Impact of Extracellular Vesicle Isolation Methods on Downstream miRNA Analysis in Semen: A Comparative Study

2.1. Characterization of Nanovesicles by Size and Concentration

In order to characterize the different exosome-EV isolation methods, we first performed an evaluation of sEVs by measuring size and concentration of the isolated particles from healthy control (HCt) semen samples (Table 1).

The profiles of the particles isolated by four isolation methods with alternative precipitation conditions, compared to ultracentrifugation (UC) as the reference method (Figure 1), were evaluated.

Nanoparticle tracking analysis (NTA) (Figure 2) provided evidence that isolated vesicles were within the expected size range for exosomes (50–250 nm), although an increased mean size and mode size was obtained for those extracted with ExoQuick ULTRA (A or B alternative protocols) kit, miRCURY Cell/Urine/CSF 10,000× g and miRCURY Serum (Supplementary Materials Table S1A,B) when compared with the UC method, so these methods resulted in a rather different distribution of undersized particles. ExoGAG and miRCURY Cell/Urine/CSF 1500× g exosome-EV isolation methods displayed very similar size profiles, including size mean and mode, to the ones obtained with the UC method (p > 0.05) (Supplementary Materials Table S1B, Figure 2A). In detail, ~90% of vesicles isolated by these three methods peaked below 250 nm, the average modal size (defined as the most frequently occurring EV size) being between 114–119 nm.

Additionally, NTA analysis revealed differences in the recovery rates for the methods used. Firstly, the miRCURY S/P (protocol Plasma) resulted in a low number of vesicles, below the recommended NTA detection levels (1 × 10⁷ particles/mL), making it the least efficient EV isolation method for SP. The ExoQuick Ultra_S/P protocol A (which includes the use of purification columns after the precipitation protocol) generated a low particle yield, both at 1500× g and 3000× g, isolating 266 and 25-fold fewer particles than the reference UC method, respectively, followed by the miRCURY S/P (protocol Serum) (Figure 2B) obtaining 2.8-fold fewer particles compared with UC. This suggests that the modifications made to the manufacturer’s indications have not been enough to provide a protocol with efficient EV isolation for SP. In contrast, the miRCURY Cell/Urine/CSF method, both at 10,000× g and at 1500× g, gave the best recovery rate for isolated vesicles from semen, resulting in 1.5–2-fold more particles respectively (Figure 2B), making it the most efficient EV isolation method.

Characterization of sEVs by their protein composition was additionally performed. Flow cytometry analysis of tetraspanins, as the most abundant proteins on EVs, showed positive labeling of CD63 and CD81 (Figure 3A) on vesicles isolated with either UC, ExoGAG 3500× g, miRCURY Cell/Urine/CSF 1500× g exosome-EV isolation methods (selected as they displayed very similar size profiles in the NTA analysis), although differences in the labelling signals are observed among them (Figure 3B).

Specifically, all samples were contrasted with antiIgG as unspecific marker (Figure 3A, upper panel). EVs isolated by ExoGAG 3500× g resulted in a positive CD63 and CD81 labelling represented by a displacement of fluorescence peak curve (Figure 3A, left panel). EVs purified by the UC technique showed a similar profile of CD63 and CD81 labelling with a displacement of fluorescence peak curve (Figure 3A, middle panel). However, EVs isolated by miRCURY Cell/Urine/CSF 1500× g exosome isolation kit only showed a clear positive CD81 labelling represented by a displacement of fluorescence peak curve (Figure 3A, right panel), and the CD63 fluorescence intensity was not enough to be considered as positive signal (CD63: IgG fold-change fluorescence intensity <2) (Figure 3B).

