Rigorous characterization of urinary extracellular vesicles (uEVs) in the low centrifugation pellet - a neglected source for uEVs

Urinary extracellular vesicles (uEVs) provide bio-markers for kidney and urogenital diseases. Centrifugation is the most common method used to enrich uEVs. However, a majority of studies to date have focused on the ultracentrifugation pellet, potentially losing a novel source of important biomarkers that could be obtained at lower centrifugation. Thus, the aim of this study is to rigorously characterize for the first time uEVs in the low speed pellet and determine the minimal volume of urine required for proteomic analysis (≥9.0 mL urine) and gene ontology classification identified 75% of the protein as extracellular exosomes. Cryo-Transmission Electron Microscopy (≥3.0 mL urine) provided evidence of a heterogeneous population of EVs for size and morphology independent of uromodulin filaments. Western blot detected several specific uEV kidney and EV markers (≥4.5 mL urine per lane). microRNAs quantification by qPCR was possible with urine volume as low as 0.5 mL. Particle enumeration with tunable resistive pulse sensing, nano particles tracking analysis and single EV high throughput imaging flow cytometry are possible starting from 0.5 and 3.0 mL of urine respectively. This work characterizes a neglected source of uEVs and provides guidance with regard to volume of urine necessary to carry out multi-omic studies and reveals novel aspects of uEV analysis such as autofluorescence of podocyte origin.

ueV morphology, counting and sizing. Cryogenic transmission electron microscopy (cryo-TEM). Pellet P21 from 3.0 mL of urine showed a heterogeneous population of uEVs for size and morphology ( Fig. 2A). Enlarged images (Fig. 2a1,a2) showed vesicles of 60-250 nm diameter not associated with any THP filaments (arrows, Fig. 2A,B,b1,C) which can apparently entrap (*) (Fig. 2C) and/or bind vesicles (#) (Fig. 2b1). Vesicles are delimited by a clear double electrodense phospholipidic bilayer membrane mostly round in shape although some elongated/flattened vesicles (^, Fig. 2E,F) are also visible. Vesicles can be electro dense and present a more complex structure with either some granular matter or multi-layered structures with 1 or more smaller vesicles within it (Fig. 2D-F).
Cryo-TEM analysis performed on the pellet P21 TCEP THP depleted ( Fig. 2G1-G6) confirmed the presence of a population of uEVs heterogeneous both for size (40-250 nm) and morphology (round, flattened electro dense, electro negative). In addition to multi composite EV structures, rupture of the plasma membrane, release of amorphous internal content (arrow), and bulging small vesicles from bigger ones (¥, Fig. 2G2,G3) can be seen, extending the repertoire of EV morphology.
Particle size distribution and number in tunable resistive pulse sensing (TRSP). TRPS was employed to estimate both particle size distribution (PSD) and number. A nanopore membrane NP300 (analysis range 150-900 nm)  Gallery of Cryo-TEM images of urinary EVs recovered in the relative low centrifugation pellet P21 before and after TCEP reduction (P21 TCEP ). A heterogeneous population of uEVs was observed including single layered vesicles and multiyered structure with two or more inner small vesicles encapsulated inside bigger vesicles before (A-F) and after TCEP treatment (G1-G6). Tamm-Horsfall protein (THP) long polymeric filaments (indicated by arrows) (A) either engulfing (*) (C) or adsorbing (#) vesicles (B and b1) were also visible. No filaments of THP were visible (G1-G6) after TCEP reduction. experimentally established in order to have a linear particle rate ( Supplementary Fig. S4), minimizing nanopore clogging whilst having a satisfactory blockade height of >0.05 nA ( Supplementary Fig. S5). PSD was similarly independent of the volume of urine used to enrich uEVs ( Supplementary Fig. S6). A moderate shift of 14 nm (mean) and 10 nm (mode) was seen when samples were calibrated with 200 nm (SPK200B) and 330 nm (CPC400E) standard particles respectively. As expected, the particle number increased with the volume of urine processed (Fig. 3A) for an estimated average urine concentration of 3.73 × 10 8 ± 6.40 × 10 7 (SPK200B calibration standard) and 2.04 × 10 8 ± 3.55 × 10 7 (CPC400B calibration standard) particles per mL of urine with a coefficient of variation of 17.1% and 17.4% respectively. Enumeration was possible with urine volume of 0.5 mL.
