Our deep proteomic analysis of benign fallopian tube tissues as well as low (LGSOC) and high (HGSOC) grade serous ovarian cancers shows these samples can be stratified by variably abundant protein alterations suggesting distinct proteome abundance profiles are characteristic of these tissue types. Differential analysis of tumor (LGSOC and HGSOC) and benign fallopian tube tissue identified that proteins elevated in fallopian tubes were largely enriched for pathways regulating cilium development. Recent evidence from single cell studies has identified several subpopulations of epithelial cell types exist within benign fallopian tube, including ciliated cell types [14, 15]. This cell population is considered to reflect a more differentiated subpopulation of fallopian tube epithelial cells emerging from non-ciliated, secretory epithelial cell types [14, 15]. The loss of proteins regulating cilium development that we observe in LGSOC and HGSOC tissues is consistent with evidence suggesting secretory epithelial cell types often exhibit greater neoplastic potential than other cell types in fallopian tube epithelium [14]. We also find that proteins elevated in both LGSOC and HGSOC tumors exhibit enrichment of pathways regulating protein translation, specifically ribosome and translational elongation, as well as amino acid metabolism in comparison with benign fallopian tube. Increased ribosome biogenesis and protein translation has been described as a characteristic of ovarian cancer cells and as a possible therapeutic vulnerability in both LGSOC and HGSOC [16, 17]. Proteins uniquely elevated in HGSOC tumors compared to benign fallopian tube were largely enriched for pathways regulating cell cycle and mitosis, which is consistent with the high-grade phenotype of this disease [18]. Proteins uniquely elevated in LGSOC were enriched for pathways regulating epithelial to mesenchymal transition including extracellular matrix organization, beta1 integrin cell surface interactions, and hematopoietic cell-related signaling, i.e. complement and coagulation. Beta1 integrin signaling has been implicated in regulating migration of ovarian cancer cells in the context of interactions with extracellular matrix proteins [19]. Our comparison with fallopian tube tissues has identified proteome alterations largely consistent with the malignant phenotypes of LGSOC and HGSOC tumors in comparison to benign fallopian tube.

Comparison of LGSOC and HGSOC revealed marked protein alterations between these disease subtypes, including elevation of proteins correlating with proliferation and DNA replication in HGSOC. We also observed elevation of diverse extracellular matrix proteins in LGSOC tumors, including multiple collagen isoforms which is noteworthy considering the enrichment of pathways regulating intercellular cell adhesion [20] we also observe in these tumors. Recent global proteome analysis of tumors collected from women diagnosed with LGSOC or HGSOC have been described by our team [6, 9] and others [7]and we were motivated to compare observed proteome alterations between LGSOC and HGSOC tumors from this study with these independent cohorts to identify and validate proteome alterations that are highly conserved between these disease subtypes. This comparison resulted in the identification of highly conserved protein alterations exhibiting high quantitative correlation between LGSOC and HGSOC tumors (Spearman Rho ≥ 0.82). These features demonstrated that proteins elevated in HGSOC tumors correlate with enrichment of pathways regulating cell cycle and cell mitosis signaling, while those elevated in LGSOC tumors correlated with pathways regulating intercellular adhesion and defective cellular apoptosis. We also investigated putative drug targets exhibiting large differences in abundance between LGSOC and HGSOC tumors and, among top targets, we identify cluster of differentiation 73 (NT5E/CD73) as elevated in LGSOC versus HGSOC tumors. CD73 has been shown to participate in an immune-related checkpoint pathway that promotes conversion of exogenous adenosine triphosphate to adenosine by tumor cells which can suppress immune cell function [21, 22]. CD73 expression has been shown to correlate with poor disease outcome in HGSOC and to be markedly elevated in HGSOC tumors that are enriched for a stromal-expression signature also correlated with poor disease outcome [23]. Recent efforts investigating the feasibility to target CD73 with a monoclonal antibody, i.e. Oleclumab, and the immune checkpoint inhibitor, i.e. Durvalumab, in recurrent ovarian cancer have found this combination to be well tolerated [22]. LGSOC has been recently described as exhibiting an immunosuppressive tumor microenvironment, particularly in the absence of mutations in the proto-oncogene B-Raf (BRAF) or Kirsten rat sarcoma virus (KRAS) [24]. Our findings that CD73 is significantly elevated in LGSOC versus HGSOC tumors warrants additional investigation of the role of this protein in regulating the immune microenvironment of LGSOC tumors and will be the focus of future research.

