MicroRNA-200c overexpression inhibits tumorigenicity and metastasis of CD117⁺CD44⁺ ovarian cancer stem cells by regulating epithelial-mesenchymal transition

The human EOC SKOV-3 cell line was from an ovarian cancer patient of origin, which was a well-established ovarian cancer model system. The cell line was purchased from the Cellular Institute in Shanghai, China, and was maintained in complete media consisting of RPMI 1640, 2 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% fetal bovine serum. The complete media were refreshed every 3 days to maintain the adherent cells.

The CD44⁺CD117⁺CSCs were isolated from the SKOV3 cell line by using the magnetic- associated cell sorting (MACS) method as described previously [16, 17]. Briefly, the CD44⁺subsets were isolated by using the mouse antihuman CD44 antibody coupled to magnetic microbeads (Miltenyi Biotec., Germany); the resulting cells were then depleted of CD117⁻subsets by using the mouse antihuman CD117 antibody coupled to magnetic microbeads (Miltenyi Biotec., Germany). The resulting CD44⁺CD117⁺cells were labeled ‘EOC CD44⁺CD117⁺CSCs’ [15]. These cells were further identified by using a flow cytometer (FCM, Beckman Coulter, USA) according to the manufacturer’s instructions. The anti-Human/ Mouse CD44 FITC and the anti-Human CD117 PE (eBioscience, USA) antibodies were used for the detection of the cells [18].

To generate the miR-200c expression lentivirus vector, we amplified an insert (full-length human miR-200c) by PCR from SKOV3 DNA. The lentivirus miR-200c was produced from the transient transfection of the HEK293T cells with pHAGE-CMV- miR-200c-IZsGreen, psPAX2, and pMD2.G plasmid DNAs plus Lipofectamine 2000 (Invitrogen, USA) according to the manufacturer's protocol. Forty-eight hours after the co-transfection, the lentivirus-bearing supernatants were collected and passed through a 0.45-mm filter. The CD44⁺CD117⁺CSCs were transduced with the pHAGE-CMV- miR-200c-IzsGreen lentivirus, and were selected by the IzsGreen expression [7, 19]. The stable expression colony was generated by limiting the dilution assay [20].

Total cellular RNA was isolated from a sample by using a Qiagen RNeasy Kit (Qiagen, CA). The sequences of the primers are as follows: the miR-200c-RT primer 5′-CTCAAC TGGTGTGGAGCGCATTCAGTTGAGTCCATCAT-3; the ZEB1 forward, 5′-GCACAAC CAAGGCAGAAGA-3′; reverse, 5′-CATTTGCAGTTGAGGCTGA-3′; the β-actin forward, 5′-GGACTTCGAGCAAGAGATGG-3′; reverse, 5′-AGCACTGTGTTGGCGTACAG-3′; U6: RT Primer, 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGA CAAATATGGAAC-3′;forward, 5′-TGCGGGTGCTCGCTTCGGCAGC-3′; URP Universal Reverse Primer, 5′-CCGGCAGGGTCCGAGGT-3′. The E-cadherin forward: CATTGCC ACACATACACTCTCTTCT, reverse: CGGTTACCGTGATCAAAATCTC; the Vimentin forward: GGAACAGCATGTCCAAATCG, reverse: GCACCTGTCTCCGGTACTCA. The qRT-PCR analysis was performed on an ABI step one plus real-time system (Applied Biosystems, USA) [19].

A short hairpin RNA sequence of the human ZEB1 was designed according to the ZEB1 DNA sequence (GenBank NO.NM_001128128.2)by the siDESIGN design software of Dharmacon Company. The shRNA sequences were as follows: The ZEB1-siRNA: forward, 5′- GATCCCCAG GAAGAGGAGGAGGATAATTCAAGAGATTATCCTCCTC CTCTTCCTTTTTTGGA AA-3′; reverse, 5′-AGCTTTTCCAAAAAAGGAAGAGGAGG AGGATAATCTCTTGAATTATCCTCCTCCTCTTCCTGGG-3′; the scramble-siRNA: forward, 5′-GATCCCCTTCT CCGAACGTGTCACGTTTCAAGAGAACG TGACACGT TC GGAGAATTTTTGGAAA −3′; reverse, 5′-AGCTTTCCAAAAATTCTCCGAACGTG TCACGTTCTCTTGAAACGTGACACGTTGGAGAAGGG-3′. All the primers were synthesized by Company of Gene and Technology of China in Shanghai. A pSUPER- EGFP1 (enhancement green fluorescent protein 1) vector was used to construct recombinants. The recombinant pSUPER-EGFP1- ZEB1-shRNA (shZEB1) was developed using the method described in a according to previous reports [1, 13]. A pSUPER- EGFP1-scrambled shRNA (scramble) was used for negative control. These recombinants were transfected by using the Lipofectamine™ 2000 reagent (Invitrogen, USA) per the manufacturer’s protocol [9].

