Extracellular Vesicle Transmission of Chemoresistance to Ovarian Cancer Cells Is Associated with Hypoxia-Induced Expression of Glycolytic Pathway Proteins, and Prediction of Epithelial Ovarian Cancer Disease Recurrence

Simple Summary Ovarian cancer is one of the most lethal cancers affecting women worldwide. Its high mortality rate is often attributed to the non-specific nature of early symptoms of the disease. Developing a better understanding of the disease progression and identifying clinically useful biomarkers that aid in clinical management are requisite to reducing the mortality rate of ovarian cancer. Reduced oxygen tension (i.e., hypoxia) is not only a characteristic of solid tumors but may also enhance the metastatic capacity of tumors by inducing the release of tumor growth promoting factors. Recently, it has been proposed that small tumor-derived extracellular vesicles (sEVs) facilitate cancer progression. In this study, we established that sEVs produced under low oxygen tension induce a metabolic switch in ovarian cancer cells associated with changes in glycolytic pathway proteins that promote resistance to carboplatin. Significantly, we identified a suite of sEV-associated glycolysis pathway proteins that are present in patients with ovarian cancer that can predict disease recurrence with over 90% accuracy. Abstract Hypoxia is a key regulator of cancer progression and chemoresistance. Ambiguity remains about how cancer cells adapt to hypoxic microenvironments and transfer oncogenic factors to surrounding cells. In this study, we determined the effects of hypoxia on the bioactivity of sEVs in a panel of ovarian cancer (OvCar) cell lines. The data obtained demonstrate a varying degree of platinum resistance induced in OvCar cells when exposed to low oxygen tension (1% oxygen). Using quantitative mass spectrometry (Sequential Window Acquisition of All Theoretical Fragment Ion Mass Spectra, SWATH) and targeted multiple reaction monitoring (MRM), we identified a suite of proteins associated with glycolysis that change under hypoxic conditions in cells and sEVs. Interestingly, we identified a differential response to hypoxia in the OvCar cell lines and their secreted sEVs, highlighting the cells’ heterogeneity. Proteins are involved in metabolic reprogramming such as glycolysis, including putative hexokinase (HK), UDP-glucuronosyltransferase 1–6 (UD16), and 6-phosphogluconolactonase (6 PGL), and their presence correlates with the induction of platinum resistance. Furthermore, when normoxic cells were exposed to sEVs from hypoxic cells, platinum-resistance increased significantly (p < 0.05). Altered chemoresistance was associated with changes in glycolysis and fatty acid synthesis. Finally, sEVs isolated from a clinical cohort (n = 31) were also found to be enriched in glycolysis-pathway proteins, especially in patients with recurrent disease. These data support the hypothesis that hypoxia induces changes in sEVs composition and bioactivity that confers carboplatin resistance on target cells. Furthermore, we propose that the expression of sEV-associated glycolysis-pathway proteins is predictive of ovarian cancer recurrence and is of clinical utility in disease management.


