Blood was collected from participants after providing written informed consent. We included a cohort comprising 100 plasma samples from patients with ESLD and 50 plasma samples from healthy kidney donors serving as controls, procured from individuals enlisted in the TransplantLines Cohort and Biobank Study (ClinicalTrials.gov identifier: NCT02811835) at the University Medical Center Groningen, which was approved by the Medical Ethics Committee (METc 2014/077) and in accordance with the guidelines of the Declaration of Helsinki. This is an ongoing, prospective study that aims to provide a better understanding of the causes of disease-related and aging-related outcomes and health problems, both physical and psychological, in solid organ transplant recipients and donors. From June 2015, all solid organ transplantation patients, and donors (aged ≥ 18 years) of the University Medical Center Groningen (UMCG, The Netherlands) were invited to participate.

Additionally, serum samples were obtained from 10 Hispanic and 10 Caucasian healthy volunteers to evaluate the applicability and compare between samples.

Blood was drawn after an overnight fasting period of 8–12 h in the morning using a 0.75-inch (19 mm) butterfly collector (#367284, BD Vacutainer, Becton, Dickinson, NJ, US) in 3 mL sodium citrate tubes (#369714, BD Vacutainer). Immediately after being collected, samples were centrifuged at 2,500 g for 10 min and the liquid was stored in 0.5 mL aliquots at –80°C (−112°F) until use, avoiding freeze-thaw cycles.

Plasma citrate samples (1 mL) were thawed at room temperature (24°C) and transferred into 5 mL (13 × 51 mm) tubes (#344057, Beckman Coulter Inc., Brea, CA, US). Samples were diluted using cold phosphate-buffered saline [phosphate buffered saline (PBS), 1×] (#10010023, Gibco, Thermo Scientific, Carlsbad, CA, US) and tubes were placed into a SW 55 Ti swinging-bucket rotor, k factor 48 (#342196, Beckman Coulter Inc).

All centrifugation steps were performed at 4°C. The UC consisted of a three-step protocol adapted from Helwa et al. [20] (Figure 1). The first step was a 30-minute cycle at 20,000 g. The supernatant was collected, and the pellet was discarded. The supernatant was placed in 5 mL ultra-clear tubes and centrifuged at 110,000 g for 2 h. After centrifugation, the supernatant was discarded, and the pellet was resuspended in cold PBS. Finally, samples were submitted to a final “washing” step at 110,000 g for 1 h. The pellet was resuspended in 100 μL of PBS (PBS-EVs) or directly into lysis buffer or RNA extraction solution for further analysis.

For samples without washing (WW), the last centrifugation step (110,000 g for 1 h) was omitted. For contaminant-depleted plasma (CDP) samples, separation of plasma from whole blood was achieved by centrifugation at 2,500 g for 15 min. The resulting plasma samples were subjected to a second centrifugation step at 2,500 g for 15 min and then stored under identical conditions [21]. In addition, samples from the first centrifugation step (30 min, 20,000 g) were collected and used as a negative control, since they contained bigger vesicles as well as cell debris.

Fifty microliters of PBS-EVs was mixed with 50 µL of lysis buffer (HEPES 25 mmol/L, Kac 150 mol/L, EDTA 2 mmol/L, NP-40 0.1%, NaF 10 mmol/L, PMSF 1 mmol/L, aprotinin 1 µg/µL, pepstatin 1 µg/µL, leupeptin 1 µg/µL, and DTT 1 mmol/L). Samples were subjected to a series of 4 frost-defrost cycles, followed by centrifugation at 12,000 g for 10 min. The supernatant was collected and processed to be quantified.

Protein concentration was quantified using a colorimetric serial dilution of bovine serum albumin (BSA) assay kit following manufacturer’s instructions (#5000116, Bio-Rad, Hercules, CA, US). Absorbance was determined at 750 nm. The concentration of each sample was determined relative to the BSA curve.