The colony formation capabilities of the different CD44⁺CD117⁺CSCs were investigated. Colonies larger than 75 μm in diameter or containing more than 50 cells were counted as 1 positive colony according to our previous reports [20, 21]. About 500 cells per well were added into a six-well culture plate, with three wells per sample. After12-day incubation, the cells were washed twice with PBS and stained with the Giemsa solution. The plate clone formation efficiency was calculated as (number of colony /number of cells inoculated) × 100%.

The cells from the above-mentioned colonies were grown to confluence and were wounded by dragging a 1-mL pipette tip across their monolayer. The cells were washed to remove any cellular debris and were then allowed to migrate for 0, 24, and 48 hours, respectively, in a humidified 5% CO2 incubator at 37°C. Images were taken, using a DMI6000 inverted microscope (Leica Microsystems GmbH, Germany), at 0, 24, and 48 hours, respectively, after the wounding procedure [22, 23].

The Matrigel invasion assay was done using the BD Biocoat Matrigel Invasion Chamber (pore size: 8 mm, 24-well; BD Biosciences, USA) following the manufacturer's protocol [17, 24]. From five randomly selected fields, the cells that had invaded through the membrane to the lower surface were counted under a light microscope.

Thirty-six five-to six-week-old female Balb/c athymic nu/nu mice were ordered from the Animal Center of Yang Zhou University of China and were raised at the Experimental Animal Center, Southeast University. All the animal experiments were performed in compliance with the guidelines of the Animal Research Ethics Board of Southeast University. The 36 mice were randomly divided into six groups of equal size (6) as follows: group 1 for CD44⁺CD117⁺CSCs with lentivirus miR-200c; group 2 for CD44⁺CD117⁺CSCs with lentivirus mock; group 3 for CD44⁺CD117⁺CSCs without lentivirus infection; group 4 for CD44⁺CD117⁺CSCs transfected with the shZEB1; group 5 for CD44⁺CD117⁺CSCs transfected with the scrambled control siRNA; group 6 for CD44⁺CD117⁺CSCs without transfection. Each mouse was subcutaneously (s.c.) injected in the back with 5 × 10⁴ cells for the group it was in. Tumor formations in the mice were monitored every three days. Evaluation was also done of the tumor volume, tumor-free mice, and survival rates, respectively. A mouse was sacrificed when any of its tumors was over 1.4 cm in the largest diameter. The necropsy was performed on each animal for further analysis of the primary tumor along with possible metastases [18, 25].

Total cell lysates were prepared and analyzed by using the Western blot method as described before [26, 27]. Briefly, 1 × 10⁶ CD44⁺CD117⁺CSCs were collected and lyzed in a protein extraction buffer according to the manufacturer’ s protocol. The PVDF membrane was blocked with 4% dry milk in the Tris-buffered saline with Tween-20 for 1 h at 20°C, and was incubated with the rabbit antibody specific to human ZEB1 (Santa cruz Biotechnology, CA, USA ) or with E-cadherin or with Vimentin (Bioworld Technology, USA) for overnight at 4°C. The membrane was then incubated with the anti-rabbit fluorescence secondary antibody for 1 h at 20°C. Immunoreactive bands were detected by the Odyssey scanning instrument (LICOR Odyssey, USA ).

4 μm-thin formalin fixed and paraffin-embedded slides were incubated with the rabbit antibody specific to human E-cadherin overnight at 4°C. The samples were then labeled with HRP-conjugated streptavidin (Invitrogen) and the chromogenic reaction that was developed using Liquid DAB Substrate Pack according to the manufacturer's instructions. E-cadherin-stained cells from random and non-overlapping fields were counted under a magnification of × 200 [28].

The lung tissues were removed from the mice and fixed in 10% formalin, and then embedded in paraffin. Tissue sections of 4 μm thin were cut and mounted on SuperFrost Plus glass slides, fixed in methanol, and stained in hematoxylin and eosin (HE) [29, 30].

Statistical comparisons of the results between groups were performed using the Student’s t-test method. P < 0.05 was considered significant statistically.

