Exploring Small Extracellular Vesicles for Precision Medicine in Prostate Cancer

Tumor microenvironment constitutes a complex network in which tumor cells communicate among them and with stromal and immune cells. It has been shown that cancer cells are able to exchange genetic materials through small extracellular vesicles (EVs), a heterogeneous group of vesicles with different size and shape, cargo content, and function. The importance to investigate populations of circulating EVs would be of great importance as prostate cancer (PCa) biomarkers. In several neoplasms as well as in PCa, nanometer-sized EVs of endosomal origin are implicated in supporting tumor growth and metastatic spread by both altering local stroma cells and creating a protumor environment that favors the formation of pre-metastatic niches. Several techniques are applicable for the isolation and analysis of PCa-derived small EVs and are illustrated in this article. Due to the high sensitivity and specificity of these techniques, small EVs have become ideal candidates for early diagnosis. Moreover, we discuss the role of small EVs during PCa carcinogenesis, as well as in modulating the development of drug resistance to hormonal therapy and chemotherapy, thus underlining the potential of EV-tailored strategies in PCa patients.

bodies with the cell membrane and could be found in blood, urine, and other biological fluids. Urine EVs have been shown to contain RNA, DNA, and miRNA, lipid and proteins/transporters specific to cells of the kidney and urogenital tract. Thus, urine represents a potentially useful noninvasive material for diag nosis and prognosis of genitourinary tract neoplasms (7,8). Due to their involvement in the early phases of tumor development and dissemination, future diagnostic and therapeutic EVtailored strategies are emerging in the armamentarium of cancer patients. In genitourinary tumors, EVs represent potential novel bio markers. Indeed, small EVs have been involved in kidney diseases and cancer, as well as in bladder cancer growth and progression (9). In prostate cancer (PCa), small EVs have been reported to contain high levels of molecules that stimulate cancer cell inva sion, such as integrins β4 (ITGB4) and αvβ6, vinculin (VCL), transmembrane glycoprotein Trop2, vimentin, and Ncadherin, wellknown epithelial-mesenchymal transition markers (6,10).
Furthermore, the efficacy of urine assays in PCa diagnostic could be improved by simple procedures like prostate massage to increase the amount of tumorspecific nucleic acids released (11). This review illustrates the main techniques for isolation and analysis of small EVs and emphasizes their role in the carcino genesis, progression, and drug response in patients with prostate tumor, thus underlying the potential of emerging EVtailored approaches in this setting.

iSOlaTiOn OF PCa SMall evs
The currently adopted methods for small EVs isolation include differential and density gradient ultracentrifugation, sizebased methods (ultrafiltration and sizeexclusion chromatography), polymerbased precipitation, and immunoaffinity capture. The choice of the isolation technique depends on the sample source, the sample amount, the type of the downstream molecular analyses, the desired purity level, and final concentration. Each method shows both advantages and disadvantages, summarized in Table 1. It should be noted that some isolation methods can be more easily integrated with analysis of vesicular RNAs and proteins, so fewer steps are needed. This can therefore improve the reliability and efficiency of small EVs analysis.

Ultracentrifugation
Ultracentrifugation is the most frequently used approach for small EVs isolation, mainly since it does not require high level of technical experience. This technique is based on the different sedimentation rates (depending on size and density) of small EVs from the other sample components (12). Generally, before ultracentrifugation, some steps of lowspeed centrifugation (3,000-20,000 × g) are carried out to clean the sample from large particles, including cells, platelets, apoptotic bodies, and microvesicles. Typically, for ultracentrifugation, the used force ranges from 100,000 to 200,000 × g. Due to the highspeed centrifugation, the small EVs rupture represents a high risk and leads to the yield reduction of intact particles suitable for downstream analyses. It also requires high volumes of samples, it is timeconsuming and not suitable for large number of samples (13).
Moreover, pellet obtained by this method may contain also EV aggregates and protein complexes similar to small EVs in size (14). A particular case regards the uromodulin (THP), which is the most abundant protein in urine and can form a polymeric network that facilitates small EVs' aggregation. Uromodulin removal and the consequent increase of small EVs' yield can be achieved by using the detergent CHAPS or the reducing agent dithiothreitol in urine samples (15). Finally, unlike free RNA that is degraded by urinary RNAses, DNA contamination is frequent, and a DNAse should be used if small EVs' DNA is the target of downstream analyses (16).