The presence of specific protein markers for EVs [CD9 (#13174S, Cell Signaling, Danvers, MA, US), CD63 (#15363, Santa Cruz Biotechnology, Dallas, TX, US), CD81 (#10630D, Invitrogen, Waltham, MA, US), TSG-101 (#7964, Santa Cruz Biotechnology)] and control proteins [albumin (#A0001, Agilent Technologies, Santa Clara, CA, US), CD41 (#13807, Cell Signaling), and anti-acetyl-histone H4 (HDAC4, #06-866, Sigma-Aldrich, San Luis, MO, US)] were analyzed. Forty micrograms of protein was loaded on 10% SDS-PAGE gels. After electrophoresis at 100 V for 90 min, proteins were transferred onto a nitrocellulose membrane by semi-dry blotting for 30 min and blocked with 5% BSA. Membranes were incubated overnight with the antibodies and after incubation secondary antibodies were added [goat anti-rabbit horseradish peroxidase-labeled secondary antibody (#P0448, Agilent Technologies)] or polyclonal rabbit anti-mouse immunoglobulins/horseradish peroxidase (#Cat P0260, Agilent Technologies). Finally, the blots were analyzed using a ChemiDoc XRS system (Bio-Rad).

Total RNA was isolated from 16 samples using the AllPrep DNA/RNA kit (#80004, Qiagen, Hilden, Germany) or TRI-reagent (#T9424, Sigma-Aldrich) following the manufacturer’s instructions. Samples WW, CDP, or those subjected to the 3-step UC procedure were either directly resuspended in TRI-reagent or RLT buffer (lysis buffer) or resuspended in 50 μL PBS mixed with 50 μL of RLT or TRI-reagent. For the AllPrep DNA/RNA kit protocol, 600 µL of RLT buffer with β-mercaptoethanol was immediately added to the samples, followed by vigorous vortexing for at least 1 min. The lysate was then transferred to a gDNA eliminator spin column and after centrifugation, the flow-through was saved. Subsequently, 600 µL of 70% ethanol was added to the flow-through and mixed by pipetting. The mixture was then transferred to an RNeasy spin column, centrifuged and the flow-through was discarded. This step was repeated, and additional washes were performed using RW1 buffer and RPE buffer. Finally, the column was subjected to high-speed centrifugation and the RNA was eluted with 30 µL of RNAse-free water.

For TRI-reagent RNA isolation, 0.2 mL of chloroform per mL of TRI-reagent was added, shaken vigorously for 15 s and allowed to stand for 2–15 min at room temperature. The mixture was centrifuged at 12,000 g for 15 min at 2–8°C. The aqueous phase was then transferred to a fresh tube and 0.5 mL of isopropanol was added per 0.75 mL of TRI reagent used for the initial homogenization. The sample was allowed to stand at room temperature for 5–10 min to precipitate the RNA. After centrifugation, the supernatant was carefully removed and the RNA pellet was washed with 75% cold ethanol. Following another centrifugation step, the RNA pellet was briefly dried and then solubilized using an appropriate volume RNase free water (20 μL). The solubilization process involved pipetting and incubation at 55–60°C for 10–15 min. The RNA concentration and purity were measured using a NanoDrop 2000 Spectrophotometer (ND-2000, Thermo Scientific, Wilmington, DE, US), and the samples were stored at –80°C for further analysis.

Size, size distribution, and quantity of EVs were analyzed by nanotracking analysis (NTA) using a Nanosight LM14 (Malvern Panalytical, Malvern, UK) with blue laser (404 nm, 70 mW) and a sCMOS camera (Hamamatsu Photonics, Hamamatsu, Japan). For NTA, samples were resuspended in 1 mL of PBS. Before each measurement, the equipment was calibrated with 100 nm polystyrene particles and a blank measurement was taken. Samples were injected using a 1 mL syringe and the focus was adjusted to avoid blur particles. Five captures of 60 s each were taken. Once completed, the analysis was done using the NTA software version 3.0. Measurements were performed using a threshold setting of 5 and blue cross count was limited to 5.

Freshly isolated samples were fixed with 2% paraformaldehyde in 0.1 mol/L PBS in a ratio of 1:10 and stored at 20°C until viewing. For transmission electron microscopy (TEM), samples were placed onto a Formvar/Carbon 300 Mesh (#FCF300CU50, Sigma-Aldrich) grid and allowed to air dry for 10 min. Next, the grids underwent initial contrast with a uranyl-oxalate solution, followed by further contrast and embedding using a 4% uranyl acetate mixture. After air drying, visualization of the grids was carried out using a Philips Tecnai 12 (BioTwin, Eindhoven, The Netherlands) electron microscope at 80 kV. The images were then captured utilizing the Item Olympus Soft Imaging Solutions software (Windows NT 6.1).

