Background
Ovarian cancer has the highest mortality rate among gynecological neoplasms. Of 314,000 women diagnosed worldwide in 2020, approximately 65% would die from the disease [
1]. Among the various histotypes, high-grade serous ovarian carcinoma (HGSOC) is associated with the poorest prognosis [
2].
Epithelial-mesenchymal transition (EMT) is a biological process associated with the development and progression of malignant tumors, involving cell motility and invasion, resistance to apoptotic stimuli, and acquisition of stemness. However, the precise connection between EMT and in vivo metastases remains unclear [
3].
Exosomes, which are nanometer-sized vesicles, serve as mediators of intercellular communication by transferring proteins, genes, chemokine receptors, soluble factors, miRNAs, and other bioactive material. Tumor-derived extracellular vesicles are believed to contribute to the development and progression of cancer through various mechanisms, including the activation of hematopoietic and stromal cells, stimulation of angiogenesis [
4], promotion of tumor dissemination, influence on the EMT process, remodeling of extracellular matrix, evasion of immune detection, and facilitation of the formation of pre-metastatic niches. Furthermore, exosomes hold promise as markers for the detection of metastases [
5].
The discovery of proteins that regulate neoplastic growth and metastasis holds the potential to yield new biomarkers for early detection and prognosis of ovarian cancer, as well as targets for treatment. In this study, our objective was to identify potential biomarkers present in exosomes derived from ovarian cancer cells during the EMT process. We achieved this by employing mass spectrometry to identify proteins and comparing the results with publicly available gene expression data.
Methods
Ovarian cancer cell culture
CAOV3 (ATCC® HTB-75™) and SKOV3 (ATCC® HTB-77™) cells lines were cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) (Thermo Scientific, Marietta, OH, USA). The OVCAR3 (ATCC® HTB-161™) cells were cultured in Roswell Park Memorial Institute Medium (RPMI-1640) (Gibco) supplemented with 20% Fetal Bovine Serum (FBS). Both media were further supplemented with 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco). The cells were maintained at 37◦C in a humidified incubator in an atmosphere of 5% CO
2. Before use, the cell lines were authenticated by short tandem repeat (STR) profiling and confirmed to be mycoplasma negative [
6]. All cell lines were acquired from the American Type Culture Collection (ATCC, Gaithersburg, MD).
EMT induction
The cell lines were initially seeded in the supplemented medium. After 24 h, the cell lines were washed twice with PBS, and the media was replaced with an FBS-free medium. Following additional 24-hour incubation, the cells were treated with an FBS-free medium containing 10 ng/mL Epidermal Growth Factor (EGF) (Cat#236-EG-200, R&D Systems, Minneapolis, Minnesota, USA). The EGF-containing medium was replenished every 24 h for a total duration of 96 h [
6].
The cells were washed with PBS twice and subsequently disrupted in lysis buffer (Cat#9803, Cell Signaling, Danvers, MA, USA) using three sonication cycles. Each sonication cycle lasted for 5 min with cooled water in an ultrasonic bath (Unique, São Paulo, SP, Brazil). Following sonication, the cell lysate was centrifuged at 20,000×g for 30 min at 4◦C. The supernatant containing the proteins of interest was collected, and the protein concentration was determined using the Bradford method (Bio-Rad, Hercules, CA). Finally, the samples were stored at -80◦C.
Western blotting
The proteins were separated by SDS–PAGE and then electrotransferred onto PVDF membranes (GE Lifesciences, Pittsburgh, PA, USA). Next, the membranes were incubated with blocking buffer (25 mM Tris-HCl (pH 7.5), 0.5 M NaCl, and 0.1% Tween-20) containing 5% non-fat dry milk. Primary antibodies were added to the membranes, followed by incubation with a secondary antibody, specifically horseradish peroxidase-conjugated goat anti-rabbit IgG (Cat#7074, Cell Signaling), as per the manufacturer’s instructions. The primary antibodies used are listed in Supplementary Table
1. The membranes were then developed using ECL Western Blotting detection reagents (GE Lifesciences) and images were captured using a CCD-Camera (ImageQuant LAS 4000 mini, Uppsala, Sweden). Dosimetric analyses were performed using ImageJ software. The values are presented as the EMT/CT ratio, normalized according to the constitutive protein glyceraldehyde 3-phosphate dehydrogenase (GAPDH). GAPDH was chosen as a loading control because it is involved in glycolysis, a fundamental metabolic pathway in most cells, and its expression levels tend to be relatively stable under many experimental conditions.