Introduction
Ovarian cancer is the most lethal gynecological cancer [1]. Delayed diagnosis, high metastatic frequency, and resistance to chemotherapy contribute to the lethality of ovarian cancer [1]. Carboplatin, paclitaxel, or a combination of both are the first line chemotherapy treatments for ovarian cancer [2]. Survival rates following treatment are less than 20%, as ovarian cancer often recurs with chemoresistance, making this type of cancer extremely difficult to treat [2]. Carboplatin is a DNA-binding alkylating agent (non-specific cell phase) that displays response rates of~70% in newly diagnosed patients with advanced disease [3,4]. In the presence of tissue hypoxia, the response to carboplatin is significantly reduced [5]. Hypoxia has emerged as a significant feature of the malignant tumor microenvironment and is considered a critical factor in promoting tumor metastasis and chemoresistance associated with poor prognosis [6,7].
Typically, hypoxic regions develop within tumors, as their growth cannot be fully supported by their vascular supply. This results in a spatial disorganization of tumor vascular networks to meet cell demand for oxygen and nutrition [8]. Disorganized and immature blood vascular networks can cause unstable blood flow and limit oxygen diffusion [9,10]. Under a hypoxic environment, cancer cells modulate their metabolic programming to meet their oxygen and nutrient requirements which accelerates cancer progression and increases cell survival and adaptation. For example, during hypoxic exposure, cancer cells induce the activity of glycolytic enzymes to convert glucose into lactate, to provide sufficient energy and biomass supporting cell survival and proliferation [11,12]. This process causes the cancer cells to secrete CO 2 and carbonic acid (low pH) and thus limits the effectiveness of chemotherapeutic agents [13,14]. The acidic environment produced by hypoxic cells increases the activity of P-glycoprotein (Pgp) in cancer cells [15]. Pgp is an efflux pump protein that reduces the cytotoxicity and accumulation of chemo drugs in the cytosol [13].
Recent studies highlight the role of small extracellular vesicles (sEVs) in ovarian cancer progression [16][17][18]. sEVs are lipid bilayer-encapsulated vesicles that are naturally released from a wide range of cells including cancer cells [19]. sEVs are selectively packaged with bioactive molecules, including proteins and small non-coding RNA, that are secreted by exocytosis into biofluid compartments [20]. EVs modulate the activity of both proximal and distal cells to modify several biological processes such as angiogenesis [19], proliferation [21], and metabolism [22]. EVs signaling, thus, represents an integral pathway mediating cell to cell communication in many cancers, including ovarian cancer [16]. Small EVs like exosomes released from cancer cells exposed to low oxygen tension reduced apoptosis in target cells [23], suggesting that exosomes might be involved in chemotherapy resistance.
The available evidence suggests that the development of tumor resistance to chemotherapy is, in part, due to cell-cell transfer of effectors within the tumor microenvironment [24,25]. For example, hypoxia increases the release of exosomes from cancer cells, alters their cargo, and enhances cell migration and invasion in target cells [26]. Similarly, cells isolated from ascites obtained from patients with ovarian cancer and cultured under hypoxic conditions release sEVs enriched with STAT3 and FAS; oncogenic proteins can induce both cell migration (in vitro) and tumor invasion in vivo [27].
The aim of this study was to test the hypothesis that carboplatin-resistant cancer cells release sEV-associated proteins, under hypoxic conditions, that confer chemoresistance to recipient cells. This hypothesis was further tested by establishing the capacity of sEVassociated proteins, isolated from ovarian cancer patients, to predict chemo-resistance and disease recurrence.

Hypoxia Induced Carboplatin-Resistance Is Characterized by Metabolic Pathway Dysregulation
Hypoxia may induce proteomic changes that control cell growth, activate the survival pathway of cells, mediate cell-cycle arrest, and promote cancer progression. We, therefore, investigated whether hypoxia-induced changes in carboplatin resistance in COVA-3 cells are associated with changes in the proteomic profile. Changes in protein abundance were determined by quantitative mass spectrometry analysis using a sequential window acquisition of all theoretical fragment ion spectra mass spectrometry (SWATH). Informationdependent acquisition (IDA) and SWATH profile were generated from CAOV-3 cultured at 8% and 1% O 2 using independent samples (n = 6, per each group). The IDA library was used to identify peptide ions present in SWATH ion profiles. Proteins were identified and quantified by comparing SWATH-generated peptide ion profiles for each individual sample against the IDA library (PeakView). IDA of mass spectra from all individual samples was initially performed and identified 1233 total proteins and analyzed using IDA and SWATH. The variation in the relative abundance of proteins between cells (8% O 2 or 1% O 2 ) was established by comparison with the SWATH profile against the IDA library and presented as volcano plot (Figure 2A). A total of 30 proteins were significantly differentially expressed in cells that were cultured under hypoxic conditions when compared with normoxic cells (i.e., 8% O 2 ) (Supplemental Table S1). From the MS/MS data, we identified a significant increase in the detoxification and ATP production enzymes: 3-hydroxyacyl-CoA dehydrogenase type-2 (HCD2) and Succinate-CoA ligase (SUCB1) (p = 0.037 and p = 0.002, respectively).  -I): SKOV-3, OVCAR-3, OVCA-420, TOV-112D,  CAOV-3, OVTOKO, OV-90, HEY, and OVCA-429) were exposed to either 1% O 2 or 8% O 2 and treated with carboplatin at different concentrations (0, 0.001 µM, 0.01 µM, 0.1 µM, 1 µM, 10 µM, 100 µM, and 1000 µM) in the presence of Caspase-3/7 reagent. The viability of the cells was assessed using the IncuCyte™ real-time cell-imaging system every 2 h for 72 h. The data are represented by the mean ± SEM (n = 6). The IC50 of all EOC cells in this study were higher under hypoxia than under normoxic in their parental cells. *** p < 0.0005 at 1% oxygen compared with 8% oxygen.