Density Gradient Centrifugation
The density gradient centrifugation method allows small EVs separation from other sample components by exploiting their typical density, which ranges between 1.13 and 1.19 g/ml (17). By using preconstructed columns composed of solutions with different densities, after ultracentrifugation small EVs will be at a specific column layer.
It allows purer small EVs isolation than simple ultracentrifu gation; in fact, it is generally used as a second step after other isolation methods to purify the extracts (18,19). However, highdensity lipoproteins represent a contaminant frequently coisolated with small EVs; therefore, plasma and serum are not suitable sample types for density gradient centrifugation. In addition, it cannot distinguish small EVs from other EVs with similar density (20).

Ultrafiltration
Ultrafiltration represents an efficient alternative to ultracen trifugation. It involves the use of nanomembrane filters of poly ethersulfone or polyvinylidene difluoride with an approximately 50-100 kDa molecular mass cutoff (12). Generally, some steps of filtration with membrane filters having larger pores can be performed for the depletion of floating cells and large cell debris (21). Membranes allow the concentration of very diluted samples and therefore they are especially suited for cell culture media and urine samples.
Extracellular vesicles size can be also exploited by the sizeexclusion chromatography that, using sepharose packed columns, allows the separation of the small EVs fraction from biofluids, with high purity and reproducibility (22). However, both ultrafiltration and sizeexclusion chromatography have long run times that limit their scalability for clinical routine laboratory applications.

Precipitation Methods
Some isolation kits for small EVs precipitation are commercially available (23,24). They are based on waterexcluding polymer mixtures that force insoluble components out. When these polymers are added to the sample and after a lowspeed ultracen trifugation (<20,000 × g), they allow the precipitation of small EVs. Since precipitation method is very fast, easy to use, and no equipment is necessary, it is a method scalable for large sample sizes and therefore is suited for the clinical use.
Recently, performances of ultracentrifugation and four dif ferent precipitation methods have been assessed in urine sam ples. In this comparison, a precipitation method outperforms the others in detecting alterations of PCa small EVs' markers (25). However, other EVs are frequently coisolated, making this procedure not so specific for small EVs. Moreover, abundant proteins, including uromodulin, can also be recovered.

immunoaffinity isolation Methods
Finally, immunoaffinity isolation methods exploit the presence of specific proteins in the small EVs' surface (26). Antibodies conjugated with magnetic beads or other materials can recog nize the vesicular antigens and facilitate the precipitation by lowspeed centrifugation or the isolation by magnetic tech niques. This method is highly specific, allowing the isolation of only small EVs by using general small EV surface markers (e.g., CD81, CD9, and CD63) and even only small EV subpopula tions, for example, tumor small EVs by using a tumorspecific marker.
Recently, specific methods for immunobased small EVs isola tion and detection for PCa have been proposed, both measuring total small EVs (27) and exploiting the antiprostatespecific membrane antigen (PSMA) antibody (28). Moreover, it has been observed that immunoaffinitybased methods have better performance than purification methods in isolating PCa small EVs from patient plasma samples (29). However, this generally leads to low yields. In addition, immunoaffinity isolation requires several elaboration steps making it subjected to potential errors by operators.