PathScan EGFR signaling array kit
The PathScan EGFR signaling array kit (Cat#12622, Cell Signaling) was utilized, which contains fixed antibodies specific to phosphorylated proteins in a chemiluminescent sandwich immunoassay format. Experiments were conducted following the manufacturer’s instructions. Densitometric analyses of the obtained data were performed using the protein array analyzer plugin for ImageJ software.
ROS detection by fluorescence assay
The production of reactive oxygen species (ROS) was assessed using the intracellular fluorogenic reagent CM-H2DCFDA (C6827, ThermoFisher Scientific) according to the manufacturer’s instructions. To serve as controls, cell lines were incubated with PMA (50 nM) for 1 h to induce ROS accumulation through PKC activation. For the experimental groups, CAOV3, SKOV3, and OVCAR3 (both CT and EGF-treated) cell lines were incubated with 5 µM CM-H2DCFDA for 1 h before analysis. The analysis was performed at 37 °C in a 5% CO2 atmosphere. ROS using a FACSCalibur cytometer (Becton-Dickinson), and the fluorescence was detected in the FL1/FL2 channel. The acquired data were analyzed using FlowJo software (Treestar, Inc).
Mitochondrial membrane potential (MMP) by fluorescence assay
The assessment of mitochondrial membrane potential was conducted using intracellular tetramethylrhodamine ethyl ester perchlorate (TMRE) (#13296, Cell Signaling), following the manufacturer’s instructions. To establish control, cell lines were incubated with carbonylcyanide 3-chlorophenylhydrazone (CCCP) (50 µM) at 37 °C for 15 min to disrupt the mitochondrial membrane potential. For the experimental groups, CAOV3, SKOV3, and OVCAR3 (both CT and EGF-treated) cell lines were incubated with 200 nM of TMRE for 20 min before analysis. The analysis was performed at 37 °C in a 5% CO2 atmosphere using a Varioskan LUX Multimode Microplate Reader. The reader settings included excitation around 550 nm and emission around 580 nm.
Isolation of exosomes
Following the EMT induction, the culture medium was collected and subjected to centrifugation at 300×g for 10 min, followed by an additional centrifugation step at 3000×g at 4oC for 30 min. The resulting supernatant was then filtered using a 0.22 μm syringe filter for decellularization. Subsequently, the filtered solution was concentrated to a minimum volume of approximately 8 mL using an Amicon ultrafiltration system with a 100-kDa cutoff (Millipore, Billerica, MA, USA). The exosomes were isolated and separated from the concentrated solution using the exoEasy Maxi kit (Qiagen; Valencia, CA) as per the manufacturer’s instructions. Once isolated and separated, the, collected exosomes were stored in a freezer − 80oC.
Quantification of exosomes by nanoparticle tracking analysis
The size, distribution, and quantification of the exosomes obtained from independent experiments were assessed using nanoparticle tracking analysis (NTA) performed on a NanoSight NS300 system (Malvern Instruments, Malvern, United Kingdom). The NTA data were analyzed using NanoSight software (version 3.2.16) [
7].
Characterization of exosomes by transmission electron microscopy (TEM)
Freshly isolated exosomes were fixed in phosphate buffer containing 3% glutaraldehyde (v/v) and 4% paraformaldehyde (v/v) at 4
oC for 2 h. After centrifugation at 16,500×g for 30 min, the exosomes were resuspended in PBS and applied onto Formvar/carbon-coated electron microscopy grids. The exosomes were then visualized by negative contrast using a transmission electron microscope JEM-100 CX II (Jeol) equipped with a digital camera Hamamatsu ORCA-HR at magnifications of 50000 ×, 100000 ×, and 200000 × [
7].
The proteins were extracted from exosomes by disrupting them using a lysis buffer (8 M urea, 150 mM Tris (pH 8.0), 0.5% Octyl β-D-glucopyranoside (O.G) (Cat#O9882, Sigma, St Louis, MO, USA), and protease and phosphatase inhibitors (MSSAFE, Sigma). The disruption process involved three sonication cycles, with each cycle lasting 5 min using cooled water in an ultrasonic bath (Unique, São Paulo, SP, Brazil). After sonication, the samples were centrifuged at 20,000×g for 30 min at 4◦C. The supernatant was collected and the protein concentration was determined by the BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA).
