EVs have been postulated as a valuable source of potential biomarkers in PCa that at some point would complement or replace the routine diagnostic procedures [15, 27]. Urinary EVs take special relevance since their cargo reflect changes in the cellular biology of the tumour in a specific point of its progression and can be isolated without any invasive procedure. However, translation of these biomarkers into the clinical setting is not exempt of limitations, including the irreproducibility of results as one of the most important . In this sense, EVs isolation and characterization approaches still constitute a scientific challenge . Hence, with the aim of deepening in the knowledge of EVs isolation methods, we have evaluated three different methodologies including the classical UC and SEC, as well as a commercial kit (Exolute®). EVs characterization was performed using NTA, TEM, spectrophotometry (Nanodrop), AlphaScreen™ Technology and whole miRNA transcriptome expression analysis with the EdgeSeq System (HTG Molecular Diagnostics). For this purpose, a series of urines collected from 10 individuals (6 PCa patients and 4 HDs) were analysed with each isolation method.
Absorbance measurements showed similar protein concentrations in UC and SEC, however a significant increase was noticed in Exolute® samples (Figure S1A). This effect can be explained by TEM analysis, in which preparations from EVs obtained with Exolute® showed a background of precipitated proteins (Fig. 2A) consistent with the presence of protein aggregates, as it has been previously described . NTA evaluation showed that UC provided the higher number of particles per mL and particles per µg protein ratio, in comparison with SEC and Exolute®, suggesting a higher EV yield obtained by this technique (Figs. 1A and 1C). Besides that, a significant increase in size of particles obtained by SEC was noticed (~ 190 nm), when compared to the other EVs isolation procedures (~ 165 nm) (Fig. 1B). However, no differences in size were appreciated when EVs were analysed by TEM with a median in size of around 90 nm (range: 30–200 nm). Interestingly, approximately 95% of EVs were ranged from 30–120 nm, and around 40% were between 60–90 nm of diameter (Fig. 2C). These findings correlate with previous reports showing that the size of urine EVs varies from 30 to 100 nm [30–33]. Discordances between NTA and TEM herein reported may be due to different aspects including: the difficulty of NTA to resolve EVs aggregates (a correct dilution of the sample is crucial to avoid this) ; the limitation of NTA in detecting particles which dimeter is lower than 100 nm  and finally, to the size overestimation of NTA [36, 37]. Additionally, and as mentioned above, TEM analyses revealed high variability in EVs yield obtained by Exolute®, with some samples showing the presence of protein aggregates that would explain the highest protein content of the spectrophotometric analysis (Fig. 2A). Interestingly, SEC-isolated EVs had a broader size distribution than those isolated with Exolute® and UC, which could be due to the growing evidences that SEC minimally alters the physical properties of EVs, whereas UC might cause vesicle rupture or fusion with proteins because of the high speed used in centrifugations . As a whole, no significant differences in particle concentration and size (measured by both NTA or TEM) were detected between PCa samples and HDs (Fig. 1D) which is in accordance with previous studies .
Once characterized through NTA and TEM, EVs were analysed with AlphaScreen™ Technology, a strategy lately used to improve the typical immunoassays  through the simultaneous detection for two specific EV-tetraspanins, CD9 and CD63[22, 24] and designed to specifically detect EVs lower than 200 nm. Our AlphaScreen data revealed that all three isolation methods obtained CD9 and CD63 positive EVs, as can be appreciated in Fig. 3A where luminescent signal was higher in purified EVs compared to crude urine. Furthermore, SEC provided the highest luminescent intensity followed by UC and Exolute®, which luminescent intensity was significantly lower. Interestingly, a high signal variation was noted among the analysed cases, especially in Exolute® and UC isolated EVs, for which in some cases the luminescent signal was low or null, whereas SEC isolated EVs provided a measurable signal in most of the cases (Fig. 3B). These differences would be related to those herein noticed with regards EVs size distribution, or as suggested by some reports, as consequence of differences in the membrane proteomic content of small and large EVs . Moreover, EVs rupture due to UC high-speed centrifugations or the reported variability on EVs yield depending on the equipment and operator technical variability could have affected the results . Hence, and according to the AlphaScreen™ Technology, our results highlight SEC as the most efficient method to isolate CD9 and CD63 positive microvesicles, followed by UC and Exolute®, respectively. Similar to the other characterization methods, no differences of luminescent signal were appreciated between EVs isolated from PCa patients and HDs for any of the isolation methods tested.
Forward characterization of EVs was carried through a whole miRNA transcriptomic analysis by using one of the newest and most reproducible RNA quantification platforms using the EdgeSeq Technology (HTG Molecular Diagnostics), currently used in many studies [42, 43]. Among the advantages of this system are that it does not require an RNA-extraction step which, reduces the extraction-associated data bias and sample loss; and the low input of sample necessary for being analysed . Our results have shown that the sum of the normalized miRNA counts was higher in EVs isolated from SEC, followed by UC and Exolute® (Fig. 4), suggesting that the isolation methods influence on the yield of the transcriptomic analysis. Many -omic studies have been found to be highly dependent on the EVs isolation procedures, so that different methods produce EVs and EV sub-fractions of variable homogeneity, which makes difficult to extrapolate findings between different studies of EVs . This is in line with the results we have obtained, in which samples were classified based more on the EVs isolation method than on their origin (PCa patients or HDs) (Figure S3). Despite this, correlation of miRNA profiles between the different isolation approaches was high in all cases, being better between UC and SEC (Table 2). Interestingly, when miRNA expression was divided into quartiles, the best correlation coefficients (R2) were obtained in q4 (higher number of miRNAs) with a R2 = 0.97 between SEC and UC, the correlation in q2 and q1 being lower (Figure S15). In order to demonstrate that the isolation procedures influence at biological level, two sets of RNA were evaluated: the housekeeping genes provided by the assay and the Let-7 family of miRNAs. This approach showed significant differences with regards number of reads for any of the two RNA sets of UC and SEC with Exolute® (Fig. 5), indicating that Exolute® provides the lowest performance from the biological point of view.
DEA between PCa cases and HDs for each EVs isolation method showed that a total of 21 and 3 miRNAs were differentially expressed between groups for SEC and UC respectively, and none for Exolute®. The only miRNA that was differentially expressed in both methods was miR-8052 , a miRNA not previously described in PCa but in serum from sepsis patients with different outcomes . Remarkably, two of the 3 miRNAs differentially expressed in EVs isolated by UC from urine of PCa patients were miR-142-5p and miR-223-3p, two miRNA that have been recently described in EVs from urine isolated also by UC as non-invasive PCa diagnostic biomarkers .
As a whole, our results point to what other authors have suggested, the need of methodological standardization of EVs isolation and characterization in order to guaranty the success and reproducibility of the subsequent analysis, especially for clinical settings, where a large number of samples should be analysed [48–50]. In this line, we suggest developing a codification system focused on the EV isolation and characterization variables similar to the SPREC codification system for pre-analytical conditions  that we have introduce with our samples (Table 1) and that provides information on the handling of biological specimens before analysis, another critical point that not always comprehensively considered [52, 53].