Distinct prostate cancer-related mRNA cargo in extracellular vesicle subsets from prostate cell lines

Background Multiple types of extracellular vesicles (EVs), including microvesicles (MVs) and exosomes (EXOs), are released by all cells constituting part of the cellular EV secretome. The bioactive cargo of EVs can be shuffled between cells and consists of lipids, metabolites, proteins, and nucleic acids, including multiple RNA species from non-coding RNAs to messenger RNAs (mRNAs). In this study, we hypothesized that the mRNA cargo of EVs could differ based on the EV cellular origin and subpopulation analyzed. Methods We isolated MVs and EXOs from PC-3 and LNCaP prostate cancer cells by differential centrifugation and compared them to EVs derived from the benign PNT2 prostate cells. The relative mRNA levels of 84 prostate cancer-related genes were investigated and validated using quantitative reverse transcription PCR arrays. Results Based on the mRNA abundance, MVs rather than EXOs were enriched in the analyzed transcripts, providing a snapshot of the tumor transcriptome. LNCaP MVs specifically contained significantly increased mRNA levels of NK3 Homeobox 1 (NKX3-1), transmembrane protease serine 2 (TMPRSS2), and tumor protein 53 (TP53) genes, whereas PC-3 MVs carried increased mRNA levels of several genes including, caveolin-2 (CAV2), glutathione S-transferase pi 1 (GSTP1), pescadillo ribosomal biogenesis factor 1 (PES1), calmodulin regulated spectrin associated protein 1 (CAMSAP1), zinc-finger protein 185 (ZNF185), and others compared to PNT2 MVs. Additionally, ETS variant 1 (ETV1) and fatty acid synthase (FASN) mRNAs identified in LNCaP- and PC-3- derived MVs highly correlated with prostate cancer progression. Conclusions Our study provides new understandings of the variability of the mRNA cargo of MVs and EXOs from different cell lines despite same cancer origin, which is essential to better understand the the proportion of the cell transcriptome that can be detected within EVs and to evaluate their role in disease diagnosis. Electronic supplementary material The online version of this article (doi:10.1186/s12885-017-3087-x) contains supplementary material, which is available to authorized users.


Background
Extracellular vesicles (EVs) are membrane-derived particles released under normal conditions by many cell types to eliminate non-desired cell components and to share information between cells and their environment [1,2]. EVs are considered to play important roles in cell-to-cell communication, contributing to changes in the recipient cell phenotypes by influencing their functions, which is of special relevance for instance in the hallmarks of cancer [3,4]. Cells release a diverse mixture of EVs that can be considered as the cellular EV secretome [5][6][7]. Most notable subpopulations of EVs are microvesicles (MVs) and exosomes (EXOs), both released by viable cells under normal physiological conditions. Current state of the art classification considers MVs to have a diameter of 100-1,000 nm, and being primarily derived from the cell surface by budding of the plasma membrane, whereas EXOs are smaller, 30-150 nm, and are formed within the multivesicular bodies which liberate EXOs to the extracellular space by fusion with the plasma membrane [8]. Importantly, EVs contain a broad variety of signaling molecules that reflect the composition of their originator cells including lipids, proteins, metabolites, sugars, and particularly nucleic acids [2,8]. These molecules can be functionally delivered mediating biological activities and functional changes in the recipient cells [9][10][11][12]. Interestingly, the EV cargo only partly reflects the cells of origin, and especially EXOs have been shown to carry different RNA species compared to the producing cells [9,10,[13][14][15].
Several RNA molecules, including messenger RNAs (mRNAs) and non-coding RNAs (ncRNAs), particularly microRNAs (miRNAs) but also transfer RNAs and other ncRNAs, have been investigated from EVs of different origins [15][16][17]. MVs and EXOs share functional features, and to a large extent, molecular composition. However, they are still distinct subpopulations of EVs that are likely to mediate different actions. As the most common methods of vesicle isolation do not allow separation of EV subsets, most EV studies have not compared MVs and EXOs, and hence a very limited number of reports have so far compared the cargo differences of the EV subtypes [5,6,[18][19][20].
We hypothesized that the mRNA content of EVs differs based on the EV subpopulation analyzed and the cellular origin of the EVs. Moreover, we focused on the comparison of the mRNA cargo of MVs and EXOs from different prostate cancer (LNCaP and PC-3) and benign (PNT2) cells, since such comparison has so far not been performed. Quantitative reverse transcription PCR (RT-qPCR) mRNA arrays were used to characterize and validate the mRNA signatures of 84 prostate cancer (PCa) related genes in the EVs. We analyzed mRNAs known to be related to PCa, as such data could provide new insights for understanding the differential mRNA content of EV subpopulations which may be of future clinical use.