Proteomic analysis
The exosome proteins (5 µg) were first reduced with DTT (1:1 mg/mg) for 5 min at 95oC and alkylated (5:1 mg/mg). The proteins were separated by SDS-PAGE using 4–20% Mini-PROTEAN TGX Precast Protein Gels, Cat # 4561093, Bio-Rad). After SDS-PAGE, each gel lane was divided into 4 equal-sized pieces, and in situ digestion was performed individually for each piece.
For each gel slice, SDS and dye were removed using NH4HCO3 (50 mM) containing 50% acetonitrile (ACN), followed by a wash with pure ACN. The slices were dried in SpeedVac (Savant) and then rehydrated with 20 µL of NH4HCO3 (100 mM) containing 0.6 µg of trypsin (Promega) for approximately 30 min. The gel slices were then covered with sufficient NH4HCO3 (100 mM) and maintained at 37oC for 16–18 h for digestion.
After digestion, the peptides were extracted from the gel using an ACN gradient (50%, 70%, and 100%) containing 0.1% formic acid. The extracted peptides were successively transferred to a clean microtube and dried using a SpeedVac (Thermo Scientific). The dried samples were resolubilized in ACN 5% containing 0.1% formic acid and desalted using C18 Zip Tips™ columns (Sulpelco, Sigma) following the manufacturer’s instructions.
For capillary liquid chromatography, an equivalent of 1 µg digested proteins was injected onto a column with dimensions of 25 cm length x 100 μm internal diameter of column). The peptides were separated using a linear gradient of ACN (5 to 35%) containing 0.1% formic acid over a period of 90 min at a flow rate of 250 nL/min. The chromatographic system was coupled to a high-resolution mass spectrometer, Q-Exactive HF (Thermo Scientific). The mass spectrometer was operated with a capillary voltage of 3.2 kV, a capillary temperature of 200 °C, a resolution of 100,000, and an FT-target value of 1,000,000.
The spectra were acquired in a dependent mode, selecting the 15 most abundant ions with a + 2 or + 3 charge state in the range of 400 to 1600 m/z for HCD (Higher-energy C-trap dissociation) fragmentation and MS/MS analysis. To avoid peptide sequencing redundancy, a 45-second exclusion window was applied.
Proteomic data analysis
The six separate files generated from the LC-MS/MS analysis were processed using the Maxquant quantitative proteomics software tool [
8]. This software was used to handle and process the data resulting in the identification and the relative label-free quantification (LFQ) of the detected proteins [
9].
The search criteria used during the analysis were as follows: trypsin enzyme with a tolerance of two lost cleavages, mass error tolerance for precursor peptide was 6 ppm in the main search, mass tolerance of 20 ppm for fragment ions (MS/MS); and false positive rate (FDR) of 1% for both proteins and peptides.
The identified proteins were subjected to relative LFQ, where the normalized intensity profiles (LFQ intensity) were calculated. For paired comparisons, at least 1 peptide identified by MS/MS was required (LFQ minimum ratio count = 1). The normalized intensity values (LFQ intensity) obtained from the analysis were used for statistical analyses using Perseus software version 1.6.7.0 [
10].
Data visualization and gene ontology analysis
The list of quantified proteins that showed regulation (< 0.5 and > 1.9) from our large-scale proteomics study was subjected to further analysis using public databases and available free open-source software. For the identification, clustering, and visualization of protein-protein interaction networks, we utilized the database STRING (
http://string-db.org/). This allowed us to explore the functional relationships and interactions among the identified proteins.
To gain insights into the biological processes and functions associated with the identified proteins, we performed a Gene Ontology (G.O.) analysis using FunRich [
11]. FunRich enables us to annotate the proteins based on their involvement in specific biological processes, molecular functions, and cellular components. Additionally, we used FunRich to identify proteins known to be secreted by cell vesicles, providing us with valuable information about the potential role of exosomes in the context of our study [
11].