A total of 20 samples were isolated using the 3-step UC protocol. Half of the samples underwent protein lysis (10 samples), following the previously described protein isolation method and the other half did not (10 samples).

For proteomic analysis, samples with a fixed total protein amount (25 µg) were loaded onto a precast 4–12% Bis-Tris Protein Gels (#NP0335BOX, Invitrogen) and electrophoresis was performed for a maximum of 5 min at 100 V. The gel was stained with Biosafe Coomassie G-250 Stain (#610786, Bio-Rad) and one band containing all proteins was excised from the gel. Gel pieces were subjected to multiple washes. First, they were washed with a solution containing 70% 100 mmol/L NH₄HCO₃ in water and 30% acetonitrile, followed by a wash with 50% 100 mmol/L NH₄HCO₃ in water and 50% acetonitrile. The final wash involved 100% acetonitrile for 5 min, and the gel pieces were then dried at 37°C. For reduction and alkylation, gel pieces were covered with 10 mmol/L DTT in 100 mmol/L NH₄HCO₃ and incubated at 55–60°C for 30 min. Next, 55 mmol/L iodoacetamide in 100 mmol/L NH₄HCO₃ was added and the mixture was shaken in the dark for 30 min. Subsequently, the gel pieces were washed again with 100% acetonitrile, followed by another drying step at 37°C. For trypsin digestion, a 1:100 ratio of trypsin to gel pieces was prepared in 100 mmol/L NH₄HCO₃, ensuring the gel plugs were fully submerged. The mixture was centrifuged briefly and then incubated overnight at 37°C. The next day, peptides were eluted from the gel pieces with 50 µL 75% v/v acetonitrile plus 5% v/v formic acid (incubation 20 min at room temperature, mixing at 500 rpm). The organic solvents were removed using a SpeedVac Vacuum (Thermo Scientific) and resuspended in 40 µL 0.1% v/v formic acid for further analysis.

Discovery mass spectrometric analyses were performed on a quadrupole orbitrap mass spectrometer (MS) equipped with a nano-electrospray ion source (Orbitrap Exploris 480, Thermo Scientific). Chromatographic separation of the peptides was performed by liquid chromatography (LC) on an Evosep system (Evosep One, Evosep, Odense, Denmark) using a nano-LC column (EV1137 Performance column 15 cm × 150 µm, 1.5 µm, Evosep; buffer A: 0.1% v/v formic acid, dissolved in Milli-Q H₂O, buffer B: 0.1% v/v formic acid, dissolved in acetonitrile). The digests were injected in the LC-MS with 20 µL of the peptide digests and the peptides were separated using the 30SPD workflow (Evosep). The MS was operated in positive ion mode and data-independent acquisition (DIA) mode using isolation windows of 16 m/z with a precursor mass range of 400–1,000, switching the field asymmetric waveform ion mobility spectrometry (FAIMS) between compensation voltage (CV) 45 V and 60 V with three scheduled MS1 scans during each screening of the precursor mass range. LC-MS raw data were processed with Spectronaut (version 17.5.230413, Biognosys) using the standard settings of the direct DIA workflow except that quantification was performed on MS1, with a human Swiss-Prot database (www.uniprot.org, 20,350 entries). For the quantification, local normalization was applied, and the Q-value filtering was set to the classic setting.

Data with ≥ 75% of missing values (0) was filtered, and missing value was imputed using the missForest package (version 1.5). Differential expression analysis was conducted using the DESeq2 package (version 1.40.2) in R (version 3.6.0) using a threshold of false discovery rate (FDR) P < 0.1 and fold-change >  2. The “plotPCA” function from the “DESeq2” package in R was utilized to perform a principal component analysis (PCA). EVs-related markers and contaminants were identified and plotted using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, US). To visualize the patterns of protein expression and identify significant changes between conditions, a heatmap analysis was employed. The “heatmap.2” function was used based on samples using a threshold of log2-fold change (LFC) > 1.5 and a baseMean > 100. Row scaling was applied to emphasize relative differences in expression patterns.

