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% CO2. 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].

Protein extraction

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 4oC 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].

Extraction of proteins from exosomes

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.