analYSiS OF PCa SMall evs
After isolation, small EVs quantity, size, purity, and other prop erties can be assessed. A popular method to quantify total small EVs is the nanoparticle tracking analysis (NTA) (12,(30)(31)(32). It also allows the determination of a size distribution profile of particles in solution and, by verifying that the observed distribu tion ranges within typical small EV size, also the purity degree can be estimated. In particular, a laser beam makes particles visible and therefore their concentration can be assessed. The size distribution can be evaluated by measuring the velocity of the Brownian motion of each particle and, according to the Stokes-Einstein equation, speed of particles can be correlated to the particle size. However, protein complexes, lipoprotein particles, and vesicle aggregates similar to small EVs in size can be revealed by NTA and therefore the small EVs concentration could be overestimated. Flow cytometry is another popular method for both small EVs quantification and characterization (33). In particular, particles are illuminated by a laser beam, and the scattered light is detected. However, currently, there is a detection limit of 150 nm; therefore, the use of latex or magnetic beads coupling with small EVs is suggested to make these aggre gates visible by flow cytometry (34). Flow cytometry has been used to perform "liquid biopsy" in PCa patient plasma samples. Biggs et al. measured circulating prostate microparticles (PMPs), a type of EVs, with a size ranging from 100 to 1,000 nm, immunoreactive to antiPSMA mAb. PMP enumeration levels in plasma permit to identify patients with Gleason score ≥8 PCa independently from PSA value (35).
Regarding small EVs quantification, researchers frequently adopt the enzymelinked immunosorbent assay (ELISA). Similar to immunoaffinity isolation methods, this approach exploits small EV surface antigens that can be detected by specific antibodies. To quantify total small EVs, CD63, CD9, and CD81 vesicularspecific proteins are generally measured by ELISA (34). The evaluation of the enzymatic activity of the acetylCoA acetylcholinesterase, which is enriched in small EVs, is another method used for quantification and purity assess ments. Alternatively, to evaluate the purity degree of a small EVs preparation, Western blotting can be used to assess the presence of typical small vesicular proteins (CD63, CD9, CD81, Alix, and Tsg101) (12). However, this assay is not quantitative and gives no information on the particle size.
Upon small EVs isolation and quantification, their RNA, DNA, protein, and lipid content can be analyzed. RNA profiling can include mRNAs and miRNAs and can be achieved both by PCRbased assays for single RNA molecule profiling and by nextgeneration sequencing (RNAseq) for largescale analysis. For example, after isolation of small EVs in urine specimens, the levels of PCaassociated mRNA transcripts in these vesicles have been assessed by quantitative realtime PCR analysis (36,37). In addition, quantification of vesicular miRNAs can be assessed by RNAseq, as carried out for urinary small EVs from PCa patients (38). However, vesicular miRNA quanti fication can be affected by inadequate procedural choices, including the choice of optimal endogenous miRNA for expres sion normalization (39). Differential expression analysis of vesicular mRNAs or miRNAs between samples from primary PCa, metastatic cancers, and healthy subjects can lead to the identification of candidate biomarkers (40). In addition, other gene expression analysis methods, including coexpression network analysis, can suggest potential diagnostic and prognos tic biomarkers (41)(42)(43). Analysis of urinary vesicular RNA and DNA has also been performed in prostate tumors to identify the presence of mutations (44). It has been observed that small EVs released by different PCa cell lines harbor specific DNA mutations (45). These results may allow the development of noninvasive tests for mutational status assessment in PCa. Interestingly, the specific molecular effects due to the identified mutations on transcription, splicing, miRNA binding at 3′UTR, and nucleocytoplasmic export of mRNA can be effectively predicted (46)(47)(48)(49).
Furthermore, methods for single protein and multiple pro tein profiling exist. Western blot, cytofluorimetry, and ELISA represent popular methods for the assessment of specific small vesicular proteins and for validation of results obtained by other proteomic methods, also regarding urinary small EVs of PCa patients (50). Recently, a new aptamerbased method for the simultaneous analysis of around 1,000 proteins from small EVs of PCa patients has been developed (51). Moreover, mass spectrometry has been frequently used to identify all small EVs' proteins from many sample types and, recently, it has also been exploited for the analysis of small EVs' proteins in plasma samples of PCa patients (52). Interestingly, mass spectrometry can be also used to assess the lipid content of small EVs. In par ticular, lipid species in small EVs released by PCa cell lines and present in urine of PCa patients have been evaluated (53,54). It should be also taken into account the heterogeneous nature of cancer EVs; different types of PCa cells are more prone to shed heterogeneous populations of EVs compared with other PCa cells. Indeed, ElSayed and her group have recently showed mesenchymallike prostate carcinoma cells have a propensity to release EVs of varied sizes with diameters ranging from 100 to 300 nm approximately, while epitheliallike PCa cells principally generate small vesicles (50-150 nm). The isolation method could be a technical limitation; further studies are needed to properly characterize and classify these vesicles and to investigate the potential effects of EVs as a whole, including small EVs' fraction and others EVs' subgroups (10).