Cell culture
LNCaP and PC-3 (ATCC, Manassas, VA, USA) were grown in RPMI 1640 and DMEM/F12 media (Sigma-Aldrich, St Louis, MO, USA), respectively. Both media were supplemented with 10% (v/v) EV-depleted FBS (Sigma-Aldrich). The FBS was EV-depleted by ultracentrifugation at 118,000 x g avg for 18 h using a type 45 Ti rotor k-factor 178.6 (Beckmann Coulter, Brea, CA, USA), followed by filtration through a 0.22 μm filter (Merck Millipore, Billerica, Massachusetts, USA). Immortalized human benign prostate epithelial cells PNT2 (ECACC, Sigma-Aldrich) were grown in serum-free defined keratinocyte media, supplemented with bovine pituitary extract and human recombinant epidermal growth factor (Life Technologies, Carlsbad, CA, USA). All media were supplemented with 100 IU/mL of penicillin and 100 μg/mL streptomycin (HyClone, Logan, UT, USA). Cells were cultivated at 37°C and 5% CO 2 . Cell viability was measured by Trypan Blue solution (Sigma-Aldrich). Cells were routinely checked for mycoplasma contamination using the MycoAlert™ PLUS (Lonza Walkersville, MD, USA).

Extracellular vesicle isolation
Three hundred mL of cell cultured conditioned media was harvested from LNCaP, PC-3, and PNT2 cells at 80% confluence, and centrifuged at 1,000 x g at 4°C for 10 min to remove remaining cells and cellular debris. The remaining supernatant was centrifuged at 2,500 x g for 25 min at 4°C to pellet larger vesicles such apoptotic bodies. The supernatant was transferred to new tubes and centrifuged at 20,000 x g avg for 25 min at 4°C to pellet the MV-enriched fractions. The final supernatant was ultracentrifuged at 110,000 x g avg , for 2 h using an Optima XE 90 ultracentrifuge, 45 Ti rotor and polyallomer centrifuge tubes (Beckman Coulter) k-factor 191.3, to pellet the EXO-enriched fractions. All vesicle pellets were resuspended in PBS or lysis buffer depending on the downstream analysis and stored at -80°C.

Transmission electron microscopy
MV and EXO samples (6 μL) were incubated onto glow discharged 200 mesh formvar copper grids (Electron Microscopy Science) for 2 min at 4°C. The grids were washed and blotted dry with filter paper. Next, the grids were negatively stained with 2% aqueous uranyl acetate (Sigma-Aldrich), washed with distilled water and dried in darkness. The grids were visualized using a transmission electron microscope (FEI Tecnai Spirit G2) at 80 kV. Images were taken by a digital camera (Soft Image System, Morada).

Western blot analysis
EVs and cells were lysed with RIPA buffer (Pierce, Thermo Scientific, Rockford, IL, USA) and sonicated three times for five minutes with intermittent vortexing in between. Protein concentration was determined by using the BCA Protein assay following manufacturer´s recommendations (Pierce, Thermo Scientific). Then, 30 μg of protein of cellular lysates and EVs were loaded on a 10% polyacrylamide gel and transferred onto a nitrocellulose membrane (Bio-Rad laboratories, Hercules, CA, USA). Membranes were blocked with 5% Blotting-Grade Blocker Non-Fat Dry Milk The membranes were washed three times with TBST, and incubated for 1 h at room temperature with the secondary antibody (1:10,000) ECL donkey anti-rabbit IgG horseradish peroxidase-linked F(ab') 2 fragment (GE Healthcare, Buckinghamshire, UK), diluted in 0.25% Milk in TBST. The membranes were washed three times with TBST and analyzed with ECL Prime Western Blotting Detection (GE Healthcare) and a VersaDoc 4000 MP (Bio-Rad Laboratories).

RNA extraction and profiling
Total RNA from cells and EV subpopulations was isolated using the miRNAeasy micro kit (Qiagen, Hilden, Germany), according to the manufacturer´s instructions. RNA samples were eluted in 14 μL of RNAse-free water, aliquoted, and stored at -80°C. The quality and sized of the isolated RNA was measured by capillary electrophoresis using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). RNA from each sample was denatured at 72°C for 2 min and loaded into RNA 6000 Nano and Pico total RNA kits (Agilent Technologies) to analyze RNA profile and concentration. The RNA concentration was used as a normalizing loading factor for all EV samples.