Integrative and statistical analysis
To prioritize target candidate proteins, we used gene expression data of 489 HGSOC available in The Cancer Genome Atlas (TCGA) database. The samples were clustered into the molecular subtypes immunoreactive (110), mesenchymal (108), proliferative (136), and differentiated (133) [
12], The gene expression data was accessed using the cBioPortal repository (
http://www.cbioportal.org/).
To integrate and analyze the data, we employed R 4.1.1 (The CRAN project,
https://www.r-project.org/) software. We utilized the singular value decomposition for principal component analysis (PCA) to identify major sources of variation in the gene expression data. This allowed us to gain insights into the overall patterns and relationships among the samples based on their gene expression profiles.
Furthermore, we employed a conditional inference tree algorithm to identify the main potential biomarkers associated with the different molecular subtypes of HGSOC. This approach enabled us to uncover key proteins that could serve as promising biomarkers for distinguishing between the subtypes and potentially providing important clinical insights.
Discussion
Our study revealed that EMT induction enhances the concentration of exosomes secreted by ovarian cancer cells, and these exosomes carry potential tumor markers associated with the mesenchymal HGSOC subtype. Through evaluating changes in the proteome of exosomes derived from EGF-stimulated CAOV3 cells and integrating the data with the mRNA expression profile of the HGSOC from the TCGA-OV cohort, we identified PLAU, LAMB1, COL6A1, and TGFB1 as potential markers of mesenchymal HGSOC subtype.
The CAOV3 EMT induction model was previously developed by our research group to emulate key aspects of cancer progression and metastasis [
6,
14]. The induced EMT phenotypes in this model include enhanced motility, invasiveness, and resistance to cisplatin which have been shown to arise from the stimulation of the ERK/MAPK signaling pathway based on mechanistic studies [
15]. EGF stimulation plays a role in matrix remodeling through the ERK and ILK/GSK-3 pathways in ovarian surface epithelial cells [
16], and the combined targeting of endothelin and EGF receptors has been shown to increase antitumor activity in ovarian cancer [
6,
17]. Furthermore, the activation of NOX enzymes, some of which are located in the plasmatic membrane, triggers EGFR signaling, leading to increased H2O2 levels. This, in turn, results in sulphenylating Cys797 residues of EGFR, creating a positive feedback loop and amplifying kinase activity [
18‐
20].
In a study by Grassi et al. [
6] the cell proliferation rate was measured using BrdU incorporation, and it was demonstrated that the proliferation rate of Caov-3 cells remained unchanged after EGF-induced EMT. This finding indicated that proliferation does not directly influence the production and release of exosomes.
EMT potentially plays a significant role in the outcomes of HGSOC, which is a highly lethal gynecological malignancy [
2]. EMT endows tumor cells with metastatic characteristics, enhancing their mobility, invasiveness, resistance to apoptosis, and stem cell-like properties [
3]. The cancer cell secretome plays a crucial role in maintaining the tumor microenvironment, facilitating disease progression, and metastasis [
21], with exosomes emerging as key players in this context [
5]. Exosome-derived proteins from tumor cells mediate binding, motility, and degradation of the extracellular matrix, activation of hematopoietic cells and stroma, angiogenesis, and promotion of EMT in neighboring non-metastatic tumor cells [
4].
Exosomes serve as carriers for a diverse range of molecules, including proteins, nucleic acids, lipids, and metabolites. In our study, among the differentially accumulated proteins in exosomes from EMT-induced cells and controls, a significant number belonged to lysosomes, nucleosomes, and proteasomes. The analysis of vesicles derived from tumor cells, as well as non-classical secretion processes, revealed intracellular proteins within extracellular vesicles [
9,
22,
23]. Additionally, extracellular vesicles transmit EMT signals. Our study identified several binding-associated proteins, particularly involved in extracellular matrix (ECM) restructuring, cell growth, and motility and migration. These proteins may play a role in tumor EMT induction and metastasis. Overexpression or activation of these pathways in the tumor microenvironment can lead to increased resistance to treatment. The ECM of tumors is recognized as a significant barrier to effective cancer treatment [
24,
25]. High expression of ECM remodeling genes distinguishes the mesenchymal molecular subtype from the other three HGSOC subtypes. The mesenchymal subtype is associated with upper abdominal and omental metastases, suboptimal surgical debulking, residual macroscopic disease after surgery, severe post-operative complications, and reduced overall survival [
26]. Interactions between circulating tumor cells and the distant microenvironment can influence the formation of a metastatic lesion and colonization, contributing to the development of a metastatic niche through the extracellular matrix. In murine cancer models, combination therapies targeting global ECM deposition, such as TGF inhibitors, or specific components of ECM, such as in situ hyaluronic acid breakdown, have shown the potential to enhance responsiveness to therapy [
6,
27‐
30].
