This study showed that a CD133+ sphere-forming subpopulation of OVCAR3 cells had a distinct microRNA expression profile compared with that of OVCAR3 cells grown in adherent culture conditions. These findings might contribute to understanding miRNA-driven pathways related to chemoresistance in ovarian cancer.
Initially, we tested whether various ovarian cancer cell lines can make tumor sphere. Interestingly, tumor spheres were observed in OVCAR3, SKOV3, and TOV112D cells. Previous studies revealed that a small subpopulation of CSCs exists within tumor spheres [13
]. MTT assay showed that tumor spheres from the OVCAR3 and SKOV3 cells were resistant to paclitaxel. Given these results, we may speculate that tumor spheres may contain an increased proportion of CSCs. On the other hand, there is a possibility that these findings could also be partly explained by the phenomenon of a worse drug penetration into the inner layer of the spheroid as compared to monolayer cultures. Therefore, we tested whether spheroid-forming cells express stem cell-associated surface markers. Reproducible isolation using distinct cell surface antigens is now regarded as a required characteristic of CSCs [14
]. Importantly, enrichment of the cell surface marker CD133+
was observed in tumor spheres, which shows that spheroids contain more CSCs than adherent cultures. However, CD44+
cells were initially high in the SKOV3 cell line in the conventional culture system. Curley and colleagues showed a relatively high tumorigenic potential of CD133+
cells derived from primary human ovarian tumors in an in vivo
serial transplantation model [7
]. Likewise, Ferrandina and colleagues reported that CD133+
ovarian tumor cells exhibited higher clonogenic efficiency compared with that of CD133-
]. Therefore, these results suggest that CD133 may be a good marker for identifying ovarian CSC populations in vitro
. Then, for miRNA expression analysis, we picked the OVCAR3 cell line because the CD133+
subpopulation was the most efficiently enriched in OVCAR3 cells with the sphere culture system.
Then, the expression of genes specific to embryonic stem cells was examined in spheroids. Poorly differentiated tumors exhibit over-expression of genes that are normally enriched in embryonic stem cells [16
]. We measured the expressions of the Oct-4, Sox-2, and Nanog using RT-PCR and Western blot analysis. Intriguingly, the over-expressions of stem cell-related genes were noted in CD133+
spheroid-forming populations of OVCAR3 cells, but not in a CD133-
spheroid-forming subpopulation or in cancer cells grown in adherent culture conditions. These results suggest that only CD133+
spheroid-forming cells are associated with the expressions of stem cell-related genes and presumably have stem cell-like properties.
Therefore, we sorted CD133+
cells from tumor spheres using a FACSAria flow cytometer for miRNA analysis. In this study, we identified 37 miRNAs that were differentially expressed in CD133+
spheroid-forming cells compared with the levels in cancer cells in conventional adherent culture conditions of OVCAR3. These miRNAs may be associated with stem cell-like characteristics of ovarian cancer stem cells. While our manuscript was under review, a study of miRNA expression profiling the CD133+
subpopulation of OVCAR3 was published by Guo and colleagues [12
]. They found that the expressions of 40 miRNAs were
two-fold higher and those of 112 miRNAs were
two-fold lower in CD133+
cells compared to those in CD133-
cells. They showed up-regulation of the expressions of miR-204
and down-regulation of miR-9, miR-100, miR-223,
cells using RT PCR. However, their results were not consistent with our miRNA expression profiling results.
We speculate that this contradiction may be explained by differences in the methods used to expand the CSC population prior to sorting. We enriched CD133+
cells using a tumor sphere assay; however, Guo and colleagues sorted CD133 cells using FACS only. Differences in culture conditions may have also affected miRNA microarray results. Ince and colleagues used differential culture methods to isolate two separate cell populations from the same initial source [17
]. The two populations had differential tumorigenic potential following injection into immunocompromised mice and generated histologically different tumors with distinct metastatic potential. Taken together, these data suggest that the characteristics of putative CSCs may vary depending on the techniques used to isolate the CSCs. Indeed, it is becoming evident that more stringent and clear methodologies for the isolation of CSCs should be developed.
The limitation of this study is that only one cell line-OVCAR3 was used to analyze miRNA expression patterns for CD133+ spheroid forming subpopulations. Therefore, we should note that differential expression patterns of miRNAs in CSCs of OVCAR3 cells in this study are not applicable for ovarian cancer in general. However, it is worth noting the reason why only one cell line was chosen for miRNA analysis. During this work, we found that various cell lines from single organ ovary could have differential stem cell characteristics.
In the present study, miR-205
was present at a level of approximately 10-fold higher in a CD133+
spheroid-forming subpopulation of OVCAR3 cells compared with the levels in cells in the adherent culture condition. Recent studies have demonstrated that miR-205
has a role in both normal and cancer development, but these results are controversial. Specifically, the expression of miR-205
is significantly suppressed in ovarian cancer [18
], whereas miR-205
is significantly up-regulated in bladder cancer [19
]. Inhibition of the Src family can cause cell cycle arrest and growth suppression of ovarian cancer cells, which may paradoxically result in chemo-resistant characteristics of ovarian CSCs. Additionally, miR-205
was found to be highly expressed in stem cell-rich populations and thus may have a function in normal mammary stem cell maintenance [20
]. However, down-regulation of miR-205 was noted in prostate cancer cell lines resistant to chemotherapy [21
was second most differentially expressed (7.33 fold) in CD133+
spheroid-forming subpopulations of OVCAR3 cells. This miRNA is mainly expressed in primitive hematopoietic stem cells and T lymphocytes [22
]. Reduced expression of miR-146a
in pancreatic cancer cells was observed compared with the level in normal human pancreatic duct epithelial cells [23
]. On the other hand, higher expression of miR-146a
in NK/T cell lymphoma demonstrated better prognosis [24
]. Taken together, further studies are needed to validate whether these miRNAs are associated with chemo-resistance and regulation of CSC function.