Figure 2.
Quantitative proteomic analysis of CAOV-3 cells exposed to 1% O 2 and 8% O 2 . (A). Volcano plot of the quantified proteins showing the differentially expressed proteins in the CAOV-3 cells exposed to 1% oxygen tension compared with cells exposed to 8% oxygen tension (upregulated on the right side and downregulated on the left side). 30 proteins were significantly expressed in the hypoxic cells (1% O 2 CAOV-3) than in the normoxic cells. Gene set enrichment analysis (GSEA) of proteomic data illustrate four gene clusters that are enriched in the 1% O 2 CAOV-3 cells compared cells at 8% O 2 namely, (B) hypoxia, (C) epithelial-mesenchymal transition, (D) glycolysis, and (E) Myc target genes. The protein subset was evaluated using the normalized enrichment score (the green line). In the GSEA figures, the black vertical lines identify the positions where the members of a particular pathway appear in the ranked list of genes. The upregulated genes in red are localized on the left, while the downregulated proteins in blue are on the right.
To gain further insights into the biological functions of the dysregulated proteins, MS/MS profile data were subjected to gene enrichment analysis (GSEA), revealing several gene sets to be significantly enriched in hypoxic cells compared with normoxic cells. This is demonstrated by the normalized enrichment score that is the primary statistic that determines whether a gene set is overrepresented in a ranked list of genes and has an FDR <0.01. GSEA showed an enrichment of proteins involved in hypoxia, epithelial mesenchymal transition, glycolysis, and MYC targets ( Figure 2B-E). These results establish that hypoxia changes the proteomic profile of CAOV-3 cells, involved in the expression of the glycolysis signaling pathway and proteins that are important in the metabolic requirements associated with cellular response to chemotherapy.

Analysis of sEVs from CAOV-3 Cells
This study further investigated the effects of hypoxia on cellular protein abundance and if these changes are reflected in sEVs released from ovarian cancer cells. CAOV-3 cells were used because a significant switch in cell apoptosis in response to carboplatin in 1% oxygen compared with 8% oxygen was observed (i.e., from 11 to 28 µM, Figure 1). Extracellular vesicles were isolated from CAOV-3 cell-conditioned media and cultured at 8% and 1% O 2 for 48 h. The size distribution of the vesicles is presented in Figure 3A. No significant differences were observed in the size distribution of the EVs from cells cultured with 8% O 2 and those cultured with 1% O 2 . The EV preparations contained vesicles~100 nm in diameter and were consistent with the size range and morphology of sEVs ( Figure 3A,C). sEVs were positive for CD9 and TSG101 ( Figure 3B and Figure S4), proteins associated with vesicles from an endocytic origin-like sEVs. The EV preparation was negative for Grp94 ( Figure 3B and Figure S4); an endoplasmic reticulum marker used to establish the purity of EV isolates. The release of EVs from cells incubated under hypoxic conditions was 2-fold greater (p < 0.05) than that observed under 8% O 2 ( Figure 3D).
The proteomic profile of sEVs from the CAOV-3 cultured at 8% and 1% O 2 was determined using the IDA of all individual sEV samples as well as SWATH. A total of 1233 EV-associated proteins were identified and subjected to functional enrichment analysis using FunRich (http://www.funrich.org (accessed on 20 December 2019)) software to further elucidate their putative biological relevance. Proteins were characterized as cytoplasmic (67.3%) and exosomal proteins (52.2%), respectively ( Figure 3E). Functionally, these proteins were found to be involved in protein metabolism (21%), metabolism (18%), and energy pathways (17%) ( Figure 3F). A total of 35 proteins from the 100 exosomeassociated proteins present in the ExoCarta database were identified in the sEV samples. Overall, these results confirmed the reproducibility of the present method to enrich a specific type of EVs, sEVs. sEVs were defined by the enrichment of proteins associated with sEVs (i.e., CD9, and TSG101) as well as a lack of Grp94 (negative control); size distribution of~100 nm (within the size of the sEVs); and spherical morphology. SWATH analysis matched the sEVs database and revealed that an enriched population of sEVs was isolated in the present study.