2.2. Nanovesicle RNA Quality and Quantity

Quality and quantity of RNA obtained from the nanoparticles was also assessed to evaluate the influence of each of the exosome-EV isolation methods on the RNA after EV recovery. Consistent with our results on EV characterization, RNA quality from the ExoGAG 1500× g; ExoGAG 3500× g; and miRCURY Cell/Urine/CSF 1500× g exosome-EV isolation methods displayed similar 260/280 ratios (1.76 ± 0.05; 1.79 ± 0.08 and 1.63 ± 0.11 respectively) to the UC technique (1.78 ± 0.17). Interestingly, the miRCURY Cell/Urine/CSF 1500× g exosome-EV isolation method gave a higher RNA concentration (Table 2), this is probably related to the higher vesicle yield obtained by this method (Figure 2); on the contrary the application of high speed (10,000× g) in the miRCURY Cell/Urine/CSF procedure resulted in a low concentration of RNA of bad quality.

Given the limited number of vesicles obtained by the miRCURY S/P (protocol Plasma), we were unable to extract sufficient RNA for RNA analysis, and thus this method was definitively excluded from further analyses.

2.3. Pre-Study miRNA Expression Pattern

Next, we assessed whether SP sEV RNA of similar quality, which had been obtained from the different EV isolation methods, would result in similar miRNA quantitative real-time PCR (qPCR) amplification patterns from samples in physiological conditions. Accordingly, we subsequently analysed three miRNAs, which had previously been described as presenting quantifiable levels in semen exosomes (miR-30d-5p, miR-93-5p and miR-449a [7]), in samples from healthy control individuals. Although an equivalent amount of RNA from each of the EV isolation approaches was used, an increase of at least four raw Cp cycles was measured for the miRCURY Cell/Urine/CSF at 10,000× g, probably due to the poor quality of the RNA obtained with this approach. Slight differences in the Cp values were found among the rest of the different methods of exosome extraction (data not shown) and specifically among the most efficient exosome-EV extraction methods (Figure 4), that is UC, ExoGAG 3500× g and miRCURY Cell/Urine/CSF 1500× g.

Thus, according to the vesicle measurements, RNA assessment and miRNA quantification data, we chose the ExoGAG 3500× g and the miRCURY Cell/Urine/CSF 1500× g EV isolation methods for the next phase of the study, in which they were tested as equivalent alternative methods to UC for isolating exosomes/sEVs from semen.

2.4. Exosome/sEV-miRNA Expression Pattern in Prostate Cancer

In order to assess whether EV isolation methods influence the results of the miRNA analysis in semen for PCa, we analysed the expression of 16 human miRNAs (Supplementary Materials Table S2) in SP vesicles obtained by using the three exosome isolation methods mentioned above (UC, miRCURY Cell/Urine/CSF 1500× g and ExoGAG 3500× g methods). All of these miRNAs had been previously tested in semen exosomes isolated by UC [7,8] and some of them had been found to be differentially expressed in the prostates of PCa individuals (Supplementary Materials Table S2). MiRNA levels were assessed in semen vesicle RNA samples from 5 healthy control (HCt), 5 benign prostatic hyperplasia (BPH) and 9 prostate cancer (PCa) individuals (Table 1).

First, our results showed that miRNA levels showed a slight variation pattern in the different patient and control groups between the three methods. Specifically, the expression values for miR-142-3p, miR-196b-3p, miR-30c-5p and miR-34a-3p were statistically different between the PCa and HCt samples from vesicles obtained with the UC technique. None of the miRNAs presented differences in expression between the PCa and HCt groups from miRCURY purified EVs, whereas miR-142-3p and miR-92a-3p were found to be differentially expressed between PCa and HCt samples from ExoGAG extracted vesicles (Table 3; Figure 5). All of the differentially expressed miRNAs presented a fold-change increase or decrease of >1.5 between PCa and HCt samples. Thus, there is only one differentially expressed miRNA, miR-142-3p, between PCa and HCt EV samples for both the UC and ExoGAG methods.

Interestingly, we observe similar expression behaviour for some miRNAs tested when we used the different EV isolation methods. Some miRNAs, such as miR-142-3p, miR-142-5p and miR-223-3p showed the same trends in level of expression for all three methods (Figure 5), although the differences between PCa and HCt samples were not statistically significant for the three procedures. In contrast, some miRNAs were seen to be over-represented in vesicles obtained by one of the isolation methods. This is the case for miR-663b in vesicles extracted by ExoGAG (Figure 5), which indicates that the method has some kind of selectivity for vesicles containing this miRNA.