Nano tracking analysis (NTA) of particle size distribution before and after TCEP treatment. Size and particle concentration of P21, P21 TCEP and SN21 TCEP were measured by nanoparticle tracking analysis (NTA). As previously seen for TRPS, the PSD and particle concentration results were similar, independent of the volume of urine for P21, P21 TCEP and SN21 TCEP (Supplementary Fig. S7A-C). Not surprisingly, the particle number increased with the volume of urine processed (Fig. 3B, Supplementary Table S2) for an estimated average urinary concentration of 1.58 × 10 10 ± 3.97 × 10 9 particles per mL of urine with a coefficient of variation of 25.1%. After P21 TCEP reduction the particle number of P21 TCEP was 5.57 × 10 9 ± 1.18 × 10 9 particles per mL of urine with a coefficient of variation of 21.1% ( Fig. 3C; Supplementary Table S3) and SN21 TCEP was 1.08 × 10 9 ± 2.21 × 10 8 particles per mL of urine with a coefficient of variation of 20.4% ( Fig. 3D; Supplementary Table S4). PSD was the same for P21, P21 TCEP and SN21 TCEP (Supplementary Fig. S7D). Taking into account that there was a minimal signal detection for uEV markers (Fig. 1), the relatively high SN21 TCEP particle concentration might come from THP. Enumeration was possible with urine volume of 0.5 mL. ueV cargo analysis. EV Proteomic analysis by mass spectrometry. A bottom-up proteomic approach was adopted to determine the protein composition of uEV pellet P21 TCEP without THP ( Supplementary Fig. S8A). Overall, we found 1254 non-redundant gene name proteins with 2 or more unique peptides and 99% protein confidence (Supplementary Table S5). We compared our data set with the list of the proteins deposited in the most recently updated (Version 4.1 8 15 2018) vesiclepedia repository and the subset of protein identifications specific for uEVs (Supplementary Table S5).
Analysis of GO terms with DAVID recapitulate the results obtained with PANTHER. DAVID algorithm showed that the cellular component annotation with the highest percentage of proteins and low P value was extracellular exosome (GO:0070062; Supplementary Fig. S11A). Molecular function and KEGG pathway analysis ( Supplementary Fig. S11B,C) highlighted the key role of these proteins in metabolic pathways, endocytosis and actin cytoskeleton regulation. Finally, in spite of the TCEP denaturation step to eliminate uromodulin, THP was still present in the MS list of the proteins identified. Interestingly, among the 15 unique tryptic peptides we found 2 peptides which matched the amino acid sequence of the domain between the catalytic cleavage -serine S in position 589 and the serine in position 614 bound to the glycophosphatidylinositol (GPI) anchor ( Supplementary  Fig. S12) [43].
Protein pattern and EV protein analysis by WB. Electrophoresis separation of P21 pellets obtained from 3 independent urine collections showed a very similar protein pattern for both silver staining (Fig. 4A) and Coomassie staining (Fig. 4B,C). THP was found to be the most abundant protein, the amount of which increased proportionally with increasing volume of urine.