sEVs Protein Cargo Is Altered by Oxygen Tension and Is Selectively Enriched for Chemoresistant Factors
The difference in the proteomic profiles of sEVs from 1% versus 8% O 2 is summarized in Figure 4A, as a volcano plot. A total of 65 proteins were differentially expressed in sEVs that were cultured under hypoxic conditions when compared with normoxic cells (8% O 2 ) (Supplemental Table S2). Moreover, glycolysis pathway-associated proteins and detoxification enzymes were enriched in sEVs from 1% when compared with 8% O 2 ( Figure 4B and Supplemental Table S3). A specific set of proteins was selectively enriched within the sEVs when compared with their cells of origin at 8% and 1% O 2, indicating the specific packaging of proteins into sEVs ( Figure 4C-E). 8% sEVs were enriched with proteins associated with epithelial mesenchymal transition ( Figure 4D), and 1% sEVs were enriched with hypoxia association proteins ( Figure 4F). Data were analyzed by GSEA and are presented in Figure 4B,D,F.  Ingenuity Pathway Analysis (IPA) was used to identify signaling pathways associated with changes in the proteomic profile in sEVs from cells cultured under hypoxic compared with normoxic conditions. The main network activated by hypoxic sEVs compared with normoxic sEVs is mitochondrial dysfunction and NRF2-mediated oxidative stress response and synaptogenesis signaling pathway (Supplemental Figure S1A). The top regulator effect networks were associated with movement of endothelial cells and invasion of cancer cells (Supplemental Figure S1B). Overall, this study suggests that when compared with sEVs from normoxic cells, sEVs from hypoxic conditions were enriched with signals associated with metastasis and glycolysis-key factors in chemoresistance [29]. This indicates that sEVs from hypoxic cells may play a role in regulating recipient-cell responses to carboplatin and induce chemoresistance in ovarian cancer.

Targeted Proteomic Analysis of Glycolysis in Ovarian Cancer Cells
The effects of hypoxia on glycolytic pathway proteins were assessed using liquid chromatography multiple reaction monitoring (LC-MRM). Proteins involved in core glycolysis, glycolysis regulation, pentose phosphate and other metabolism pathways were selected based on previously published data [30] and are detailed in the Supplemental Table S4 and Figure 5. To ensure assay reproducibility, a Hela digest was injected at the start and in the middle of the sample set. Figure 5 show the peak areas and retention times for each of the selected peptides.

Effect of Hypoxia on Glycolytic Pathway Proteins in Ovarian Cancer Cell and Their Secreted Extracellular Vesicles
To evaluate whether low oxygen tensions (i.e., 1% oxygen) modified the glycolysis proteome in ovarian cancer cell lines, and their secreted sEVs, we performed three experimental analyses: (i) Cellular Proteome: The effect of hypoxia (1% O 2 for 48 h) on the proteome of CAOV-3, SKOV-3, TOV-112D, OVTOKO, HEY and OV-90 cell lines is presented in as Figure 6A-F. Significant cell-specific differential responses (p < 0.05) in the expression of glycolytic pathway proteins (e.g., ENO1, LDHA, ENO, PGK1, PKM2 TP1) were identified; (ii) Extracellular Vesicle Proteome: The effect of hypoxia on extracellular vesicle-associated proteins isolated from CAOV-3, SKOV-3, TOV-112D, OVTOKO, HEY, and OV-90 cell-conditioned media are presented as volcano plot in Figure 6A-F. Specific changes associated with hypoxia for each cell line were identified and included, TPI, PGK, ENO, PKM, and LDH (p < 0.05); and (iii) Comparison of Cellular and Extracellular Vesicle Proteomes: A cell-specific comparison of the cellular and extracellular vesicle proteomes is present in Figure 7. In general, changes in the glycolysis proteome of cells are reflected in their secreted EVs under both 8%, and 1% oxygen ( Figure 7A,B). Specific changes associated with each cell type were identified, in cells (e.g., TP1, and PGK1 in CAOV-3), and in EVs (e.g., PKM2 in HEY) at 8% oxygen. Specific differentially abundant proteins in cells compared with EVs were identified for each cell line (i.e., comparison analysis between cells and EVs for COV-3, SKOV-3, TOV-112D, OVTOKO, HEY, and OV-90), suggesting that proteins associated with glycolysis proteome in EVs do not reflect the cell of origin, perhaps due to the specific packaging of proteins into EVs ( Figure 8C,D).