2.5. Diagnostic Performance of Combined sEV-miRNA-Based Diagnostic Classifiers for PCa

In our previous study [8], we showed semen exosome miRNA signatures that combined with blood Prostate specific antigen (PSA) concentration, can be used as molecular biomarkers with the potential to improve PCa diagnosis efficiency of PSA. The miRNAs included in the models (miR-142-3p, miR-142-5p and miR-223-3p) were obtained from sEVs isolated by the ultracentrifugation procedure. Thus, in the present study we further assessed whether these miRNA models were useful as biomarkers from sEVs obtained from the different EV isolation methods.

As mentioned above, miR-142-3p, miR-142-5p and miR-223-3p were over-expressed in PCa and BPH sEVs, showing the same trends in the level of expression for all three methods (Figure 5), although the differences between PCa and HCt samples were not statistically significant for the three procedures. Accordingly, the predictive accuracy of the miRNA multiplex model (miR-142-3p + miR-142-5p) was not statistically significant to discriminate PCa from non-PCa samples (Table 4), and the (miR-142-3p + miR-142-5p + miR-223-3p) model only showed diagnostic efficiency for the miRCURY procedure (p = 0.014) to discriminate PCa from BPH samples (Table 4). However, when PSA was introduced in the miRNA-based model, it resulted in higher predictive accuracy than PSA as biomarker as previously described [8], both to discriminate PCa from nonmalignant individuals (miR-142-3p + miR-142-5p + PSA model: AUC > 0.85, p < 0.009) and to discriminate benign from malignant tumours (miR-142-3p + miR-142-5p + miR-223-3p + PSA model: AUC > 0.933, p < 0.009) for the three sEV extraction methods (Table 4). Interestingly, a better use for diagnosis of these combined miRNA-PSA models than PSA as biomarker was also obtained, resulting in a higher sensitivity and specificity to discriminate not only PCa from healthy individuals, but, more clinically relevant, to discriminate malignant from benign tumours (Table 4) for the three sEV extraction methods.

4.1. Subjects of Study

Semen specimens were collected from healthy individuals without tumours (control group 1: HCt; n = 12) consulting for vasectomy or infertility at the Andrology Service of Fundació Puigvert and individuals consulting for diagnosis of PCa who underwent routine prostate screening including PSA testing and DRE (digital rectal examination) at the Urology Service of the Bellvitge Hospital; only patients that presented moderate PSA levels (4–18 ng/mL) and consented to undergo prostate biopsy were selected. The patient group (PCa) comprised 9 men with biopsy-proven PCa including those who were previously successfully vasectomised (PCa-V, n = 2) and non-vasectomised individuals (PCa-noV, n = 7). Additionally, samples from 5 non-vasectomised individuals with benign prostatic hyperplasia (BPH) or prostate enlargement (control group 2: BPH) who presented elevated PSA levels (>4 ng/mL) but no detectable cancer on biopsy (Table 1), were collected. The study was approved by the Ethical Review Board of both institutions, Fundació Puigvert and Bellvitge Hospital (Feb 2015, 2018). The participants signed an informed consent form. All methods were performed in accordance with the relevant guidelines and regulations.

Samples were analysed in two phases: the pre-study, to optimize and select the proper exosome-EV isolation methods for semen, and the study, to analyse the miRNA expression levels in the different groups of study and compare the results among the suitable EV isolation methods for semen. The pre-study included 7 HCt samples, whereas the study included 5 HCt, 5 BPH and 9 PCa samples (Table 1). In order to avoid miRNA expression differences due to inter-individual variations, the same semen samples were used for the different isolation methods within the pre-study as well as within the study phase.

4.2. Sample Collection and Processing

Semen samples were obtained by masturbation after 3–5 days of sexual abstinence. They were allowed to liquefy for 30 min at 37 °C. Samples were centrifuged twice (1600× g for 10 min, then 16,000× g for 10 min) at 4 °C to obtain seminal plasma (supernatant) as previously described [7,8,21]. Seminal plasma (SP) was carefully collected and immediately stored at −80 °C until use.