As was described in the method section, the CD44⁺CD117⁺CSCs were isolated from the SKOV-3 cell line using MACS and were identified for cell phenotype using FCM to study the miR-200c overexpression in CD44⁺CD117⁺CSCs. Figure 1A shows that the CD44 and CD117 double-positive cells accounted for 3.1% of the SKOV-3 cell line, and were labeled ‘CD44⁺CD117⁺CSCs’ as was done before [15]. Next, we detected the miR-200c expression in the CD44⁺CD117⁺CSCs and in the lentivirus miR-200c transduced CD44⁺CD117⁺CSCs after the cells were stably selected using a single clone screening method. Figure 1B indicates that the miR-200c expression analyzed by qRT-PCR was markedly lower in the CD44⁺CD117⁺CSCs than in the non CD44⁺CD117⁺CSCs. The difference was statistically significant (p < 0.01). Morphologically, the CD44⁺CD117⁺CSCs transduced with the lentivirus miR-200c appeared to be fusiform-shaped and closely connected clones in the media, whereas the CD44⁺CD117⁺CSCs transduced with the lentivirus mock seemed to have a looser and more dispersed structure (Figure 1C). Figure 1D presents the results of the miR-200c expression analyzed by qRT-PCR. The CD44⁺CD117⁺CSCs transduced with the lentivirus miR-200c (the rightmost bar) exhibited a significantly higher level of miR-200c overexpression than the CD44⁺CD117⁺CSCs transduced with the lentivirus mock (18 ± 5 vs 3 ± 1, P < 0.01) or the CD44⁺CD117⁺CSCs (18 ± 5 vs 2 ± 1, P < 0.01). Figure 1E presents the E-cadherin expression (brown cells) stained with immunohistochemistry. The results show that more brown cells were found in miR-200c-CD44⁺CD117⁺CSCs of tumor tissue than the control cells.

To characterize the function of miR-200c in the CD44⁺CD117⁺CSCs, we examined the effects of miR-200c overexpression on the CD44⁺CD117⁺CSCs with regard to colony forming, cell migration, cell invasive ability, and cell proliferation ability, respectively. The colony forming capability was analyzed by the plate colony forming assay. The plating colony formation rates were about 60% and 50% for the CD44⁺CD117⁺CSCs and the CD44⁺CD117⁺CSCs transduced with lentivirus mock, respectively; the colony formation rates were less than 20% for the CD44⁺CD117⁺CSCs transduced with the lentivirus miR-200c (Figure 2A, 2D). The cell migration ability was assessed with the wound healing assay; the results are displayed in the pictures in Figure 2B. The overexpression of miR-200c clearly resulted in a significant reduction in cell migration in comparison the control cells (CD44⁺CD117⁺CSCs with lentivirus mock and CD44⁺CD117⁺CSCs without lentivirus infection); the differences were statistically significant (Figure 2E). The cell invasive ability was studied using the transwell invasive assay. The overexpression of miR-200c resulted in fewer CD44⁺CD117⁺CSCs with lentivirus miR-200c (mean ± SD: 70.81% ± 2.16%), in the bottom of the chamber insert, than in the CD44⁺CD117⁺CSCs with lentivirus miR-200c compared with the CD44⁺CD117⁺CSCs with lentivirus mock (125.92% ± 2.14%), or than the CD44⁺CD117⁺CSCs without lentivirus infection (162.26% ± 6.78%) (Figure 2C, 2F). The differences were statistically significant. Figure 2G shows that the proliferation ability of CD44⁺CD117⁺CSCs with lentivirus miR-200c detected by MTT assay was slower than the control cells.

Because the miR-200c overexpression exhibited significant effects on the colony forming and on the migratory and invasion of CD44⁺CD117⁺CSCs in vitro, we sought to determine whether the miR-200c overexpression could affect the establishment and progression of ovarian cancer in vivo nude mouse model. Figure 3A shows that the growth curves of the tumors in the mice injected with the aforementioned CSCs. The images of the tumors in Figure 3B were taken from the mice injected with the different CSCs when the tumor tissues were dissected on Day 56. Figure 3C indicates the survival time of the tumor-bearing mice. The mice injected with the 5 × 10⁴ CD44⁺CD117⁺CSCs with lentivirus miR-200c showed much higher tumor free rates than the mice injected with the 5 × 10⁴ CD44⁺CD117⁺CSCs with lentivirus mock or 5 × 10⁴ CD44⁺CD117⁺CSCs without lentivirus infection on different days after the injection. The differences were statistically significant (p < 0.01 in both comparisons). The results suggested that the stable miR-200c overexpression in established tumors delayed tumor progression.

