We have found that over 2000 genes are significantly differentially expressed between OSE and CEPI. Many of these genes are known to be involved in the canonical cell cycle pathway, as well as, signaling pathways previously implicated in development (i.e., the TGFB/BMP, WNT, NOTCH, Hedgehog and Retinoid pathways). The fact that many of the genes highly expressed in OSE have previously been associated with the maintenance of stem cells in a quiescent state is relevant to hypotheses on the cellular origin of ovarian cancer.
Under the dedifferentiation hypothesis of cancer development, cancer cells are postulated to be less differentiated than their progenitor cells but often resemble their tissue of origin phenotypically . The fact that CEPI appear more rather than less differentiated than OSE and do not phenotypically resemble OSE has been offered as evidence that these cells are not the source of CEPI . An alternative hypothesis is that OSE are stem-cell like and maintain a degree of pluripotency sufficient to allow them to morphologically transform during the process of CEPI development. Our results indicate that many, if not all, of the OSE cells on the surface of the ovary are not terminally differentiated but arrested in a quiescent state characteristic of most adult stem cell populations . While our findings are consistent with the hypothesis that OSE may retain properties of relatively uncommitted pluripotent cells until undergoing neoplastic transformation, they do not preclude the possibility that at least some tumors currently classified as ovarian may arise from related, yet anatomically distinct sources such as cancer initiating cells embedded within the fallopian tube . Indeed, our finding that the OSE molecular profile so closely resembles that of previously identified somatic stem cells and cancer stem cells suggests that all ovarian cancer initiating cells, regardless of their proposed tissue of origin, will likely share many essential characteristics.
The processes by which stem cells self-renew and differentiate are accomplished by a combination of cellular division strategies known as symmetric and asymmetric division [17, 70]. Symmetric division gives rise to two identical daughter stem cells. In contrast, asymmetric division results in one stem cell and one progenitor cell with limited self-renewal potential. Our results, showing that LHX2 and LHX9 are expressed in OSE supports the notion that asymmetric cellular division is occurring in OSE. Progenitor cells can subsequently go through several rounds of cell division before terminating into a mature differentiated cell. Whether or not stem cells self-renew or differentiate is regulated by the microenvironment. A microenvironment that is conducive to stem cell self-renewal is referred to as a stem cell niche [16, 68]. Stem cell progeny that remain in a stem cell niche typically display arrested cell growth/replication and are described as being quiescent [68, 71]. In contrast, stem cell progeny that exit a stem cell niche typically enter a transient period of accelerated cell division resulting in large numbers of cells prior to terminal differentiation.
The above description of stem cells and the niches that control their division is relevant to OSE because during the period between ovulations, OSE are quiescent. While arrested cell growth and division are associated with terminal differentiation, at least some OSE must not terminally differentiate because they reactivate their cell cycle and proliferate in response to ovulation. Evidence recently has been presented showing that cells on the surface of the macaque ovary transition from quiescent to a replicating phenotype in response to ovulation . Similar phenomena have been previously observed in mice [73–75] and generally support the notion that ovulation temporarily disrupts the ovarian surface niche resulting in controlled proliferation.
Our results are consistent with the hypothesis that the surface of the ovary is an interovulatory or facultative stem cell niche and this suggests that all or many of the resting cells on the surface of the ovary are not terminally differentiated but arrested in a quiescent state characteristic of adult stem cell populations [68, 71]. We find that a variety of signaling molecules (including TGFB/BMP and TGFBR family members, antagonists of the WNT and hedgehog signaling pathways, as well as members of the retinoid signaling pathway) is expressed at high levels in OSE. The fact that these molecules previously have been shown to be integral for the maintenance of stem cell niches in other organ systems [31, 35, 71, 76, 77] indicates that they are likely performing a similar function on the surface of the ovary between ovulations. Also consistent with the hypothesis that the surface of the ovary is a type of facultative stem cell niche is our finding that transcription factors previously implicated in self-renewal and asymmetric division are expressed in OSE [64, 78]. In contrast, genes expressed in CEPI have been associated with the entry and progression of cells through the cell cycle [79–84]. Of particular note, is the high expression of the Cyclin E family genes (CCNE) in CEPI. It has recently been shown that the constitutive expression of CCNE in mouse embryonic stem cells results in an almost nonexistent G0/G1 phase of the cell cycle . The lack of an extended G0/G1 results in less time for cells to respond to mitogens that stimulate cellular differentiation. Thus, the elevated expression of CCNE in CEPI may contribute to cancer growth in a similar fashion.
The complex molecular processes underlying the onset and development of epithelial ovarian cancer is only beginning to be unraveled. Our results indicate that a number of key developmental pathways are involved in the establishment and development of ovarian cancer. While many of these pathways have previously been either directly or indirectly implicated in ovarian cancer, detailed network analyses of our gene expression data led to the identification of linkages between these pathways attributable to the altered expression of key regulatory genes. We believe that the type of detailed network analyses of gene expression data presented in this paper when combined with next generation sequencing for mutation analyses of individual ovarian adenocarcinoma genomes will help expand our understanding of the origins of ovarian cancer and facilitate the development of more effective therapies.