sEVs Released from Hypoxic Cells Confer Carboplatin Resistance to Recipient Normoxic Cells
Next, we evaluate whether hypoxic sEVs can transfer resistance to carboplatin in target cells. Effect of sEV on carboplatin-induced apoptosis: The effects of hypoxia on the capacity of sEVs to inhibit carboplatin-induced apoptosis in CAOV-3 cells is presented in Figure 9. EVs isolated from CAOV-3 cells cultured under 1% significantly inhibited carboplatin-induced [0-100µM] apoptosis in normoxic CAOV-3 cells. Effect of sEV on recipient cell proteome: The effect of EVs isolated from CAOV-3 cells cultured under 1% on the proteome of recipient normoxic CAOV-3 cells is summarized in Supplemental Table  S5 and Figure 9C. A total of 132 proteins were differentially expressed (p < 0.05) in cells incubated in the presence of hypoxic sEVs compared with cells incubated with normoxic sEVs. Gene enrichment analysis (GSEA) identified genes related to oxygen-monitoring mechanisms, including hypoxia-inducible factors (HIFs), the major components of hypoxia signaling pathways, glycolysis, fatty acid synthesis, and protein secretion ( Figure 9D). The data obtained are consistent with the hypothesis that oxygen tension regulates the packing of specific proteins into sEVs that are involved in the metabolic adaptation of cancer to enhance glycolytic pathways and that they play a role in inducing carboplatin resistance in recipient cells. Figure 5. Peptides in the glycolytic enzyme structure. Left: The peptides are colored by the number for each glycolytic enzyme structure from the protein data bank (PDB) with the best resolution and coverture available. For each peptide, the solvent accessible surface (SASA) was calculated with the software freesasa, the average for residue is in parenthesis (A-P). The structures were prepared with pymol in surface mode and in biological assembly available from PDB (monomer, dimer, and tetramer). Right: MRM chromatogram of the transition used to trigger MS/MS for each protein associated with "glycolysis proteome".

Analysis of Circulating sEVs in Patients with Ovarian Cancer
To determine if changes in protein expression observed in vitro experiments are recapitulated in vivo, sEVs were isolated from the plasma of patients with ovarian cancer recurrence (n = 9), and controls (i.e., without ovarian cancer, n = 22) after 12 months from initial diagnosis (Figure 10) (Table S6). No differences were identified in the concentration, mean, and mode of the circulating sEVs from patients with cancer recurrence and controls ( Figure 10A-C). Morphological analysis using electron microscopy showed vesicles of around 100 nm ( Figure 10D). Information-Dependent acquisition (IDA) mass spectrometry analysis was performed and identified 353 total proteins (Supplemental Table S7). Interestingly, of the 353 proteins identified in sEVs from plasma samples, a total 89 proteins are present in sEVs from CAOV-3 cells cultured at 1% oxygen (Supplemental Figure S2). An Ingenuity Pathway Analysis library identified 4 exosome-associated glycolysis pathway proteins: pyruvate kinase M1/2, enolase 1, glyceraldehyde-3-phosphate dehydrogenase, and aldolase fructose-bisphosphate (Supplemental Figure S3). The relative abundance of these proteins in patients with ovarian cancer recurrence and controls was established by comparison using SWATH and expression profiling against the IDA library ( Figure 10E-H). The expression of pyruvate kinase M1/2, enolase 1, and aldolase fructose-bisphosphate were significantly higher in circulating sEVs of patients with ovarian cancer recurrence compared to the controls ( Figure 10E-I). No difference in the expression of glyceraldehyde-3-phosphate dehydrogenase was identified ( Figure 10G).