Western blot analysis was performed to evaluate both EV markers according to the MISEV guidelines 18 and markers of interest for downstream analysis. Tumor susceptibility gene 101 (TSG101, Fig. 4J), as part of the Endosomal Sorting Complex Required for Transport (ESCRT) machinery and CD9 (Fig. 4F), as it is one of the three tetraspanins, were selected as positive controls. Calreticulin (CALR, Fig. 4I) and calnexin (CALX, Fig. 4L) were targeted to exclude potential contaminants mimicking EVs from the intra cellular membrane compartments (endoplasmic reticulum). Two positive samples -rat kidney and saliva epithelial cells, were added to check the antibodies reactivity and cross-species specificity respectively. We did not detect either CALX or CALR at the level of sensitivity set for the acquisition; this suggests no major contaminations from cellular debris and intracellular membranes. Podocalyxin (PODXL, Fig. 4K), podocin (NHPS2, Fig. 4H) and collectrin (TMEM27, Fig. 4D) were selected as nephron-specific markers originating from podocyte, proximal and distal tubule cells respectively. Insulin-like growth factor binding protein 7 (IGFBP-7, Fig. 4G) and tissue inhibitor of metalloproteinases 2 (TIMP-2, Fig. 4E), which are secreted soluble proteins as part of the uEV proteome, are of great interest as biomarkers to predict the risk of developing acute kidney injury (AKI). All the antibodies have good specificity to recognize their own antigen at the right molecular weight with the exception of podocin, which shows a higher molecular weight band consistent with the ubiquitinated isoform 20 . The lower molecular weight fragment could be a splicing isoform lacking the PHB domain 21 . Most antibodies (TMEM-27, IGFB-7, CD9 and NPHS2) detected the respective antigen in P21 starting from 4.5 mL of urine, while PODXL, TSG101 and TIMP-2 could be found in as little as 1.0 and 0.5 mL of urine respectively. EV Surface protein analysis by imaging flow cytometry (iFC). iFC was employed as a tool for high-throughput single EV targeted protein analysis to analyse the surface distribution of uEV markers, namely, podocalyxin (PODXL), collectrin (TMEM27), insulin-like growth factor binding protein 7 (IGFBP-7), tissue inhibitor of metalloproteinases 2 (TIMP-2) and annexin V (AV). The antibody clones used in this analysis were also used to detect the same antigens in western blot. Analysis and gating strategy were established utilizing: buffer only, buffer plus reagents, buffer plus uEVs only and detergent lysis (Supplementary Figs. S13 and S14). Molecules of equivalent soluble fluorochrome (MESF) beads were used when available as a tool to provide standardized comparable results with different flow cytometry platforms ( Supplementary Fig. S15). The analysis of the uEVs showed a unique natural auto-fluorescence proportional to the amount of uEVs ( Supplementary Fig. S16) with a peak of emission in the red (CH 11) for camera 2, which captures the emission from both 405 nm and 640 nm excitation lasers, and for for camera 1(CH 5) which captures the emission for both the 488 nm and 561 nm excitation lasers. Since AV-APC emission is in CH 11, we created an additional gate to delimit this autofluorescence (AF). We also used the whole set of uEVs (0.5-13.5 mL) to evaluate if the median fluorescence intensity (MFI) was stable with the decrease of the particle counts ( Supplementary Fig. S17), demonstrating that increased particle concentration did generate coincident (or aggregate) events. Application of morphology and intensity masks for the highest volume (13.5 mL) combined with the spot count feature on the positive gate for each antigen confirmed that the majority of the events are single events ( Supplementary Fig. S17). Detergent lysis by Triton X-100 at concentration of 0.8% for 30 minutes at room temperature reduced (68.2%) the particle concentration in both TRPS ( Supplementary Fig. S14A,B,C,E; Supplementary Table S7) and iFC. Concentration of PODXL, AV and AF (CH 11) decreased by 64.4%, 72.0% and 96.9% respectively ( Supplementary Fig. S14F-H).
Concentration of IGFBP7, TIMP2, TMEM27, PODXL, AV and AF (CH 5 and 11) increased proportionally to the volume of processed urine (Fig. 5, Supplementary Table S8). When results were reported as object per mL of urine we noticed that for IGFBP7, TIMP2, TMEM27 and AV positivity the coefficient of variation (CV) was 36.9%, 27.7%, 16.0% and 27.8% respectively (Fig. 5A-D). Conversely, for PODXL (Fig. 5G) the CV was 98.5%. This trend seems to follow the amount of AF particles detected in the uEVs only sample in channels 5 and 11 (Fig. 5E,F). When we applied the Boolean algorithm to exclude AF, the coefficient of variation dropped to 53.4% (Fig. 5H), thus suggesting a co-localization with PODXL. In fact among all possible combinations of antigens ( Supplementary Fig. S18), the most prominent double staining occurred between PODXL and AF. Single staining for PODXL confirmed the co-localization of PODXL and AF at low (Fig. 6A), medium (Fig. 6B) and high (Fig. 6C) scatter intensities respectively. When we applied the Boolean mask for both for morphology and intensity, we found that more than 55% of the counts were double staining single events (Fig. 6D,E). Overall, as with the western blot, ideal urinary volume to process in order to enrich uEVs reaching a sufficient concentration to carry out a multiparametric characterization depends on the abundance or level of expression of the target marker. For all tested volumes we did not experience any swarming or coincidence effect. However, the presence of AF can be problematic when uEVs are enriched from large volume of urine (> 9.0 mL). Taking into account all these factors we conclude that for imaging flow cytometry, one of the more sensitive Flow Cytometric methods for EV detection, the best volume of urine for detection of uEV surface proteins is between 1.5 and 4.5 mL.