Platelet-poor plasma and serum samples were isolated following the 3-step UC protocol (20,000 g × 30 min, 110,000 g × 2 h, and 110,000 g × 1 h). Randomized samples were used for EVs characterization and representative experiments from plasma or serum are shown in Figure 2. Small EVs were isolated in both serum and plasma samples from patients and controls as demonstrated by the presence of EVs specific markers (CD63, CD9, CD81, TSG-101). Contaminants such as platelets (CD41) or cell debris (HDAC4) were absent in plasma; however, HDAC4 was present in serum EVs (Figure 2A and B). Albumin, which is present at high abundance, was not completely removed from either group of samples (Figure 2A and B), and was not significantly different between healthy control plasma or serum samples (Figure 2C). NTA was employed to determine the size distribution of the particles, revealing the presence of particles with an average size of 181.4 nm and a peak at 125.1 nm from plasma and 207.7 nm and 182.4 nm respectively for serum (Figure 2E and F). TEM images demonstrate the presence of ‘collapsed’ cup-shaped structures as a consequence of dehydrated bilipid membranes of the EVs, with an average size below 200 nm in plasma samples (Figure 2D). These findings allude to successful small-EVs isolation.

In our study, the isolation of EVs from plasma and serum exhibited similarities; however, after careful consideration, we opted for plasma as our preferred starting material. The decision was based on two key advantages. First and foremost, the preservation of EVs in their native state is crucial for accurate analysis, and plasma minimizes alterations in the EV population that may occur during clotting in serum. Secondly, the simplified and quicker handling of plasma contributes to the overall efficiency of our experimental procedures.

Since our plasma samples showed variation in protein markers despite loading the same amount of protein (40 μg), we considered there could be technical differences regarding the isolation procedure as well as variance depending on the health status of each subject (ESLD diagnosis, stage, medication, etc.). Hence, to improve the isolation yield of EVs and thereby increasing the total amount of protein, the protocol was modified and the last “washing” centrifugation step was omitted (WW), additionally, samples from healthy controls were used. As a control, the pellet from the first centrifugation step (1st), containing apoptotic bodies and cell debris, was used for characterization.

Twenty WW samples from healthy controls (kidney donors or volunteers) were analyzed (Figure 3). The presence of EV markers was similar in the samples from the 3-step centrifugation protocol (Figure 3A), except for CD9, which showed an increased intensity in the WW samples. However, platelets (CD41) and cellular debris (HDAC4) showed no reduction in the WW samples. Samples from the first centrifugation step were also analyzed and although they did show the presence of EVs-related proteins, other contaminants were also increased (Figure 3B). Albumin contamination was also higher in both groups compared to the 3-step UC samples, although not significantly different (Figure 3C). Overall, although it appeared that the yield of EVs was higher in the WW samples, there was a reduction in the purity of the samples. NTA analysis and TEM also showed the presence of artifacts with different particle sizes in the WW samples (Figure 3D–F).

Since the omission of the washing step did not lead to a significant improvement of the yield and reduced the purity of the EVs, the 3-step centrifugation protocol was maintained. However, a modification to the plasma collection process to eliminate contaminants was introduced. Twenty samples from healthy controls were collected and centrifuged at 2,500 g for 10 min to separate plasma from whole blood. Next, a second centrifugation at 2,500 g for 15 min was performed to obtain platelet-poor plasma before proceeding with the 3-step isolation protocol. These samples were named CDP samples. The results showed that EVs markers were present, while platelets and cell debris were no longer observable (Figure 4). However, albumin was still detectable to the same extent as the samples isolated with the original 3-step centrifugation protocol (Figure 4C). The NTA analysis (Figure 4B) indicated that the particles had a mean size of 191.8 nm and a peak of 184.6 nm, confirming the presence of small EVs. TEM (Figure 4D) showed a reduction in contaminants, but also the EVs quantity was decreased. Therefore, although EVs were of higher purity than the EVs obtained with the original 3-step UC procedure, the yield of EVs was slightly compromised. Considering these findings, it was concluded that the 3-step plasma UC protocol is the most suitable approach to obtain EVs for further analysis.

For protein quantification, samples from healthy controls (kidney donors or volunteers) that underwent the first-step centrifugation, WW, CDP, and with the complete 3-step UC protocol were quantified using BSA protein assay kit. Differences within samples revealed that samples from the first step (20,000 g for 30 min) of UC and WW had higher protein concentration compared to the CDP and 3-step UC samples (Figure 5). However, the isolated amount in all procedures was sufficient to demonstrate the presence of EVs via western blot.

