altered expression in ALDH1(+) cells compared with ALDH1(−) cells to validate the ALDEFLUOR assay and the microarray data. The qRT-PCR for ALDH 1 was conducted with ALDH1(+) or ALDH1(−) cells from SKpac-12, 16, and 17 cells. One μg of total RNA was converted to cDNA with the SuperscriptIII First-strand Synthesis System (Invitrogen) and qRT-PCR was conducted with specific primers and a probe for ALDH1. For miRNAs, one μg of total RNA was reverse transcribed to cDNA using stem-loop specific RT-primers of mature miRNAs and a TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). The qRT-PCR was performed with a Taqman Universal Master Mix (Applied Biosystems) using the Bio-Rad CFX96 Real-Time PCR Detection System (Bio-Rad). All PCR reactions were run in triplicate, and gene expression relative to GAPDH (for ALDH1) or RNU48 (for miRNAs) was calculated using the comparative threshold method (2^-ΔΔCt). Results for the ALDH1 (+) cells are expressed as fold-change relative to ALDH1 (−) cells.

We also conducted the qRT-PCR in various ovarian cancer cell lines (SKOV3, SKpac-8, 10, 11, 12, 13, 16, and 17) and 34 ovarian serous carcinoma samples including 15 chemoresistant and 19 chemosensitive carcinomas to assess the association ALDH1 and ALDH1(+)-related miRNA expression with chemoresistance. The results for the cancer cell lines are expressed as fold-changes of chemoresistant SKpac cells relative to the level of parent SKOV3 cells. The results for carcinoma samples are expressed as fold-change relative to the mean level of three normal ovarian surface epithelial cells.

We examined the population of ALDH1(+) cells, a potential cancer stem cell marker, in various ovarian cancer cell lines (SKOV3, A2780, and OVCAR 3), chemoresistant sublines (SKpac-12,-16,-17, A2780pac, and A2780cis), and primary tumor cells (SCN-1–7). The ALDH1(+) cells varied considerably among ovarian cancer cell lines with a range of 0.9–17.2% and with a range of 0.4–4.0% among the primary cancer cells (Table 3).

The ALDH 1(+) population percentage by FACS analysis in chemoresistant SKpac cells (4.46 ± 0.71%) increased significantly (Figure 1A,B) compared with that in parental SKOV3 cells (0.9 ± 0.2%), suggesting an association between ALDH 1(+) cells and chemoresistance.

We performed qRT-PCR for ALDH1 mRNA to confirm whether FACS-sorted ALDH1 (+) cells showed increased ALDH1 expression. The FACS-sorted ALDH1(+) population showed 2–5 fold higher expression of ALDH1 mRNA compared with that in the ALDH1(−) population (Figure 1C).

The stem cell phenotype of the ALDH1(+) cells was confirmed by expression of several stemness-associated markers for Notch3, Oct-4, NAC1, c-Kit, BMI-1, Nanog, and SOX2 using semiquantitative RT-PCR. The mRNA expression of Notch3 (1.73-fold) was significantly higher (t-test, p = 0.04) in ALDH1(+) cells than that in ALDH1(−) cells (Figure 2A). Oct-4 (1.19-fold), NAC1 (1.39 -fold), and Sox2 (1.26-fold) mRNA levels tended to be elevated in ALDH1(+) cells compared with those in ALDH1(−) cells, but the difference was not significant. BMI-1 (1.02-fold), c-Kit (1.16-fold), and Nanog (1.05-fold) mRNAs were not differentially expressed.

We compared the spheroid formation capability between ALDH1 (+) and ALDH 1(−) cells to test the stem cell-like property. ALDH1(+) cells showed increased size and number (106 ± 37) of spheroids compared with ALDH1(−) cells (63 ±15) (Figure 2B) (t-test, p = 0.05).