EV mi-RNA cargo analysis. miRNA miR-16, miR-155, miR-200b, miR-203, have been previously reported to be found abundantly in cell-free fraction of urine from healthy volunteers by deep sequencing techniques [47][48][49]. These miRNAs were isolated from P21 pellet from urine collected on 3 different days from the same subject. Spike-in controls cel-miR-39 was added before the RNA extraction to normalize. The expression of miRNAs (dCT) was confirmed and it was observed to increase proportionally with increasing volume of urine that was . Nitrocellulose membranes were hybridized respectively with: collectrin (TMEM27) (D) and tissue inhibitor of metalloproteinases 2 (TIMP-2) and CD9. After the first image acquisition the same membranes were incubated again with insulin-like growth factor binding protein 7 (IGFBP-7) (G), podocin (NHPS2) (H) and calreticulin (CALR) (I) respectively. Finally membranes were incubated a 3 rd time with tumor susceptibility gene 101 (TSG101) (J), podocalyxin (PODXL) (K) and calnexin (L) respectively. Samples for gel in C and western blots in (F,I,L) were run without DTT. K rat kidney, S Saliva pellet 4.600 g. (2020) 10:3701 | https://doi.org/10.1038/s41598-020-60619-w www.nature.com/scientificreports www.nature.com/scientificreports/ processed to obtain the P21 urine supernatant pellet (Fig. 7). Even though the relative amount of miRNAs was lower in the 0.5 mL urine fraction, all the miRNAs tested were detectable in this lowest tested volume of urine fraction. It was also confirmed that these 4 miRNA that we tested were expressed in both P21 and the P21 TCEP and their expression level was higher than SN21 TCEP (Supplementary Fig. S20). These results are in line with WB and NTA analysis.
Characterization and recovery of uEV proteins from 8 healthy donors. In order to validate the minimal volume needed to characterize EVs, we studied 8 different healthy donors. We performed uEV analysis from 4 female and 4 male healthy donors and compared protein pattern and expression levels of TSG101. Based on the previous analysis for the proteomic validation we used a pellet (P21) from 9.0 mL of urine (Fig. 8A) treated with TCEP to eliminate THP interference (Fig. 8C,E). Western blot analysis of TSG101 confirmed the presence of markers in   (Fig. 8B) and in P21 TCEP (Fig. 8D) with minimal loss in SN21 TCEP (Fig. 8F). Particle concentration measured in NTA using a P21 originating from 3.0 mL of urine showed a moderate inter-individual variability in particle concentration and PDS. After TCEP reduction a higher variability for P21 TCEP and SN21 TCEP than P21 was noted (Fig. 8G, Supplementary Tables S9-S11).