Discussion
The development of chemoresistance in patients with cancer is a major factor limiting survival [31]. Elucidating the fundamental mechanisms by which cancer cells acquire resistance to chemotherapeutic agents, therefore, is requisite to developing more efficacious treatments. Hypoxia is a common feature of malignant tumors [32] and is associated with reprogramming of cellular metabolism to favor glycolytic energy production, increased invasiveness, metastatic potential, angiogenesis, and reduced apoptosis and chemoresistance [32][33][34][35]. Interestingly, small extracellular vesicles like exosomes have been associated with platinium resistance in ovarian cancer [36]. This study aimed to test the hypothesis that hypoxia-induced chemoresistance is propagated within ovarian tumors by extracellular vesicle signaling between cancer cells. In this study, we used the terminology sEVs as recommended by the International Society of Extracellular Vesicles, due to the mixed origin of EV preparations and the lack of specific EV subtype markers [37].
The data obtained using a panel of nine epithelial ovarian cancer cell lines, confirm that carboplatin resistance (as defined by cell the IC 50 for cell survival) is regulated by oxygen tension but varies up to 3-fold between cell lines (IC 50 = 8.7 µM for OVCA-420 to IC 50 = 28.5 µM for CAOV-3). The incubation of cells under low oxygen tension induces the cellular expression of a cassette of proteins associated with hypoxia (e.g., HIF-1), epithelial mesenchymal transition (e.g., GPC1, PPIB), glycolysis (e.g., CPC1, FKBP4) and MYC targets (e.g., XP01, and G3BP1), and increases the release of extracellular vesicles by 2-fold (as indicated by the concentration of extracellular vesicles in cell-conditioned media). Hypoxia-induced cellular proteins were enriched in the extracellular vesicles released from these cells. The treatment of carboplatin-sensitive ovarian cancer cells with extracellular vesicles isolated from cells incubated under low oxygen tension conferred carboplatin resistance. These data support the hypothesis that ovarian cancer cells utilize extracellular vesicle signaling between cancer cells to propagate chemoresistance. Furthermore, the data obtained are consistent with recent studies that implicate extracellular vesicles in hypoxia-induced metabolic changes within the pre-metastatic milieu [38][39][40] and in the acquisition of chemoresistance [41][42][43][44][45][46].
While the mechanisms by which extracellular vesicles derived from hypoxic ovarian cancer cells induce chemoresistance in recipient cells remains to be elucidated, the protein cargo they carry has been implicated in metabolic and oncogenic pathways. For example, HIF-1, increases the flux of the glycolysis pathway, which in turn decreases the flux of the tricarboxylic acid cycle (TCA). Decreased TCA flux reduces mitochondrial reactive oxygen species (ROS) production but maintains matrix ATP concentrations under energy-limited conditions [47]. HIF-1 also increases the flux of the serine synthesis pathway, inducing the production of antioxidant products, such as NADPH and glutathione, to neutralize ROS [47]. In this study, 3-hydroxyacyl-CoA dehydrogenase (HAD) was significantly upregulated. HAD belongs to a family of oxidoreductases that catalyzes 3-hydroxy-2methylbutanoyl-CoA in the third step of beta-oxidation and produces NADPH and 2methyl-3-COA.
Similarly, hexokinase (HK), UDP-glucuronosyltransferase 1-6 (UD16), 6-phosphogluconolactonase (6PGL), and CTP synthase 1 (PYRG1), the key enzymes involved in the glycolysis and biosynthesis of phospholipids and nucleic acids [48], were significantly enriched in the sEVs from hypoxic cells. Patra et al. demonstrated that HK positively correlates with cancer cell survival through the initiation of glycolysis, an effect that is abolished when HK is knocked out using an in vivo model [49]. HK expression is a key factor in inducing tumor progression whilst depletion of HK assists in restoring the balance between glycolysis and OXPHOS pathways, leading to the upregulation of mitochondrial biogenesis and induction of intrinsic apoptotic pathways [50]. HK is also involved in chemoresistance via the enhancement of the autophagy of cisplatin in ovarian cancer [51]. The enrichment of proteins that are involved in NADH/NADPH productions (i.e., alcohol dehydrogenase class-3 (ADH5), NAD-dependent malic enzyme (MAMO), bifunctional methylenetetrahydrofolate dehydrogenase/cyclohydrolase (MTDC) and peroxisomal multifunctional enzyme type 2 (DHB4)) was also detected in hypoxic sEVs. This enrichment could potentially act as a protective mechanism for the targeted cells as it has been shown that these enzymes reduce oxidative stress under low oxygen consumption. We, therefore, suggest that the presence of oncogenic molecules secreted from hypoxic cells via sEVs initiate the metabolic changes of normoxic cells, upregulate the glycolysis pathway, and drive physiological responses such as cell survival and chemoresistance.
In addition to proteins, EVs can carry a diverse range of molecules, such as lipids or microRNAs that can mediate chemoresistance by interacting with the immune system [52]. Cytotoxic chemotherapy induces dynamic changes in the cellular immune landscape from within the tumor microenvironment, impacting overall survival outcomes [53,54]. EVs obtained from ovarian tumors were found to be immunosuppressive [55], causally linked by their cargoes of lipid phosphatidylserine (PS) [56,57] and ganglioside GD3 [58,59]. Furthermore, the microRNA-940 was found to be highly expressed in EVs derived from ovarian cancer cells under hypoxic conditions [60], inducing macrophages to remodel a protumorigenic microenvironment. Future analyses investigating the role of EVs from hypoxic ovarian cancer cells could also characterize non-protein cargoes such as PS, ganglioside GD3 or microRNAs. Thus, further studies are required to determine the role of other molecules within sEVs from hypoxic conditions on conferring resistance to carboplatin in target cells.
The data obtained define an extracellular vesicle chemoresistance protein profile that may be of clinical utility as an aid in the management of patients with disease recurrence. More than 60% of ovarian cancer patients that present with advanced stage disease develop recurrent chemoresistant disease [61]. In this study, we identified an increased expression of pyruvate kinase M1/2, enolase 1, and aldolase, fructose-bisphosphate A, in circulating extracellular vesicles of patients with ovarian cancer recurrence compared to the controls. Previously, these enzymes have been implicated in metabolic adaption, tumorigenesis, chemoresistance and patient survival [62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78]. Multivariate models based on the expression of these proteins in extracellular vesicle delivered classification efficiencies of 73-100%. A larger patient study is warranted to validate these findings clinically.
In conclusion, this study advances our understating of the involvement of extracellular vesicle signaling pathways in the transfer of chemoresistance between ovarian cancer cells. Such pathways may be targeted for therapeutic intervention and/or improved patient management.