Small evs and PCa Diagnosis
In the past 10 years, small EVs have emerged as a novel effec tive and noninvasive clinical tool for the screening and early diagnosis of PCa. This evidence is supported by the specificity and sensitivity showed by small EVs analysis in this disease, together with the possibility of a noninvasive assessment of gene expression and mutations (36), thus leading to a yearby year growing number of studies in this context. Indeed, it has been shown that only PCa patients are characterized by high levels of nanovesicles (125-180 nm, i.e., small EVs) expressing both CD81 and PSA (55). Such results were obtained analyzing 1 ml of plasma samples of 15 healthy donors, 15 benign prostatic hyperplasia, and 15 PCa patients by nanoscale flow cytometry and ELISA assay. Moreover, McKiernan and his group revealed that gene expression assay in urine small EVs was able to dis criminate highgrade (Gleason score ≥7) from lowgrade (GS6) cancer and benign disease. Small EVs' RNA cycle threshold values of ERG, PCA3, and SPDEF were used to derive urine small EVs gene expression assay score that was tested in 255 men and then was prospectively validated in an independent cohort of 519 men. In this way, they identified patients with highergrade PCa among men with elevated PSA levels, thus potentially reducing the number of unnecessary biopsies (56). In the same view, high Claudin 3 levels from isolated small EVs were able to predict GS ≥8 (57), while small EVs' levels of gammaglutamyltransferase 1, a cellsurface enzyme that regulates the catabolism of extracellular glutathione, were significantly higher in PCa patients compared with benign prostatic hyperplasia patients (58).
Interestingly, several miRNAs from isolated urinary small EVs, such as miR2909 (59), miR19b (60), miRNA21, and miR375 (61), have demonstrated to be effective as diagnostic biomarkers for prostate tumors. The isolation method used in these studies included precipitation of small EVs in fluids with lowspeed centrifugation step (59) or differential centrifugation (60,61) with selection of vesicles of 30-100 nm obtained by 0.1µm filtration (60). The authenticity of these isolated small EVs was confirmed by subjecting them to morphological analysis by electronic microscopy (59,61), immunostaining with anti bodies against CD63, CD9, and CD24 (60), or Western blot analysis using CD63 antibody coupled with standard immuno detection procedure (59).
More recently, the analysis of the lipidome of urinary small EVs showed statistically significant difference between PCa patients (n = 15) and healthy volunteers (n = 13), in particular for the high presence of phosphatidylserine and lactosylceramide associated with PCa diagnosis (53). These pioneering studies, even if limited to a small number of samples, are promising and should be validated in larger independent cohorts.

Small evs and PCa Development and Progression
Prostate cancerderived small EVs are directly involved in PCa carcinogenesis and metastasis. Indeed, small EVs are able to significantly reduce apoptosis, increase cancer cell prolifera tion, and induce cell migration in LNCaP and RWPE1 cells (62). Moreover, they can modulate bone cell formation by affec ting the fusion and differentiation of osteoclasts in the meta static sites (63), thus favoring the formation of premetastatic niches.
Hypoxia plays a crucial role in regulating small EVs activity. In fact, it has been shown that small EVs secreted by PCa cells under hypoxic conditions showed higher metalloproteinases (MMPs) activity and increased levels of proteins primarily implicated in the remodeling of epithelial adherens junction pathway compared with small EVs released from normoxic cells. Therefore, this enhanced the invasiveness and stemness of naïve PCa cells (64).
Beyond their potential use for PCa diagnosis, miRNAs are directly implicated in PCa development and progression. It has been shown that normal prostate fibroblasts (WPMY1) trans fected with miR1005p, miR215p, and miR1395p augmented their migration and metastatic invasion by increasing the expres sion of MMP2, 9, and 13 and RANKL (65). Furthermore, adipocyte differentiationrelated protein can be detected in small EVs released by PCa cells and is able to induce neuroen docrine differentiation of these cells in a paracrine fashion (66). Interestingly, small EVs are also involved in modulating PCa induced immunosuppression of dendritic cell functions (67) and promote immune evasion by downregulating NKG2D expression on natural killer cells and CD8+ T cells (68).
Even if most of the studies on biological effects of small EVs are conducted in vitro, they lay the foundations for following clinical investigations.
Recently, it has been shown that the expression of CD9 is increased in PCa of patients who suffer from disease recurrence in 5 years, indicating the role of CD9 in the progression of recur rent advanced PCa (69).
Using the established immunocapture and immunodetection method, Soekmadji and colleagues reported that the level of CD9+ EVs in plasma is increased in PCa patients compared to those with benign prostate hyperplasia and, on the contrary, CD63+ EVs level does not show a significant difference between the two groups. Moreover, they showed that in plasma obtained from a metastatic PCa patient cohort the level of CD9+ EVs were higher in circulating tumor cell (CTC)positive PCa patients compared with CTCnegative patients. Instead, the CD63+ EVs level did not show significant differences between the cohorts (70,71).
Such findings underline the importance to investigate par ticularly subpopulations of circulating EVs that would be more informative as PCa biomarkers.