Discussion
Urinary extracellular vesicles have been extensively investigated for their novel role in cell-to cell communication, shuttling informative molecular cargo along the nephron and being a novel source of biomarkers [22][23][24] . We opted to analyse the first pellet of the differential centrifugation protocol as it has not been characterized rigorously and is mostly neglected by researchers. Our analysis also addresses the volume of urine necessary to provide EVs for multi-omic analysis on the same specimen, testing the limit of detection of instrument as well as the technical variability. In fact the 7 uEVs P21 pellets have to be considered as a septuple (7 replicas) of the same specimen. Our objective was to perform this rigorous characterization in keeping with "Minimal information for the study of EVs (MISEV) guidelines by the International Society for Extracellular Vesicles 17 . In particular, we applied several controls for each downstream analysis; this is especially important for flow cytometry analysis as many particles can mimic EVs. One of the major drawbacks of enriching uEVs is the co-sedimentation of THP, which can sediment readily at very low speed 25 and entrap uEVs in its filaments 18 or bind EVs 26 . Thus, one of our main goals was to reduce THP interference for MS analysis, RNA extraction and miRNA quantification, and NTA enumeration. We used Tris (2-carboxyethyl) phosphine hydrochloride (TCEP-HCl) as an alternative reducing agent to dithiothreitol (DTT) for its high reducing energy 27,28 . In fact, TCEP was able to quickly reduce the 24 disulfide bonds of THP at a 10 mM concentration. A key step for a successful release of THP in the supernatant (SN21 TCEP ) and recovery of uEVs in the centrifugation pellet (P21 TCEP ) is dilution prior to centrifugation. This step decreases the probability of unfolded THP to aggregate and precipitate following centrifugation 29 . Western blot analysis of EVs positive markers (TSG101, CD9) confirmed the expected recovery in P21 TCEP rather than the SN21 TCEP whilst two soluble protein like THP and ALB were released mainly in SN21 TCEP . It is worth noting that two markers which have attracted a lot of interest in the prediction of acute kidney injury, IGFBP-7 and TIMP-2, seem to be differently affected by TCEP reduction. While IGFBP7 was completely retained in the pellet, TIMP2 was partially released in the supernatant after reduction. This suggests either a protective role of the EVs for some protein biomarkers (being packaged in EVs) or that the disulfide bonds do not play a key role in the tridimensional structure of the protein which is preserved. Unfortunately, no reports are available on the precise mechanism of their secretion as either soluble proteins (EV free) or adsorbed on the surface of the EV since both antigens were detected in flow cytometry. Further studies are required to establish the exact secretory pathway, the percentage of distribution between uEVs and the soluble urinary protein fraction. Of note, for many disease processes and biomarker discovery studies it is not known whether proteins like IGFBP-7 and TIMP-2 are superior biomarkers in soluble form or as packaged in EVs.
Our Cryo-TEM analysis is the first to investigate uEVs in cryo-TEM in the P21 pellet before and after denaturation of THP. We confirmed a wide variety (or heterogeneity) of different types of vesicles of different sizes (40-500 nm) and morphology (round, oval/flattened) with complex architecture made of a set of vesicles enclosed inside some larger ones independent of the presence of filament of uromodulin. This is consistent with Cryo-TEM analysis of EVs enriched from other biofluids, such as plasma 30,31 , synovial fluid 32 , ejaculate 33 and urine 34 .
Particle size distribution and particle concentration determination was done with TRPS and NTA. Particles were detectable with as little as 0.5 mL urine for both TRPS and NTA with an overall coefficients of variation of 16% and 23% respectively. PSD distribution between TRPS and NTA was very close with an average diameter of 227.8 nm (TRPS) and 229.3 nm (NTA) and a mode diameter of 174.2 nm (TRPS) and 172.8 nm (NTA) respectively. However, NTA P21 detected a particle concentration two order of magnitude higher than TRPS. We note that it was beyond the aim of this study to compare NTA and TRPS. These two techniques are based on completely different principles and instrument settings. However, similar discrepancies were observed in other studies on different biofluids 32,33 . www.nature.com/scientificreports www.nature.com/scientificreports/ We can explain this discrepancy in count numbers between TRPS and NTA on the basis that NTA is more sensitive to detect small particles made also of soluble proteins such as albumin, which can scatter light 34 . For example, the particle counts in the SN21 TCPE , P21 TCEP and SN21 TCEP is still significant which might be due to the fact that THP can scatter and be detected as a particle like albumin. www.nature.com/scientificreports www.nature.com/scientificreports/ Mass spectrometry protein analysis of P21 TCEP after THP elimination revealed a relatively complex protein composition with 1251 identified proteins. Gene ontology protein classification with both Panthers and DAVID