sEVs Isolation and Characterization
Cells grown to 70% confluence in T175 flasks were washed twice with phosphatebuffered saline (PBS) before FBS-free media was added. The cells were then incubated for 48h before conditioned media was collected for sEV isolation. sEV were isolated by differential centrifugation and ultrafiltration as previously published [16]. Briefly, cell conditioned media or plasma were centrifuged at 800× g for 10 min, 2000× g for 10 min at 4 • C, and 12,000× g for 10 min at 4 • C to remove cells and debris. The resultant supernatant was centrifuged at 100,000× g for 2 h at 4 • C with the Type 70.1 Ti fixed angle rotor (Beckman Coulter, CA, USA) to pellet the sEVs. The resultant pellet was resuspended in 10 mL PBS, filtered through a 100 kDa Amicon ® Ultra-15 Centrifugal Filter Units (MERCK, Bayswater, VIC, Australia), and centrifuged at 4000× g for 30 min. The sEVs were collected and stored at −80 • C for further experiments. sEVs were characterized according to the recommendations of the International Society of Extracellular Vesicles, by size distribution, abundance of proteins (i.e., CD63 and TSG101) associated with sEVs, and by a morphological assessment involving Nanoparticle Tracking Evaluation (NTA), Western blot, and electron microscopy [37].

Ion Library Generation and SWATH
Protein (20 µg) from individual samples was reduced, alkylated, and trypsinized in triplicate using the filter-aided sample preparation (FASP) method in order to generate an ion library for SWATH mass spectrometry (MS) analysis. The resulting peptide samples were combined and processed in an information-dependent acquisition (IDA) on an AB Sciex 5600 TripleTOF MS with the top 20 precursor ions automatically selected for fragmentation and processed. For SWATH analysis, a global false discovery rate (FDR) of 1% was used as the threshold for the number of proteins for import to ensure the highest quality ion library. SWATH acquisition was as described by Menon et al., [79], operating in a looped product ion mode. Using an isolation width of 26 Da (25 Da of optimal ion transmission efficiency and 1 Da for the window overlap), a set of 32 overlapping windows (1 m/z overlap) was constructed covering the mass range 400 to 1200 m/z.

Data Processing
To generate the protein library, ProteinPilot (version 4.5b) software and the Paragon™ Algorithm were used to search against a human SwissProt database. For SWATH processing, the SWATH Acquisition Microapp (version 2.0) within PeakView (version 2.2) was used. An FDR threshold of 1% was applied by setting three peptides per protein with the retention time manually realigned with a minimum of five peptides with consistently high signal intensities and distributed along the time axis. The resulting peak area for each protein after SWATH processing was exported to MakerView version 1.3.1 (SCIEX) for statistical analysis. The resulting data was normalized using Total Area Sum (TAS). The coefficient of variation in the peptide abundance across the samples was established by comparing the SWATH peptide ion against the IDA library. Finally, unpaired Student's t-tests were performed with p < 0.05 considered statistically significant.

Target Peptide Set Selection for "Glycolysis Proteome"
The proteins involved in core glycolysis, glycolysis regulation, pentose phosphate, and other metabolism pathways were selected based on a previous study [30]. Also included were five commonly used reference proteins. The FASTA protein sequence for all proteins were imported into Skyline software (version 20.2.1.404) for method development and data processing. For selecting the optimal fragment ions, a targeted MRMHR experiment was performed using the pre-selected peptides and their corresponding masses as per the above study. A commercially available Hela tryptic digest (500 ng) was injected two times into a TripleTOF 5600 mass spectrometer (AbSciex) connected to a NanoLC 400 system with an analytical MicroLC column ChromXP C18CL, 120A, 150 × 0.3 mm (Eksigent). Chromatography was performed in a trap-elution mode with solvent A (100% H 2 O, 0.1% FA) and solvent B (100% ACN, 0.1% FA). Peptide separation was performed at a total flow rate of 5 µL/mL according to the gradient conditions listed the Supplemental Table S8. Afterwards, a full scan data extraction from the raw. wiff files were conducted in Skyline. The fragment ions were manually inspected, and the presence of consecutive high mass y ion series gives evidence of the correct identification of the peptide. The top three fragment ions were elected for the MRM experiment based on the highest co-eluting y ions' intensities.
The MRM experiments were performed on a 5500 QTRAP hybrid triple quadrupole/linear ion trap mass spectrometer (ABSciex), equipped with an Eksigent MicroLC 200 System. Hela digest was again injected, with the retention times of the peptides used to build a scheduled method with a two-minute window. Cells or cell derived EVs were digested using FASP as previously described by our group [80]. Tryptic digests of cells or exosomes were resuspended in 20 uL of H 2 O, 0.1% FA of which 5 uL was injected and separated by RP-HPLC in an analytical MicroLC column HALO™ C18 2.7 µm, 90A 0.5 × 100 mm (Eksigent). Chromatography was performed with solvent A (100% H 2 O, 0.1% FA) and solvent B (100% ACN, 0.1% FA) at a total flow rate of 15 µL/mL according to the gradient conditions listed in Supplemental Table S9. To ensure the assay's reproducibility, a commercially available hela peptide digest (ThermoFisher) was injected at the start and in the middle of the sample set ( Figure 5). The peak areas and retention times for each of the selected peptides.