4.3. Exosome/sEV Isolation (Figure 1)

SP (200 µL) was first passed through a 0.22 μm filter to enrich for sEVs [6]. The resulting filtrate was processed by either ultracentrifugation, three commercial exosome-EV isolation reagents [miRCURY^® Exosome Cell/Urine/CSF kit (Qiagen, NV; Germany), miRCURY^® Exosome Serum/Plasma kit (Qiagen, NV; Germany), ExoQuick^® ULTRA EV Isolation kit for Serum and Plasma (System Biosciences, Palo Alto, CA, USA)] or a non-commercial exosome-EV isolation reagent [ExoGAG (NasasBiotech, Santiago de Compostela, Spain)] previously tested for isolating plasma EV from endometrial cancer patients [22]. They all were precipitation-based methods, although ExoQuick^® ULTRA EV Isolation kit for Serum and Plasma can also include a final solid extraction phase (precipitation + column-based method). Different experimental conditions were tested for each isolation reagent (Figure 1). Ultracentrifugation (UC).

The filtered SP fluid plus 9 mL of PBS was ultra-centrifuged at 100,000× g in a SW40 rotor for 2 h at 4 °C to sediment the sEVs, which mainly contain exosomes, as described elsewhere [17]. The pellet was resuspended in 100 µL PBS and frozen at −80 °C.

• -

  miRCURY ^® Exosome Cell/Urine/CSF kit (miRCURY Cell/Urine/CSF)

The exosomal fraction from SP filtered fluid was isolated according to the manufacturer’s recommendations. Briefly, filtered fluid was supplemented with PBS till 1mL and subsequently, 400 µL of Precipitation buffer B was added. Samples were incubated at 4 °C for 1 h. The mixture was centrifuged at either 10,000× g (manufacturer’s protocol) or 1500× g for 30 min at room temperature. Pellet fraction was resuspended in 100 µL of Resuspension buffer and frozen at −80 °C.

• -

  miRCURY ^® Exosome Serum/Plasma kit (miRCURY S/P)

Both protocols (for serum and for plasma) were independently followed. SP filtered fluid was supplemented with PBS till 0.5 mL. Then the corresponding amounts of reagents were added according to the manufacturer’s instructions. Samples were incubated at 4 °C for 1 h and centrifuged at 1500× g 30 min (protocol serum) or 500× g 5 min (protocol plasma) at room temperature. Pellet fraction was resuspended in 270 µL of Resuspension buffer and frozen at −80 °C.

• -

  ExoQuick^® ULTRA EV Isolation kit for Serum and Plasma (ExoQuick Ultra_S/P)

The corresponding amounts of ExoQuick^TM EV isolation reagent was added to 0.2 mL of filtered fluid according to the manufacturer’s instructions. Samples were incubated at 4 °C for 30 min or overnight. The seminal plasma/ExoQuick mixture was centrifuged at 1500× g [23] or 3000× g for 10 min at 4 °C and the pellet resuspended in 100 µL PBS (protocol B: precipitation) and frozen at −80 °C. For several samples, a column step was additionally followed (protocol A: precipitation + column) for comparison.

• -

  ExoGAG

A suitable volume of precipitation reagent A of ExoGAG kit was added to 0.2 mL of filtered seminal plasma (reagent A/sample ratio 2:1) [22]. The mixture was incubated for 5 min at 4 °C and subsequently centrifuged at either 1500× g or 3500× g for 30 min at 4 °C. The pellet was resuspended in 100 µL PBS and frozen at −80 °C.

4.4. Characterization of EVs

Vesicles were characterized in terms of physical properties (size, concentration) by Nanoparticle tracking analysis (NTA) using the NanoSight NS300 (Malvern Instruments).