algorithms [35][36][37] showed that class distribution of proteins was not dissimilar from the vesiclepedia data sets. Some differences compared to the vesiclepedia data sets were noted when subcategorizing protein class with less nucleolar protein and more protein binding actin filaments. GO classification confirmed the presence of exosomes and apical plasma membrane vesicles carrying a variety of plasma membrane proteins specific to every type of epithelial cell forming the nephron. Podocalyxin (PODXL), nephrin (NPHS1), podocin (NPHS2) originating from podocytes and acquaporin-2 (AQP2) originating from collecting cells are just a few examples. Overall, independently from THP we confirmed that a low RCF can pellet different type of vesicles including exosomes (as defined by exosomes markers like TSG101 and CD9 for example) or small EVs [38][39][40] . Interestingly, our proteomic analysis revealed the presence of THP with a characteristic peptide pattern which include two peptides originating from the domain of the protein between the serine (S 614 ) glycosylphosphatidylinositol (GPI) lipidation anchor site and hepsin cleavage site ( 586 RFRS 589 ) 41 . Hence, it is plausible that THP could be anchored to the membrane of uEVs secreted by the cells ascending Henle's loop limb. Therefore, THP traces in uEV preparation might not be simply a mere contamination of the predominantly cleaved THP secreted form, but rather as part of the EV protein cargo.
Western blot analysis of P21 and P21 TCEP EV positive markers confirmed the presence of several EV markers such as TSG101 and tetraspanin CD9. The volume of urine to enrich uEVs and detect a marker of interest depends both on the abundance and the affinity of the antibody for the antigen. Ideally this should be shown by each study, however in many studies it has not been performed nor transparently reported. Overall, we estimate that 4.5 mL of urine is the minimum (or minimal) amount required to provide enough material to detect an antigen in western blot analysis but higher amount may be necessary for the detection of nephrin, where uEVs P21 were enriched from 20 mL of urine.
Imaging flow cytometry (iFC) was used as a high-throughput single EV analysis to detect uEV surface markers using multiple antigens. The advantages of iFC with respect to conventional flow cytometry have been already described 42,43 . iFC is one of the few highly sensitive flow-cytometers currently available for EV research. In particular iFC can provide increased sensitivity for the detection of smaller (including <100 nm) EVs which almost all conventional flow cytometers are incapable of detecting 44 . In addition, iFC provides robust population statistics and imaging confirmation of EVs utilizing a single technology 42,43 . We found an unexpected and never reported complication of the natural auto-fluorescence (AF) in the uEVs pellet without any reagents. Interestingly, AF was associated with EVs in general, but in particular with PODXL positive EVs. AF interferes with the counts of PODXL positive particles which resulted in a lack of proportionality to the amount of uEVs (Fig. 5). When the count was normalized by the volume of urine we found the coefficient of variation (CV 95%) was partially reduced when AF was excluded applying Boolean logic function (CV 54.3%) which offers a solution to correct for AF. However, AF was overall a rather complex factor which complicates the analysis and is amplified with the amount of volume of urine used to enrich EVs (Fig. 5E,F). For the aforementioned reason we think that a volume between 3 and 4.5 mL is optimal to perform a multiparametric analysis which includes 2 washing steps as per this study. This is the first report which highlights the difficulties associated with evaluations in the presence of AF in urinary EVs. This phenomenon is documented by some researchers 14 , but mostly ignored or not detected as it is below the instrument's detection limit. However, the phenomenon of AF needs to be addressed as it can mimic artefactual EV counts by FC or lead to quenching of other fluorescent antibodies (or affect antibody performance). The biological relevance of the source of this AF is of interest and requires further studies.
EVs have been known as shuttles that also carry microRNAs that are crucial upstream regulators of gene expression. The role of urinary cell free miRNA in association with kidney disease and function has been reported [45][46][47] . It was observed that the detection was achieved with as little as 0.5 mL volume of urine for the lower abundant miRNA of the four miRNAs that were tested. This was indicated as an increased dCt in the reaction of RNA isolated from 0.5 mL of urine as compared to the other volumes. Furthermore, the same set of miRNAs was tested in the pellet P21 and P21 TCEP . Again, the majority of the miRNA carried by uEVs was collected in the pellet after TCEP treatment with minimal release in the SN21 TCEP , THP did not interfere with the RNA extraction and therefore TCEP treatment is not really necessary, as previously reported 48 . From our own experience it is likely that lower volumes could be sufficient for miRNA detection (data not published). However this particular study presented here did not assess volumes lower than 0.5 mL.