Additional investigation of putatively conserved protein alterations between LGSOC and HGSOC tumors identified mucin-16 (MUC16)/cancer antigen 125 (CA125) as significantly elevated in LGSOC versus HGSOC tumor tissues. MUC16 is a transmembrane glycoprotein that is highly abundant in the female reproductive tract and can be actively shed from ovarian cancer cells into blood plasma [8]. Circulating MUC16/CA125 protein levels thus represents a measurable biomarker that can assess evidence of disease and is a facet of routine diagnosis and disease monitoring in the clinical management of ovarian cancer [3, 8]. Comparison of the putative MUC16 interactome and highly conserved protein alterations observed between LGSOC and HGSOC tumors identified nesprin-1 (SYNE1) as elevated in LGSOC tumors. SYNE1 and MUC16 have been observed as commonly mutated gene targets in stomach cancer and tumors with mutations in these genes have been shown to exhibit decreased immune cells as well as to correlate with lower response to immunotherapy [25]. Shedding of MUC16 has been shown to occur as a product of proteolytic cleavage of the carboxy-terminal ectodomain in the acidifying Golgi/post-Golgi [26], possibly by membrane-type 1 matrix metalloproteinase [27] followed by release of the N-terminal CA125 region [27] into interstitial fluid and ultimately into circulating blood [8]. As we quantify membrane-type 1 matrix metalloproteinase (MT1-MMP/MMP-14/P50281) protein in our discovery cohort, we compared levels between LGSOC and HGSOC tumors, but we do not observe this protein as significantly altered between these disease subtypes (MWU p = 0.48). Recent evidence investigating mechanisms regulating cell surface retention of MUC16 have implicated N-glycosylation mediated by N-acetylglucosaminyltransferase I (MGAT1) which has been shown to promote increased retention of MUC16 at the cell surface [28]. This study further identified that N-glycosylation of MUC16 promotes interaction with the protein galectin-3 (LGALS3) at the cell surface [28]. Notably, we identify LGALS3 as significantly elevated in LGSOC versus HGSOC tumor tissues (p < 0.05 across all cohorts assessed) and, although we quantify MGAT1, we see this protein is significantly elevated in HGSOC tumors in our discovery cohort (Supplementary Table 7), but we do not see this protein as significantly altered between LGSOC versus HGSOC tissues in validation cohorts. If we consider this evidence in the context of our findings, this suggests LGSOC tumor cells perhaps exhibit increased retention of MUC16 at the cell surface. Circulating CA125 levels have been measured in both LGSOC and HGSOC patients, although it has been noted that these levels are more commonly lower in LGSOC patients [3, 29]. We also find that circulating CA125 levels were lower in our LGSOC versus HGSOC patients (Table 1), providing further evidence in support of this hypothesis. A possible mechanism underlying altered dynamics of MUC16 retention may be due to differential N-glycosylation of MUC16 in LGSOC tumor cells, possibly due to altered enzymatic activity of MGAT1. Alternatively, differences in MUC16 expression and shedding could relate to greater cleavage of the extracellular domain in HGSOC, accounting for the greater shedding of CA125 and decreased membrane expression. Future studies will investigate these hypotheses, including the comparison of cleaved versus non-cleaved MUC16 levels in LGSOC and HGSOC tumors.

We also observe that LGSOC tumor cells exhibit a predominantly apical staining pattern for MUC16 in comparison to HGSOC tumor cells. Similar findings were observed in murine ovarian carcinomas where more intense apical expression of MUC16 was observed in grade 1 tumors [30]. Recent evidence has shown that MUC16 is maintained within apical membranes of epithelial cells, such as in ocular epithelium [31]and to be dependent on glycosylation [28]which may explain why we observe increased apical staining patterns for MUC16 in LGSOC tissues. Beyond its role as a biomarker, MUC16 has been shown to promote cellular proliferation, invasion of ovarian cancer cells [32]and has previously been shown to discourage immune surveillance, and to be a potential target for immunotherapy [33]. Future efforts will focus on investigating mechanisms of membrane retention of MUC16 in LGSOC and improving our understanding of how this characteristic impacts LGSOC tumor cell biology.

Limitations for this study include small sample sizes for normal fallopian tube, LGSOC, and HGSOC tissues in our discovery cohort. We aimed to address this limitation by prioritizing protein alterations between LGSOC and HGSOC tumors that are also validated within global proteomic data generated for independent cohorts of LGSOC and HGSOC patients. Another limitation may be that samples were obtained by formalin-fixed, paraffin-embedded tissue instead of fresh frozen tissue, however, comparison of fresh tissue samples and fixed frozen paraffin embedded samples has not demonstrated differences in peptide identification and quantification [34]. Strengths of this study include the foundation provided for future proteomic research and the contributions of our findings towards improving our understanding of molecular alterations underlying ovarian cancer pathogenesis.