An additional flow cytometry analysis (BD FACSAria^TM IIu, BDbiosciences, SanJosé, CA, USA) was performed and tetraspanins CD63, and CD81 protein levels were analysed as the most usually employed markers for EV characterization [19,24]; their presence demonstrates the lipid-bilayer structure specific of EVs. Semen EVs isolated by UC, ExoGAG or miRCURY cell/urine/CSF were incubated for 1 h with antiCD81 (1:50, sc7637, Santa Cruz Biotechnology, Santa Cruz, CA, USA), antiCD63 (1:50, sc5275, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and antiIgG (1:50, 400120, BioLegend) as negative control, followed by anti-mouse Alexa488 (1:1000, ab150113, Abcam, Cambridge, UK) as the secondary antibody as described [22]. Quantification of fluorescence was calculated as the fold-change of fluorescence intensity median with respect to IgG (immunoglobulin G) negative control. Positive signals were considered as those with a fold-change fluorescence intensity of two or higher than two.

4.5. Small RNA-Containing Total RNA Isolation from Semen Exosomes/sEVs

With the purpose of degrading the residual RNA outside the vesicles, the vesicle suspension was treated with RNAse A (Qiagen; NV, Germany) (100 μg/mL final reaction concentration; 15 min at 37 °C). Total RNA was obtained from exosomes/sEVs using the miRNeasy Micro Kit (Qiagen; NV, Germany). RNA concentration was calculated by using the QUBIT fluorometer and the Quant-iT RNA Assay kit (Invitrogen; CA, USA). RNA quality was determined by evaluating the OD 260/280 nm ratio when using a Nanodrop UV-Vis spectrophotometer (Thermo Fisher Scientific; MA, USA).

4.6. Exosomal miRNA Quantitative Real-Time PCR (qPCR)

First-stranded cDNA specific for miRNA was obtained by reverse transcription (RT) of 50 ng of exosomal RNA in 10 µL, using the miRCURY^® LNA^® RT Kit (Qiagen; NV; Germany). For qPCR analysis, cDNA was diluted (12×) and assayed in 10 µl PCR reactions containing miRCURY LNA SYBR^® Green PCR Kit (Qiagen; NV; Germany). Duplicate amplification reactions of 16 individual miRNA assays (LNA™-enhanced miRNA qPCR primers; Supplementary Materials Table S2), including miRNAs previously described as altered in PCa exosome semen samples [8], were carried out on a Lightcycler^® 96 Instrument (Roche; Basilea, Switzerland). Target miRNA expression in semen sEV samples was calculated relative to the mean expression value of the most stable assays as normalizers, to correct for potential overall differences between the samples. Normalizers (miR-576-5p and miR-181c-5p) were selected because they presented the lowest coefficient of variation (CV < 0.030) of Cp values among the samples in the study (Table 3), as well as having no statistical differences in absolute expression levels between groups from all the extraction methods used (Table 3). The relative quantification (RQ) miRNA expression values were calculated using the 2^dCp strategy.

4.7. Statistical Analysis

Statistical analysis was carried out with nonparametric tests such as the Kruskal–Wallis and Mann–Whitney tests to evaluate differences among the groups of study in all exosome isolation methods.

Receiver operating characteristic (ROC) curve analysis of the RQ values was used to distinguish the samples showing malignancy of prostate tumour. Accuracy was measured as the area under the ROC curve (AUC) [25]. The threshold value was determined by Youden’s index, calculated as sensitivity plus specificity-1 [26]. A multivariate binary logistic regression analysis was used to assess whether previously described miRNA combined models [8] were useful as PCa biomarkers from sEVs obtained from the different EV isolation methods. An enter method was used to include in the equation all the variables of each nested model. The binary logistic regression model provides the following estimation of the logit function:

where p = P (presence of prostate cancer), Logit (p) = log(p/(1 − p)) = log(odds), B = logOR and Xn = the expression values of the miRNAs. Therefore, if we use this estimated model as a prediction model, with the standard classification cutoff of 0.5, we would classify individuals with a positive Logit function estimation as “positive for PCa” and individuals with negative Logit function estimation as “negative for PCa”.

Only p-values ≤ 0.05 were considered significant. All data analyses were performed using SPSS software, version 15 (SPSS Inc, Chicago, USA).