We examined ALDH1 mRNA expression in ovarian cancer cells and 34 cases of ovarian cancer tissue samples using qRT-PCR. The expression levels of various chemoresistant SKpac cells were compared with those of the chemosensitive SKOV3 parent cell line. ALDH1 mRNA expression in chemoresistant SKpac cells was markedly elevated (17 ~ 124-fold) compared with that in SKOV3 (Figure 3A). The ALDH1 mRNA expression level relative to that of normal epithelial cells was calculated and compared between chemosensitive and chemoresistant groups in the human cancer tissue samples. The relative expression level of ALDH1 mRNA (Figure 3B) in the chemoresistant group of human cancers (11-fold) was significantly higher (t-test, p = 0.024) than that of the chemosensitive group (4.29-fold).

To assess the role of ALDH1 in ovarian cancer, we analyzed the relationship of ALDH1 mRNA expression with clinicopathological parameters, including clinical stages, lymph node metastasis and distant metastasis. For comparison, patients were classified into high (≥2-fold) and low (<2-fold) ALDH1 expression groups. The high expression of ALDH1 mRNA was significantly (p = 0.019) associated with advanced clinical stage (Figure 3C). The ALDH 1 expression was not correlated with lymph node or distant metastases.

We performed the Affymetrix GeneChip microarray analysis containing 4,560 human precursor and mature miRNA oligonucleotide probes to characterize the miRNA expression profile of ALDH1 (+) cells. As a result, six miRNAs were differentially overexpressed more than 1.5-fold in ALDH1 (+) cells compared with that in ALDH1 (−) cells (Table 4, Figure 4A) (miR-424 [1.98-fold], miR-346 [1.95-fold], miR-503 [1.86-fold], miR-27a [1.66-fold], miR-23b [1.53-fold], and miR-27b [1.50-fold]). No miRNAs were downregulated in ALDH1 (+) cells.

We compared expression of six differentially expressed miRNAs between ALDH1(+) and ALDH1(−) cells using real-time RT-PCR to validate the microarray results. As a result, miR-424 (1.62-fold), miR-346 (3.25-fold), miR-503 (1.66-fold), miR-27a (2.08-fold), miR-23b (1.98-fold), and miR-27b (3.09-fold) were upregulated in ALDH1 (+) cells relative to ALDH1 (−) cells (Figure 4B).

We examined the expression of these miRNAs in ovarian cancer cells and 34 cases of ovarian cancer tissue samples using qRT-PCR to assess the impact of six ALDH1(+)-related miRNAs on chemoresistance in ovarian cancers. The expression levels of various chemoresistant SKpac cells were compared with those of the chemosensitive SKOV3 parent cell line. Among the miRNAs examined, the expression levels of miR-23b (2.8-fold, p = 0.039), miR-27b (3.5-fold, p = 0.007), miR-346 (2.7-fold, p = 0.02), and miR-503 (2.2-fold, p = 0.049) were significantly higher than those in SKOV3 cells (Figure 5A).

The expression levels of miRNAs in the human cancer tissue samples relative to those of normal epithelial cells were calculated and compared between the chemosensitive and chemoresistant groups. The expression levels of miR-23b (3.5-fold vs. 0.6-fold p = 0.037), miR-27b (5.5-fold vs. 1.6-fold, p = 0.040) and miR-424 (2.0-fold vs. 0.7-fold, p = 0.047) were significantly higher in the chemoresistant group than those in the chemosensitive group (Figure 5B).

We analyzed the correlation between the expression of these miRNAs and clinicopathological parameters, including clinical stage, lymph node metastasis, and distant metastasis to verify the role of ALDH1(+)-associated miRNAs in ovarian cancer. Patients were classified into high (≥1.5-fold) and low (<1.5-fold) expression groups. High expression of miR-503 (Figure 5C) was significantly associated (p = 0.033) with advanced clinical stage (stage III and IV), and upregulation of miR-27a (Figure 5D) was related to distant metastasis (p = 0.046). No other miRNAs were associated with the clinicopathological parameters.