Although this study does not include any functional assay we believe that the utility of using a reducing agent to eliminate the interference of soluble proteins is useful for mass spectrometry analysis. More in general a reducing agent can have a detrimental impact on enzymatic activity 49 particularly for those proteins which have key disulfide bonds in maintaining the tertiary and quaternary structure.
Finally, after uEV P21 characterization, we extended the analysis from uEVs enriched from different healthy donors. Sample protein patterns were very similar, even more so after THP removal. We investigated TSG101 as an EV marker because it was the most sensitive marker to detect the antigen in SN21 TCEP , where only traces were detected. Particle concentration measured with NTA was consistent with a coefficient of variation (P21 CV 14.4%) in the same order of the technical variation.
In conclusion this study provides a detailed characterization of uEVs recovered at a centrifugation speed of 21,130 g with different urine volumes. We set up a new protocol to eliminate THP reducing the disulphide bonds with TCEP, which allowed recovery of the majority of uEVs in pellet P21 TCEP . For each downstream analysis tool used, we had 7 replicas demonstrating the technical variability and repeatability of the enriched uEV samples from the different urine volumes studied. Proteomic analysis of P21 TCEP free of uromodulin confirmed the presence of a heterogeneous population of uEVs including smaller EVs such as exosomes (TSG101 and CD9 positive markers), supporting that the low centrifugation pellet is a rich source of EV biomarkers deriving from a heterogeneous group of EVs, not just larger EVs. We opted to analyse the first pellet of the differential centrifugation Scientific RepoRtS | (2020) 10:3701 | https://doi.org/10.1038/s41598-020-60619-w www.nature.com/scientificreports www.nature.com/scientificreports/ protocol as this approach has been ignored but can be equally informative without requiring the use of expensive instruments such as ultracentrifuges and associated rotors. We showed for the first time electron microscopy pictures of uEV P21 with and without THP filaments. We carried out multiparametric analysis with imaging flow cytometry, a very sensitive and high-throughput single EV analysis tool, describing the natural auto fluorescence of uEVs. This AF (not previously reported or overlooked) of the uEV prep influences the analysis of uEVs of podocyte origin, however we can offer a solution for an analysis algorithm to overcome this phenomenon of AF. We therefore conclude and summarize that the minimal volume of urine necessary to perform a multi -OMICs study and rigorous EVs characterization is: 9 mL for mass spectrometry; 1.0 mL for NTA TRPS; and 4.5 mL per lane Western blot but more urine volume could be necessary depending on antibody affinity and antigen abundancy on the uEV; 0.5 mL (and possibly lower) for qPCR, 3.0 mL for Cryo-TEM for P21and and 9.0 mL for P21 TCEP , and between 1.5 and 4.5 mL for imaging flow cytometry.
This work makes significant contributions to the study of urine as a biofluid for EV research and provides uEV researchers guidance with regard to volume of urine necessary to carry out studies in keeping with MISEV guidelines and downstream analysis of interest. In addition, this work reveals novel aspects of uEV analysis such as AF in urine and interaction of uEV proteins with soluble proteins such as THP/Albumin, which need to be further studied and considered when doing uEV research.

Materials and Methods
Additional detailed material and methods are provided in supplementary information. Chemical reagents were purchased from Sigma-Aldrich (Saint Louis, MO) unless otherwise specified.
Urine samples. Urine samples were collected from a healthy volunteer aged 20-51 with no history of renal diseases, diabetes and hypertension. First morning void urine was processed within 3 h without adding any protease inhibitors. Written informed consent was obtained from the participant. This study was approved by The Research Ethics Committee of the University of Virginia (IRB HSR # 17192). All the experiments were performed in accordance with the declaration of Helsinki.