Here, we describe a novel approach to model normal primary fallopian tube secretory epithelial cells (FTSECs) in an in vitro three-dimensional (3D) spheroid system. Culturing FTSECs as spheroids restores the 3D architecture of the tissue in vivo, as well as gradients of nutrients, oxygen, carbon dioxide and other macromolecules. We observed molecular and cellular features of FTSECs cultured in 3D more closely resembled fresh FTSEC tissue samples than monolayer cultured fallopian tube secretory epithelial cells. One striking change associated with the transition to 3D was the reduced proliferation rate of cells in 3D compared to 2D, as demonstrated by MIB1 and p53 staining. Cells in 3D were less proliferative which was also reflected in the changing patterns of gene expression following transition from 2D to 3D. This is consistent with a previous study of normal ovarian surface epithelial cells cultured in 3D cultures , and is also true for normal breast cells . Since proliferation of the fallopian tube mucosa occurs in pre-malignant or malignant lesions , these data suggest that these 3D models more closely reflect the quiescent status of normal FTSECs in vivo and are more biologically relevant models of normal FTSECs than 2D monolayers for studying normal fallopian tube biology and tumorigenesis. Furthermore, 3D culturing enhanced the production of secretory products by FTSECs. Oviduct specific glycoprotein 1 (OVGP1), also known as mucin 9, is normally secreted by non-cilated tubal epithelia and improves in vitro fertilization rates by reducing polyspermy and increasing blastocyst formation rates [21, 22]. We found OVGP1 to be upregulated ~2-4 fold in FTSECs cultured in 3D. Similarly, a second glycoprotein, pregnancy associated plasma protein A (PAPPA) was also significantly upregulated in 3D. Increased expression of these bioactive glycoprotein molecules suggests FTSECs grown in 3D have enhanced functional differentiation compared to their 2D counterparts.
We compared global expression profiles of 2D and 3D cultured cells with biomarker expression in primary fresh fallopian tube tissue samples. We showed that gene profiles in 2D cultured cells cluster with follicular phase fallopian epithelial tissue, whereas 3D cultured cells cluster with luteal phase fallopian tube samples. This result may also be driven by the proliferative signature of the 2D cultured cells, as the follicular phase of the menstrual cycle is the proliferative phase, when raised levels of estradiol stimulate proliferation of the epithelia lining the endometrium and fallopian tube . We found that gene expression profiles of 3D cultured FTSECs cluster with those of luteal phase fallopian tube tissues. This phase of the cell cycle is the secretory phase, which may indicate a commitment to secretory differentiation FTSECs cultured in 3D. Consistent with this, we observed upregulation of an secreted proteins as well as an FTSEC marker (PAX8) when one FTSEC line was cultured in 3D. These data strongly suggest that culturing in 3D enhances functional differentiation of FTSECs to a secretory phenotype.
Previous studies have reported culture of human fallopian tube epithelia ex vivo, on collagen gel and alginate matrices [4, 5]. These models have significantly advanced our ability to model human and murine polarized fallopian tube epithelia in vitro. However, one limitation of ex vivo models is the restricted ability to sub-culture the cells. Using a growth factor rich media we were able to subculture the fallopian tube epithelial cells we isolated. We then selected a spheroid culture method to establish 3D cultures because this approach offers flexibility for downstream molecular analysis, and can be scaled up or down to perform high-throughput molecular screening or large-scale mass cultures. Although we did not supply matrix proteins in the cultures, fallopian tube secretory epithelial cells produced a matrix of which laminin was a major component. Laminin is the major protein in the basal lamina, the aspect of the basement membrane to which epithelial cells are adhered in vivo via integrin-mediated interactions. We hypothesize that altered cell-matrix interactions may contribute to the altered gene expression patterns we observed. While the 3D FTSEC cultures presented here do not recreate the complex convoluted architecture of the lumen of a fallopian tube in vivo, in FTSEC spheroids the epithelial cell-basement membrane interaction is restored. We observed that the outer surface of the spheroid is reminiscent of the lumen of the fallopian tube in that cells are in contact with other mucosal epithelia throughout the lateral domains of the cell, and basal domains of the cells are in contact with a basement membrane-type matrix. In contrast, cells trapped within the spheroid cores are surrounded by matrix, which is an ectopic microenvironment for normal epithelial cells. We hypothesize that this may induce programmed cell death, resulting in the high frequency of apoptotic cell debris observed within the cores but not at the periphery of FTSEC spheroids. Alternatively the physiological conditions within spheroids could have contributed to the changes survival of cells at the centre of the multicellular aggregates, since mathematical modeling suggests that deficiencies in ATP , glucose, hydrogen and oxygen  may all induce necrotic cell death of cells within spheroid cores.
Many key cellular processes are now known to be differently regulated between 2D and 3D cultures, and various factors can induce differential gene expression in 3D, including altered cell-cell and/or cell-matrix communications, nutrient and oxygen gradients, and reduced rates of proliferation. We propose that the 3D models are more biologically relevant tools of FTSECs than traditional 2D monolayers with which to study fallopian tube epithelial cell biology and pathogenesis. Perhaps the greatest potential for clinical impact of these models will come from their use in studies of tumor initiation. This has become particularly significant since it was established recently that the epithelia lining of the fallopian tube likely represents the cell of origin for a proportion of HGSOCs. HGSOCs bear morphological resemblance to Müllerian epithelia and over 80% of this tumor type overexpress PAX8 , an FTSEC marker that can be used to distinguish ovarian serous tumors from other, morphologically similar neoplasms [27, 28]. We identified additional FTSEC biomarkers that represent novel candidate HGSOC biomarkers. These include LRRK2, a gene that encodes a kinase involved in Parkinsons Disease [29, 30]. LRRK2 has not previously been implicated in ovarian cancer development but analyses of The Cancer Genome Atlas (TCGA) data suggests ~3% of primary HGSOCs harbor somatic mutations in this gene [31–33]. Other novel FTSEC biomarkers that are overexpressed in HGSOCs include CELSR3, an atypical cadherin; ABCC3, an ABC transport protein implicated in drug resistance ; and CTHRC1, a secreted protein shown to be a candidate biomarker for breast and pancreatic cancer [35, 36]. Analyses of primary HGSOC specimens and sera collected from ovarian cancer patients will be required to determine whether any of these novel biomarkers have clinical utility in the early detection of HGSOC.
While it is now widely accepted that a proportion of HGSOCS originate in the fallopian tube, the early stages of disease development are poorly understood and many questions remain to be answered. Reports show differences in the proportions of ciliated and secretory epithelial cells, marker expression and hormone responsiveness between the epithelia found in fimbrial and ampullary regions of the fallopian tube [23, 37, 38]. However, as yet we do not yet know why FTSECs in the fimbrial region of the fallopian tube are more prone to neoplastic transformation. One hypothesis is that the proximity to the mitogenic environment of the ovarian stroma may influence the phenotype of fimbrial FTSECs. Alternatively the region of transition between FTSECs and ovarian mesothelial-type epithelial cells is inherently more prone to neoplastic transformation. In the future, these 3D models of FTSEC transformation that incorporate common somatic genetic alterations characteristic of HGSOC (BRCA1, p53 ) or even recently discovered susceptibility alleles that confer low-risk of EOC in the general population [39–45] will be vital tools in answering some of the key questions regarding EOC initiation and development.