Reverse transcription and pre-amplification
RNA samples were subjected to genomic DNA removal; cDNA synthesis, and instant preamplification using the RT 2 PCR System PreAMP and Human Prostate Cancer Pathway mix (Qiagen, Hilden, Germany), following the manufacturer's recommendations. Next, pre-amplified cDNA was quantified by using Bioanalyzer 2100 and Nanodrop 1000 (Thermo Scientific) and input to the PCR arrays was normalized to 100 ng/μL for all samples.

Reverse transcription quantitative PCR (RT-qPCR)
The PAHS-135Z Human Prostate Cancer Pathway RT 2 Profiler PCR Arrays (SABiosciences, Qiagen) were used to analyze all vesicle samples, following the protocol´s instructions. For each sample type, three assays were carried out as independent biological replicates. In summary, 100 ng/μL of pre-amplified cDNA from LNCaP-, PC-3-and PNT2-derived MVs and EXOs were mixed with the RT 2 SYBR Green Mastermix and RNAse-free water. Aliquots of the PCR component mix (25 μL) were distributed across the array. The plates were sealed, centrifuged 1 min at 1,500 x g to remove air bubbles, and run in a CFX96 thermocycler (Bio-Rad). The cycling conditions were as follows: 95°C for 10 min; 40 cycles (95°C for 15 sec, 60°C for 1:00 min). The Ramp rate between the 95°C to 60°C step was 1°C/sec and the same threshold was used across the arrays. Each array contained inter-plate and reverse transcription calibrators as well as a gDNA contamination control.

mRNA data analysis
Raw cycle threshold (Ct) values were exported and analyzed by using the PCR Array Data Analysis Software version 3.5, provided by SABiosciences. Gene mRNA level was related to the mean mRNA levels of all the genes present across the samples. Only Ct values < 35 were included in the analysis. Calculations of relative expression were performed with the 2 −ΔΔCT method [21]. Student's t-test was used to calculate the P values of the replicate 2 −ΔCt values for each gene in the control and sample groups. P values of less than 0.05 were considered statistically significant. A fold-change threshold of 10 was used for the stringent analysis. Results are shown as the mean ± SEM of three samples for each condition in relation to the mean ± SEM of three control samples for each group. All statistical analyses were performed using the statistical software package, GraphPad Prism 7.0 (GraphPad Software, Inc., San Diego, CA, USA).

Differential RNA profiles in the subpopulations of EVs from PCa and non-cancerous cells
Using differential centrifugation, we isolated both MVs and EXOs from two common PCa cell lines (LNCaP and PC-3), and a benign prostate epithelial cell line (PNT2). Transmission electron microscopy characterization showed that MVs were more heterogeneous in size and morphology than EXOs, ranging in sizes from > 200 nm, while the EXOs varied between 30-150 nm (Fig. 1a). Additionally, no evident differences were observed in EV morphology among the three cell lines analyzed. The membrane proteins CD81 and flotillin-1 (common EV markers) were detected in both EV subpopulations by Western blotting with a higher enrichment in the EXO fractions compared to MVs. The endoplasmic reticulum marker calnexin was highly enriched in cell samples, compared to EV fractions (Fig. 1b).
To determine whether EVs isolated from the PCa and non-cancerous cell lines had distinct RNA profiles, we analyzed the total RNA content and length from their MVs and EXOs using a Bioanalyzer 2100 instrument (Fig. 1c). MVs from all three samples had clear and distinct peaks at around 2,000 and 4,000 nucleotides (nt) which represent the 18S and 28S ribosomal RNA (rRNA) respectively, with relatively moderate levels of small RNAs. In contrast, EXOs were primarily enriched in small RNAs but also contained larger RNA molecules, especially in the PNT2 EXOs RNA profile. MV and EXO RNAs from both PCa cell lines had similar profiles and size distributions. On the other hand, MVs and EXOs derived from non-cancerous cells had very similar RNA profiles between subpopulations but different RNA profiles compared with the cancer-derived EVs, suggesting that the RNA content of EVs from distinct cellular origins significantly varies.