It is known that the feedback loop model links ZEB1 to miR-200c in melanoma and breast cancer cells, and that ZEB1 and miR-200c repress each other in the loop that impacts the change in EMT-MET[10, 23]. Therefore, we wanted to find out whether the miR-200c overexpression would also impact the ZEB1 expression in the tumors of the nude mice that were injected with the CD44⁺CD117⁺CSCs with lentivirus miR-200c. The results revealed that the miR-200c overexpression in the tumors led to a marked reduction of ZEB1 mRNA (Figure 4A), Vimentin mRNA (Figure 4B) and the protein expressions (Figure 4D) compared with the CD44⁺CD117⁺CSCs with lentivirus mock and with the CD44⁺CD117⁺CSCs without lentivirus infection. Figures of 4E to 4G present the quantities of the molecular expression from the gradation scanning analysis. The miR-200c overexpression in the tumors significantly increased the E-cadherin expression (Figure 4C, 4F). Because the stable miR-200c overexpression in the established tumors delayed tumor progression and extended the survival of the tumor-bearing mice, we wanted to find out if the miR-200c overexpression would inhibit tumor metastasis. Figure 4H shows that the tumor lung metastasis was significantly reduced in the mice injected with the CD44⁺CD117⁺CSCs with lentivirus miR-200c in comparison with the mice of the two control groups (Figure 4I).

Figure 5A displays the results from our experiment that investigated the effects of the down-regulated ZEB1 expression on the CD44⁺CD117⁺CSCs. The shZEB1 CD44⁺CD117⁺CSCs significantly decreased the colony forming rates compared with the CD44⁺CD117⁺CSCs and the scrambled CD44⁺CD117⁺CSCs, respectively; Figure 5D indicates that the differences were statistically significant (P < 0.01). Further, we evaluated the cell migration ability in the shZEB1 CD44⁺CD117⁺CSCs. Figure 5B gives the representative images of the cell migration in the flat plate. The shZEB1 CD44⁺CD117⁺CSCs obviously slowed the cell migration due to the scratchy ‘wounds’ at the cell edges. The results from the quantitative analysis at 48h showed a statistically significant reduction in the wound closures in the shZEB1 CD44⁺CD117⁺CSCs compared with the CD44⁺CD117⁺CSCs and the scrambled CD44⁺CD117⁺CSCs, respectively. Figure 5C illustrates the cell invasive ability of the shZEB1 CD44⁺CD117⁺CSCs as was measured by the transwell invasive assay; the differences between the shZEB1 CD44⁺CD117⁺CSCs group and the other two groups were both statistically significant (P < 0.05 and P < 0.01 as shown in Figure 5F).

In light of the observed effects of the decreased ZEB1 expression on colony forming, cell migration and invasion in the shZEB1 CD44⁺CD117⁺CSCs in vitro, we further validated whether the effects would impact the tumorigenicity and metastatic potential of the shZEB1 CD44⁺CD117⁺CSCs in vivo. Figure 6A shows that all the 6 mice in this group developed tumors in 20 days after the injection of the 5 × 10⁴ CD44⁺CD117⁺CSCs; 5 of the 6 mice in the group receiving the 1 × 10⁶ scrambled cells developed tumors in 26 days after the injection. In contrast, only 3 of the 6 mice injected with the 5 × 10⁴ shZEB1 CD44⁺ CD117⁺CSCs developed tumor on Day 22, Day 26 and Day 28, respectively; the other 3 mice did not grow tumors throughout the 56-day observation period (Figure 6B). Figure 6C displays the photos of the tumor sizes and quantity in the mice when the resultant xenograft tumors were harvested.

To characterize the function of the downregulation of ZEB1, we examined the expression of ZEB1, E-cadherin and Vimentin, respectively, in the tumor tissues of the EOC-bearing nude mice. The Western blot results indicated that the expression of ZEB1 and Vimentin was much lower, and the E-Cadherin expression was much higher in the mice injected with the shZEB1 CD44⁺CD117⁺CSCs than in the mice injected with the CD44⁺CD117⁺CSCs or with the scrambled CD44⁺CD117⁺CSCs (Figure 6D). Figures 6E to 5G show the quantities of the molecular expression from the gradation scanning analysis.

To assess the effect of the down-regulation of ZEB1 on tumor metastasis, we performed H&E staining of lung tissue sections. Compared with the mice of the two control groups, a few tumor metastasis were found in the lungs of the nude mice injected with the shZEB1 CD44⁺CD117⁺CSCs (Figure 6H and 6I). It is thus evident from the results that the EMT-phenotype tumor cell growth and metastasis were significantly inhibited in the mice injected with the shZEB1 CD44⁺CD117⁺CSCs compared with the mice injected with the CD44⁺CD117⁺CSCs or the scrambled CD44⁺CD117⁺CSCs.