Two methods were evaluated for RNA isolation, TRI-reagent, and the RNA isolation kit AllPrep DNA/RNA kit (Qiagen). Sixteen plasma samples from healthy controls (kidney donors or volunteers) using the UC protocol WW, CDP and 3-step UC procedures were selected. Additionally, since samples were resuspended in PBS, we evaluated whether the RNA purity (260/280 nm ratio) was affected. Six samples were resuspended directly in TRI-reagent or RLT buffer and 10 samples were first resuspended in PBS (50 μL). RNA concentration was determined by measuring absorbance at 260 nm on a spectrophotometer, quantification shows that the AllPrep DNA/RNA kit (RLT buffer) was not suitable for the RNA extraction of our samples regardless of the RNA isolation procedure (Table 1). TRI-reagent alone or in combination with PBS allowed the extraction of high amounts of RNA in WW and 3-UC step prepared samples (Table 1). These findings confirm successful protein and RNA isolation from our EV samples.

Once the UC isolation method (3-step) was selected, samples from ESLD-patients were analyzed via proteomics. First, we determined the necessity of traditional cell lysis for protein isolation, considering that EVs are already extracellular components. Since the proteomics sample preparation process involves the use of solvents that can disrupt the lipid membrane surrounding EVs, a comparative analysis by testing 10 samples with traditional lysis (L) and 10 samples without lysis (NL) was performed, as described in the Materials and methods section of our study.

A total of 1,171 proteins (average) were identified. Data with ≥ 75% of missing values (0) were filtered and 688 proteins remained. To gain insights into the overall variation in the data and the clustering of samples, PCA was conducted (Figure 6A). Our PCA analysis revealed a distinct separation between L and NL along the first principal component (PC1), which explained 22% of the total variance. This clear separation suggests that the two groups exhibit significant differences in their underlying data patterns. At the same time, it shows that there is variability within each group since no cluster was observable, indicating that there are specific characteristics of each sample that contribute to it (i.e., ESLD etiology).

To properly select one of the methods (NL or L), a comprehensive analysis was conducted to delineate the distinctions between them. Figure 6B illustrates the presence of EV-specific markers, including CD9, ALIX, and HSP90AB1, alongside various contaminant proteins such as albumin and lipoproteins [apolipoprotein E (APOE), APOB, APOA2, APOM]. While the expression levels of HSP90AB1 and ALIX were relatively comparable across both groups, it was noteworthy that lysed samples exhibited a twofold increase in CD9 expression compared to non-lysed samples. This finding suggests that lysed samples demonstrate heightened specificity in detecting EV-related markers, indicating enrichment of EV-related proteins in this group. Conversely, the levels of lipoproteins exceeded those of albumin in both groups and mirrored the patterns observed with specific EV markers. Notably, other lipoproteins, including APOL1, APOA1, APOC1, APOC2, APOC3, APOH, APOD, APOA4, APOC4, APOF (data not shown), were also expressed, with only the more substantial increases being highlighted.

Finally, in the protein expression analysis, of the 688 proteins with an adjusted P-value < 0.1, we observed significant changes in protein expression levels based on the LFC values between groups (L versus NL, Figure 6C). Among the significant proteins, 27 (3.8% of the total) showed overexpression, while only 3 (0.43% of the total) exhibited under expression when comparing the L samples to the NL ones. To conclude whether their increase or reduction is significant for liver-related diseases it is necessary to correlate the clinical data of every individual with its corresponding output.

From these results, we concluded that L group has an increased protein yield, with an increased abundance of EVs-specific markers, making lysis of samples the preferred method for subsequent analysis.

In conclusion, our study has made significant strides in advancing the understanding and utility of EVs, particularly in the context of ESLD. The study successfully detected specific protein markers in plasma and serum samples from healthy controls and ESLD patients within the TransplantLines cohort, demonstrating the efficacy of DUC in isolating EVs (Figure 7). However, challenges in achieving absolute purity, particularly with persistent albumin contamination, highlight the need for further refinement and exploration of alternative techniques to enhance the isolation protocol.

Moreover, our findings underscore the complex nature of EVs, with origins and cargo contributing to significant variability in protein concentration and profiles among samples. The multifaceted nature of EVs, coupled with potential technical errors, emphasizes the importance of a comprehensive patient cohort characterization, considering clinical parameters and ESLD etiology. While promising associations with HCC markers have been identified, additional validation and integration of clinical data are imperative. Overall, our study provides a robust foundation for future investigations into the potential clinical applications of EVs in ESLD, encouraging further research to unlock their diagnostic and therapeutic potential.