MVs and EXOs from different cell lines have unique mRNA signatures
As the RNA profiles found in the EV subpopulations seem to represent a diverse selection of RNA species, we decided to focus further analyses on the mRNA content of EVs. For that purpose, we used a human PCa pathway qPCR array approach, which covers 84 mRNA transcripts of genes known to be involved in PCa ( Fig. 2 and Additional file 1: Figure S1). First, we determined the percentage of genes detected in the different groups by the presence of specific mRNAs. Close to 100% of these mRNAs were detected in MVs from both LNCaP and PC-3 cells, whereas less than 50% of the mRNAs studied were identified in MVs from the PNT2 cells. Additionally, the abundance of the different transcripts analyzed was more variable in EXOs than in MVs among the different donor cells, as observed by the Ct values (Additional file 1: Figure S1). Overall, the Ct values obtained from the analysis were lower in the MVs compared to the EXOs in all samples, suggesting an enrichment of these mRNAs in MVs.
To examine whether distinct mRNA patterns were observed in EVs derived from different PCa cells, we compared the transcripts present in LNCaP and PC-3 cellderived MVs and EXOs (Fig. 2). The differences in mRNAs abundances are shown as a scatter plots in Fig. 2a, b, with the cut-off lines being 10-fold change. We used this strict limit to focus on transcripts with obvious differences among the samples. The mRNA transcripts from ten genes were shown to be unique for LNCaP MVs, and ten different mRNA transcripts were also specific for PC-3 MVs (Fig. 2c). On the other hand, six transcripts were unique for LNCaP EXOs, and seven other mRNA transcripts were exclusive of PC-3 EXOs (Fig. 2d).
Among the commonly found mRNAs that were differentially present in both MVs and EXOs for each cell line, LNCaP samples showed significant increases of kallikrein-related peptidase 3 (KLK3) (also known as prostate-specific antigen: PSA) related to PCa neoplastic growth and metastasis [22]; NK3 Homeobox 1 (NKX3-1), an androgen-regulated transcription factor indicating poor prognosis [23]; the transmembrane protease serine 2 (TMPRSS2), a gene which is up-regulated by androgenic hormones [23]; the tissue inhibitor of metalloproteinase-3 gene (TIMP3), a tumor suppressor gene frequently downregulated in PCa [24]; and tumor protein 53 (TP53), the well-known tumor suppressor gene. In contrast, for PC-3 MVs and EXOs a common increase in the mRNA levels of several genes was found, including caveolin-1 (CAV1), reportedly overexpressed in PCa and associated with disease progression [25,26]; caveolin-2 (CAV2), a gene involved in cell cycle [27]; glutathione S-transferase pi 1 (GSTP1) and cyclin-dependent kinase inhibitor 2A (CDKN2A), both genes showing differentially methylated promoters and involved in cancer progression [28,29]; and tissue factor pathway inhibitor 2 (TFPI2), a gene usually down-regulated in PCa cells [30]. These differences between EVs based on their source of origin defined their specific mRNA signature.

mRNA transcripts associated with PCa can be variably detected in EV subpopulations
A hierarchical clustering analysis of the most significantly differing mRNAs in MVs and EXOs across the groups was performed in order to evidence the abundance pattern of each sample (Fig. 3a). The PCa cellderived EV subpopulations had more similarities with each other compared with the non-cancerous EVs, and the obtained clustering pattern reflected their differences of cellular origin (cancerous vs noncancerous cells) (Fig. 3a). Next, the relative mRNA levels between MVs and EXOs in comparison with their respective EV control from the benign cells were examined. Volcano plots showed significances (as a negative log P values) vs means of differential fold changes for the comparisons of LNCaP MVs vs PNT2 MVs (Fig. 3b), LNCaP EXOs vs PNT2 EXOs (Fig. 3c), PC-3 MVs vs PNT2 MVs (Fig. 3d), and PC-3 EXOs vs PNT2 EXOs (Fig. 3e). Using a cut-off threshold of 10-fold change and a P value of 0.05, the number of genes classified based on their relative abundance on the study groups compared with the control group was similar [LNCaP MVs: 21/9, LNCaP EXOs: 4/3, PC-3 MVs: 25/12, PC-3 EXOs; 5/3 (up/down)].
Comparison of MVs and EXOs showed that MVs were enriched in the studied mRNAs compared to the EXOs for both cell lines (Fig. 3b-e). Furthermore, the statistically significant mRNA transcripts in EXOs samples were also present in the MVs. In view of both the RNA concentration and the transcript enrichment, MVs were selected to further study the differences in the mRNA levels known to be associated with PCa.
Next, we determined which of the common mRNAs for both the PCa MVs had statistically significant different levels compared to the control PNT2 MVs (Table 1). PCa vs control comparisons revealed some startling increases of mRNA levels of: ETV1, a gene that directs androgen metabolism and confers aggressive PCa [23]; FASN, a fatty acid metabolism gene highly up-regulated in PCa [31]; RBM39, a gene implicated in colorectal and breast cancer progression [32,33]. On the other hand, we also observed decreases in the mRNA levels of genes such as CCND2, a crucial cell cycle-regulatory gene down-regulated in PCa cells [34]. Interestingly, the levels of the transcription factor ERG whose up-regulation is a poor prognosis indicator [23], and GNRH1, HAL, and LGALS4 genes were decreased in the PCa-derived MVs.