Therapeutic Potential of Small evs and Drug Resistance in Patients with PCa
The stability and low immunogenicity of EVs support their use as therapeutic delivery agents for cancer drugs and small molecules. The methodology for loading EVs with a thera peutic cargo consists into two different approaches: the first one is based on the indirect modification of EV membranes via the genetic engineering of their parental cell (72). The second one needs the direct encapsulation of a cargo into purified exosomes through active (i.e., sonification and electroporation) or passive (i.e., the introduction of hydrophobic drugs into EVs, the utilization of multivalent electrostatic interactions, permeabilization with saponin) loading methods (72). These technologies will represent a major step forward in the era of precision medicine for PCa patients and should be tested into future clinical trials.
Small EVs have been shown to be crucial for the develop ment of drug resistance in patients with prostate tumor. In 2017, Del Re et al. assessed ARV7 as a predictor of resist ance to hormonal therapy by highly sensitive digital droplet polymerase chain reaction in plasmaderived small EVs' RNA. They found that both median progressionfree survival (20 vs. 3 months; p < 0.001) and overall survival (8 months vs. not reached; p < 0.001) were significantly longer in ARV7negative vs. ARV7positive patients (73).
Small EVderived microRNAs contribute also to PCa chem oresistance (74) and can act as surrogate biomarker of tumor response to taxanes (75). It has been observed that the transfer of small EVs (in particular, small EVs' MDR1/Pgp) from docetaxelresistant cell lines to DU145, 22Rv1, and LNCap PCa cell lines induces acquired resistance to this drug (76). In the same view, Kawakami et al. reported that β4 (ITGB4) and VCL in small EVs could be useful markers of PCa progression correlated with taxane resistance (77).
Interestingly, serum small EVs' Pglycoprotein high levels are associated with resistance to docetaxel but not to cabazitaxel (78), thus representing a potential biomarker to guide the decisionmaking process in PCa patients.
Another important therapeutic use of EVs is vaccination to treat cancer. It has already been studied on mice injected with autologous tumorderived nanovesicles, enriched with tumor specific antigens. The highlevel immunogenicity of these nano vesicles induced an antitumor immune responses in both primary and metastatic melanoma mouse models (79).
Some questions have to be answered before these preclini cal evidences on EVbased therapy can be applied in a clinical scenario. Major problems are the EV's production, purification, and concentration, the biodistribution of EVs, the targeting of recipient cells, the uptaking by the recipient cells, and the effects on the recipient immune system (80).

DiSCUSSiOn anD COnClUSiOn
For years, Gleason score has represented the most relevant prognostic factor in patients with PCa, along with PSA levels measured at the time of diagnosis and the TNM score. Plasma quantification of PSA has markedly improved the early detec tion of PCa, but still lacks the required specificity. More recently, a new generation of potentially predictive and prognostic parameters has grown, opening the way to molecularly tailored approaches in PCa patients. Small EVs have been reported to act as mediators of celltocell communication due to the regulatory functions of their content. The high sensitivity and specificity of data obtained from small EVs analysis, their presence in almost all human fluids, and the variety of their functions exerted during tumor carcinogenesis, progression, and response to treatments, support the notion that small EVs exploration will represent a cornerstone of future approaches in cancer patients. Further investigations are needed, particularly regarding the content of the different EVs and their possible use in clinical and therapeutic settings. The current application of EVs engineering is limited by a series of key factors that should be overcrossed to introduce these methodologies into daily clinical practice. Among them, a standardization of present methods of isolation and analysis results fundamental. Interestingly, EVs is a wide term that includes different vesicle types such as exosomes, microvesi cles, and apoptotic bodies. This nomenclature, based on size, biogenesis, and cellular release mechanisms, is currently not consistent throughout the literature (17, 81,82). Although the International Society of EVs (ISEV) is making an effort to unify this controversial nomenclature, favoring the term EVs instead of exosomes and of other terms, a consensus criterion does not still exist (82). ISEV is involved also in the standar dization of the methodologies of EVs isolation and charac terization (81). Also in this case, there are not com monly accepted procedures. The lack of a consensus regarding EVs isolation and even storing conditions (83) can result in impure exosome preparations, raising a possible explanation regarding some inconsistent results in literature.
In conclusion, the search for effective predictive and prognostic factors in prostate neoplasms is still ongoing. The growing knowledge about molecular biology of this tumor is bringing us very rapidly to a new age of therapeutic possibilities. Nevertheless, the cost-benefit ratio of the massive application of these new potential prognostic factors represents a crucial point in the choice of the "best one. " Based on this scenario, small EVs may represent a cornerstone in the future diagnostic and treatmentdecision processes of PCa, leading to a more tailored and personalized approach for patients with advanced disease (Figure 1) (84).

FUnDinG
This paper was not funded. Declaration of interest: The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the production of this manuscript.