Patient Derived Specimens
Using an Oregon Health & Science University (OHSU) Institutional Review Board approved protocol (OHSU #0921), the participating women consented for blood draw (10mls in sodium-citrate tube) before surgery for suspected ovarian adenocarcinoma. This study was approved by the Human Research Ethics Committee of the the University of Queensland (2021/HE000319). Platelet poor plasma was made by centrifugation at 2500× g for 10 min and 0.5 mL aliquots were banked at −80 • C. These banked coded plasma samples were provided by OHSU for experimental analysis, linked to a coded database without patient health identifiers extracted from the electronic medical record. Study subject coded data included patient age, pathological diagnosis (e.g., benign cystadenoma, borderline tumor, primary ovarian carcinoma including differentiation and grade, or metastatic carcinoma including classification) that was confirmed by a board certified anatomic pathologist with expertise in gynecologic pathology (Dr. Morgan, OHSU) [82], clinical stage, whether the patient was optimally debulked (yes or no), interval from blood collection to follow up assessment, and clinical outcome at follow up (recurrent disease (RD), dead of disease (DOD), and alive without disease (AWOD)).

Statistical Analysis
Data are represented as mean ± SEM. Comparisons between two groups were performed by unpaired Student's t-tests. Multiple groups were compared using the analysis of variance (ANOVA). Post hoc analyses were used for pairwise comparisons (Bonferroni correction test). Statistical significance was defined as at least p < 0.05. Statistical analyses were preformed using commercially available programs Stata 11 (StatCorp, College Station, TX, USA) and Prism 6 (GraphPad Inc, La Jolla, CA, USA).

Conclusions
Hypoxia induces changes in the composition and bioactivity of small EVs like exosomes, that confer carboplatin resistance in recipient target cells. These findings provide a new insight into tumor cell EV signaling and identify a putative theranostic target for the prediction and prevention of disease recurrence.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13143388/s1, Figure S1: Ingenuity Pathway Analysis (IPA) was used to identify signaling pathways associated with changes in proteomic profile in sEVs from cells cultured under hypoxic compared with normoxic conditions. (A) Canonical Pathways, and (B) Diseases and Bio Functions. Figure S2. Analysis of the protein identified in small extracellular vesicles isolated from cells and plasma. sEVs were isolated from CAOV-3 cells cultured at 1% oxygen, and plasma obtained from patients with ovarian cancer, and global proteomic identification was performed. Venn diagram shown the unique and common protein identified in sEVs from plasma, and from cell-conditioned media. Figure S3. An Ingenuity Pathway Analysis library in circulating small extracellular vesicles identified 4 sEV-associated glycolysis pathway proteins: pyruvate kinase M1/2, enolase 1, glyceraldehyde-3-phosphate dehydrogenase, and aldolase fructose-bisphosphate. Figure  S4. Uncropped Western blot images. Table S1. Differentially expressed proteins in 1% oxygen CAOV-3 cells compared to 8% oxygen CAOV-3 cells. Table S2: Differential expressed proteins in sEVs derived from 1% CAOV-3 cells in compared to exosomes derived from 8% CAOV-3 cells. Table  S3. A comparison of proteins differentially expressed in 1% O2 sEVs and 8% O2 sEVs. Table S4. Target Set of Metabolic and Reference Proteins. Table S5. Differentially expressed proteins in cells incubated with hypoxic exosomes compared to cells incubated with normoxic sEVs. Table S6. Clinical characteristics of the patients involved in this study. Table S7. Proteins identified in circulating sEVs isolated from plasma. Table S8. Peptide separation was performed at a total flow rate of 5 µL/mL according to the gradient conditions listed below. Table S9 Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
All data generated during this study are available within the article and its supporting information. Further details are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.