Differential expression of the mRNAs in other EVs, prostate cancer cell lines, and prostate biopsies
To evaluate the presence of the common PCa mRNAs identified in LNCaP and PC-3 MVs in EVs from different Bars represent the mean ± SEM of three independent experiments per group. *P <0.05, **P <0.01, ***P <0.001. d Venn diagram representing a summary the statistically significant mRNAs of the analyzed genes after comparison with the Ctrl, exclusively present in LNCaP MVs and PC-3 MVs. The mRNAs that were significantly common for both samples are also represented sources, a comparison was performed with all the studies collected in EVpedia were these mRNAs were identified in EVs [35] (Fig. 5a). A very limited number of reports (2-4 out of 11) identified the mRNAs in EVs from different cancer and non-cancerous cell sources, with the exception of FASN, which was found in 6/10 studies. Interestingly, LGALS4 mRNA has been first identified and validated in this study. To further assess whether the statistically significant common mRNAs found in PCa MVs were cellline specific and represented specific PCa signatures, we compared them with a publically available dataset (GSE36720) in the GEO database of mRNAs expressed in different PCa cell lines (Fig. 5b). CDH1, NKX3-1, and TP53 genes were over-expressed in LNCaP cells compared with PC-3 cells. Contrary, CAV2, GSTP1, TFPI2, and ZNF185 genes were over-expressed in PC-3 cells in comparison with LNCaP cells, and ETV1, FASN, CCND2, ERG, GNRH1, HAL, and LGALS4 genes exhibited similar expression in both cell lines. Finally, as a proof of concept to evaluate the possible clinical relevance of the common mRNAs for both PC-3 and LNCaP MVs, the results were compared with a publically available microarray dataset in the GEO database containing information of mRNA expression on benign and malignant prostate tissue (GSE55945) (Fig. 5c). Here, some of the significant common mRNAs discovered in the PCa MVs, including ETV1 and FASN genes, were significantly differentially expressed in the malignant prostate tissue in comparison with the benign tissue. This finding suggests that the mRNAs from ETV1 and FASN genes found to be significantly enriched in MVs from LNCaP and PC-3 cells correlated with prostate cancer in patient samples, demonstrating the possible future applicability of the mRNA analysis using MVs.