The integrated analysis of EMT-induced exosome proteomic data and TCGA-OV cohort gene expression z-scores [
12] identified PLAU, LAMB1, COL6A1, and TGFB1 as potential markers to classify mesenchymal HGSOC subtype. The PLAU gene encodes a secreted serine protease known as plasminogen activator urokinase. Plasminogen activator urokinase converts plasminogen to plasmin promoting the degradation of the basement membrane and facilitating tumor invasion and metastasis. The urokinase-type plasminogen activator system has been associated with aggressiveness and poor prognosis in various solid tumors, including ovarian cancer [
31]. Ovarian cancer patients, especially those in the advanced stage, often exhibit increased serum levels of PLAU and PLAUR [
32,
33], but while its elevated level is linked to shorter progression-free and overall survival, it does not represent a standalone prognostic indicator [
34].
The gene LAMB1 encodes laminin subunit beta 1. Laminins play a role in cell differentiation, migration, adhesion, and tumor cell invasiveness [
35],. LAMB1 has been identified in the EMT process of chemoresistant ovarian cancer cell lines, contributing to a more invasive phenotype [
35,
36]. Overexpression of LAMA4, LAMB1, and LAMC1 mRNAs has been associated with lower overall survival (OS) and progression-free survival (PFS) in ovarian cancer [
37].
In osteosarcoma and lung metastasis, COL6A1 is associated with a worse prognosis. Additionally, COL6A1 can be transported by exosomes and influence the differentiation of normal fibroblasts into cancer-associated fibroblasts. This process, driven by TGF/COL6A1 signaling, promotes cell metastasis, invasion, and migration [
38]. In ovarian cancer cells, COL6A1 has been identified as an effector gene that promotes invasion and metastasis [
39,
40]. The upregulation of COL6A1 and TGFB1 expression in a hypoxic microenvironment may enhance cell proliferation [
41].
The role of TGFBI in cancer is ambiguous. Hypermethylation of the TGFB1 promoter has been associated with the suppression of TGFB1 expression [
42]. TGFB1 is positively regulated by a feedback loop involving FXYD5, TGF β/SMADs signaling drives EMT during ovarian cancer progression [
43]. Macrophage-produced TGFB1 supports an immunosuppressive microenvironment in ovarian cancer [
44]. TGFB1 is also associated with pathogenesis [
45], facilitating migration and invasion [
46]. Furthermore, TGFB1 enhances autophagy and mitophagy through PI3K/Akt and Ras-Raf-MEK-ERK pathways in ovarian cancer [
47].
While our study has provided valuable insights into the proteomic analysis of exosomes secreted during EMT in CAOV3 cells and the identification of potential biomarkers for mesenchymal HGSOC, we recognize the limitation of utilizing a single cell line in the first step of our investigation. Future studies should consider incorporating multiple cell lines in all steps of the study to enhance the generalizability of our findings and to validate the clinical relevance of these potential biomarkers.
Conclusions
Proteomic analysis of exosome proteins from multiple cell lines offers great potential for advancing our understanding of ovarian cancer biomarkers. This approach provides valuable tools for characterizing biomarkers associated with the disease. In our study, we successfully identified and validated candidate biomarkers in EMT-induced TP53 mutant ovarian cancer cell lines, shedding light on their relevance in ovarian cancer progression. The integration of experimental induction of EMT, exosome purification, large-scale proteomic analysis, and integration with gene expression data from the TCGA database allowed us to identify potential markers of the mesenchymal subtype of HGSOC. The identified markers, including PLAU, LAMB1, COL6A1, and TGFB1, warrant further investigation regarding their roles in prognosis, as potential predictors of debulking, and as targets for new therapeutic drugs. Continued research in this field holds the potential to drive advancements in the diagnosis, prognosis, and treatment of ovarian cancer, ultimately leading to improved outcomes for patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.