Discussion
In this study, we analyzed the mRNAs of 84 genes known to be involved in PCa from EV subpopulations derived from human PCa cell lines. LNCaP and PC-3 cells were chosen for EV isolation due to their extensive use as PCa cell models. To compare the relative abundance of mRNAs in EVs, the same subpopulations were isolated from PNT2 cells, a benign human prostatic epithelial cell line [36]. Differential centrifugation, the gold standard technique in the EV isolation [37], allowed the separation of enriched MVs and EXOs fractions, as previously reported [20,38]. Despite its drawbacks, differential centrifugation is still the only technique that results in the separation of enriched, larger and smaller EV subpopulations albeit overlapping and nonhomogeneous. PCa MVs and EXOs carried common EV proteins irrespective of being derived from malignant or non-malignant cells, and they had similar morphology and size range independent of their cellular origin, as described by our group [5,6,20,38] and others [7,39].
EVs from both normal and cancer cells can carry both intact and fragmented mRNA with a size range between 25 and 700 nt, which is considerably different from the intact cellular mRNAs, sized between 400 nt to 12,000 nt [40][41][42]. The presence of 18S and 28S rRNA in EVs has been reported rather ambiguously. In our study, and consistent with what others have previously published, MVs contained significantly more rRNA than EXOs, which were primarily enriched in smaller RNAs of < 700 nt, despite the presence of larger RNA fragments [5][6][7]. However, the PNT2-derived EXOs had a bigger proportion of rRNA than the EXOs from PC-3 and LNCaP cells, which may imply that specific RNA sorting signals and mechanisms exist for distinct subtypes of EVs or originator cells [14,16,[43][44][45][46].
RNA signatures of EVs can be specific for the cells of their origin [36,42] discriminating e.g. the cancerous transformation status of cells. In this study, MVs were found to be significantly enriched in several of the mRNAs analyzed when compared to EXOs. Based on these results, MVs rather than EXOs might better represent a Log 2 fold-change (mean of the differential mRNA levels); P value for the differential mRNA levels population of vesicles with an advantage for translational studies. Since MVs are larger vesicles, they might also contain more tumor-derived molecules including metabolites, proteins, and nucleic acids. Moreover, the RNA profile of MVs seems to be more similar to the donor cells than that of EXOs [5,6], thereby giving a snapshot of the tumor transcriptome. This is an important finding, since the emphasis of EVs in cancer research has so far heavily been on EXOs and only a few studies have highlighted the advantages of larger vesicles, including large oncosomes [47]. However, more research is needed in order to elucidate whether tumor cells modulate the EV cargo at different stages of the disease and how the specific sorting of molecules into EVs takes place.
Here, the relative abundance of mRNAs was analyzed for different types of genes associated with PCa: differentially methylated promoters, androgen-signaling pathway, PI3 kinase/AKT and PTEN signaling pathways, apoptosis, cell cycle, and metastatic potential. Among them, statistical significant mRNA levels of 16 genes were at least 10-fold differentially regulated in the LNCaP MVs and 32 in PC-3 MVs when compared to the control PNT2 MVs. Interestingly, mRNAs sequences of only three genes were uniquely increased in LNCaP MVs in comparison with PNT2 MVs. These included the NKX3-1 gene, which is one of the earliest markers for prostate development during embryogenesis and is a regulator of cell proliferation, differentiation, and apoptosis [23]; the TP53 gene, which is well-known for tumor recurrence and metastasis [48], and the androgen-regulated TMPRSS2 gene, which is highly expressed in PCa [49,50]. In contrast, eight mRNAs were exclusively present in PC-3 MVs, some of them include CAV2 gene, a cell cycle regulator involved in PCa progression [27]; the GSTP1 gene, whose hypermethylation is one of the most frequently observed aberrations in PCa [28]; PES1, CAMSAP1, and CCND1 genes, up-regulated or correlated with PCa development [51][52][53] and the ZNF185 gene, which is frequently reported to be dysregulated in PCa [54]. Overall, the unique mRNA signature of LNCaP and PC-3 MVs allows the discrimination c Heat map displaying the differential expression of the mRNAs of eight genes in malignant prostate tissue as compared to benign prostate tissue. Dataset with accession number GSE55945. Results between different microarray studies are not comparable of the characteristics of their parental cells, such as androgen sensitivity or metastatic potential, since MVs represent their source of cell origin. The common mRNAs in MVs were first identified and validated in PCa EVs in our study, and very limited number of reports previously identified those mRNAs in other cancer-derived EVs [35]. Furthermore, another interesting finding was the possible clinical relevance of ETV1 and FASN mRNA transcripts. As determined by cluster analysis, ETV1 and FASN were differential over-expressed in malignant prostate tissue in comparison with benign biopsies, highlighting their possible relevance as future cancer markers.
The majority of the transcriptomic reports studying EVs utilize microarray analyses and deep sequencing to simultaneously examine the expression of thousands of genes [6,[15][16][17]. While these technologies represent a great opportunity for clinical research, the results need further validation [55,56]. To our knowledge, this is the first study showing and validating the changes in mRNA levels of 84 genes associated with PCa by RT-qPCR in different subsets of PCa cell-derived EVs. The presented results provide important leads to identify specific markers in MVs and EXOs isolated from both cancer cells and biofluids of cancer patients, which could have a utility for cancer diagnosis and response monitoring of cancer treatments.

Conclusions
The mRNA cargo in EVs considerably differs based on the EV subpopulation analyzed and the EV cellular origin, as different PCa cells and benign prostate epithelial cells release very different mRNAs in their respective EVs. The mRNA analyses performed in this study therefore provide new insights into the proportion of the cell transcriptome that can be detected within EVs. The data also emphasize the need to better dissect the role of EVs and their cargo in cancer cell phenotype, putatively in relation to progression of disease. Since the molecular cancer signatures can be identified in EVs, their analysis can contribute to the understanding of differences in PCa heterogeneity. Whilst many previous studies have focused on EV-associated miRNAs, this work demonstrates the potential of the selective analysis of the mRNA content of MVs and EXOs and highlights their applicability in biomarker research.